The Cost-Effectiveness of Interventions in Health and

MedicineWilliam H. Herman, M.D., M.P.H.

University of Michigan

• Rationale for conducting cost-effectiveness analyses

• How is cost-effectiveness assessed?

• What is the cost-effectiveness of diabetes prevention?

Barriers to diffusion of new medical treatments

• Patient• Provider• System

Patient level barriers

• Demographic (age, gender, race)• Socioeconomic position (education,

income)• Health status (including depression)• Self-efficacy• Cost

Provider level barriers• Demographics (age, training, CME,

experience)• Knowledge of guidelines and critical

pathways• Attitudes to innovation• Opinions of key opinion leaders• Peer practices• Cost

System level barriers

• Practice structure and organization• Information systems• Time barriers• Cost

Why perform CEAs?

• Resources are limited

• Choices must be made

• Choices should consider costs and outcomes

Value for Money

Essential Elements of Economic Analyses of Health-Care Programs

• Type of analysis• Perspective• Type and definition of costs• Description and valuation of outcomes• Choice of comparator• Modeling• Discounting• Sensitivity analyses

Types of Economic Analyses• Descriptive cost analysis• Cost-benefit• Cost-effectiveness• Cost-utility

Perspective of Economic Analyses

• Payer• Society

Type of Costs• Direct medical• Direct nonmedical• Indirect

Definition of Direct Medical Costs• Cost of intervention• Cost of side-effects of intervention• Cost of outcomes

Description and Valuation of Outcomes

• Beneficial outcomes produced• Adverse outcomes averted

Outcomes• Clinical• Years of life• Quality-adjusted life-years

QALYQuality-Adjusted Life-Year

adjusts length of life forquality of life

Quality-Adjusted Life-Yeartime in health state x

quality of life in health state

where quality of life = health utility

1.0 = excellent health0 = death

Calculation of QALYs20 years of life/excellent health

20 x 1.0 = 20 QALYs

20 years of life/10 excellent health10 with blindness

(10 x 1.0) + (10 x 0.51) = 15.1 QALYs

Approaches to Measuring Health Utilities

• Standard gamble• Multiattribute utility models• Rating scales

Multiattribute Utility Models• Health Utilities Index (HUI)• Quality of Well-Being Index (QWB)• EuroQol (EQ-5D)

21

22

Choice of ComparatorNew therapy

vs.? all relevant alternatives?

? usual therapy?? substandard therapy?

? placebo?

Choice of ComparatorFailure to compare a new therapy with a strong alternative will result in a deceptively favorable cost-effectiveness picture.

Modeling• When direct empirical data are not

available, methods of imputation and extrapolation are used to estimate outcomes

• No model generates new data, it merely combines existing information within an explicit framework

Discounting• Even in a world of zero inflation,

there are advantages to receiving benefits earlier and incurring costs later.

• Discounting adjusts future costs and benefits to current value.

Sensitivity AnalysesThe values of one or more of the key parameters are varied singly or simultaneously to evaluate the robustness of the results to the underlying assumptions.

What is the cost-effectiveness of diabetes

prevention?

Interventions Proven to Delay or Prevent the Development of Type 2 Diabetes

Intervention % Risk ReductionLifestyle (4 trials) 29-58%Metformin (2 trials) 26-31%Lifestyle & Metformin (1 trial) 28%Acarbose (1 trial) 25%Troglitazone (1 trial) 55%Rosiglitazone (1 trial) 60%

DPP Study Population• 3,234 subjects with impaired

glucose tolerance (IGT)–Fasting plasma glucose 95 - 125 mg/dl–2 hour plasma glucose 140 - 199 mg/dl

• Age > 25 years (mean 51 years) • BMI > 24 kg/m2 (mean 34 kg/m2)• 68% women• 45% minorities

DPP Interventions• Lifestyle

– healthy, low-calorie, low-fat diet & physical activity of moderate intensity (brisk walking for 150 min/week) to achieve and maintain 7% loss of body weight

– 16 session core curriculum over 6 months then monthly follow-up

• Metformin– 850 mg daily increasing to 850 mg twice daily– standard lifestyle recommendations– quarterly follow-up

• Placebo– standard lifestyle recommendations

0 1 2 3 4

0

10

20

30

40 Placebo (n=1082)Metformin (n=1073, p<0.001 vs. Plac)Lifestyle (n=1079, p<0.001 vs. Met , p<0.001 vs. Plac )

Percent developing diabetes

All participants

All participants

Years from randomization

Cum

ulat

ive

inci

denc

e (%

)

Placebo (n=1082)Metformin (n=1073, p<0.001 vs. Placebo)Lifestyle (n=1079, p<0.001 vs. Metformin , p<0.001 vs. Placebo)

Incidence of Diabetes

Risk reduction31% by metformin31% by metformin58% by lifestyle58% by lifestyle

The DPP Research Group, NEJM 346:393-403, 2002

Analyses• Health system perspective• Cost per Quality-Adjusted Life-Year (QALY)• Lifetime time horizon• Interventions as implemented in the DPP• Year 2000 US dollars

DPP. Ann Intern Med 142:323, 2005

Data Sources

Treatment of IGT Treatment of Diabetes

Costs DPP Cost Model

Quality of Life DPP Quality of Life Model

Health Outcomes DPP Type 2 Diabetes Model

DPP. Ann Intern Med 142:323, 2005

1684

1900

2200

2700

4600

1400

$0 $1,000 $2,000 $3,000 $4,000 $5,000

IGT (Placebo)

Diabetes (Diet &Exercise)

Diabetes (Oral Agent)

Diabetes withMicroalbuminuria

Diabetes with MA andHigh BP

Diabetes with MA,High BP, and Angina

Annual Direct Medical Costs in a Man Progressing from IGT to Diabetes with Complications

Brandle et al. Diabetes Care 26:2300, 2003.

0.69

0.67

0.60

0.59

0.52

0.70

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

IGT (Placebo)

Diabetes (Diet &Exercise)

Diabetes (Oral Agent)

Diabetes withNeuropathy

Diabetes with Neuroand High BP

Diabetes with Neuro,High BP, and Stroke

Health Utility Scores in a Man Progressing from IGT to Diabetes with Complications

Coffey et al. Diabetes Care 25:2238, 2002.

Diabetes Cost-Effectiveness Model• Markov model structure• Follows a patient cohort from diagnosis of IGT to

death• IGT transition probabilities based on DPP• Diabetes, microvascular and macrovascular

transition probabilities based on UKPDS and literature

• Assumes 10 year interval between DPP onset and UKPDS clinical diagnosis of type 2 diabetes mellitus

• Tracks costs, QALYs, disease progression, 5 complications, and survival

CDC Diabetes Cost-effectiveness Group. JAMA 287:2542, 2002

Years Since IGT Diagnosis

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

3.4 yrs 8.7%

22.3%

11.1 yrsPlacebo

Metformin

Lifestyle

Simulated Cumulative Incidence of Diabetes in the DPP

Herman et al. Ann Intern Med 142:323, 2005

20%

8%

Simulated Lifetime Clinical Outcomes in the DPP

Outcome Lifestyle Metformin PlaceboDiabetes (%) 63 75 83Blindness (%) 3 5 6ESRD (%) 0.6 0.8 1.0Amputation (%) 1.3 1.6 1.9Stroke (%) 19 21 21CHD (%) 39 41 42Life expectancy (yrs) 24.7 24.3 24.1

Herman et al. Ann Intern Med 142:323, 2005

IGT Intervention -Summary Lifetime Outcomes*

Outcome Lifestyle Metformin Placebo

Lifetime Costs $51,974 $55,261 $51,339

Lifetime QALYs 10.89 10.45 10.32

Cost v. Pbo $635 $3,922 ——

QALY v. Pbo 0.57 0.13 ——

Cost/ QALY $1,124 $31,286 ——* costs and QALYs discounted at 3% per year

Herman et al. Ann Intern Med 142:323, 2005

How Attractive Does a New Technology Have to be to Warrant Adoption and Utilization?

more costly

less costly

Increase in QALYsDecrease in QALYs

more effective & more costly

less effective & less costly

less effective & more costly

more effective & less costly

Distribution of Cost-Effectiveness Ratios for Preventive Measures and Treatments for

Existing Conditions

Cohen JT. N Engl J Med 2008; 358:661-663

Cost-Effectiveness of Selected Interventions in the Medicare Population

InterventionCost-Effectiveness

(Cost/QALY)Influenza vaccine Cost-savingBeta-blockers after myocardial infarction <$10,000Mammographic screening $10,000-$25,000Hypertension medication (DBP >105 mmHg) $10,000-$60,000Cholesterol management, as secondary

prevention$10,000-$50,000

Dialysis for end-stage renal disease $50,000-$100,000Left ventricular assist devices $500,000-$1.4 million

PJ Neumann. N Engl J Med 2005; 353:1516-1522

How Attractive Does a New Technology Have to be to Warrant Adoption and Utilization?

more costly

less costly

Increased QALYsDecreased QALYs

$100,000/QALY

$100,000/QALY

$20,000/QALY

$20,000/QALY

ConclusionInterventions for diabetes preventionrepresent a good value for money inpeople with IGT.

But...• An alternative analysis suggested a

substantially higher cost per QALY-gained for the lifestyle intervention ($200,000 per QALY-gained).

Eddy DM. Ann Intern Med 2005; 143:251-264

Purpose

To assess the cost-effectiveness of the lifestyle and metformin interventions relative to the placebo intervention with an intent-to-treat analysis spanning the combined 10 years of DPP/DPPOS.

Background• The DPPOS followed participants for an additional 7 years

during which time those in the lifestyle and metformin interventions were encouraged to continue those interventions.

• During DPPOS, lifestyle participants received extra lifestyle support and all participants were offered a 16 session group lifestyle intervention and 4 healthy lifestyle program sessions per year.

• A recent intent-to-treat analysis demonstrated a persistent benefit of the lifestyle and metformin interventions on the incidence of type 2 diabetes for at least 10 years after randomization.

Cumulative Incidence of Diabetes during DPP/DPPOS

DPP Research Group. Lancet. 2009; 374:1677-1686

52%47%42%Risk reduction vs

PlaceboDPP – 3 yearsLifestyle 58%Metformin 31%

Risk reduction vs PlaceboDPP/DPPOS – 10 yearsLifestyle 31%Metformin 19%

10-year incidence

Methods• Data on resource utilization, cost, and quality-of-life were

collected prospectively during DPP and DPPOS.• To estimate the cost of lifestyle if it had been administered

in a group format rather than individually as it was during DPP, we recalculated costs assuming that the core curriculum and monthly follow-up visits with the lifestyle case managers were conducted as group sessions with ten participants

• Economic analyses were performed from a health system perspective that considered direct medical costs.

Cumulative, Undiscounted, Per-participant, Direct Medical Costs of the DPP/DPPOS

Interventions by Intervention Group and Year

Cumulative, Undiscounted, Per-participant, Direct Medical Costs of Medical Care Received Outside the

DPP/DPPOS by Intervention Group and Year

Undiscounted, Per-participant, 10-year Direct Costs of Medical Care Received Outside the DPP/DPPOS by Intervention Group and Type

Costs ($) by category Lifestyle Metformin PlaceboOutpatient visits 6,845 7,145 7,325Inpatient care 5,631 5,817 6,856ER visits 1,941 1,690 1,825Urgent care visits 1,697 1,945 1,811Calls to physicians 712 742 712Prescription medications 6,490 6,619 6,959Self monitoring supplies and

laboratory tests* 1,248 1,628 1,978TOTAL 24,563 25,615 27,468* diabetic participants only

Cumulative, Undiscounted, Per-participant, Total Direct Medical Costs of the DPP/DPPOS

Interventions and Medical Care Received Outside the DPP/DPPOS by Intervention Group and Year

Cumulative, Undiscounted, Per-participant, Quality-Adjusted Life-Years Gained by Intervention Group and Year

Incremental Cost-Effectiveness Ratios over 10 Years by Intervention Group –

Health System Perspective

Differences in costs ( cost )Lifestyle vs

placebo

DPP group lifestyle vs

placeboMetformin vs

placeboHealth system perspective1      

Undiscounted 928 -650 -321Discounted2 1,226 -323 -159

       Differences in QALYs ( QALY)      

Undiscounted 0.14 0.14 0.02Discounted 0.12 0.12 0.02

       Incremental cost-effectiveness ratios ( Cost / QALY)Health system perspective1      

Undiscounted 6,651 Cost-saving Cost-savingDiscounted2 10,037 Cost-saving Cost-saving

1 Includes total direct medical costs2 Both costs and QALYs discounted at 3%

SummaryOver 10 years, from a payer perspective:•The lifestyle intervention was cost-effective and the group lifestyle intervention was cost-saving compared to the placebo intervention

– The group lifestyle intervention was approximately 1/3 less expensive than the lifestyle intervention

– The increased cost of the lifestyle intervention relative to the placebo intervention was largely offset by the reduced costs of non-intervention-related medical care

– The lifestyle intervention was associated with better quality-of-life than the placebo intervention

•The metformin intervention was cost-saving or at least, cost-neutral compared to the placebo intervention

– The increased cost of the metformin intervention relative to the placebo intervention was entirely offset by the reduced costs of non-intervention-related medical care

ConclusionHealth policy and societal policy in the United States should support the funding of intensive lifestyle and metformin interventions for diabetes prevention