1. Value-Based Insurance Design:Value-Based Insurance Design:
Preserving Quality While Containing CostPreserving Quality While Containing Cost
Presented at the Leonard Davis Institute, University of PennsylvaniaPresented at the Leonard Davis Institute, University of Pennsylvania
April 3, 2009April 3, 2009
Funding from University of Michigan Health System, Hartford Foundation, and theFunding from University of Michigan Health System, Hartford Foundation, and the
Michigan Diabetes Research Training Center (NIDDK)Michigan Diabetes Research Training Center (NIDDK)
Allison B. Rosen, MD, ScDAllison B. Rosen, MD, ScD
University of Michigan Center for Value-Based Insurance DesignUniversity of Michigan Center for Value-Based Insurance Design

2. Outline of TalkOutline of Talk
• BackgroundBackground
• A story of serendipity (and a bit of hard work)A story of serendipity (and a bit of hard work)
• UM intervention study – primary outcomesUM intervention study – primary outcomes
• Preliminary secondary analysesPreliminary secondary analyses
• Policy implicationsPolicy implications
– LocallyLocally
– More broadlyMore broadly

3. Health Care Cost CrisisHealth Care Cost Crisis
““The nation’s long-term fiscal balance will beThe nation’s long-term fiscal balance will be
determined primarily by the future rate of healthdetermined primarily by the future rate of health
care cost growth.”care cost growth.”
-- Peter Orszag, Director, Congressional Budget Office-- Peter Orszag, Director, Congressional Budget Office
Testimony before Senate Budget Committee, June 21, 2007.Testimony before Senate Budget Committee, June 21, 2007.

4. Value of Care is PoorValue of Care is Poor
• Substantial underutilization of high value
health care services
• U.S. adults receive only about half of
recommended care*
• For some chronic diseases, like diabetes,For some chronic diseases, like diabetes,
patients get fewer than half of needed clinicalpatients get fewer than half of needed clinical
services*services*
*McGlynn et al. N Engl J Med, 2003.

5. Fundamental Health Policy QuestionFundamental Health Policy Question
How do we organize and finance the healthHow do we organize and finance the health
care system to achieve maximum value forcare system to achieve maximum value for
what we spend?what we spend?
***Not: how do we save money?***Not: how do we save money?

6. Health Care Cost Crisis:Health Care Cost Crisis:
Remedies Must Recognize Cost / Quality TradeoffRemedies Must Recognize Cost / Quality Tradeoff
• Financial incentives to moderate utilization will beFinancial incentives to moderate utilization will be
a fundamental part of the solutiona fundamental part of the solution
• Yet, if not carefully aligned, they may worsenYet, if not carefully aligned, they may worsen
already pervasive problems in quality of carealready pervasive problems in quality of care

7. Today’s Talk Will Focus on Chronic MedicationsToday’s Talk Will Focus on Chronic Medications

8. Recent Drug Copayment Trends Based Largely on CostRecent Drug Copayment Trends Based Largely on Cost

9. 429
63
0
100
200
300
400
500
Primary Prevention Secondary Prevention
Different Patients Get Different BenefitsDifferent Patients Get Different Benefits
from Medicationsfrom Medications
NNTtopreventCVevent
Ellis JJ. J Gen Intern Med 2004;19:639-646.
Example: Number needed to
treat with statins to prevent
one cardiovascular event

10. No Difference in Statin Compliance Stratified byNo Difference in Statin Compliance Stratified by
Prevention CategoryPrevention Category. Survival Curves for Persistence to Statin Therapy Stratified by Prevention Category
Ellis JJ. J Gen Intern Med 2004;19:639-646.
Secondary prevention cohort
Primary prevention cohort

11. Statin Discontinuation Rates StratifiedStatin Discontinuation Rates Stratified
by Mean Prescription Copaymentby Mean Prescription Copayment
Ellis JJ. J Gen Intern Med 2004;19:639-646.
$0 to <$10
$10 to <$20
>$20
Copay amount was the most
important predictor of drug
discontinuation rate
No difference between
primary and secondary
prevention groups

12. Does Cost-Related Medication Underuse Matter?Does Cost-Related Medication Underuse Matter?
For Employers:
• Shifting costs to employees reduces employer drug spending
• However, evidence growing that overall costs may increase
*For excellent review, see Goldman et al., JAMA 2007;298:61.
Fendrick AM, et al. Am J Managed Care, 2001; Rosen AB. Med Care, 2006.
For Consumers:
• Growing evidence* shows that cost-related underuse:
– Increases adverse outcomes (chronically ill & poor most at risk)
– Increases health care costs in some cases

13. Getting Services to People WhoGetting Services to People Who NeedNeed Them:Them:
Should the Patient Decide?Should the Patient Decide?
• If increased cost sharing decreases the use of essentialIf increased cost sharing decreases the use of essential
medications & leads to worse outcomes, is it appropriatemedications & leads to worse outcomes, is it appropriate
to place the burden of weighing the benefits and costs ofto place the burden of weighing the benefits and costs of
medical interventions on the patient?medical interventions on the patient?
• If not, the system should provide some guidance andIf not, the system should provide some guidance and
incentives to promote better decisionsincentives to promote better decisions

14. Getting Services to People WhoGetting Services to People Who NeedNeed Them:Them:
Value-Based Insurance DesignValue-Based Insurance Design
• Value-based insurance design has been proposed toValue-based insurance design has been proposed to
realign incentives for valuerealign incentives for value
• Cost sharing is based on likelihood of benefit, notCost sharing is based on likelihood of benefit, not
(solely) the acquisition cost(solely) the acquisition cost
− The greater the benefit, the lower the co-payThe greater the benefit, the lower the co-pay
• Such a system would provide financial incentives toSuch a system would provide financial incentives to
targetedtargeted patients most likely to benefit frompatients most likely to benefit from specificspecific
therapiestherapies
Fendrick AM. Am J Managed Care, 2001.
Rosen AB. Med Care, 2006. Chernew M. Health Affairs, 2007.

15. Numerous VBID Experiments Ongoing:Numerous VBID Experiments Ongoing:
But in Need of Rigorous EvaluationBut in Need of Rigorous Evaluation
• Several employer-based experiments underway with variousSeveral employer-based experiments underway with various
forms of VBID for different diseases and/or drugsforms of VBID for different diseases and/or drugs
– These efforts are largely coming out of the private sectorThese efforts are largely coming out of the private sector
• Reported ‘results’ are excellentReported ‘results’ are excellent  over a dozen companiesover a dozen companies
reporting a “positivereporting a “positive financialfinancial return on investment (ROI)”return on investment (ROI)”
• Few rigorous evaluations exist to support these claimsFew rigorous evaluations exist to support these claims

16. • Two basic approaches in useTwo basic approaches in use
1.1. TargetTarget servicesservices that are high value (e.g., beta blockers)that are high value (e.g., beta blockers)
2.2. TargetTarget patientspatients with select clinical diagnoses (e.g., diabetes)with select clinical diagnoses (e.g., diabetes)
• Most employers have taken services approachMost employers have taken services approach
• Example: Marriott InternationalExample: Marriott International
– Waived copays for generics and cut branded copays in halfWaived copays for generics and cut branded copays in half
– Adherence increased from 2% to 4%Adherence increased from 2% to 4%
– Medication costs increased → medical claims decreased byMedication costs increased → medical claims decreased by
roughly same amountroughly same amount
• Rigorous evaluation needed of the second flavor of VBIDRigorous evaluation needed of the second flavor of VBID
Targeting is Critical to Attain ValueTargeting is Critical to Attain Value

17. Benefit Based Copay forBenefit Based Copay for
ACE-Inhibitors and Angiotensin ReceptorACE-Inhibitors and Angiotensin Receptor
Blockers for UM employees with DiabetesBlockers for UM employees with Diabetes
Proposal to the Michigan HealthyProposal to the Michigan Healthy
Community Initiative Task ForceCommunity Initiative Task Force
July 14, 2005July 14, 2005

18. Why Diabetes?Why Diabetes?
Sources: ADA. Economic costs of diabetes in the U.S., 2007. Diabetes Care. 2008;31:1-20.
CDC. Diabetes Public Health Resource. http://www.cdc.gov/diabetes/ Accessed on Oct 7, 2008.
• Almost 24 million Americans have diabetes (90–95% Type 2)
• Diabetes and its complications are a leading source of
morbidity, mortality, and costs, as well as lost productivity
• In 2007, direct medical costs of diabetes were $116 billion, &
indirect costs (from reduced productivity) totaled $58 billion
• Several therapies markedly reduce the risk of complications
• Yet, these therapies are underutilized in practice
• Increasing out-of-pocket (OOP) costs an important cause

19. Value Based Insurance DesignValue Based Insurance Design (VBID)(VBID)
Example: Predictive ModelingExample: Predictive Modeling
• Diabetes Mellitus*Diabetes Mellitus*
– Medicare first-dollar coverage (co-pays waived) of ACEMedicare first-dollar coverage (co-pays waived) of ACE
inhibitors resulted in nearly one million life years gainedinhibitors resulted in nearly one million life years gained
and a net savings of $7.4 billion over the cohort lifetimeand a net savings of $7.4 billion over the cohort lifetime
*Rosen AB, et al. Ann Intern Med. 2005;143:89.

20. Proposal for a Value-Based InsuranceProposal for a Value-Based Insurance
Design for UM Employees with DiabetesDesign for UM Employees with Diabetes
Michigan Healthy Community InitiativeMichigan Healthy Community Initiative
Task ForceTask Force
February 15, 2006February 15, 2006

21. Timing is EverythingTiming is Everything

22. FOD Objectives and OutcomesFOD Objectives and Outcomes
• Objectives:
– To examine the impact of targeted value-based copayment
reductions on the use of evidence-based medications by UM
employees and dependents with diabetes
– To successfully implement VBID program in real world setting
• Primary outcomes:
– Medication uptake (or utilization)
– Medication adherence
• Secondary outcomes: Health care utilization & expenditures

23. FOD InterventionFOD Intervention
• Targeted copayment reductions for evidence-based therapies:Targeted copayment reductions for evidence-based therapies:
− Antihypertensives (lower blood pressure)Antihypertensives (lower blood pressure)
− ACE-Inhibitors & Angiotensin Receptor Blockers (ACE/ARBs)ACE-Inhibitors & Angiotensin Receptor Blockers (ACE/ARBs)
− Statins (lower cholesterol)Statins (lower cholesterol)
− Glycemic agents (lower blood sugar)Glycemic agents (lower blood sugar)
− AntidepressantsAntidepressants
• VBID designed to maintain underlying incentive structure:VBID designed to maintain underlying incentive structure:
− Tier 1 (Generics)Tier 1 (Generics) Copays waivedCopays waived
− Tier 2 (Preferred Brand)Tier 2 (Preferred Brand) Copays reduced 50%Copays reduced 50%
− Tier 3 (Non-preferred brand)Tier 3 (Non-preferred brand) Copays reduced 25%Copays reduced 25%

24. Original Study TimelineOriginal Study Timeline
July 1, 2005July 1, 2005 –– June 30, 2007June 30, 2007
Q5 Q6 Q7 Q8 Q1 Q2 Q3 Q4
6/30/077/1/05 6/30/087/1/04 July 1, 2006
FOD Intervention
Q5 Q6 Q7 Q8Q1 Q2 Q3 Q4
**Disenrollment rates increased**Disenrollment rates increased
after M-CARE sale announced.after M-CARE sale announced.
XX
Sale of M-CARE to BCNSale of M-CARE to BCN
announcedannounced
XX
BCN officially takesBCN officially takes
ownership of M-CAREownership of M-CARE
UM Open
Enrollment
Control Firms Open
Enrollment Windows

25. Revised Timeline Used for EvaluationRevised Timeline Used for Evaluation
July 1, 2005 to June 30, 2007July 1, 2005 to June 30, 2007
Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8
6/30/077/1/05 6/30/087/1/04 July 1, 2006
FOD Intervention
Q5 Q6 Q7 Q8Q1 Q2 Q3 Q4
X
POST-PERIOD
7/1/06–6/30/07
PRE-PERIOD
7/1/05–6/30/06
Look-back window
to ID study population
Evaluation WindowEvaluation Window
X

26. Study Design and PopulationStudy Design and Population
• Study Design:Study Design: Interrupted time series with concurrent control groupInterrupted time series with concurrent control group
• Intervention GroupIntervention Group (N = 1777)(N = 1777)
– UM employees and dependents with at least one glucose lowering drugUM employees and dependents with at least one glucose lowering drug
filled in the prior yearfilled in the prior year
– Continuous enrollment in drug benefit for the period of interestContinuous enrollment in drug benefit for the period of interest
• Control GroupControl Group (N = 3273)(N = 3273)
– Employees & dependents of other (non-UM) employers, enrolled in M-Employees & dependents of other (non-UM) employers, enrolled in M-
CARE, and meeting same inclusion criteria as intervention groupCARE, and meeting same inclusion criteria as intervention group
• Sample Identification:Sample Identification: SeparateSeparate for PRE and POST periodsfor PRE and POST periods
– Example:Example: Intervention group sample in PRE period (7/05Intervention group sample in PRE period (7/05––6/06)6/06)
includes:includes:
» UM actives with glycemic agent filled in 7/04 – 6/05 (year before PRE period)UM actives with glycemic agent filled in 7/04 – 6/05 (year before PRE period)
» Continuously enrolled from 7/05 to 6/06 (entire year of PRE period)Continuously enrolled from 7/05 to 6/06 (entire year of PRE period)

27. Analytic ApproachAnalytic Approach
• Difference-in-Difference Regression Model
– Means of predicted values for intervention and control groups
» Uptake: measured once in PRE year & once in POST year
» Adherence: at quarterly intervals before and after program
– Bootstrapped SEs for differences & difference-in-differences
• Generalized Estimating Equations (GEEs)
– Accounts for correlation of multiple measures in same person
• Control Variables:
– Baseline age, gender, employee/dependent, comorbidity score,
neighborhood median household income, # medications

28. Co-Pay
Reductions
Begin
Pre-Period Post-Period
Medication
Adherence
Control GroupCo-Pay Reduction Group (FOD)
Feb 2002
Difference-in-Difference FrameworkDifference-in-Difference Framework
EFFECT =
Difference in
Difference
(D-in-D)
Diff.
Diff.
50%50%
100%100%

29. Definitions – Primary OutcomesDefinitions – Primary Outcomes
• Utilization:Utilization: At least one pharmacy fill of theAt least one pharmacy fill of the
medication at any time during the one year periodmedication at any time during the one year period
– Pre – year: 7/1/05 to 6/30/06Pre – year: 7/1/05 to 6/30/06
– Post – year: 7/1/06 to 6/30/07Post – year: 7/1/06 to 6/30/07
• Adherence:Adherence: Use prescription and days suppliedUse prescription and days supplied
data to assess days with available medications perdata to assess days with available medications per
quarter (Medical Possession Ratio, MPR)quarter (Medical Possession Ratio, MPR)
– Adjust for partial eligibility over the quarterAdjust for partial eligibility over the quarter
– Adjust for inpatient admissionAdjust for inpatient admission
– Adjust for medication switchingAdjust for medication switching

30. Standard Adherence ApproachStandard Adherence Approach
This group is ignored
regardless of any
indications for treatment
Standard approach requires at
least one pharmacy claim to
calculate adherence.

31. Focus on Diabetes: Baseline CharacteristicsFocus on Diabetes: Baseline Characteristics
Evaluation POST GroupEvaluation POST Group
InterventionIntervention ControlControl
Sample SizeSample Size 17771777 32733273
Female GenderFemale Gender 57.6%57.6% 54.0%**54.0%**
Mean AgeMean Age 46.246.2 46.846.8
EmployeeEmployee 63.0%63.0% 64.2%**64.2%**
ChildChild 4.7%4.7% 4.6%4.6%
SpouseSpouse 30.8%30.8% 30.8%30.8%
OtherOther 1.4%1.4% 0.4%0.4%
Income*Income* $55,447$55,447 $55,608$55,608
Charlson ScoreCharlson Score 1.431.43 1.461.46
*Median household income by zip code **Significant at p<0.05

32. FOD InterventionFOD Intervention Significantly Increased UptakeSignificantly Increased Uptake
of Medications in All Drug Classesof Medications in All Drug Classes
BaselineBaseline
UptakeUptake
(FOD)(FOD)
AbsoluteAbsolute
IncreaseIncrease
% reduction% reduction
in non-usersin non-users
MetforminMetformin 65.4%65.4% 3.2%3.2% 9.1%9.1%
ACE/ARBACE/ARB 55.0%55.0% 4.7%4.7% 10.4%10.4%
StatinsStatins 55.9%55.9% 5.2%5.2% 11.8%11.8%
SSRIsSSRIs 22.0%22.0% 1.8%1.8% N.A.N.A.
*All significant (p<0.001)
RELATIVE UPTAKE INCREASED BY 5% TO 9%RELATIVE UPTAKE INCREASED BY 5% TO 9%
0
2
4
6
8
10
M
etform
in
AC
E/AR
B
Statins
SSRIs
IncreaseinUptake(%)

33. BaselineBaseline
MPR (%)MPR (%)
AbsoluteAbsolute
Increase inIncrease in
MPR (%)MPR (%)
% Reduction% Reduction
in non-in non-
adherenceadherence
Metformin*Metformin* 71.3%71.3% 2.5%2.5% 8.6%8.6%
ACE/ARBACE/ARB◊◊ 82.3%82.3% 7.2%7.2% 40.6%40.6%
StatinsStatins†† 78.3%78.3% 4.1%4.1% 18.6%18.6%
SSRIs*SSRIs* 69.0%69.0% 5.1%5.1% 16.5%16.5%
*NS, ◊ p<0.001, † p=0.067
ADHERENCE TO ACE/ARBs & STATINs INCREASED SUBSTANTIALLYADHERENCE TO ACE/ARBs & STATINs INCREASED SUBSTANTIALLY
0
2
4
6
8
10
M
etform
in
AC
E/AR
B
Statins
SSR
Is
IncreaseinAdherence(%)
FOD InterventionFOD Intervention Increased Adherence to the HighestIncreased Adherence to the Highest
Value MedicationsValue Medications (ACE/ARBS & Statins)(ACE/ARBS & Statins)

34. Secondary Outcomes: CostsSecondary Outcomes: Costs
• ExpendituresExpenditures
− Pharmacy, non-pharmacy, totalPharmacy, non-pharmacy, total
− Expenditure analyses complicated by co-interventionsExpenditure analyses complicated by co-interventions
− I.e. other policy changes occurring during the study timeI.e. other policy changes occurring during the study time
frame which impact UM expenditures but not controls’frame which impact UM expenditures but not controls’

35. UM PBM Renegotiated
(Pharm Costs )
1/1/061/1/06
Unanticipated ‘Co-Interventions’ Made Evaluation of ActualUnanticipated ‘Co-Interventions’ Made Evaluation of Actual
Costs DifficultCosts Difficult
Q1 Q2 Q3 Q4Q1 Q2 Q3 Q4
6/30/077/1/05 July 1, 2006
FOD Intervention
BCN Pricing Went Into Effect
(Non-pharm Costs )
1/1/071/1/07
*Intervention group OOP costs unaffected; evaluation ∴ applied industry
standardized unit prices to changes in utilization due to FOD intervention
Q5 Q6 Q7 Q8Q5 Q6 Q7 Q8
OOP = Out-of-pocket

36. Focus On DiabetesFocus On Diabetes
Financial EffectsFinancial Effects  VeryVery
PreliminaryPreliminary
$0
$200
$400
$600
$800
Intervention Control Intervention Control
MeanQuarterlyExpenditures
($permemberperquarter)
pre post pre post pre post pre post
Pharmacy Spending Non-Pharmacy Spending
$63
$35
–$112 –$95
$28
–$18
Pharmacy was a little
more costly
-Diff-of-Diff +$28
Non-Pharmacy was a
little less costly
-Diff-of-Diff –$18
Overall was a little
more costly
-Diff-of-Diff +$10
Preliminary Story:
*Note: Cost estimates are
per member per quarter

37. Focus on Diabetes: ConclusionsFocus on Diabetes: Conclusions
• Targeted co-pay reductions increased uptake of medicationsTargeted co-pay reductions increased uptake of medications
• Among those on the medications, non-adherence ratesAmong those on the medications, non-adherence rates
declined substantially for ACE/ARBs and statinsdeclined substantially for ACE/ARBs and statins
• VBID-type interventions is a useful adjunct to efforts aimedVBID-type interventions is a useful adjunct to efforts aimed
at increasing patient initiation of and adherence to highat increasing patient initiation of and adherence to high
value medicationsvalue medications
• Impact on cost remains to be seenImpact on cost remains to be seen

38. LimitationsLimitations
• Ideal rate of use is not knownIdeal rate of use is not known
– Most diabetics should be on statin and ACE/ARBMost diabetics should be on statin and ACE/ARB
• Using ‘supply of medications filled’ as a proxy forUsing ‘supply of medications filled’ as a proxy for
adherence may overstate actual adherenceadherence may overstate actual adherence
• Comorbidity measured using claims data may beComorbidity measured using claims data may be
underestimatedunderestimated
• ROI not knownROI not known

39. A Comment on ROI and Cost-EffectivenessA Comment on ROI and Cost-Effectiveness
• Cost-saving:Cost-saving: intervention is effective and costs less inintervention is effective and costs less in
the long run than the cost of not interveningthe long run than the cost of not intervening
• Cost-effective:Cost-effective: intervention provides a health benefit atintervention provides a health benefit at
an acceptable costan acceptable cost
• High-value:High-value: intervention prevents a significant amountintervention prevents a significant amount
of illness and deathof illness and death andand is cost-effectiveis cost-effective
Most clinical preventive services are cost-effective;Most clinical preventive services are cost-effective;
Very few are cost savingVery few are cost saving

40. Value Based Insurance DesignValue Based Insurance Design
Maximizing Return On InvestmentMaximizing Return On Investment
Incremental costs of increased use of high value servicesIncremental costs of increased use of high value services
can be subsidized by:can be subsidized by:
1.1. Medical cost offsetsMedical cost offsets
− Amount saved by preventing adverse events will beAmount saved by preventing adverse events will be
directly related to level of clinical targetingdirectly related to level of clinical targeting
1.1. Higher cost sharing for services of lower valueHigher cost sharing for services of lower value
2.2. Enhanced productivityEnhanced productivity
3.3. Reduced disability costsReduced disability costs

41. What Was the University’s Response?What Was the University’s Response?

42. Reporting Results at UMReporting Results at UM
UM the university?UM the university?
OrOr
UM the employer?UM the employer?

43. Reporting Results at UM: Take 2Reporting Results at UM: Take 2

44. UM as a National Leader of VBID adoptionUM as a National Leader of VBID adoption
• In response to UM program, other employers haveIn response to UM program, other employers have
adopted programs almost identical to FODadopted programs almost identical to FOD
– Health Alliance Medical Plan of IllinoisHealth Alliance Medical Plan of Illinois
– Disney CorporationDisney Corporation –– Abbott LaboratoriesAbbott Laboratories
– QuadgraphicsQuadgraphics –– Diabetes AmericaDiabetes America
““Setting an example is not the main means ofSetting an example is not the main means of
leading others. It is the only means.”leading others. It is the only means.”
–– Albert EinsteinAlbert Einstein

45. VBID PractitionersVBID Practitioners . . .. . .
4646

46. How Did Employees Respond?How Did Employees Respond?
"As a member of the U-M community for nearly 20 years and"As a member of the U-M community for nearly 20 years and
the mother of a daughter who has suffered from Type 1the mother of a daughter who has suffered from Type 1
Diabetes for 15 years, I celebrate this initiative! Thank you!“Diabetes for 15 years, I celebrate this initiative! Thank you!“
““I was recently diagnosed with Type II diabetes by my primaryI was recently diagnosed with Type II diabetes by my primary
care physician. Since I already have 4 meds that I fill eachcare physician. Since I already have 4 meds that I fill each
month at $20.00 co-pay per month, I couldn't take on two moremonth at $20.00 co-pay per month, I couldn't take on two more
prescriptions. She called them into my pharmacy and I neverprescriptions. She called them into my pharmacy and I never
had them filled. I cannot afford them. I felt it more importanthad them filled. I cannot afford them. I felt it more important
to take the Blood Pressure meds.”to take the Blood Pressure meds.”

47. Which Elements of the Intervention Are Most Important?Which Elements of the Intervention Are Most Important?
Number Needed to Treat (NNT) to Prevent One Macrovascular (CVD) EventNumber Needed to Treat (NNT) to Prevent One Macrovascular (CVD) Event
*Tight glucose control*Tight glucose control doesdoes have importanthave important micromicrovascular benefits (but macrovascular protection higher priority)vascular benefits (but macrovascular protection higher priority)
4040
3535
1616
1414
1212
99
AspirinAspirin
StatinStatin
22ryry
PreventionPrevention
ACE/ARBsACE/ARBs
StatinStatin
11ryry
PreventionPrevention
MetforminMetformin
In ObeseIn Obese
Other BP MedsOther BP Meds
Glycemic AgentsGlycemic Agents**
Blood pressure medications with extra CVD & renal benefits
Only glycemic agent with macrovascular benefitsOnly glycemic agent with macrovascular benefits
No evidence ofNo evidence of macromacrovascular benefits (unless A1c ~10)vascular benefits (unless A1c ~10)
For reference onlyFor reference only

48. Annual Cost to Continue VBID for Actives,Annual Cost to Continue VBID for Actives,
by Number Needed to Treat (NNT) to Prevent One CVD Eventby Number Needed to Treat (NNT) to Prevent One CVD Event
NNTNNT Drug CostsDrug Costs CumulativeCumulative
Costs*Costs*
ACE (HOPE)ACE (HOPE) 99 84-118k84-118k ——
Tight BP ControlTight BP Control 10-1410-14 96-121k96-121k 180-239180-239
Statins (1Statins (1ryry
⇒⇒ 22ryry
prevention)prevention) 1414 ⇒⇒ 3535 73-11973-119 253-358253-358
Other Lipid Lowering AgentsOther Lipid Lowering Agents —— 18-3818-38 271-396271-396
Metformin in ObeseMetformin in Obese 14-1614-16 87-11087-110 358-506358-506
Other Glycemic AgentsOther Glycemic Agents NSNS 68-15568-155 427-661427-661
AntidepressantsAntidepressants —— 56-9656-96 483-757483-757
*If a budget threshold is reached, covering all drugs above that line will maximize health benefits relative to costs*If a budget threshold is reached, covering all drugs above that line will maximize health benefits relative to costs

49. Preliminary Assessment ofPreliminary Assessment of
Secondary Outcomes of InterestSecondary Outcomes of Interest

50. The Adverse Impact of Cost Sharing
Is More Pronounces in Vulnerable Groups
Chernew, Rosen, Fendrick. JGIM. 2008

51. Could VBID Be a Tool to Reduce Disparities?Could VBID Be a Tool to Reduce Disparities?

52. Income DataIncome Data
• Patient household income derived from zip-Patient household income derived from zip-
code matched census datacode matched census data
• Split into two groups: above and belowSplit into two groups: above and below
median household incomemedian household income
• Comparison by income of:Comparison by income of:
– Mean number of medicationsMean number of medications
– Medication uptakeMedication uptake

53. SES Effect on Mean Number Medications
-0.16
-0.12
-0.08
-0.04
0.00
0.04
All Glucose Lowering BP Lowering
ChangeinMean#MedsbyIncomeStatus
Intervention Controls

54. Impact of FOD Intervention by Income StatusImpact of FOD Intervention by Income Status
Metformin ACE-ARB Statin SSRI
ChangeinUptakeDuetoFOD(∆–∆)
Income below mean (Q1-Q2) Income above mean (Q3-Q4)
∆–∆–∆∆–∆–∆∆–∆–∆∆–∆–∆
∆–∆–∆∆–∆–∆
0%
1%
2%
3%
4%
5%
6%

55. ConclusionsConclusions
• Among those with incomes below median, a VBIDAmong those with incomes below median, a VBID
intervention, significantly increased:intervention, significantly increased:
– Mean number of medicationsMean number of medications
– Uptake of three of the four medication classesUptake of three of the four medication classes
• Among those with incomes above median, the VBIDAmong those with incomes above median, the VBID
intervention did not have a significant impactintervention did not have a significant impact
• Big Caveat:Big Caveat:
– Sample size is smallSample size is small
– Analyses are prelimaryAnalyses are prelimary

56. Conclusions To DateConclusions To Date
• Targeted co-pay reductions increased uptake of medicationsTargeted co-pay reductions increased uptake of medications
• Among those on the medications, non-adherence ratesAmong those on the medications, non-adherence rates
declined substantially for ACE/ARBs and statinsdeclined substantially for ACE/ARBs and statins
• VBID-type interventions is a useful adjunct to efforts aimed atVBID-type interventions is a useful adjunct to efforts aimed at
increasing patient initiation of and adherence to high valueincreasing patient initiation of and adherence to high value
medicationsmedications and possibly an avenue for addressingand possibly an avenue for addressing
disparities?disparities?

57. Future DirectionsFuture Directions
• Evaluate impact on intermediate outcomeEvaluate impact on intermediate outcome
• Assess for differential impact in other vulnerableAssess for differential impact in other vulnerable
populationspopulations
• Evaluate productivity outcomesEvaluate productivity outcomes
– Work-related disability, absenteeism, presenteeismWork-related disability, absenteeism, presenteeism
• Continue efforts to educate employers that positive ROI isContinue efforts to educate employers that positive ROI is
unrealistic expectationunrealistic expectation
– When that fails, revisit importance of productivity measurementWhen that fails, revisit importance of productivity measurement
• Initiating a smoke-free work place intervention at UMInitiating a smoke-free work place intervention at UM
– Looking for control universityLooking for control university

58. The Power of Financial IncentivesThe Power of Financial Incentives

62. Extra SlidesExtra Slides
mean number of medicationmean number of medication

63. Intermediate OutcomesIntermediate Outcomes
• Have incomplete lab data on cohortHave incomplete lab data on cohort
• Analyses not prespecifiedAnalyses not prespecified
A1c LDL
FOD Diff -0.30 -5.68
Control Diff -0.06 -2.17
FOD-Control Diff -0.23 -3.51

66. FOD Study PopulationFOD Study Population
M-CARE*
(~200,000)
U of M*
(~70,000)
UM FOD
(2,507)
a
b
Controls
(8,637)
c
*Estimates from 2006, UM includes actives and dependents only
Diabetics
Analyses restricted
to M-CARE if used:
-Medical claims
-Lab data
-Survey data

67. Extra Slides FollowExtra Slides Follow

68. Value-Based Insurance Design (VBID) inValue-Based Insurance Design (VBID) in
the Medicare Prescription Drug Benefit:the Medicare Prescription Drug Benefit:
An Analysis of Policy OptionsAn Analysis of Policy Options

69. Policy Options for ImplementingPolicy Options for Implementing
VBID in Part DVBID in Part D
Option 1:Option 1: Reduce cost sharing for specific drugs or drug classesReduce cost sharing for specific drugs or drug classes
Option 2:Option 2: Exempt specific drugs or drug classes from 100% costExempt specific drugs or drug classes from 100% cost
sharing in the coverage gapsharing in the coverage gap
Option 3:Option 3: Reduce cost sharing for enrollees with chronic conditionsReduce cost sharing for enrollees with chronic conditions
Option 4:Option 4: Reduce cost sharing for enrollees participating inReduce cost sharing for enrollees participating in
medication therapy management programs (MTMPs)medication therapy management programs (MTMPs)
Option 5:Option 5: Reduce cost sharing for chronic condition special needsReduce cost sharing for chronic condition special needs
plans (CC-SNPs)plans (CC-SNPs)

70. Political Support
Ability to Implement
Policy
CMS Authority to
Change Policy
Feasibility
Size of Medicare
Population Affected
Potential to Improve Medicare
Option 5
Policy Options
Option 1 Option 2 Option 3 Option 4
Political Support
Ability to Implement
Policy
CMS Authority to
Change Policy
Feasibility
Size of Medicare
Population Affected
Potential to Improve Medicare
Option 5
Policy Options
Option 1 Option 2 Option 3 Option 4
Greatest Potential / Most Feasible
Moderate Potential / Feasibility
Least Potential / Feasible
Greatest Potential / Most Feasible
Moderate Potential / Feasibility
Least Potential / Feasible
We Evaluated Each Option’s Impact And FeasibilityWe Evaluated Each Option’s Impact And Feasibility
According To Four CriteriaAccording To Four Criteria

71. Greatest Potential / Most Feasible
Moderate Potential / Feasibility
Least Potential / Feasible
Size of MedicareSize of Medicare
PopulationPopulation
AffectedAffected
Policy ChangePolicy Change
RequiredRequired
OperationalOperational
Change RequiredChange Required Political SupportPolitical Support
Option 1: Reduce Cost Sharing for SpecificOption 1: Reduce Cost Sharing for Specific
Drugs or Drug ClassesDrugs or Drug Classes
 Potential to reach a large number of Part D beneficiaries
Can be implemented in the current policy environment
 Plans can create new formulary tier for targeted drugs
 Incentives needed to encourage plans to take advantage of the option
Low or no cost sharing for high-value drugs would encourage adherence among all
enrollees who may benefit from a drug in these classes, regardless of their chronic
condition diagnosis
Most Promising

72. Greatest Potential / Most Feasible
Moderate Potential / Feasibility
Least Potential / Feasible
Size of MedicareSize of Medicare
PopulationPopulation
AffectedAffected
Policy ChangePolicy Change
RequiredRequired
OperationalOperational
Change RequiredChange Required Political SupportPolitical Support
Option 2: Exempt Specific Drugs or Drug Classes fromOption 2: Exempt Specific Drugs or Drug Classes from
100% Cost Sharing in the Coverage Gap100% Cost Sharing in the Coverage Gap
 Affects fewer beneficiaries, but targets patients with high drug
spending
Can be implemented in the current policy environment
 Incentives needed to encourage plans to add gap coverage for
targeted drugs or drug classes
Because adherence may decline as enrollees are exposed to high cost sharing, this
option would offer protection when costs are generally the greatest—during the
coverage gap
Promising, But Smaller Impact

73. Greatest Potential / Most Feasible
Moderate Potential / Feasibility
Least Potential / Feasible
Size of MedicareSize of Medicare
PopulationPopulation
AffectedAffected
Policy ChangePolicy Change
RequiredRequired
OperationalOperational
Change RequiredChange Required Political SupportPolitical Support
Option 3: Reduce Cost Sharing for EnrolleesOption 3: Reduce Cost Sharing for Enrollees
with Chronic Conditionswith Chronic Conditions
Potential to reach a large number of Part D beneficiaries
May require exemption from non-discrimination clause and uniform
benefit requirement
Requires process to identify enrollees with specific chronic condition
diagnoses and to select drugs eligible for reduced cost sharing
Targeting enrollees with a specific chronic condition for lower cost sharing—for all
drugs or just those that treat the particular condition—would lessen the out-of-pocket
burden associated with the chronic condition
Effective Targeting, But Legislative Barriers

74. Size of MedicareSize of Medicare
PopulationPopulation
AffectedAffected
Policy ChangePolicy Change
RequiredRequired
OperationalOperational
Change RequiredChange Required Political SupportPolitical Support
Option 4: Reduce Cost Sharing for EnrolleesOption 4: Reduce Cost Sharing for Enrollees
Participating in MTMPsParticipating in MTMPs
Targets small percentage of Medicare population
Positively reinforces MTMP efforts to improve beneficiaries’
medication adherence
May require exemption from non-discrimination clause and uniform
benefit requirement
 Plans may wish to monitor MTMP participation to identify beneficiaries
who qualify for low cost sharing
Enrollees participating in Part D’s MTMP would benefit from reduced cost sharing for
specific drugs, in addition to other patient outreach and counseling on medication
use
Greatest Potential / Most Feasible
Moderate Potential / Feasibility
Least Potential / Feasible
MTMP: Medication Therapy Management Program
Potential to Improve Adherence, But Legislative Barriers

75. Greatest Potential / Most Feasible
Moderate Potential / Feasibility
Least Potential / Feasible
Size of MedicareSize of Medicare
PopulationPopulation
AffectedAffected
Policy ChangePolicy Change
RequiredRequired
OperationalOperational
Change RequiredChange Required Political SupportPolitical Support
Option 5: Reduce Cost Sharing for CC-SNPOption 5: Reduce Cost Sharing for CC-SNP
Enrollees Based on the Plan’s Target ConditionEnrollees Based on the Plan’s Target Condition
Affects fewest beneficiaries
Can be implemented in the current policy environment
 No operational changes required
 May be an ideal first step in implementing VBID in the Medicare Part
D program
SNPs designed for a specific chronic condition could reduce cost sharing for drugs
treating the target condition as part of an overall model of care aimed at better
disease management
Path of Least Resistance

76. While VBID Can Be Implemented in Part D,While VBID Can Be Implemented in Part D,
Policymakers Should Consider How toPolicymakers Should Consider How to
Encourage Plan AdoptionEncourage Plan Adoption
• Several options are now available to Part D plansSeveral options are now available to Part D plans
– No legislative or regulatory changes needed for Options 1, 2, or 5No legislative or regulatory changes needed for Options 1, 2, or 5
– Plans may requirePlans may require incentives to adopt VBIDincentives to adopt VBID
• Additional analysis is needed to further exploreAdditional analysis is needed to further explore
VBID in MedicareVBID in Medicare
– Identify incentives that would be most attractive to Part D plansIdentify incentives that would be most attractive to Part D plans
– Define high-value drugs or chronic conditionsDefine high-value drugs or chronic conditions
– Project costs or savings from VBID implementationProject costs or savings from VBID implementation
– Examine VBID opportunities for other Medicare services (Parts A&B)Examine VBID opportunities for other Medicare services (Parts A&B)

77. 35%
58%
19%
27%
0%
25%
50%
75%
100%
2006 2009
5 tier+
4 tier
N = 1,429
Could VBID Take Off As Other Part D BenefitCould VBID Take Off As Other Part D Benefit
Designs Have?Designs Have?
Source: Avalere Health analysis using DataFrame®, a proprietary database of Medicare Part D plan features. 2009 data from
November 2008. 2006 data from July 2006.
*N = Total number of PDPs offered each year.
N = 1,648
Percent of Medicare Prescription Drug Plans (PDPs) with Four or More Tiers