Comparative Effectiveness: p New Initiatives from AHRQ Michael - - PowerPoint PPT Presentation

Comparative Effectiveness: p

New Initiatives from AHRQ

Michael Fischer, M.D., M.S. , , Division of Pharmacoepidemiology & Pharmacoeconomics Brigham & Women’s Hospital Harvard Medical School

Spo tte d a t the Co me dy Ce ntra l ra lly in DC:

Changes in the era of health care reform Changes in the era of health care reform

Payment system reform Medical homes Accountable care organizations Accountable care organizations Many other new changes Comparative effectiveness p

3

Comparative effectiveness Comparative effectiveness

What comparative effectiveness is: What comparative effectiveness is not:

4

Comparative effectiveness Comparative effectiveness

What comparative effectiveness is: Head‐to‐head evaluation of real therapeutic choices Aimed at providing clinically relevant data Aimed at providing clinically relevant data What comparative effectiveness is not: Cost‐effectiveness analysis Aimed at denying care to patients y g p

5

What does the new emphasis on comparative What does the new emphasis on comparative effectiveness mean?

Increased role for AHRQ More research to generate clinical CE data DEcIDE network other efforts DEcIDE network, other efforts New areas of emphasis for CE Translating CE into clinical practice Expanding CE beyond traditional clinical studies

6

Expanding CE beyond traditional clinical studies

6

Tools for improving medication use

Changing prescribing habits/patterns Creating incentives through payment policy Creating incentives through payment policy

7

Tools for improving medication use

Changing prescribing habits/patterns Academic detailing Creating incentives through payment policy Creating incentives through payment policy Prior authorization and other reimbursement policy tools tools

8

Key questions for an intervention or policy

Is it effective? Is it cost effective? Is it cost‐effective? Can it be implemented sustainably?

9

A tale of two approaches

Academic detailing Evidence of effectiveness and cost‐effectiveness Limited implementation Limited implementation Prior authorization/payment policy Widespread implementation Limited evidence of effectiveness or cost‐

10

Limited evidence of effectiveness or cost effectiveness

10

Turning comparative effectiveness into practice

New AHRQ program: Innovative Adaptation and Dissemination of AHRQ Innovative Adaptation and Dissemination of AHRQ Comparative Effectiveness Research Products iADAPT iADAPT

11 11

iADAPT background

AHRQ supported comparative effectiveness AHRQ supported comparative effectiveness Original research Evidence summaries Clinician guides, patient education materials Clinician guides, patient education materials BUT: Not having an impact on health care system or

12

g p y medical practice

12

Goals of iADAPT

Stimulate the development of approaches to increase the use and application of comparative effectiveness findings d f h h d k h ld identify the right settings and stakeholders explore different methods of adaptation disseminate evidence to improve care evaluate the interventions implemented

13 13

iADAPT‐funded projects iADAPT funded projects

Informatics interventions Adaptation of consumer guides Interventions in nursing home settings Interventions in nursing home settings Outreach to vulnerable populations Our project: Academic detailing p j g

14 14

A National Academic Detailing Resource to Adapt and Disseminate CER Findings (NaRCAD)

What is academic detailing? What will be provided by this new national resource?

15 15

Academic detailing Academic detailing

Pharmaco‐epistemology

How do we know what we know about drugs?

Educating physicians g p y

Understanding effective learning

Academic detailing Academic detailing

History and research Current programs NaRCAD

16 16

The problem The problem

Limited data when drugs first approved with limited relevance to many patients Physician data overload hundreds of important drug‐related papers published each month Imbalanced information Need for non‐product‐driven overviews delivered in a clinically relevant, user‐friendly way

17 17

Clinical trials Clinical trials

Usually don’t provide head‐to‐head comparative data about relevant Rx choices relevant Rx choices A drug that achieved a surrogate outcome may not produce expected clinical benefit expected clinical benefit e.g., Avandia (rosiglitazone) and M.I. d d ff l k l Unanticipated adverse effects are likely e.g., Vioxx (rofecoxib) Use differs in trials vs. actual practice Efficacy vs. effectiveness

18

Efficacy vs. effectiveness

18

Information overload

Dozens of biomedical journals Physician time constraints Physician time constraints Systematic overviews

cover selected fields but cover selected fields, but… are lengthy, abstruse hard to wade through may not be recently updated

y y p

Some important findings not in journals

FDA alerts, ‘Dear Doctor’ letters important trial data presented at clinical meetings unpublished results

19 19

Information imbalance

Trial design, promotion, CME favor use of new, costly drugs Needed head‐to‐head comparative studies often not

performed performed

Most drug information comes from industry

g y

$30 billion per year on promotion 2/3rds of continuing medical education is industry‐funded

NB: this is in the process of changing

20 20

I d t t d i f ti Industry‐generated information

A dominant source of drug information A dominant source of drug information

• ften only available source for new products

Main purpose is to increase sales, so promotes positives not

negatives

Selective about which comparisons are presented

21 21

Clinical trial: Drug A vs. Drug B for reducing blood

sugar levels in diabetes

After 6 months of therapy, Drug A was better than

Drug B g

After 12‐48 months, no difference between the 2

drugs drugs

What was the message delivered to doctors by the

industry?

22 22

• Drug A vs. Drug B for reducing HbA1c in diabetes
• After 6 months of therapy, Drug A was significantly

better than Drug B g

• After 12‐48 months, no difference between the 2

drugs drugs

23 23

Does promotion work? Does promotion work?

Yes! Clear evidence that sales reps and samples change

prescribing prescribing

Social science literature shows the persuasive effects of

l ti hi ift relationships, gifts

symbolic power of even small gifts reciprocal obligation

p g

Marketing promotes costliest products

24 24

The rationale for academic detailing

• FDA has limited data when drugs are first

d

The rationale for academic detailing

approved

– with limited relevance to many patients

• physician data overload
• physician data overload

– hundreds of important drug‐related papers are published each month

• imbalanced communication

– manufacturers provide much of the information

• the need for non‐product‐driven overviews

– delivered in a relevant, user‐friendly way

25 25

The goal of academic detailing The goal of academic detailing

to close the gap g p between the best available evidence and actual prescribing practice, p g p so that each prescription is based

• nly on the most current and accurate

evidence about efficacy, safety, and cost‐effectiveness.

26 26

Two different worlds

• Academia:

– MD comes to us

• Drug industry:

– Go to MD MD comes to us – Didactic – Content ornate, not – Interactive – Content is simple, i h f d l clinically relevant – Visually boring N id f MD’ straightforward, relevant – Excellent graphics – MD‐specific data informs – No idea of MD’s perspective – Evaluation: minimal MD specific data informs discussion – Outcome evaluated, drives salary Evaluation: minimal – Goal: ???? salary – Goal: behavior change

27 27

Th ti l f d i d t ili The rationale for academic detailing

Academic detailing

Medical school

g

Medical school faculty: tr uste d sour c e s Dr ug maker s: gr e at

• f c linic al

infor mation gr e at c ommunic ator s

28 28

Academic detailing

Synthesizes up‐to‐date evidence about comparative efficacy,

y p p y, safety, and cost‐effectiveness of commonly used drugs

Content independently created by independent clinical experts

and practitioners

Academic detailers provide information interactively, in

physicians’ own offices

A time‐efficient way to keep up with new findings

29 29

Th t t f d i d t ili

Well trained clinicians (pharm RN MD) visit

The content of academic detailing

Well trained clinicians (pharm, RN, MD) visit

prescribers in their offices and offer a service that provides independent, unbiased, non‐ p p , , commercial, non‐product‐driven, evidence‐based information about the comparative benefit, safety, and cost‐ effectiveness of drugs used for common clinical problems problems.

30 30

The method of academic detailing

Information is provided interactively

The method of academic detailing

generally in the doctor’s own office

This enables the educator to

understand where the MD is coming from in terms of understand where the MD is coming from in terms of

knowledge, attitudes, behavior

modify the presentation appropriately keep the prescriber engaged

The visit ends with specific practice‐change

recommendations recommendations.

Over time, the relationship becomes more trusted

and useful.

31 31

and useful.

What academic detailing is not

• memos or brochures provided through

What academic detailing is not p g the mail

• lectures delivered in the doctor’s office

lectures delivered in the doctor s office

• about policing

i il b d i

• primarily about cost reduction
• “counter‐detailing”
• effective drugs should be used
• effective drugs will be cost effective

32 32

g

The history of academic detailing The history of academic detailing

• Developed in early 1980’s

– “un‐ads” for physicians with clinical background and specific prescribing recommendations specific prescribing recommendations – patient educational materials

• Effective from the start

– 92% MD acceptance rate from ‘cold calls’ to physicians – Significant 14% reduction in inappropriate prescribing

A orn & Soumerai NEJM 1983 Avorn & Soumerai, NEJM 1983

33 33

Current status of the evidence

• A mass of AD literature has developed in last 25 years

A l t ti i i bi d 6 t di d

• A large systematic review in 2007 combined 69 studies and

confirmed efficacy of AD

• O’Brien MA, Rogers S, et al. Educational outreach visits:

effects on professional practice and health care outcomes effects on professional practice and health care outcomes. Cochrane, Database of Systematic Reviews 2007, Issue 4

Eff i i i h li f i

• Effectiveness varies with quality of execution

– like brain surgery

34 34

Is it cost‐effective? Is it cost effective?

Economic analysis of the original 1983 research which coined the

term ‘academic detailing’ found that for each $1 spent on their academic detailing program $2 was saved in Medicaid drug academic detailing program $2 was saved in Medicaid drug expenditures.1

When evaluating global primary care clinical practice changes in a

large British study of academic detailing, cost effectiveness was ill d d h l d ll ff i still demonstrated even where only modest overall effect sizes were observed.2

Independent economic study of an Australian DATIS service‐

• riented academic detailing program showed that between $5 and

g p g 5 $6.50 of direct health expenditure was saved for each $1 spent delivering the program.3

1 So ume ra i SB Avo rn J E c o no mic a nd po lic y a na lysis o f unive rsity b a se d drug

• 1. So ume ra i SB, Avo rn J. E

c o no mic a nd po lic y a na lysis o f unive rsity-b a se d drug "de ta iling ". Me d Ca re 1986;24(4):313-31.

• 2. Ma so n J, F

re e ma ntle N, Na za re th I , E c c le s M, Ha ine s A, Drummo nd M. Whe n is it c o st- e ffe c tive to c ha ng e the b e ha vio r o f he a lth pro fe ssio na ls? JAMA 2001;286(23):2988-92.

• 3. Co o pe rs & L

yb ra nd Co nsulta nts. Drug a nd T he ra pe utic s I nfo rma tio n Se rvic e - Upda te o f the e c o no mic e va lua tio n o f the NSAI D pro je c t I n: Ma y F W Ro we tt D e ds DAT

I S pro gre ss re po rt to the

35 35 e c o no mic e va lua tio n o f the NSAI D pro je c t. I n: Ma y F W, Ro we tt D, e ds. DAT

I S pro gre ss re po rt to the De partme nt o f He alth and F amily S e rvic e s Oc to be r to Marc h 1995-96. Canbe rra: Australian Co mmo nwe alth De partme nt o f He alth and F amily S e rvic e s 1996. .

P bj ti Program objectives

Optimize therapy for patients

Optimize therapy for patients

Safety, efficacy, and cost of therapeutic options

Provide evidence‐based resources Facilitate good therapeutic decision‐making by physicians Establish a viable, sustainable educational model

b d b d

Lower costs by providing better medicine Provide a service to prescribers Emphasis on quality of care, not just cost

36 36

E i ti Existing programs

Multiple states

Multiple states

PA, DC, MA

Independent Drug Information Service (iDiS)

p g ( )

ME, SC, VT, others

Different funding models

g

CDC chronic disease prevention funds User fee on private sector detailers

p

Lottery funds

37 37

E t bli hi Establishing a program

Needs assessment

Needs assessment

Patient populations affected Clinicians for likely outreach

y

Clarity about goals of program What academic detailing is and is not

Establish a source of funding Develop a management structure

p g

Identify clinical areas to be targeted

38 38

D l i th i t ti Developing the intervention

Training personnel

Training personnel

Social marketing techniques Specific topic content

p p

Developing materials

Must be based on solid clinical evidence Literature summaries Un‐advertisements Patient‐directed materials

39 39

G tti i th fi ld Getting in the field

Access clinicians

Access clinicians

Harder than it seems Problem of “no detailers” policies

p

Being part of a larger effort can help

Massachusetts: collaboration with CDC

40 40

A ff t Assess program effects

Are we meeting the prescriber’s needs?

Are we meeting the prescriber s needs?

Qualitative

Experiences of providers

Experiences of providers

Process measures of success getting into offices

Quantitative: think carefully about metrics

Q y

Often suggest less medications, or less costly

medications

But sometimes:

Suggest more medications or increased awareness,

41

Suggest more medications or increased awareness, testing, treatment

41

N RCAD 5 t NaRCAD: 5 components

Establish a network of interested programs

Establish a network of interested programs

Provide training in the techniques and content of

AD AD

Adapt CE materials into effective AD tools

p

Evaluate outcomes of AD programs Communicate findings and lessons to promote

improved AD practice over time

42 42

E t bli h t k f i t t d Establish a network of interested programs

Identify organizations that could benefit from AD

Identify organizations that could benefit from AD

Medicaid programs

S

State coverage programs Community health organizations

P i t i

Private insurers

Provide input and advice on establishing programs

Funding sources, establishing collaborations Identifying clinicians

43

Understanding goals of program

43

P id i i i h h i d f Provide training in the techniques and content of academic detailing

Host multiple training sessions

Ed i i i i l f i l k i d

Education in principles of social marketing and

persuasive communication

Role play with clinicians to master techniques Role play with clinicians to master techniques

Cost of developing sessions covered by AHRQ

Will be offered 4‐6 times in next 3 years Can train new or existing detailers

Al l f h i

44

Also relevant for pharmacists, managers

44

Ad i ff i i l i Adapt comparative effectiveness materials into effective academic detailing tools

Rigorous clinical content is the baseline Develop materials for doctors and patients

Concise Concise Visually engaging Designed for detailers to use

Examples: www.rxfacts.org

45

Examples: www.rxfacts.org

45

E l f d i Evaluate outcomes of academic detailing programs

Qualitative evaluation of implementation

processes

Surveys and interviews with detailers and

y physicians

Quantitative evaluation of prescribing patterns Quantitative evaluation of prescribing patterns Repeated evaluations over time

46 46

C i t fi di d l Communicate findings and lessons

Build on the network of programs

Build on the network of programs

Foster discussion of different approaches Provide feedback on what works

D l i i d l f

Develop new innovations and elements for

improved academic detailing

47 47

The future of academic detailing: NaRCAD

• Opportunities to establish new AD

The future of academic detailing: NaRCAD programs

– subsidized by iADAPT support for NaRCAD:

– establishing programs – developing materials – training detailers training detailers – evaluating outcomes – network collaboration to improve quality and id if b i identify best practices

Remember the evidence: Effectiveness depends on

48 48

Remember the evidence: Effectiveness depends on implementing high‐quality academic detailing

Changing gears

So far: Efforts to translate existing comparative effectiveness findings into practice Next: Comparative effectiveness studies in different Next: Comparative effectiveness studies in different settings

49 49

Expanding comparative effectiveness beyond Expanding comparative effectiveness beyond traditional clinical studies

Comparative effectiveness of delivery systems Public and private sector try new ways to deliver care Public and private sector try new ways to deliver care Many ideas theoretical, or with limited impact Need for rigorous evaluation of which approaches are most effective Both evaluation and demonstration projects

50 50

Comparative effectiveness delivery system grants Comparative effectiveness delivery system grants (AHRQ)

Comprehensive care for mentally ill adults Primary care practice transformation Primary care practice transformation Coordinated care programs Medical home in pediatric practice Physician quality reporting and patient outcomes y q y p g p f d d l d l d

51

Impact of Medicaid policy on cardiovascular drug use and outcomes

51

Medicaid policy for cardiovascular drugs

Conceptual overview Brief review of prior studies Brief review of prior studies New research plan Input from Medicaid stakeholders p

52 52

The policy‐maker’s challenge

Baseline drug use patterns and clinical outcome rates Application of drug cost-containment policy More targeted use of expensive drugs No change in drug use Reduced use of beneficial drugs Cost reduction Clinical events Cost reduction from baseline Clinical events due to non-use Overall Adverse Overall spending reduction Adverse clinical

• utcomes

success failure

The policy‐maker’s challenge

Baseline drug use patterns and clinical outcome rates Application of drug cost-containment policy

Requires detailed, patient-level data

More targeted use of expensive drugs No change in drug use Reduced use of beneficial drugs Cost reduction Clinical events Cost reduction from baseline Clinical events due to non-use Overall Adverse Overall spending reduction Adverse clinical

• utcomes

success failure

Prior studies of Medicaid policy

Focused on prior authorization Gathered data on policies from Medicaid programs Gathered data on policies from Medicaid programs and websites Two case studies: Coxibs/NSAIDs ARBs/ACE‐Is

55

ARBs/ACE Is

55

D t l i Data analysis

CMS d d t CMS drug use data Prescriptions and units dispensed and dollars paid by Medicaid for all medications by paid by Medicaid for all medications, by calendar quarter, aggregated by state Units converted to defined daily doses (DDDs) Units converted to defined daily doses (DDDs) for coxibs and ACE/ARBs Main outcome measures: Main outcome measures: DDD Spending Spending

Models of policy impact

I d i i l i Interrupted time‐series analysis Evaluate level of drug use and spending before/after prior authorization before/after prior authorization implementation States with no program serve as controls p g Two types of effects, controlling for secular trends: Level effect: Immediate change in outcome Slope effect: Change in trend over time

Coxibs: the clinical scenario

Compared to non‐selective NSAIDs Similar efficacy Much more expensive Clinical benefit for properly selected patients p p y p well defined risk factors Widely variable use across states y Goal for policymakers Target coxibs to high‐risk patients Target coxibs to high risk patients Avoid overuse in others

Prior authorization programs

8 states implemented immediately 22 states implemented between 2000 and early 2003 20 states provided control data 6 i h h d l d b i f 6 states with programs scheduled to begin after study period

Proportion of NSAID defined daily doses accounted for by coxibs before and after implementation of a prior authorization program

60% No Prior Authorization Prior Authorization Difference 40%

se

14.6% reduction:

(95% CI:10.8 to 18.4 p<0.001)

20%

cent coxib us

Prior authorization start

0%

Perc

1.4% increasing slope

(95% CI: 0.0% to 2.8%; p = 0.052)

• 20%
• 6
• 5
• 4
• 3
• 2
• 1

1 2 3 4 5 6

Quarters before and after prior authorization start p

Fischer et al, NEJM, 2004

Coxib prior auth ‐ summary

36 states had prior authorization programs for coxibs at end of study Implementation of program associated with one‐ time decrease in coxib use of 14.6%, slow increase subsequently No difference in impact by adherence to clinical evidence

Renin‐angiotension axis blockers: ACE –I vs. ARB

Important component of HTN therapy Both classes effective at blocking RAA ACE‐I can cause cough or angioedema Rate of ACE‐I intolerance ~10% (5%‐20%) 5 ARBs much more expensive than ACE‐I ACE‐I: Many generics, $8‐$15 per month y g , $ $ 5 p ARB: All brand‐name, $46‐$58 per month

ARB prior authorization: key variables

Prior trials of ACE inhibitors Preferred drug lists (PDL) Preferred drugs in class can be prescribed without prior authorization O h d i h l i i Other drugs in the class require prior authorization

P i th i ti Prior authorization programs

19 states with prior authorization for ARBs 9 p implemented and 3+ quarters of post‐ intervention data 15 using PDL only 4 with requirement for ACE‐I trial 18 states with no prior authorization for ARBs Control states 13 states with prior authorization for ARBs 3 p scheduled or recently implemented but inadequate post‐implementation data

ARB DDDs as a proportion of RAA‐blocker DDD’s before and after prior authorization

35% Level effect: -0.3%; p=NS Slope effect: 0 5%; p=0 007

before and after prior authorization

25% 30% Slope effect: 0.5%; p=0.007 20% 10% 15%

Control PA using PDL

Level effect: -1.6%; p=0.026 Slope effect: -1.3%; p<0.001 0% 5%

PA using PDL PA - ACE req

0%

• 6
• 5
• 4
• 3
• 2
• 1

1 2 3 4 5 6

Fischer et al, Health Affairs, 2007

ARB spending as a proportion of anti‐hypertensive spending ARB spending as a proportion of anti hypertensive spending before and after prior authorization

20% 25% Level effect: 0.4%; p=0.049 Slope effect: 0.3%; p<0.001 15% 10%

Control

Level effect: -1.0%; p=0.003 Slope effect: -0.7%; p<0.001 0% 5%

PA using PDL PA - ACE req

p ; p 0%

• 6
• 5
• 4
• 3
• 2
• 1

1 2 3 4 5 6

Fischer et al, Health Affairs, 2007

ARBs‐ summary

32 states had prior authorization programs for ARBs at end of study Implementation of program with only PDL increased ARB use and spending li i i i AC i l d d A Policies requiring ACE‐I trial reduced ARB use and spending by a small amount

Limitations of aggregate analyses

Cannot distinguish between reductions in overuse and Cannot distinguish between reductions in overuse and reductions in clinically beneficial use Cannot assess the impact of policy on clinical

• utcomes

68 68

The policy‐maker’s challenge

Baseline drug use patterns and clinical outcome rates Application of drug cost-containment policy

Requires detailed, patient-level data

More targeted use of expensive drugs No change in drug use Reduced use of beneficial drugs Cost reduction Clinical events Cost reduction from baseline Clinical events due to non-use Overall Adverse Overall spending reduction Adverse clinical

• utcomes

success failure

New comparative effectiveness study

Focus on cardiovascular medications Focus on cardiovascular medications anti‐hypertensives anti‐diabetics anti diabetics anti‐platelets statins Clinical area with opportunities for better care that can cost less thiazide diuretics rosiglitazone

70

ezetimibe

70

Medication reimbursement policy approaches

Clinical guidance: Requires specific diagnosis test Clinical guidance: Requires specific diagnosis, test result, or other clinical factor for approval

e.g. coxibs and GI risk factors; biologics e.g. coxibs and GI risk factors; biologics

Therapeutic algorithm: Recommended treatment sequence, must be implemented in order

e.g. stepped therapy approaches

Economic preference: Choosing a preferred di i i h l i i medication without more complex criteria

e.g. PDL’s, generic drug requirements

71 71

Evaluate drug use outcomes

Patient level data (MAX) Patient‐level data (MAX) Changes in overall drug use Changes in new starts Changes in new starts Switching of regimens Changes in patients with strong indications Changes in patients without indications Changes in patients without indications

72 72

Evaluate clinical outcomes

Patient level data (MAX) Patient‐level data (MAX) Clinical outcomes in overall population MI MI Vascular intervention Kidney failure Kidney failure Diabetes complications Clinical outcomes in patients at high risk Multiple comorbidity

73

p y Switching/stopping of therapy

73

Structure of new study

74 74

Input from Medicaid stakeholders

Review and discussion of policy approaches Review and discussion of policy approaches Are the lists complete and accurate? Are the classifications reasonable? Are the classifications reasonable? Input on choice of drug use and clinical endpoints Input on choice of drug use and clinical endpoints Are the right metrics being measured? Review of preliminary results and feedback Are we providing the right data?

75

p g g

75

To learn more: mfischer@partners.org

www.rxfacts.org

Michael Fischer, M.D., M.S. Division of Pharmacoepidemiology & Pharmacoeconomics Brigham & Women’s Hospital Harvard Medical School