Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4...

28
Ian Creek Head of Projects Dr Deborah Neal Training & evaluation lead Symphony Vanguard Programme Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

Transcript of Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4...

Page 2: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

A little about me…

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

He came from the dark side…. or some people call it a ‘Commercial background’

Page 3: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

A little about me…

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

• Currently working in south Somerset on the NHSE Vanguard programme

• Joined the NHS 3 years ago after 12 years in marketing

• Spoilt in the past by great data and free flowing access

• A data lover not hater!

Page 4: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Background

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

• Changing demographic• Financial & workforce challenges• Need to change to less reactive, acute hospital

based system• Widespread stakeholder consultation –

patient, staff, carers• Symphony data set – includes social care,

primary, secondary, community & mental health data.

Page 5: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Proportion of the population

aged 65+ by LSOA - 2033

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

Page 6: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

0

1

2

3

4

5

6

7

8

£m

Gain share

CC: otherCC: IPMH: I/PMH: O/PSC: otherSC: ResiSC:Home+Day

AC: I/P urgent

AC: I/P elect

AC: O/PAC: A&EPrimary care

GPPrescribing

Socialcare

Mentalhealth

Acutecare

List size 6,380 4,340 660 1,990 630 490 70 30 130 120 10 30 1,010

Cost per patient

£160 £210 £140 £340 £1,940 £2,770 £5,100 £12,500 £2,550 £2,230 £20,600 £9,790 £250

Communitycare

Pote

nti

al t

o b

e g

ain

sh

ared

Pre

scri

bin

g2

Total: £1.02m £0.93m £3.35m £1.02m £0.48m £0.58m £7.38m

1. Mean figures averaging across 19 South Somerset GP practices.2. Prescribing cost is extrapolated from Mar 2015 (HSCIC), with prescribing list for 2013–14 (Symphony data)

Distribution of spending across a typical GP Practice

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

Page 7: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized
Page 8: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized
Page 9: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Our new care model and

extended teams

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

Page 10: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Brief overview of the Symphony care model

The symphony new care model was developed following consultation with

patients, carers and staff and using learning from research conducted and

established care models from other countries – namely IORA health, CareMore

and Alignment Health in the USA but also work in Jönköping in Sweden. The

model is closely aligned to the NHS England General Practice Forward View

and includes but is not limited to:

• Complex care for the highest need patient cohort

• Health Coaches – as a new role into Primary care

• Pharmacists, MSK practitioners and Mental Health Link Workers

• New data flows and data analysis/reporting tools

• MDT meetings and development of the huddle approach

• Patient stratification and identification

• An ethos of person centred, coordinated care, self-care and prevention

Page 11: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Population segments Cost breakdown Care models

Complex patients with many conditionsHigh Cost (over £7k/yr)

Person-centred, holistic coordinated care

→ Complex Care (CC)

Less complex patientswith fewer conditionsModerate cost (£1-7k/yr)

Proactive, person centredcondition management

→ Enhanced Primary Care (EPC)

Mainly healthy patientsLow cost (under £1k/yr)

Efficient primary care, proactive health and well-being services

Population cost pyramid, South Somerset

78%~90k

18%~20k

4%~5k

~15%~£20m

~35%~£55m

~50%~£75m

Previously 85% of resources were used by ~20% of the population, the challenge therefore was to develop new care models that deliver high quality, effective, cost

efficient care for all of the population.

Source: South Somerset Symphony project data 12/13, Oliver Wyman analysis

Note: 1 Community service activity (e.g. district nursing) data not allocated to individual patients, therefore not included here

~£150m total cost; Primary, secondary, community1, mental health and social care

The top 4% of population drive 50% of the cost; the top 22% drive 85% of the cost

Page 12: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Our new care

model and

extended

teams

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

Complex Care

Health Coaching

Specialist nursing

Virtual clinics

MSK

Exte

nd

ed t

eam

Mental Health

QIRapid

access to advice

Page 13: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Numbers through the New

Care Model

• The complex care team has worked with 1,030 individuals

• 10,848 people have had contact with a health coach.

• There are also knock on benefits for the remainder of the patients in a practice, even where there is no health coach or complex care involvement, through the greater time and proactive approach which the practice is able to take.

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

Page 14: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Data, data, data…

Local data collection by practices & YDH

External evaluation partners – SWAHSN, SWCSU, York university

Three main strands to the evaluation

• Quantitative

• Qualitative

• Econometric

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

Page 15: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

Data, data, data…

• Quantitative Dashboard: PAM, A&E attendances, Emergency Admissions, Emergency Bed Days and Outpatients.

• Qualitative: Views of patients, Health coaches, GPs and Practice Mangers.

• Econometric: ‘difference in differences’ analysis using Symphony dataset.

Visit: www.symphonyintegratedhealthcare.com Follow us @SymphonyProj

Page 16: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Year to date compared to the previous year, based on CSU data for the South Somerset registered population at all providers

• Non-elective overnight admissions (non-ambulatory care) are down 7.5%

• Non-elective bed days are down 15.2%

• Excess bed days are down 48.9%

• In addition, a year on year historic trend of an increase in non-elective admissions of 4.5% has been avoided, giving a net benefit of 12% compared to the models not being in place.

Page 17: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Impact on Emergency Admissions

Page 18: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Impact on Non-Elective Bed days

Page 19: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Change in PAM scores

• Health coaches and the complex care team record PAM scores at the start of an intervention and then repeat the score 6 months later

• More activated patients are more likely to adhere to treatment protocols, to eat healthily and to have fewer hospital admissions and A&E attendances.

Page 20: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Evaluation of impact of health coach role

Admission avoidance

• Helped to avoid demand on hospital services

• Improved communicationbetween primary and secondary care

Health outcomes

• Supported patient weight loss

• Supported diabetes self-management with reductions in HbA1c level, due to support of health coaches to attend exercise and diet support.

• Confidence to self-manage improved

• Positive general lifestyle change

• Reduction in social isolation

• Patient resilience improved

Page 21: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Evaluation of impact of health coach role

Impact on primary care

• Allowed GPs to change their appointment structure with longer appointments

• Helped to improve GPs’ working day & reduce GP stress

• Could repeat/relay GP advice to patients & support continuity of care

• Could engage in other work to support the practice

• Contributed to care plan development

• Prepared to engage with ‘difficult cases’

• Act as the GPs eyes and ears

Patient experience

• Carers felt supported

• Reduction in patients bouncing around the system

• Positive and tailored outcomes

• Identifying unseen problems (detective role)

• Health coaches easy to access and flexible providing multiple roles; Signposting, Empowering, Liaising, Activating, Persuading and Patient advocate.

Page 22: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Matching process - CCT

• Data Preparation• Identify patients in data every year 2012/13 to

2015/16, i.e. 4 years

• Split CCT cases into 6m periods by enrolment date: 5 cohorts

• Generate covariates

• Two types of matching• Matching by propensity score

• Nearest neighbour, no replacement

• Coarsened exact Matching

Page 23: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

20132012 2014 2015 2016 2017

MAY 2015(cohort midpoint)

1

COHORT 122-Feb-15 to 22-Aug-15

MAY 2012

Count of emergency admissions each year

y3 y2 y1

Page 24: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

20132012 2014 2015 2016 2017

MAY 2015(cohort midpoint)

MAY 2012

1

COHORT 122-Feb-15 to 22-Aug-15

Count of emergency admissions each year

y3 y2 y1

2

COHORT 222-Aug-15 to 22-Nov-15

Count of emergency admissions each year

y3 y2 y1

Page 25: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Covariates used for matching cases & controls

• Utilisation• 3 x lagged years of emergency admission counts• Readmissions for LTC• Risk ratio

• Clinical• Count of 8 comorbidities used for eligibility

assessment

• Demographics• 4 age groups (<18s are excluded)• Deprivation quintiles• Gender

Page 26: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Balance graphs – by enrolment cohort

-100 0 100200300Standardized % bias across covariates

IMD_quint5male

IMD_quint3agecat3

IMD_quint4IMD_quint2

agecat5agecat4

ac_ip_cnt_lag3ac_ip_cnt_lag2

readmac_ip_cnt_lag1

rratiosumcond8

-50 0 50100150200Standardized % bias across covariates

IMD_quint5IMD_quint3

agecat3IMD_quint4

agecat4IMD_quint2

malereadm

ac_ip_cnt_lag3ac_ip_cnt_lag2ac_ip_cnt_lag1

agecat5rratio

sumcond8

0 50 100150200Standardized % bias across covariates

IMD_quint4IMD_quint5

agecat3IMD_quint3

maleIMD_quint2

agecat4ac_ip_cnt_lag2

agecat5readm

ac_ip_cnt_lag3ac_ip_cnt_lag1

sumcond8rratio

-50 0 50 100150Standardized % bias across covariates

agecat3IMD_quint5IMD_quint4

maleIMD_quint3IMD_quint2

agecat4readm

ac_ip_cnt_lag3ac_ip_cnt_lag2

agecat5ac_ip_cnt_lag1

sumcond8rratio

-50 0 50 100 150Standardized % bias across covariates

agecat3IMD_quint5IMD_quint4

maleIMD_quint3IMD_quint2

agecat4readm

ac_ip_cnt_lag3ac_ip_cnt_lag2

agecat5rratio

sumcond8ac_ip_cnt_lag1

From cohort 1 (top left) across then down to 5 (bottom centre)

Page 27: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Challenges

• Information Governance

• Challenges in getting an appropriate not just accurately matched cohort

• Patterns of utilisation rather than average utilisation in previous year.

• Trigger for intervention is often a change in social context. Limited social care data has been available as yet to be used for matching purposes due to IG.

• Low patient activation impacts significantly on utilisation of health and social care services .There is PAM data for the cohort group but very limited PAM data for controls.

Page 28: Ian Creek - The South West Academic Health Science Network · IMD_quint2 agecat5 agecat4 ac_ip_cnt_lag3 ac_ip_cnt_lag2 readm ac_ip_cnt_lag1 rratio sumcond8-50 0 50 100150200 Standardized

@SymphonyProj

Questions

• ?