NICE OBSERVATIONAL DATA UNIT (ODU) · 2019-04-04 · Providing consultancy and research in health...

Providing consultancy and research in health economics for the

NHS, pharmaceutical and health care industries since 1986

NICE OBSERVATIONAL DATA UNIT

(ODU)

Commissioning through Evaluation (CtE)

Percutaneous mitral valve leaflet repair for

mitral regurgitation (MitraClip)

FINAL LINKED DATA REPORT

Newcastle and York External Assessment Centre

(NY EAC)

Dr. Iain Willits

Kim Keltie

Dr. Sam Urwin

Helen Cole

Joyce Craig

Dr. Mark de Belder, NHS England Cardiac CtE Clinical Lead (MitraClip)

Professor Nicholas Linker, NHS England Cardiac CtE Clinical Lead (LAAO)

Dr. Rob Henderson, NHS England Cardiac CtE Clinical Lead (PFOC)

Dr. Andrew Sims

24th December 2018

Project Name: Commissioning through Evaluation (cardiovascular procedures)

Project Number: RX085

Start Date: 12 September 2014

Output Date: 28 November 2018 FINAL LINKED DATA REPORT

Correspondence to: Dr. Iain Willits

NICE External Assessment Centre

Medical Physics Directorate

Freeman Hospital

Newcastle upon Tyne

NE7 7DN

Tel: +44 (0)191 2137787

Commercial in confidence data have been redacted. Declared interests of the authors: None

The views expressed in this report are those of the authors and not those of NICE. Any errors are the responsibility of the authors. HES and ONS data held by the UK NHS Health and Social Care Information Centre have been used to help complete this analysis (© 2018. Reused with the permission of the Health and Social Care Information Centre. All rights reserved)

Contents

Contents

Executive summary i

Abbreviations iv

Section 1: Introduction 1 1.1 NHS England Commissioning through Evaluation – MitraClip 1 1.2 Description of the procedure 3

Section 2: Methods 5 2.1 CtE MitraClip Providers and Programme Governance 5 2.2 CtE MitraClip Commissioning details 5 2.3 Inclusion and exclusion patient selection criteria 6 2.4 Primary data collection (CtE registry) 6

2.4.1 Database details and information governance arrangements 6

2.4.2 Active surveillance 7

2.4.3 Case eligibility criteria 9

2.4.4 Data cleaning 9

2.4.5 Outcomes indicators 9

2.4.6 Statistical analysis 10

2.5 Primary data collection (linked data) 11 2.5.1 Linked Dataset Details 11

2.5.2 Linked Dataset Cleaning 11

2.5.3 Linked Dataset Outcomes 12

2.5.4 Statistical Analysis of Healthcare Resource Use Data 14

2.6 Secondary data collection (Literature review) 14 2.7 Research design 17

Section 3: Results 23 3.1 Primary data collection (CtE Registry) 23

3.1.1 Numbers of patients treated at each centre 23

3.1.2 Summary statistics of patient and procedural characteristics 23

3.1.3 Active surveillance (Evaluation of Coverage) 23

3.1.4 Outcomes – using CtE registry data only 25

3.2 Primary data collection (Data Linkage) 28 3.2.1 Cohort characteristics 28

3.2.2 In-hospital complications 29

3.2.3 All-cause mortality 30

3.2.4 Infective endocarditis 34

3.2.5 Additional interventions 34

3.2.6 Hospital resource usage 34

Contents

3.3 Secondary data collection (Literature review) 38 3.4 Emerging new evidence 42

3.4.1 TVT registry study 42

3.4.2 RCTs 42

3.4.3 On-going randomised controlled trials 44

3.5 NHS England questions 46 3.5.1 Question One 49

3.5.2 Question Two 55

3.5.3 Question Three 60

3.5.4 Question Four 66

3.5.5 Question Five 69

3.5.6 Question Six 70

3.5.7 Question Seven 71

3.5.8 Question Eight 74

3.5.9 Question Nine 79

3.5.10 Question Ten 79

3.5.11 Question Eleven 83

3.5.12 Summary of answers to NHS England questions 85

Section 4: Discussion 87 4.1 Summary of findings from primary data collection (CtE Registry) 87 4.2 Summary of findings from primary data collection (Data Linkage) 87 4.3 Results in the context of other studies 88 4.4 Limitations and strengths 89

4.4.1 Limitations of primary evidence 89

4.4.2 Strengths of primary evidence 90

Section 5: Conclusion 92

Section 6: Acknowledgements 94

Section 7: References 95

Appendix 1 – Data flow diagram – CtE Registry 100

Appendix 2 – Patient characteristics – CtE Registry 101

Appendix 3 – Procedural characteristics – CtE Registry 104

Appendix 4 – Outcomes – CtE Registry 106

Appendix 5 – Outcomes for elective/emergency cases – CtE Registry 107

Appendix 6 – Outcomes (time to event analysis) – CtE Registry 108

Appendix 7 – Kaplan-Meier curves – CtE Registry 109

Appendix 8 – Mitral regurgitation over time – CtE Registry 111

Contents

Appendix 9 – Cost of a MitraClip procedure 112

Appendix 10 – Data Flow Diagram – Linked Data 115

Appendix 11 – Linked Data reported in-hospital complications 116

Appendix 12 – Kaplan-Meier analysis by procedural urgency and by aetiology of mitral regurgitation 118

Appendix 13 – Mitral valve interventions subsequent to MitraClip implantation - linked dataset 119

Appendix 14 - Hospital resource (HES only) by procedural urgency of index MitraClip implantation – a) elective and b) urgent 122

Appendix 15 - Hospital resource (HES only) by aetiology of mitral regurgitation – a) Functional/Ischaemic and b) Degenerative 126

Executive Summary i

Executive summary

Mitral valve regurgitation (MR) occurs when the mitral valve loses competency, resulting in

retrograde flow of blood from the left ventricle into the atrium, which in turn reduces the

efficiency of the heart. There are two principal causes of MR. Degenerative (primary) MR is

caused by deterioration of the valve itself, whereas in functional (secondary) MR, the valve

itself is structurally normal, but is functionally compromised as the leaflets fail to coapt,

usually secondary to left ventricular enlargement. Mitral valve regurgitation leads to

worsening heart failure and associated symptoms. In people with MR who are considered at

too high risk for open heart surgery, percutaneous leaflet repair with the MitraClip system

may be the only treatment option other than palliative medical management.

In order to evaluate the MitraClip procedure, NHS England has set up a multi-centre

observational registry using the process of Commissioning through Evaluation (CtE). The

registry was designed to include patients who had moderate or severe MR of degenerative

or functional aetiology, and for whom conventional surgery was deemed to be an

excessively high risk intervention. The registry recorded a range of clinical outcomes with a

maximum follow up of 2 years. The aims of the CtE registry were to provide data on the

safety, efficacy and costs of MitraClip in a real-world setting, and specifically to answer 11

pragmatic questions concerning these issues. To complement/validate its content, the

registry was linked to two routine administrative datasets: Hospital Episode Statistics (HES)

Admitted Patient Care (APC) in terms of hospital admissions, and the Office of National

Statistics (ONS) mortality dataset in terms of reported deaths, thus providing more

comprehensive coverage in terms of follow-up. Additionally, as the registry was single-

armed, a parallel literature search was undertaken in order to present the registry findings in

the context of published studies in comparable populations, and to assess whether

procedural outcomes were consistent with previously reported studies. Information gained

from the registry will be used to inform future commissioning.

From the CtE registry alone, out of a total of 272 enrolled patients, 199 were eligible for CtE

data analyses. The 199 patients included in the CtE analyses had functional (60%) or

degenerative (40%) MR with a mean age of 76.2 years. Most patients were men (69%) and

most patients (66%) had moderate or severe left ventricular impairment. The majority of

patients were recruited electively (84.4%), with 13.6% admitted urgently and 2.0%

undergoing the procedure as an emergency. Nearly all patients had moderate or severe MR

(grade 3+ or 4+), which caused significant dyspnoea in 92% of cases (New York Heart

Association [NYHA] class 3 or 4). The mean EuroSCORE II (per cent risk of dying from

cardiac surgery) was 6.4 (median 4.8; range 0.7 to 42.5).

A MitraClip device was implanted in 187 patients (94%). The procedural success rate,

defined as device successfully implanted with no major complications, was 85.9% (95%

confidence interval 80.3 to 90.4%), with 8.2% of procedures being associated with an in-

hospital major complication, including ten deaths (5.1%) and four additional interventions

(2.0%). There was no association between procedural success rate and procedural urgency

(elective success rate 158/168 [94.0%] vs. urgent/emergency success rate 29/31 [93.5%]).

In patients successfully treated, there was an immediate and significant improvement in MR,

with a reduction from 100% MR grade ≥3+ to 7% MR grade ≥3+. These peri-procedural

outcomes were consistent with observational studies identified in the literature with similar

Executive Summary ii

populations, and emphasise the early clinical benefits, but also the high mortality associated

with the MitraClip procedure in this high-morbidity cohort.

In the medium term, the MR benefits of MitraClip were largely sustained, with 76% of

patients having mild or absent MR (grade ≤2+) at 1 year. This was reflected in significantly

improved patient symptoms, with 82% having mild or no symptoms of dyspnoea (NYHA

class ≤2), and significant improvements in quality of life (QoL), as measured by EQ-5D. The

mortality rate at 1 year, reported directly from the registry, was 11.6%. The overall mortality

rate using the linked data was 13.6 (95% CI 9.8 to 18.8) deaths per 100 person years (PY);

11.5 (95% CI 7.7 to 16.6) per 100 PY in the elective subgroup (n=145), 31.3 (95% CI 14.3 to

59.5) per 100 PY in the urgent/emergency subgroup (n=21). These data were consistent

with published studies, which reported a range of mortality at 1 year of between 6% and

26%. Linked data also reported that 67% of patients had at least one hospital admission in

the year following the MitraClip procedure, with some having multiple admissions, resulting

in a readmission rate of 176.4 per 100 PY. This was a higher rate than had been reported in

the literature by observational studies and randomised controlled trials (RCTs).

There was no statistically significant difference in mortality rate, readmission rate, MR grade,

NYHA class, or adverse events in patients with degenerative compared with functional MR

aetiologies, although it is likely the study was underpowered to detect such differences.

Using registry data, patients receiving MitraClip as an urgent or emergency case had a

greater risk of death, with 68.2 dying per 100 person years (PY) compared with 22.1 per 100

PY in the elective cohort (p = 0.0105). This increased risk of death was driven by in-hospital

mortality. When only post-discharge mortality was considered, there was no significant

difference.

The cost of the MitraClip procedure has been estimated at £32,910 (range £29,000 to

£34,500). This includes pre-operative assessment; peri-procedural and device costs (which

are per procedure rather than per device); and post-operative management. De novo

economic analysis was performed using HES-linked data from the 1 year prior to the

procedure compared with 1 year following the procedure. This reported that, overall,

MitraClip was associated with post-procedure reductions in hospital readmission rates, total

hospitalisation days and mean per patient hospitalisation costs. The aggregated pre-

procedural hospitalisation cost was £1,267,321, with a median per patient cost of £5196, and

mean cost per patient of £8448 (statistical deviation [SD] £11,033). This compared with an

aggregated post-procedure hospitalisation cost of £889,721 (median per patient cost £1402,

mean per patient cost £5931 [SD £10,688]). Using a mixed effects model this gave a cost

ratio of 0.62 (95% CI 0.50 to 0.79), which represented a statistically significant reduction in

costs following the MitraClip procedure. Subgroup analysis showed that reductions in

healthcare resource use and associated costs were driven by reductions in admissions and

related costs for cardiac and heart failure causes, and these were not offset by small

increases in the rate of readmissions for non-cardiac causes. However, the overall cost

savings related to reduced hospital readmissions were unlikely to be offset by the device and

procedural costs associated with MitraClip. The likely cost-effectiveness of MitraClip when

used in the NHS is not known.

A limitation of the CtE registry, common to all device registries, was that it did not report a

standard care control arm (medical management without MitraClip). Recently, two RCTs

enrolling patients with severe functional MR who were at prohibitive risk of surgery have

been published that compared MitraClip with medical management. The MITRA-FR study

reported that MitraClip did not improve mortality or readmission rates after 1 year follow up.

Executive Summary iii

In contrast, MitraClip was associated with statistically and clinically significant improvements

in mortality and readmissions after 2 years in the COAPT trial. The reasons for these

discordant results are unclear, but may emphasise the importance of careful patient

selection for this procedure.

In conclusion, the CtE registry has reported data that show MitraClip is associated with an

immediate reduction in MR. This effect is sustained in the medium term (1 year) and is

associated with significant improvements in symptoms and QoL, largely consistent with other

published data. Recent data from the COAPT trial suggests that MitraClip is likely to be

effective at reducing mortality and patient readmissions compared with medical management

in carefully selected patients with functional MR. However, there remains a lack of relevant

comparative evidence in patients with degenerative MR.

Abbreviations iv

Abbreviations

ACC American College of Cardiologists ACE-I Angiotensin-converting enzyme inhibitor AF Atrial fibrillation AKI APC

Acute kidney injury Admitted Patient Care

ARB Angiotensin II receptor blockers BNP B-type natriuretic peptide CABG Coronary artery bypass graft CCS Canadian Cardiovascular Society CG Clinical guideline CI Confidence interval CPB Cardiopulmonary bypass CRT Cardiac resynchronisation therapy CSHA Canadian Study on Health and Ageing CT Computed tomography CtE Commissioning through Evaluation CVA Cerebrovascular accident DMR Degenerative mitral valve regurgitation eGFR Estimated Glomerular Filtration Rate EAC External Assessment Centre EACTS European Association for Cardio-Thoracic Surgery ECG Electrocardiograph EQ-5D EuroQoL 5 dimensions ESC European Society of Cardiology FBC Full blood count FMR Functional mitral valve regurgitation Fr French FU Follow up HF Heart failure HR Hazard ratio IABP Intra-aortic balloon pump ICD-10 International Statistical Classification of Diseases and

Related Health Problems - 10th revision ICE Intracardiac echocardiogram ICER Incremental cost-effectiveness ratio IPG Interventional procedures guidance IQR Inter-quartile range ITT Intention to treat ITU Intensive therapy unit JVP Jugular venous pressure LAAO Left atrial appendage occlusion LoS Length of stay LVEF Left ventricular ejection fraction MACCE Major Adverse Cardiac and Cerebrovascular event MDS Minimum data set MDT Multi-disciplinary team MI Myocardial infarction MM Medical management MR Mitral valve regurgitation MRI Magnetic resonance imaging MV Mitral valve MVA Mitral valve area NHS National Health Service

Abbreviations v

NHSE NHS England NICE National Institute for Health and Care Excellence NICOR NY

National Institute for Cardiovascular Outcomes Research Newcastle and York

NYHA ONS OPCS-4

New York Heart Association Office of National Statistics Office of Population Censuses and Surveys Classification of Surgical Operations and Procedures – 4th revision

OR Odds ratio PCI Percutaneous coronary intervention PFO Patent foramen ovale PFOC Patent foramen ovale closure PRISMA Preferred Reporting Items for Systematic Reviews and

Meta-Analyses PVD Peripheral vascular disease PY Person years QALY Quality adjusted life year QoL Quality of life RCT Randomised controlled trial RR Relative risk SD Standard deviation SE Standard error SE Systemic embolism SF-36 Short form 36 [dimensions] STS Society of Thoracic Surgeons TAPSE Tricuspid annular plane systolic excursion TIA Transient ischaemic attack TOE Transoesophageal echocardiogram TR Tricuspid regurgitation TTE Transthoracic echocardiogram U&E Urea and electrolytes

Introduction 1

Section 1: Introduction

1.1 NHS ENGLAND COMMISSIONING THROUGH EVALUATION – MITRACLIP

NICE provides support to NHS England in Commissioning through Evaluation (CtE):

“NHS England’s Commissioning through Evaluation (CtE) programme enables a limited

number of patients to access treatments that are not funded by the NHS, but nonetheless

show significant promise for the future, while new clinical and patient experience data are

collected within a formal evaluation programme.”

The work commissioned by NICE (‘Project RX085’) from Newcastle and York (NY) EAC

comprises evaluation of three percutaneous cardiac procedures:

• Percutaneous occlusion of the left atrial appendage in non-valvular atrial fibrillation

for the prevention of thromboembolism (NICE IPG349, June 2010). Shortened term

used is ‘LAAO’.

• Percutaneous Closure of Patent Foramen Ovale to prevent recurrent cerebral

embolic events (NICE IPG472, December 2013). Shortened term used is ‘PFO

Closure’.

• Percutaneous mitral valve leaflet repair for mitral regurgitation (MitraClip) (NICE

IPG309, August 2009) [1]. Shortened term used is ‘MitraClip’.

A Cardiology CtE Steering Group is established as a subgroup of the NHS England

Cardiothoracic Services Clinical Reference Group (CRG). It reports to the Programme of

Care Board for Internal Medicine for NHS England. Three Individual Technology Groups

report to the CtE Steering Group on the progress of the above three specialised cardiology

interventions which form the cardiac CtE programme.

The National Institute for Cardiovascular Outcomes Research (NICOR) was contracted by

NY EAC to design and host an on-line registry for MitraClip procedures, to provide a project

management function to promote data entry quality and completeness by commissioned CtE

provider sites and to link registry data with Hospital Episode Statistics (HES) and Office of

National Statistics (ONS) mortality datasets. NICOR and the EAC consulted the MitraClip

Individual Technology Group in the design of the MitraClip registry. NICOR were the formal

data owner of the registry, and were the applicant to NHS Digital for data linkage with HES

and ONS.

NY EAC’s objectives in Project RX085 from NICE were to:

• review existing register data fields in each dataset and advise on their suitability for

updating and developing NICE guidance;

• advise on the appropriateness of register data fields for each dataset being proposed

or considered in relation to clinical and cost effectiveness outcomes to enable NICE

to provide NHS England with further data to help inform future commissioning

decisions for the procedures;

• establish processes to a) ensure on-going review of the MitraClip dataset quality,

completeness and coverage, with action plans for improvements where needed and

b) deliver regular evaluative reports that are useful for decision making;

• update the literature searches since publication of each NICE interventional

procedures guidance (IPG) in order to identify publications of relevance;

https://www.england.nhs.uk/commissioning/spec-services/npc-crg/comm-eval/

http://www.nice.org.uk/guidance/IPG349




https://www.england.nhs.uk/commissioning/spec-services/npc-crg/group-a/a05/

https://www.england.nhs.uk/commissioning/spec-services/npc-crg/group-a/

https://www.england.nhs.uk/commissioning/spec-services/npc-crg/group-a/

Introduction 2

• manage the contract with NICOR and participate in the CtE Steering Group for

cardiovascular procedures;

• develop a protocol for analysis of data and consult with key partners (listed above) to

gather views on the proposed methodology and proposed outputs;

• produce a final report (not intended for publication) answering the CtE evaluation

questions set by NHS England (tabulated below);

• present findings in the form of a publishable paper (to be submitted for peer review

for a high impact journal). This should be of a standard to be included as an input in

the evidence base of the NICE technology appraisals programme

(http://www.nice.org.uk/article/PMG9/chapter/Foreword);

• advise on further research that might be needed to generate clinical and cost

effectiveness evidence in line with methods used in NICE evaluation programmes,

including suitable study designs for such research.

Outputs required by NICE from NY EAC and delivered prior to this final report were:

• Output One [2] - a report for presentation to the CtE Steering Group on all three

procedures, analysing the coverage, quality and completeness of the register to date,

and making preliminary recommendations about the definitive dataset to inform NHS

England’s contracts for the procedures with the specialist centres, and to meet

NICE’s needs in relation to updating guidance. Completed 28/11/2014.

• Output Two [3] - a report for submission to the CtE Steering Group for cardiovascular

procedures and collaborating partners proposing: a) a process to ensure on-going

review of the database quality, completeness and coverage, with action plans for

improvements where needed and b) the format of evaluative reports designed to be

useful in informing decision making for guidance development. Completed

04/02/2015.

• Output Three [4] - a report for submission to NICE and the CtE Steering Group

proposing a draft protocol for analysis of data that describes the methods that will be

used to compare effectiveness of each of the procedures between propensity-

matched cohorts of patients undergoing the range of treatment options (including

cost analysis). This will have been circulated for consultation with key partners (listed

above) and adjusted as appropriate prior to presentation to NICE. Completed

31/03/2015.

The above three outputs from the project, all of which were shared with the CtE Steering

Group and approved by them, are used as source material for the general background and

methods sections of this final report from NY EAC to NICE.

The NHS England questions for CtE of MitraClip were originally presented to NICE,

discussed with NY EAC, and edited to the final form presented in the Table 1.

http://www.nice.org.uk/article/PMG9/chapter/Foreword

Introduction 3

Table 1. Percutaneous mitral valve leaflet repair for mitral regurgitation (MitraClip)

Questions from NHS England Final version of question, as amended by NICE following discussion with EAC

1. Can UK clinical teams undertaking MitraClip reproduce the reduction in mitral regurgitation seen in the early clinical trials?

NICE agrees the question is appropriate

2. Do any reductions in mitral regurgitation lead to reduced symptoms and improved quality of life (compared to pre-procedure)?

Is reduction in mitral regurgitation mediated by MitraClip associated with improvements in quality of life?

3. Is there any evidence of improved survival compared to expected survival from natural history studies or risk models?

Does MitraClip improve survival rates?

4. Is there evidence of reduced frequency of hospital admission following MitraClip?

Does MitraClip reduce the frequency of subsequent hospital admissions?

5. Are the early benefits in reduction in mitral regurgitation maintained in the longer-term? Is there a need for repeat treatment over time (either by a repeat percutaneous procedure or surgery)?

Are the early benefits in reduction in mitral regurgitation maintained in the medium-term? Is there a need for repeat treatment over time (either by a repeat percutaneous procedure or surgery)?

6. How many patients who are not candidates for conventional mitral valve surgery would benefit from a reduction in the degree of mitral regurgitation that could be achieved with successful MitraClip therapy? (i.e. if MitraClip is to become routinely funded, what is the likely clinical need in England?)

What proportion of patients referred to a specialist MitraClip service as defined in the CtE documents were assessed by the MDT as suitable for the intervention? What proportion of the patients considered suitable for the procedures received it and what proportion of them benefitted (refer also to Question 5)?

7. What are the short-term and long-term complications of MitraClip therapy? Is there a risk of longer-term mitral stenosis? Is the frequency of complications sufficiently low to provide a positive risk-benefit ratio?

What are the short and medium term risk of complications from MitraClip use? Answers to Questions 7 and 8 will be considered by the NICE Interventional Procedures Advisory Committee when NICE updates the IP guidance on this procedure.

8. What are the characteristics of patients who are successfully treated compared to those in whom treatment is unsuccessful? Are there subsets of patients who get a particularly advantageous result? Conversely, are there subsets of patients for whom this treatment is not effective? Do patients of different gender or from different ethnic origins respond equivalently?

Are clinical outcomes with MitraClip associated with particular patient characteristics (clinical or demographic)?

9. What is the true procedural cost of MitraClip therapy in the NHS?

What are the full procedural costs of using MitraClip to the NHS?

10. What costs savings might occur in the NHS as a result of MitraClip therapy?

What are the potential cost savings for the NHS arising from patients receiving MitraClip?

11. What is the cost-effectiveness of MitraClip therapy based on UK procedural and follow-up costs?

Is MitraClip cost-effective from the perspective of the NHS?

1.2 DESCRIPTION OF THE PROCEDURE

The MitraClip procedure is described in NICE IPG309:

“Percutaneous mitral valve leaflet repair is undertaken with the patient under general

anaesthesia. Under fluoroscopy and transoesophageal echocardiography guidance, a

catheter is advanced through the femoral vein to the right atrium and via a transseptal

puncture into the left atrium. The mitral leaflets are partially clipped to each other (more than


Introduction 4

one clip may be used). Imaging is used to assess whether the reduction in mitral

regurgitation is adequate. The clips may be repositioned if necessary.”

In the 2009 NICE IP Guidance, evidence on the safety and efficacy of percutaneous mitral

valve leaflet repair for mitral regurgitation (MitraClip) was considered inadequate in quality

and quantity. Therefore, the procedure should only be used:

• with special arrangements for clinical governance, consent and research for patients

who are well enough for surgical mitral valve leaflet repair to treat their mitral

regurgitation, or

• in the context of research for patients who are not well enough for surgical mitral

valve leaflet repair to treat their mitral regurgitation.

5

Section 2: Methods

2.1 CTE MITRACLIP PROVIDERS AND PROGRAMME GOVERNANCE

Hospitals providing the CtE procedures in the 3 centres participating in the MitraClip scheme

are:

• The University Hospital of South Manchester NHS Foundation Trust (Wythenshawe

Hospital)

• University Hospitals Bristol NHS Foundation Trust (Bristol Royal Infirmary)

• Royal Brompton and Harefield NHS Foundation Trust (Royal Brompton Hospital).

The criteria used to select the hospitals for the CtE work considered a number of competing

factors and are described in the NHS England Specialised Services Circular (SSC) 1454 for

MitraClip [5]. An advisory panel made recommendations to NHS England as to which

providers should be selected to be CtE centres. The final selection of centres was

undertaken by the regional Medical Directors.

The NHS England Cardiac CtE Clinical Lead for MitraClip is Dr. Mark de Belder, Consultant

Cardiologist, South Tees Hospitals NHS Foundation Trust. Dr. de Belder is Chair of the NHS

England MitraClip Individual Technology Group. The role of the Group, set out in its Terms

of Reference (ToR), is to:

• Work with the EAC and the National Institute for Cardiovascular Outcomes Research

(NICOR) on the development of the relevant dataset.

• Define and clarify patient access criteria, where required, within the terms of the

published policy statements / specification.

• Ensure that all participating centres are collecting, verifying and uploading data in a

timely manner.

• Ensure that all participating centres are collecting follow-up data appropriately.

• Monitor performance of all centres performing procedures as part of CtE and report

any concerns to the Steering Group.

• Monitor referrals, patient pathways and waiting times for the relevant procedure at all

participating centres (including pathways for patients who do not receive the CtE

treatment).

2.2 CTE MITRACLIP COMMISSIONING DETAILS

NHS England commissioned a total of 180 MitraClip procedures in the cardiac CtE scheme.

Each of the 3 centres was required to do no more than 40 procedures per financial year. As

CtE commenced on 01/10/2014, each centre could do no more than 20 MitraClip procedures

in those last 6 months of 2014/15. Funding was made available by NHS England for each

centre to do 40 procedures in 2015/16, making 60 procedures per centre in total.

Owing to slower than anticipated roll-out of the programme, centres were permitted, by NHS

England Specialised Services Circular SSC 1669 (November 2016), to carry on with their

2015/16 activity plans in financial year 2016/17, up to the contracted number of 180

procedures in total for the MitraClip CtE programme.

6

2.3 INCLUSION AND EXCLUSION PATIENT SELECTION CRITERIA

According to the NHS England Specialised Services Circular (SSC) 1454 for MitraClip [5],

patient selection criteria were:

“Patients with severe grade 3 or 4 mitral regurgitation, deemed by an MDT to be at too high

risk or at high risk for conventional mitral valve surgery:

a. Symptomatic patients (in spite of optimal medical therapy, which includes the

appropriate use of device therapy [implantable cardioverter defibrillator or cardiac

resynchronisation therapy (CRT)]).

b. Appropriate mitral valve anatomy (mitral valve area ≥4 cm2, leaflet flail width <15

mm and gap <10 mm; patients with leaflet tethering - coaptation depth >11 mm and

length <2 mm – are excluded).

c. Patient fully informed and consent provided.

Patients with mitral valve disease and heart failure who might be considered eligible for

mitral valve replacement or repair or MitraClip can be referred from cardiologists or other

specialists in district general hospitals to cardiac surgeons or cardiologists in a specialist

cardiac centre. This must be in line with the specialised service specification for cardiology

and cardiac surgery. Direct referrals to cardiac centres from primary care and general

practice requesting consideration for MitraClip will not be accepted. Selection of patients for

a MitraClip procedure must be via the MDT process.”

2.4 PRIMARY DATA COLLECTION (CTE REGISTRY)

2.4.1 Database details and information governance arrangements

NICOR worked with the CtE MitraClip Individual Technology Group and NY EAC to produce

the final dataset for MitraClip.

NY EAC produced ‘RX085 Output One - Recommendations on three NHS England

Commissioning through Evaluation (CtE) registry draft datasets for MitraClip, LAAO and

PFO Closure cardiovascular procedures’ [2] (November 2014). This identified and appraised

new evidence added to the literature base and public domain since the original NICE IP664

overview [6] was published and compared findings against the data fields contained in the

draft MitraClip CtE dataset.

The final MitraClip dataset was developed into the online database by NICOR and the latest

version may be downloaded as a Microsoft Excel spread sheet (last updated 27/08/2015).

Regarding information governance arrangements, as Data Controller, NICOR’s

responsibilities were:

• To ensure that a dataset being proposed or used for national data collection has

appropriate independent oversight, and that all relevant data will be made available

to NICE for use in developing guidance.

• To provide NY EAC with a monthly download of episode level full raw data sets from

each registry (out with normal NICOR data sharing policy and following the ‘Use of

Data’ principles agreed with NY EAC). Data cleansing will happen to usual NICOR

schedule. Monthly downloads may be aggregated or incremental. The EAC will

provide feedback to NICOR on any data quality / completeness issues observed in

the monthly raw data downloads.

https://www.nice.org.uk/guidance/ipg309/evidence/overview-pdf-312650893

http://www.ucl.ac.uk/nicor/registries/mitraclip/datasets

7

• To arrange and undertake data linkage with Hospital Episode Statistics (HES) and

Office of National Statistics (ONS) mortality data and provide complete data

extract(s) to NY EAC in order to check for extra safety and efficacy, clinical

effectiveness or resource utilisation information.

• To arrange and maintain appropriate EQ-5D-5L licensing arrangements to cover all

projected patient volumes commissioned by NHS England in its CtE programme.

This should include all commissioned follow up visits.

• To provide a telephone helpdesk service for answering technical enquiries / requests

and for individual registration and access to each registry web portal. Clinical

enquiries will need to go to the NICOR project manager and NY EAC may be co-

opted to help NICOR respond to clinical or scientific queries.

• To operate within the general principles of Good Clinical Practice (GCP) in research,

as outlined in the Research Governance Framework for Health and Social Care

2005.

• To make all necessary applications to comply with information governance

requirements. These include but are not restricted to:

i complete the Information Governance Statement of Compliance

process to the satisfaction of the NHS Health and Social Care Information

Centre

ii demonstrate compliance with the Data Protection Act 1998. This is

also particularly relevant when data will leave or enter the EU. Appropriate

regard needs to be paid to international regulations

iii complete the Confidentiality Advisory Group application process to

comply with the NHS Health Research Authority requirements for Section 251

approval.

2.4.2 Active surveillance

NICOR provided NY EAC with their Minimum Data Standard (MDS) Summary Document for

Cardiac CtE (Confidential). Some of the background detail is extracted in the following

summary:

“While NICOR undertakes a number of manual and automated data quality control

processes, the responsibility for data quality is shared with clinicians and organisations

undertaking procedures in the NHS. It is particularly important that data are collected for

patients who experience adverse outcomes (such as death, stroke, bleeding) and harm.

NICOR aims to further assist organisations in their data submissions by defining a minimum

data standard (an acceptable standard for data submissions to be measured against), to

provide feedback to the provider organisations on the data quality of their quarterly

submissions and to give organisations the opportunity to improve and resubmit the data

should improvements be required.”

The final NICOR MDS for MitraClip CtE baseline data completeness monitoring contained

21 key fields. Six additional fields were monitored for patient completion of EQ-5D

questionnaires and EuroQoL data entry (NICOR field identifiers 4.08 to 4.13). These are

summarised in Table 2:

8

Table 2. Fields in registry used for monitoring of data completeness.

NICOR Field identifier Data Field

1.03 NHS Number

1.06 Birth date

1.10 Postcode

2.03 Reason for treatment

4.05 New York Heart Association (NYHA) status

4.06 Killip class

4.07 Date EuroQoL form filled

4.08 EuroQoL Mobility

4.09 EuroQoL Self-care

4.10 EuroQoL Usual activities

4.11 EuroQoL Pain / discomfort

4.12 EuroQoL Anxiety / depression

4.13 EuroQoL Health state today

4.15 CSHA

5.02 Severity of mitral valve regurgitation (MR)

5.03 Aetiology of MR

5.64 Logistic EuroSCORE

7.01 Date of admission

7.04 Date and time of procedure

7.06 Responsible consultant

7.27 No of clips successfully deployed

8.01 Device embolisation/detachment

8.12 Mitral valve (MV) surgery

8.26 Pre-discharge mitral regurgitation (echo)

8.29 Successful procedure

9.01 Life status

9.02 Discharge date

The six follow-up MDS fields for MitraClip data completeness monitoring at 6 weeks were:

10.01 Life status

10.05 New York Heart Association (NYHA) status

10.06 Mitral valve regurgitation (MR) severity

10.35 Additional mitral valve (MV) intervention

10.40 Major Adverse Cardiac and Cerebrovascular event (MACCE)

10.41 Date 6 week EuroQoL form filled

The equivalent variables were also monitored for follow up data at 6 months, 12 months and

24 months.

Summary reports were submitted to NICE by NY EAC on a quarterly basis, to a standard

reporting template agreed with NHS England for all CtE projects. Key parameters for each

CtE provider were:

• Contracted activity to date: the amount of CtE activity the centre should have

performed by this point, according to their contract with NHS England.

• Actual activity to date, as identified through both register entries and active

surveillance by NICOR collating a ‘SurveyMonkey’ questionnaire from the CtE

providers.

• Number of cases submitted to the NICOR registry to date. This number could be

lower than the above actual activity to date, since active surveillance could identify

cases that had not yet been registered.

• Number of cases identified through active surveillance but for which data were not

yet submitted to the registry (i.e. the difference between the two previous figures).

https://www.surveymonkey.co.uk/

9

• Initially, data completeness (%) was calculated for the subset of all MitraClip records

where the CtE provider had selected the ‘CtE=Yes’ check box when submitting the

case to the NICOR dataset (this is the ‘Number of cases’ denominator, below):

Data completeness (%) = Number of completed entries in MDS data fields x 100

Number of cases

• Later, queries from the CtE providers on this denominator led to refined definitions for

Activity, Coverage, Completeness and Follow up (FU) reported. The final defined

measures were:

o Activity: The number of CtE procedures recorded with a procedure date

between 01/10/2014 and the date of raw data extract that had an eligible

reason for treatment.

o Coverage: The percentage of patient follow ups reported out of the number of

patients reaching the follow up time point in question. A ‘reported’ follow up

had data in any of the MDS follow up fields for the time point in question.

o Completeness: The percentage of fields with any data out of the number of

MDS fields for the time point in question.

o FU reported: This number included patients reported to have died since the

previous follow up visit.

2.4.3 Case eligibility criteria

Inclusion criteria: All pseudonymised NHS procedures recorded in the MitraClip CtE registry

conducted between 1st October 2014 to 24th October 2017 for patients with Grade 2 (mild-

moderate), Grade 3 (moderate-severe) and Grade 4 (severe) mitral regurgitation with

recorded reasons for treatment including surgical turn-down, high risk for surgery (bail-out

surgery would be offered) and high-risk for surgery (bail-out surgery would not be offered). A

further final data extract was drawn down by NICOR on 13th November 2017, as centres had

been asked to complete cause of death against recorded mortality events, where possible.

Exclusion criteria: Patients with rheumatic aetiology (contraindicated for MitraClip),

procedures with missing procedure date, discharge date preceding procedure date.

2.4.4 Data cleaning

Detailed methods of variable cleaning are described in Supplementary Material - Table 1.

Data completeness and summary statistics, in terms of distribution of responses, were

conducted for each of the data fields available and used to inform variables used and

definition of outcomes during the statistical analysis.

2.4.5 Outcomes indicators

a. Clinical

Primary outcome measures included: device implantation, in-hospital major complications,

in-hospital minor complications, new requirement for pacemaker, post-discharge major

complications, post-discharge minor complications and mitral regurgitation over time.

Secondary outcome measures included: death, embolisation, requirement for additional

intervention, myocardial infarction (MI), pericardial effusion or tamponade requiring

intervention, major vascular complication requiring intervention, mitral valve complication,

neurological event, cardiogenic shock, major bleed, acute kidney injury (AKI) stage 1/2/3,

endocarditis, oesophageal damage, device failure, partial detachment, pericardial effusion or

tamponade treated conservatively, thrombus, new moderate or severe mitral stenosis, minor

10

bleed, minor vascular complication. Detailed definitions of included outcome measures are

described in Supplementary Material – Table 2.

b. Cost / resource

A bottom-up costing study of each stage in the pathway to insert the MitraClip device was

conducted. NY EAC firstly reviewed the draft Excel® costing template provided by the NHS

England MitraClip Individual Technology Group. Amendments were agreed with the Chair of

the Group and the final template provided to the three NHS England CtE-funded centres with

detailed instructions on inputting the resources required to conduct each of three stages in

the relevant pathway being:

• Pre-operative assessment;

• Peri-operative procedure;

• Post-operative management.

The findings from the completed templates on resource use were reviewed by all authors

and compared with existing clinical pathways. Where possible, outcomes reported in the

MitraClip dataset such as number of clips implanted, type of imaging conducted at each

stage in the pathway, procedure duration, primary and secondary operator and length of stay

were used. Where such information was not available the three clinicians reached a

consensus view on the appropriate resources required. Unit costs from NHS national

datasets and other English national cost sources were applied to the resources and

aggregated to give a total procedural cost. Sensitivity analyses were conducted to provide a

high and low range of estimated costs. Full details are provided at Appendix 9, with a

summary of results in Tables 7 and 8.

c. Patient experience

From the outset of the cardiac CtE project, it was intended that EQ-5D-5L questionnaires

would be issued to all patients at baseline procedure and all subsequent follow-up visits.

This was to allow pairwise analysis of results over the follow up period.

2.4.6 Statistical analysis

All scripts for case ascertainment, cleaning, processing and statistical analysis were written

in the statistical programming language R [7].

Patient demographics, pre-operative clinical scores and procedural details were compared

between the whole, eligible cohort (n = 199) and the subgroup of patients discharged alive

and with any information recorded from at least one follow-up appointment (n=177) at 6

weeks, 6 months, 1 year or 2 years. Fisher’s exact tests or Mann Whitney U-tests were used

as appropriate. Bonferroni correction was used to adjust the level of significance to take into

account multiple comparisons.

Exploratory univariate and multivariate analyses were conducted for the defined outcome

measures. Univariate analysis was conducted for each outcome measure and up to 36

covariates. Bonferroni correction was used to adjust the level of significance to take into

account multiple comparisons (between outcome measure and each covariate of interest).

Multivariate analysis used generalised linear modelling with binomial error distribution in

order to estimate the effect size of covariates. Numeric covariates were centred on their

median before inclusion in multivariate analysis, if appropriate. Binary logistic regression

analyses were checked for convergence and over-fitting, and either modified (e.g. by

11

reducing the number of covariates) or reported as not valid. Coefficients of the covariates

were expressed as odds ratios (OR) with 95% confidence intervals.

Crude incidence rates for death, additional mitral valve intervention, and infective

endocarditis events recorded during the study period were calculated as the number of

events per 100 person years of follow-up. Kaplan-Meier analysis was applied to the time

from procedure to the time of the death, and first additional MV intervention and diagnosis of

endocarditis. Patients who suffered no events and were alive at the end of the study were

considered censored.

Severity of mitral regurgitation was also recorded at each follow-up (6 weeks, 6 months, 1

year, 2 years) and compared to pre-operative severity using Fisher’s exact test (paired

analysis).

Paired quality of life scores and utilities were compared at each time interval (6 weeks, 6

months, 1 year, 2 years) against pre-operative scores using Fisher’s tests or t-tests where

appropriate.

2.5 PRIMARY DATA COLLECTION (LINKED DATA)

2.5.1 Linked Dataset Details

In order to check for extra safety and efficacy, clinical effectiveness or resource utilisation

information, patient identifiers (i.e. NHS number, date of birth, gender, postcode and study

number; a pseudonymised patient identifier) were extracted from the CtE registry by NICOR,

as the data controller, on 5th April 2018 and sent to NHS Digital. CtE records were linked by

NHS Digital to the Hospital Episode Statistics (HES) Admitted Patient Care (APC) dataset

and the Office of National Statistics (ONS) mortality dataset, using an 8 step matching

algorithm [8]. Data from HES included all admissions from matched patients with hospital

discharge dates between 1st April 2008 and 1st March 2018. Data from ONS included all

deaths from matched patients reported until 1st March 2018. Records from patients with type

2 opt-outs (i.e. those not wishing for their patient information to be used for purposes other

than that of their individual care) were removed from both extracts by NHS Digital before

releasing the linked data to NICOR. NICOR then removed patient identifiers, replacing with a

“Study ID” and provided NY EAC with a pseudonymised linked dataset, as a data co-

processor.

2.5.2 Linked Dataset Cleaning

Pseudonymised data from all three data sources (CtE registry, HES and ONS) were

analysed by NY EAC. Records with conflicting demographic and administrative details

between the data sources were flagged by the EAC to indicate potential error in matching

(i.e. matching to incorrect patient). All flagged records were reported to the data controller,

NICOR, via a technical issue log. Technical issues and specific examples of inconsistencies

within HES or between HES and ONS datasets were discussed directly with NHS Digital.

The demographics were compared between cohorts with a potential mismatch (‘red flag’)

and those which were consistently matched between data sources (‘green flag’), in order to

confirm that exclusion of potentially mismatched records would not introduce bias into the

linked data results. These potentially mismatched patient records were then excluded from

the linked data analysis.

12

2.5.3 Linked Dataset Outcomes

All eligible patients (section 2.4.3) were analysed.

In-hospital complications (major and minor combined) from the registry were defined as

described in section 2.4.5a and Supplementary Material – Table 2. In-hospital complications

from HES were identified from the ICD-10 codes captured in the 20 diagnosis fields using

the method recommended by NHS Clinical Classifications Service [9, 10].

Total length of follow-up using the CtE registry was calculated from the date of the MitraClip

procedure and the date of last recorded follow-up (or discharge date in cases where the

device was not successfully implanted). Total length of follow-up using the linked dataset

was calculated from the date of the MitraClip procedure until the date of extract (1st March

2018), or date of death, or date of discharge (in cases where the device was not successfully

implanted).

Total all-cause mortality from the linked dataset combined all in-hospital deaths, and all

deaths occurring post-MitraClip procedure in those discharged from hospital with a device

successfully implanted. All linked data outcomes were firstly determined for the registry and

HES (and ONS where applicable) separately. The total event rates (for each outcome) were

then determined using the union of all datasets. Inconsistencies were raised by NY EAC with

the data controller, NICOR, within a technical issue log. Responses from CtE providers (to

queries raised by NY EAC, via NICOR) were also logged within the technical issue log.

Kaplan-Meier analysis was conducted for total all-cause mortality. Patients were followed

from date of procedure until date of first event, date of discharge (in cases where the device

was not successfully implanted), or if no event, then date of extract of HES and ONS

datasets (i.e. 1st March 2018). In patients with multiple events (i.e. transient ischaemic attack

(TIA) followed by haemorrhagic event; or neurological event followed by death), the time to

the earliest event following MitraClip admission were used in subsequent Kaplan-Meier

analysis.

Additional outcomes determined from the linked dataset included instances of infective

endocarditis identified through presence of ICD-10 codes from the I33 chapter “Acute and

subacute infective endocarditis” appearing in any diagnosis field (20 available) in any

readmission occurring after the MitraClip implantation. Also presence of additional mitral

valve intervention which were identified and categorised using the following OPCS-4 code

appearing in any procedure field (24 available) in any readmission occurring after the

MitraClip implantation:

• Replacement

• K25.1: Allograft replacement of mitral valve

• K25.2: Xenograft replacement of mitral valve

• K25.3: Prosthetic replacement of mitral valve

• K25.4: Replacement of mitral valve NEC

• Repair:

• K25.5: Mitral valve repair NEC

• K25.8: Other specified plastic repair of mitral valve

• K25.9: Unspecified plastic repair of mitral valve

• K30.1: Revision of plastic repair of mitral valve

• Other:

• K31.1: Open mitral valvotomy

13

• K32.1: Closed mitral valvotomy

• K34.1: Annuloplasty of mitral valve

• K35.1: Percutaneous transluminal mitral valvotomy

• K37.5: Excision of supramitral ring

• K38.3: Operations on mitral subvalvar apparatus

• Any OPCS-4 code followed by site code Z32.1: Mitral valve

Insertion of a cardiac pacemaker system was also determined from the following OPCS-4

codes:

- K60.1:Implantation of intravenous cardiac pacemaker system NEC

- K60.5: Implantation of intravenous single chamber cardiac pacemaker system

- K60.6: Implantation of intravenous dual chamber cardiac pacemaker system

- K60.7: Implantation of intravenous biventricular cardiac pacemaker system

- K60.8: Other specified cardiac pacemaker system introduced through vein

- K60.9: Unspecified cardiac pacemaker system introduced through vein

- K61.1: Implantation of cardiac pacemaker system NEC

- K61.5: Implantation of single chamber cardiac pacemaker system

- K61.6: Implantation of dual chamber cardiac pacemaker system

- K61.7: Implantation of biventricular cardiac pacemaker system

- K61.8: Other specified other cardiac pacemaker system

- K61.9: Unspecified other cardiac pacemaker system

Insertion of a cardioverter defibrillator was determined from the following OPCS-4 codes:

- K59.1: Implantation of cardioverter defibrillator using one electrode lead

- K59.2: Implantation of cardioverter defibrillator using two electrode leads

- K59.6: Implantation of cardioverter defibrillator using three electrode leads

- K59.7: Renewal of cardioverter defibrillator using three electrode leads

- K59.8: Other specified cardioverter defibrillator introduced through the vein

- K59.9: Unspecified cardioverter defibrillator introduced through the vein

Hospital resource usage was determined only for patients discharged alive with successful

MitraClip implantation (i.e. per protocol analysis). Hospital admissions, as recorded in the

HES APC dataset, were analysed in the year prior to (i.e. pre-admissions) and year following

(i.e. post-admissions) the MitraClip implantation; excluding the procedural admission itself.

Primary outcomes of interest included admission rate, hospitalisation days and cost

associated with hospitalisation in the 12 months before and after the MitraClip procedure.

Admission rates and hospitalisation days were reported per 1000 person years follow-up in

order to account for the variable person time contributed by each patient during each time

period of interest. Hospitalisation costs were determined by converting each finished

consultant episode of the HES APC extract to Healthcare Resource Groups (HRGs) by

applying the HRG4+ 2017/18 Reference Costs Grouper (National Casemix Office), and

referring to the National Schedule of Reference Costs 2017/18 at aggregate level using the

mean costs from all NHS trusts and NHS foundation trusts (NHS Improvement). Costs per

episode were then aggregated into spells to give a cost per admission. Outcomes were

described by reason for admission: non-cardiac, cardiac, and heart failure including mitral

insufficiency (ICD-10 codes I340, I500, I501, I509).

https://digital.nhs.uk/services/national-casemix-office/downloads-groupers-and-tools/costing-hrg4-2017-18-reference-costs-grouper

https://improvement.nhs.uk/resources/reference-costs/

14

All outcomes analysis were repeated for the following subgroups: a) elective, b)

urgent/emergency, c) ischaemic/functional MR, d) degenerative MR and e) those alive at 12

months. MR aetiology and procedural urgency were taken from the MitraClip implantation

procedural admission as recorded by the registry (more clinically rich dataset), mortality was

determined using linked ONS data.

Differences pre- and post-MitraClip were compared via mixed effects modelling (pre-/post-

procedure phase as fixed effect, patient ID as random effect, offset by follow-up time) to

account for the before and after samples being non-independent (involving the same

patients) and multiple admissions for some patients. Fixed effects were expressed as

incidence rate ratios with associated 95% confidence intervals. A log-Poisson distribution

was used when comparing admission rates and hospitalisation days, and a log-gamma

distribution used when comparing hospitalisation costs in the 1 year pre- and post-MitraClip

(Vemulapalli et al. 2017).

2.5.4 Statistical Analysis of Healthcare Resource Use Data

To account for some patients incurring zero in-patient cost, the distributions of hospitalization

cost were fitted with Tweedie distributions [11]. These are a family of distributions with

exponential dispersion which are suitable for use in generalized linear models and account

for the presence of a positive probability mass at zero. This approach is similar to the

conventional practice of fitting costs with a Gamma distribution, but accounts for the

presence of some patients incurring zero cost. Parameters of Tweedie distributions were

fitted with maximum likelihood estimates using the R package ‘tweedie’ [12].

Observed in-patient costs were summarized as means and SDs for each period. Cost ratios

and their 95% confidence intervals were estimated using compound Poisson generalized

linear mixed models with R package ‘cplm’ [13], with a random intercept for each patient.

These mixed-effects models fitted Tweedie distributions to the observed costs.

2.6 SECONDARY DATA COLLECTION (LITERATURE REVIEW)

The aim of the final MitraClip literature review for CtE was to identify key published studies in

patients with mitral valve regurgitation (MR) of any severity or aetiology and summarise

results so they align with the requirements of the outputs of NY EAC project RX085,

including the 11 questions set by NHS England. A brief summary of the review methods is

presented here. A standalone literature review document is available for further information

(Willits et al., MitraClip literature review document, November 2017) [14].

Firstly, a literature search was performed from March 2009, which was the updated search

date of the original NICE IP664 overview that informed NICE IPG309 (this was consistent

with the requirements from NHS England). The NICE search strategies for replication were

sourced through documents supplied by NICE and through communication with the Senior

Information Manager at NICE Guidance Information Services. The EAC team and NICE

agreed that no quality assessment would be made of the NICE strategies and the intention

was to use the NICE-designed strategies as supplied. Some minor edits were made (for

example, the correction of some search structure and syntax errors identified in the original

NICE strategy). Apart from these minor changes, the terms used in the update strategies

reflected those used in the original strategies of NICE IPG309.




15

The scope of the literature review was intended to broadly reflect the population and

intervention covered in the CtE registry. The scope, described in PICO (Population,

Intervention, Comparator, Outcomes) format, is summarised in the Table 3.

Table 3. Scope of literature review.

Domain Terms identified from title or abstract

Comment

Population Patients with mitral regurgitation

Includes degenerative or functional mitral regurgitation.

Intervention Percutaneous mitral valve leaflet repair

Specifically interested in MitraClip. If unable to ascertain from abstract, retrieve. If other device or procedure, flag.

Comparator Any or none Comparative studies preferred but in practice may not be available. Studies with comparisons versus medical management of particular relevance.

Outcomes Clinical outcomes Utility and resource use outcomes*

Studies reporting only surrogate or non-clinical outcomes will be excluded.

Study type All primary studies Secondary studies* (systematic reviews and meta-analyses) Economic studies*

Non-systematic reviews, editorials and opinion pieces excluded. Abstracts included only if reporting an adverse effect not already known. All RCTs included. Single armed observational studies included if n≥100. Comparative observational studies of any size included if comparator is conservative medical management.

* Systematic reviews and meta-analyses not included in final PRISMA selection but flagged. Economic studies and associated outcomes to be identified for possible future reference.

Given the timelines of the project and the purpose of the update search, the EAC team and

NICE agreed that only the bibliographic databases listed in Table 4 would be searched. In

addition, it was agreed that strategies would be limited to results published in English

language only, and that conference-related publication types would be excluded from the

Embase search.

Where database functionality allowed, results were limited to records added to the database

since the date of the last search, using appropriate fields such as the entry date field in

MEDLINE. Where database functionality did not allow this, results were limited by

publication date, reflecting the pragmatic context of the search.

Table 4. Bibliographic databases searched.

Database / information source Interface / URL

MEDLINE and MEDLINE In-Process OvidSP

EMBASE OvidSP

Cochrane Database of Systematic Reviews

(CDSR)

Cochrane Library/Wiley Interscience

Cochrane Central Register of Controlled Cochrane Library/Wiley Interscience

16

Database / information source Interface / URL

Trials (CENTRAL)

Database of Abstracts of Reviews of

Effects (DARE)

http://www.crd.york.ac.uk/CRDWeb/

Health Technology Assessment Database

(HTA)


NHS Economic Evaluation Database

(EED)


Relevant studies were sifted by two reviewers according to the predefined scope, and these

studies were then combined with those reported in NICE IPG309. As this approach identified

an unmanageable number of studies, a further selection process was employed to identify

studies on the basis of methodological quality and size, with randomised controlled trials

(RCTs) and single-armed registry studies with 500 or more participants or cohort studies

with a combined total of 100 or more participants selected for full review [14]. Systematic

reviews and economic studies were also identified.

Figure 1 is a flow diagram of this pragmatic literature review strategy, including the inclusion

and exclusion criteria applied to sifting.


17

Figure 1. Flow chart illustrating the literature review strategy for MitraClip.

A brief summary of the results of the literature review is presented in the results section of

this final CtE report on MitraClip, with full details available in the standalone literature review

document [14].

2.7 RESEARCH DESIGN

The study was a procedural registry designed with a maximum 2 years of follow up. The

registry was single armed with no comparator or control arm. Data were collected

prospectively in accordance with best practice [15, 16].

18

There has been some uncertainty regarding the efficacy and safety of MitraClip (see Section

3.2 for further details). In particular, there is an issue concerning the generalisability of

published comparative trial evidence to this population in the CtE registry, and how effective

and safe the procedure is in real-world practice. To help clarify this uncertainty, NHS

England has requested that the answers to 11 clinical and economic questions should be

addressed, using data reported by the CtE registry and supported by published studies in

the literature. These questions have been revised and adjudicated by NICE (see Table 1).

The EAC performed a pragmatic literature review, which identified the key experimental and

observational studies performed to date on MitraClip, as described in Section 2.5. As the

CtE registry was non-comparative, data from the literature has been used as a proxy control.

Table 5 summarises the a priori intended methods for answering each question [17].

However, due to issues with data quality and reporting of published literature, the original

methods were not always possible. These limitations have subsequently been annotated in

the table.

The relationship between the registry and published literature in answering the NHS England

questions is illustrated in Figure 2. Inference has been made by comparing point estimates

and confidence intervals where available. Additionally, in some instances where the registry

was not sufficiently robust to answer the questions, published evidence was used to directly

answer questions.

19

Table 5. Methods used to analyse and report CtE registry data.

Question (NICE modified where applicable)

Can it be answered using registry data?

Key registry data required Type of analysis – CtE registry data

Type of analysis – Linked data (CtE registry/HES/ONS)

Comment

1) Can UK clinical teams undertaking MitraClip reproduce the reduction in mitral regurgitation seen in the early clinical trials?

Yes, fully. Change in mitral regurgitation grade.

Pairwise (‘before and after’) analysis of registry data. Comparison with published RCTs and observational studies.

None (MR grade not recorded in HES).

Registry does not provide comparative data so this will be matched with published data. Depending on the goodness of fit, a narrative summary or statistical analysis may be possible.

2) Is reduction in mitral regurgitation mediated by MitraClip associated with improvements in quality of life?

Yes, fully. Severity of mitral valve regurgitation. 6 minute walk test. NYHA score. Quality of life.

Pairwise (‘before and after’) analysis of registry data. Correlation and regression analysis.

None (QoL not recorded in HES).

Association between reduction in mitral regurgitation and improved quality of life likely to be causal association.

3) Does MitraClip improve survival rates?

Yes, partly. Mortality rate. Kaplan-Meier analysis. Comparison with published RCTs and observational studies.

Proportion of people having event with confidence intervals. Survival analysis (Kaplan-Meier).

Data from registry limited by low patient enrolment and short follow up.

4) Does MitraClip reduce the frequency of subsequent hospital admissions?

Yes, partly. Additional mitral valve intervention. MitraClip detachment/embolisation/ retrieval. Electrical device therapy. Other MACCE. [Update: follow up not extended, could feasibly be addressed in future using HES analysis].

Comparison with literature on natural history of mitral valve disease. Control arm of suitable RCTs, observational studies, expert opinion.

Pairwise (‘before and after’) analysis.

Most fields are not core so risk of missing data in registry.

5) Are the early benefits in reduction in mitral regurgitation

Yes, partly. Severity mitral valve regurgitation.

Kaplan-Meier analysis with extrapolation.

Proportion of people having additional MV

Short, medium, and long (or longer) term time

20





Comment

maintained in the medium-term? Is there a need for repeat treatment over time (either by a repeat percutaneous procedure or surgery)?

Additional mitral valve intervention.

Comparison with published literature.

intervention.

periods require defining and feasibility of longer term analysis will be dependent on contract extension. [Update: follow up not extended, could feasibly be addressed in future using HES analysis].

6) What proportion of patients referred to a specialist MitraClip service as defined in the CtE documents were assessed by the MDT as suitable for the intervention? What proportion of the patients considered suitable for the procedures received it and what proportion of them benefitted?

Yes, partly. Decision to treat by MDT. Severity of mitral valve regurgitation (change over time). NYHA (change over time).

Proportion of patients considered by MDT/patient received MitraClip.

None. Difficult to unravel selection process, unknown how many potential candidates do not make the registry at all.

7) What are the short and medium term risk of complications from MitraClip use?

Yes, partly. Additional mitral valve intervention. MitraClip detachment/embolisation/ retrieval. Electrical device therapy. Other MACCE.

Kaplan-Meier analysis. Comparison with published literature.

Proportion of people having event with confidence intervals.

Some complications specifically attributable to MitraClip. Risk-benefit ratio of MitraClip will require longer-term studies and/or modelling. Registry data considered more robust for measurement of in-hospital complications.

8) Are clinical outcomes with MitraClip associated with particular patient characteristics (clinical or demographic)?

Yes, partly. Patient characteristics. Efficacy outcomes. Complication outcomes. Mortality.

Subgroup analysis. Bonferroni correction if hypotheses not pre-specified.

Proportion of people having events with confidence intervals. Survival analysis (Kaplan-

Limitations with patient enrolment (power), patient selection and confounding variables (generalisability

21





Comment

Meier). Comparison of healthcare resource use.

issues).

9) What are the full procedural costs of using MitraClip to the NHS?

No. In hospital resources including length of stay, investigations, theatre staff, procedure duration and type of anaesthetic.

Process costing with separate costs for each stage of the clinical pathway.

Not applicable. Procedural costs were estimated by combining information from the registry and data from sites, collected using a pro forma.

10) What are the potential cost savings for the NHS arising from patients receiving MitraClip?

No. As 9) plus efficacy post discharge, adverse events and mortality.

Mean and SD for each key event on pathway.

De novo “before and after” analysis.

The EAC performed a de novo economic analysis, based on the methodology of Vemulapalli et al. (2018) [18].

11) Is MitraClip cost-effective from the perspective of the NHS?

No. As 9) plus efficacy post discharge, adverse events and mortality.

Mean and SD for each parameter.

De novo “before and after” analysis.

The EAC performed a de novo economic analysis, based on the methodology of Vemulapalli et al. (2018) [18].

22

Figure 2. Relationship between NHS England registry clinical data and published evidence identified in the literature review (questions 1 to 8

[Q1 to Q8]). [Note: two additional RCTs were published subsequent to the literature search (see Section 3.4.2). These have informed Q1 to Q5 and Q

7 and Q8].

23

Section 3: Results

3.1 PRIMARY DATA COLLECTION (CTE REGISTRY)

3.1.1 Numbers of patients treated at each centre

A total of 272 MitraClip procedure records were extracted from the registry by NICOR on 13th

November 2017. Seventy three did not meet the eligibility criteria, Appendix 1, 57 of which

did not include eligible reasons for MitraClip treatment (multiple reasons permitted): 8 patient

preference, 3 clinician preference, 28 other, and 18 not providing a reason for treatment. A

total of 199 MitraClip procedures were eligible for analysis.

Follow-up information for CtE-commissioned procedures was recorded in 98.7% of eligible

MitraClip procedures at 6 weeks, 95.9% at 6 months, 117.7% at 1 year and 29.7% at 2

years, Supplementary Material – Table 4. The reason for a figure in excess of 100% follow

up at 1 year may possibly relate to data from other follow-up time points being entered in the

“3rd follow-up” registry page incorrectly, since this was intended for 12 month data only.

3.1.2 Summary statistics of patient and procedural characteristics

The MitraClip registry enrolled 272 patients, of whom 199 were eligible for CtE data analyses

(Appendix 1). The 199 patients included in the CtE analyses had functional (60%) or

degenerative (40%) MR and a mean age of 76.2 years. Most patients were men (68.8%).

Most of these patients were recruited electively (84.4%) with 13.6% admitted urgently and

2.0% undergoing the procedure as an emergency. Nearly all patients had moderate or

severe MR (grade 3+ or 4+) that was symptomatic in 92% of cases (New York Heart

Association [NYHA] class 3 or 4). The mean EuroSCORE II (risk of dying from cardiac

surgery) was 6.4 (standard deviation 5.7). Complete patient demographics and procedural

characteristics for the cohort are summarised in Appendix 2 and Appendix 3 respectively. No

statistical differences were identified between the whole cohort and those with reported

follow-up information for any of the tabulated variables.

3.1.3 Active surveillance (Evaluation of Coverage)

The data coverage and completeness results for CtE commissioned procedures only, for the

27 MitraClip MDS baseline fields and the 6 specified follow-up fields (to 24/10/2017) are

reported in Table 6.

24

Table 6. Data completeness by CtE-funded provider.

CtE provider Baseline

MDS

completeness

Coverage†

& completeness‡

at 6 weeks FU

Coverage†

& completeness‡

at 6 months FU

Coverage†

& completeness‡

at 12 months FU

Coverage†

& completeness‡

at 24 months FU

The University Hospital of

South Manchester NHS

Foundation Trust

97% 100% (42/42) coverage

91.7% FU data completeness

97.6% (40/41) coverage


100% (35/35) coverage


63.6% (7/11) coverage


University Hospitals Bristol

NHS Foundation Trust 93%

97.9% (47/48) coverage


95.7% (45/47) coverage


97.7% (43/44 coverage


12.5% (1/8) coverage


Royal Brompton and Harefield

NHS Foundation Trust 94%

98.6% (72/73) coverage


95.8% (68/71) coverage


95.8% (68/71) coverage


21.1% (4/19) coverage


Total 95%

163/181 (90.1%) with device

implanted, discharged alive

and reaching 6 weeks since

procedure date.

161/163 (98.8%) with some

degree of FU data.

Completeness of FU MDS

(versus expected) = 88.0%.

159/163 (97.5%) with device

implanted, still alive at 6 weeks

FU and reaching 6 months

since procedure date.

153/159 (96.2%) with some

degree of FU data.



153/159 (96.2%) with device

implanted, still alive at 6

months FU and reaching 12

months since procedure date.

146/150 (97.3%) with some

degree of FU data.



147/153 (96.1%) with device

implanted, still alive at 12

months FU and reaching 24

months since procedure date.

12/38 (31.6%) with some

degree of FU data.



FU Coverage† = Actual no. of MitraClip procedures with some degree of FU data entered / No. of MitraClip procedures eligible for FU for the stated period (%).

NB FU Coverage can only be calculated for cases with a procedure date entered. This is the case for 181/181 (100%) of CtE-funded MitraClip cases in the registry, to 24/10/2017.

FU Completeness‡ = Average completeness of the 6 specified MitraClip MDS-FU data fields (%).

25

3.1.4 Outcomes – using CtE registry data only

a. Clinical

A total of 182 procedures (91.5%) recorded both admission and discharge dates, showing a

median length of stay of 5 nights (inter-quartile range [IQR] 3.3 to 8.0 nights, range 0 to

46 nights). Most of the 199 eligible patients were recruited electively (84.4%) with 13.6%

admitted urgently and 2.0% undergoing the procedure as an emergency. The impact of

procedural urgency on clinical outcomes is specifically addressed in NHS England Question

Eight. Device implantation was conducted in 187 (94.0%) of procedures; reasons for non-

use of device included: 3 unable to clip valve at all, 3 procedures ended successfully (but

number of clips implanted = 0), 2 ending before completion due to complication, 2 ended

early due to problem with left atrial access (by registry data field definition), 1 ended early

due to inadequate imaging and 1 ended early due to inadequate device stability.

In-hospital major complications were reported in 16 patients (8.2%, multiple events

permitted) including: 10 deaths, 4 AKI (stage 2/3), 4 requiring additional interventions

(comprising 1 percutaneous retrieval of embolised device, 1 bailout/urgent percutaneous

coronary intervention (PCI), 1 MV surgery before discharge and 1 conversion to open heart

surgery before discharge), 3 major bleeds, 2 myocardial infarctions, 2 cardiogenic shock, 1

embolisation, 1 new neurological event, and 1 oesophageal damage. There was no

association between major complications and procedural urgency when Bonferroni

correction was applied (elective: 10/165, 6.1%; urgent/emergency: 6/31, 19.4%).

In-hospital minor complications were reported in 15 patients (7.6%), including: 7 minor

bleeds, 3 new moderate/severe mitral stenosis, 3 pericardial effusion/tamponade treated

conservatively, 1 partial detachment, and 1 AKI (stage 1). Frequencies of in-hospital

outcomes for all eligible MitraClip patients and post-discharge outcomes for all eligible

MitraClip patients with a device implanted are described in Appendix 4. Procedural success

(device implanted and no major complications) was achieved in 171 MitraClip procedures

(85.9%).

No variables were significantly associated with device implantation, in-hospital major

complication, or in-hospital minor complication at the Bonferroni adjusted alpha levels during

exploratory univariate or multivariate analysis.

In the 182 procedures (91.5%) which recorded both admission and discharge dates, length

of stay was significantly different between elective (median 5 nights, IQR 3 to 7 nights, range

0 to 37 nights) and urgent/emergency cases (24 nights, IQR 8 to 32 nights, range 0 to 46

nights), p < 0.0001. In-hospital and outcomes recorded after discharge are described

separately for all elective (n = 168) and emergency/urgent (n = 31) cases in Appendix 5.

For those with a MitraClip device implanted (n = 187), follow-up was reported in 170 patients

(90.9%). Major complications occurring post-MitraClip procedure discharge were recorded in

25 patients, including: 20 deaths, 4 additional interventions (including 2 surgical MV

interventions, 1 percutaneous MV intervention, 1 coronary artery bypass graft (CABG) and 2

strokes. Twenty-two patients reported minor complications, including: 21 with new mitral

stenosis and 1 with partial detachment.

26

Total crude incidence rates of adverse events are described in Appendix 6. A total of 30

deaths have been reported (15.1%); 10 of which were in hospital (5.0%) and considered

directly attributable to the procedure. There were 22 (13.1%) deaths in elective patients and

8 (26.7%) deaths in urgent/emergency 8 patients. Cause of death was reported in 23 cases

(76.7%), and included: 8 multi-organ failure, 1 ventricular fibrillation with severe ventricular

impairment, 1 biventricular failure, 1 pulmonary oedema/cardiogenic shock, 1 oesophageal

damage, 1 acute renal failure, 1 clip abandoned and patient palliated, 1 peri-prosthetic

fracture of knee joint (died in theatre), 1 respiratory arrest, 1 lung cancer, 1 old age, 1

attributed to a combination of acute kidney injury, heart failure and ischaemic heart disease

and 4 not otherwise specified. Kaplan-Meier curves for time to death and additional MV

intervention are shown in Appendix 7.

Severity of mitral regurgitation over time is described in Appendix 8. Patients in the CtE

registry experienced immediate and dramatic improvements in MR grade, with the proportion

of patients with moderate-severe or severe MR (≥3+) reduced from 99.5% to 6.7%, post-

procedure. However, after 1 year, some deterioration in mitral valve function was observed,

with 24.4% of patients reporting moderate-severe or severe MR.

Improvements in MR were matched by improvements in NYHA scores, with 92.4% of

patients reporting a score of NYHA class III or IV at baseline, compared with 17.9% at

1 year. Change in NYHA dyspnoea score over time is described in Supplementary Material –

Table 8.

b. Cost / resource

Three well-completed MitraClip CtE Excel® costing workbooks were received. These were

synthesised, together with information from the LAAO and PFOC pathways and the final

dataset from the MitraClip registry, to create a list of the resources required at each stage of

pathway. In November 2017 these were sent to the three Clinical Leads for review.

Following subsequent changes in light of comments received, the NY EAC updated the

template and included cost information. Unit costs were taken from published national

datasets (primarily NHS Reference Costs [19] and PSSRU [20]). The unit price was provided

by the manufacturer as ‘commercial in confidence’ (Personal communication, Dr Steven

Fearn, Health Economics & Reimbursement Manager, Abbott; 4 October 2017) who also

advised that manufacturers charge per procedure, not per device. This is a key assumption

as the device is the highest cost in the pathway and the registry recorded that the mean

number of clips opened was 1.9 per procedure.

The resultant estimated central cost and high and low cost scenarios for a MitraClip

procedure conducted in NHS England were updated from 2015/16 costs to 2017/18 costs

using the hospital & community health services index [21]. This was estimated to have

increased by 1.8% in each of the two intervening years. This increase was not applied to the

cost of the device, the price of which was agreed with manufacturer for the duration of the

registry. The resultant estimated central cost and high and low cost scenarios for a MitraClip

procedure conducted in NHS England are shown in Table 7.

27

Table 7. Central cost and range of cost for a MitraClip procedure.

Pathway stage Central cost Low cost High cost

Pre-operative assessment £819 £427 £1,034

Peri-operative procedure £27,885 £26,822 £28,928

Post-operative management £4,209 £1,759 4,566

Total £32,912 £29,007 £34,528

Table 8 analyses the estimated costs by component and stage for the central case. The

device accounts for **% of the cost, with investigations forming the second largest cost

component (**%), length of stay comprises 8%, staff comprise 5%, consumables and theatre

use contribute **% each to the cost, with outpatient follow-up being 1%.

Table 8. Estimated costs by component by stage for central case.

Pre-op Peri-op Post-op Total % of Total

Device £****** £****** **%

Investigation £657 £****** £1,146 £****** **%

Staff £147 £1,647 £1,794 5%

Consumables £15 £****** £****** **%

Length of stay £2,865 £2,865 9%

Theatre

£****** £****** **%

Outpatient £197 £197 1%

Total £819 £27,885 £4,209 £32,912 100%

Data from the registry show that elective patients have a shorter length of stay at 5.6 days

compared with 20.5 days for patients admitted as urgent or emergencies; the mean for all

patients was 7.8 days. Applying these lengths of stay gives a cost for an elective patient of

£32,115, rising to £37,616 for a non-elective patient.

A full summary of all resources and unit costs is provided in Appendix 9. This also describes

the assumptions underpinning the sensitivity analyses.

Hospital resource usage over time is described in Supplementary Material – Table 10.

c. Patient experience

Significant improvements in mobility, self-care, usual activities, pain/discomfort and

anxiety/depression quality of life (EQ5D) components were observed at varying time points

when compared to pre-procedure scores, Supplementary Material – Table 9.

Pre-procedure, EQ-5D values were available for 163 patients. At 6 weeks, 136 paired scores

were available and these showed a mean gain in utility of 0.18, with 83.1% of patients

reporting improved quality of life, 9% no change and 10.3% deterioration. At 6 months,

paired data for 113 patients were available. These showed a similar marginal improvement

in the utility score of 0.20, with 79.6% of patients reporting improved quality of life, 7.1% no

change and 13.3% deterioration. The mean baseline value was 0.55 ± 0.23 (SD); however,

the median value of 0.60 was adopted as a measure of central tendency because of the

skewed distribution [22].

28

Full EQ-5D results are presented in Supplementary Material - Table 9.

3.2 PRIMARY DATA COLLECTION (DATA LINKAGE)

3.2.1 Cohort characteristics

Records from a total of 278 MitraClip patients were extracted from the NICOR registry on

05/04/2018, of which 199 were eligible for analysis (i.e. meeting CtE criteria in section 2.4.3).

All 199 of these records preceded 01/03/2018 (i.e. date of data linkage). A total of 187

(94.0%) were matched with the HES APC dataset, of which 21 (11.2%) did not pass internal

quality checks with the registry and ONS data, casting serious doubt on their matching to the

correct patient (see Appendix 10). Issues associated with all 21 patients not passing internal

quality checks were raised to the data controller. As of 14th September 2018 no further

validation was available. However, there were no significant differences in baseline

characteristics between those which passed the internal quality checks (n=166 ‘green flags’)

and those who did not (n=21 ‘red flags’) in terms of age, sex, body surface area, diabetes,

critical pre-op status, CCS angina status, NYHA dyspnoea status, Killip class of heart failure,

CSHA frailty score, aetiology of MR, Katz index of independence – see Supplementary

Material – Table 11. Procedural urgency was significantly different (i.e. there were only 21

red flags and 9 were urgent/emergency). Therefore there is an association between data

quality and procedural urgency (as expected). All 21 patients were removed from all

subsequent analysis.

The patient demographics of 166 patients included in the linked data analysis are described

in Table 9.

Table 9: Patient demographics of records included in the linked dataset.

Patient characteristic† Eligible patients in linked dataset (n=166)

Female 52 (31.3%)

Age, years median (Q1,Q3) [range]

78.5 (69,84.8) [37-94]

Body surface area, m2

median (Q1,Q3) [range] 1.9

(1.7,2.0) [1.3-2.4]

Diabetes 30 (18.1%)

Critical pre-op status 3 (1.9%)

Katz (index of independence) median (Q1,Q3) [range]

6 (6,6) [2-6]

Aetiology of MR - Functional/Ischaemic 99 (61.1%)

- Degenerative 63 (38.9%)

Procedural urgency - Elective 145 (87.3%)

- Urgent/emergency 21 (12.7%)

NYHA (dypnoea status): - No limitation of physical activity - Slight limitation of ordinary physical activity - Marked limitation of ordinary physical activity - Symptoms at rest or minimal activity

2 (1.2%) 10 (6.0%)

106 (63.9%) 48 (28.9%)

Killip class (of heart failure): - No clinical signs of heart failure - Rales in lungs/S3/elevated JVP but not class 3

107 (57.7%) 39 (24.7%)

- Acute pulmonary oedema 11 (7.0%)

- Cardiogenic shock 1 (0.6%)

29

CSHA frailty score: - Very fit/Well/Well with treated comorbidities 38 (26.2%)

- Apparently vulnerable/Mildly frail 52 (35.9%)

- Moderately/Severely frail 55 (37.9%) † Not all data fields were complete for every patient at baseline and follow up. The percentages

presented in this table are calculated using the number of patients with each characteristic reported

as the denominator.

3.2.2 In-hospital complications

The registry reported 25 in-hospital complications, HES reported 29. From the union of both

datasets, a total of 42/166 (25.3%) MitraClip procedures had some degree of complication

reported, see Table 10.

Fourteen major discrepancies in complications reported by one source (either the CtE

registry or HES) and not validated in the other were raised by the NY EAC to NICOR. As of

14th September 2018 no further validation was available.

Table 10: Contingency table of in-hospital complications recorded in the linked

dataset.

HES Total

No complication Complication

CtE Registry

No complication 124 17 141

Complication

13 12 25

Total 137 29 166

Major complications recorded in the CtE registry included: 7 deaths during the MitraClip

procedural admission (main cause reported as pulmonary sepsis, oesophageal damage,

pulmonary oedema, three cases due to multi-organ failure, and cause unreported in one

case ), 1 neurological events (of ischaemic type), 2 MI, 1 requiring bailout PCI, 1 requiring

MV surgery, 1 requiring conversion to open heart surgery (other than for a MV procedure), 3

major/life threatening bleeds, 2 cases of cardiogenic shock, see Appendix 11.

Complications recorded in HES could not be robustly categorised as major and minor due to

a lack of granularity in the ICD-10 codes. However the most common complications reported

in HES included: 11 instances of haemorrhage and haematoma complicating a procedure, 2

accidental punctures and laceration during a procedure not elsewhere classified, 2 cardiac

septal defects acquired, and 8 deaths during the MitraClip procedural admission (cause of

death from ONS included 5 cases due to mitral (valve) insufficiency, 1 case of surgical

operation and other surgical procedures as the cause of abnormal reaction of the patient, or

of later complication, without mention of misadventure at the time of the procedure: other

reconstructive surgery, 1 case of accidental puncture or laceration during a procedure, and 1

case of chronic ischaemic heart disease), see Appendix 11.

30

In-hospital complications are reported separately by procedural urgency (elective and

urgent/emergency), and MR aetiology (functional/ischaemic and degenerative) in Table 11.

Table 11: In-hospital outcome analysis by procedural urgency and MR aetiology.

All MitraClip procedures

Procedural urgency MR Aetiology*

Elective Urgent/ Emergent

Functional/ Ischaemic

(FMR)

Degenerative (DMR)

n, patients 166 145 21 99 63

In-hospital death (%)

8† (4.8) 5 (3.4) 3 (14.3) 5 (5.1) 3 (4.8)

Total procedural complications‡ (%)

42 (25.3) 35 (24.1) 7 (33.3) 33 (33.3) 9 (14.3)

*4 procedures with MR aetiology not reported †all of whom had at least 1 successful MitraClip inserted ‡minor and major combined as reported by registry or HES (union)

3.2.3 All-cause mortality

Of the 166 patient records available for analysis, 8 deaths were recorded during the

MitraClip procedural admission (i.e. 4.8% in-hospital mortality), and 8 different patients did

not have a successful MitraClip device implantation (i.e. 4.8% in-hospital procedural failure).

Of the 150 patients discharged from hospital alive with successful implantation of the MitraClip device:

- 5 (3.3%) had no follow-up entered into the registry

- 11 (7.3%) had 1 follow-up entry fully or partially completed

- 5 (3.3%) had 2 follow-up entries fully or partially completed




The total length of follow-up for the 166 records calculated from the CtE registry data was

40,148 days (110 person years) with a mean (SD) 241.9 (231.9), median (Q1,Q3) 184.5

(40.0,384.8) range 0 to 898 days per patient.

Of the 150 patients discharged alive with successful MitraClip implantation:

- 25 (16.7%) had no subsequent hospital admissions recorded in HES following the

MitraClip procedure

- 37 (24.7%) had 1 hospital admission

- 30 (20.0%) had 2 hospital admissions



- 27 (18.0%) had 5+ hospital admissions (the maximum number of admissions was

17 occurring in one individual with iron deficiency anaemia requiring regular

continuous intravenous infusion of a therapeutic substance and intravenous blood

transfusion of packed cells).

31

The total length of follow-up calculated from the linked dataset was 99,241 days (271.9

person years) with a mean (SD) 597.8 (283.1), median [Q1;Q3] 610.0 [408.8 to 807.0],

range 2 to 1128 days per patient.

A total of 22 deaths were reported by the registry, 23 in-hospital deaths reported in HES, and

36 total deaths reported in ONS, see Figure 3.

Figure 3: All-cause mortality as reported across the three datasets.

The total number of deaths (all in-hospital, post-MitraClip discharge in those with successful

MitraClip implantation) occurring across all three data sources (i.e. union of the registry,

HES, ONS) was 37; 8 of which occurred during the MitraClip procedural admission. Total

event rate for all-cause mortality (using n=37 events) was 13.6 [9.6 to 18.8] per 100 PY

follow up, see Table 12 and Figure 4. Kaplan Meier analysis by procedural urgency of index

MitraClip implantation procedure and aetiology of mitral regurgitation are shown in Appendix

12.

16

CtE Registry (n=22)

HES (n=23) ONS (n=36)

0

0

9

1

6

5

21 deaths

matching in CtE

Registry and ONS

22 deaths

matching in ONS

and HES

17 deaths

matching in CtE

Registry and HES

32

Figure 4: Kaplan-Meier analysis for all-cause mortality.

33

Table 12: All-cause mortality by procedural urgency and MR aetiology.

Procedural urgency* MR aetiology**

All MitraClip procedures Elective Urgent/ Emergency

Functional/ Ischaemic (FMR)

Degenerative (DMR)

No. of patients followed 166 145 21 99 63

Total days followed† 99,241 88,758 10,483 60,105 36,405

All-cause deaths 37 28 9 23 13

All-cause mortality event rate, per 100 person years follow-up

13.6 [9.6:18.8] 11.5 [7.7:16.6] 31.3 [14.3:59.5] 14.0 [8.9:21.0] 13.0 [6.9:22.3]

1 year survival rate 0.873 [0.823 to 0.925], n=133

0.896 [0.847 to 0.947], n=119

0.708 [0.536 to 0.935], n=14

0.868 [0.804 to 0.938], n=82

0.888 [0.813 to 0.970], n=48

2-year survival rate 0.773 [0.706 to 0.847], n=58

0.810 [0.743 to 0.884], n=53

0.523 [0.334 to 0.820], n=5

0.759 [0.672 to 0.858], n=34

0.796 [0.692 to 0.915], n=22

* of index MitraClip procedure **4 procedures with MR aetiology not reported † calculated from MitraClip procedure date until date of death or date of HES/ONS extract (1st March 2018).

34

3.2.4 Infective endocarditis

From the linked dataset there were 3 patients readmitted to hospital with a diagnosis of

infective endocarditis (none were recorded in the registry). All 3 patients required emergency

hospital admissions (2 via A&E), which occurred 129, 155 and 392 days after discharge from

the MitraClip procedure.

3.2.5 Additional interventions

From the linked dataset 9 patients required additional MV intervention following the MitraClip

implantation: 3 required additional MV repair and 7 patients required MV replacement (with 1

patient requiring repair followed by replacement) – see Appendix 13. Additional MV

intervention was not recorded in the registry for 4 of these patients.

A total of 10 patients required an insertion of a cardiac pacemaker system and 2 required

implantation of a cardioverter defibrillator after the MitraClip implantation (see

Supplementary Material – Table 12).

3.2.6 Hospital resource usage

A total of 150 patients had the MitraClip device implanted and were discharged alive. Of

these 56 had DMR and 90 FMR (4 procedures with MR aetiology unreported). A total of 137

patients were alive at 1 year. A total of 146 patients were admitted at least once during the

year prior to their index procedure, accumulating a total of 470 pre-procedural admissions in

that time. A total of 100 patients were admitted at least once during the year after their index

procedure, accumulating a total of 251 post-procedural admissions in that time, see Table

13.

Total all-cause hospital admissions per 1000 person years were decreased in the year

following MitraClip (1763.6 post-procedure vs. 3133.3 pre-procedure, incidence rate ratio

0.53 [95% CI 0.46 to 0.62]), as were total hospitalisation days (9415.2 post- vs. 10900.0 pre-

procedure, Table 13). The mean hospitalisation days per patient decreased to 8.9 post-

procedure vs. 10.9 pre-procedure.

The aggregated hospitalisation cost (pre-procedure) was £1,267,321, with median £5,196,

[IQR £2,575 to £10,292] mean £8,449 (SD £11,033), minimum £0, maximum £82,676. There

were 4 patients (3%) who incurred no in-patient costs in the year prior to their index

procedure. The aggregated hospitalisation cost (post-procedure) was £889,721, with median

£1,402, [IQR £0 to £7,451] mean £5,931 (SD £10,688), minimum £0, maximum £90,625.

There were 50 patients (33%) who incurred no cost in the year after their index procedure.

Using a mixed effects model, the cost ratio was 0.63 (95% CI 0.50 to 0.79). This showed

MitraClip was associated with a significant reduction in hospital costs.

Hospital admissions by procedural urgency of the index MitraClip procedure (i.e. elective

and urgent) are described in Appendix 14a and b respectively. The proportion of emergency

admissions was significantly different (and numerically higher) in the 1 year post- MitraClip,

Fisher’s exact P-value <0.0001, see Table 14. Hospital admission by aetiology of mitral

regurgitation reported at the index MitraClip procedure (i.e. functional/ischaemic and

degenerative) are described in Appendix 15a and b respectively. For the DMR subgroup,

there was a significant decrease in admissions (mean 1.6 post- and 2.7 pre-procedure)

incidence rate ratio 0.64 [95%CI 0.49:0.83], no significant change in total hospitalisation

days (mean 8.4 post- and 8.1 pre-procedure) ratio 0.85 [0.49:1.47]. However the

hospitalisation cost was still reduced (mean £4608 post- and £6322 pre-procedure) ratio

35

0.65 [0.43:0.97], which just reached significance. Whereas for the FMR subgroup, there was

a decrease in total hospitalisation days per 1000 person years (8969 post- vs. 11778 pre-

procedure) ratio 0.66 [0.46:0.95], and mean hospitalisation days per patient (8.5 post- vs.

11.8 pre-procedure). The mean hospitalisation cost per patient was also reduced to £6506

post vs £9134 pre-procedure, ratio 0.62 [0.47 to 0.83]. Changes in mean hospitalisation

costs were similar for FMR and DMR.

A large reduction in heart failure related events, in terms of readmissions, hospitalisation

days and hospitalisation costs, were observed across all subgroups. However an increase in

non-cardiac admissions, hospitalisation days, and costs partly offsets reductions in cardiac

admissions post-procedure. The main cause for non-cardiac readmission was iron deficiency

anaemia (10 admissions), followed by lobar pneumonia (6), cataract (5) and acute renal

failure (5). The main cause of cardiac readmission was congestive heart failure (28

admissions), followed by chest pain (8), dilated cardiomyopathy (7), left ventricular failure (7)

(for comprehensive list see Supplementary Material – Table 13).

Additional interventions coded in the 1 year following MitraClip implantation also indicate that

5 patients required additional MV intervention (1 requiring mitral valve repair, 4 requiring

mitral valve replacement) – see Supplementary Material – Table 14.

36

Table 13: Admission and hospitalisation days in the 1 year pre- and post- MitraClip implantation obtained from linked dataset (CtE

Registry/HES/ONS).

1 year

pre-MC

admission*

(n=150)

1 year

post-MC

discharge*

(n=150)

Incidence rate

ratio [95% CI]

1 year

pre-MC

discharge*

all deaths

removed

(n=137)

1 year

post-MC

discharge*

all deaths

removed

(n=137)

Incidence rate

ratio [95% CI]

Patients with admissions 146 100 134 88

Total all-cause admissions

(rate per 1000 person years)

470 (3133.3) 251 (1763.6)† 0.57 [0.49:0.67] 418 (3051.1) 221 (1613.1) 0.53 [0.45:0.62]

Non-cardiac‡ 119 (793.3) 151 (1061.0) 1.40 [1.10:1.78] 98 (715.3) 136 (992.7) 1.39 [1.07:1.80]

Cardiac‡ 351 (2340.0) 100 (702.6) 0.30 [0.24:0.38] 320 (2335.8) 85 (620.4) 0.27 [0.21:0.34]

Cardiac (HF)‡ 195 (1300.0) 45 (316.2) 0.24 [0.18:0.34] 177 (1292.0) 38 (277.4) 0.21 [0.15:0.30]

Per patient admissions

Mean (SD) 3.1 (1.8) 1.7 (2.2) 3.1 (1.8) 1.6 (2.2)

Median [Q1;Q3] 3 [2;4] 1 [0;2] 1 [0;2] 1 [0;2]

Range 0 to 11 0 to 11 0 to 11 0 to 11

Total all-cause hospitalisation days


1635 (10,900.0) 1340 (9415.2) 0.73 [0.54:0.98] 1432 (10,452.6) 1092 (7970.8) 0.67 [0.49:0.93]

Non-cardiac‡ 338 (2253.3) 689 (4841.1) 1.62 [0.96:2.75] 259 (1890.5) 549 (4007.3) 1.71 [0.98:3.00]

Cardiac‡ 1297 (8646.7) 651 (4574.1) 0.45 [0.31:0.64] 1173 (8562.0) 543 (3963.5) 0.40 [0.27:0.60]

Cardiac (HF)‡ 784 (5226.7) 449 (3154.8) 0.53 [0.32:0.86] 677 (4941.6) 369 (2693.4) 0.50 [0.29:0.87]

37

1 year

pre-MC

admission*

(n=150)

1 year

post-MC

discharge*

(n=150)

Incidence rate

ratio [95% CI]

1 year

pre-MC

discharge*

all deaths

removed

(n=137)

1 year

post-MC

discharge*

all deaths

removed

(n=137)

Incidence rate

ratio [95% CI]

Per patient days in hospital

Mean (SD) 10.9 (19.2) 8.9 (17.2) 10.5 (19.4) 8.0 (16.3)

Median [Q1;Q3] 4 [1;12] 1 [0,8] 4 [1;11] 0 [0;6]

Range 0 to 134 0 to 81 0 to 134 0 to 81

No. of admissions with HRG codes

available (%)

463 (98.5%) 249 (99.2%) 413 (98.8%) 219 (99.1%)

Total aggregated hospitalisation cost £1,267,321 £889,721 0.62 [0.50:0.79] £1,130,425 £743,260 0.57 [0.45:0.73]

Non-Cardiac‡ £253,261 £379,176 1.42 [0.96:2.12] £189,163 £311,188 1.54 [1.01:2.35]

Cardiac‡ £1,014,060 £510,545 0.41 [0.31:0.54] £941,261 £432,072 0.36 [0.26:0.48]

Cardiac (HF)‡ £597,322 £320,061 0.39 [0.27:0.58] £543,814 £273,808 0.35 [0.23:0.52]

Per patient hospitalisation cost

Mean (SD) £8448 (£11,033) £5931 (£10,688) £8251 (£11,098) £5425 (£10,724)

Median [Q1;Q3] £5196 [£2574; £10,292]

£1402 [£0; £7451]

£4896 [£2544; £10,206]

£938 [£0; £5799]

Range £0 to £82,675 £0 to £90,625 £0 to £82,675 £0 to £90,625

Abbreviations: HF – heart failure; HRG – healthcare resource group; MC – MitraClip.

*does not include MC procedural admission

†followed until death or 1 year post-MC discharge

‡using DIAG 1 only, interpreted as reason for admission

38

Table 14: Admissions categorised by urgency in the 1 year pre- and post-MitraClip

implantation obtained from linked dataset (CtE Registry/HES/ONS).

PRE-admission

(1y pre-MitraClip)

RE-admission

(1y post-MitraClip)

Elective 311 (74.0%) 104 (41.5%)

Emergency 159 (26.0%) 147 (58.5%)

TOTAL 470 251

Fisher’s exact P-value <0.0001

3.3 SECONDARY DATA COLLECTION (LITERATURE REVIEW)

The CtE MitraClip literature search retrieved 1472 potentially relevant articles. Abstracts

from these articles were independently assessed for relevance by two EAC researchers. Of

these, 1382 were excluded immediately after screening as being not relevant to the scope.

Of the remaining 90 records, 17 were excluded for various reasons, including 8 studies that

were identified as systematic reviews and/or meta-analyses and 9 studies that reported on

economic or QoL outcomes. These were used in the economic evaluation. The remaining 73

records were identified as in scope. Where possible, full papers were retrieved and from

these, key studies were identified for full analysis by one researcher (IW for clinical evidence

review and JC for economic evidence review). Sixteen papers (12 clinical and 4 economic)

were selected for focussed review.

The process of sifting using PRISMA methodology (Preferred Reporting Items for Systematic

Reviews and Meta-Analyses) [23] is illustrated in Figure 5.

39

Figure 5. PRISMA schematic of literature search for clinical evidence.

[Note: due to the publication of two important RCTs in August and September 2018,

this review is substantively out of date. Thus conclusions may no longer be applicable.

See Section 3.4].

The CtE MitraClip literature search identified 66 publications in scope of the clinical evidence

review. Of these, 12 publications pertaining to 8 primary studies were selected for focussed

review. Selection was based on study quality, relevance (in particular concerning studies

with a relevant comparator), and study size. As a large number of single-armed

observational studies were identified, studies of this nature were only selected if they

reported data on more than 500 patients.

One of the studies was a randomised controlled trial (RCT) [24]. This study was reported in

several publications which included associated registries. This study was selected on the

basis of study design and quality. Three of the studies were relatively small prospective

observational studies. These studies were selected because they made matched

comparisons with the most appropriate comparator, conservative medical treatment. The

remaining four studies were observational registries or databases (n > 500).

40

Thirty-six primary studies identified by the searches were deemed to be technically in scope

but were not selected for focussed review. These were primarily observational studies with

more than 100 but less than 500 participants, which, due to the existence of several other

much larger studies, was subjectively considered as a threshold for selection. In addition to

the primary studies, 8 secondary studies (systematic reviews and/or meta-analyses) were

identified; 4 of which were selected for further consideration. The first sift also identified 16

economic or costing studies. Full papers were obtained for 15 of these studies, with 7

meeting the PICO criteria and 4 included in the focussed review.

The RCT identified was the EVEREST II RCT, which reported primary outcomes at 1 year

[24] and further follow up at 3 and 5 years [25, 26]. The authors of this study had previously

published the EVEREST I registry [27] (which was identified by the NICE IP664 overview [6])

and additionally the EVEREST HRS registry of high risk patients [28]. The EVEREST II RCT

randomised relatively low risk patients to receive treatment with MitraClip (n = 184) or open

surgery (n = 95). Two pre-specified primary outcomes concerning medium-term efficacy and

short-term safety were employed as well as several secondary outcomes. The EAC critically

appraised the RCT and considered it was at high risk of performance and detection bias due

to the lack of blinding. However, the study was at low risk of selection bias and there was no

evidence of selective reporting or attrition bias. In terms of external validity, the study lacked

generalisability to the NHS because of the population enrolled, who were relatively well, and

the comparator, which was open surgery.

Three comparative observational studies were identified that compared the use of MitraClip

with conservative medical management. The prospective EVEREST II HRS study [28]

enrolled patients with a high degree of surgical risk and compared them with historical

controls. The prospective study by Giannini et al. (2016) [29] enrolled patients consecutively

and assigned MitraClip to indicated patients and conventional medical management to

patients not fulfilling the anatomical criteria; otherwise the groups were closely matched.

Propensity matching was performed and the two cohorts (n = 60 each) were compared

reporting clinical outcomes. The study by Velazquez et al. (2015) [30] identified high-risk

patients from the EVEREST studies and propensity matched these with applicable patients

receiving conservative management from a hospital database. This study only reported

mortality at 1 year as an outcome.

Four large single-armed studies were identified. These were all large European registries

and were the German TRAMI registry [31, 32], the ACCESS-EU registry [33], the Pilot

European Sentinel Registry [34, 35] and the retrospective study by Rahhab et al. (2017) [36].

These studies reported real-life procedural and clinical data on MitraClip and presented

subgroup analyses.

In addition to the primary studies, the EAC identified eight systematic reviews of relevance.

Four of these studies were selected for further analysis on the basis of methodology, date of

search, and included studies. Two of the reviews reported meta-analyses that compared

MitraClip with surgical repair [37, 38]. One systematic review and meta-analysis provided a

comparative analysis of the efficacy and safety of MitraClip in patients with degenerative and

functional MR. The final systematic review focussed on the relative benefits of MitraClip in

low and high risk MR populations [39].

The EVEREST II study was a non-inferiority RCT with 12 months follow up [24].The primary

efficacy outcome (freedom from death, from surgery for mitral-valve dysfunction, and from

grade 3+ or 4+ mitral regurgitation) was achieved in 55% of MitraClip patients compared with


41

73% in the control (mitral valve surgery) arm, which was statistically inferior (p = 0.007). This

suggests that MitraClip is probably not a good option in patients who are well enough to

receive open surgery. For patients who are not well enough to undergo open surgery, there

is evidence from comparative observational studies that they may benefit from MitraClip,

with lower mortality at up to 3 years, and lower rates of readmission to hospital. However,

due the non-experimental methodology employed, these benefits have not been shown

definitively. The German TRAMI registry [31, 32], the ACCESS-EU registry [33], the Pilot

European Sentinel Registry [34, 35] and the retrospective study by Rahhab et al. (2017) [36]

reported high procedural success rates in excess of 90%. Longitudinal data reported a

mortality rate of between 15 to 20% at 1 year.

Data reported from meta-analyses of studies identified by systematic reviews indicated that, compared with open surgery, there was a trend towards MitraClip having lower short-term mortality (≤ 30 days) and equivalent longer term mortality rates, but that MitraClip was inferior at reducing MR grade. One systematic review did not report any significant difference in outcomes between patients with degenerative or functional MR with the exception of the need for MitraClip re-intervention, which was lower in patients with DMR (4% vs 10%; relative risk [RR] 0.60; 95% CI 0.38 to 0.97; p=0.04) [40]. The findings from four economic studies (considered to be of medium validity) reported that,

although MitraClip could save costs associated with hospital readmission due to congestive

heart failure, these did not offset the procedural costs (most of which are associated with the

device itself). However, there remain key uncertainties from the clinical evidence base

concerning this patient group and the extent of disease progression, and the relationship of

this with hospitalisations, life expectancy and patient symptoms. In particularl, there is a lack

of evidence concerning the longer-term clinical outcomes in high surgical risk populations.

These uncertainties will not be resolved until high-quality, comparative trials are published in

this cohort. Newcastle and York EAC is aware that there are three on-going RCTs that may

provide these data (see Section 3.3.2).

In conclusion, NY EAC found that there is now a substantial evidence base on the MitraClip

system. However, despite this, there remain key uncertainties in the use of MitraClip in

populations who are at high-risk of open surgery:

• Evidence from an RCT suggests that, in patients in whom it is a relatively safe option,

open surgery is a better option than MitraClip.

• Limited observational evidence suggests that MitraClip may confer benefits in

symptom reduction and life expectancy in patients unsuitable for open surgery.

Mortality is about 10 to 25% in the first year following MitraClip in this group.

• The costs of MitraClip are likely to outweigh the costs associated with continued

conservative medical management. However, because of a lack of direct

comparative evidence, particularly concerning longer-term outcomes in a high-risk of

surgery group, there is also considerable uncertainty about the cost-effectiveness of

MitraClip. Further research is required to acquire clinical data to inform economic

analysis that will reduce this uncertainty.

Further details are available in the standalone literature review document by Willits et al.

(unpublished, November 2017) [14].

42

3.4 EMERGING NEW EVIDENCE

3.4.1 TVT registry study

The cut-off date for the EAC literature search was August 2017. Since this date, following

completion of the RX085 literature review document [14], a new study on MitraClip has been

published that has been brought to the EAC’s attention. The EAC has briefly reviewed this

study and has concluded that, had it been published in time for the literature review

document, it would have been included for full review. Additionally, the EAC considers the

data reported by the study is of sufficient importance to be retrospectively included in this

report. Consequently, although the study has not been fully appraised, the study is briefly

described in this section, and results have been included to inform the NHS England

questions (Section 3.4), where relevant.

The study by Sorajja et al. (2017, henceforth referred to as the TVT registry) was based on

published data from the Society of Thoracic Surgeons/American College of Cardiology

Transcatheter Valve Therapy (STS/ACC TVT) registry [41]. Patients were eligible for

inclusion if they had severe (grade 3 or 4) primary (degenerative) MR and were considered

to be at prohibitive surgical risk, as measured by STS predictive risk of mortality criteria. The

study reported data from all patients registered in the TVT registry who underwent

commercially funded therapy with the MitraClip system between the initial U.S. Food and

Drug Administration approval (October 2013) and September 2015. Data reported included

patient characteristics, procedural outcomes, and clinical and resource use outcomes at up

to 1 year follow up. Subgroup comparisons were made between functional and degenerative

MR, degree of tricuspid regurgitation, and post-procedural MR grade.

The TVT registry enrolled 2952 patients, 85.9% of whom had a diagnosis of DMR. The

broad characteristics of the population are reported in Table 16. The registry reported an

acute procedural success rate of 92% and that the large majority of patients were

discharged with mild or absent MR (grade ≤2 of 95.5%). There was an overall 30 day

mortality rate of 5.2%. The death rate at 1 year was 25.8% and the rate of re-hospitalisation

due to heart failure was 20.2%; with repeat mitral valve procedures required in 6.2% of

patients. Multivariate subgroup analysis indicated inferior outcomes (mortality and

readmission to hospital) associated with FMR compared with DMR (see Question 8).

Tricuspid regurgitation and persistence of MR at discharge were also reported as being

associated with significantly increased risk of death and re-hospitalisation at 1 year follow

up.

3.4.2 RCTs

Two important RCTs have recently been published that investigated the use of MitraClip in

the treatment of people with secondary MR (FMR). These studies were published after the

substantive work for this document was complete, including data linkage and the publication

of the TVT registry. Because of this timing, these studies are only reported briefly here.

However, the publications are highly relevant to the CtE project and should be fully

considered in the context of NHS England decision making.

MITRA-FR trial

The MITRA-FR trial (Percutaneous Repair with the MitraClip Device for Severe

Functional/Secondary Mitral Regurgitation) by Obadia et al. was published on August 27th

2018 [42]. This study enrolled 304 patients (from 37 centres) with severe secondary MR

(defined as an effective regurgitant orifice area of >20 mm2 or a regurgitant volume of >30 ml

43

per beat), and randomised them to receive percutaneous mitral valve repair using the

MitraClip system or medical therapy in a 1:1 ratio (n = 152 in each arm). The primary

efficacy outcome was a composite of death from any cause or unplanned hospitalization for

heart failure at 12 months. The trial was supported by the French Ministry of Health and

Research National Program and Abbott Vascular (who supplied and funded most of the

devices).

Using ITT analysis, the rate of the primary outcome was 54.6% (84/152) in the intervention

group compared with 51.3% (78/152 patients) in the control group (OR 1.16; 95% CI 0.73 to

1.84, p = 0.53). There was no difference in the rate of death from any cause, with 24.3%

(37/152) dying in the intervention group compared with 22.4% (34/152) in the control group

(HR 1.11, 95% CI, 0.69 to 1.77). There was also no difference in the rate of re-

hospitalisation, with 48.7% being readmitted in the intervention group compared with 47.4%

in the control group (HR 1.13; 95% CI, 0.81 to 1.56). Thus there were no significant

difference in the primary outcome or the components of this outcome, and in fact the point

estimate was slightly worse in the MitraClip arm. The device was not successfully implanted

in 9.2% (12/152) patients; however, per protocol analysis also did not identify any significant

differences between the groups. The authors considered that other patient data, such as MR

grade, NYHA class, and quality of life, had insufficient follow up to allow for publication.

The authors concluded that the use MitraClip did not improve the rate of mortality or

rehospitalisation in this patient population. The reasons for this lack of measured benefit

were unknown, but the authors speculated the enrolled population may have had heart

failure that was too advanced to be successfully treated with this technology. That is,

secondary MR may have been a marker of the progression of the underlying

cardiomyopathy, rather than a direct contributor to the pathophysiology of heart failure. It

was also noted that the introduction of angiotensin receptor–neprilysin inhibitor drugs to

France at the time of the trial may have conferred additional benefit to the control group.

COAPT trial

The COAPT study (Cardiovascular Outcomes Assessment of the MitraClip Percutaneous

Therapy for Heart Failure Patients with Functional Mitral Regurgitation) was published on the

23rd of September 2018 in the same journal as the MITRA-FR study (New England Journal

of Medicine) [43]. The study recruited 614 patients with heart failure and moderate-to-severe

or severe FMR from 78 sites, and randomised them in a 1:1 ratio to receive treatment with

MitraClip (n = 302) or medical therapy alone (control, n = 312). There were two primary

endpoints. The primary effectiveness outcome was all hospitalisations for heart failure within

24 months of follow-up. The primary safety outcome was freedom from device-related

complications at 12 months (compared with a pre-specified objective performance goal of

88.0%).

The intervention group reported a 24 month hospitalisation of 35.8 per 100 PY compared

with 67.9 per 100 PY for the control group; the difference of which significantly favoured

MitraClip (HR 0.53; 95% CI 0.40 to 0.70; p < 0.001). The rate of freedom from device-related

complications was 96.6% (lower 95% CI 94.8%, p < 0.001 compared with performance

goal). At 24 months, there were 29.1% deaths from any cause in the intervention group. The

corresponding figure for the control group was 46.1%, which was significantly higher (HR

0.62, 95% CI, 0.46 to 0.82; p < 0.001). At 12 months follow up, MitraClip was non-inferior to

medical management for mortality (19.1% vs. 23.2%, p < 0.001 for non-inferiority). The

authors also reported that MitraClip was superior to medical management alone for a series

44

of secondary outcomes including MR grade of ≤ 2+ at 12 months; change in KCCQ score

(Kansas City Cardiomyopathy Questionnaire) at 12 months; NYHA class at 12 months;

6 minute walk test at 12 months; and change in left ventricular end-diastolic volume

(12 months compared with baseline).

The authors concluded that percutaneous mitral valve repair using MitraClip was associated

with a decreased risk of death and hospitalisation, and had an acceptable rate of device-

related complications.

Comparison of RCTs

The two recently published RCTs appeared to enrol similar populations (patients with MR

secondary to heart failure) and had related primary effectiveness outcomes. However, whilst

the MITRA-FR study reported similar point estimates for the intervention and control groups

at 12 months (that is, MitraClip did not confer additional benefit) [42], the COAPT study

found that MitraClip was associated with significantly fewer deaths or cases of

rehospitalisation over a 2 year period [43]. Furthermore, these differences were large and

clearly of clinical and economic benefit.

The reasons for the conflicting results are not fully understood but have been speculated on

[44]. Often, in situations where one trial reports negative results in contrast to another which

reports positive results, there are differences in the characteristics of the population. Whilst

the populations of both trials were nominally similar, analysis of baseline characteristics

shows there may have been important differences. Investigators in the COAPT study found

that two subgroups did not benefit: this was those with a relatively small amount of MR

(effective regurgitant orifice area [EROA] of < 30 mm2), and those with a large left ventricular

volume (> 96 mL/m2). In the MITRA-FR trial 52% of patients in MITRA-FR had an EROA of

< 30 mm2 compared with 14% in the COAPT. It has been hypothesised that patients won’t

benefit if they have a small amount of volumetric MR from a particularly large left ventricle

(that is, correction of MR in these patients is not great enough to make a clinically significant

difference).

Other possible reasons for the differences seen in the trials might relate to operator

experience and differences in the use of heart failure drugs in the intervention and control

arms. The more pragmatic nature of the MITRA-FR study perhaps underlines the need for

very careful patient selection if the results of the COAPT study are to be achieved in clinical

practice. Additionally, it must be remembered that neither trial has reported additional

information on the efficacy and safety of MitraClip with DMR aetiology.

3.4.3 On-going randomised controlled trials

The EAC also identified protocols for four RCTs that were currently on-going (two of which

have subsequently been published (Section 3.4.2). [14]. These are summarised in Table 15.

Upon completion and publication, these studies are expected to address some of the

uncertainties in the extant evidence base.

45

Table 15. Summary of on-going RCTs on MitraClip (identified from clinicaltrials.gov).

Study characteristics Setting and Population Intervention and Comparator Key outcome(s)*

Comment

ReShape-HF2 trial [45] Parallel open-label RCT Follow up 12 months

Europe (Denmark, Germany, Greece, Italy, Poland, Portugal, Spain) Patients with clinically significant functional MR (moderate to severe or severe MR) Estimated enrolment: 420 subjects

I: MitraClip C: Standard medical care

Composite rate of recurrent heart failure hospitalisations and CV death MR grade MYHA grade QoL

Started: May 2015 Status: recruiting Estimated primary completion date: July 2018** Sponsor: Institut fuer anwendungsorientierte Forschung und klinische Studien GmbH

MATTERHORN trial [46] Parallel open-label RCT Follow up 12 months

Germany Patients with moderate-to-severe MR of primarily functional pathology and reduced left ventricular function considered to be at high surgical risk Estimated enrolment: 288 subjects

I: MitraClip C: Mitral valve surgery

Composite of death, rehospitalisation for heart failure, re-intervention (repeat operation or repeat intervention Recurrence of grade 3 or 4 MR within 12 months NYHA class Length of hospital stay

Started: February 2015 Status: recruiting Estimated primary completion date: December 2019* Sponsor: Abbott Vascular

Abbreviations: C – comparator; COAPT - Cardiovascular Outcomes Assessment of the MitraClip Percutaneous Therapy for Heart Failure Patients with Functional Mitral Regurgitation Trial; CV – cardiovascular; HF – heart failure; I – intervention; MACCE – Major Adverse Cardiac and Cerebrovascular events; MATTERHORN - Multicenter, Randomised, Controlled Study to Assess Mitral vAlve reconsTrucTion for advancEd Insufficiency of Functional or iscHemic ORigiN trial; MITRA-FR - Multicentre Randomised Study of Percutaneous Mitral Valve Repair MitraClip Device in Patients With Severe Secondary Mitral Regurgitation trial; MR – mitral valve regurgitation; QoL – quality of life; ReShape HF2 - RandomisEd Study of tHe MitrACliP DEvice trial; RCT – randomised controlled trial. * Primary outcome in italics. ** Final data collection date for primary outcome measure.

https://clinicaltrials.gov/

46

3.5 NHS ENGLAND QUESTIONS

The aims of the CtE registry were to provide data on the safety, efficacy and costs of

MitraClip in the real-world NHS setting and specifically to answer 11 pragmatic questions

concerning these issues. In this section, the findings from the CtE registry are used to

answer these questions and are presented in the context of published studies in other

populations. Table 16 summarises some key characteristics that illustrate the risk profile of

patients undergoing MitraClip in the CtE programme, key clinical trials and large

observational studies.

Qualitative analysis of the baseline characteristics of the studies highlights that the

EVEREST II RCT (2011) [24] was conducted in a younger, healthier population (using

surgical risk and NYHA grade as proxies for overall health) than the other studies, including

the CtE registry. Patients enrolled into the EVEREST II study were therefore not

representative of the patients in the selected observational studies or the CtE registry.

The CtE registry was conducted in a similar population to the observational studies. In

general terms, the population of the observational studies was comprised predominantly of

men in their mid to late 70s, with FMR the most common diagnosis (with the notable

exception of the large TVT registry [41]). Nearly all patients enrolled in MitraClip studies had

moderate or severe MR at baseline, and this was reflected in the observational studies by

moderate to severe symptoms of dyspnoea. Finally, the surgical risk reported in the CtE

registry was consistent with those reported in other observational studies; that is, the surgery

was not an appropriate option in most of these patients.

47

Table 16. Summary of patient characteristics in the CtE registry and published literature (for comparative studies, MitraClip group reported).

Study Mean age ± SD (years)

Sex (% male) Aetiology MR grade (% ≥ 3+)

NYHA grade (% ≥ III)

Urgency Surgical risk ± SD*

CtE registry (n=199)

76.2 ±10.5 68.8 FMR: 60% DMR: 40%

100.0 92.4 Elective: 84.4% Urgent: 13.6% Emergency: 2.0%

Mean ES: 20.1 ± 15.6 Mean ES2: 6.4 ± 5.7

Linked dataset (CtE registry/HES/ONS) (n=166)

76.1 ±10.6 68.7 FMR: 61% DMR: 39%

NA N/A Elective: 87.3% Urgent/emergency: 12.7%

Mean ES: 20.0 ± 15.2 Mean ES2: 6.2 ± 5.5

EVEREST II RCT** (2011) [24] (n=279)

67.3 ± 12.8 62 FMR: 27% DMR: 77%

96 52 Elective: 100% NR

EVEREST II HR study ***(2012) [28] (n=114)

76.7 ± 9.8 62.8 FMR: 41.0% DMR: 59.0%

98.7 89.7 Elective: 100% Mean STS: 14.2 ± 8.2

Giannini (2016) [29] (n=120)

75 ± 8 67 FMR: 100% 95 75 Elective: 100% Mean ES: 17 (IQR 11 to 28) Mean ES2: 6 (IQR 4 to 120

Velazquez (2015) [30] (n=478)

75.7 ± 10.5 61.0 FMR: 70.1% DMR: 29.9%

100 84.9 Elective: 100% NR

TRAMI registry (2012) [31] (n=828)

Median 75 (IQR 70 to 80)

59 FMR: 67% DMR: 33%

100** 93 Elective: 91% Emergency: 9%

Median ES: 23 (IQR 12 to 38)

ACCESS-EU study (2013) [33] (n=567)

73.7 ± 9.6 63.8 FMR: 69% DMR: 31%

100 84.9 NR Mean ES: 23.0 ± 18.3

European Pilot Sentinel study (2014) [34] (n=628)

73.7 ± 9.7 63.1 FMR: 72% DMT: 28%

93.3** 85.5 NR Mean ES: 24.7 ± 16.7

Rahhab (2017) [36] (n=1151)


59 FMR: 72% DMR: 17%

99 NR NR NR

TVT registry (2017) [41] (n=2952) †


55.8 FMR: 8.6% DMR: 85.9% Mixed: 8.9% Post-inflammatory:

93.0 85.0 NR Median STS for MV repair: 6.1 (IQR 3.7 to 9.9) Median STS for MV

48

0.7% Other/indeterminate: 2.8%

replacement: 9.2 (6.0 to 14.1)

MITRA-FR RCT (2018) [42] (n=152) †

70.1 ± 10.1 78.9 FMR: 100% >99% (estimate from graph)

53.1 Elective: 100% Median ES2: 6.6 (IQR 3.5 to 11.9)

COAPT RCT (2018) [43] (n=301) †

71.7 ± 11.8 66.6 FMR: 100% 100% 57.0 Elective: 100% Mean STS: 7.8 ± 5.5

Abbreviations: NA – not applicable; DMR – degenerative MR; ES – Logistic EuroSCORE; ES2 – Logistic EuroSCORE II; FMR – functional MR; IQR – inter-quartile range; MR – mitral valve regurgitation; MV – mitral valve; NR – not reported; SD – standard deviation; STS – Society of Thoracic Surgeons (score).

* * EuroSCORE II is a risk-stratification model that estimates the mortality risk following cardiac surgery [47]. It has replaced the EuroSCORE and logistic EuroSCORE algorithms which overestimate the risk of death from cardiac surgery [48]. The STS calculator is a US algorithm that provides similar estimates of risk of surgical mortality [49]. ** Data reported from MitraClip arm. No significant differences were reported between MitraClip and surgery arms. *** MR graded as I (mild), II (moderate), and III (severe). Value refers to proportion of patients who were grade II or III. † Studies published subsequent to first drafts of this document and not included in original literature review. Studies not critically appraised.

49

3.5.1 Question One

“Can UK clinical teams undertaking MitraClip reproduce the reduction in mitral

regurgitation seen in the early clinical trials?”

The mechanism of action of percutaneous repair of the mitral valve using MitraClip, through

which clinical and patient benefits are derived, is to reduce MR. The severity of MR has been

measured in the registry by categorisation into MR grades (of none, 1+, 2+, 3+, or 4+) [50],

which have been dichotomised into grades of ≤MR 2+ and ≥MR 3+, which broadly speaking

demarcates patients with absent or mild MR from patients with moderate or severe MR.

Summary data reporting MR grades of this nature has been used extensively in the

published literature allowing for direct comparisons with other relevant studies (see EAC

Literature Review for further detail [14]).

All except one of the patients recruited into the CtE registry were in the ≥MR 3+ category at

baseline. This was in full concordance with the inclusion criteria which specified “Patients

with severe grade 3 or 4 mitral regurgitation” [5]. Following the MitraClip procedure, there

was an immediate clinical and statistically significant impact on MR, with 93% of patients

being classified as ≤MR 2+ on discharge from hospital. This effect was only slightly

diminished at later time points, with 76%, 81%, 76%, and 90% being classified as ≤MR 2+ at

6 week, 6 months, 12 months and 24 months follow up respectively.

The distribution of MR grades of patients in the registry is illustrated in Figure 6.

Dichotomised MR grades are illustrated in Figure 7. As can be seen, there is an immediate

and substantial improvement in MR grade associated with the MitraClip procedure. Figure 8

plots how the MR grade of individual patients changed over the course of follow up. Nearly

all patients are observed to move down at least one MR grade at discharge. Subsequent to

this, most patients remained at the lower grade of MR, but a minority also moved up one or

more classes.

A comparison of dichotomised MR grades of patients in the CtE registry compared with data

from published studies is reported in Table 17. One of these studies, the EVEREST II trial

[25, 27], was an RCT with controlled comparative data. This study, performed in a population

at lower surgical risk than those enrolled in the CtE registry, reported similar efficacy in terms

of MR grade reduction for MitraClip and open surgery, and both were similar to the CtE

registry. Since the publication of EVEREST II and subsequent to earlier drafts of this

document, two new RCTs have been published exclusively in patients with FMR. Reporting

of this outcome was limited in the MITRA-FR study, with poor follow up of MitraClip patients

and no follow up of control patients for this outcome [42]. However, in the COAPT study, a

large effect in MR reduction was observed in the MitraClip arm only, with patients receiving

medical management reporting no benefit (94.8% MR grade ≤2+ vs. 46.9%, p < 0.001) [43].

The observational studies that included a medical management arm did not report

longitudinal MR data on the comparator [28-30]. However, single armed observational

studies all reported similar MR outcomes to the CtE registry [31, 33-36]. That is, there was

an immediate and significant reduction in MR grade at discharge (typically with over 90% of

patients achieving mild or absent MR). At 12 months follow up, around 20% of patients in the

observational studies had moderate or severe MR, indicating that most patients had

sustained mitral valve integrity.

50

Figure 6. Proportional distribution of MR grades at baseline and following the

MitraClip procedure observed in the CtE registry. Numbers above bars report raw data

count.

51

Figure 7. Dichotomised proportional MR grades at baseline and following the

MitraClip procedure. Numbers above bars report raw data count.

52

Figure 8. Graphical representation of individual patients moving between MR grades

at follow up periods.

53

Table 17. Severity of MR reported in the CtE registry and selected published literature before and after the MitraClip procedure.

Study Proportion of patients with significant MR (MR grade of 3+ or 4+ or equivalent) (%)

Baseline Post-procedure 6 weeks 6 months 12 months

CtE registry

100 7 24 19 24

EVEREST II RCT [25, 27]*

MitraClip arm (n = 194) 96 21

Surgery arm (n = 95) 93 20

EVEREST HR study [28]

98.7 22.2

TRAMI registry [31, 32]**

99 3

ACCESS-EU study [33]

97.7 7.8 21.1

Pilot European Sentinel Registry [34, 35]**

99.3 27.2

Netherlands study [36]

99 6

TVT registry*** [41] 93.0 7.0 5.1

MITRA-FR RCT***† [42]

MitraClip arm 100 9 17

COAPT RCT*** [43] MitraClip arm (n = 302)

100 5.2

Surgery arm (n = 312) 100

53.1

Abbreviations: MR – mitral valve regurgitation (grade). * Five year follow up reported as significant MR of 12% in MitraClip arm and 2% in surgery arm. ** In these instances, MR grade was classified as none, mild, moderate or severe. The EAC classified moderate or severe MR as significant. Shaded areas report absence of data reported. *** The TVT registry, MITRA-FR RCT and COAPT RCT were identified after the cut off dates for the literature search and have not been critically appraised (see Section 3.3.1). † Data estimated from graph in supplementary material. Data for control arm not reported.

54

Although the EVEREST II study reported that MitraClip was as effective as surgery at

reducing MR, differences (in favour of surgery) were reported in some comparative

observational studies. For instance, in the study by Alozie et al. (2017) [51], none of the

surgically treated patients had residual post-operative MR (MR grade 2 or 3), compared with

23.8% of patients treated with MitraClip (p < 0.001). Similarly, the study by Taramasso et al.

(2012) [52] reported freedom from MR grade ≥3+ at 1 year was 79.1 ± 8% (SD) for MitraClip,

compared with 94 ± 2% for surgery (p = 0.01). The study by De Bonis (2016) [53] reported

freedom from MR grade ≥3+ at 4 years follow up was 75 ± 7.6% (SD) in the MitraClip group,

compared with 94 ± 3.3% in the surgical cohort (p = 0.04). This is discussed further in

Question Five.

Evidence from systematic reviews reflects that of the constituent studies. For instance, the

systematic review by Vakil et al. (2014) reported that the mean MR at baseline of the

included studies was 3.56 (± 0.1 SD). Following the MitraClip procedure, this reduced to 1.8

(± 0.1) [39]. The systematic review by Takagi et al. (2017) [37] compared percutaneous

repair with MitraClip with surgical repair. It reported freedom from recurrent MR (≥3+) at

1 year was 84.0% for MitraClip compared with 97.3% for surgery. At 3 years, the respective

figures were 75.0% and 96.0%. This difference was significant, with a pooled hazard ratio

(HR) for moderate or severe MR recurrence of 4.80 (95% confidence interval [CI] 2.58 to

8.93; p = 0.00001) for MitraClip.

Conclusion

Evidence reported from the CtE registry shows that percutaneous mitral valve leaflet repair

using the MitraClip system results in immediate and significant improvements in mitral valve

function, as measured by MR grade. Immediately following the procedure, the proportion of

patients with moderate-severe or severe MR (≥3+) reduced from 99.5% to 6.7%. This large

effect is consistent with all the studies identified in the literature. However, after 1 year there

is evidence that there is some deterioration in mitral valve function, with 24.4% of patients

reporting moderate-severe or severe MR. Again, this is fully consistent with the published

literature.

The new COAPT study has confirmed no improvement in MR grade results from continuing

optimal medical management alone [43]. The EVEREST II RCT reported that reduction in

MR grade was similar for both MitraClip and surgery patients. However, this is contradicted

by several observational studies which have reported surgery is superior to MitraClip in

reducing MR, especially in the longer term.

55

3.5.2 Question Two

“Is reduction in mitral regurgitation mediated by MitraClip associated with

improvements in quality of life?”

Quality of life (QoL) has been captured in the registry using the validated EQ-5D-5L

(EuroQoL 5 dimension 5 levels) system [54] favoured by NICE. This reports data of personal

wellbeing in five domains: mobility, self care, usual activities, pain and/or discomfort, and

anxiety and/or depression. Five point scores from these domains are used to calculate a

single utility score measuring from negative values (indicating wellbeing worse than death),

zero (death), to one (perfect health). It also comprises a visual analogue scale (VAS) which

can be used as a validatory measure. Quality of life data were captured at baseline,

6 weeks, 6 months, 12 months, and 24 months. Changes in QoL were analysed statistically

using paired analysis, and summary data were also presented.

The New York Heart Association (NYHA) classification system is a measurement of

dyspnoea (symptom of heart failure), which is closely related to physical QoL, and as such

can be considered a surrogate outcome. An NYHA class I represents no limitation of

physical activity, and class IV represents inability to conduct any activity without physical

discomfort [55]. The NYHA scale is particularly useful as it is widely reported in studies of

mitral valve dysfunction.

Quality of life

The utility scores for patients who received MitraClip in the registry and reported EQ-5D

outcomes at baseline and follow up (paired data) is reported in Table 18, with additional

detail in Supplementary Material – Table 9. It can be seen that there is a significant

improvement observed in utility measured at 6 weeks, and this improvement is sustained

throughout the course of the study (although significance is lost at 2 years because of limited

follow up). This upward trend and plateau is illustrated in Figure 9.

Table 18. Utility scores of patients in CtE registry over time (paired data).

Time point Paired (with baseline) median utility score (Q1, Q3 quartiles) Number of participants n

Paired (with baseline) mean utility score (SD)

Mean change in utility score (SD)

Statistical significance compared with baseline (paired test)

Baseline (reference) 0.6 (0.45, 0.72) 163

0.55 (0.23) Reference Reference

6 weeks 0.78 (0.66, 0.88) 136

0.75 (0.21) 0.18 (0.23) p < 0.0001

6 months 0.81 (0.71, 0.88) 113

0.77 (0.20) 0.20 (0.24) p < 0.0001

1 year 0.78 (0.67, 0.90) 48

0.73 (0.24) 0.15 (0.22) p < 0.0001

2 years 0.74 (0.68, 0.85) 7

0.75 (0.18) 0.21 (0.24) p = 0.059

56

Figure 9. Changes in utility scores over time (paired data) observed in the CtE

registry.

Analysis of individual EQ-5D domains showed trends for improvement at all time-points,

which were mostly significant up to 1 year (Table 19). Additionally, there were more patients

reporting improved rather than deteriorating utility compared to baseline at all follow up time

points. Improvements in utility were matched by changes in the visual analogue score (VAS)

[56], which was measured as 50 mm (IQR 25 to 65 mm) at baseline, and 70 mm (50 to

80 mm), 70 mm (60 to 80 mm) and 80 mm (57.5 to 85 mm) at 6 weeks, 6 months and

12 months respectively. This change at 1 year is illustrated in Figure 10.

Table 19. Statistical improvement from baseline in each EQ-5D domain reported in the

CtE registry.

Time point Statistical improvement compared with baseline*

Mobility Self care Usual activities

Pain/discomfort Depression/anxiety

6 weeks Yes p < 0.0001

Yes p < 0.0001

Yes p < 0.0001

Yes p = 0.0001

Yes p < 0.0001

6 months Yes p < 0.0001

Yes p = 0.0001

Yes p < 0.0001

Yes p < 0.0001

Yes p = 0.0001

12 months Yes p = 0.016

No p = 0.391

Yes p = 0.0001

Yes p = 0.0009

Yes p = 0.0061

24 months No p = 0.254

No p = 1.000

No p = 0.45

Yes p = 0.0128

No p = 1.000

* Paired comparison with baseline data using Fisher’s Exact test.

57

Figure 10. VAS at baseline and 1 year.

There are only limited data on QoL associated with MitraClip reported in the literature. The

EVEREST II study [24] measured QoL using Short Form-36 (SF-36) at pre-procedure

(baseline), 30 days, and 12 months. It reported significant positive changes in both mental

and physical components associated with MitraClip at these time points. In contrast, there

was a significant decrease in the physical component at 30 days associated with surgery.

However, physical and mental components were significantly improved with surgery after

12 months. The generalisability of these results should be treated with caution because the

EVEREST II trial recruited a healthier population than the CtE registry. However, similar

improvements in QoL (SF-36) were observed in the EVEREST II HR observational study

[28], and the TRAMI registry reported significant improvements in utility at 12 months as

measured by EQ-5D-3L and VAS [31].

The systematic review by Vakil et al. (2014) [39] reported an increase in the physical

component of QoL, as measured by SF-36, from 32.5 ± 1.3 standard error (SE) at baseline

to 40.1 ± 2.0 at follow up (3 studies). For the mental component, the change was from

44.3 ± 0.5 to 52.2 ± 2.6. The significance of these changes was not stated.

NYHA

NYHA dyspnoea scores observed by the CtE registry are reported in Table 20. Scores are

dichotomised such that I and II represents patients with no or moderate limitations to activity,

whereas III and IV represents marked or extreme limitation and discomfort. As can be seen,

the MitraClip procedure is associated with an immediate improvement of NYHA scores at

6 weeks, which persists for up to 2 years and which is significant. This improvement is

represented graphically in Figure 11.

58

Table 20. NYHA class at baseline and follow up reported in the CtE registry.

Time point NYHA class I or II % (frequency, n)

NYHA class III or IV % (frequency, n)

Statistical significance compared with baseline*

Pre-procedure (baseline)

7.5 (14) 92.4 (172) N/A

6 weeks 81.9 (122) 18.1 (27) p < 0.0001

6 months 82.5 (94) 17.5 (20) p < 0.0001

1 year 82.1 (55) 17.9 (12) p < 0.0001

2 years 100 (10) 0 (0) p < 0.0001

* Fisher’s Exact test (paired data).

Figure 11. NYHA class at baseline and follow up reported in the CtE registry. Numbers

above bars report raw data count.

The CtE results for improvements in NYHA class are consistent with those seen in the

published literature, as reported by Table 21. At 12 months, all the studies that included

NYHA class as an outcome reported statistically significant results compared with baseline.

Similar results were observed in other retrospective [57-59] and prospective [60-64]

observational studies.

59

Table 21. Comparison of CtE registry NYHA data with published literature.

Study Proportion of patients with NYHA class III/IV at baseline (%)

Proportion of patients with NYHA class III/IV at 12 months follow up (%)

Absolute change from baseline (%)

CtE registry 92.4 17.9 74.5

EVEREST II [27]


Surgery arm 47 13 34

EVEREST HR study [28] 74.1 25.9 48.2

TRAMI study [31] 93 37 56

ACCESS-EU study [33] 84.9 28.6 56.3

European Sentinel study [34] 84.1 25.8 58.3

MITRA-FR RCT [42]*†


Medical management arm

68 33 35

COAPT RCT [43] *

MitraClip arm 57.0 27.8 29.2

Medical management arm

64.6 50.4 14.1

* Studies published subsequent to original draft of this document. These have not been critically appraised. † Data reported graphically in supplementary material.

Conclusion

The CtE registry reported significant improvements in QoL, as measured by utility and VAS,

at all follow up time points up to 1 year, compared with baseline. Baseline EQ-5D index

scores were similar to those reported for 75 year old people with heart failure in the literature

[65]. There were statistically significant improvements in all domains of EQ-5D at 6 months.

These results were consistent with results reported in the EVEREST II RCT [66] and two

observational studies [28, 31] which measured QoL at baseline and 12 months. The use of

MitraClip was also associated with an immediate and sustained (up to 2 years) improvement

in NYHA class, with the large majority of patients experiencing resolution from moderate or

severe dyspnoea to mild or absent dyspnoea. Similar improvements in NYHA class were

also widely demonstrated in the published literature.

60

3.5.3 Question Three

“Does MitraClip improve survival rates?”

The CtE registry was unable to directly answer this question because it did not have a

comparator group (of patients who did not receive MitraClip). The mean age of patients in

the CtE registry was 76.2 years (± 10.5 SD). The age-specific annual mortality rate of people

aged 75 to 79 years of age is 3.4% [67]; this value can be considered as a “background”

annualised incidence of death in the general population; however it is not representative of

the population receiving MitraClip in the registry, who were generally in much poorer health

than an average individual of this age (as indicated by their high EuroSCORE values, see

Section 3.1.2). However, indirect comparisons with limited published data have been made

(see below).

Using data directly reported form the registry, ten patients died in-hospital; these deaths can

be considered to be directly attributable to the procedure (see Section 3.1.4 for individual

recorded causes of death). This resulted in an in-hospital death rate of 5.0% (95% CI 2.4 to

9.0%). The death rate at 30 days (12 individuals) was slightly higher at 6.0% (95% CI 3.2 to

10.3%). A further 20 patients were recorded as dying during follow up of up to 2 years, giving

an overall death rate of 15.1% (95% CI 10.4 to 20.8%). The death rate at 1 year was 11.6%

(95% CI 7.5% to 16.8%). However, the annualised death rate using incidence data derived

from a total of 111.2 recorded person years (PY), was 27.0 (95% CI 18.2 to 38.5%) per 100

PY. This ostensibly higher rate was likely due to early deaths associated with the procedure

and censoring of patients at later time points. Kaplan-Meier analysis of these data is

reported in Appendix 7.

Using the linked data, a mortality rate of 12.7% (95% CI 7.5 to 15.9%) was reported at

1 year, and 22.7% (95% CI 15.3 to 29.4%) at 2 years. The latter value was numerically

higher than the registry estimate, reflecting the identification of several more deaths

(principally through ONS). However, the improved coverage (272 PY) meant that the

annualised mortality rate was lower, at 13.6% (95% CI 9.8 to 18.8) per 100 PY. As linked

data were considered to be more comprehensive than the registry data, these data are

preferred for analysis and comparisons with other studies.

The rate of death reported in the registry is compared with those in the published literature in

Table 22. Four of the studies reported comparative data. The EVEREST II study was an

RCT which compared MitraClip with surgery [24]. However, participants in this trial were

significantly healthier than those enrolled into the CtE registry, which were patients “deemed

by an MDT to be at too high risk or at high risk for conventional mitral valve surgery” [5]. Two

recent RCTs have been published that compared MitraClip with medical management in

patients with FMR [42, 43]. These studies are considered highly generalisable to the

population of interest as they were unsuitable for surgery.

Three observational studies used optimal medical management as the comparator. These

comparator groups were derived from retrospective [28, 30] or prospective [29] cohorts and

were subject to considerable confounding and potential for bias. The remaining studies were

single armed observation studies which enrolled similar populations to that of the CtE

registry, and thus provide a relevant comparison of clinical performance.

The rate of in-hospital death, or death within 30 days, was 6% in the CtE registry, which is

broadly comparable with results reported for MitraClip in the observational studies, which

ranged from 0% to 7.7%. Longer-term follow up mortality data was most comprehensively

61

reported at 12 months. The 1 year rate of death of around 12% in the CtE registry was within

the expected range as reported in published studies, which were between 6% and 24.6%.

The 1 year mortality rate was significantly lower in the MitraClip cohorts than medical

management cohorts in the studies where it was reported. In the study by Giannini et al.,

which prospectively matched patients undergoing medical management, 10.3% died after

1 year in the MitraClip cohort compared with 35.7% in the medical management cohort [29].

This study also reported a mortality rate of 38.6 % and 65.1% at 3 years for MitraClip and

medical management respectively (HR 2.31 [95% CI 1.30 to 4.09, p = 0.007]).

For comparison, the EAC did not identify any epidemiological studies that reported on

patients closely equivalent to those enrolled in the CtE registry. However, a meta-analysis of

individual data of UK patients with heart failure (n = 39,372) reported a death rate of 40.2%

after a median follow up of 2.5 years [68].

62

Table 22. Comparison of mortality data reported by the CtE registry and published literature.

Study Mortality rate (%) Comment

30 days (unless

otherwise stated)

1 year 2 years Longer

follow up

CtE registry

(n=199)

6.0 (95% CI 3.2 to

10.3%)

11.6 (95% CI

7.5 to 16.8)

15.1 (95% CI

10.8 to 21.4)

Annualised mortality rate:

27.0 (95% CI 18.2 to 38.5)

per 100 PY

Linked dataset (CtE registry/HES/ONS)

(n=166)

4.8 (in hospital) 12.7 (95% CI

7.5 to 15.9)

22.7 (95% CI

15.3 to 29.4)

Annualised mortality

rate: 13.6 (95% CI 9.8 to

18.8) per 100 PY

Preferred dataset.

Com

para

tive

stu

die

s

EVEREST II RCT

[24, 25]

(n=279)

MC arm (n=184) 1 6 11 21 (5 years) No significant difference at

1 year (p=1.00) Surgery arm (n=95) 2 6 11 27 (5 years)

EVEREST II HR

study [28]

(n=114)

MC cohort (prospective,

n=78)

7.7 24.6 MC associated with

significantly reduced

mortality at 1 year

(p=0.047) MM/surgery cohort

(retrospective, n=36)

44.7

Giannini (2016) [29]

(n=120)

MC cohort (n=60) 0.0 10.3 28.8 38.6 (3 years) HR (MM vs. MC at

3 years): 2.31 (95% CI

1.30 to 4.09, p=0.007) MM cohort (n=60) 1.7 35.7 48.3 65.1 (3 years)

63


30 days (unless

otherwise stated)


follow up

Velazquez (2015)

[30]

(n=478)

MC cohort (n=239)

4.2 22.4 Adjusted HR (MC vs. MM

at 1 year) 0.66 (95% CI

0.45 to 0.99; p=0.043)

MM cohort (historical,

n=239)

7.2 32.0

MITRA-FR RCT [42]*

(n=304)

MC arm (n=152) 3.3 24.3 HR at 1 year (MC vs. MM):

1.11 (95% CI 0.69 to1.77,

p=ns) MM arm (n=152) 2.6 22.4

COAPT RCT [43]*

(n=614)

MC arm (n=302) 5.3 19.1 29.1 HR at 1 year (MC vs. MM):

0.81 (95% CI 0.57 to1.15,

p=ns)

HR at 2 years (MC vs.

MM): 0.62 (95% CI 0.46 to

0.82, p<0.001)

MM arm (n=312) 23.2 46.1

Sin

gle

arm

ed s

tudie

s TRAMI registry [31, 32]

(n=828)

2.0 (in-hospital)

4.5 (30 days)

20.3

ACCESS EU study [33]

(n=567)

3.4 17.3

Pilot European Sentinel Registry [34, 35] 2.9 (procedural) 15.3

64


30 days (unless

otherwise stated)


follow up

(n=628)

TVT registry [41]*

(n=2952)

5.2 25.8

Abbreviations: CI – confidence intervals; HR – hazard ratio; MC – MitraClip; MM – medical management; ns – not significant; PY – person years.

Shaded areas indicate data not available.

* The TVT registry, MITRA-FR RCT, and COAPT RCT were identified after the cut off dates for the literature search, and have not been critically appraised (see

section 3.4).

65

The EVEREST II RCT did not report significant differences in overall mortality rate between

MitraClip and surgery [24]. One observational study that compared MitraClip with surgery

found that surgery was associated with an increased risk of death at 30 days, but over time

this changed, such that at 1 year the risk of death was higher in the MitraClip cohort. Longer-

term data from the study by De Bonis et al. (2016) [69] found no significant difference in

mortality rates between MitraClip and surgery at 4 years, with a survival rate of 77% ± 5.6%

for surgery compared with 74% ± 5.1% for MitraClip (p = 0.2).

Both the MITRA-FR RCT [42] and COAPT RCT [43] reported similar rates of all-cause

mortality for MitraClip and optimal medical management at 1 year, with no significance

difference observed between arms. However, the COAPT trial also reported data at 2 years

which showed MitraClip was associated with a statistically significant and clinically significant

reduction in mortality (29.1% vs. 46.1%, HR 0.62 (95% CI 0.46 to 0.82, p<0.001). Data from

comparative observational studies were consistent with the COAPT study [29, 30].

Regarding synthesised data, the systematic review by Takagi et al. (2017) compared

mortality rates in patients treated with MitraClip with open surgery [37]. The authors reported

that MitraClip was associated with a significantly lower short-term (≤ 30 days) mortality rate

(1.5% vs. 3.1%, relative risk [RR] 0.54 [95% CI 0.27 to 1.08, p = 0.08]). However, there was

no significant difference in survival rate over the longer term (> 6 months), with 91.5% of

MitraClip patients surviving compared with 92.7% of surgery patients, giving a reported HR

of 1.17 (95% CI 0.77 to 1.77, p = 0.46). The systematic review by Philips et al. (2014)

similarly reported an increased early death rate associated with surgery, with 3% (95% CI 3

to 4%) of deaths occurring at ≤ 30 days in the MitraClip cohorts, compared with 16% (95%

CI 13 to 20%) in the surgery cohort. This study reported that the 1 year mortality rate

associated with MitraClip was 13.0 % (95% CI 9 to 18.3%). Synthesised data from the study

by Vakil et al. (2014) [39] reported a procedural death rate of 0.1%, a 30-day death rate of

4.2%, and a longer-term death rate (mean 310 days follow up) of 15.8%.

Conclusion

The CtE registry reported an in-hospital mortality rate of 5.0% (95% CI 2.4% to 9.0%) and a

1 year mortality rate of 11.6% (95% CI 7.5% to 16.8%). Linked data reported an annualised

mortality rate of 13.6 (95% CI 9.8 to 18.8) per 100 PY. These data appear consistent with

recent results from observational and experimental studies performed in patients with similar

characteristics. The EAC identified three observational studies that included optimal medical

management as a historic or matched control which all indicated MitraClip reduces mortality

in those patients who survive the procedure. Although interpretation of these studies is

limited by bias and confounding, the recent COAPT study appears to confirm these findings

at a later time point (2 years) [43], but, as with the MITRA-FR RCT [42], did not identify a

significant reduction in mortality at 1 year. Studies that have compared MitraClip with

surgery, including the EVEREST II RCT [24], have indicated that MitraClip is a safer

procedure in the short-term, with fewer death-related procedural complications, but the

evidence that either MitraClip or surgery is superior at reducing longer-term mortality is

equivocal. It is likely the reason for this uncertainty is because adequately powered studies

of sufficient methodological rigour have not been performed; that is, current research is

prone to type II error.

The EAC notes that the company that manufactures MitraClip have made the following

statement (EAC italics) [70]:

66

“The major clinical benefits of MitraClip are reduction of MR to ≤2+ resulting in

reduced hospitalisations, improved quality of life, reverse LV remodelling and

symptomatic relief in patients who have no other therapeutic option. No mortality

benefit following MitraClip therapy has been demonstrated”.

The EAC would therefore conclude that, at the time of the literature review, there was an

absence of good evidence that supports the hypothesis that MitraClip reduces patient

mortality rates compared with alternative management strategies. However, the publication

of the COAPT study might alter this conclusion, provided the device is used in the correct,

carefully selected, population.

3.5.4 Question Four

“Does MitraClip reduce the frequency of subsequent hospital admissions?”

This question cannot be adequately answered by the CtE registry alone as there are no

comparator data on patients receiving optimal medical management. Additionally, although

readmission to hospital was a (non-required) field, data reported from this is unlikely to be

reliable, because patients being readmitted would not have their data routinely entered into

the registry. HES data linkage did allow for estimation of subsequent hospital admissions.

Following discharge from hospital, there were 151 admissions in 90 identified patients in the

following year (over the course of 31,294 days follow up). This equated to a rate of 176.1

(95% CI 151.9 to 198.8) admissions per 100 PY, with 62 patients (69%) having at least one

admission. The biggest single reason (modal) for readmission identified by HES was for

cardiac reasons, with an adjusted rate of 70.3 admissions per 100 PY (including 31.6

admission per 100 PY for reasons associated with heart failure). Cardiac causes accounted

for 66% of the total readmission rate. The mean hospital stay following readmission was

5.1 ± 8.1 days (range 0 to 44 days).

Surgery for mitral valve dysfunction was a component of the primary composite outcome of

the EVEREST II RCT [24]. The authors reported 20% of the MitraClip patients required

repeat surgery after 12 months, compared with 2% in the surgery arm (p < 0.001). Thus, one

fifth of patients were effectively crossed over to the other arm of the trial and reported as

treatment failures; this is the principal reason that MitraClip was found to be inferior

compared with surgery in the primary outcome. This effect persisted at 5 years follow up,

with 28% of MitraClip patients requiring surgical intervention, compared with 9% of patients

initially requiring surgery (p = 0.003). However, in patients who were event-free (of the

primary outcome) at 12 months, there was no significant difference (6% in both arms,

p = 0.99). It should be noted however that repeat (open) surgery would not be an option for

most of the population enrolled into the CtE registry.

One prospective comparative observational study reported data on readmission to hospital

due to cardiac disease in patients receiving MitraClip or optimal medical management [29].

The authors reported in the MitraClip cohort that the rate of freedom from readmission for

cardiac disease was 98.2% at 30 days, 76.5% at 1 year, 71.5% at 2 years, and 57.2% at

3 years. This compared with 96.4%, 66.9%, 48.2%, and 36.5% for time points at 30 days

and 1, 2, or 3 years respectively for the medical management cohort. This difference was

significant in favour of MitraClip, with a HR of 1.86 (95% CI 1.05 to 3.29, Log-rank test

p = 0.04).

67

Three single armed studies in high-risk patients were identified in the literature that reported

outcomes relating to readmission. The TRAMI registry [31, 32] reported a high

rehospitalisation rate of 64.3% at 12 months follow up, with 14.1% of the registry cohort

being readmitted for reasons related to cardiac decompensation, 17.8% for other cardiac

reasons, and the remaining patients (25.8%) having non-cardiac reasons for readmission.

The Pilot European Sentinel Registry reported a readmission rate of 22.8% at 12 months

follow up [71]. In contrast, the EVEREST II HR study reported that no patients required

readmission for “reoperation for failed MV surgical repair or replacement” or “urgent or

emergency cardiovascular surgery for adverse event” at 30 days or 12 months.

The TVT registry, which was published after the EAC literature review was completed,

reported data on readmission to hospital for heart failure in a cohort of 2,952 patients with

mainly DMR. The overall incidence of readmission for heart failure at 1 year was 20.5%. The

rate of rehospitalisation with heart failure was significantly associated with MR of functional

aetiology (p = 0.008).

Recently, two RCTs have reported the rate of rehospitalisation for heart failure as their

primary outcome (COAPT trial [43]) or as a component of their primary outcome (MITRA-FR

trial [42]). The COAPT study reported that over 2 years there were 160 admissions in the

MitraClip arm compared with 283 in the medical management arm. This equated to an

annualised rate of 35.8% compared with 67.9% and was statistically significantly lower (HR

0.53, 95% CI 0.40 to 0.70, p < 0.001). The MITRA-FR trial reported the annualised

unplanned hospitalisation rate for heart failure was 48.7% in the MitraClip group compared

with 47.4% in the medical management group; there was no significant difference (HR 1.13,

95% CI 0.81 to 1.56).

Hospital readmission data are summarised in Table 23.

68

Table 23. Proportion of patients requiring re-intervention or re-admission following

MitraClip procedure.

Study Readmission (or proxy) rate (%, unless otherwise stated)

Definition/comment

Early 12 months

CtE registry 10.4 (between discharge and 6 weeks)

9.3 (in addition to 6 week data)

Data considered unlikely to be reliable.

Linked dataset (CtE registry/HES/ONS)

Total: 176.1 per 100 PY For heart failure: 36.2 per 100 PY

Annualised readmission data from HES.

EVEREST II RCT (2011) [24]

MC 2 (≤ 30 days) 20 “Early” data is “Reoperation for failed surgical repair or replacement” or “Urgent or emergency cardiovascular surgery for adverse event”. 12 month data is “Surgery for mitral-valve dysfunction: first mitral-valve surgery in the MC group and the rate of reoperation for mitral-valve dysfunction in the surgery group”.

Surgery 4 (≤ 30 days) 2

Giannini (2016) [29]

MC 1.8 (≤ 30 days) 23.5 Readmission to hospital for cardiac indication. MM 3.6 (≤ 30 days) 33.1

MITRA-FR RCT* (2018) [42]

MC 48.7 per 100 PY Unplanned readmission to hospital for heart failure. MM 47.4 per 100PY

COAPT RCT* (2018) [43]

MC 35.8 per 100 PY All hospitalisations for heart failure. Significantly fewer events in MC group (HR 0.53).

MM 67.9 per 100 PY

TRAMI registry [31, 32] 3.5 (peri-procedural)

64.3 “Early” data is procedural indicating “surgery for failed percutaneous intervention”.

Pilot European Sentinel Registry [71]

0.7 (peri-procedural)

22.8 “Early” data reported as “Vascular complication requiring intervention”.

EVEREST II HR study [28]

16% Rehospitalisation for heart failure.

TVT registry* [41] 6.2 20.5 “Early” data reported as repeat transcatheter mitral valve repair.

* Studies published subsequent to the literature search. These studies have not been critically appraised. Abbreviations: MC – MitraClip; MM – medical management; PY – person years.

Conclusion

There is no direct evidence from the CtE registry concerning the relative rate of readmission

in patients following the MitraClip procedure. However, linked data reported that there was

an overall readmission rate of 176 per 100 PY, and 36 per 100 PY for heart failure

specifically. Evidence from the EVEREST II RCT [24] suggests that about one fifth of

patients require open surgery following the MitraClip procedure, with the majority happening

in the first year. However, this study was conducted in a population well enough to tolerate

surgery. A recent RCT reported that the rate of unplanned readmission for heart failure was

close to 50% in both MitraClip and medical management arms, with no significant difference

between them [42]. This contrasts with another RCT, which reported MitraClip significantly

69

reduced the rate of readmission for heart failure compared with medical management alone

[43]. Evidence from observational studies indicated that the rate of readmission for MitraClip

patients is high in the first year, ranging from approximately 20 to 60%.

Readmission rates associated with MitraClip are discussed further in Question Ten.

3.5.5 Question Five

“Are the early benefits in reduction in mitral regurgitation maintained in the medium-

term? Is there a need for repeat treatment over time (either by a repeat percutaneous

procedure or surgery)?”

The efficacy of MitraClip in reducing MR is discussed in Question One. The benefits of this

reduction, in terms of improvements to QoL and reducing symptomatic dyspnoea, are

discussed in Question Two. Reductions in MR grade and related health benefits were

sustained for the duration of the CtE registry (up to 2 years).

The CtE registry recorded that four people required an additional intervention (three having a

surgical intervention and one having a repeat percutaneous intervention to retrieve an

embolised device); out of 168 patients followed for a total of 96.8 person years. This was

probably an underestimate because of incomplete reporting (see Question Four). Additional

interventions coded in HES APC dataset in the 1 year following MitraClip implantation

indicated that 5 patients required additional MV intervention (1 requiring mitral valve repair, 4

requiring mitral valve replacement).

The EVEREST II RCT follow up at 3 years [26] and 5 years [25] reported that reduction in

MR grade was sustained in the longer-term. At 5 years, 87.7% of patients had ≤MR 2+; this

compared with 98.2% of patients who had surgery (p = 0.02). However patients enrolled into

the EVEREST II RCT were not representative of those in the CtE registry. The COAPT RCT

recently published data from 2 years of follow up [43]. This study showed that improvements

in overall mortality rate were non-significant over 1 year but became significant after 2 years,

with the Kaplan Meier appearing to diverge with time. The study was restricted to patients

with FMR, which was representative of 60% of the registry patients.

An observational study, which compared treatment using MitraClip with surgery reported

contradictory evidence to the EVEREST II RCT. The study by De Bonis et al. (2016) [69]

was a retrospective review of MitraClip patients who had had an optimal procedural outcome

(i.e. successful MitraClip procedure with MR reduction and without significant complications,

n = 85), which was compared to a cohort (n = 58) who had undergone surgical repair. Both

groups were similar at baseline. The authors reported that, although freedom from cardiac

death was not statistically significant at 4 years follow up, the initial results of MitraClip did

not remain stable, with evidence of deteriorating MR over time. Freedom from MR grade ≥3+

at 4 years was 75% ± 7.6% (SD) in the MitraClip group and 94% ± 3.3% in the surgical

cohort (p = 0.04). Freedom from MR grade ≥2+ at 4 years was 37% ± 7.2% compared with

82% ± 5.2% for MitraClip and surgery respectively (p = 0.0001). The authors concluded that

the results did not confirm previous observations from the Everest II RCT that reported

sustained results for MitraClip at 4 years.

Surgery has consistently been reported as resulting in superior MR outcomes when

compared with MitraClip in a number of observational studies [52, 69, 72-74]. Data from a

systematic review by Takagi et al. (2017) [37] reported rates of freedom from ≥MR3+ was

70

75% for MitraClip compared with 96% for surgery after 3 years. A pooled analysis indicated

significantly higher incidence of recurrent MR in patients receiving MitraClip compared with

surgical repair, with a pooled HR/OR of 4.80 (95% CI, 2.58 to 8.93; p = 0.00001). This

suggests that for people who can tolerate it, surgery is the better option to achieve a

sustained reduction in MR. The EAC did not identify any studies that reported MR outcomes

that compared MitraClip with conservative medical management.

Conclusion

The benefits of MitraClip were sustained for the follow up duration of the CtE registry. A

recent RCT reported that the benefits of MitraClip in terms of reduction in mortality and

rehospitalisation accrue over a period of 2 years. There is conflicting evidence from the

literature about the longer-term reduction in MR associated with MitraClip compared with

surgery, but most observational evidence suggests that surgery – when it is deemed

appropriate (feasible and with acceptable risk) - is a better long-term option.

3.5.6 Question Six

“What proportion of patients referred to a specialist MitraClip service as defined in the

CtE documents were assessed by the MDT as suitable for the intervention? What

proportion of the patients considered suitable for the procedures received it and what

proportion of them benefitted (refer also to Question 5)?”

The CtE registry cannot directly answer this question. Centres did not use the registry to

capture all cases presented to the MDT meeting; that is, patients not selected for the

procedure were not subsequently enrolled into the registry; therefore this data cannot be

used to reliably answer the question. The patient flow of patients following the MDT meeting

is reported in Appendix 1. Of the 272 patients indicated for treatment with MitraClip, a total of

69 (25%) were excluded from the registry on grounds of non-eligibility. A further 4 patients

were excluded because the procedure was not conducted within the timeframes, or because

procedure dates were contradictory to the discharge dates.

The registry reported that 15.6% of cases were urgent or emergency procedures, and thus

would have bypassed elective MDT consultation. Patients who underwent the procedure

urgently or as an emergency had a significantly different length of stay compared with

elective patients, as illustrated in Figure 12, which will have an important impact on NHS

resources. Differences in outcomes between these groups are reported in Question Eight.

The eligibility criteria for percutaneous leaflet repair using MitraClip is clearly stated by NHS

England, including anatomic measurements [5]. However, it is unknown what proportion of

potentially eligible patients fulfil these criteria. One cohort study consecutively recruited 160

patients with severe FMR. Patients were assessed for eligibility for MitraClip based on

anatomical suitability, with those who were judged unsuitable, or who withheld consent,

receiving optimal medical management. Ninety (56%) of the patients received optimal

medical management. This suggests about half of patients with severe FMR unable to

tolerate surgery may be suitable for the MitraClip procedure.

Conclusion

The registry cannot be used to answer this question. Extrapolated data from a prospective

cohort study suggests around half of elective patients may be suitable candidates for

71

MitraClip treatment. However, there remains considerable uncertainty regarding the

prevalence of the indicated population.

Figure 12. Average length of stay by urgency.

3.5.7 Question Seven

“What are the short and medium term risk of complications from MitraClip use?

Answers to Questions 7 and 8 will be considered together by the NICE Interventional

Procedures Advisory Committee when NICE updates the IP guidance on this

procedure.” [Note: IPG309 is currently under review].

To answer this question, the EAC has defined “short term” as peri-procedural or in-hospital

(or ≤ 30 days for some studies in the literature). “Medium term” refers to follow up out of

hospital (up to 2 years). Complications were classified as being major, which would have a

significant impact on the patient’s wellbeing, or minor, which generally were transient and

non-serious in nature.

The CtE registry reported that there was successful deployment of at least one clip in 94% of

patients. The procedural success rate, defined as successful device deployment with no

major complications, was 86%. This was somewhat lower (worse) than most values reported

in the literature. However, a direct comparison with the literature is not possible due to the

heterogeneous definitions of procedural success used in the studies. A comparison of

procedural success rates is reported in Table 24.

72

Table 24. Procedural success in CtE registry compared with values reported in

published studies.

Study Procedural success rate (%) Definition/comment


94 Successful deployment of at least one clip

85.9 (95% CI 80.3 to 90.4) Device implanted and no major complications

EVEREST II study [24] (n=184)

100 Implied from 0% re-operation/repair rate following procedure

EVEREST II HR study [28] (n=78)

96.0 Successful clip deployment

79.5 Successful deployment with MR reduction ≤1 class

Giannini (2016) [29] (n=60)

98 “Acute procedural success” (not further defined)


97.0 Successful implant and MR ≤2+ achieved


99.6 Successful implantation

Pilot European Sentinel study (2014) [34] (n=628)

95.4 “Acute procedural success” (not further defined, but implied that procedure must reduce MR by ≤1 class)

Rahab (2017) [36] (n=1151)

95 Successful deployment of the device with absence of procedural mortality

91 Placement of the device and MR ≥1 reduction to an absolute level of moderate MR

TVT registry (2017) [41] (n=2952)

91.8 Successful deployment with reduction to moderate or less MR, in the absence of cardiac surgery or in-hospital mortality

MITRA-FR RCT (2018) (n=304) [42]

95.8% Inverse of “device-implantation failure”. Procedural success (successful implant and discharged with MR grade ≤1+) was 76.4%.

COAPT RCT (2018) (n=614) [43]

95.0% 82.3% in whom device was implanted were discharged with MR grade1+.

Abbreviations: MR – mitral valve regurgitation (grade)

Sixteen major in-hospital complications were reported in the CtE registry, with ten of these

being deaths. There was one recorded event of device embolisation requiring percutaneous

retrieval, three additional surgeries, and two myocardial infarctions. There were 25 additional

events occurring after discharge from hospital, including twenty additional deaths. Major and

minor complications are listed in Appendix 4. The in-hospital major complication rate was

8.2% (95% CI 4.7 to 12.9%) and the major complication rate after discharge was 14.7%

(95% CI 9.7 to 20.9%). The respective overall in-hospital rates (major and minor combined)

were 15.2% (95% CI 10.5 to 20.9), compared with 25.9% (95%CI 19.5 to 33.1%) following

discharge.

Major in-hospital complications are compared with values published in the literature in Table

25. Published values ranged from 3.1% to 27.8%. This wide range was probably due to

differing definitions of major complications (including rules for multiple events and event

73

hierarchy), and populations studied. For instance, the TRAMI registry [31] only reported

Major Adverse Cardiac and Cerebrovascular events (MACCEs), and reported a particularly

low in-hospital mortality rate (despite 9% of procedures being described as emergency).

Major complication events occurring after discharge are reported in Table 26. Again,

comparison is limited by differing definitions but also by different follow up durations. The

major complication rate of 14.7% in the CtE registry is relatively low compared to values

published in these studies. The main contributory component was death, discussed in

Question Three.

Table 25. Comparison of major in-hospital adverse events reported by the CtE registry

and published literature.

Study Proportion with major in-hospital complications/adverse event (%)

Definition / comment


8.2 (95% CI 4.7 to 12.9) Includes death, neurological event, additional surgery, device embolisation, MI, endocarditis, pericardial effusion/tamponade (requiring intervention), major vascular injury (requiring intervention), MV complication, oesophageal rupture, major bleed, AKI (stage 2/3), cardiogenic shock


15 30 days post-procedure Similar events included compared with CtE registry.


26.9 30 days post-procedure Similar events included compared with CtE registry. 17.9% had requirement for ≥2 units of blood

Giannini (2016) [29] (n=60)

19.9 Calculated by EAC* including incidences of sepsis, new onset AF, AKI, bleeding requiring transfusion, and partial clip detachment (5%). No in-hospital deaths recorded.


3.1 In-hospital MACCE rate.


17.1 Calculated by EAC (30 day event rate)*. Includes MACCE, renal failure, respiratory failure, need for resuscitation, cardiac tamponade, and bleeding complications.


27.8 Calculated by EAC*. Includes death, tamponade, stroke, severe bleeding, need for transfusion, vascular complication, new onset AF.

MITRA-FR RCT (2018) [42] (n=304)

14.6% Total peri-procedural complications.

Abbreviations: AF - atrial fibrillation; AKI – acute kidney injury; MACCE – major adverse cardiac and cerebrovascular events; MI – myocardial infarction; MV – mitral valve; NR – not reported. * Crude count of sum of events (which may not be independent).

74

Table 26. Comparison of major adverse events after discharge reported by the CtE

registry and published literature.

Study Proportion with major post-discharge complications/adverse events (%)

Definition / comment


14.7 (95% CI 9.7 to 20.9) Most events (11.8%) were deaths.


45 Inverse of composite primary endpoint at 12 months. Includes death (6%), surgery for MV dysfunction (20%), and recurrence of grade ≥3+ MR (21%)*.


42.3 12 months post-procedure. Similar events included compared with CtE registry.*


20.3 MACCE at 12 months.


36.2 Calculated by EAC (12 month event rate)**. Includes MACCE, renal failure, respiratory failure, need for resuscitation, cardiac tamponade, and bleeding complications.


31.0 Survival from death or readmission.

COAPT RCT (2018) [43] (n=302)

4.4 Freedom from device-related complications at 12 months.

Abbreviations: AF - atrial fibrillation; AKI – acute kidney injury; MACCE – major adverse cardiac and cerebrovascular events; MI – myocardial infarction; MV – mitral valve; NR – not reported. * Total events post-procedure (that is, includes in-hospital and post-discharge events). **Crude count of sum of events (which may not be independent).

Conclusion

In the CtE registry, MitraClip was successfully deployed in 94% of the patients it was

attempted in, with a procedural success rate of 86%. Around 8% of people admitted to

hospital for the MitraClip procedure experienced a major complication, with 5% dying.

Following discharge, there was a major complication rate of around 15%, with death being

the most common contributor (12%). Superficially, these values appear to be only broadly

consistent with those in the published literature. However, comparisons are limited because

studies have used different terminology and methodology, and were conducted in different

populations.

3.5.8 Question Eight

“Are clinical outcomes with MitraClip associated with particular patient characteristics

(clinical or demographic)?”

Due to the relatively low sample size of the CtE registry, and subsequently, the relatively low

number of outcome observations, it was necessary to limit the number of covariates to two

within the multivariate analysis modelling (Cox proportional hazards modelling and binary

logistic regression modelling). A general rule of thumb in multivariate analyses is to allow a

minimum number of 10 outcome observations per covariate included in the model. In all

75

outcomes apart from one (n = 22), observations did not exceed 20, suggesting a maximum

number of two covariates should be used in all cases. The two covariates were selected as

being the most clinically relevant following consultation with the Clinical Leads (see Section

2.4.6).

The two covariates selected for multivariate analysis were aetiology of MR (functional or

degenerate) and urgency of procedure (elective or urgent/emergency). Both these

characteristics were deemed as fundamentally important and potentially could demarcate

two different populations of patients. In turn, this could inform which patients benefit most

from the MitraClip procedure and therefore who the procedure should be indicated for.

MR aetiology

The potential impact of MR aetiology on mortality rate, MR grade, NYHA class, and adverse

events (major and minor complications) was assessed using multivariate analysis. For this

analysis, patients were dichotomised as having functional (or ischaemic) MR, or

degenerative MR. Patients with DMR were significantly older (mean age 80.2 years) than

those with FMR (mean age 73.7 years, p < 0.0001); this is illustrated in Figure 13.

Using data from the registry, there was no significant difference in mortality observed

between groups when time to event analysis was used (Cox proportional hazards model;

p = 0.404). Additionally, there was no significant difference observed between patients with

functional or degenerative MR when binary logistic regression was performed on the

outcomes of MR grade (6 and 12 months), NYHA class (6 and 12 months), and adverse

events (major or minor complications). Using linked data, the mortality rate was 13.2% at

1 year and 24% at 2 years in patients with FMR. In patients with DMR, these figures were

11% and 20% respectively (see Table 12). There was no significant difference in death

between aetiologies.

The EAC performed subgroup analysis on mitral valve dysfunction aetiology for the use of

hospital resources and costs, described fully in Question Ten and Appendix 15. For the DMR

subgroup, there was an increase in total hospitalisation, from 8125 days per 1000 PY in the

year before the procedure, to 8857 days per 1000 PY in the year after the procedure. This

gave a incidence rate ratio of 0.85 (95% CI 0.49 to 1.47), which was not significant. There

was also a slight increase in the mean hospitalisation days per patient (8.1 pre-procedure vs.

8.4 post-procedure). However the mean hospitalisation cost per patient was still reduced

post-procedure (£6322 pre-procedure vs. £4608 post-procedure). In contrast, for the FMR

subgroup, there was a decrease in total hospitalisation days from 9459 per 1000 PYpre-

procedure to 8600 days per 100 PY post-procedure), and a mean hospitalisation days per

patient reduction of 9.5 days pre-procedure compared with 8.2 post-procedure. The mean

hospitalisation cost per patient was also reduced from £9858 pre-procedure to £5967 post-

procedure. Changes in mean hospitalisation costs were similar for FMR and DMR. These

results may indicate there is a greater potential to reduce hospital resource use and costs for

DMR (although absolute hospitalisation post-procedure were similar). However, an important

limitation of the de novo economic model was that there were insufficient data to effectively

run statistical analyses in multiple subgroups (e.g. admissions for heart failure in FMR

patients). Instead of separate subgroup analysis to determine the effect of MR aetiology

(FMR/DMR) and the procedural urgency of the index MitraClip procedure (which reduce

sample size), a more robust method would be to incorporate variables within the mixed

effects model to determine if they significantly contributed to outcome: this work is on-going.

76

MitraClip is licensed for use for the management of both DMR and FMR in Europe. Post-

marketing data suggests that around two thirds (65%) of MitraClip procedures have been

performed on patients with FMR, compared with 22% with solely DMR and 13% with mixed

aetiologies [75]. These proportions are reflected in the data reported by the CtE registry and

by the composition of participants in most the studies identified by the EAC (Table 16). In the

US, where use of MitraClip in patients with FMR is technically off label, most, but not all,

patients treated with MitraClip have DMR. This may change with the publication of the

COAPT RCT [43].

Figure 13. Distribution of age in patients with functional or degenerative MR.

Currently, the only direct comparative evidence for the benefit of MitraClip in these patient

groups comes from the EVEREST II RCT. Subgroup analysis of the participants in this study

reported that surgery was a better option than MitraClip in terms of the primary outcome in

patients with DMR. For patients with a diagnosis of FMR, the evidence of benefit was

equivocal [24].

The US TVT registry was the largest (single armed) observational study published to date,

enrolling 2,952 patients [41]. Most (85.9%) of the patients had DMR, with 8.6% having solely

FMR and 8.9% have mixed functional and degenerative aetiology (classed as functional for

dichotomous analysis). The authors of this registry reported that the cumulative incidences

of mortality (24.7%), re-hospitalisation for heart failure (20.5%), and the combined endpoint

for both of these 2 outcomes (35.7%) were significantly lower for patients with DMR, in

comparison to those observed with FMR (31.2%, 32.6%, and 49.0%, respectively). Thus the

prognosis in patients with FMR treated with MitraClip was worse than those with DMR.

The European Pilot Sentinel Study reported that patients with FMR were more likely to be re-

hospitalised compared with patients with DMR [34]. The ACCESS-EU study reported no

77

significant differences in clinical outcomes between patients with functional and degenerative

MR after 1 year follow up [33].

A recent systematic review and meta-analysis identified and pooled nine studies (one RCT

and eight prospective observational studies) with the aim of identifying prognostic

differences between functional and degenerative MR patients who had been treated with

MitraClip [40]. The authors reported that FMR was associated with a significantly lower rate

of re-intervention after 1 year (4%) compared with DMR (10%), with a RR of 0.60 (95% CI:

0.38 to 0.97, p = 0.04). This may have been because patients with FMR were less suitable

clinically to receive re-intervention (which was poorly defined in the study).

Urgency of procedure

Most of the patients (84.4%) enrolled into the CtE registry were elective patients, fulfilling the

criteria set by NHS England (see Section 2.3). However, 13.6% of patients were treated with

MitraClip urgently, and 2.0% were treated as emergencies. The EAC considered that these

were likely to represent a distinct cohort of patients compared with the elective group (in

terms of overall health and reporting). Because of this, the EAC performed subgroup

analysis in these cohorts on key outcomes.

Using registry data, patients receiving MitraClip as an urgent or emergency case had a

greater risk of death, using time to event analysis, than those admitted as elective cases.

This is illustrated in the Kaplan-Meier analysis in Figure 14. For elective patients, there was

an annualised event rate of 22.1 (95% CI 13.9 to 33.5) per 100 PY, compared with 68.2

(95% CI 29.5 to 134.3) per 100 PY for urgent/emergency cases. This difference was

significant (p = 0.0105). As might be expected, this increase is driven by an increased rate of

in-hospital mortality; when only post-discharge mortality was considered, there was no

significant effect.

78

Figure 14. Kaplan-Meier analysis comparing patients treated electively or urgently (or

as an emergency).

Using linked data, patients electively treated had a mortality rate of 10.4% (95% CI 5.3 to

15.3%) at 1 year compared with 29.2% (95% CI 6.5 to 46.4%) in patients receiving MitraClip

urgently or as an emergency. At 2 years, the corresponding values were 19.0% (95% CI

11.6 to 25.7%) for elective patients and 47.7% (95% CI 18.0 to 66.6%). See Table 12 and

Appendix 12. No important differences were observed in the overall use of hospital

resources when elective or urgent procedures were compared (see Appendix 14), with the

incidence rate ratio for admissions 0.53 (95% CI 0.44 to 0.65) for elective procedures

compared with 0.64 (95% CI 0.49 to 0.83) for urgent or emergency procedures; however,

this may reflect the sample size was underpowered to detect differences.

There was no statistically significant effect of the urgency of the procedure on MR and NYHA

grades at 6 months and 12 months. Urgent and emergency procedures were associated with

increased in-hospital adverse events compared with elective procedures (p = 0.0242),

although this was not significant when Bonferroni correction for multiple outcome analysis

was applied. There was no effect observed of procedural urgency on minor adverse events.

The incidence rate ratio for readmission for elective procedures was 0.54, compared with

0.79 for urgent or emergency procedures, showing a trend to reduced readmissions

favouring MitraClip when used planned as an elective procedure.

Concerning the literature, only the TRAMI registry reported that a proportion of the patients

were enrolled as emergency cases (9%, as opposed to 91% recruited electively) [31].

However, no subgroup analysis was performed. The remaining studies either exclusively

enrolled elective patients or did not specify what proportion were managed urgently or as

emergencies (see Table 16).

79

Other factors

Univariate analysis and multivariate analysis was performed on device implanted (one or

more clips successfully deployed) in Supplementary Material - Table 5 and on in-hospital

major and minor complications in Supplementary Material - Table 6 and Table 7. After

application of Bonferroni correction for multiple outcomes, no significant associations were

identified.

Conclusion

Subgroup analysis of data reported from the CtE registry did not report any differences in

clinical outcomes in patients with functional or degenerative MR. There was a trend for

increased hospital resource use in patients with DMR (relative to the year prior to the

MitraClip procedure), but absolute rates were similar to post-procedure for both aetiologies.

Data from published studies, including the recently published large TVT registry [41],

indicate that patients with FMR have an increased mortality rate and are more likely to be re-

admitted to hospital than patients with DMR. Patients in the CtE registry who were admitted

urgently or as emergencies had an increased risk of in-hospital mortality. This would be

expected as these patients are at greater immediate risk by definition. However, the EAC,

did not identify any published literature that performed comparable analysis.

3.5.9 Question Nine

“What are the full procedural costs of MitraClip to the NHS?”

Table 7 reports that the forecast cost for a MitraClip procedure ranges from about £29,000 to

about £34,500, with a central estimate of around £32,910. The device cost included in each

scenario is £****** per procedure and assumes the manufacturers charge per procedure not

per device. This is a key assumption given that 1.9 devices were opened per procedure (this

has been clarified through communication with the device manufacturer). The device

accounts for **% to **% of the total cost, depending on the scenario. The cost of the Abbott

device pre-value added tax was provided by the manufacturer as ‘commercial in confidence’

(Personal communication, Dr Steven Fearn, Health Economics & Reimbursement Manager,

Abbott; 4 October 2017). An additional 15% overhead was added to the post VAT costs to

meet the overheads of the NHS Trusts associated with procurement, stores and onsite

delivery.

Under the central cost scenario the pre-operative pathway accounted for 2%, the procedure

85% and subsequent management 13% of total costs respectively. A more detailed analysis

of the cost components is provided at Appendix 9.

3.5.10 Question Ten

“What are the potential cost savings for the NHS arising from patients receiving

MitraClip?”

De novo analysis

Background to de novo economic analysis

To answer this question, the EAC performed a de novo economic analysis, based on the

methodology of Vemulapalli et al. (2018) [18]. This study identified that the main potential

80

resource benefit from the MitraClip procedure was a reduction in hospitalisation episodes,

reflecting the patients’ improved clinical status, as evidenced by improved NYHA class (see

response to Question Two). Additionally, medication use per patient may also decline, but

the CtE registry did not adequately capture change in use of medicines. The de novo

analysis used a “before and after” approach, using linked data to identify hospital admissions

1 year prior to the procedure being undertaken, and 1 year after the procedure was

undertaken.

The methods of the analysis are described in Section 2.5.3. Results of the analysis (hospital

resource use) are reported in Section 3.2.6. Full results are presented in tabular form in

Appendix 14 and Appendix 15. Limitations of this analysis are discussed in Section 4.4.1.

The main results and “take home messages” are discussed below.

Results of de novo economic analysis

From the linked dataset, a total of 150 patients were alive at discharge following successful

MitraClip implantation (which were conducted between 28th January 2015 and 23rd August

2017). Of these 56 had DMR and 90 FMR (4 procedures with MR aetiology unreported). A

total of 137 patients were alive at 1 year. A total of 146 patients were admitted at least once

during the year prior to their index procedure, accumulating a total of 470 pre-procedural

admissions in that time. A total of 100 patients were admitted at least once during the year

after their index procedure, accumulating a total of 251 post-procedural admissions in that

time. The proportion of admissions which were single episode spells were 81.9% and 73.3%

within each of these periods respectively. Of the 137 patients who survived to 1 year, 134

experienced at least one admission in the year prior to their index procedure (total of 418

pre-procedural admissions), and 88 experienced at least one admission in year following

their procedure (total of 221 post-procedural admissions).

The total all-cause hospital admissions per 1000 PY were decreased in the year following

MitraClip from 3133.3 pre-procedure to 1763.6 post-procedure. This gave an incidence rate

ratio of 0.57 (95% CI 0.49 to 0.67), which indicated a significant decrease. Similarly the total

hospitalisation days decreased from 10900.0 pre-procedure to 9415.2 post- procedure

(incidence rate ratio 0.73 [95% CI 0.54 to 0.98]). Regarding costs, the aggregated baseline

hospitalisation cost (pre-procedure) was £1,267,321, with median £5196 (IQR £2574 to

£10,292), mean £8448 (SD £11,033). This compared with an aggregated post-procedural

hospitalisation cost of £889,721, with median £1402, (IQR £0 to £7451), and mean £5931

(SD £10,688). The incidence rate ratio, calculated using a mixed effects model (see Section

2.5.4), was 0.62 (95% CI 0.50 to 0.79). Thus, using linked data, MitraClip was associated

with decreases in hospital admission rates and total hospitalisation days, and reduced

hospitalisation costs of around 30% (aggregate). These differences were statistically

significant.

Sensitivity analysis was also performed where patients who had died post-procedure were

excluded from analysis. This was done to address the “positive” economic consequences of

healthcare resource reduction due to death, bearing in mind the high mortality rate

associated with the procedure [18]. Removing patients who died did not change the direction

of results.

Hospital admission was also classified into specific sub-categories: these were non-cardiac,

all-cause cardiac, and heart failure reasons for admission. There was a significant increase

in the admission rate in the non-cardiac group, from 793.3 per 1000 PY pre-procedure to

1,061.0 per 1000 PY, giving an incidence ratio rate of 1.40 (95% CI 1.10 to 1.78). The main

81

cause for non-cardiac post-admission was iron deficiency anaemia (10 admissions), followed

by lobar pneumonia (6), cataract (5) and acute renal failure (5). In contrast, there was a large

decrease in the admission rate for cardiac causes, from 2340.0 to 702.6 per 1000 PY, giving

an incidence rate ratio of 0.30 (95% CI 0.24 to 0.38). Even more pronounced was the

decrease in admission due to heart failure and related causes, from 1300.0 pre-procedure,

to 316.2 per 1000 PY. This resulted in an incidence rate ratio of 0.24 (95% CI 0.18 to 0.34).

The main causes of cardiac post-admission was congestive heart failure (28 admissions),

followed by chest pain (8), dilated cardiomyopathy (7), left ventricular failure (7).

Differences in hospital resource use and costs between subgroups of patients (by urgency or

aetiology) are discussed in Question Eight.

Economic evidence from the literature Hospital resource use Four economic studies were identified in the literature review that informed, or partly

informed, Question Ten [18, 76-78] (see Section 4.2). These studies were generally well

conducted but did not have external validity and hence the findings cannot be directly

generalised to NHS England. However, they do provide additional evidence on the impact of

the MitraClip procedure on hospitalisation rates and costs: these are the key economic

variables. These findings, together with relevant clinical evidence, can help inform on the

likelihood of MitraClip being cost-saving.

The study by Vemulapalli et al. (2017) [18], the methodology of which informed the EAC de

novo economic analysis, was considered to be well conducted. It compared admission rates,

inpatient days and inpatient costs in the year before and year after the MitraClip procedure.

The study included 403 patients in the USA, recruited from the EVEREST II High-Risk

Registry and REALISM Continued-Access Study [79]. In the year after the MitraClip

procedure, all-cause admissions decreased from 1,854 to 1,435 per 1,000 PY (hazard rate

0.82), with a statistically significant reduction in heart failure admissions (749 versus 332 per

1,000 PY, hazard rate 0.46), but bleeding increased (199 versus 298 per 1,000 PY, hazard

rate 1.72). The single most common cause for a hospital day remained heart failure.

However, the number of days patients spent in hospital for all causes increased from 9,385

per 1,000 PY in the year before the procedure to 10,312 per 1,000 PY after it (hazard rate

1.36). A survivor analysis reported annual all-cause hospitalisation days reduced from 2,673

per 1,000 PY to 1,881 per 1,000 PY, rate 60%, after the procedure.

Two of the other studies only included hospitalisation for heart failure [76, 77], omitting

changes in admissions for other causes. As Vemulapalli et al. (2017) [18] demonstrated, as

well as the de novo analysis, heart failure admissions alone are not a good measure of the

impact of the procedure on total hospital resources.

The fourth study by Armeni et al. (2016) [78] was set in Italy. The authors extracted

admission and length of stay data from hospital records for 232 patients treated with

MitraClip and a control group of 151 patients treated with medical therapy only. The study

reported a materially lower rate of annual hospitalisations for the MitraClip cohort (0.16 per

patient versus 0.70 per patient in the medical therapy arm, ratio 0.23). However, these data

are confounded because the cohorts were not matched at baseline. For example, the

82

MitraClip cohort had fewer hospital admissions in the previous year than patients in the

medical therapy group (1.5 versus 2.0).

The clinical evidence presented in response to question 4 from studies in high-risk patients

reported a range of readmission rates from 64% at 12 months from the TRAMI registry [31,

32] to 23% from the Pilot European Sentinel Registry [71] and 21% from the TVT registry

[41]. However, the last two studies did not report all-cause admissions but only admissions

for specified vascular events. These will understate all-cause admission rates in patients

after a MitraClip procedure.

Inpatient costs

The study by Vemulapalli et al. (2017) [18] reported mean inpatient costs for all patients of

$19,006 in the year before MitraClip and $16,989 in the year after (cost ratio 0.89). A sub-

group analysis showed the change in costs did not vary between those with improvement in

MR (cost ratio = 0.65) or with no improvement (cost ratio = 0.64). A survivor analysis

reported a statistically significant reduction in mean hospital costs in the year after MitraClip

(from $18,131 to $11,679, cost ratio 0.64). It should be noted that none of these costs

include the cost of the MitraClip procedure.

The only other economic study reporting inpatient costs was Armeni et al. (2016) [78]. This

study assumed annual hospitalisation costs of 495€ [£396]1 for the MitraClip arm and 4,338€

[£3,470] for those managed on medical therapy (cost ratio 0.11).

Total costs per patient

Vemulapalli et al. (2017) [18] only reported inpatient costs, excluding cost of the MitraClip

procedure.

Mealing et al. (2013) [76] only reported incremental costs of £26,989 per patient (2011

prices) for the MitraClip cohort compared to patients receiving medical therapy over five

years. The MitraClip device accounted for £20,500 of this difference (76%) and short-term

hospitalisation for a further £3,800 (14%). The cost of medicines was also higher in the

MitraClip cohort (+£1,400) (5%) because these patients were modelled to have a longer

mean survival than the medical management cohort.

Cameron et al. (2014) [77] reported that over a lifetime, the mean cost per patient receiving

a MitraClip was 62,510 Canadian dollars (C$) [£34,730)2 compared to C$21,893 (£12,160)

for medical therapy. All of the incremental cost of C$40,617 (£22,565) was accounted for by

device related costs. These were C$43,439 (£24,130). Savings in hospitalisation costs for

heart failure and surgery following the MitraClip procedure offset slightly higher disease

management costs post procedure compared with medical therapy only.

Armeni et al. (2016) [78] only reported the incremental lifetime cost of 23,342€ (£18,670) per

patient for the MitraClip cohort compared to those managed on medical therapy. Almost all

of this cost is accounted for by the procedure (23,069€ [£18,455]).

Conclusions to Question Ten

1 Exchange rate of £1 to 1.25€ adopted for exchange rate in 2012. Source: http://www.x-rates.com/average/?from=GBP&to=EUR&amount=1&year=2012 2 Exchange rate of £1 to 1.80 Canadian $ for exchange rate in 2014. Source: http://www.x-rates.com/average/?from=GBP&to=CAD&amount=1&year=2014

http://www.x-rates.com/average/?from=GBP&to=EUR&amount=1&year=2012

http://www.x-rates.com/average/?from=GBP&to=EUR&amount=1&year=2012

http://www.x-rates.com/average/?from=GBP&to=CAD&amount=1&year=2014

http://www.x-rates.com/average/?from=GBP&to=CAD&amount=1&year=2014

83

The de novo economic analysis performed by the EAC, based on the methodology of the

study by Vemulapalli et al. (2017) [18], found that, overall, the MitraClip procedure was

associated with a decrease in post-procedure hospital resource use and NHS costs. This

was driven by a decrease in the rate of cardiac admissions, especially those associated with

heart failure. Although admissions for non-cardiac reasons increased slightly, possibly

because patients were well enough to receive treatment for other conditions post-procedure

(e.g. cataract operations), this did not offset the overall rate of hospital readmission.

However, this analysis comes with some important caveats, the most important of which

being that the analysis did not include the cost of the MitraClip device or procedure itself.

This is important as these costs are large (ranging from £29,000 to almost £34,500, with a

central estimate of around £32,910, see Question Nine), and very likely to exceed feasible

cost reductions associated with reduced future hospital admissions.

The evidence from four economic studies identified in the literature is consistent with:

• A material reduction in admissions for heart failure in the years immediately after a

MitraClip procedure;

• However, this reduction may be offset by increases in hospital admissions for other

causes including bleeds [18], such that the use of hospital bed-days does not

decline;

• A reduction in annual inpatient costs of around 10% in the first year following the

procedure [18];

• The reduction in inpatient costs should be higher in the second year as the cost

reduction for survivors was materially greater than for all patients (cost ratio 0.64).

As with the de novo analysis, no study found that the incremental cost of the MitraClip

procedure would be expected to be offset from savings from fewer hospitalisations or

avoided surgery compared with managing patients on medical therapy. To answer this

question with more confidence, prospective parallel studies comparing MitraClip directly with

medical therapy are required. Two such studies, both set in patients in FMR, have recently

been published which do this. However, the MITRA-FR study [42] and the COAPT study [43]

have reported contradictory evidence with regard to this.

3.5.11 Question Eleven

“Is MitraClip cost-effective from the perspective of the NHS?”

As noted in response to Question Ten, currently there are no comparator data on English

patients receiving optimal medical management to enable this question to be answered.

The response to Question Ten has set out evidence which indicates that MitraClip is likely to

cost the NHS more per patient than the costs to manage similar patients with medical

therapy, when the device and procedural costs are included. Three studies [76-78] indicated

that, over the lifetime of a patient, the incremental cost of the MitraClip pathway per quality

adjusted life year (QALY), compared with managing patients on medical therapy, could be

similar to recognised willingness to pay thresholds. In the context of NHS England this could

be around £32,560 (see Question Nine). A further three economic studies were identified but

these were judged to have low internal and external validity and hence not used in further

analyses (Asgar, 2017, Guerin, 2016, Palmieri, 2015) [80-82].

84

Three studies conducted a cost utility analysis, calculating the additional quality adjusted life

years gained with the MitraClip procedure compared to medical therapy. The incremental

cost effectiveness ratios were calculated as:

• 7,900€ (£6,320) by Armeni et al. (2016) [78]

• C$23,433 (£13,020) by Cameron et al. (2014) [77]

• £14,800 by Mealing et al. (2013) [76].

All authors concluded that the intervention was cost-effective at the relevant willingness to

pay threshold for a quality adjusted life year in their country. However, due to weaknesses in

the conduct of the studies, particularly the use of admissions for heart failure only, these

results are not reliable to inform decisions by NHS England.

Relevant evidence from the registry to inform a decision on cost-effectiveness includes the

improved QoL reported by patients with a MitraClip device(s). The mean increase in QoL at

two years, as measured by the EQ-5D instrument, was 0.21 (see Table 18).

There was no strong evidence to support improved survival rates following the procedure

compared with optimal medical management (see response to Question Four). However,

recently the COAPT study has reported MitraClip may improve survival rate in carefully

selected patients with FMR [43].

Conclusion

The published evidence, using cost-utility analysis, suggests that the MitraClip procedure

may be cost-effective at conventional willingness to pay thresholds. However, that evidence

is not sufficiently robust to inform NHS England’s procurement decision. The EAC has

undertaken economic analysis which has indicated that the cost-saving potential of

MitraClip, through reduced future hospital admissions, is unlikely to offset the cost of the

device and procedure itself (Question Ten). However, this did not take into account gains in

QoL utilities used in cost utility analysis. To calculate this, more research would need to be

undertaken.

85

3.5.12 Summary of answers to NHS England questions

Answers to the NHS England questions are summarised in Table 27.

Table 27. Summary of NHS England answers.

Final version of question, as amended NICE following discussion with EAC

Summary answer (from registry data supplemented by published literature)

1) Can UK clinical teams undertaking MitraClip reproduce the reduction in mitral regurgitation seen in the early clinical trials?

Evidence reported from the CtE registry shows that the MitraClip system results in immediate and significant improvements in MR grade with the proportion of patients with moderate-severe or severe MR (≥3+) reduced from 99.5% to 6.7%. This large effect is consistent with all the studies identified in the literature. However, after 1 year there is evidence that there is some deterioration in mitral valve function, with 24.4% of patients reporting moderate-severe or severe MR. Again, this is fully consistent with the published literature. Evidence from the EVEREST II RCT reported that reduction in MR grade was similar for both MitraClip and surgery patients; however, this is contradicted by several observational studies which have reported surgery is superior, especially in the longer term.

2) Is reduction in mitral regurgitation mediated by MitraClip associated with improvements in quality of life?

The CtE registry reported significant improvements in QoL, as measured by utility and VAS, at all follow up time points compared with baseline. There were statistically significant improvements in all domains of EQ-5D at 6 months. The use of MitraClip was also associated with an immediate and sustained improvement in NYHA class. These results were consistent with those published in the literature.

3) Does MitraClip improve survival rates? The CtE registry reported an in-hospital mortality rate of 5.0% (95% CI 2.4% to 9.0%) and a 1 year mortality rate of 11.6% (95% CI 7.5% to 16.8%). Linked data reported a 1 year mortality rate of 12.7% (95% CI 7.5 to 15.9%) and 22.7% (95% CI 15.3 to 29.4%) at 2 years. These data appear consistent with results from observational studies performed in patients with similar characteristics. One recent RCT in patients with FMR reported that MitraClip was associated with a significant clinical reduction in mortality after 2 years follow up. This was not observed in another RCT of FMR patients with 1 year follow up.

4) Does MitraClip reduce the frequency of subsequent hospital admissions?

There is no reliable direct evidence from the registry concerning the relative rate of readmission in patients following the MitraClip procedure. Linked data reported that 69% of patients had at least one admission to hospital in the year following the procedure, with an overall annualised rate of 176.1 (95% CI 151.9 to 198.8) admissions per 100 PY). Evidence from observational studies indicated that the rate of readmission for MitraClip patients is high in the first year, ranging from approximately 20 to 60%.

5) Are the early benefits in reduction in mitral regurgitation maintained in the medium-term? Is there a need for repeat treatment over time (either by a repeat percutaneous procedure or surgery)?

The benefits of MitraClip were sustained for the follow up duration of the CtE registry. There is conflicting evidence on the benefits of MitraClip reducing MR compared with surgery in the longer term.

86

Final version of question, as amended NICE following discussion with EAC

Summary answer (from registry data supplemented by published literature)

6) What proportion of patients referred to a specialist MitraClip service as defined in the CtE documents were assessed by the MDT as suitable for the intervention? What proportion of the patients considered suitable for the procedures received it and what proportion of them benefitted?

The CtE registry cannot be used to answer this question. The EAC did not identify any robust UK based data which would inform this question.

7) What are the short and medium term risk of complications from MitraClip use?

MitraClip was successfully deployed in 94% of the patients it was attempted in, with a procedural success rate of 86%. Around 8% of people admitted to hospital for the MitraClip procedure experienced a major complication, with 5% dying. Following discharge, there was a major complication rate of around 15%, with death being the most common contributor (12%). These values appear to be broadly consistent with those reported in the literature, although comparisons of complications are limited due to issues with generalisability.

8) Are clinical outcomes with MitraClip associated with particular patient characteristics (clinical or demographic)?

Limited data volume meant that subgroup analysis was limited to procedural urgency and disease aetiology. Patients in the CtE registry who were admitted urgently or as emergencies had an increased risk of in-hospital mortality. There were no significant differences in outcomes in patients with functional or degenerative MR. Data from published observational studies indicate that patients with FMR have an increased mortality rate and are more likely to be re-admitted to hospital than patients with DMR.

9) What are the full procedural costs of using MitraClip to the NHS?

The central estimate of the cost of a MitraClip procedure is about £32,910, range £29,000 to £34,500.

10) What are the potential cost savings for the NHS arising from patients receiving MitraClip?

De novo economic analysis from the EAC, derived from linked data, has reported that, overall, MitraClip is associated with decreases in hospital admission rates, total hospitalisation days, gross hospitalisation costs, and mean per patient hospitalisation costs. These savings were driven by a reduction in admission for cardiac reasons and heart failure. However, these are unlikely to offset the cost of the procedure and device.

11) Is MitraClip cost-effective from the perspective of the NHS?

This question will require de novo economic analysis using robust NHS England generated data.

Discussion 87

Section 4: Discussion

4.1 SUMMARY OF FINDINGS FROM PRIMARY DATA COLLECTION (CTE

REGISTRY)

From the patients enrolled in the CtE registry, a full analysis was possible on 199 patients.

The mean age of patients was 76.2 years (10.5 years SD); most (68.8%) were men; and

most (60%) had a diagnosis of functional or ischaemic MR. Patients with degenerative MR

were significantly older than those with functional or ischaemic MR. This distribution

probably reflects the inclusion criterion of being too high risk for conventional mitral valve

surgery. The majority of patients were recruited electively (84.4%) with 13.6% admitted

urgently and 2.0% undergoing the procedure as an emergency. All but one patient enrolled

had moderate or severe MR (grade 3+ or 4+), and the large majority of patients had

significant dyspnoea symptoms as measure by NYHA class (92.4% being class 3 or 4). The

mean EuroSCORE II was 6.4 (5.7 SD).

One hundred and eighty seven patients (94%) were recorded as having a MitraClip device

implanted. Both admission and discharge dates were reported for 182 procedures (91.5%),

showing a median length of stay of 5 days (IQR 3.3 to 8.0 days, range 0 to 46 days).

Procedural success, defined as device implanted with no major in-hospital complications

was 85.9% (95% CI 80.3 to 90.4%). In-hospital major complications were reported in 16

patients (8.2%, multiple events permitted), the majority of which were deaths (10 patients).

Four patients required an additional intervention. In-hospital minor complications were

reported in 15 patients (7.6%). MitraClip was associated with an immediate improvement in

MR class, with 93% of patients being ≤MR 2+ on discharge. There were corresponding

improvements in NYHA class.

Post-discharge follow-up was reported in 170 patients (90.9%). Significant improvements in

MR and NYHA grades compared with baseline persisted for up to 2 years, and there was

significant improvement in QoL during this period. Major complications occurring post-

MitraClip procedure discharge were recorded in 25 patients (including 20 deaths) and minor

complications in 22 patients. In total 30 deaths were reported pre- and post-discharge,

accounting for 15.1% of the cohort. Patients undergoing urgent or emergency procedures

were associated with significantly increased occurrence of in-hospital death, with no other

significant associations detected.

The central estimate of procedural cost for MitraClip was £32,910 (range £29,900 to

£34,500). The pre-operative pathway accounted for **%, the procedure **% and subsequent

management **% of total costs respectively. Device cost itself accounted for **% to **% of

total costs.

4.2 SUMMARY OF FINDINGS FROM PRIMARY DATA COLLECTION (DATA

LINKAGE)

By linking the registry to HES APC and ONS routine datasets, a total of 37 deaths were

identified, however given the extended length of follow up at 271.9 PY, the total event-rate of

this outcome was reduced to 13.6 (95% CI 9.6 to 18.8) events per 100 PY. By conducting

data linkage the EAC were able to identify 17 additional deaths, 3 additional patients with a

diagnoses of infective endocarditis and 5 additional patients requiring additional mitral valve

interventions occurring after the MitraClip implantation, which were not captured in the

registry. This was related to the poor completion of follow up records in the registry.

Following discharge from hospital, there was an annualised rate of 176.1 (95% CI 151.9 to

Discussion 88

198.8) readmissions per 100 PY for any cause, with 100 patients (67%) having at least one

admission.

4.3 RESULTS IN THE CONTEXT OF OTHER STUDIES

The EAC performed a literature search which informed a literature review document that

reported on the current evidence base for MitraClip [14]. All the studies included for analysis

were published subsequent to the IP664 overview document of IPG309 [6]. One RCT was

identified in the original literature review. The EVEREST II RCT [24] failed to report

equivalence of the MitraClip device with open surgery in its primary composite outcome.

Data from this trial were not considered generalisable to the CtE registry because the

patients were healthy enough to be surgical candidates, whereas the only option available to

the CtE cohort patients was conventional medical management. Recently, a further two

RCTs have been published, both of which studied MitraClip in high risk patients with FMR

compared with medical management. The COAPT trial published positive results in both its

primary outcome (hospitalisation) and all its secondary outcomes [43]. In contrast, the

MITRA-FR study reported MitraClip did not benefit the patients it was used in [42].

Three observational studies were identified which included comparator groups receiving

medical management [28-30]. However, the comparator arms used in these studies were

historical or actively selected for, thus results were considered to be at high risk of bias and

confounding. The EAC also identified three large prospective single-armed registries [31, 33,

34] and one large retrospective study [36]. Additionally, the TVT registry [41], which was

published subsequent to the literature review [14], was included for analysis due to its large

size (n = 2952) and reporting of relevant outcomes. All the observational studies included

high-risk populations that largely reflected those of the CtE registry.

The CtE registry provided good data on the procedural efficacy and in-hospital safety of

MitraClip. Although statistical comparisons could not be performed due to data

heterogeneity, peri-procedural outcomes were generally consistent with those reported in the

literature from observational studies. These studies indicated MitraClip is associated with an

immediate and significant improvement in MR grade, which appears to positively reduce

symptoms, as evidenced by improved NYHA class and QoL. However, MitraClip was also

associated with a substantial in-hospital risk of death, reported as 5.1% in the CtE registry

which is consistent with observational studies. This rate included patients admitted urgently

or as an emergency, which was poorly reported in the literature.

In the medium term, the CtE registry concurred with observational studies that the reduction

in MR and improvements in NYHA associated with MitraClip were sustained. However, the

mortality rate at 1 year was high, ranging from 10.3% in the Giannini study [29] to 25.3% in

the TVT registry [41]. In addition to readmission, there is observational evidence that a high

proportion of patients receiving MitraClip are subsequently readmitted for cardiac indications,

although the COAPT study reported that MitraClip reduced the readmission rate compared

with patients receiving only medical therapy. The CtE registry could not provide data on the

longer-term safety and efficacy of MitraClip. However, limited observational evidence

suggests that the device may not be as effective in reducing MR as surgery after 4 years

[69].

The EAC identified four economic studies with medium internal validity that informed the

economics of the MitraClip procedure [18, 76-78]. These reported that costs per patient are

higher in patients who receive a MitraClip device compared to those who do not, and that

savings in future years from reduced hospital admissions from heart failure are probably not

sufficient to offset the initial cost of the procedure. However, there were key uncertainties

Discussion 89

relating to overall clinical effectiveness of MitraClip compared with other treatment

strategies, and the impact of this on resource use, particularly in the longer-term. This

limitation has meant there is currently no robust economic data to reliably estimate the cost-

effectiveness of the procedure.

In summary, the EAC has analysed data collected from the CtE registry and qualitatively

compared it with published data. Procedural and peri-procedural data from patients were

comparable with the registry, as were medium-term efficacy and safety data. Importantly,

MitraClip appears to be associated with significant improvements in QoL. However, this

population of patients have a high mortality rate and requirement for further hospital

treatment. There remains considerable uncertainty concerning the longer-term efficacy of

MitraClip and how it compares with conservative medical management. This uncertainty also

means it is unclear what the true costs of this intervention are in this high-morbidity

population.

4.4 LIMITATIONS AND STRENGTHS

4.4.1 Limitations of primary evidence

The CtE registry was a single armed study and thus comparisons had to be made implicitly

with results published in the literature [83]. This had two limitations. Firstly, no statistical or

quantitative comparisons could be made with the comparator of interest, which was

conservative medical management (not surgery). Secondly, some of the published literature

was not directly comparable to the registry. Specifically, evaluation of CtE results with trial

data was limited by differences in outcome terminology and measurement, and possible

issues with generalisability of the population (this was particularly with respect to the single

RCT identified in the original literature review, the EVEREST II trial [24]). Thus inferences of

equivalence (or not) of CtE and trial data are subject to considerable uncertainty.

Other specific and non-specific limitations with the registry include the following:

• Two of the key outcomes were MR grade and NYHA class, neither of which are

“hard” outcomes. Measurement of MR grade has frequently involved subjective

judgement on behalf of the assessor [84], and over optimism in interpretation could

have led to detection bias. Similarly, NYHA class can lead to performance bias on

behalf of the patient [55].

• Although initially designed for 5 years follow up, CtE registry follow up was limited to

a maximum of 2 years. This meant that longer-term efficacy outcomes or data on

longer-term complications were not available.

• In addition to the 2-year cut off point, most patients were not eligible for assessment

at this time point because of the timeframe of the study and associated deadlines.

For instance, only 45 patients were eligible for follow up at 2 years (28.3% of the total

cohort), because most patients had the procedure within the previous 2 years, and

were therefore not qualified for the 2-year follow up. Of these 45 patients, only 11

had follow up data reported (24.4%).

• Kaplan-Meier analysis assumed “no event” status of patients unless an event was

recorded. Thus the analysis relies on complete reporting of all event data. Patients

who are lost to follow up are censored from the analysis, but it is unclear if these are

representative of the overall cohort. Finally, patients may have multiple events

(excluding death), but the Kaplan-Meier protocol only analyses time to first event.

• The CtE registry study reported on a case mix of patients admitted electively or

urgently/emergencies. Although it was clear these represented different cohorts with

Discussion 90

different patient characteristics (discussed in Question Eight), sample sizes

(particularly in the latter category) were insufficient to report separate results

throughout the manuscript. Therefore aggregate data has generally been reported.

There were similar issues with other defining patient characteristics (e.g. patient

aetiology).

In conducting data linkage the following assumptions were also made:

• The EAC assume that the identifiers and matching process were correct for the 166

patients who did not have a red flag (i.e. who had no conflicting data fields appearing

across the 3 data sources to suggest otherwise).

• The proportion of patients with a “red flag” associated with their data linkage (21/187;

11.2%) was higher than expected in a well-characterised cohort. In excluding these

patients, the EAC have assumed that the remaining 166 are i) representative of all

those successfully implanted and ii) have had their data successfully linked.

• ONS is regarded as the gold standard of death reporting, however there is a delay in

some deaths appearing in the ONS mortality dataset due to lack of death certification

(i.e. those under investigation). Therefore additional deaths recorded in the registry

and confirmed in HES but not appearing in ONS were also included in the total

number of deaths.

• Long-term follow-up in the registry was poor, therefore long-term analysis (post-

MitraClip discharge) relied on HES/ONS data.

Specific limitations concerning the identification of hospital resource use, and monetisation

of these, included the following:

• The analysis was based on a “before and after” design. Whilst this allowed patients

to act as their own controls, this design does not prove causality of the MitraClip

procedure on the outcomes measured.

• This was a novel approach to costing, the validity of which has not yet been

extensively analysed.

• Admissions in the 1 year prior to MitraClip may include work-ups for the procedure

which were conducted in an inpatient setting, however the EAC were unable to

differentiate these in the HES extract. The consequence of this is that the true cost

difference in admitted patient care between the year prior to and year following

MitraClip implantation will be smaller than that estimated by this analysis.

• The cost of the MitraClip procedure was not included in this cost analysis (ranging

from £29,000 to £34,500 with a central estimate of £32,910). Therefore the cost

difference in admitted patient care between the year prior to and year following

MitraClip implantation is significantly overshadowed by the cost of the procedure

itself.

• Costs only represent those accrued in admitted patient care. Other healthcare

resources (such as outpatient, accident and emergency costs) were not included in

this analysis.

4.4.2 Strengths of primary evidence

The CtE registry had several strengths. Firstly, the registry enrolled indicated patients

consecutively and represented a pragmatic real-world cohort of patients receiving treatment

with MitraClip as performed in the NHS. Thus the external applicability of the registry to

Discussion 91

future practice is high, although improvements in the procedure protocol and the learning

curve effect may ultimately lead to improved outcomes.

Secondly, following an initial disappointing response from centres in providing follow-up

data, this improved considerably such that there was 111.2 PY follow up available for

analysis. Follow up was particularly robust up to 1 year, with 79.4% of eligible patients

reporting follow up data at this time point. The completion of individual data fields varied, but

overall, data completeness was regarded as good. The number reporting results for each

data field are presented in the Supplementary Material Table 4 of this report.

Thirdly, patients in the registry were linked to HES APC and ONS data, which provided a

more comprehensive estimate of mortality and additional MV intervention event rates due to

increased coverage and longer duration of follow up (271.9 person years). By analysing

these data in combination with the data directly reported from the registry, the EAC can be

confident the outcome data accurately reflects real clinical event rates.

Finally, the CtE registry reported important clinical outcomes, which also allowed for limited

subgroup analysis. In addition, the registry captured quality of life data and, through the use

of pro formas directed at a centre level, estimated the cost of the procedure.

Conclusion 92

Section 5: Conclusion

The MitraClip CtE registry enrolled 199 patients with a mean age of 76.2 years (10.5 years

SD). Nearly all patients had clinically significant symptomatic MR, with the majority (60%)

being of functional or ischaemic aetiology. A minority of patients (16%) were treated urgently

or as an emergency. Patients were considered to be at high risk from conventional cardiac

open surgery, with a mean EuroSCORE II of 6.4 (5.7 SD). The patient characteristics were

broadly similar to those identified in observational studies in the published literature, but

were different from those recruited to the only published RCT against surgery to date, the

EVEREST II study [24], with the patients in the RCT being well enough to have open heart

surgery.

One hundred and eighty seven patients (94%) had a MitraClip device successfully

implanted, with a procedural success rate (device implanted and no major complications) of

85.9% (95% CI 80.3 to 90.4%). There was a major in-hospital adverse rate of 8.2% (95% CI

4.7 to 12.9%) which included 10 deaths (about 5%). This was broadly comparable to data

reported in the literature, although direct comparisons could not be made due to differences

in the terminology used. The MitraClip device was associated with a median length of

hospital stay of 5 days (IQR 3.3 to 8.0, range 0 to 46 days).

In patients who had a MitraClip successfully implanted, there was an immediate and

significant reduction in MR grade at discharge, from 100% MR grade ≥3+ to 7% MR grade

≥3+. This benefit diminished slightly over time, with 24% of patient reporting MR grade ≥3+

after 1 year. Reduction in MR was associated with a corresponding improvement in

dyspnoea symptoms as measured by NYHA class. Furthermore there was a significant

improvement reported in QoL as measured by EQ-5D, both in terms of overall health utility

and all individual domains. These results are consistent with those reported in the literature.

The registry reported an overall death rate of 15.1% (95% CI 10.4 to 20.8%). The death rate

at 1 year was 11.6% (95% CI 7.5% to 16.8%). The annualised death rate using time to event

analysis, derived from a total of 111.2 recorded person years (PY), was 27.0 (95% CI 18.2 to

38.5%) per 100 PY; this higher mortality rate was likely artefactual due to issues with

censoring. Using linked data, an annualised rate of 13.6% (95% CI 9.8 to 18.8) per 100 PY

was reported, and this is considered likely to be closer to the true value. The mortality rate in

the CtE MitraClip cohort generally compared favourably with similar patients reported in the

literature, with annual rates of between 10% and 25% reported. Three observational studies

reported that MitraClip improved survival compared to matched comparators receiving

optimal medical management [28-30], but these were subject to confounding. Recently, two

new RCTs have compared MitraClip with medical management exclusively in patients with

FMR. These reported contradictory findings, with one RCT reporting MitraClip was

associated with a large and significant reduction in mortality at 2 years [43], whilst the other

reported that there was no difference at 1 year follow up [42]. It is possible these conflicting

results were due to differences in patient selection and their heart failure management.

MitraClip is an expensive procedure, with an estimated cost ranging from about £29,000 to

£34,500, with a central estimate of around £32,910. De novo economic analysis performed

by the EAC, and reported in published literature [18], suggests that whilst MitraClip reduces

future hospital admissions and healthcare resource use, this is unlikely to offset the initial

device and procedural costs. Robust economic analysis is required to determine the cost-

effectiveness of the procedure.

Conclusion 93

In conclusion, the CtE registry has demonstrated that MitraClip, when successfully

implanted, is causally associated with clinically important reductions in MR, associated

symptoms, and improved QoL. MitraClip has been the subject of extensive observational

research since the publication of IPG309 in August 2009 [1], and the results of the registry

broadly concur with the results reported in this observational evidence base. Recently, two

pivotal RCTs exclusively in patients with FMR have reported contradictory results in the

comparison of MitraClip with medical management [42, 43]. It will be important to understand

the reasons for the differences in these trials to inform commissioning, particularly in terms

of patient selection. Two further trials, also both recruiting patients with FMR, are on-going

and may provide some clarification on the efficacy and safety of MitraClip in the FMR

population [45, 46]. However, there are currently no published protocols for an RCT in high-

risk patients with DMR, so the efficacy of MitraClip in this population may remain uncertain

for the foreseeable future.

Acknowledgements 94

Section 6: Acknowledgements

External Assessment Centre (EAC) Administrator in Newcastle, Emma Belilios, acted as

secretary and produced minutes of meetings and teleconferences throughout the duration of

the cardiac CtE project.

EAC literature searches were conducted by Mick Arber, Senior Information Specialist at York

Health Economics Consortium.

Roseanne Jones, Research Scientist within the Newcastle EAC team, supported literature

sifting as a second reviewer to Iain Willits and Helen Cole.

Julie Burn, Clinical Scientist in Clinical Computing at Newcastle, acted as an adviser to Kim

Keltie and Sam Urwin, in support of their data analysis work for cardiac CtE.

References 95

Section 7: References

1. National Institute for Health and Care Excellence. Percutaneous mitral valve leaflet repair for mitral regurgitation (IP309) 2009. 2. Willits I, Jones R, Bousfield D, Arber M, Cole H, Sims AJ. OUTPUT ONE: Recommendations on three NHS England Commissioning through Evaluation (CtE) registry draft datasets for MitraClip, LAAO and PFO Closure cardiovascular procedures. In: EAC NY, editor. Unpublished; 2014. 3. Cole H, Willits I, Craig J, Sims AJ. OUTPUT TWO: Data quality monitoring process. Unpublished; 2015. 4. Willits I, Craig J, Cole H, Urwin S, Sims AJ. OUTPUT THREE: Draft registry data analysis protocol - MitraClip. In: EAC NY, editor. Unpublished; 2015. 5. NHS England. Specialised Services Circular: Percutaneous mitral valve leaflet repair (using MitraClip) CtE Provider Selection Confirmed (SSC 1454). 2014. 6. National Institute for Health and Care Excellence. Interventional procedure overview of percutaneous mitral valve leaflet repair for mitral regurgitation (IP664). 2008. 7. R Core Team. . R Foundation for Statistical Computing V, Austria, . URL http://www.R-project.org/. R: A language and environment for statistical computing. Vienna; 2016. 8. NHS Digital. Delivering linked datasets to support health and care delivery and research v1.0. 2018 [cited 2018 18th September]; Available from: https://digital.nhs.uk/binaries/content/assets/website-assets/services/dars/linked-datasets-in-nhs-digital-final.pdf 9. Aylin P, Tanna N, Bottle A, Jarman B. How often are adverse events reported in English hospital statistics? BMJ. 2004;329:369. 10. Aylin P, Tanna N, Bottle A, Jarman B. How often are adverse evenets reported in English hospital statistics? BMJ. 2004;329:369. 11. Tweedie MCK. An Index Which Distinguishes Between Some Important Exponential Families. In: Ghosh JK, Roy J, editors. Statistics: Applications and New Directions Proceedings of the Indian Statistical Institute Golden Jubilee International Conference. Calcutta; 1984. 12. Dunn PK, Smyth GK. Evaluation of Tweedie Exponential Dispersion Model Densities by Fourier Inversion. Statistics and Computing. 2008;18:73-86. 13. Zhang J. Likelihood-Based and Bayesian Methods for Tweedie Compound Poisson Linear Mixed Models. Statistics and Computing. 2013;23:743-57. 14. Willits I, Craig J, Arber M, Cole H. RX085 Commissioning through Evaluation (CTE) registry of 3 cardiovascular procedures. MitraClip literature review document. Unpublished; 2016. 15. Agency for Healthcare Research and Quality (AHRQ). Registries for evaluating patient outcome: a user's guide: Volume 1. 3rd ed. Rockville, MD; 2012. 16. Agency for Healthcare Research and Quality (AHRQ). Registries for evaluating patient outcome: a user's guide: Volume 2. 3rd ed. Rockville, MD; 2012. 17. Willits I, Craig J, Cole H, Urwin S, Sims AJ. OUTPUT THREE: Draft registry data analysis protocol - LAAO. In: EAC NY, editor. Unpublished; 2015. 18. Vemulapalli S, Lippmann SJ, Krucoff M, Hernandez AF, Curtis LH, Foster E, et al. Cardiovascular events and hospital resource utilization pre- and post-transcatheter mitral valve repair in high-surgical risk patients. American Heart Journal. 2017 Jul;189:146-57. 19. Department of Health. NHS reference costs 2015 to 2016. 2016. 20. Curtis L, Burns A. Unit Costs of Health and Social Care 2016. In: Personal Social Services Research Unit UoK, editor.; 2016. 21. Curtis L, Burns A. Unit Costs of Health and Social Care 2017. 2016 [cited 2018 1st November]; Available from: https://doi.org/10.22024/UniKent/01.02/65559 22. Van Reenen M, Janssen B. EQ-5D-5L User Guide: Basic information on how to use the EQ-5D-5L instrument. 2015. 23. Moher D, Liberati A, Tetzlaff J, Altman G, Prisma PRISMA. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS medicine. 2009;6(7).

http://www.r-project.org/

http://www.r-project.org/

https://digital.nhs.uk/binaries/content/assets/website-assets/services/dars/linked-datasets-in-nhs-digital-final.pdf

https://digital.nhs.uk/binaries/content/assets/website-assets/services/dars/linked-datasets-in-nhs-digital-final.pdf

https://doi.org/10.22024/UniKent/01.02/65559

References 96

24. Feldman T, Foster E, Glower DD, Kar S, Rinaldi MJ, Fail PS, et al. Percutaneous repair or surgery for mitral regurgitation.[Erratum appears in N Engl J Med. 2011 Jul 14;365(2):189 Note: Glower, Donald G [corrected to Glower, Donald D]]. New England Journal of Medicine. 2011 Apr 14;364(15):1395-406. 25. Feldman T, Kar S, Elmariah S, Smart SC, Trento A, Siegel RJ, et al. Randomized Comparison of Percutaneous Repair and Surgery for Mitral Regurgitation: 5-Year Results of EVEREST II. Journal of the American College of Cardiology. 2015 Dec 29;66(25):2844-54. 26. Feldman T, Foster E, Qureshi M, Whisenant B, Williams J, Glower D, et al. The everest ii randomized controlled trial (RCT): Three year outcomes. Journal of the American College of Cardiology; 2012. p. B229-b30. 27. Feldman T, Kar S, Rinaldi M, Fail P, Hermiller J, Smalling R, et al. Percutaneous mitral repair with the MitraClip system: safety and midterm durability in the initial EVEREST (Endovascular Valve Edge-to-Edge REpair Study) cohort. Journal of the American College of Cardiology. 2009 Aug 18;54(8):686-94. 28. Whitlow PL, Feldman T, Pedersen WR, Lim DS, Kipperman R, Smalling R, et al. Acute and 12-month results with catheter-based mitral valve leaflet repair: the EVEREST II (Endovascular Valve Edge-to-Edge Repair) High Risk Study. Journal of the American College of Cardiology. 2012 Jan 10;59(2):130-9. 29. Giannini C, Fiorelli F, De Carlo M, Guarracino F, Faggioni M, Giordano P, et al. Comparison of Percutaneous Mitral Valve Repair Versus Conservative Treatment in Severe Functional Mitral Regurgitation. American Journal of Cardiology. 2016 Jan 15;117(2):271-7. 30. Velazquez EJ, Samad Z, Al-Khalidi HR, Sangli C, Grayburn PA, Massaro JM, et al. The MitraClip and survival in patients with mitral regurgitation at high risk for surgery: A propensity-matched comparison. American Heart Journal. 2015 Nov;170(5):1050-9.e3. 31. Baldus S, Schillinger W, Franzen O, Bekeredjian R, Sievert H, Schofer J, et al. MitraClip therapy in daily clinical practice: initial results from the German transcatheter mitral valve interventions (TRAMI) registry. European Journal of Heart Failure. 2012 Sep;14(9):1050-5. 32. Puls M, Lubos E, Boekstegers P, von Bardeleben RS, Ouarrak T, Butter C, et al. One-year outcomes and predictors of mortality after MitraClip therapy in contemporary clinical practice: results from the German transcatheter mitral valve interventions registry. European Heart Journal. 2016 Feb 21;37(8):703-12. 33. Maisano F, Franzen O, Baldus S, Schafer U, Hausleiter J, Butter C, et al. Percutaneous mitral valve interventions in the real world: early and 1-year results from the ACCESS-EU, a prospective, multicenter, nonrandomized post-approval study of the MitraClip therapy in Europe. Journal of the American College of Cardiology. 2013 Sep 17;62(12):1052-61. 34. Nickenig G, Estevez-Loureiro R, Franzen O, Tamburino C, Vanderheyden M, Luscher TF, et al. Percutaneous mitral valve edge-to-edge repair: in-hospital results and 1-year follow-up of 628 patients of the 2011-2012 Pilot European Sentinel Registry. Journal of the American College of Cardiology. 2014 Sep 2;64(9):875-84. 35. Pighi M, Estevez-Loureiro R, Maisano F, Ussia GP, Dall'Ara G, Franzen O, et al. Immediate and 12-Month Outcomes of Ischemic Versus Nonischemic Functional Mitral Regurgitation in Patients Treated With MitraClip (from the 2011 to 2012 Pilot Sentinel Registry of Percutaneous Edge-To-Edge Mitral Valve Repair of the European Society of Cardiology). American Journal of Cardiology. 2017 Feb 15;119(4):630-7. 36. Rahhab Z, Kortlandt FA, Velu JF, Schurer RAJ, Delgado V, Tonino P, et al. Current MitraClip experience, safety and feasibility in the Netherlands. Netherlands Heart Journal. 2017 Jun;25(6):394-400. 37. Takagi H, Ando T, Umemoto T, Group A. A review of comparative studies of MitraClip versus surgical repair for mitral regurgitation. International Journal of Cardiology. 2017 Feb 01;228:289-94. 38. Philip F, Athappan G, Tuzcu EM, Svensson LG, Kapadia SR. MitraClip for severe symptomatic mitral regurgitation in patients at high surgical risk: A comprehensive systematic review. Catheterization & Cardiovascular Interventions. 2014 Oct 1;84(4):581-90.

References 97

39. Vakil K, Roukoz H, Sarraf M, Krishnan B, Reisman M, Levy WC, et al. Safety and efficacy of the MitraClip system for severe mitral regurgitation: a systematic review. Catheterization & Cardiovascular Interventions. 2014 Jul 1;84(1):129-36. 40. Chiarito M, Pagnesi M, Martino EA, Pighi M, Scotti A, Biondi-Zoccai G, et al. Outcome after percutaneous edge-to-edge mitral repair for functional and degenerative mitral regurgitation: a systematic review and meta-analysis. Heart. 2017 Jun 29;29:29. 41. Sorajja P, Vemulapalli S, Feldman T, Mack M, Holmes DR, Jr., Stebbins A, et al. Outcomes With Transcatheter Mitral Valve Repair in the United States: An STS/ACC TVT Registry Report. J Am Coll Cardiol. 2017 Nov 7;70(19):2315-27. 42. Obadia JF, Messika-Zeitoun D, Leurent G, Iung B, Bonnet G, Piriou N, et al. Percutaneous Repair or Medical Treatment for Secondary Mitral Regurgitation. N Engl J Med. 2018 Aug 27. 43. Stone GW, Lindenfeld J, Abraham WT, Kar S, Lim DS, Mishell JM, et al. Transcatheter Mitral-Valve Repair in Patients with Heart Failure. N Engl J Med. 2018 Sep 23. 44. Nishimura RA, Bonow RO. Percutaneous Repair of Secondary Mitral Regurgitation — A Tale of Two Trials. N Engl J Med. 2018 Dec 13. 45. Anker S, Schillinger W. A Clinical Evaluation of the Safety and Effectiveness of the MitraClip System in the Treatment of Clinically Significant Functional Mitral Regurgitation (Reshape-HF2). 2017 [cited 2017 3rd October]; Available from: https://clinicaltrials.gov/ct2/show/NCT02444338 46. Volker R, Baldus S. A Multicenter, Randomized, Controlled Study to Assess Mitral vAlve reconsTrucTion for advancEd Insufficiency of Functional or iscHemic ORigiN (MATTERHORN). 2017 [cited 2017 11th October]; Available from: https://clinicaltrials.gov/ct2/show/NCT02371512 47. Prins C, de Villiers Jonker I, Botes L, Smit FE. Cardiac surgery risk-stratification models. Cardiovasc J Afr. 2012 Apr;23(3):160-4. 48. European System for Cardiac Operative Risk Evaluation. EuroSCORE calculator. 2011 [cited 2018 22nd February]; Available from: http://www.euroscore.org/ 49. The Society of Thoracic Surgeons. Risk calculators. [cited 2018 22nd February]; Available from: https://www.sts.org/resources/risk-calculators 50. Apostolakis EE, Baikoussis NG. Methods of estimation of mitral valve regurgitation for the cardiac surgeon. J Cardiothorac Surg. 2009 Jul 15;4:34. 51. Alozie A, Paranskaya L, Westphal B, Kaminski A, Sherif M, Sindt M, et al. Clinical outcomes of conventional surgery versus MitraClip therapy for moderate to severe symptomatic mitral valve regurgitation in the elderly population: an institutional experience. BMC Cardiovascular Disorders. 2017 Mar 20;17(1):85. 52. Taramasso M, Denti P, Buzzatti N, De Bonis M, La Canna G, Colombo A, et al. Mitraclip therapy and surgical mitral repair in patients with moderate to severe left ventricular failure causing functional mitral regurgitation: a single-centre experience. European Journal of Cardio-Thoracic Surgery. 2012 Dec;42(6):920-6. 53. De Bonis M, Lapenna E, Buzzatti N, La Canna G, Denti P, Pappalardo F, et al. Optimal results immediately after MitraClip therapy or surgical edge-to-edge repair for functional mitral regurgitation: are they really stable at 4 years? European Journal of Cardio-Thoracic Surgery. 2016 Sep;50(3):488-94. 54. Janssen MF, Pickard AS, Golicki D, Gudex C, Niewada M, Scalone L, et al. Measurement properties of the EQ-5D-5L compared to the EQ-5D-3L across eight patient groups: a multi-country study. Qual Life Res. 2013 Sep;22(7):1717-27. 55. Raphael C, Briscoe C, Davies J, Ian Whinnett Z, Manisty C, Sutton R, et al. Limitations of the New York Heart Association functional classification system and self-reported walking distances in chronic heart failure. Heart. 2007 Apr;93(4):476-82. 56. Whitehead SJ, Ali S. Health outcomes in economic evaluation: the QALY and utilities. Br Med Bull. 2010;96:5-21. 57. Taramasso M, Maisano F, Denti P, Latib A, La Canna G, Colombo A, et al. Percutaneous edge-to-edge repair in high-risk and elderly patients with degenerative mitral regurgitation: midterm outcomes in a single-center experience. Journal of Thoracic & Cardiovascular Surgery. 2014 Dec;148(6):2743-50.

https://clinicaltrials.gov/ct2/show/NCT02444338

https://clinicaltrials.gov/ct2/show/NCT02371512

http://www.euroscore.org/

https://www.sts.org/resources/risk-calculators

References 98

58. Estevez-Loureiro R, Settergren M, Pighi M, Winter R, D'Allara G, Jacobsen P, et al. Effect of advanced chronic kidney disease in clinical and echocardiographic outcomes of patients treated with MitraClip system. International Journal of Cardiology. 2015 Nov 1;198:75-80. 59. Downs EA, Lim DS, Saji M, Ailawadi G. Current state of transcatheter mitral valve repair with the MitraClip. Annals of Cardiothoracic Surgery. 2015 Jul;4(4):335-40. 60. Yeo KK, Yap J, Yamen E, Muda N, Tay E, Walters DL, et al. Percutaneous mitral valve repair with the MitraClip: early results from the MitraClip Asia-Pacific Registry (MARS). EuroIntervention : journal of EuroPCR in collaboration with the Working Group on Interventional Cardiology of the European Society of Cardiology. 2014 Sep 22;10(5):620-5. 61. Lesevic H, Sonne C, Braun D, Orban M, Pache J, Kastrati A, et al. Acute and Midterm Outcome After MitraClip Therapy in Patients With Severe Mitral Regurgitation and Left Ventricular Dysfunction. American Journal of Cardiology. 2015 Sep 1;116(5):749-56. 62. Glower D, Ailawadi G, Argenziano M, Mack M, Trento A, Wang A, et al. EVEREST II randomized clinical trial: predictors of mitral valve replacement in de novo surgery or after the MitraClip procedure. Journal of Thoracic & Cardiovascular Surgery. 2012 Apr;143(4 Suppl):S60-3. 63. Cheng R, Tat E, Siegel RJ, Arsanjani R, Hussaini A, Makar M, et al. Mitral annular calcification is not associated with decreased procedural success, durability of repair, or left ventricular remodelling in percutaneous edge-to-edge repair of mitral regurgitation. EuroIntervention : journal of EuroPCR in collaboration with the Working Group on Interventional Cardiology of the European Society of Cardiology. 2016 Oct 20;12(9):1176-84. 64. Braun D, Lesevic H, Orban M, Michalk F, Barthel P, Hoppe K, et al. Percutaneous edge-to-edge repair of the mitral valve in patients with degenerative versus functional mitral regurgitation. Catheterization & Cardiovascular Interventions. 2014 Jul 1;84(1):137-46. 65. Calvert MJ, Freemantle N, Cleland JG. The impact of chronic heart failure on health-related quality of life data acquired in the baseline phase of the CARE-HF study. Eur J Heart Fail. 2005 Mar 2;7(2):243-51. 66. Feldman T, Young A. Percutaneous approaches to valve repair for mitral regurgitation. Journal of the American College of Cardiology. 2014 May 27;63(20):2057-68. 67. Office for National Statistics. Statistical bulletin: Deaths registered in England and Wales: 2016. 2017 [cited 2017 6th December ]; Available from: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsregistrationsummarytables/2016#decrease-in-deaths-at-ages-75-and-over 68. Pocock SJ, Ariti CA, McMurray JJ, Maggioni A, Kober L, Squire IB, et al. Predicting survival in heart failure: a risk score based on 39 372 patients from 30 studies. Eur Heart J. 2013 May;34(19):1404-13. 69. De Bonis M, Taramasso M, Lapenna E, Denti P, La Canna G, Buzzatti N, et al. MitraClip therapy and surgical edge-to-edge repair in patients with severe left ventricular dysfunction and secondary mitral regurgitation: mid-term results of a single-centre experience+. European Journal of Cardio-Thoracic Surgery. 2016 Jan;49(1):255-62. 70. Abbot Vascular. MitraClip Transcatheter Mitral Valve Repair. 2017 [cited 2017 7th December ]; Available from: https://www.vascular.abbott/us/products/structural-heart/mitraclip-mitral-valve-repair.html 71. Nickenig G, Schueler R, Dager A, Martinez Clark P, Abizaid A, Siminiak T, et al. Treatment of Chronic Functional Mitral Valve Regurgitation With a Percutaneous Annuloplasty System. J Am Coll Cardiol. 2016 Jun 28;67(25):2927-36. 72. Swaans MJ, Bakker ALM, Alipour A, Post MC, Kelder JC, de Kroon TL, et al. Survival of transcatheter mitral valve repair compared with surgical and conservative treatment in high-surgical-risk patients. Jacc: Cardiovascular Interventions. 2014 Aug;7(8):875-81. 73. Conradi L, Treede H, Rudolph V, Graumuller P, Lubos E, Baldus S, et al. Surgical or percutaneous mitral valve repair for secondary mitral regurgitation: comparison of patient characteristics and clinical outcomes. European Journal of Cardio-Thoracic Surgery. 2013 Sep;44(3):490-6; discussion 6.

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsregistrationsummarytables/2016#decrease-in-deaths-at-ages-75-and-over

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsregistrationsummarytables/2016#decrease-in-deaths-at-ages-75-and-over

https://www.vascular.abbott/us/products/structural-heart/mitraclip-mitral-valve-repair.html

https://www.vascular.abbott/us/products/structural-heart/mitraclip-mitral-valve-repair.html

References 99

74. Buzzatti N, Maisano F, Latib A, Taramasso M, Denti P, La Canna G, et al. Comparison of outcomes of percutaneous MitraClip versus surgical repair or replacement for degenerative mitral regurgitation in octogenarians. American Journal of Cardiology. 2015 Feb 15;115(4):487-92. 75. Deuschl F, Schofer N, Lubos E, Blankenberg S, Schafer U. Critical evaluation of the MitraClip system in the management of mitral regurgitation. Vasc Health Risk Manag. 2016;12:1-8. 76. Mealing S, Feldman T, Eaton J, Singh M, Scott DA. EVEREST II high risk study based UK cost-effectiveness analysis of MitraClip in patients with severe mitral regurgitation ineligible for conventional repair/replacement surgery. Journal of Medical Economics. 2013 Nov;16(11):1317-26. 77. Cameron HL, Bernard LM, Garmo VS, Hernandez JB, Asgar AW. A Canadian cost-effectiveness analysis of transcatheter mitral valve repair with the MitraClip system in high surgical risk patients with significant mitral regurgitation. Journal of Medical Economics. 2014 Aug;17(8):599-615. 78. Armeni P, Boscolo PR, Tarricone R, Capodanno D, Maggioni AP, Grasso C, et al. Real-world cost effectiveness of MitraClip combined with Medical Therapy Versus Medical therapy alone in patients with moderate or severe mitral regurgitation. International Journal of Cardiology. 2016 Apr 15;209:153-60. 79. Glower DD, Kar S, Trento A, Lim DS, Bajwa T, Quesada R, et al. Percutaneous mitral valve repair for mitral regurgitation in high-risk patients: results of the EVEREST II study. Journal of the American College of Cardiology. 2014 Jul 15;64(2):172-81. 80. Palmieri V, Baldi C, Di Blasi PE, Citro R, Di Lorenzo E, Bellino E, et al. Impact of DRG billing system on health budget consumption in percutaneous treatment of mitral valve regurgitation in heart failure. Journal of Medical Economics. 2015 Feb;18(2):89-95. 81. Guerin P, Bourguignon S, Jamet N, Marque S. MitraClip therapy in mitral regurgitation: a Markov model for the cost-effectiveness of a new therapeutic option. Journal of Medical Economics. 2016 Jul;19(7):696-701. 82. Asgar AW, Khairy P, Guertin M-C, Cournoyer D, Ducharme A, Bonan R, et al. Clinical outcomes and economic impact of transcatheter mitral leaflet repair in heart failure patients. Journal of Medical Economics. 2017 Jan;20(1):82-90. 83. AHRQ. Role of Single Group Studies in Agency for Healthcare Research and Quality Comparative Effectiveness Reviews. 2013. 84. McAlister NH, McAlister NK, Buttoo K. Understanding cardiac "echo" reports. Practical guide for referring physicians. Can Fam Physician. 2006 Jul;52:869-74. 85. National Clinical Guideline Centre. Preoperative tests Routine preoperative tests for elective surgery. Appendix M: Economic considerations for Delphi. 2016 [cited 2018 3rd January]; Available from: https://www.nice.org.uk/guidance/ng45/evidence/appendix-m-economic-considerations-for-delphi-pdf-87258149464 86. Information service Division. Theatre Services. October 2016. 2016 [cited 2018 3rd January]; Available from: http://www.isdscotland.org/Health-Topics/Finance/Costs/Detailed-Tables/Theatres.asp 87. BNF. British National Formularly Online. 2017 June 2017 [cited 2017 18th December]; Available from: https://www.bnf.org/products/bnf-online/

https://www.nice.org.uk/guidance/ng45/evidence/appendix-m-economic-considerations-for-delphi-pdf-87258149464

https://www.nice.org.uk/guidance/ng45/evidence/appendix-m-economic-considerations-for-delphi-pdf-87258149464

http://www.isdscotland.org/Health-Topics/Finance/Costs/Detailed-Tables/Theatres.asp

http://www.isdscotland.org/Health-Topics/Finance/Costs/Detailed-Tables/Theatres.asp

https://www.bnf.org/products/bnf-online/

100

Appendix 1 – Data flow diagram – CtE Registry

177 patients discharged alive from hospital

272 patients indicated for MitraClip

203 patients indicated for MitraClip eligible for CtE

Exclusions (multiple reasons permitted): - 57 patients with non-eligible reasons for treatment - 50 patients with plan for treatment including: alongside other valve procedure as a staged procedure, medical therapy, surgical procedure, other or unknown - 6 patients with unknown/none/mild severity of MR - 3 patients with rheumatic aetiology

199 MitraClip patients (incl. 32 non-CtE)

Exclusions: - 2 procedures on waiting list or procedure not conducted - 0 procedures conducted before 01/10/2014 - 2 procedures with discharge date preceding procedure date

187 patients with MitraClip device

implanted

Lost to follow-up: - 12 not implanted successfully/unconfirmed successful implantation (incl. 0 deaths)

Lost to follow-up: - 10 in-hospital deaths

6 week follow-up record: CtE: eligible=177, recorded=165

Lost to follow-up: - 0 not reaching 6 weeks since procedure

6 month follow-up record: CtE: eligible=167, recorded=159

Lost to follow-up: - 8 not reaching 6-months since procedure - 2 deaths reported at 6-weeks

1 year follow-up record: CtE: eligible=131, recorded=145

Lost to follow-up: - 25 not reaching 1-year since procedure - 11 deaths reported at 6-months

2 year follow-up record: CtE: eligible=45, recorded=11

Lost to follow-up: -79 not reaching 2-years since procedure - 7 deaths reported at 1-year

101

Appendix 2 – Patient characteristics – CtE

Registry

Patient characteristics for all eligible MitraClip patients and those with recorded information

from follow-up appointments.

Patient characteristic† All eligible MitraClip patients (n=199)

All patients with device implanted and follow-up recorded (n=170)

P-value

Female 62 (31.2%) 51 (30.0%) 0.83

Age, years median (Q1,Q3) [range]

78 (69,85) [37-94]

79 (69,85) [37-94]

1.00

Body surface area median (Q1,Q3) [range]

1.9 (1.7,2.0) [1.4-2.4]

1.9 (1.7-2.0) [1.4-2.4]

0.95

Ethnic origin: 1.00 Caucasian 179 (91.3%) 152 (91.0%) Black 3 (1.5%) 3 (1.8%) Asian 11 (5.6%) 9 (5.4%) Other 3 (1.5%) 3 (1.8%)

eGFR median (Q1,Q3) [range]

57.0 (42.0, 72.2) [6.1-117.1]

57.0 (41.8,71.8) [6.1-117.1]

0.83

Dialysis 5 (2.5%) 3 (1.8%) 0.73

Smoking status: 1.00 Never smoked 93 (53.4%) 81 (54.0%) Ex-smoker 77 (44.3%) 66 (44.0%) Current smoker 4 (2.3%) 3 (2.0%)

Diabetes 36 (18.1%) 32 (18.8%) 0.89

Hypertension 104 (53.3%) 87 (52.4%) 0.92

Previous neurological disease 26 (13.1%) 24 (14.1%) 0.88

Peripheral vascular disease (PVD) 28 (14.4%) 26 (15.6%) 0.77

Previous myocardial infarction (MI) 0.96 No 117 (59.7%) 98 (58.3%) Yes (within 90 days) 8 (4.1%) 4 (4.27%) Yes (> 90 days) 71 (36.2%) 63 (37.5%)

Critical pre-op status 5 (2.6%) 3 (1.8%) 0.73

CCS angina status 0.99 No angina 147 (75.0%) 125 (73.5%) No limitation of physical activity 22 (11.2%) 19 (11.2%) Slight limitation of ordinary activity 17 (8.7%) 16 (9.4%) Marked limitation of ordinary physical activity 9 (4.6%) 9 (5.3%) Symptoms at rest or minimal activity 1 (0.5%) 1 (0.6%)

NYHA dyspnoea status 0.90 No limitation of physical activity 3 (1.5%) 3 (1.8%) Slight limitation of ordinary physical activity 12 (6.1%) 10 (5.9%) Marked limitation of ordinary physical activity 124 (62.6%) 112 (65.9%) Symptoms at rest or minimal activity 59 (29.8%) 45 (26.5%)

Killip class 0.93 No clinical signs of heart failure 128 (67.7%) 114 (69.9%) Rales in lungs/S3/elevated JVP but not class 3 46 (24.3%) 39 (23.9%) Acute pulmonary oedema 14 (7.4%) 10 (6.1%) Cardiogenic shock 1 (0.5%) 0 (0.0%)

CSHA frailty score 0.86 Very fit/Well/Well with treated comorbidities 48 (28.1%) 46 (30.1%) Apparently vulnerable/Mildly frail 54 (31.6%) 50 (32.7%) Moderately/Severely frail 69 (40.4%) 57 (37.3%)

Katz index of independence 0.98 0 4 (2.4%) 3 (2.1%) 2 3 (1.8%) 1 (0.71%) 3 3 (1.8%) 3 (2.1%) 4 12 (7.2%) 10 (6.8%) 5 6 (3.6%) 5 (3.4%) 6 139 (83.2%) 124 (84.9%)

102



P-value

6 minute walk test, m median (Q1,Q3) [range]

190 (108,294) [0-450]

195 (115,291) [0-450]

0.85

Previous PCI 45 (23.1%) 39 (23.2%) 1.00

Previous cardiac surgery 78 (39.6%) 66 (39.1%) 1.00

Severe liver disease 1 (0.5%) 1 (0.6%) 1.00

History of pulmonary disease 51 (26.0%) 43 (25.6%) 1.00

Previous electric device therapy 51 (25.8%) 48 (28.2%) 0.64

Heart rhythm pre-op 0.97 Sinus rhythm 80 (40.6%) 66 (39.1%) Atrial fibrillation/flutter 93 (47.2%) 80 (47.3%) Paced 17 (8.6%) 16 (9.5%) Other 7 (3.6%) 7 (4.1%)

Severity of mitral regurgitation (MR) 0.93 Grade 2 (mild-moderate) 1 (0.5%) 1 (0.6%) Grade 3 (moderate-severe) 20 (10.1%) 18 (10.6%) Grade 4 (severe) 178 (89.4%) 151 (88.8%)

MR aetiology 0.91 Functional/ischaemic 117 (60.0%) 101 (60.8%) Degenerative 78 (40.0%) 65 (39.2%)

MR jet area 0.99 Grade 1 1 (0.6%) 1 (0.7%) Grade 2 3 (1.8%) 3 (2.1%) Grade 3 31 (18.6%) 26 (18.2%) Grade 4 121 (72.5%) 102 (71.3%) Unknown or not applicable 11 (6.6%) 11 (7.7%)

Heart rate at time of echo, beats per minute median (Q1,Q3) [range]

71 (65,83) [44-114]

70 (65,81) [44-114]

0.83

LVIDd, mm median (Q1,Q2) [range]

57 (51,63) [35-81]

57 (51,63) [35-81]

0.87

LVIDs, mm median (Q1,Q2) [range]

43 (32,50) [18-76]

45 (33,51) [18-76]

0.74

Peak TR velocity, m/s median (Q1,Q2) [range]

3.2 (2.8,3.6) [0.0-4.6]

3.2 (2.8,3.6) [0.0-4.6]

0.84

PA systolic pressure, mmHg median (Q1,Q2) [range]

46 (37,62) [20-100]

48 (38,62) [20-100]

0.74

TAPSE, cm median (Q1,Q2) [range]

1.6 (1.3,1.9) [0.5-3.0]

1.6 (1.3,1.9) [0.5-3.0]

0.77

LVEF 0.94 Good (≥55%) 66 (39.5%) 53 (36.1%) Mild impairment (45-54%) 31 (18.6%) 28 (19.0%) Moderate impairment (30-44%) 39 (23.4%) 37 (25.2%) Severe impairment (<30%) 31 (18.6%) 29 (19.7%)

Estimate of TR severity 0.99 No TR 14 (8.6%) 13 (9.2%) Mild 75 (46.0%) 67 (47.2%) Moderate 50 (30.7%) 41 (28.9%) Severe 24 (14.7%) 21 (14.8%)

Aortic regurgitation (regardless of prev. surgery) 0.97 None 88 (53.7%) 77 (54.2%) Mild 72 (43.9%) 61 (43.0%) Moderate 4 (2.4%) 4 (2.8%)

Aortic stenosis (regardless of prev. surgery) 1.00 None 147 (91.3%) 127 (90.7%) Mild 11 (6.8%) 10 (7.1%) Moderate 3 (1.9%) 3 (2.1%)

Coronary vessel disease 0.96 No vessel with >50% diameter stenosis 68 (65.4%) 59 (62.1%) 1 vessel with >50% diameter stenosis 12 (11.5%) 12 (12.6%) 2 vessels with >50% diameter stenosis 9 (8.7%) 9 (9.5%) 3 vessels with >50% diameter stenosis 15 (14.4%) 15 (15.8%)

Left main stem (LMS) disease 1.00 No LMS disease or LMS disease ≤50% diameter

stenosis 92 (92.0%) 81 (91.0%)

LMS>50% diameter stenosis 8 (8.0%) 8 (9.0%)

103



P-value

BNP (pre-op), pg/ml median (Q1,Q3) [range]

376 (162,846) [13-2000]

353 (156-831) [13-2000]

0.78

Logistic EuroSCORE median (Q1,Q2) [range]

15.6 (9.6,27.1) [1.9-74.7]

15.0 (9.6,26.3) [1.9-74.7]

0.69

Logistic EuroSCOREII median (Q1,Q2) [range]

4.8 (3.0,7.6) [0.7-42.5]

4.7 (2.9,7.6) [0.7-42.5]

0.92

Betablocker (pre-op) 136 (73.9%) 120 (75.5%) 0.80

ACE-I/ARB (pre-op) 133 (72.3%) 114 (71.7%) 0.90

Aldosterone antagonist (pre-op) 49 (26.6%) 47 (29.6%) 0.55

Loop diuretic (pre-op) 146 (79.3%) 126 (79.2%) 1.00

Metolazone/thiazide diuretic 26 (14.1%) 23 (14.5%) 1.00

Digoxin 40 (21.7%) 34 (21.4%) 1.00

Ivabradine 8 (4.3%) 8 (5.0%) 0.80

Note: † Not all data fields were complete for every patient at baseline and follow up. The percentages presented in this table are calculated using the number of patients with each characteristic reported as the denominator.

104

Appendix 3 – Procedural characteristics – CtE

Registry

Procedural details for all eligible MitraClip patients and those with recorded information from

follow-up appointments.

Procedural characteristic† All eligible MitraClip patients (n=199)


P-value

Treating hospital 0.59 Bristol Royal Infirmary 46 (23.1%) 42 (24.7%) Royal Brompton Hospital 94 (47.2%) 83 (48.8%) University Hospital of North Staffordshire (non-CtE)

15 (7.5%) 7 (4.1%)

Wythenshawe Hospital 44 (22.1%) 38 (22.4%)

Procedural urgency 0.50 Elective 168 (84.4%) 148 (87.1%) Urgent 27 (13.6%) 21 (12.4%) Emergency 4 (2.0%) 1 (0.6%)

Anaesthesia 1.00 General anaesthesia 193 (98.0%) 164 (97.6%) Local anaesthesia ± sedation 4 (2.0%) 4 (2.4%)

Intra-operative imaging‡ 1.00 3D TOE 194 (97.5%) 166 (97.6%) 2D TOE 3 (1.5%) 2 (1.2%) None 2 (1.0%) 2 (1.2%)

Delivery approach 1.00 Femoral transvenous trans-septal

(percutaneous) 191 (97.9%) 163 (98.2%)

Femoral transvenous trans-septal (surgical cut-down)

2 (1.0%) 2 (1.2%)

Femoral arterial (retrograde) 1 (0.5%) 1 (0.6%) Unknown or not applicable 1 (0.5%) 0 (0.0%)

Large venous sheath, Fr median (Q1:Q3) [range]

18 (18,24) [1-26]

18 (18,22) [1-24]

0.66

IABP/Impella/CPB 0 (0.0%) 0 (0.0%) NA

Cerebral protection 0 (0.0%) 0 (0.0%) NA

Arterial sheath 0.89 Not used 117 (72.2%) 102 (72.9%) Used – femoral 37 (22.8%) 33 (23.6%) Used – radial/brachial 8 (4.9%) 5 (3.6%)

No. of clips opened 0.86 1 54 (28.0%) 44 (25.9%)

2 111 (57.5%) 103 (60.6%) 3 24 (12.4%) 21 (12.4%) 4 4 (2.1%) 2 (1.2%)

No. of clips successfully deployed 0.96 1 58 (31.0%) 52 (30.6%) 2 103 (55.1%) 96 (56.5%) 3 26 (13.9%) 22 (12.9%)

Venous closure technique 0.94 Manual/Compression 17 (8.8%) 13 (7.8%) Percutaneous/Device 166 (86.0%) 145 (86.8%) Surgical 1 (0.5%) 0 (0.0%) Other 9 (4.7%) 9 (5.4%)

Arterial management 1.00 Manual/Compression 63 (87.5%) 56 (86.2%) Device 5 (6.9%) 5 (7.7%) Other 4 (5.6%) 4 (6.2%)

Fluoroscopy time, mins median (Q1:Q3) [range]

32 (22,43) [8-79]

32 (22,43) [10-79]

0.87

105

Procedural characteristic† All eligible MitraClip patients (n=199)


P-value

X-ray dose, mGray.cm2 median (Q1:Q3) [range]

3879 (3000,6948) [3000-

20,000]

3757 (3000,6543) [3000-

18,260]

0.66

Contrast dose, ml median (Q1:Q3) [range]

0 (0,0) [0-105]

0 (0,0) [0-20]

0.67

Procedural duration, mins median (Q1:Q3) [range]

180 (137,221) [54-300]

184 (140,222) [62-300]

0.82

Time from procedure to extubation, mins median (Q1,Q3) [range]

210 (160,276) [72-

10140]

210 (160,280) [78-935]

0.73

ITU stay 38 (31.4%) 31 (28.2%) 0.67

Length of stay, days Median (Q1:Q3) [range]

5 (3.25,8) [0,46]

- NA

Note: † Not all data fields were complete for every patient at baseline and follow up. The percentages presented in this table are calculated using the number of patients with each characteristic reported as the denominator. ‡ multiple choices permitted

106

Appendix 4 – Outcomes – CtE Registry

Outcomes for all eligible MitraClip patients.

In-hospital (n=199)

After discharge (6w, 6m, 1y, 2y combined)

(n=170)

Total no. of procedures

(multiple events

permitted)

% [95% CI] Total no. of procedures

(multiple events

permitted)

% [95% CI]

Major complications: 16 8.2 [4.7:12.9] 25 14.7 [9.7:20.9] Death 10 5.1 [2.5:9.2] 20 11.8 [7.3:17.6] Neurological event 1 0.5 [0.0:2.8] 2 1.2 [0.1:4.2] Additional surgery 3 1.5 [0.3:4.5] 4 2.4 [0.6:5.9] Device embolisation (percutaneous retrieval) 1 0.5 [0.0:2.9] 0 0.0 [0.0:2.1] MI 2 1.0 [0.1:3.7] 0 0.0 [0.0:2.1] Endocarditis 0 0.0 [0.0:1.9] 0 0.0 [0.0:2.1] Pericardial effusion/tamponade (requiring

intervention) 0 0.0 [0.0:1.9] NA NA

Major vascular injury (requiring intervention) 0 0.0 [0.0:1.9] NA NA MV complication 0 0.0 [0.0:1.9] NA NA Oesophageal rupture 1 0.5 [0.0:3.0] NA NA Major bleed 3 1.6 [0.3:4.6] NA NA AKI (stage 2/3) 4 2.1 [0.6:5.3] NA NA Cardiogenic shock 2 1.1 [0.1:3.9] NA NA

Minor complications: 15 7.6 [4.3:12.2] 22 12.9 [8.30:18.9] Device failure 0 0.0 [0.0:1.9] NA NA

Partial detachment 1 0.5 [0.0:2.9] 1 0.6 [0.0:3.2] Pericardial effusion/tamponade (treated

conservatively) 3 1.6 [0.3:4.5] NA NA

Thrombus 0 0.0 [0.0:1.9] NA NA New moderate/severe mitral stenosis 3 1.8 [0.4:5.2] 21 12.4 [7.8:18.3] Minor bleed 7 3.7 [1.5:7.5] NA NA AKI (stage 1) 1 0.5 [0.0:3.0] NA NA Minor vascular complication 0 0.0 [0.0:1.9] NA NA

Any complication 30 15.2 [10.5:20.9]

44 25.9 [19.5:33.1]

Device implanted 187 94.0 [89.7:96.8]

NA NA

Procedural success (device implanted in absence of major complications)

171 85.9 [80.3:90.4]

NA NA

New requirement for permanent pacing 3 1.6 [0.3:4.7] NA NA

Note: NA Not applicable

107

Appendix 5 – Outcomes for elective/emergency

cases – CtE Registry

Elective (n=168)

Urgent/Emergency (n=31)

Total no. of

patients

% [95% CI] Total no. of

patients

% [95% CI]

In-hospital major complications 10 6.1 [2.9:10.9] 6 19.4 [7.5:37.5] In-hospital minor complications 14 8.4 [4.7:13.7] 1 3.2 [0.1:16.7] Device implanted 158 94.0 [89.3:97.1] 29 93.5 [78.6:99.2] New requirement for permanent pacing 3 1.9 [0.4:5.6] 0 0.0 [0.0:11.9]

Major complications (after discharge) 21 14.2 [9.0:20.9] 4 18.2 [5.2:40.3] Minor complications (after discharge) 21 14.2 [9.0:20.9] 1 4.5 [0.1:22.8]

108

Appendix 6 – Outcomes (time to event analysis) –

CtE Registry

Patient outcomes (in-hospital and after discharge as reported at any follow-up combined) for all eligible MitraClip patients. No. of

patients with event

Total follow-up (person years)

Event rate (per 100 person years follow-up) [95%CI]

No. of patients at risk at 1-year

1-year event-free probability (95%CI)

Death 30 (15.1%) 111.2 27.0 [18.2:38.5] 49 0.818 (0.750 to 0.894) MV intervention 3 (1.8%) 96.8 3.1 [0.6:9.1] 42 0.967 (0.927 to 1.000) Infective endocarditis 0 (0.0%) - - - -

109

Appendix 7 – Kaplan-Meier curves – CtE Registry Kaplan-Meier curve for death (a) and MV intervention (b): Time to event (solid lines),

corresponding 95% confidence limits (shaded area), and proportions of patients event-free

at 1 year (red line).

a. Mortality.

b. MV intervention.

110

Kaplan-Meier curve for death in the elective versus urgent/emergency cohorts (c), plus the

functional/ischaemic MR versus degenerative MR cohorts (d): Time to event (solid lines),

corresponding 95% confidence limits (shaded area), and proportions of patients event-free

at 1 year (red line).

c. Urgency of intervention.

d. Aetiology of MR.

111

Appendix 8 – Mitral regurgitation over time – CtE Registry

Mitral regurgitation grade over time [* denotes significant difference when compared to pre-procedure grade] Pre-op

(n=187) Post-op (n=178)

Discharge (n=174)

6 weeks (n=144)

6 months (n=107)

1 year (n=62)

2 years (n=10)

None 0 (0.0%) 21 (11.8%) 38 (21.8%) 2 (1.4%) 2 (1.9%) 1 (1.6%) 0 (0.0%) Mild 0 (0.0%) 102 (57.3%) 60 (34.5%) 47 (32.6%) 35 (32.7%) 24 (38.7%) 7 (70.0%) Mild/Moderate 1 (0.5%) 43 (24.2%) 59 (33.9%) 61 (42.4%) 50 (46.7%) 22 (35.5%) 2 (20.0%) Moderate/Severe 19 (10.2%) 10 (5.6%) 13 (7.5%) 25 (17.4%) 16 (15.0%) 12 (19.4%) 1 (10.0%) Severe 167 (89.3%) 2 (1.1%) 4 (2.3%) 9 (6.2%) 4 (3.7%) 3 (4.8%) 0 (0.0%)

Fisher’s exact P-value [n pairs]

reference P<0.0001* [n=178]

P=0.0005* [n=174]

P<0.0001* [n=144]

P<0.0001* [n=107]

P<0.0001* [n=62]

P<0.0001* [n=10]

112

Appendix 9 – Cost of a MitraClip procedure

Table A.9.1 identifies all the inputs and sources used to calculate the central cost. Table

A.9.2 provides information on the sensitivity analyses conducted to provide high and low

cost ranges. These are stated at 2015/16 price levels. Subsequently the costs were

increased to 2017/18 costs using the hospital & community health services index [85]. This

was estimated to have increased by 1.8% in each of the two intervening years. This increase

was not applied to the cost of the device, the price of which was agreed with the

manufacturer for the duration of the registry. Both tables report the unit costs at 2015/16

prices to enable readers to track the unit cost back to the original source.

Table A.9.1 Cost of pathway for a MitraClip procedure (2015/16 prices)

Parameter Usage Unit cost % patients Total cost Source

MitraClip pre-operative assessment costs

Consultant cardiologist 50 mins

£104.00 per hr 100% £86.67

1 MDT of 2 cardiologists and 1 nurse for 15 mins per patient + pre-assessment clinic taking 20 mins cardiologist and 60 mins nurse time advised by clinical experts. Costs from PSSRU [20]. Nurse band 6 75 mins

£44.00 per hr 100% £55.00

Echocardiogram with contrast 1 £87.83 100% £87.83 Imaging use from clinical experts;

costs from NHS Reference costs [85]

ECG 1 £40.35 100% £40.35

TOE (day case) 1 £506.30 100% £506.30

Blood gases 1 £6.42 - £9.84 100% £8.13

Tests from clinical experts; costs from ‘Preoperative tests’ by National Clinical Guideline Centre [85]

FBC 1 £3.00 100% £3.00 Tests from clinical experts; costs from NHS Reference costs [19] U&E 1 £3.00 100% £3.00

Sub-total pre-operative assessment costs £790 All costs include overheads.

Peri-operative costs:

Cardiologist

1.726 for 3.34 hours

£104.00 per hr 100% £600.74

Operators from registry, cost PSSRU [20]. Registrar

0.274 for 3.01 hours

£40.00 per hr 100% £33.03

Anaesthetist

1 for for 3.01 hours

£105.00 per hr 100% £316.40

Staffing structure from clinical experts; cost PSSRU [20]

Cath lab assistant band 3

1 for 3.01

hours s £25.00 per

hr 100% £75.33

Echocardiographer

0.75 for 3.01 hours

£46.00 per hr 100% £103.96

Nurse Band 6

1 for 3.34 hours

£44.00 per hr 100% £147.25

Nurse band 5

1 for 3.01 hours

£35.00 per hr 100% £105.47

Cardiac physiologist

0.75 for 3.01 hours

£46.00 per hr 100% £103.96

Radiographer

0.75 for 3.01 hours

£46.00 per hr 100% £103.96

Procedural time in theatre

180.80 mins. £*** per hr 100% £******

Time from registry; cost from Information Services Division (ISD)

113

Parameter Usage Unit cost % patients Total cost Source

cost of theatres excluding staff and consumables costs [86]

TOE or ICE 1 £1,437 100% £1,437

Use from database; costed as EY502 complex echocardiogram for an elective inpatient from NHS Reference costs [19]

X-ray 1 £27.00 45% £12.15 Portsmouth NHS Trust

Anaesthetic drugs - desflurane & remifentanil 1 £82.18 100% £82.18

Drugs agreed with clinical experts; price from a submitted template.

Heparin 2 hrs per surgery and 8/12 hrs after. 1 £5.80 100% £5.80

Drugs from clinical experts; costs from BNF [87]

Cefuroxime X 2 1.5 g, 8 hours apart 1 £10.10 100% £10.10

Consumables 1 £*** 100% £*** Two well-completed templates totalled £*** and £***; used £***.

Device cost includes VAT and 15% overheads per procedure 1 £****** 100% £****** Cost of devices from manfacturer.

Sub-total peri-operative costs £27,707 All costs include overheads.

Post-operative management

Inpatient stay 7.76 days

£356 per day 100% £2,765

Stay mean value from dataset; costed using mean cost for codes EY23A to C for Standard Other Percutaneous Transluminal Repair of Acquired Defect of Heart. Reference costs [19].

Transthoracic Echocardiogram 1 £1,437 0.77 £1,106

Use from database; costed as EY502 complex echocardiogram for an inpatient from NHS Reference costs [19].

Outpatient follow-up 1 £191 100% £191

Use from clinical experts; cost Cardiac Surgery consultant-led outpatients. Reference costs [19]

Sub-total post-operative management £4,062

All costs include overheads. GRAND TOTAL £32,560

http://www.portsmouthccg.nhs.uk/Default.aspx.LocID-033new00v.RefLocID-01p009.Lang-EN.htm

114

Table A.9.2 Low and high cost scenarios for pathway for a MitraClip procedure.

Scenarios Changes from central case New cost

Pre-operative assessment central cost £790

Low cost Use first quartile cost for TOE (£185 vs central value £506) and 20% decrease in all other costs.

£412

High cost Use third quartile cost for TOE (£657 vs central value £506) and 20% increase in all other costs.

£998

Peri-operative costs central cost £27,707

Low cost Use quartile 1 for procedure time, 20% decrease in all other costs except device.

£26,681

High cost Use quartile 3 for procedure time, 20% increase in all other costs except device.

£28,714

Post-operative management central cost £4,062

Low cost Use length of stay time of 3 days and tariff cost for complex echocardiogram for congenital heart disease elective patient.

£1,697

High cost Use 2 days length of stay and 20% increase in all other costs. £4,407

2015/16 prices

Total cost central case and % accounted for by device: £32,560 (**%)

Total case low cost and % accounted for by device £28,790 (**%)

Total cost high cost and % accounted for by device £34,119 (**%)

2017/18 prices

Total cost central case and % accounted for by device: £32,910 (**%)

Total case low cost and % accounted for by device £29,000 (**%)

Total cost high cost and % accounted for by device £34,500 (**%)

Thus the forecast cost for a MitraClip procedure, at 2017/18 prices, ranges from about £29,000 to

£34,500 with the device cost per procedure (£******) accounting for between **% to **% of the total

cost.

115

Appendix 10 – Data Flow Diagram – Linked Data

MitraClip Registry

HES (Admitted Patient Care)

278 procedures (275 patients)

Exclusions: - 39 non-eligible reasons for treatment (clinician/patient

preference, other) - 18 missing reasons for treatment - 7 with stage 1 mitral regurgitation (MR) - 4 with rheumatic MC (contra-indicated) - 2 not started/on waiting list - 2 with discharge date preceding procedure date 199 procedures

(197 patients) )

Exclusions: - 0 procedures conducted after linkage date (01/03/2018)

199 procedures (197 patients)

187 patients (3646 episodes of care)

Exclusions: - 6 type 2 opt-outs - 4 patients unmatched by NHS Digital

RED FLAG: 21 patients Inconsistencies in data between sources casting doubt on patient matching

- 5 with overlapping admissions (i.e. might be multiple patients merged into single identification number)

- 2 with no hospital admission +/- 7 days of MitraClip admission date

- 11 with no cardiac procedure coded (any OPCS-4 code from the “K” Heart chapter)

- 2 with gender mismatch - 2 with age discrepancy larger than +/- 1 year - 2 with death not validated by HES/ONS

GREEN FLAG: 166 patients Passes all internal QA checks

116

Appendix 11 – Linked Data reported in-hospital

complications

Complication recorded* CtE Registry

Acute kidney injury (within 72 hours of procedure) 2

Bailout PCI 1

Cardiogenic shock 2

Clinical bleed 9

- Minor 6

- Major 2

- Life threatening or disabling 1

Device complications 2

- Partial detachment 1

- Embolisation 1

New mitral stenosis 2

- Moderate (MVA <1.5 cm2, >1 cm2) 1

- Severe (MVA <1 cm2) 1

Neurological event 1

- Ischaemic 1

Oesophageal damage during procedure 1

Pericardial effusion 3

- New effusion without tamponade – conservative treatment 2

- New effusion without tamponade – drained 1

Permanent pacing (post-procedure) 2

Post-procedure electrial device therapy (CRT-P) 1

Renal replacement (up to 309 days post-procedure) 2

Requirement of MV surgery 1

Conversion to open heart surgery (other than for a MV procedure) 1

Retrieval of embolisations/detached device 1

- Percutaneous retrieval - successful 1

MI 2

Death 7

117

Complication recorded* HES

D64.9: Anaemia, unspecified 1

D69.6: Thrombocytopenia, unspecified 1

I31.2: Haemopericardium, not elsewhere classified 1

I42.7: Cardiomyopathy due to drugs and other external agents 1

I51.0: Cardiac septal defect, acquired 2

R04.0: Epistaxis 1

R05X: Cough 1

R09.0: Other symptoms and signs involving the circulatory and respiratory systems 1

T81.0: Haemorrhage and haematoma complicating a procedure, not elsewhere

classified

11

T81.2: Accidental puncture and laceration during a procedure, not elsewhere

classified

2

T81.4: Infection following a procedure, not elsewhere classified 1

T82.1: Mechanical complication of cardiac electronic device 1

Death 8

*Note patients can experience multiple complications

†qualified by a supplementary ICD-10 code indicating complication

118

Appendix 12 – Kaplan-Meier analysis by procedural urgency and by

aetiology of mitral regurgitation

a) elective (n=145 patients, 28 deaths) b) urgent (n=21 patients, 9 deaths)

c) FMR (n=99 patients, 23 deaths) d) DMR (n=63 patients, 13 deaths)

119

Appendix 13 – Mitral valve interventions

subsequent to MitraClip implantation - linked

dataset

Admission

method

Primary

diagnosis

[DIAG01]

Primary procedure

[OPERTN01]

Time to

event,

days

(counted

from MC

discharge

date)

Aetiology Urgency of

index MC

implantation

Elective,

booked

I34.0: Mitral

(valve)

insufficiency

K25.3: Prosthetic

replacement of

mitral valve

Y73.1:

Cardiopulmonary

bypass

[Note: Implantation

of cardiac

pacemaker system

NEC within same

hospital visit]

5 FMR Elective

Transfer from

another

hospital (not

emergency)

T82.0:

Mechanical

complication of

heart valve

prosthesis

K25.3: Prosthetic

replacement of

mitral valve

13 FMR Elective

Elective,

waiting list

I08.1 Disorders

of both mitral

and tricuspid

K30.1: Revision of

plastic repair of

mitral valve

Y02.1: Implantation

of prosthesis into

organ NOC

Y49.4: Transapical

approach to heart

Z94.3: Left sided

operation

74 DMR Elective

Elective, via

waiting list

I34.1: Mitral

(valve)

prolapse

K25.2: Xenograft

replacement of

mitral valve

[Note: Implantation

of dual chamber

144 DMR Elective

120

Admission

method

Primary

diagnosis

[DIAG01]

Primary procedure

[OPERTN01]

Time to

event,

days

(counted

from MC

discharge

date)


index MC

implantation

cardiac pacemaker

system within same

hospital visit]

Elective,

booked

I08.0:

Disorders of

both mitral and

aortic valves

K25.2: Xenograft

replacement of

mitral valve

Y49.1: Median

sternotomy

approach

Y71.2: Secondary

operations NOC

[Note: Implantation

of biventricular

cardiac pacemaker

system within same

hospital visit]

196 FMR Elective

Elective,

waiting list

I34.0: Mitral

(valve)

insufficiency

K25.5: Mitral valve

repair NEC

Y02.1: Implantation

of prosthesis into

organ NOC

Y53.4: Approach to

organ under

fluoroscopic control

Y71.1: Subsequent

stage of staged

operations NOC

292 DMR Elective

Emergency T82.6: Infection

and

inflammatory

reaction due to

cardiac valve

prosthesis

K40.1: Saphenous

vein graft

replacement of one

coronary artery

Y18.1: Freeing of

adhesions of organ

NOC

[Note: Prosthetic

replacement of

mitral valve, and

Implantation of

394 FMR Elective

121

Admission

method

Primary

diagnosis

[DIAG01]

Primary procedure

[OPERTN01]

Time to

event,

days

(counted

from MC

discharge

date)


index MC

implantation

cardiac pacemaker

system NEC and

Implantation of

cardioverter

defibrillator using

three electrode

leads all conducted

within same hospital

visit]

*Elective,

planned

I34.0: Mitral

(valve)

insufficiency

K25.5: Mitral valve

repair NEC

Y07.2: Clipping of

organ NOC

Y79.3: Transluminal

approach to organ

through femoral

artery

Y53.4: Approach to

organ under

fluoroscopic control

428 FMR Elective

*Elective,

waiting list

I08.1:

Disorders of

both mitral and

tricuspid valves

K25.2: Xenograft

replacement of

mitral valve

[Note: Implantation

of cardiac

pacemaker system

NEC within same

hospital visit]

513 FMR Elective

Elective,

waiting list

I08.1:

Disorders of

both mitral and

tricuspid valves

K25.3: Prosthetic

replacement of

mitral valve

479 DMR Urgent

*same patient

122

Appendix 14 - Hospital resource (HES only) by procedural urgency of index

MitraClip implantation – a) elective and b) urgent

a) Definition: “X7.05_Procedural.Urgency” of index MC procedure = “1. Elective” 1 year

pre-MC admission*

(n=133)

1 year

post-MC

discharge*

(n=133)

Incidence rate

ratio [95% CI]

1 year

pre-MC

discharge*

all deaths

removed

(n=123)

1 year

post-MC

discharge*

all deaths

removed

(n=123)

Incidence rate

ratio [95% CI]




406 (3052.6) 205 (1617.5)

0.54 [0.45:0.63] 366 (2975.6)

183 (1487.8)

0.50 [0.42:0.60]

Non-cardiac‡ 99 (744.4) 120 (946.8) 1.33 [1.02:1.74] 82 (666.7) 111 (902.4) 1.35 [1.02:1.80]

Cardiac‡ 307 (2308.3) 85 (670.7) 0.29 [0.23:0.37] 284 (2308.9) 72 (585.4) 0.25 [0.20:0.33]

Cardiac (HF)‡ 174 (1308.3) 38 (299.8) 0.23 [0.16:0.33] 162 (1317.1) 33 (268.3) 0.20 [0.14:0.30]


Mean (SD) 3.1 (1.6) 1.5 (2.0)

3.0 (1.6) 1.5 (2.0)

Median [Q1;Q3]

3 [2;4] 1 [0;2]

3 [2;4] 1 [0;2]

Range 0 to 7 0 to 11

0 to 7 0 to 11



1258 (9458.6) 1090 (8600.3) 0.79 [0.57:1.08] 1102 (8959.3) 924 (7512.2) 0.74 [0.53:1.04]

Non-cardiac‡ 286 (2150.4) 533 (4205.5) 1.60 [0.91:2.84] 209 (1699.2) 465 (3780.5) 1.90 [1.02:3.52]

Cardiac‡ 972 (7308.3) 557 (4394.8) 0.50 [0.34:0.72] 893 (7260.2) 459 (3731.7) 0.44 [0.29:0.65]

123

1 year

pre-MC admission*

(n=133)

1 year

post-MC

discharge*

(n=133)

Incidence rate

ratio [95% CI]

1 year

pre-MC

discharge*

all deaths

removed

(n=123)

1 year

post-MC

discharge*

all deaths

removed

(n=123)

Incidence rate

ratio [95% CI]

Cardiac (HF)‡ 614 (4616.5) 400 (3156.1) 0.56 [0.33:0.95] 552 (4487.8) 330 (2682.9) 0.51 [0.29:0.89]


Mean (SD)

Median [Q1;Q3]

Range

9.5 (17.8)

3 [1;10]

0 to 134

8.2 (15.8)

1 [0;8]

0 to 81

9.0 (17.6)

3 [1;9]

0 to 134

7.5 (15.6)

0 [0;6]

0 to 81


available (%)

399 (98.3%) 204 (99.5%) 361 (98.6%) 182 (99.5%)

Total aggregated hospitalisation

cost

£1,037,636 £757,646 0.64 [0.50:0.82] £931,507 £651,401 0.59 [0.46:0.77]

Non-Cardiac‡ £213,942 £303,130 1.41 [0.90:2.19] £155,405 £263,145 1.62 [1.02:2.58]

Cardiac‡ £823,693 £454,516 0.44 [0.33:0.59] £776,101 £388,257 0.38 [0.28:0.52]

Cardiac (HF)‡ £489,404 £291,233 0.41 [0.27:0.61] £459,897 £257,194 0.36 [0.24:0.55]


Mean (SD) £7802 (£9858) £5697 (£10,669) £7573 (£9676) £5296(£10,849)

Median [Q1;Q3] £4811 [£2449; £8713]

£1404 [£0; £6304] £4811 [£2449; £8706]

£1024 [£0; £5543]

Range £0 to £82,676 £0 to £90,625 £0 to £82,676 £0 to £90,625


124

b) Definition: “X7.05_Procedural.Urgency” of index MC procedure = “2. Urgent” or “3. Emergency” 1 year

pre-MC admission*

(n=17)

1 year

post-MC

discharge*

(n=17)

Incidence rate

ratio [95% CI]

1 year

pre-MC discharge*

all deaths removed

(n=14)

1 year

post-MC

discharge*

all deaths

removed

(n=14)

Incidence rate

ratio [95% CI]




64 (3764.7) 46 (2951.8)

0.79 [0.54:1.16] 52 (3714.3)

38 (2714.3)

0.73 [0.48;1.11]

Non-cardiac‡ 20 (1176.5) 31 (1989.3) 1.75 [1.00:3.09] 16 (1142.9) 25 (1785.7) 1.56 [0.83:2.93]

Cardiac‡ 44 (2588.2) 15 (962.6) 0.37 [0.21:0.67] 36 (2571.4) 13 (928.6) 0.36 [0.19:0.68]

Cardiac (HF)‡ 21 (1235.3) 7 (449.2) 0.36 [0.15:0.86] 15 (1071.4) 5 (357.1) Insufficient data for model


Mean (SD) 3.8 (2.8) 2.7 (3.2)

3.7 (3.1) 2.7 (3.4)

Median [Q1;Q3]

3 [2;5] 1 [0;5]

3 [2;5] 1 [0;5]

Range 0 to 11 0 to 10

0 to 11 0 to 11


(rate per 1000 person years) 377 (22,176.5)

250 (16,042.5)

0.60 [0.30:1.22] 330 (23,571.4)

168 (12,000)

0.47 [0.21:1.07]

Non-cardiac‡ 52 (3058.8) 153 (10,010.5) 1.97 [0.50:7.86] 50 (3571.4) 84 (6000.0) 1.11 [0.30:4.14]

Cardiac‡ 325 (19,117.6) 94 (6032.0) 0.28 [0.10:0.78] 280 (20,000.0) 84 (6000.0) 0.30 [0.08:1.08]

Cardiac (HF)‡ 170 (10,000.0) 49 (3144.3) 0.29 [0.08:1.02] 125 (8928.6) 39 (2785.7) 0.31 [0.06:1.68]


Mean (SD) 22.6 (26.0) 14.7 (25.8) 23.6 (28.3) 12.0 (21.8)

125

1 year

pre-MC admission*

(n=17)

1 year

post-MC

discharge*

(n=17)

Incidence rate

ratio [95% CI]

1 year

pre-MC discharge*

all deaths removed

(n=14)

1 year

post-MC

discharge*

all deaths

removed

(n=14)

Incidence rate

ratio [95% CI]

Median [Q1;Q3] 14 [7;25] 0 [0;14] 15 [6;24] 0 [0;12]

Range 0 to 104 0 to 78 0 to 104 0 to 67


available (%)

64 (100.0%) 45 (97.8%) 52 (100.0%) 37 (97.4%)

Total aggregated hospitalisation cost £229,685 £132,075 0.55 [0.27:1.13] £198,918 £91,859 0.45 [0.19:1.04]

Non-Cardiac‡ £39,319 £76,046 1.69 [0.69:4.17] £33,758 £48,044 1.40 [0.52:3.80]

Cardiac‡ £190,366 £56,029 0.28 [0.11:0.73] £165,160 £43,815 0.23 [0.07:0.79]

Cardiac (HF)‡ £107,918 £28,828 0.27 [0.08:0.86] £83,917 £16,614 0.20 [0.04:0.97]


Mean (SD) £13,511 (£17,353) £7769 (£10,982)

£14,208 (£19,165) £6561 (£9852)

Median [Q1;Q3] £9718 [£4594; £11,988]

£1009 [£0; £12,282]

£9453 [£3955; £13,949] £961 [£0; £11,603]

Range £0 to £73,591 £0 to £32,929 £0 to £73,591 £0 to £32,929


*does not include MitraClip procedural admission

†followed until death or 1 year post-MitraClip discharge


126

Appendix 15 - Hospital resource (HES only) by aetiology of mitral

regurgitation – a) Functional/Ischaemic and b) Degenerative a) Definition: “X5.03_Aetiology.of.Mitral.Regurgitation” recorded at index MC procedure = “1. Functional” or “3. Ischaemic”. [Note 4 procedures did not complete X5.03 datafield].

1 year

pre-MitraClip

admission*

(n=90)

1 year

post- MitraClip

discharge*

(n=90)

Incidence rate

ratio [95% CI]

1 year

pre- MitraClip

discharge*

all deaths removed

(n=82)

1 year

post- MitraClip

discharge*

all deaths

removed

(n=82)

Incidence rate

ratio [95% CI]




301 (3344.4) 151 (1761.2)

0.53 [0.44:0.65] 270 (3292.7)

126 (1536.6)

0.47 [0.38:0.58]

Non-cardiac‡ 73 (811.1) 73 (851.4) 1.10 [0.79:1.52] 62 (756.1) 61 (743.9) 0.98 [0.69:1.40]

Cardiac‡ 228 (2533.3) 78 (909.8) 0.36 [0.28:0.47] 208 (2536.6) 65 (792.7) 0.31 [0.24:0.41]

Cardiac (HF)‡ 124 (1377.8) 35 (408.2) 0.30 [0.20:0.43] 114 (1390.2) 30 (365.9) 0.26 [0.18:0.39]


Mean (SD) 3.3 (1.8) 1.7 (2.2) 3.3 (1.7) 1.5 (2.1)

Median [Q1;Q3] 3 [2;5] 1 [0;2] 3 [2;5] 1 [0;2]

Range 0 to 7 0 to 11 1 to 7 0 to 11



1060 (11,777.8)

769 (8969.3)

0.66 [0.46:0.95] 903 (11,012.2)

586 (7146.3)

0.59 [0.40:0.88]

Non-cardiac‡ 163 (1811.1) 231 (2694.3) 1.14 [0.55:2.37] 113 (1378.0) 144 (1756.1) 1.09 [0.51:2.35]

127

1 year

pre-MitraClip

admission*

(n=90)

1 year

post- MitraClip

discharge*

(n=90)

Incidence rate

ratio [95% CI]

1 year

pre- MitraClip

discharge*

all deaths removed

(n=82)

1 year

post- MitraClip

discharge*

all deaths

removed

(n=82)

Incidence rate

ratio [95% CI]

Cardiac‡ 897 (9966.7) 538 (6275.0) 0.54 [0.35:0.83] 790 (9634.1) 442 (5390.2) 0.49 [0.31:0.78]

Cardiac (HF)‡ 601 (6677.8) 360 (4198.9) 0.57 [0.30:1.08] 508 (6195.1) 292 (3561.0) 0.55 [0.27:1.11]


Mean (SD)

Median [Q1;Q3]

Range

11.8 (20.3)

5 [1; 13]

0 to 134

8.5 (16.0)

1 [0; 8]

0 to 78

11.0 (20.1)

5 [1; 12]

0 to 134

7.1 (14.2)

0 [0; 6.8]

0 to 72]


available (%)

295 (98.0%) 150 (99.3%) 266 (98.5%) 125 (99.2%)

Total aggregated hospitalisation cost £822,057 £585,554 0.62 [0.47:0.83] £723,721 £471,820 0.55 [0.40:0.75]

Non-cardiac‡ £141,826 £171,545 1.13 [0.67:1.89] £98,047 £122,347 1.13 [0.66:1.93]

Cardiac‡ £680,230 £414,009 0.50 [0.36:0.71] £625,674 £349,473 0.44 [0.30:0.64]

Cardiac (HF)‡ £426,525 £263,115 0.46 [0.28:0.75] £386,037 £230,799 0.41 [0.24:0.70]


Mean (SD) £9134 (£11,164) £6506 (£12,019) £8826 (£11,034) £5754 (£12,039)

Median [Q1;Q3] £5664 [£2669; £11,716]

£1615 [£0; £8915]

£5574 [£2669; £10,892]

£1029 [£0; £6295]

Range £0 to £82,676 £0 to £90,625 £603 to £82,676 £0 to £90,625

128

1 year

pre-MitraClip

admission*

(n=90)

1 year

post- MitraClip

discharge*

(n=90)

Incidence rate

ratio [95% CI]

1 year

pre- MitraClip

discharge*

all deaths removed

(n=82)

1 year

post- MitraClip

discharge*

all deaths

removed

(n=82)

Incidence rate

ratio [95% CI]


129

b) Definition: “X5.03_Aetiology.of.Mitral.Regurgitation” recorded at index MC procedure = “2. Degenerative” appearing alone or in any combination [Note 4 procedures did not complete X5.03 datafield].

1 year

pre-MC admission*

(n=56)

1 year

post-MC discharge*

(n=56)

Incidence rate

ratio [95% CI]

1 year

pre-MC discharge*

all deaths removed

(n=52)

1 year

post-MC

discharge*

all deaths removed

(n=52)

Incidence rate

ratio [95% CI]




149 (2660.7) 89 (1673.5)

0.64 [0.49;0.83]

131 (2519.2)

86 (1653.8)

0.66 [0.50;0.86]

Non-cardiac‡ 44 (785.7) 72 (1353.9) 1.81 [1.24:2.65]

35 (673.1) 70 (1346.2) 2.00 [1.33:3.00]

Cardiac‡ 105 (1875.0) 17 (319.7) Insufficient data for model

96 (1846.2) 16 (307.7) Insufficient data for model

Cardiac (HF)‡ 62 (1107.1) 8 (150.4) Insufficient data for model

56 (1076.9) 7 (134.6) Insufficient data for model


Mean (SD)

2.7 (1.6) 1.6 (2.3)

2.5 (1.5) 1.7 (2.3)

Median [Q1;Q3] 2.5 [2;3] 1 [0;2]

2 [2;3] 1 [0;3]

Range

0 to 8 0 to 10

0 to 8 0 to 10


(rate per 1000 person years) 455 (8125.0)

471 (8856.6)

0.85

[0.49:1.47] 409 (7865.4)

413 (7942.3)

0.83 [0.46:1.49]

Non-cardiac‡ 168 (3000.0) 386 (7258.3) 1.96

[0.88:4.37]

139 (2673.1) 338 (6500.0) 2.25 [0.94:5.39]

Cardiac‡ 287 (5125.0) 85 (1598.3) 0.27

[0.14:0.54]

270 (5192.3) 75 (1442.3) 0.26 [0.12:0.53]

Cardiac (HF)‡ 137 (2446.4) 66 (1241.0) 0.44 123 (2365.4) 56 (1076.9) 0.42 [0.17:1.03]

130

1 year

pre-MC admission*

(n=56)

1 year

post-MC discharge*

(n=56)

Incidence rate

ratio [95% CI]

1 year

pre-MC discharge*

all deaths removed

(n=52)

1 year

post-MC

discharge*

all deaths removed

(n=52)

Incidence rate

ratio [95% CI]

[0.20:0.99]


Mean (SD) 8.1 (12.9) 8.4 (17.9) 7.9 (13.3) 7.9 (17.7)

Median [Q1;Q3] 2.5 [1;9] 0 [0;4] 2 [0;8]

0 [0;3]

Range 0 to 54 0 to 81 0 to 54 0 to 81


available (%)

148 (99.3%) 89 (100.0%) 130 (99.2%) 86 (100.0%)

Total aggregated hospitalisation cost £354,048 £258,055 0.65

[0.43:0.97]

£317,001 £229,229 0.63 [0.41:0.96]

Non-Cardiac‡ £106,062 £190,582 1.79

[0.93:3.46]

£86,153 £173,650 2.16 [1.08:4.35]

Cardiac‡ £247,986 £67,473 0.24

[0.14:0.40]

£230,848 £55,580 0.21 [0.13:0.36]

Cardiac (HF)‡ £127,919 £41,621 0.27

[0.14:0.51]

£116,002 £29,728 0.22 [0.11:0.44]


Mean (SD) £6322 (£6661) £4608 (£7998) £6096 (£6768) £4408 (£8023)

Median [Q1;Q3] £3755 [£2521; £7653]

£727 [£0; £4754] £3600 [£2449; £7503]

£670 [£0; £4206)

Range £0 to £33,621 £0 to £33,915 £0 to £33,621 £0 to £33,915

131

1 year

pre-MC admission*

(n=56)

1 year

post-MC discharge*

(n=56)

Incidence rate

ratio [95% CI]

1 year

pre-MC discharge*

all deaths removed

(n=52)

1 year

post-MC

discharge*

all deaths removed

(n=52)

Incidence rate

ratio [95% CI]


*does not include MC procedural admission

†followed until death or 1 year post-MC discharge


NICE OBSERVATIONAL DATA UNIT (ODU) · 2019-04-04 · Providing consultancy and research in health...

Documents

Transcript of NICE OBSERVATIONAL DATA UNIT (ODU) · 2019-04-04 · Providing consultancy and research in health...