A Framework for Leveraging Health Information

Technology (HIT) in Clinical Trial Planning and Execution

15-Nov-2012

Otis JohnsonAssociate Director, Clinical Research

Brendan O’NeillDirector, Clinical Research

OutlineLearning objectivesBusiness problemResponse framework

HIT and EHR as part of the solution Expectations and success

indicatorsLessons learned from early EHR

experiencePerformance Measurement

2

Learning Objectives

Learn how to: Incorporate HIT in clinical trial planning and execution Partner with IT to advance trial operations Include EHR evaluation in standard study feasibility

assessment process Avoid common pitfalls when using EHR information Measure performance against pre-established goals

3

Patient Enrollment Period ExtendedPercent of sites extending patient enrollment period on average

Source: CenterWatch Investigative Global Site Survey 2011

66% of sites extend

enrollment timeline 20-80%

of the time

4

Why Do we Care? Problem

Clinical trial recruitment often slow and unpredictable

Result Unmet expectations Delayed filings Lost revenue Damaged relationships

Value of 1 day in drug development

$37,000 in operational cost $1.1 M in prescription revenue

Source: Gartner. 2007. Case study: Boosting the predictability of clinical trial performance.

NA

SA

EU

ASIA

5

Traditional Response to Slow EnrollmentTypical response to

under-enrollment Add countries Add sites Revise protocol Implement costly

remediation programs Recruitment campaigns;

involve vendors Rejuvenation workshops

DisruptiveCostly

Not sustainable

Did we set the right

expectations?

Can we accurately

predict trial performance?Can we do a better job setting expectations?

6

Opportunity for Improved Response Root cause analysis Insights into trial

performance Factors that impact trial

performance Organizational culture

that supports innovation and sound change

Key finding Trials often designed

without consideration of real world clinical data

Can we leverage EHR in planning and execution of clinical trials?

7

Response Framework Standard feasibility assessment process Incorporation of EHR analyses in study planning

and execution Allocation of study optimization resources Enrollment modeling technology

8

FTEs Dedicated to Addressing ProblemGTO (Global Trial Optimization)

Optimize clinical trial feasibility and execution through data-driven analyses

Scope and Objectives Operational insight into trial design Geographic input Enrollment modeling R & R strategies and tactics

Data resources

Analytical tools

Technical expertise

Data analyses

9

Understanding Trial Performance- Factors Impacting Trial Enrollment

Screening c Rate

Screen Failure

Site Initiation

Period

Site Failure

Site Enrollment

Capacity

10

Extensive Research- Data Sources and Types

ProprietaryCTMSIRT/IVRS/IWRS

ProprietaryCTMSIRT/IVRS/IWRS

Public Clinicaltrials.govLiterature reports

Public Clinicaltrials.govLiterature reports

Purchased Industry consortia: CMR, KMR, DecisionViewAggregators: TrialTrove, SiteTrove, BioPharm Clinical, Clinical Trials Insight, etc.

Purchased Industry consortia: CMR, KMR, DecisionViewAggregators: TrialTrove, SiteTrove, BioPharm Clinical, Clinical Trials Insight, etc.

11

How We Gain Insights From Internal Data

PA2FPE

AREARGAUSAUTBELBGRBRACANCHECHLCHNCOLCRICZEDEUDNKECUESPESTFIN

FRAGBRGRCGTMHKGHRVHUNINDIRLISLISRITA

JORJPN

KORLBNLTULVA

MEXMYSNLDNORNZL

PANPERPHLPOLPRTROMROURUSSAUSGPSRBSVNSWETHATUR

TWNUSAVENZAF

0 200 400 600 800

12

Early Assumption-Based Enrollment Model

0

50

100

150

200

250

300

350

400

450

500

0 5 10 15 20 25 30 35 40 450

50

100

150

200

250

300

350

400

450

500

0 5 10 15 20 25 30 35 40 45

0

50

100

150

200

250

300

350

400

450

500

0 5 10 15 20 25 30 35 40 45

0

50

100

150

200

250

300

350

400

450

500

0 5 10 15 20 25 30 35 40 45

No. of R

andomized P

atients

Months After Protocol Approval

13

In-life Enrollment ModelingSite Ready Planned

SiteReady

Actual & Projected

Screening and Randomization

Planned

Screening and Randomization

Actual & Projected

14

Understanding Trial Performance- Factors Impacting Trial Enrollment

Screening c Rate

Screen Failure

Site Initiation

Period

Site Failure

Site Enrollment

Capacity

15

Page | 16Page | 16Page | 16

Leveraging EMR for Secondary Use

Determine the value of EHR-enabled providers in facilitating improved clinical trial execution Analysis of EHR to determine patient availability Identification of provider-affiliated investigators Randomization of patients

16

Disease EtiologyDisease Etiology

Data Types Desired for Clinical Research

• Family• Illness• Social/

psychological• Allergies• Medication• Economic• Exposure• Smoking/

Alcohol• Diet, Etc.

• Family• Illness• Social/

psychological• Allergies• Medication• Economic• Exposure• Smoking/

Alcohol• Diet, Etc.

Symptoms• Onset• Chronology• Location• Radiation• Severity• Duration• Context• Modifying

factors (+/-) Etc.

Symptoms• Onset• Chronology• Location• Radiation• Severity• Duration• Context• Modifying

factors (+/-) Etc.

Patient Patient HistoriesHistoriesPatient Patient

HistoriesHistories HPI/DD*HPI/DD*HPI/DD*HPI/DD*

• Physical Exam

• ICD Primary Diagnoses + comorbidity

• Pharmacy• Labs• Imaging

remarks Etc.

• Physical Exam

• ICD Primary Diagnoses + comorbidity

• Pharmacy• Labs• Imaging

remarks Etc.

Physical Physical DiagnosisDiagnosisPhysical Physical

DiagnosisDiagnosis

• CPT(s)• Assessments• Allergy• Adverse events• Orders

o Lab resultso Rx dispensedo Radiologyo Supplies

Etc.

• CPT(s)• Assessments• Allergy• Adverse events• Orders

o Lab resultso Rx dispensedo Radiologyo Supplies

Etc.

TherapyTherapyTherapyTherapy

• Patient/disease progress

• Side effects• Adverse

events• Outcomes• Compliance• Adherence• Quality of life,

Etc.

• Patient/disease progress

• Side effects• Adverse

events• Outcomes• Compliance• Adherence• Quality of life,

Etc.

OutcomesOutcomesOutcomesOutcomes

*HPI/DD = History of Present Illness / Differential Diagnosis 17

When is EMR Analysis Needed?

Wk-16Wk -12

Wk -11

Wk -10

Wk -09

Wk -08

Wk -07

Wk -06

Wk -05

Wk -04

Wk -03

Wk -02

Wk -01

PA FPE LPI LPLV

Project Initiation

Team Meeting

Team Meeting

Review Team Meeting

PA

Protocol Operational Feasibility

Protocol Concept Available

Protocol Operational Feasibility to Team

Input into Review

InitiateVendors

as needed

Input Feasibility Report Addendum, as

needed

18

Does EHR Have Data Needed?

19

What Insights Can you Obtain from EHR Analysis?

20

Potential Patients Identified in Last Year

Criteria Pts Comments

Diagnosed with disease and met age range requirement 3085 Matched selected ICD9 codes

Seen in last year 1316 Patient considered active

# of the above who also have their most

recent A1C < 11 1125Several patients with no A1C data

in last year# of the above who also pass exclusion

criteria #1 819 See source for details.# of the above who also pass exclusion

criteria #2 370 See source for details.# of the above who also pass exclusion

criteria #3 338 See source for details.

Acceptable BMI 300 See source for details.21

Reached 40 of 300 Potential Patients

Pre-screening Activity

No. of Patients Remaining

Patients approached 40

Found ineligible upon contact

19

Refused to participate 4

Did not return phone call 4

Pre-screened and ready for further processing 13

22

Success MeasuresSuccess Measures EHR-enabled sites vs. traditional

sites Enrollment rate, enrollment

duration, data quality, site startup dynamics, percent screen failure, retention, etc.

Comparison to industry benchmarks

Ability to flag enrollment challenges

Expectations Protocol Refinement Investigator Identification Patient Recruitment

23

EHR Criteria Assessed: Protocol 1

Protocol 1 Screen Failure Reasons

0

50

100

150

200

250

300

350

Incl1

Incl2

Incl3

Incl4

Incl5

Incl6

Incl7

Incl8

Incl9

Excl1

Excl2

Excl3

Excl4

Excl5

Excl6

Excl7

Excl8

Excl9

Excl10

Excl11

Excl12

Excl13

Excl14

Excl15

Excl16

Excl17

Excl18

Excl19

Excl20

Excl21

Excl22

Excl23

Excl24

Excl25

Excl26

Excl27

Excl28

Incl10V2

Incl11V2

Incl12V2

Excl29V2

Excl30V2

Excl30V2

Incl13V3

Incl14V3

Incl15V3

Incl16V3

Incl17V3

Excl32V3

Excl33V3

Excl34V3

Reason

No.

of P

atie

nts

•EHR does not always have all information needed for clinical trial decision making

Risk: Inaccurate conclusion

24

EHR vs. Actual Screen Failure Reasons

Protocol 1 Screen Failures: EHR Assessment Vs Actual

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

40.0%

45.0%

50.0%

ICS/LABA orSABA

Smokers Hypertension Recent C linicalTrial

Lung Disease Renal orHepaticDisease

LabAbnormalities

OtherMedications

Screen Failure Reason

Per

cen

tag

e o

f A

sses

sed

Scr

een

F

ailu

res

EMR Actual

Some parameters change over time

25

Actual Screen Failures as a Proportion of all Patients Screened

Frequency of Inclusion and Exclusion Criteria in Protocol 2

0

1

2

3

4

5

6

7

8

9

Incl

1In

cl2In

cl 3

Incl

4In

cl 5

Incl

6In

cl 7

Excl 1

Excl 2

Excl 3

Excl 4

Excl 5

Excl 6

Excl 7

Excl 8

Excl 9

Excl 1

0

Excl 1

1

Excl 1

2

Excl 1

3

Excl 1

4

Excl 1

5

Excl 1

6

Excl 1

7

Excl 1

8

Excl 1

9

Excl 2

0

Excl 2

1

Excl 2

2

Excl 2

3

Exclu

24

Excl 2

5

Inclusion or Exclusion Criteria

Nu

mb

er o

f P

atie

nts

Exc

lud

ed

•May find what’s needed in certain disease areas•Know what you need

26

Lessons Learned EHR Analysis

Provides information on number of patients in EHR database who match specific inclusion and exclusion criteria

Limitations:Can assess only criteria typically captured in medical

recordsDoes not predict screen failure ratio for a trialUnsure of the relationship of EHR database to prevalence in

the general population. Unsure of applicability of data on a subset of the US

population to the rest of the world. Study Initiation Process must be optimized in

parallel Accelerate budget and contract negotiations Prenegotiate contract language and budget elements

27

Results

28

Enrollment Period PredictionGTO-Supported Trials (Enrollment Period)

222018161412108642

16

14

12

10

8

6

4

2

0

Study

FPE-

LPE

(Month

s)

FPE-LPE-PredictedFPE-LPE-Actual

Variable

Predicted Vs. Actual Enrollment Period

Correlation Coefficient = 0.829

Enrollment Period Prediction

Difference of 1 month

FPE-LPE-ActualFPE-LPE-Predicted

16

14

12

10

8

6

4

2

0

Enro

llment

Period (

Month

s)

Predicted vs. Actual FPE-LPE

30

Randomization RateTrials Achieving LPE Since 2009

Correlation Coefficient > 0.9

Predicted vs. Actual randomization Rate

0

5

10

15

20

25

30

Study

Ran

do

miz

ati

on

Rate

Predicted

Actual

31

50454035302520151051

0.3

0.2

0.1

0.0

-0.1

-0.2

-0.3

-0.4

Index

Diffe

rence

0.1

-0.2

Predicted vs. Actual SFR for Studies With PA in 2007 and 2008

Had multiple amendments to facilitate better enrollment. For example, several

incl/excl criteria removed.

Restrictive eligibility criteria listed as challenge. Protocol clarification letter needed.

Study cancelled during enrollment

period.

Screen Failure – Prior Performance

50% of studies within variance32

Percent Screen Failure for Completed Trials

24222018161412108642

0.1

0.0

-0.1

-0.2

-0.3

Study

SFR

-Diff

0.1

-0.1

Predicted vs. Actual SFR for Studies Completed Since 2009

80% of studies within 10% of prediction, N = 25 trials 33