The Cost-Effectiveness of
Interventions in Health and
Medicine
William H. Herman, M.D., M.P.H.
University of Michigan
• Rationale for conducting costeffectiveness 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
QALY
Quality-Adjusted Life-Year
adjusts length of life for
quality of life
Quality-Adjusted Life-Year
time in health state
x
quality of life in
health state
where quality of life = health utility
1.0 = excellent health
0 = death
Calculation of QALYs
20 years of life/excellent health
20 x 1.0 = 20 QALYs
20 years of life/10 excellent health
10 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 Comparator
New therapy
vs.
? all relevant alternatives?
? usual therapy?
? substandard therapy?
? placebo?
Choice of Comparator
Failure to compare a new therapy with
a strong alternative will result in a
deceptively favorable costeffectiveness 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 Analyses
The 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 costeffectiveness of diabetes
prevention?
Interventions Proven to Delay or Prevent
the Development of Type 2 Diabetes
Intervention
% Risk Reduction
Lifestyle (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
Percent developing
diabetes
Incidence
of Diabetes
Placebo (n=1082)All
Metformin (n=1073, p<0.001 vs. Placebo)
Lifestyle (n=1079, p<0.001 vs. Met , p<0.001 vs. Plac )
Lifestyle
(n=1079,
p<0.001
vs.
,
Metformin
(n=1073,
p<0.001
vs.Metformin
Plac)
Placebo (n=1082) p<0.001 vs. Placebo)
Cumulative incidence (%)
40
30
participants
Risk reduction
31% by metformin
58% by lifestyle
20
10
0
0
1
2
3
4
Years from randomization
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
Annual Direct Medical Costs in a Man Progressing
from IGT to Diabetes with Complications
$0
IGT (Placebo)
Diabetes (Diet &
Exercise)
Diabetes (Oral Agent)
Diabetes with
Microalbuminuria
Diabetes with MA and
High BP
Diabetes with MA,
High BP, and Angina
$1,000
$2,000
$3,000
$4,000
$5,000
1400
1684
1900
2200
2700
4600
Brandle et al. Diabetes Care 26:2300, 2003.
Health Utility Scores in a Man Progressing from IGT
to Diabetes with Complications
0
0.1
0.2
0.3
0.4
0.5
0.6
0.8
0.70
IGT (Placebo)
Diabetes (Diet &
Exercise)
0.69
0.67
Diabetes (Oral Agent)
Diabetes with
Neuropathy
0.60
Diabetes with Neuro
and High BP
Diabetes with Neuro,
High BP, and Stroke
0.7
0.59
0.52
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
Simulated Cumulative Incidence of Diabetes
in the DPP
90%
11.1 yrs
Placebo
80%
8.7%
8%
3.4 yrs
Metformin
22.3%
20%
70%
Lifestyle
60%
50%
40%
30%
20%
10%
0%
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
Years Since IGT Diagnosis
Herman et al. Ann Intern Med 142:323, 2005
Simulated Lifetime Clinical Outcomes
in the DPP
Outcome
Diabetes (%)
Blindness (%)
ESRD (%)
Amputation (%)
Stroke (%)
CHD (%)
Life expectancy (yrs)
Lifestyle
Metformin
Placebo
63
3
0.6
1.3
19
39
24.7
75
5
0.8
1.6
21
41
24.3
83
6
1.0
1.9
21
42
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
——
$1,124
$31,286
——
 Cost/  QALY
* 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 effective &
more costly
more effective &
more costly
Decrease in QALYs
Increase in QALYs
less effective &
less costly
more effective &
less costly
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
Intervention
Influenza vaccine
Beta-blockers after myocardial infarction
Cost-Effectiveness
(Cost/QALY)
Cost-saving
<$10,000
Mammographic screening
$10,000-$25,000
Hypertension medication (DBP >105 mmHg)
$10,000-$60,000
Cholesterol management, as secondary
prevention
$10,000-$50,000
Dialysis for end-stage renal disease
$50,000-$100,000
Left ventricular assist devices
PJ Neumann. N Engl J Med 2005; 353:1516-1522
$500,000-$1.4 million
How Attractive Does a New Technology Have to be to
Warrant Adoption and Utilization?
more costly
$100,000/QALY
$20,000/QALY
Decreased QALYs
Increased QALYs
$20,000/QALY
$100,000/QALY
less costly
Conclusion
Interventions for diabetes prevention
represent a good value for money in
people with IGT.
But...
• An alternative analysis suggested a
substantially higher cost per QALY-gained for
the lifestyle intervention ($200,000 per QALYgained).
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
10-year incidence
52%
47%
42%
Risk reduction vs
Placebo
DPP – 3 years
Lifestyle 58%
Metformin 31%
Risk reduction vs Placebo
DPP/DPPOS – 10 years
Lifestyle 31%
Metformin 19%
DPP Research Group. Lancet. 2009; 374:1677-1686
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
Outpatient visits
Inpatient care
ER visits
Urgent care visits
Calls to physicians
Prescription medications
Self monitoring supplies and
laboratory tests*
TOTAL
* diabetic participants only
Lifestyle Metformin Placebo
6,845
7,145
7,325
5,631
5,817
6,856
1,941
1,690
1,825
1,697
1,945
1,811
712
742
712
6,490
6,619
6,959
1,248
24,563
1,628
25,615
1,978
27,468
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, Perparticipant, 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 )
Health system perspective1
Undiscounted
Discounted2
Differences in QALYs ( QALY)
Undiscounted
Discounted
Lifestyle vs
placebo
DPP group
lifestyle vs
placebo
Metformin vs
placebo
928
1,226
-650
-323
-321
-159
0.14
0.12
0.14
0.12
0.02
0.02
Cost-saving
Cost-saving
Cost-saving
Cost-saving
Incremental cost-effectiveness ratios ( Cost /  QALY)
Health system perspective1
Undiscounted
6,651
Discounted2
10,037
1
2
Includes total direct medical costs
Both costs and QALYs discounted at 3%
Summary
Over 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
Conclusion
Health policy and societal policy in the
United States should support the
funding of intensive lifestyle and
metformin interventions for diabetes
prevention