Correspondence

Geographic Variations in Payments to Physicians

N Engl J Med 1993; 329:666-667August 26, 1993

Article

To the Editor:

In their article (March 4 issue), Welch et al. analyze geographic variations in expenditures for physicians' services in the United States1. They cite a positive correlation between age-adjusted mortality among Medicare patients and the rate of hospital admissions. This finding is not consistent with the results of other studies.

An analysis of 1990 data collected by the Health Care Financing Administration (HCFA) on standardized mortality 30 days after admission and on population-adjusted admissions rates shows that the relative likelihood of admission for any cause, according to state, was negatively correlated with standardized rates of mortality from any cause (r = -0.35, P<0.013)2. There were also negative correlations for congestive heart failure (r = -0.47, P<0.001), stroke (r = -0.40, P<0.003), and pneumonia (r = -0.26, P<0.06).

Manheim et al.3 found that hospitals had lower death rates in areas where there are more Medicare admissions per Medicare enrollee, as did hospitals with a higher volume of Medicare patients in relation to all admissions.

Welch et al. studied urban patients; the other studies included urban and rural patients. There may be disease-related differences between urban and rural areas in mortality rates in relation to admissions rates. For instance, the HCFA reported higher mortality rates in urban areas in 19904. Types of admissions may also vary between rural and urban areas. How do the authors reconcile these seemingly contrary findings?

Edwin Huff, Ph.D.
New Hampshire Foundation for Medical Care, Dover, NH 03820

4 References

1
Welch WP, Miller ME, Welch HG, Fisher ES, Wennberg JE. Geographic variation in expenditures for physicians' services in the United States. N Engl J Med 1993;328:621-627
Full Text | Web of Science | Medline

2
Health Care Financing Administration. Medicare hospital information report. 1992 Technical supplement. Section D. Hospitalization and mortality data for states. Washington, D.C.: Department of Health and Human Services, 1992.

3
Manheim LM, Feinglass J, Shortell SM, Hughes EFX. Regional variation in Medicare hospital mortality. Inquiry 1992;29:55-66
Web of Science | Medline

4
Health Care Financing Administration. Special report: hospital data by geographic area for aged Medicare beneficiaries: selected diagnostic groups, 1986. Washington, D.C.: Department of Health and Human Services, 1990.

To the Editor:

Welch et al. expressed surprise that the supply of physicians was not correlated with the level of Medicare payments.

Almost all the population over 65 years of age is covered by Medicare. In many areas, however, much of the population under 65 is uninsured or is covered by Medicaid. In areas with low rates of private insurance coverage in the population under 65, physicians may devote relatively more time to Medicare patients. The reverse may be true in areas with high rates of private insurance among people under the age of 65.

A rough analysis supports this view. Overall payments per beneficiary as presented in the authors' 3 are negatively correlated with the rate of private insurance for those under 651 (r = -0.47, N = 59, P<0.001 by two-sided t-test). This relation remains significant if the analysis is limited to the 25 largest metropolitan statistical areas (P<0.01). Per-capita income2 (as adjusted according to the Medicare prevailing-charge level3) is also negatively correlated with overall payments per beneficiary, but it adds little to the private-insurance variable.

This analysis supports the idea that universal health insurance for the population under 65 may be a necessary step -- along with global budgeting, as the authors suggest -- in reducing geographic practice variations for the over-65 (Medicare) population.

William N. Brownsberger
Office of the Attorney General Public Protection Bureau, Boston, MA 02108

3 References

1
Sources of health insurance and characteristics of the uninsured. Washington, D.C.: Employee Benefit Research Institute, January 1993.

2
Slater CM, Hall GE. Annual metro, city and county data book. 1992 County and city extra. Lanham, Md.: Bernan Press, 1993.

3
Pope GC, Hurdle S, Posner JG, Henderson M. An index of Medicare prevailing charges: report to the Health Care Financing Administration. Waltham, Mass.: Center for Health Economics Research, 1989.

To the Editor:

Three Florida metropolitan statistical areas were among the top five with regard to Medicare payments to physicians per beneficiary -- Miami, Fort Lauderdale, and Tampa. Florida has had a high influx of retirees, who commonly come from states with unkind winter climates. People migrate during retirement for a number of reasons; health commonly plays a part in their decisions. Age adjustment alone is insensitive to health as a motivation for migration and the resulting Medicare cost. The map and bar graph presented by Welch et al. show a general tendency to higher expenditures in more southerly areas, with the notable exception of Michigan.

Donald E. McMillan, M.D.
Gordon Brunhild, Ph.D.
University of South Florida, Tampa, FL 33612

To the Editor:

I was struck by the fact that the metropolitan statistical areas that were either nearly or more than 3 SD above the mean with regard to overall payments to physicians per Medicare beneficiary are located in two areas of the country (Dade County, Florida; and Wayne County, Michigan) where the malpractice situation is particularly bad. Was there any attempt to adjust the data for the frequency of malpractice actions filed, malpractice premiums, or some other measure that might take this factor into account? If this type of correction explains a substantial portion of the geographic variation in Medicare expenditures, it would provide strong evidence of the role of “defensive medicine” in driving up health care costs.

Donald R. Peven, M.D.
Oakwood Hospital, Dearborn, MI 48123

Author/Editor Response

The authors reply:

To the Editor: Dr. Huff identifies apparently contradictory findings in the correlation of admission rates and mortality rates. In doing so, he highlights the importance of the type of mortality data being analyzed. Dr. Huff is apparently using rates of mortality 30 days after admission, whereas our analysis used population-based mortality rates. Since the former includes only patients who are hospitalized, it is not surprising that in metropolitan areas where patients are admitted more often (and where presumably inpatients are on average less severely ill), lower mortality rates are found among hospitalized patients. Our analysis, in contrast, was based on mortality for the entire Medicare population, both inpatients and outpatients. To estimate the influence of hospital admissions on the health of the population, the overall mortality rate is the more appropriate measure.

Many factors can influence geographic variation in expenditures for physicians. Mr. Brownsberger reports a negative correlation between insurance coverage rates among those under 65 and expenditures for physicians' services among those over 65 in the 59 metropolitan areas reported in our Table 3. This finding suggests that physicians may be more likely to treat Medicare patients when there is a relative shortage of patients with private health insurance. Our work was restricted to one segment of the population (the elderly). A uniform national data base is a critical first step in evaluating the U.S. health system as a whole.

Dr. Peven suggests that malpractice patterns may be an important factor in explaining our results. We do not know whether malpractice premiums or lawsuits and expenditures for physicians are correlated. If such a correlation were found, we would be uncertain about the direction of causality. Dr. Peven's suggestions are certainly worthy of further study. Finally, Drs. McMillan and Brunhild suggest that southward migration of retirees is an important consideration. They hypothesize that retirees moving away from the unkind winter climates in the north do so primarily for health reasons and are therefore more likely to require health care. Alternatively, retirees who migrate may be healthier than average, because the healthy may be better able to enjoy outdoor activities. Although patients' migration may be relevant, so may be the southward migration of specialty physicians.

H. Gilbert Welch, M.D., M.P.H.
Department of Veterans Affairs, White River Junction, VT 05009

Mark E. Miller, Ph.D.
W. Pete Welch, Ph.D.
Urban Institute, Washington, DC 20037

Citing Articles (1)

Citing Articles

1
Feyza Darendeliler, Georgios Karagiannis, Patrick Wilton. (2007) Headache, Idiopathic Intracranial Hypertension and Slipped Capital Femoral Epiphysis during Growth Hormone Treatment: A Safety Update from the KIGS Database. Hormone Research 68:5, 41-47
CrossRef