Graduate medical education is precariously positioned in the rapidly evolving American health care system. Increased scrutiny of teaching hospitals is one effect of ongoing efforts to reform the delivery of health care and to control its costs. In pursuing the goals of teaching, providing patient care, and conducting research, teaching hospitals generate costs that average 20% more than other hospitals (1). The value obtained for these additional costs is now being questioned as stakeholders respond to financial pressures. Some observers have invoked war metaphors, as they perceive academic medicine and teaching hospitals "under siege" by the cost-control strategies referred to as managed care (2). In this environment, teaching hospitals will be forced to account for their differences in costs and will be pressured to reduce them (34).
Graduate medical education (GME) is financed with revenues derived from patient care and through grants in a "complex web of cross-subsidization" (3). The federal government is the single most important source of GME financing. Medicare payments to teaching hospitals for GME amount to more than $6 billion per year through direct and indirect mechanisms (3). Medicaid contributes a small additional amount in some states (5).
The purchasers of private insurance subsidize GME through higher premiums necessary to pay for the larger bills at teaching hospitals. Managed care organizations (MCOs), because they operate in a fiercely competitive market, are increasingly unwilling to subsidize GME through these higher payments. Furthermore, MCOs are now less willing than in the past to have their patients treated by resident physicians. Among the medical specialties, psychiatry has been identified by MCOs as a specific target of cost-control efforts (6). Increased cost-consciousness is a major trend in health care that has intensified with the privatization and corporatization of the American health care system (78).
Proponents of teaching hospitals have argued that in carrying out their social missions, they offer a wide array of services (9) and provide a disproportionate amount of indigent care (10). This argument suggests that their high costs are caused, in part, by the severity of their case mix, their specialized services, and their roles as providers of care to many people who are uninsured or underinsured.
The aim of the study reported here was to examine the clinical services provided by psychiatric teaching hospitals and their responsiveness to community need. Data for the study are taken from the 1988 National Mental Health Facilities Survey (NMHFS), a national survey of providers of inpatient psychiatric care in the United States. A limitation of this data is that it does not allow examination of the teaching and research activities of the hospitals surveyed. Instead, the study data examines variables related to the provision of clinical services. Clinical services are increasingly the only concerns of third-party payers, but they are only one of the important missions of teaching hospitals.
The data used in this analysis represent the state of psychiatric teaching hospitals at the beginning of the managed care era. Although the data are from 1988, this is the most comprehensive existing database available that can answer important questions about the services provided by psychiatric teaching hospitals. There are no published reports comparing the services provided by psychiatric teaching hospitals with those provided by nonteaching hospitals. This comparison is meaningful because teaching hospitals are now being held accountable for their higher costs (10). They are now being called upon to justify their current subsidies and indicate their value to the government and third parties that pay for their services. In the current environment, government monies are particularly important, because the research and education that teaching hospitals produce are public goods that private payers see as costs that cannot be passed on to their customers (4). Although the the health care marketplace continues to evolve rapidly, the study examines issues that are relevant today and challenges assertions about the role of teaching hospitals that will be important to address in future research (10).
The NMHFS (11), conducted between 1987 and 1988, was designed to address unanswered questions about the organization and financing of mental health services in the United States. The survey consisted of a 200-item questionnaire mailed to administrators of all nonfederal psychiatric hospitals in the United States, including a 75% random sample of psychiatric units in nonfederal general hospitals. The overall response rate was 60%, ranging from 78% for public specialty hospitals to 38% for for-profit specialty hospitals, yielding a total sample of 915. Because of the relatively low response rate for for-profit specialty hospitals, we conducted a comparison of our findings on revenue sources, case mix, and staffing practices of for-profit specialty hospitals with data from national surveys by the National Institute of Mental Health (12) and the National Association of Private Psychiatric Hospitals and found overall consistency. A detailed discussion of sample selection and response rate was published previously (11).
Within this context, we used the NMHFS data to examine differences in services mix, case mix, payer mix and staffing ratios among hospitals that operate as academic medical centers, those that are teaching hospitals but not academic medical centers (AMCs), and those without a psychiatric training program.
To test common assumptions about the role of teaching hospitals, compared with other hospitals, we selected variables that would allow us to examine the breadth and quality of hospitals' clinical services and their responsiveness to community needs. The clinical services indicators included case mix, number and type of services offered, staffing patterns, and several measures associated with quality care. Indicators of response to community need included payer mix and uncollected fees. Control variables included number of psychiatric beds, hospital ownership (private or public), hospital focus (general or specialty), and location in an urban center.
Suicide review conferences, because they are an indicator of ongoing internal review of clinical treatment and procedures, were included as a measure of quality of care. Nursing staff turnover rate was also included as an indicator of quality of care. Stable staffing patterns are expected to result in more experienced staff and indicate an ability to provide the continuity of care needed by patients with mental illness. Psychiatrist-to-patient ratio was included because we expected teaching hospitals to have more professional staff than other hospitals, although this factor may have no direct relationship on quality of care. Our analysis included only staff psychiatrists, so that the comparison of staffing among hospitals excludes any substitution of residents for staff psychiatrists.
The number of specialty psychiatric services offered and the availability of case management were included to reflect both quality of care and community service. An increased number of specialty services may reflect higher quality of patient care. Furthermore, in competitive health care markets there is increasing pressure to offer only services that are profitable and that result in an overall decrease in the number of services (13. Thus, an increased number of services may reflect a greater commitment to serving community needs. Case management is specifically examined because it is widely accepted as the most effective way to ensure continuity of care and to monitor appropriateness of care, but only 25% of providers provide such care (14).
The percent of inpatients with a primary diagnosis of schizophrenia or other psychotic disorders treated in the past year was used as an indicator of a hospital's economic behavior and its commitment to community service. In inpatient settings, severely ill patients are less profitable to treat than patients with less severe illness (15). The hospitals that treat many severely ill patients will generate fewer profits than others, and this fact may justify subsidies (13). Further, willingness to treat many such patients may indicate a hospital's commitment to community service.
The proportion of a hospital's clinical revenues derived from services for Medicaid recipients is a measure of a hospital's willingness to override economic concerns with community service, because in 1988 Medicaid payment rates were low. The proportion of revenues derived from services for Medicare recipients treated at a hospital is used here to examine the relationship between Medicare funding for GME and direct service for Medicare patients.
Uncollected inpatient fees are a measure of a hospital's service to people who are uninsured or unable to pay their entire hospital bill. Previous research has indicated that some hospitals screen patients for ability to pay prior to providing services (8). Subsidies may be appropriate for hospitals that serve people who otherwise would not have access to treatment.
Multivariate analysis of covariance was performed by using the SAS (SAS Institute Inc., Cary, NC) general linear models procedure to examine each of the six continuous dependent variables (physician-to-patient ratio, nursing turnover rate, percent of revenue from Medicaid, percent of revenue from Medicare, percent of patients with schizophrenia, and number of clinical services offered). Hospital type (AMC, other teaching hospital, and nonteaching hospital) was the categorical independent variable of interest. Covariates included number of hospital beds, hospital ownership (public or private), hospital focus (general or specialty), and percent of population in an urban area.
The associations between hospital status (AMC, other teaching hospital, and nonteaching hospital) and whether the hospitals offered case management services or conducted suicide-review conferences were examined by using contingency tables and the chi-square statistic.
t1 presents the results of the analysis of covariance for the continuous dependent variables and the results of the chi-square analysis of the categorical variables representing whether hospitals offered case management and whether they conducted suicide review conferences. The statistical significance of each finding is based on an alpha set at 0.05.
Of the 584 hospitals for which we have data on GME, 118 reported that they operated psychiatric residencies. By using a list of members of the Association of Academic Medical Centers, we determined that 41 of these 118 were AMCs. AMCs are hospitals that are formally affiliated with a medical school and in which medical school faculty instruct resident physicians (10).
Of the three types of hospitals, AMCs had the highest physician-to-patient ratio. The differences were significantly different when compared with other teaching hospitals and with nonteaching hospitals. Nursing turnover rate was significantly higher at the AMCs than at other teaching hospitals. Compared with other teaching hospitals, AMCs treated a lower percentage of patients with schizophrenia. The AMCs were less likely than other teaching hospitals to hold suicide review conferences. Compared with the nonteaching hospitals, the AMCs offered a larger number of specialized clinical services. The AMCs received more of their revenues from Medicaid than did the nonteaching hospitals.
Other (non-AMC) teaching hospitals had a significantly lower physician-to-patient ratio than the AMCs, but a higher ratio than the nonteaching hospitals. Compared with nonteaching hospitals, other teaching hospitals received a lower percentage of their revenues from Medicare and collected a lower percentage of their inpatient charges. Other teaching hospitals offered a larger number of specialized clinical services than the nonteaching hospitals. Other teaching hospitals were more likely than both AMCs and nonteaching hospitals to conduct suicide-review conferences.
In the examination of clinical services, we found evidence that teaching hospitals had some advantages over nonteaching hospitals. AMCs had the highest physician-to-patient ratio, other teaching hospitals had intermediate physician-to-patient staffing ratios, and nonteaching hospitals had the lowest. Both AMCs and other teaching hospitals offered more specialty services than nonteaching hospitals. Other (non-AMC) teaching hospitals were more likely than nonteaching hospitals to conduct suicide-review conferences. On no clinical service variables were nonteaching hospitals superior to either type of teaching hospital. Among the two types of teaching hospital, findings on the clinical service variables were mixed. Compared with non-AMC teaching hospitals, AMCs had higher physician-to-patient ratios, but they also had higher nursing staff turnover and were less likely to conduct suicide-review conferences.
On the variables we examined, the teaching hospitals demonstrated more of a commitment to community service than did the nonteaching hospitals. We found evidence that psychiatric teaching hospitals serve more persons who are poor or underinsured than do nonteaching hospitals. AMCs did this by providing more care through Medicaid, whereas other teaching hospitals provided more uncompensated care.
In the context of uncertainty about the financing of teaching hospitals, others have argued persuasively that dedicated funding to support their social missions is needed (10). This study provides evidence to support claims that teaching hospitals treat more poor and indigent patients than do nonteaching hospitals. There is no evidence, however, that teaching hospitals treat sicker patients than nonteaching hospitals. This finding contradicts the notion that teaching hospitals served the most severely ill patients. Schizophrenia was our proxy for severe mental illness. Although diagnosis is a limited indicator of case mix, the argument that the high costs of psychiatric teaching hospitals are justified by expenses related to case mix is not supported.
Nonteaching hospitals received more of their revenues from Medicare than other (non-AMC) teaching hospitals. This fact is interesting because Medicare is an important source of funding for GME, and because this source of funding is in jeopardy. Like other payers, the Health Care Financing Administration will be held accountable for its expenditures.
This analysis confirms one of the most widely held beliefs about teaching hospitals: that they offer a wide array of services (9). We interpret this increased number of specialized services to mean that teaching hospitals provided their communities with greater opportunity for high-quality care. To the extent that teaching hospitals have a responsibility to train physicians to work in a variety of settings and with a range of modalities, this finding also suggests that teaching hospitals are serving an important role, for which they are subsidized by public funds, nonprofit and for-profit health plans, and philanthropic organizations.
An important limitation of this study is that it did not measure research or training activities, which are important missions of AMCs and other teaching hospitals. These research and education missions of teaching hospitals are essential to continued medical progress and to the maintenance of a highly skilled medical profession, and are worthy of financial support.
This analysis demonstrated that at the beginning of the managed care era, psychiatric teaching hospitals cared for a disproportionate share of poor and underinsured patients. This service may, in part, justify the financial subsidies these hospitals receive. As state mental health authorities continue to close state hospitals and to reduce their capacity, service to the underinsured will continue to be a crucial social mission of psychiatric teaching hospitals.
This work was supported by grants from the National Institute of Mental Health (Grant No. MH—40316 and Grant No. MH—01177).