Using Clinical Information To Project
Federal Health Care Spending
How Congress could use a diabetes spending
projection model to help inform budget decisions.
Michael J. O’Grady,
O Grady, PhD,
Principal, O’Grady Health Policy, LLC, &
Senior Fellow. National Opinion Research
Center at the University of Chicago
Potential Implications of Diabetes
Si
Simulation
l i M
Models
d l
• Diabetes is perhaps the prototypical chronic
condition for demonstrating what epidemiological
modelin can do for cost-estimating.
modeling
cost estimatin
• Natural history of diabetes has been assessed and
modeled extensively for over a decade.
– The baseline progression of major complications
are well documented.
– The effect of treatment interventions are generally
well understood (but evolving).
– Multiple
M lti l scientific
i tifi organizations
i ti
have
h
created
t d
diabetes models (NIH, CDC, UK and Europeans).
Diabetes Trials and Models
• Publication of groundbreaking trials has been
followed by
y model building
g
 Diabetes Control and Complication Trial
(Type 1 Diabetes) 1993
 DCCT first trial demonstrating microvascular
benefits of intensive glucose control in diabetes
United Kingdom Prospective Diabetes Study (Type
2 Diabetes) 1998
 UKPDS demonstrated benefits of intensive glucose
and blood pressure control in type 2 diabetes

Diabetes Trials and Models
• Trials provide us with natural history of the disease
• Trials also provide us with information regarding
 When treatments will have effects
 How large treatment effects are
 What complications are prevented
• UKPDS example
l
 Microvascular benefits observed after 9 years of
intensive glucose control
 Mortality and cardiovascular benefits during 10
years of post
post-trial
trial follow
follow-up
up (metabolic memory)
Insights on the Budget Window, Disease
Progression, and Effect of Treatment
– The NIDDK Model
Type 2 Diabetes and Glucose Control Efforts:
Average Annual Cost of Complications - 2007$
$2,500
The 10-Year
10 Year “Budget Window”
$2,000
Conventional
Protocol
$1,500
$1,000
$
,
Intensive Protocol
$500
$0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20
Year
Source:
National Changing Diabetes Programs: Federal Health Care Cost Estimating: A Look at Current
Practice and the Implications for Assessing Chronic Disease Prevention Proposals.
Prevalence and Incidence Modeling
g
Probability estimates
are used to sort the
population
l ti into
i t BMI
categories
Body Mass Index (BMI):
US Population
Over Age 24
BMI (along with
age) influences
probability of
having diabetes
Normal (< 25)
Diagnostic Categories:
Non-Diabetic
Undiagnosed Diabetic
Overweight (25-30)
(25 30)
Diagnosed Diabetic
Obese (> 30)
Deceased
Health care
service use
Probability of progressing to
disease and screening rates
determine populations in
these categories
6
Modeling Diabetes Complications
Advance in
disease
progression
one year
Retinopathy Module
(Clarke, 2004)
Assign Initial
Patient
Characteristic
s
Simulate
natural history
off di
diabetes
b t
progression
according to
patient
characteristics
Nephropathy Module
(UKPDS 33, 1998)
Neuropathy Module
(Clarke, 2004)
Coronary Heart
Disease Module
(Stevens, 2001)
Alive
Mortality
Module
(Vital
Statistics)
Dead
Stroke
St
k M
Module
d l
(Kothari, 2002)
Select next
patient
7
Policy Projections
• Modeled a prototypical diabetes treatment
improvement intervention that is similar to current
well designed disease management programs.
well-designed
programs
• Intensify the treatment of individuals with prevalent
and incident diabetes aiming to improve





Glucose control
Blood pressure control
Cholesterol control
Use of beneficial p
preventive therapies
p
((aspirin,
p
ACEI))
Expected benefits from program based on metaanalyses and national diabetes QI program experience.
Exhibit 4:
Diabetes Quality Improvement Intervention
Entry Age
Cohort
Number of patients
entering treatment
program each year
Baseline
spending (no
improvement
intervention)
Improvement
intervention
spending
New
spending
plus cost of
intervention
Net new
spending
25-year effects (2009–2033)
24-30 year old
60,000
$167 billion
$21 billion
$161 billion
-$6 billion
31-40
31
40 year old
60,000
$145 billion
$20 billion
$145 billion
$0 billion
41-50 year old
60,000
$134 billion
$19 billion
$136 billion
$2 billion
51-60 year old
80,000
$153 billion
$21 billion
$159 billion
$6 billion
61-64
61
64 year old
100 000
100,000
$160 billion
$21 billion
$165 billion
$5 billion
10-year effects (2009–2018)
24-30 year old
60,000
$21.0 billion
$3.7 billion
$22.1 billion
$1.2 billion
31 40 year old
31-40
60 000
60,000
$20 2 billion
$20.2
$3 7 billion
$3.7
$22 0 billion
$22.0
$1 9 billion
$1.9
41-50 year old
60,000
$20.7 billion
$3.6 billion
$22.1 billion
$1.5 billion
51-60 year old
80,000
$28.0 billion
$4.5 billion
$29.4 billion
$1.4 billion
61 64 year old
61-64
100 000
100,000
$34 9 billion
$34.9
$5 1 billion
$5.1
$36 5 billion
$36.5
$1 6 billion
$1.6
Dollar amounts in 2007 $.
Source: Derived from the authors’ own analyses/computations.
Exhibit 5:
Diabetes Quality
y Improvement
p
Intervention
(25-Year Spending)
110%
100%
14%
14%
90%
14%
13%
90%
89%
13%
80%
Diabetes Spending w/o
intervention (baseline)
70%
87%
82%
60%
87%
Intervention Costs
Diabetes Spending After Intervention
50%
24-30
31-40
41-50
Age
g Cohorts
Source: Derived from the authors’ own analyses/computations.
51-60
61-64
Possible Enhancements
#1 In
I selected
l
d iinstances, iinclude
l d the
h best
b
epidemiologic data and modeling in baseline
and intervention estimates:
–
–
–
Allow the modeling of obesity trends and their
interaction with chronic illness,
illness like diabetes.
diabetes
Incorporation of consensus “lessons learned”
from clinical trials.
Challenge to the epidemiological community –
g
be sure the data and trials meet rigorous
standards for inclusion in the policy debate.
Possible Enhancements
(continued)
#2 In certain instances, look beyond the
traditional 10-year
y
budget
g window,, if the data
indicates a better understanding for
policymakers.
–
–
For most proposals a 10-year window is
appropriate, but if there’s a well established
natural
t
l history
hi t
off the
th disease
di
exceptions
ti
should
h ld
be possible.
Cuts both ways – CBO may find that for many
proposal a longer window would show
g spending
p
g in the out years.
y
ballooning
Contact Information
Michael J. O
O’Grady,
Grady, Ph.D.
Principal,
O'Grady Health Policy LLC
(301) 656-7699 (v)
MOG d @
MOGrady@ogradyhp.com
d h
Elbert Huang,
Huang M
M.D.,
D M
M.P.H.
PH
Assistant Professor of Medicine
University of Chicago
((773)) 834-9143 ((v))
ehuang@medicine.bsd.uchicago.edu
James C. Capretta, MA
Principal and Director of Health
Policy Consulting
Civic Enterprises, LLC
(202) 715-3494 (v)
jcapretta@civicenterprises.net
Anirban Basu, Ph.D.
Assistant Professor
Center for Health and the Social
Sciences, University of Chicago
(773) 834
834-1796
1796 (v)
abasu@medicine.bsd.uchicago.edu
13