|
|
| |||||||||||||||||||||||||||||||||||||||
Background Expenditures per capita for hospitals are higher in the United States than in Canada. If the United States had the same spending pattern as Canada, the annual savings in 1985 would have exceeded $30 billion.
Methods We used data from published sources, computer files, and institutional reports to compare 1987 costs for acute care hospitals on three levels: national (the United States vs. Canada), regional (California vs. Ontario), and institutional (two California hospitals vs. two Ontario hospitals). Expenditures per admission were adjusted for the case mix of patients, prices of labor and other resources, and outpatient visits.
Results The United States had proportionately fewer hospital beds than Canada (3.9 vs. 5.4 per 1000 population), fewer admissions (129 vs. 142 per 1000 population), and shorter mean stays (7.2 vs. 11.2 days). Higher costs per admission in the United States were explained in part by a case mix that was more complex by 14 percent and by prices for labor, supplies, and other hospital resources that were higher by 4 percent. Hospitals in the United States provided relatively less outpatient care, particularly in emergency departments (320 vs. 677 visits per 1000 population). After all adjustments, the estimate of resources used for inpatient care per admission was 24 percent higher in the United States than in Canada and 46 percent higher in California than in Ontario. The estimated differences between the two pairs of California and Ontario hospitals were 20 and 15 percent.
Conclusions Canadian acute care hospitals have more admissions, more outpatient visits, and more inpatient days per capita than hospitals in the United States, but they spend appreciably less. The reasons include higher administrative costs in the United States and more use of centralized equipment and personnel in Canada.
Why are hospital expenditures greater in the United States? A study of patients 65 years of age or older concluded that admission rates and case mix were similar in the two countries, but that the cost per patient was much lower in Canada6. Other studies have focused on global budgeting and regulation as the principal mechanism of cost control in Canada7,8,9. Patterns of medical practice may differ because of the higher incidence of malpractice suits in the United States10. Also, the health care needs of the two populations may be dissimilar because of differing cultures and lifestyles11.
What factors contribute to the higher expenditures for hospital care in the United States? Do Americans have higher admission rates or longer hospital stays? Do U.S. hospitals treat patients with more complex diseases? Do they pay higher wages or higher prices for drugs and other resources? Do they provide more outpatient care? To answer these questions, we examined hospital costs for 1987 on three levels: national (comparing the United States with Canada); regional (comparing California, the largest state, with Ontario, the largest province); and institutional (comparing two hospitals in California with two in Ontario). The goal was to measure and explain differences in the use of resources to treat comparable patients. We did not examine the effects of differences in hospital expenditures on outcomes for patients.
Methods
Background Data
Details of the definitions, sources, and statistical adjustments are available from the National Auxiliary Publications Service (*). General background data, information on health care expenditures, and 1987 hospital data for the United States, Canada, California, and Ontario were gathered from published sources12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33. This study focused on acute care hospitals; extended care facilities, long-term chronic care units, and nursing homes were excluded. Canadian institutions report on a fiscal-year basis (April 1 to March 31), whereas U.S. institutions report on a calendar-year basis (January 1 to December 31); we adjusted the financial data for Canada to account for the three-month time difference. Canadian dollars were converted to U.S. dollars at the purchasing-power exchange rate of $1.23 Canadian to $1 U.S.; all dollar amounts mentioned in this article are expressed in 1987 U.S. dollars.
Adjustment for Case Mix
To adjust for case mix, we classified the patients in the United States according to diagnosis-related group (DRG) and the patients in Canada according to case-mix group (CMG). In 1987 these two systems were similar in design and definition32. We explored possible differences in coding between the two countries by identifying 100 pairs of DRGs and CMGs for which both complicated and uncomplicated versions of the same diagnosis have been defined and by comparing the proportions of patients classified as having the complicated type. To quantify the overall case mix, we analyzed the complete distribution of diagnoses using weights based on DRG-specific charges in California. A second summary measure was obtained with weights based on CMG-specific resource estimates in Ontario. The results obtained with the two weighting approaches were averaged to yield the case-mix ratios used in our adjustments.
Adjustment for Prices of Resources
Comparisons of wages were based on the earnings of full-time workers plus their fringe benefits. Because the comparisons differed depending on the level of skill required by the occupation, ratios were assessed at three different wage levels: high (head nurses and general-duty staff nurses), medium (laboratory technicians and x-ray technicians), and low (hospital cleaners and food-service helpers). To obtain a ratio for wages in general, we calculated a weighted average of the wage ratios for the three occupational categories. Weights of 50, 25, and 25 for the high, medium, and low wage categories, respectively, were estimated from the distribution of earnings of hospital employees according to occupation33. Alternative averages based on weights of 60, 20, and 20, and on 40, 30, and 30 were also calculated.
The price ratio for pharmacy supplies was derived from an analysis by Schieber et al30. The price ratio for nonmedical resources (such as electricity, water, and telephone service) was assumed to be the same as the ratio found by the Organization for Economic Cooperation and Development to apply to all goods and services -- i.e., the purchasing-power parity exchange rate. To determine the overall relative price of hospital resources, the ratios for labor and nonlabor resources were averaged, with weights of 0.67 for labor resources and 0.33 for nonlabor resources used to reflect an average of their relative contribution to total hospital costs in the United States and Canada12,13.
Adjustment for Outpatient Care
To adjust for hospital outpatient services, we developed a model in which "total admission equivalents" were estimated as the sum of hospital admissions, visits to the emergency department, and other outpatient visits (with the last two of these weighted according to their relative costs). When hospital admissions were divided by total admission equivalents, the quotient indicated the proportion of total expenditures that was related to inpatient service; for example, a hospital with 8000 admissions and 2000 additional admission equivalents (representing emergency and outpatient visits) would have 10,000 total admission equivalents and an inpatient share of total expenditures of 80 percent.
Emergency and outpatient visits were converted to admission equivalents on the basis of data from the American Hospital Association for individual states in 1987. We used multivariate regression to analyze these data for the 48 contiguous states and estimated the marginal costs of an average visit to the emergency department and an average outpatient visit, relative to the cost of an average admission. We then applied these estimates to Canadian hospitals, assuming that the difference between the two countries in the use of resources for each outpatient service was similar to the difference in resource use for an inpatient admission. The results of the regression model for California and Ontario were compared with estimates provided by the American Hospital Association and the Ontario Ministry of Health; the results obtained for the four specific hospitals were reviewed by the administrators of the respective facilities.
Specific Institutions
Background data for the four study hospitals came from published sources, computer files, and confidential reports. The Stanford University Medical Center and the Sunnybrook Health Science Center are large tertiary care hospitals located in northern California and southern Ontario, respectively. In 1987, both had trauma units, academic teaching programs, and extensive referral networks. St. John's Hospital and Wellesley Hospital are community-oriented hospitals in southern California and southern Ontario, respectively. Both had active obstetrics programs, small intensive care units, and large general medical programs. At all four facilities we evaluated case mix, relative prices, and the inpatient share of expenditures with the same methods used in the national and regional comparisons.
Specific Services
To examine the use of resources, we obtained confidential departmental reports detailing clinical activity at each of the four hospitals. Many services could not be compared at each of the four hospitals because of differences in accounting and reporting, but for this study we selected 12 well-specified diagnostic services chosen to illustrate diverse aspects of care, including routine and high-technology procedures.
Results
Table 1 shows selected socioeconomic factors that affect hospital expenditures. As compared with Canada, the United States had a higher percentage of persons 65 years of age or older and a smaller proportion living in cities with more than 100,000 residents. The crude death rate was higher in the United States than in Canada, but a smaller proportion of deaths occurred in hospitals. The birth rate was higher in the United States, and a higher percentage of American newborns weighed less than 2500 g. The homicide rate and the incidence of the acquired immunodeficiency syndrome were much higher in the United States. The differences between California and Ontario were proportionately similar, except that there were fewer elderly patients in California.
|
|
|
The adjustment for case mix did not take into account possible differences in coding practices between the two countries, nor did it reflect possible differences in severity within a given diagnostic group. One test of case mix showed that for each of 100 pairs of DRGs in which complicated and uncomplicated versions of the same diagnosis have been defined, the proportions classified as complicated were virtually identical in the two California and two Ontario hospitals.
The second adjustment considered the prices of the resources purchased by hospitals. Prices were, on average, slightly higher in the United States (Table 3). Persons working in high-wage occupations (such as staff nurses and head nurses) were paid considerably more there, but those in occupations at the other end of the wage scale (such as cleaners and food aides) received higher wages in Canada (Table 4). The overall price ratio depended on the choice of weights used to average the ratios for specific resources, but even substantial changes in weights resulted in shifts in the overall ratio of only about 2 percent in either direction. Adjustment for prices of resources lowered both the expenditure ratio between the United States and Canada and that between California and Ontario (Table 3).
|
The net result after adjustment for case mix, prices of resources, and inpatient share showed that hospitals in the United States used 24 percent more real resources per adjusted admission than Canadian hospitals; the difference between California and Ontario on this measure was 46 percent (Table 3).
The two pairs of California and Ontario hospitals are compared in Table 5. The lengths of stay and occupancy rates were much lower at the California hospitals; the number of admissions per bed was considerably higher at Stanford than at Sunnybrook, but at St. John's and Wellesley it was approximately equal. The two California hospitals had substantially fewer visits to the emergency department than their Ontario counterparts, although the proportion admitted was similar at all four facilities. The expenditures per admission before the three adjustments were 23 percent higher at Stanford than at Sunnybrook and 22 percent higher at St. John's than at Wellesley. Part of this difference was explained by a more complex case mix of patients, and part by higher prices of resources. The prices at Stanford were particularly high because of higher wages in the Stanford area in relation to California as a whole. The outpatient share of expenditures was higher in the two Ontario hospitals. After all adjustments, the two California hospitals used 20 percent (for Stanford vs. Sunnybrook) and 15 percent (for St. John's vs. Wellesley) more resources per adjusted admission.
|
|
Canadians spent appreciably less than Americans for acute hospital care in 1987, but they did not have fewer beds, fewer admissions, or shorter stays. The difference arises entirely in expenditures per admission, which were 39 percent higher in the United States. American hospitals had a more complicated mix of patients (as measured by DRGs), and they paid slightly more than Canadian hospitals for the resources used to provide care. The estimated inpatient share of total expenditures was higher in the United States, resulting in a difference of 24 percent between the countries in the quantity of resources per adjusted admission. The comparisons between California and Ontario were similar to the national ratios, but the difference of 46 percent in the quantity of resources per adjusted admission was considerably larger. By contrast, comparisons of specific institutions in California and Ontario yielded differences of 20 percent and 15 percent at the pair of tertiary care hospitals and the pair of community care hospitals, respectively.
What explains the apparently greater use of inpatient resources per adjusted admission in the United States? We do not think it is simply a problem of inadequate adjustment. The DRG system is far from a perfect measure of severity, but there is no evidence that within each DRG the illnesses in the United States were more acute than those in Canada. Indeed, some analysts have suggested that the prospective payment system of Medicare invites "DRG creep" in the United States, with a bias toward upgrading the classification of severity in order to obtain more reimbursement34. We also have confidence in the adjustment for the prices of resources. The ratio between the U.S. and Canada for hourly compensation in manufacturing has been estimated as 1.04, the same ratio that we found for the compensation of hospital employees35. Other investigators, using data from 1981 and 1985, concluded that hospital wages in the United States were actually lower than those in Canada36. If we overestimated the wage ratio, then the difference in resources used per patient was even larger than that reported here. The greatest uncertainty concerns the adjustment for inpatient share, but it is worth noting that for California, Ontario, and the four institutions the inpatient share was within 5 percent of the values available from other sources.
It is widely believed that American hospitals provide services more intensively. This is certainly true if the unit of comparison is a patient-day; however, the higher intensity per patient-day may be offset by the length of the average stay in Canadian hospitals -- more than 50 percent longer. As is true of physicians' services, Americans receive a different mix of services but not necessarily more of them. The longer stays in Canadian hospitals imply that Canadian patients have more frequent evaluation of their vital signs, more dressing changes, and more of other services that are provided on a daily basis. Direct measures of the quantity of 12 diagnostic services provided per admission at Stanford as compared with Sunnybrook and at St. John's as compared with Wellesley do not support the hypothesis that when comparable inpatients are treated, American hospitals provide more diagnostic services.
If U.S. hospitals do not deliver more clinical services per admission, what consumes the extra resources? We did not obtain data with which to answer this question, but there are three possibilities: administrative costs, other nonclinical services, and less intensive use of clinical equipment and personnel.
Hospitals in the United States must keep more extensive records in order to facilitate billing to the state and federal governments, insurance companies, and patients, and in anticipation of malpractice suits. According to Woolhandler and Himmelstein,3 in 1987 these administrative costs accounted for 20 percent of hospital expenditures in California but only 9 percent in Canada. If their results are correct and are representative of the corresponding country and region, we can calculate the increased costs for nonadministrative resources per admission as 10 percent for the United States as compared with Canada and as 28 percent for California as compared with Ontario. Thus, roughly half the differences reported in Table 3 would be explained by administrative costs. These costs may, however, be proportionately higher in California than in the rest of the United States; if so, the difference between the United States and Canada in nonadministrative resources per admission would be more than 10 percent.
In addition to administration, hospitals provide other nonclinical services that are frequently referred to as "hotel services" or "amenities." Most American hospitals compete for well-insured patients, raising the possibility that they provide these services at a quantity and quality greater than is found in Canada. Reliable, comparable data to test this possibility are not available, but there are several reasons for doubting that such services contribute substantially to higher costs per admission in the United States. First, patients in the United States have shorter average stays; therefore, they require fewer meals, less waste disposal, and less laundry service per admission. Second, these services are usually performed by low-wage workers, whose salaries tend to be lower in the United States than in Canada. Finally, the nonlabor resources consumed in these activities, such as the food supplies and cleaning products, represent only a small fraction of total hospital expenditures31.
Is it possible to deliver the same clinical service and to use fewer resources? In Canada, specialized procedures are performed in a relatively small number of large hospitals, whereas in the United States most community hospitals provide a wide variety of tertiary care services. For example, after adjustment for differences in population size, in 1987 there were 3 times as many hospitals with units providing open-heart surgery in California as in Ontario, 5 times as many with magnetic resonance scanners, and 10 times as many with extracorporeal lithotriptors. One consequence is much fuller use of capacity in Canada. The lithotriptor at Wellesley Hospital is used to treat almost 50 patients a week. At Stanford, as at many American hospitals, the average use is less than 1/10 as much. Among the more than 100 hospitals in California offering open-heart surgery, half have fewer than 200 procedures performed per year37. Canadian centralization, reliance on referral, and establishment of waiting lists result in less idle time for high-cost equipment and associated personnel.
Hospitals in the United States may also use relatively more resources to provide routine clinical services. The tremendous emphasis on early discharge in American hospitals creates a need for additional equipment and personnel ready to provide routine laboratory, radiologic, and other services on short notice. By contrast, the relatively long stays in Canadian hospitals are conducive to a queuing approach that probably results in better use of capacity. In the United States and California, the majority of patients are in small-to-medium-sized hospitals that do not have the relative efficiencies associated with the frequent performance of routine services. Also, greater variability in occupancy rates in American hospitals from day to day and from week to week makes it more difficult for them to schedule equipment and personnel with maximal efficiency.
Canadian patients pay a price for their higher-capacity use of equipment and personnel. The more centralized hospital system can cause delay or inconvenience in obtaining access to specialized services38,39. In some cases this results in a delay of discharge40. On the other hand, the quality of care in American hospitals may suffer when complex procedures are performed relatively infrequently in smaller hospitals41.
After all adjustments, we found that in 1987 the United States devoted more real resources to inpatients of acute care hospitals, after standardization for case mix, than did Canada. Our comparisons of a limited number of diagnostic procedures per admission in the four study hospitals did not support the hypothesis that American hospitals deliver more clinical services. We believe that the most likely explanations for the difference include higher administrative costs for billing and record keeping in American hospitals and more intensive use of centralized equipment and personnel in Canada.
Supported by the Robert Wood Johnson Foundation and the Connaught Fund.
We are indebted to Dr. Wade Aubry, Mr. Peter Ellis, Dr. Robert Fredricks, Prof. Morley Gunderson, Mr. Chris Helyar, Mr. David Hopkins, Mr. Terry Long, Mr. Cyril Nair, Mr. Scott Rowand, Mr. Narendra Shah, Prof. Evelyn Shapiro, Ms. Joanne Spetz, Blue Shield of California, the California Association of Hospitals and Health Systems, the Ontario Ministry of Health, Stanford University Medical Center, Sunnybrook Health Science Center, St. John's Hospital, and the Wellesley Hospital for their assistance in obtaining and interpreting data, and to Prof. George Pink, Prof. Noralou Roos, and Drs. Claire Bombardier, Allan Detsky, Alan Garber, John Hornberger, Douglas Owens, and Miriam Shuchman for reviewing earlier versions of this manuscript.
Source Information
From the Department of Medicine, University of Toronto, and the Division of Clinical Epidemiology, Wellesley Hospital Research Institute, Toronto (D.A.R.); and the Departments of Economics and Health Research and Policy, Stanford University, and the National Bureau of Economic Research, Stanford, Calif. (V.R.F.).
Address reprint requests to Dr. Redelmeier at Wellesley Hospital, Jones Bldg., Rm. 123, 160 Wellesley St. East, Toronto, ON M4Y 1J3, Canada.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Health Care in Canada and the United States
Schneider J. F., Roth S. L., Cornell C., Mendelssohn D.C., Skorecki K.L., Cardella C.J., Uretsky B. F., Redelmeier D. A., Fuchs V. R., Rouleau J. L., Braunwald E., Pfeffer M. A.
Extract |
Full Text
N Engl J Med 1993;
329:964-966, Sep 23, 1993.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
