|
|
| |||||||||||||||||||||||||||||||
Background Lower rates of use of resources have been reported for the treatment of hospitalized patients covered by Medicaid than for privately insured patients. Cost-containment policies may exacerbate such differences in the use of hospital resources. We studied patients with ischemic heart disease who received care at nonfederal hospitals in California in 1983 (the year a Medicaid cost-containment program was implemented), in 1985, or in 1988. Within this sample of patients, we compared the rates of coronary revascularization (coronary-artery bypass surgery or coronary angioplasty) among patients covered by Medicaid, patients with private insurance covering fee-for-service care, and patients enrolled in a health maintenance organization (HMO).
Methods Logistic-regression models were used to determine adjusted odds ratios for the use of coronary revascularization procedures in patients with different types of insurance, with control for demographic, clinical, and hospital characteristics. The study samples were made up of 49,167 patients in 1983, 47,809 in 1985, and 44,631 in 1988.
Results The frequency of revascularization increased in all three insurance groups from 1983 to 1988, but it did so much faster in the fee-for-service and HMO groups than in the Medicaid group. Patients with private fee-for-service insurance were 1.66 times as likely as Medicaid patients to undergo revascularization in 1983 (P<0.01), 2.01 times as likely in 1985 (P<0.01), and 2.33 times as likely in 1988 (P<0.01). Patients enrolled in HMOs were 0.96 times as likely as Medicaid patients to undergo revascularization in 1983 (P<0.05), 1.23 times as likely in 1985 (P<0.01), and 1.53 times as likely in 1988 (P<0.01).
Conclusions The frequency of coronary revascularization in California in 1983 was nearly twice as high for patients with private fee-for-service insurance as for patients enrolled in HMOs or for Medicaid recipients. The implementation that year of stringent cost-control measures by Medicaid may explain the slower increase in the frequency of revascularization over five years among Medicaid recipients as compared with patients in the fee-for-service and HMO groups. Different incentives in fee-for-service and HMO practice may explain the lower frequency of revascularization among patients enrolled in HMOs, although the rates of increase for these two groups were about the same from 1983 to 1988.
Did cost-containment policies implemented in the 1980s encourage lower levels of use of resources for uninsured patients and Medicaid recipients? Cross-sectional analyses of data for a single year have been unable to address the possible association between cost-containment policies and differences in treatment related to type of health insurance. We examined the effects of a Medicaid cost-containment policy on the relative rates of coronary revascularization among patients covered by Medicaid, patients with private insurance covering fee-for-service care, and patients enrolled in health maintenance organizations (HMOs) who received care at any nonfederal hospital in California in 1983, 1985, or 1988.
California implemented a Medicaid cost-containment program in 1983. The Selective Provider Contracting Program required hospitals to compete for state contracts to serve patients covered by Medicaid. Instead of cost-based reimbursement, hospitals received a fixed per diem payment arrived at through competitive bidding and negotiation7. The Selective Provider Contracting Program decreased the rates of reimbursement to hospitals caring for patients covered by Medicaid. Even though there was a real increase of 6 percent in the average cost of a day of hospital care in California between 1982 and 1988, the average real rate of reimbursement by Medicaid decreased by 18 percent. Reimbursement covered only 70 percent of average hospital costs in 1988, as compared with 91 percent in 19828,9,10,11.
The number of Californians enrolled in HMOs increased from about 20 percent of the state population in 1983 to nearly 30 percent by 198812,13. People enrolled in HMOs are an important group with which to compare the treatment of people with Medicaid coverage and fee-for-service insurance after the Selective Provider Contracting Program was implemented. The incentives for a more selective use of health care resources in prepaid health plans, as compared with fee-for-service care, are well documented14,15. These cost-containment incentives are an important element of the current debate about health care reform.
Methods
Collection of Data
We reviewed hospital-discharge abstracts for all nonfederal acute care hospitals in California for the years 1983, 1985, and 1988. The abstracts, collected by the Office of Statewide Health Planning and Development, contain information on the patient's age, sex, race or ethnic group, and ZIP Code of residence; the length of stay; the source and type of admission; the principal procedure and principal diagnosis; the destination of the patient after discharge; the expected principal source of payment; and any secondary diagnoses and procedures (up to four were reported). The number of beds in each hospital was obtained from the Annual Report of Hospitals of the Office of Statewide Health Planning and Development16. Whether hospitals were teaching facilities (defined by membership in the Council of Teaching Hospitals) was determined from the 1989 edition of the American Hospital Association's Guide to the Health Care Field17.
Selection of Patients
Patients with the following principal diagnostic codes (from the International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM]) were selected: myocardial infarction (ICD-9-CM codes 410.0 through 410.9), unstable angina (codes 411.1 and 411.8), angina pectoris (codes 413.0 through 413.9), and chronic ischemic heart disease (codes 414.0, 414.8, and 414.9). Patients in these diagnostic groups may undergo coronary-artery bypass surgery (codes 36.1 through 36.2) or coronary angioplasty (code 36.0), since their principal diagnosis is indicative of ischemic heart disease18.
Patients were included in the sample if their expected principal source of payment was either Medicaid or private insurance (defined as Blue Cross-Blue Shield, a commercial insurance company, or an HMO). The private-insurance sample was further broken down into those with fee-for-service insurance (provided by Blue Cross-Blue Shield or commercial insurers) and those enrolled in an HMO. Results are reported both for all the patients with private insurance and for the subgroups covered by fee-for-service insurance or enrolled in an HMO.
To ensure that the patients with the three types of insurance were clinically similar, we used the following additional inclusion criteria: age of 35 through 64 years, residency in the state of California (as determined by the ZIP Code of residence reported on the patient's abstract), and discharge home or to a site other than another general acute care hospital. Patients older than 64 years of age were excluded, since most elderly patients are covered by the Medicare program and since the older age of these patients might affect their treatment. Admissions that resulted in a transfer to another acute care facility were excluded to avoid double-counting a patient's treatment for a single episode of illness18. A total of 497 hospitals admitted patients who met the selection criteria during the study period.
The patients were classified according to the type of admission as follows: emergency admissions were defined as those in which the patient required immediate hospitalization for the alleviation of severe pain or the immediate diagnosis and treatment of an unforeseen medical condition that might otherwise lead to disability or death; urgent admissions were those in which the patient required hospitalization as soon as possible for medical attention; and elective admissions were routine admissions, the postponement of which would not endanger the patient.
Statistical Analysis
A logistic-regression model with a dichotomous dependent variable indicating whether or not a patient underwent coronary revascularization (either coronary-artery bypass surgery or angioplasty) was used to control for differences among patients in the following characteristics: age (age groups 35 through 44, 45 through 54, and 55 through 64), sex, race or ethnic group (white, black, Hispanic, or other), number of diagnoses (one or two, three or four, or five or more), principal diagnosis, presence of a secondary diagnosis of congestive heart failure (ICD-9-CM code 428.0), presence of a secondary diagnosis of diabetes mellitus (codes 250.0 through 250.9), type of admission (emergency, urgent, or elective), size of the hospital at which treatment took place (<200 beds vs. 
200 beds), and type of hospital (teaching vs. nonteaching). Results from the logistic regressions are presented as odds ratios, with patients covered by Medicaid used as the reference category. An odds ratio greater than 1.0 indicates that patients with private insurance were more likely than patients covered by Medicaid to undergo coronary revascularization, with demographic, clinical, and hospital characteristics held constant. The statistical significance of the year-to-year changes in the estimated odds ratios was evaluated by pooling data from two years and assessing the interaction of all covariates with a dummy variable indicating the year of the observation. The regression coefficients for the interaction of fee-for-service insurance and year and for HMO enrollment and year indicate the magnitude and statistical significance of year-to-year changes in odds ratios19.
A logistic-regression model with a dependent variable indicating whether a patient died during a hospital stay was used to estimate differences in inpatient mortality among the patients with different types of insurance, regardless of whether a patient underwent revascularization, with control for demographic, clinical, and hospital characteristics. The results are also presented as odds ratios with patients covered by Medicaid as the reference category.
All reported P values are two-tailed. A P value of less than 0.05 was considered to indicate statistical significance.
Results
Characteristics of the Patients
For each year we analyzed, patients covered by Medicaid were more likely than patients with fee-for-service insurance or patients enrolled in an HMO to be female, to be black or Hispanic, and to have been treated in a teaching hospital (Table 1). The clinical indicators that are proxies for the severity of illness -- the number of diagnoses, the presence of congestive heart failure or diabetes, and the type of admission -- suggest that patients covered by Medicaid were, on average, more severely ill than the other patients during each year we examined. Medicaid recipients had a greater average number of diagnoses than either privately insured group, and they were more likely to have a secondary diagnosis of congestive heart failure or diabetes in each of the three years. The proportion of Medicaid recipients who were admitted non-electively (urgent or emergency admissions) was greater than the proportion of the fee-for-service sample in each year, and greater than the proportion of the HMO sample in 1985 and 1988.
|
The growth in HMO enrollment during the 1980s is mirrored in the growth in the HMO sample between 1983 and 1988. In 1983, there were 7912 patients enrolled in HMOs who met the inclusion criteria for our analysis. In 1988, the HMO sample was made up of 14,915 patients. Conversely, the size of the fee-for-service sample decreased from 35,558 patients in 1983 to 23,797 patients in 1988.
Use of Procedures
The rate of revascularization (the percentage of admissions for which either bypass surgery or coronary angioplasty was listed as either the primary procedure or as one of the up to four reported secondary procedures) was similar for both patients covered by Medicaid and patients enrolled in HMOs in 1983. Between 1983 and 1988, the rate of revascularization increased significantly more in the HMO group than in the Medicaid group (Figure 1). In 1983, the rate of revascularization for patients with fee-for-service insurance was more than twice the rate for the other groups. This rate also increased significantly over the rate for patients covered by Medicaid between 1983 and 1988. The unadjusted odds ratio for revascularization among patients with fee-for-service insurance as compared with patients covered by Medicaid increased from 2.49 in 1983 to 3.68 in 1988, whereas the unadjusted odds ratio for HMO enrollees as compared with Medicaid recipients increased from 1.01 to 2.48 (Table 2). Separate analysis of the rates of coronary-artery bypass surgery and angioplasty showed similar results (data not shown).
|
|
Outcome
Inpatient mortality remained essentially unchanged between 1983 and 1988 for patients with ischemic heart disease who were covered by Medicaid, regardless of whether they underwent revascularization. There was a slight decrease in the inpatient mortality rate in both the fee-for-service and the HMO groups (Figure 2). Patients insured by Medicaid were significantly more likely than patients with fee-for-service insurance to die in the hospital in each year of the analysis, and they were significantly more likely than HMO enrollees to die in the hospital in 1983 and 1988 (Table 3). The changes in these odds ratios over time were not statistically significant.
|
|
Our results are consistent with the hypothesis that declining real rates of reimbursement by Medicaid and declining opportunities for hospitals to shift the cost of treating patients with Medicaid insurance to private insurers encouraged reductions in the use of resources for coronary revascularization in the treatment of Medicaid recipients as compared with both patients with fee-for-service insurance and patients enrolled in HMOs in California between 1983 and 1988. In 1983, patients with Medicaid coverage were about 66 percent less likely than patients with fee-for-service insurance and 4 percent more likely than patients enrolled in an HMO to undergo revascularization. In 1988, despite an absolute increase in the rate of revascularization for Medicaid patients (as the use of angioplasty became more widespread), patients covered by Medicaid were less than half as likely as patients with fee-for-service insurance and two thirds as likely as HMO enrollees to undergo revascularization.
Our findings also provide evidence in support of a central theme of present efforts to reform health care -- that the health care delivery arrangements of HMOs encourage a less resource-intensive style of medical care than fee-for-service insurance. Nevertheless, although HMO enrollees and patients covered by Medicaid underwent revascularization at similar rates in 1983, there was greater growth in the rate of revascularization in the HMO group over the next five years. A number of factors may explain this trend, including more stringent Medicaid cost-containment efforts, unobserved changes in the demographic and clinical characteristics of patients between 1983 and 1988, and a more rapid diffusion of angioplasty technology among hospitals affiliated with HMOs.
In addition to changes in hospital-reimbursement rates, real decreases in Medicaid reimbursement to California physicians may help explain our findings. These decreases may have contributed to the reluctance of some physicians to care for patients insured by Medicaid. It is noteworthy that a 1983 survey of medical and surgical specialists found that cardiologists were the least likely of all the specialists surveyed to accept patients insured by Medicaid in their practices5,22. If access to specialists within a hospital became more limited for patients covered by Medicaid during the period of our study, it is probable that these patients would have become increasingly less likely to undergo the procedures, such as coronary revascularization, provided by these specialists.
As compared with patients with private insurance, those covered by Medicaid are more likely to delay seeking health care. They may therefore present with more advanced coronary heart disease, which is less operable than the disease at early stages5,23,24. If the tendency for patients covered by Medicaid to delay seeking care became more pronounced between 1983 and 1988, the increasing differences in the rate of use of cardiac procedures might result partly from the greater severity of illness among hospitalized Medicaid recipients.
The measure of resource use in our study was the average number of procedures per admission. Changes in the types of patients who are admitted to hospitals can affect the observed rates of procedures. If the admission of privately insured patients became relatively "selective" over the five-year period, as compared with the admission of patients covered by Medicaid (that is, if patients with private insurance who were admitted in 1988 were more likely than those admitted in 1983 to require revascularization, whereas Medicaid recipients in both years had a similar need for revascularization), then this pattern might explain the differences we observed. To examine such possible confounding, we conducted a separate analysis of patients with a principal diagnosis of myocardial infarction, who had a relatively clear indication for hospitalization. It is less likely that there would be an important change in the threshold for admission among such patients over the five-year period18. In this analysis, we found a trend similar to that in the analysis of the entire sample. The adjusted odds ratio for revascularization among patients with a principal diagnosis of myocardial infarction who had fee-for-service coverage, as compared with those with Medicaid coverage, increased from 2.30 in 1983 to 2.73 in 1988 (P = 0.38), whereas the odds ratio for HMO enrollees as compared with Medicaid recipients increased from 0.61 in 1983 to 1.18 in 1988 (P<0.01). These results are consistent with the hypothesis that decisions about whether to perform coronary revascularization were more strongly associated with a patient's insurance status in 1988 than in 1983.
An important limitation of our study is that we were unable to identify multiple hospital admissions for the same patient, since unique patient identifiers were not included in the discharge abstracts. Such a limitation is especially important in analyses of the treatment of ischemic heart disease. A patient may undergo diagnostic cardiac catheterization during one hospitalization and then be discharged in order to be readmitted electively at a later date for coronary revascularization. If patients covered by Medicaid were more likely than privately insured patients to undergo diagnostic cardiac catheterization and subsequent revascularization during different hospital admissions (rather than during the same admission), that pattern might explain the lower calculated rates of revascularization among patients covered by Medicaid.
To minimize this possible source of bias, we constructed a sample that excluded patients who underwent revascularization, but who did not undergo diagnostic cardiac catheterization (ICD-9-CM codes 37.21 through 37.23 and 88.55 through 88.57) during the same hospital admission. The results of a logistic-regression analysis restricted to this sample were also similar to the results of the analysis of the entire sample; the odds ratio for revascularization among patients with fee-for-service insurance as compared with patients covered by Medicaid was 2.07 in 1983 and 2.67 in 1988 (P<0.01), a 29 percent increase in the odds ratio, whereas the odds ratio for revascularization among HMO enrollees as compared with Medicaid recipients was 0.63 in 1983 and 1.59 in 1988 (P<0.01), a 152 percent increase.
During each year we studied, patients covered by Medicaid were more likely than patients with private insurance to die in the hospital. The difference in inpatient mortality increased between 1983 and 1988. These results must be interpreted with caution, however, given the limited clinical information available on the patients and the lack of post-discharge mortality data. Although we found a statistically significant association between the type of insurance and inpatient mortality, it is difficult to ascertain the clinical importance of the difference -- 1.5 percentage points -- in death rates between Medicaid-insured and privately insured patients in 1988. The statistical association of an increase in the difference between the groups in the rate of coronary revascularization and an increase in the difference in inpatient mortality does not establish a cause-and-effect relation.
On the basis of our data, we cannot exclude the possibility that patients covered by Medicaid were no worse off, or even were better off, than others as a result of their lower rate of coronary revascularization. Inappropriate cardiac surgery has been documented25. The rate at which "discretionary" procedures are performed among patients enrolled in HMOs is lower than that among patients with fee-for-service insurance,15 perhaps indicating a more appropriate use of procedures in HMOs. Studies that review medical records to determine the appropriateness of revascularization procedures are necessary to assess whether the differences we observed represent underutilization for patients covered by Medicaid or overutilization for patients with fee-for-service or HMO coverage.
Supported by the Pew Charitable Trusts and by a grant (HS 06948-01) from the Agency for Health Care Policy and Research.
We are indebted to Christine Cassel, M.D., David Dranove, Ph.D., Arnold Epstein, M.D., Edward Lawlor, Ph.D., David Meltzer, M.D., Ph.D., Katie Merrell, James Robinson, Ph.D., Patrick Romano, M.D., and Robert Willis, Ph.D., for their helpful comments on earlier drafts of the manuscript and to the Center for Health Administration Studies at the University of Chicago for providing data.
Source Information
From the Harris Graduate School of Public Policy Studies (K.M.L.) and the Pritzker School of Medicine (K.M.L., E.J.S.), University of Chicago, Chicago. Presented in preliminary form at the national meeting of the Society of General Internal Medicine, Washington, D.C., April 30, 1992.
Address reprint requests to Dr. Sussman at Lehigh Valley Hospital, Cedar Crest and I-78, P.O. Box 689, Allentown, PA 18105-1556.
References
| |||||||||||||||||||||||||||||||
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
