The Most Expensive Medical Conditions In America

Abstract

This study uses a nationally representative survey to identify the most expensive conditions in the United States and to examine the association between spending and disability. The most expensive conditions at a population level were ischemic heart disease and motor vehicle accidents; at the per capita level they were respiratory malignancies. There was not a significant association between rank order of treatment costs and disability; the conditions with the greatest disability relative to expenditures were mood disorders, chronic obstructive pulmonary disease, and arthropathies. We use the findings to discuss the role for cost-of-illness and burden-of-disease estimates in setting priorities.

Full text

At the Intersection of Health, Health Care and Policy
doi: 10.1377/hlthaff.21.4.105
, 21, no.4 (2002):105-111Health Affairs
The Most Expensive Medical Conditions In America
Benjamin G. Druss, Steven C. Marcus, Mark Olfson and Harold Alan Pincus
Cite this article as:
http://content.healthaffairs.org/content/21/4/105.full.html
is available at:
The online version of this article, along with updated information and services,
For Reprints, Links & Permissions: http://healthaffairs.org/1340_reprints.php http://content.healthaffairs.org/subscriptions/etoc.dtlE-mail Alerts : http://content.healthaffairs.org/subscriptions/online.shtmlTo Subscribe:
reserved.
storage or retrieval systems, without prior written permission from the Publisher. All rights
form or by any means, electronic or mechanical, including photocopying or by information
may be reproduced, displayed, or transmitted in anyHealth Affairs U.S. Code), no part of
People-to-People Health Foundation. As provided by United States copyright law (Title 17,
by Project HOPE - The2002600, Bethesda, MD 20814-6133. Copyright ©
is published monthly by Project HOPE at 7500 Old Georgetown Road, SuiteHealth Affairs
Not for commercial use or unauthorized distribution
by guest on June 1, 2013Health Affairs by content.healthaffairs.orgDownloaded from

The Most Expensive Medical
Conditions In America
This nationwide study finds that the most disabling conditions are not
necessarily the ones we spend the most to treat.
by Benjamin G. Druss, Steven C. Marcus, Mark Olfson, and Harold
Alan Pincus
ABSTRACT: This study uses a nationally representative survey to identify the most expen-
sive conditions in the United States and to examine the association between spending and
disability. The most expensive conditions at a population level were ischemic heart disease
and motor vehicle accidents; at the per capita level they were respiratory malignancies.
There was not a significant association between rank order of treatment costs and disabil-
ity; the conditions with the greatest disability relative to expenditures were mood disorders,
chronic obstructive pulmonary disease, and arthropathies. We use the findings to discuss
the role for cost-of-illness and burden-of-disease estimates in setting priorities.
Estimates of cost are increasingly used by policymakers, advo-
cacy groups, and medical specialty organizations to quantify the burden of
disease.1However, critics have enumerated three important concerns that
limit the applicability of such estimates in setting health system priorities.2First,
the policy implications of cost-of-illness estimates are complicated by the fact that
while treatment costs represent a financial burden, they also are an indicator of
how society allocates its resources.3That is, a condition may incur low treatment
costs either because of a modest clinical need or because it is insufficiently funded;
high-cost illnesses may represent either opportunities for cost containment or
high-priority conditions requiring further resources.
Second, comparisons of spending across specific illnesses have required the
combination of multiple studies of single diseases, each using separate methodolo-
gies. This is particularly problematic because many of the estimates are provided
by specialty organizations and advocacy groups, which, while well intentioned,
have an incentive to highlight the importance of particular conditions. As a con-
Costly Conditions
HEALTH AFFAIRS ~ Volume 21, Number 4 105
©2002 Project HOPE–The People-to-People Health Foundation, Inc.
Benjamin Druss is associate professor in the Departments of Psychiatry and Public Health at Yale University
School of Medicine. Steven Marcus is research assistant professor at the University of Pennsylvania School of
Social Work. Mark Olfson is associate professor in the New York State Psychiatric Institute Department of
Psychiatry and the Columbia University College of Physicians and Surgeons. Harold Pincus is professor and
executive vice-chairman of the University of Pittsburgh Department of Psychiatry and senior scientist and
director of RAND–University of Pittsburgh.
by guest on June 1, 2013Health Affairs by content.healthaffairs.orgDownloaded from

gressionally mandated National Institutes of Health (NIH) report on cost of ill-
nesses warned, “disease-specific cost of illness estimates…are…essentially
noncomparable.”4
Finally, most standard economic methods, such as cost-effectiveness analysis
and cost-utility analysis, focus exclusively on the marginal costs and benefits of
treatment rather than the absolute costs and burdens of disease. These techniques
seek to identify treatments that result in the most aggregate health improvement
per dollar spent. What, if any, consideration should be given to absolute, rather
than marginal, treatment costs and burden of disease in priority setting?
This study uses a nationally representative survey to provide a consistent as-
sessment of cost and disability burden across multiple conditions. We then use
this assessment to consider the appropriate role of burden-of-disease and cost-of-
illness estimates in health care priority setting.
nStudy methods. Using the 1996 Medical Expenditure Panel Survey (MEPS), a
nationally representative sample of the U.S. civilian, noninstitutionalized popula-
tion, we determined the fifteen most expensive conditions nationally and then ex-
amined the cost and disability rankings among them.5We classified diseases based
on slightly modified Global Burden of Disease categories.6Expenditures for each health
care service were directly linked to the conditions reported by the participant as
prompting the service use and summed across all individuals. We calculated a sum-
mary disability variable for each condition at the population level as the mean of the
ranks for bed days, missed workdays, and rates of impairment in activities of daily
living (ADLs) and instrumental activities of daily living (IADLs).
First, we compared the association between national rankings of cost and dis-
ability (summed across individuals) and per capita (mean) rankings for treatment
costs and disability among affected individuals. Next, we assessed the relation-
ship between the condition ranks for costs and disability, first for national rank-
ings and then for per capita rankings. We used the nonparametric Wilcoxon
Signed Rank Test to assess the statistical significance of these comparisons.
Finally, to identify the conditions with the largest gap between spending and dis-
ability level, we calculated the difference in ranks between disability and treat-
ment costs (disability rank minus cost rank) across the fifteen conditions.7
Study Results
nPopulation-based estimates of cost and disability. Spending for the fifteen
highest-cost conditions outlined in Exhibit 1 accounted for 44.2 percent of total U.S.
health care spending in 1996. These conditions spanned a broad spectrum of both
chronicity and severity. For instance, three chronic conditions (ischemic heart dis-
ease, arthropathies, and hypertension) and two acute conditions (motor vehicle ac-
cidents and acute respiratory conditions) were the most costly. Similarly, the list in-
cluded both life-threatening conditions (malignancies) and prevalent but milder
conditions (such as respiratory infections).
106 July/August 2002
Health Care Costs
by guest on June 1, 2013Health Affairs by content.healthaffairs.orgDownloaded from

nCost and disability among affected persons. On a per capita basis, the two
most expensive conditions were respiratory malignancies and ischemic heart dis-
ease (Exhibit 2). The association between rankings of population-based spending
and spending per affected person was not statistically significant (S =–3.5, p=.85).
nAssociation between cost and disability rankings. The association be-
tween cost and disability rankings was not statistically significant either nationally
(S =6.5, p=6.63) or per affected person (S =–0.5, p=.98). The three conditions with
the lowest treatment costs relative to their associated disability level were mood
disorders, chronic obstructive pulmonary disease, and arthropathies. The three con-
ditions with the highest expenditures relative to their disability level were genitouri-
nary cancers, motor vehicle accidents, and cardiac dysrhythmias (Exhibit 3).
Discussion
The most expensive conditions during 1996 spanned a broad range of character-
istics, sharing little in common except their high national health costs. The rank
order of the conditions in terms of both cost and disability changed markedly
Costly Conditions
HEALTH AFFAIRS ~ Volume 21, Number 4 107
EXHIBIT 1
The Most Costly Conditions In The United States: National Expenditures, Bed Days,
Work-Loss Days, And Activity Impairments, 1996
Estimated
number with
condition
(millions)
National cost Bed days Work-loss days
Any ADL/IADL
impairment
Condition Billions Rank Millions Rank Millions Rank Thousands Rank
Ischemic heart
disease
Motor vehicle
accidents
3.4
7.3
$21.5
21.2
1
2
70.1
102.9
10
7
21.8
70.0
9
3
638.3
808.6
10
8
Acute respiratory
infection
Arthropathies
44.5
16.8
17.9
15.9
3
4
196.4
359.7
4
1
69.2
67.2
4
5
1,949.6
3,070.5
3
1
Hypertension
Back problems
26.0
13.2
14.8
12.2
5
6
61.1
191.6
12
5
12.0
83.0
11
1
544.3
1,380.9
12
5
Mood disorders
Diabetes
Cerebrovascular
disease
9.0
9.2
2.0
10.2
10.1
8.3
7
8
9
227.3
210.5
97.0
2
3
9
78.2
27.5
5.2
2
8
13
1,400.9
1,954.0
1,084.1
4
2
6
Cardiac dysrythmias
Peripheral vascular
disorders
2.9
3.4
7.2
6.8
10
11
66.5
55.1
11
13
7.2
12.8
12
10
528.7
591.4
13
11
COPD
Asthma
12.4
8.6
6.4
5.7
12
13
176.3
102.0
6
8
57.5
31.4
6
7
889.3
690.4
7
9
Congestive heart
failure
Respiratory
malignancies
1.1
0.3
5.2
5.0
14
15
48.7
21.7
14
15
1.1
2.5
15
14
494.6
121.5
14
15
SOURCE: Authors’ analysis of 1996 Medical Expenditure Panel Survey data.
NOTES: COPD is chronic obstructive pulmonary disease. ADL is activity of daily living. IADL is instrumental activity of daily living.
by guest on June 1, 2013Health Affairs by content.healthaffairs.orgDownloaded from

when shifting between national and individual estimates. Health expenditures
were not significantly associated with burden of disability across conditions, ei-
ther nationally or per affected person. In other words, the most costly conditions
are not necessarily the most disabling ones.
nRole of disease burden in priority setting. The potential role of disease bur-
den in priority setting can best be understood in the context of two philosophical
schools: utilitarianism and egalitarianism.8In its goal of maximizing population
health outputs for each dollar spent, cost-effectiveness analysis is rooted in a utili-
tarian value system. Under such a framework, which emphasizes aggregate health
improvement rather than how health is distributed across individuals, there is little
role for measuring total disease burden in setting priorities.9In contrast, egalitarian-
ism, an outgrowth of Rawls’s “Difference Principle,” seeks to allocate resources so as
to maximize the well-being of the worst-off. This approach serves as an implicit ba-
sis for burden-of-disease studies. Populations with the greatest burden of disease
are the “least well-off” and thus potentially deserve attention in policy and research
initiatives.
Few would take a strictly egalitarian approach in allocating health resources.
Funding ineffective therapies is inappropriate, however serious the disease is that
the therapies are intended to treat. However, a purely utilitarian approach may
also result in conclusions that appear ethically questionable. Perhaps the best ex-
ample of these potential pitfalls became evident during the initial Oregon
108 July/August 2002
Health Care Costs
EXHIBIT 2
Mean Spending, Bed Days, Work-Loss Days, And Activity Impairments Per Person
Affected By A Costly Condition, 1996
Mean annual
treatment costs
Mean annual
bed days
Mean annual
work-loss days
Any ADL/IADL
impairment
Condition Dollars Rank Days Rank Days Rank Percent Rank
Respiratory malignancies
Ischemic heart disease
$17,816
6,463
1
2
76.6
20.4
1
7
71.4
20.7
1
3
45
19
3
5
Congestive heart failure
Cerebrovascular disease
4,908
4,257
3
4
45.6
49.0
3
2
7.1
21.4
15
2
48
55
2
1
Motor vehicle accidents
Cardiac dysrythmias
3,081
2,467
5
6
14.0
22.9
12
6
14.6
8.8
5
11
11
18
11
6
Peripheral vascular disorders
Mood disorders
2,074
1,122
7
8
16.3
25.1
9
4
10.5
15.7
6
4
18
15
8
9
Diabetes
Back problems
1,097
956
9
11
22.9
14.5
5
10
8.2
9.6
12
9
21
8
4
12
Arthropathies
Asthma
954
663
10
12
18.5
11.8
8
14
9.9
9.2
8
10
14
8
7
13
Hypertension
COPD
Acute respiratory infection
569
529
415
14
13
15
13.9
14.2
8.8
13
11
15
7.6
9.9
7.7
13
7
14
14
7
4
10
14
15
SOURCE: Authors’ analysis of 1996 Medical Expenditure Panel Survey data.
NOTES: COPD is chronic obstructive pulmonary disease. ADL is activity of daily living. IADL is instrumental activity of daily living.
by guest on June 1, 2013Health Affairs by content.healthaffairs.orgDownloaded from

Medicaid priority-setting process. An initial effort to rank interventions solely on
the basis of cost-effectiveness led to several counterintuitive rankings, such as the
prioritization of tooth capping over appendectomy.10 Under a pure cost-effective-
ness approach, either high costs of treatment or low prevalence of illness can re-
sult in lifesaving treatments’ going unfunded. The second problem that emerged,
which resulted in a rejection of the initial waiver for the Oregon plan, was that the
scheme accorded less value to the well-being of disabled persons than to that of
healthy persons. Ranking schemas that rely solely on cost-effectiveness or treat-
ment benefits invariably run the risk of according lower value to the lives of per-
sons who are chronically ill or disabled.11
nAssessing disease burden. Assuming that disease burden has a role in prior-
ity setting, how should it be measured? The use of MEPS for disability and cost esti-
mates made it possible to directly compare these two domains, but also imposed
limits on the disability measures and did not make it possible to include mortality as
an indicator of burden. The majority of conditions that were highest in the MEPS
disability rankings were also among the top national causes of mortality and esti-
mated disability-adjusted life years (DALYs) in 1996.12 As with MEPS disability rat-
ings, population expenditures were not significantly associated with either the top
causes of mortality (S =–4, p=.8) or DALYs (S =3.5, p=.84). However, specific rank-
ings of disease burden varied considerably across the three measures.
Costly Conditions
HEALTH AFFAIRS ~ Volume 21, Number 4 109
EXHIBIT 3
Difference Between Rankings For Spending And Disability Levels, National And Per
Person Estimates, 1996
Rank for national estimates
Rank for mean estimates among
affected persons Total
difference
Condition Disability Cost Difference Disability Cost Difference
Mood disorders
COPD
2
6
7
12
–5
–6
4
11
8
14
–4
–3
–9
–9
Arthropathies
Diabetes
1
4
4
8
–3
–4
6
7
11
9
–5
–2
–8
–6
Asthma
Cerebrovascular disease
8
9
13
9
–5
0
14
2
12
4
2
–2
–3
–2
Back problems
Respiratory malignancies
3
15
6
15
–3
0
11
1
10
1
1
0
–2
0
Congestive heart failure
Respiratory infections
14
5
14
3
0
2
5
15
3
15
2
0
2
2
Peripheral vascular disease
Cardiac dysrythmias
11
13
11
10
0
3
9
8
7
6
2
2
2
5
Hypertension
Ischemic heart disease
Motor vehicle accidents
12
10
7
5
1
2
7
9
5
13
3
10
13
2
5
0
1
5
7
10
10
SOURCE: Authors’ analysis of 1996 Medical Expenditure Panel Survey data.
NOTES: At the top of the list, disability level exceeds spending. Moving down the list, spending exceeds disability level. COPD is
chronic obstructive pulmonary disease.
by guest on June 1, 2013Health Affairs by content.healthaffairs.orgDownloaded from

Even after a metric for assessing disease burden is chosen, there remains the
question of whether burden and treatment costs should be examined at the level
of the individual or population. Clinical studies, such as the Medical Outcomes
Study, have emphasized the former approach, whereas those taking a public
health perspective, such as the Global Burden of Disease study, have stressed the lat-
ter.13 In our study the vast difference in rankings between mean per capita and to-
tal national cost and burden estimates underscores the need to carefully consider
the perspective taken in such estimates.
Decisions about how to weigh different domains of disease burden and what
perspective to take when comparing across conditions involve more than method-
ological considerations. Rather, they involve choosing among basic values. Our
findings demonstrate that how these decisions are made is likely to have a major
impact on how priorities are set.
nCombining measurement methods. Quantitative methods have been devel-
oped to combine treatment effectiveness, cost, and illness burden into a single in-
dex.14 However, these methods may still be subject to the basic problem that differ-
ent methods of measuring illness severity and treatment benefits will inevitably
result in different prioritization rankings. Thus, it may be less helpful to use a single
quantitative method than to consider several different perspectives with an eye to-
ward the basic ethical trade-offs involved in choosing among them. For instance, the
Norwegian Health Service has explicitly identified both disease severity and treat-
ment effectiveness as guiding principles in resource allocation but has allowed
decisionmakers to decide how to weigh each of these considerations.15
nThe gap between spending and disability. We found a substantial gap be-
tween spending and disability burden overall and identified a series of outlier con-
ditions for which this gap was greatest. To understand the potential implications of
this gap, let us consider the case of mood disorders, one of the conditions with the
greatest disability relative to expenditures. If therapies for these conditions were in-
effective or prohibitively expensive, then the distribution could be regarded as effi-
cient—the free market would simply be diverting resources toward more cost-
effective treatments. However, studies have demonstrated that treatments for this
class of disorders are similarly cost-effective relative to other conditions.16 This sug-
gests that factors other than treatment effectiveness, such as stigma and under-
insurance, may be contributing to the gap.
Policy experience and common sense argue against mechan-
ically applying either cost-effectiveness or disease burden as tools in priority
setting. The ethical issues are too complex and too important for any quanti-
tative technique to remove the key role of human decisionmakers. However, that
does not mean that data cannot be used to inform and improve this process.
Norman Daniels has argued that rather than relying on any one approach, a variety
of data sources, in the context of basic ethical principles, can be used to promote a
110 July/August 2002
Health Care Costs
by guest on June 1, 2013Health Affairs by content.healthaffairs.orgDownloaded from

more fair and transparent deliberative process.17 We hope that the findings from
this study can serve as one such tool in achieving that larger goal.
This work was funded in part by a grant from the Robert Wood Johnson Foundation.
NOTES
1. World Health Organization, Investing in HealthResearch and De velopment:Report of the Ad Hoc Committee on Health
relating to Future Intervention Options (Geneva: WHO, 1996); and Institute of Medicine, Scientific Opportunities
and Public Needs: Improving Priority Setting and PublicInput at the National Institutes of Health (Washington: National
Academy Press, 1998).
2. B. Bloom et al., “Usefulness of U.S. Cost-of-Illness Studies in Healthcare Decision Making,”
Pharmacoeconomics 19, no. 2 (2001): 207–213.
3. M.L. Berk and A.C. Monheit, “The Concentration of Health Care Expenditures, Revisited,” Health Affairs
(Mar/Apr 2001): 9–18.
4. R. Kirschstein, Disease-Specific Estimates of Direct and Indirect Costs of Illness and NIH Support, National Institutes
of Health, Fiscal Year 2000 Update, February 2000, www1.od.nih.gov/osp/ospp/ecostudies/
COIreportweb.htm. (24 April 2002).
5. J. Cohen et al., “The Medical Expenditure Panel Survey: A National Health Information Resource,” Inquiry
33, no. 4 (1996): 373–389.
6. C. Murray and A. Lopez, The Global Burden of Disease (Cambridge, Mass.: Harvard School of Public Health,
1996).
7. A more detailed description of the study methods is available from benjamin.druss@yale.edu.
8. D. Brock, “Priority to the Worst Off in Health Care Resource Prioritization,” in Health Care and Social Justice,
ed. M. Battin, R. Rhodes, and A. Silvers (New York: Oxford University Press, 2001); and J. Olsen, “Theories
of Justice and Their Implications for Priority Setting in Health Care,” Journal of Health Economics 16, no. 6
(1997): 625–639.
9. A. Shiell et al., “Cost of Illness Studies: An Aid to Decision-Making?” Health Policy 8, no. 3 (1997): 317–323;
and V. Wiseman and G. Mooney, “Burden of Illness Estimates for Priority Setting: A Debate Revisited,”
Health Policy 43, no. 3 (1999): 243–251.
10. D. Hadorn, “Setting Health Care Priorities in Oregon: Cost-Effectiveness Meets the Rule of Rescue,” Jour-
nal of the American Medical Association 265, no. 17 (1991): 2218–2225.
11. D. Brock, “Health Care Resource Prioritization and Discrimination against Persons with Disabilities,” in
Americans with Disabilities: Implications for Individuals and Institutions, ed. L. Francis and A. Silvers (New York:
Routledge, 2000).
12. A. Minino and B. Smith, “Deaths: Preliminary Data for 2000,” National VitalStatistics Reports 49, no. 12 (2001):
1–40; and C. Michaud et al., “Burden of Disease—Implications for Future Research,” Journal of the American
Medical Association 285, no. 5 (2001): 535–539.
13. A. Tarlov et al., “The Medical Outcomes Study: An Application of Methods for Monitoring the Results of
Medical Care,” Journal of the American Medical Association 262, no. 7 (1989): 925–930; and Murray and Lopez,
The Global Burden of Disease.
14. E. Nord, Cost-Value Analysis in Health Care (Cambridge: Cambridge University Press, 1999).
15. Olsen, “Theories of Justice.”
16. R. Sturm and K. Wells, “How Can Care for Depression Become More Cost-Effective?” Journal of the American
Medical Association 273, no. 1 (1995): 51–58; and G. Simon et al., “Cost-Effectiveness of a Collaborative Care
Program for Primary Care Patients with Persistent Depression,” American Journal of Psychiatry 158, no. 10
(2001): 1638–1644.
17. N. Daniels and J. Sabin, “Limits to Health Care: Fair Procedures, Democratic Deliberation, and the Legiti-
macy Problem for Insurers,” Philosophyand Public Affairs 26, no. 4 (1997): 303–350; and N. Daniels, “Rationing
Fairly: Programmatic Considerations,” Bioethics 7, no. 2/3 (1993): 224–233.
Costly Conditions
HEALTH AFFAIRS ~ Volume 21, Number 4 111
by guest on June 1, 2013Health Affairs by content.healthaffairs.orgDownloaded from