COMMENTARY

Is US Health Really the Best in the World?

Barbara Starfield, MD, MPH
I
NFORMATIONCONCERNINGTHE DEFICIENCIES OF US MEDI-
cal care has been accumulating. The fact that more than
40 million people have no health insurance is well
known. The high cost of the health care system is con-
sidered to be a deficit, but seems to be tolerated under the
assumption that better health results from more expensive
care, despite evidence from a few studies indicating that as
many as 20%to 30%of patients receive contraindicatedcare.
1
In addition, with the release of the Institute of Medicine
(IOM) report To Err Is Human,
2
millions of Americans
learned, for the first time, that an estimated 44000 to 98000
among them die each year as a result of medical errors.
The fact is that the US population does not have any-
where near the best health in the world. Of 13 countries in
a recent comparison,
3
the United States ranks an average of
12th (second from the bottom) for 16 available health in-
dicators. Countries in order of their average ranking on the
health indicators (with the first being the best) are Japan,
Sweden, Canada, France, Australia, Spain, Finland, the Neth-
erlands, the United Kingdom, Denmark, Belgium, the United
States, and Germany. Rankings of the United States on the
separate indicators
3
are:
13th (last) for low-birth-weight percentages
13th for neonatal mortality and infant mortality overall
11th for postneonatal mortality
13th for years of potential life lost (excluding external
causes)
11thfor life expectancyat 1year for females, 12thfor males
10th for life expectancy at 15 years for females, 12th for
males
10thforlifeexpectancyat 40yearsforfemales, 9thformales
7thfor life expectancyat 65years for females, 7thfor males
3rdfor life expectancyat 80years for females, 3rdfor males
10th for age-adjusted mortality
The poor performance of the United States was recently
confirmed by the World Health Organization, which used
different indicators. Using data on disability-adjusted life
expectancy, child survival to age 5 years, experiences with
the health care system, disparities across social groups in
experiences with the health care system, and equality of fam-
ily out-of-pocket expenditures for health care (regardless
of need for services), this report ranked the United States
as 15th among 25 industrialized countries.
4
Thus, the fig-
ures regarding the poor positionof the UnitedStates inhealth
worldwide are robust and not dependent on the particular
measures used. Common explanations for this poor perfor-
mance fail to implicate the health system. The perception
is that the American public behaves badly by smoking,
drinking, and perpetrating violence. The data show other-
wise, at least relatively. The proportionof females who smoke
ranges from14%in Japan to 41%in Denmark; in the United
States, it is 24% (fifth best). For males, the range is from
26%in Sweden to 61%in Japan; it is 28%in the United States
(third best).
The data for alcoholic beverage consumption are similar:
the United States ranks fifthbest. Thus, althoughtobacco use
and alcohol use in excess are clearly harmful to health, they
do not account for the relatively poor position of the United
States on these health indicators. The data on years of po-
tential life lost exclude external causes associated withdeaths
due to motor vehicle collisions and violence, and it is still the
worst among the 13 countries.
3
Dietary differences have been
demonstrated to be related to differences in mortality across
countries,
5
but the United States has relatively lowconsump-
tion of animal fats (fifth lowest in men aged 55-64 years in
20 industrialized countries) and the third lowest mean cho-
lesterol concentrations among menaged50to70years among
13 industrialized countries.
6
The real explanationfor relatively poor healthinthe United
States is undoubtedly complex and multifactorial. From a
health systemviewpoint, it is possible that the historic fail-
ure to build a strong primary care infrastructure could play
some role. A wealth of evidence
3
documents the benefits of
characteristics associated with primary care performance.
Of the 7 countries in the top of the average health ranking,
5 have strong primary care infrastructures. Although bet-
ter access to care, including universal health insurance, is
widely considered to be the solution, there is evidence that
the major benefit of access accrues only when it facilitates
receipt of primary care.
3,7
The health care system also may
contribute to poor health through its adverse effects. For
example, US estimates
8-10
of the combined effect of errors
and adverse effects that occur because of iatrogenic dam-
age not associated with recognizable error include:
12000 deaths/year from unnecessary surgery
7000 deaths/year from medication errors in hospitals
20000 deaths/year from other errors in hospitals
Author Affiliation: Department of Health Policy and Management, Johns Hop-
kins School of Hygiene and Public Health, Baltimore, Md.
Corresponding Author and Reprints: Barbara Starfield, MD, MPH, Department
of Health Policy and Management, Johns Hopkins School of Hygiene and Public
Health, 624 N Broadway, Room 452, Baltimore, MD 21205-1996 (e-mail:
[email protected]).
2000 American Medical Association. All rights reserved. (Reprinted) JAMA, July 26, 2000Vol 284, No. 4 483
80000 deaths/year from nosocomial infections in
hospitals
106000 deaths/year from nonerror, adverse effects
of medications
These total to 225000 deaths per year from iatrogenic
causes. Three caveats should be noted. First, most of the data
are derived from studies in hospitalized patients. Second,
these estimates are for deaths only and do not include ad-
verse effects that are associated with disability or discom-
fort. Third, the estimates of death due to error are lower than
those in the IOM report.
1
If the higher estimates are used,
the deaths due to iatrogenic causes wouldrange from230000
to 284000. In any case, 225000 deaths per year constitutes
the third leading cause of death in the United States, after
deaths from heart disease and cancer. Even if these figures
are overestimated, there is a wide marginbetweenthese num-
bers of deaths and the next leading cause of death (cere-
brovascular disease).
One analysis overcomes some of these limitations by es-
timating adverse effects in outpatient care and including ad-
verse effects other than death.
11
It concluded that between
4% and 18% of consecutive patients experience adverse ef-
fects in outpatient settings, with 116 million extra physi-
cian visits, 77 million extra prescriptions, 17 million emer-
gency department visits, 8 million hospitalizations, 3 million
long-term admissions, 199000 additional deaths, and $77
billion in extra costs (equivalent to the aggregate cost of care
of patients with diabetes).
11
Another possible contributor to the poor performance of
the United States on health indicators is the high degree of
income inequality in this country. An extensive literature
documents the enduring adverse effects of low socioeco-
nomic position on health; a newer and accumulating litera-
ture suggests the adverse effects not only of low social po-
sition but, especially, low relative social position in
industrialized countries.
12
Among the 13 countries in-
cluded in the international comparison mentioned above,
the US position on income inequality is 11th (third worst).
Sweden ranks the best on income equality (when income
is calculatedafter taxes andincluding social transfers), match-
ing its high position for health indicators. There is an im-
perfect relationship between rankings on income inequal-
ity and health, although the United States is the only country
in a poor position on both (B.S., unpublished data, 2000).
An intriguing aspect of the data is the differences in rank-
ing for the different age groups. US children are particu-
larly disadvantaged, whereas elderly persons are much less
so. Judging fromthe data on life expectancy at different ages,
the US population becomes less disadvantaged as it ages,
but even the relatively advantaged position of elderly per-
sons in the United States is slipping. The US relative posi-
tion for life expectancy in the oldest age group was better
inthe 1980s thaninthe 1990s.
13
The long-existing poor rank-
ing of the United States with regard to infant mortality
14
has
been a cause for concern; it is not a result of the high per-
centages of lowbirth weight and infant mortality among the
black population, because the international ranking hardly
changes when data for the white population only are used.
Whereas definitive explanations for the relatively poor po-
sition of the United States continue to be elusive, there are
sufficient hints as to their nature to provide the basis for
consideration of neglected factors:
(1) The nature and operation of the health care system.
In the United States, in contrast to many other countries,
the extent to which receipt of services from primary care
physicians vs specialists affects overall health and survival
has not been considered. While available data indicate that
specialty care is associated with better quality of care for spe-
cific conditions in the purview of the specialist,
15
the data
on general medical care suggest otherwise.
16
National sur-
veys almost all fail to obtain data on the extent to which the
care received fulfills the criteria for primary care, so it is not
possible to examine the relationships betweenindividual and
community health characteristics and the type of care re-
ceived.
(2) The relationship between iatrogenic effects (includ-
ing both error and nonerror adverse events) and type of care
received. The results of international surveys document the
high availability of technology in the United States. Among
29 countries, the United States is second only to Japan in
the availability of magnetic resonance imaging units andcom-
puted tomography scanners per million population.
17
Ja-
pan, however, ranks highest on health, whereas the United
States ranks among the lowest. It is possible that the high
use of technology in Japan is limited to diagnostic technol-
ogy not matched by high rates of treatment, whereas in the
United States, high use of diagnostic technology may be
linked to the cascade effect
18
and to more treatment. Sup-
porting this possibility are data showing that the number
of employees per bed (full-time equivalents) in the United
States is highest among the countries ranked, whereas they
are very lowin Japan
17
far lower than can be accounted for
by the common practice of having family members rather
than hospital staff provide the amenities of hospital care.
How cause of death and outpatient diagnoses are coded
does not facilitate an understanding of the extent to which
iatrogenic causes of ill health are operative. Consistent use
of E codes (external causes of injury and poisoning) would
improve the likelihood of their recognition because these
ICD (International Classification of Diseases) codes permit
attribution of cause of effect to Drugs, Medicinal, and Bio-
logical Substances Causing Adverse Effects in Therapeutic
Use. More consistent use of codes for Complications of
Surgical and Medical Care (ICD codes 960-979 and 996-
999) might improve the recognitionof the magnitude of their
effect; currently, most deaths resulting from these under-
lying causes are likely to be coded according to the imme-
diate cause of death (such as organ failure). The sugges-
tions of the IOM document on mandatory reporting of
adverse effects might improve reporting in hospital set-
COMMENTARY
484 JAMA, July 26, 2000Vol 284, No. 4 (Reprinted) 2000 American Medical Association. All rights reserved.
tings, but it is unlikely to affect underreporting of adverse
events in noninstitutional settings. Only better record keep-
ing, with documentation of all interventions and resulting
health status (including symptoms and signs), is likely to
improve the current ability to understand both the adverse
and positive effects of health care.
(3) The relationships among income inequality, social dis-
advantage, and characteristics of health systems, including
the relative contributions of primary care and specialty care.
Recent studies using physician-to-population ratios (as a
proxy for unavailable data on actual receipt of health ser-
vices according to their type) have shown that the higher
the primary care physiciantopopulation ratio in a state,
the better most health outcomes are.
19
The influence of spe-
cialty physiciantopopulation ratios and of specialistto
primary care physician ratios has not been adequately stud-
ied, but preliminary andrelatively superficial analyses suggest
that the converse may be the case. Inclusion of income in-
equality variables in the analysis does not eliminate the posi-
tive effect of primary care. Furthermore, states that have more
equitable distributions of income also are more likely to have
better primary care resource availability, thus raising ques-
tions about the relationships among a host of social andhealth
policy characteristics that determine what and how re-
sources are available.
Recognition of the harmful effects of health care inter-
ventions, and the likely possibility that they account for a
substantial proportion of the excess deaths in the United
States compared with other comparably industrialized na-
tions, sheds newlight on imperatives for research and health
policy. Alternative explanations for these realities deserve
intensive exploration.
REFERENCES
1. Schuster M, McGlynn E, Brook R. How good is the quality of health care in the
United States? Milbank Q. 1998;76:517-563.
2. Kohn L, Corrigan J, Donaldson M, eds. To Err Is Human: Building a Safer Health
System. Washington, DC: National Academy Press; 1999.
3. Starfield B. Primary Care: Balancing Health Needs, Services, and Technology.
New York, NY: Oxford University Press; 1998.
4. World Health Report 2000. Available at: http://www.who.int/whr/2000/en/
report.htm. Accessed June 28, 2000.
5. Kunst A. Cross-national Comparisons of Socioeconomic Differences in Mor-
tality. Rotterdam, the Netherlands: Erasmus University; 1997.
6. Law M, Wald N. Why heart disease mortality is low in France: the time lag ex-
planation. BMJ. 1999;313:1471-1480.
7. Starfield B. Evaluating the State Childrens Health Insurance Program: critical
considerations. Annu Rev Public Health. 2000;21:569-585.
8. Leape L. Unnecessary surgery. Annu Rev Public Health. 1992;13:363-383.
9. Phillips D, Christenfeld N, Glynn L. Increase in US medication-error deaths be-
tween 1983 and 1993. Lancet. 1998;351:643-644.
10. Lazarou J, Pomeranz B, Corey P. Incidence of adverse drug reactions in hos-
pitalized patients. JAMA. 1998;279:1200-1205.
11. Weingart SN, Wilson RM, Gibberd RW, Harrison B. Epidemiology and medi-
cal error. BMJ. 2000;320:774-777.
12. Wilkinson R. Unhealthy Societies: The Afflictions of Inequality. London, En-
gland: Routledge; 1996.
13. Evans R, Roos N. What is right about the Canadian health system? Milbank
Q. 1999;77:393-399.
14. Guyer B, Hoyert D, Martin J, Ventura S, MacDorman M, Strobino D. Annual
summary of vital statistics1998. Pediatrics. 1998;104:1229-1246.
15. Harrold LR, Field TS, Gurwitz JH. Knowledge, patterns of care, and outcomes
of care for generalists and specialists. J Gen Intern Med. 1999;14:499-511.
16. Donahoe MT. Comparing generalist and specialty care: discrepancies, defi-
ciencies, and excesses. Arch Intern Med. 1998;158:1596-1607.
17. Anderson G, Poullier J-P. Health Spending, Access, and Outcomes: Trends in
Industrialized Countries. New York, NY: The Commonwealth Fund; 1999.
18. Mold J, Stein H. The cascade effect in the clinical care of patients. N Engl J
Med. 1986;314:512-514.
19. Shi L, Starfield B. Income inequality, primary care, and health indicators. J Fam
Pract. 1999;48:275-284.
EDITORIAL
Editorials represent the opinions
of the authors and THE JOURNAL and not those of
the American Medical Association.
Sex and CyberspaceVirtual Networks
Leading to High-Risk Sex
Kathleen E. Toomey, MD, MPH
Richard B. Rothenberg, MD
I
N ADDITION TO ITS POWER AND REACH FOR RAPID INFOR-
mation exchange, the Internet has generated a new de-
bate: does it fundamentally change the way we lead our
lives? That the Internet has revolutionized communi-
cations and business practices worldwide is clearly recog-
nized. That the Internet may have some psychological ef-
fects on individual behavior, molded perhaps by the
technology itself, has nowbeen noted. Newly described be-
havioral disorders possibly linked to Internet use include
Internet-related depression
1,2
and cybersex addiction.
3-5
Re-
cent reports have suggested that fully one third of adult In-
ternet visits are directed to sexually oriented Web sites, chat
rooms, and news groups,
4
where Internet users can ob-
serve sexual images or participate in online sexual discus-
See also pp 443 and 447.
Author Affiliations: Division of Public Health, Georgia Department of Human Re-
sources (Dr Toomey), and Department of Family and Preventive Medicine, Emory
University School of Medicine (Dr Rothenberg), Atlanta.
Corresponding Author and Reprints: Kathleen E. Toomey, MD, MPH, Division
of Public Health, Georgia Department of Human Resources, 2 Peachtree St NW,
Suite 15-470, Atlanta, GA 30303-3682 (e-mail: [email protected]).
COMMENTARY/EDITORIAL
2000 American Medical Association. All rights reserved. (Reprinted) JAMA, July 26, 2000Vol 284, No. 4 485