Ni H, Nauman DJ, Hershberger RE. Managed Care and Outcomes of Hospitalization Among Elderly Patients With Congestive Heart Failure. Arch Intern Med. 1998;158(11):1231-1236. doi:10.1001/archinte.158.11.1231
Copyright 1998 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1998
Little was known about the impact of the health maintenance organization–managed care on patients hospitalized for congestive heart failure. Understanding this issue is important with regards to the increasing prevalence of congestive heart failure among the elderly population as well as the growing enrollment of Medicare beneficiaries in managed care.
To examine the impact of the health maintenance organization–managed care on the outcomes of hospitalization among patients with congestive heart failure.
Patients and Methods
We analyzed the Oregon hospital discharge data set. Study subjects were all patients with congestive heart failure aged 65 years or older (N=5821) discharged from hospitals in 1995 and classified into 6 insurance groups: managed care, Medicare, Medicaid, commercial or private insurance, self-pay, and other.
The percentage of patients admitted to hospitals via emergency departments was significantly higher in the managed care patients (69%) than in other health insurance coverage groups (29.0%-58.5%; P<.001). After adjusting for age, sex, and comorbidity, the managed care patients experienced a similar length of hospital stay (3.6 days) as the commercial or private insurance patients (3.7 days; P=.67), but a shorter length of hospital stay than the Medicare patients (4.0 days; P<.001), self-pay patients (4.5 days; P<.001), and other patients (4.8 days; P<.001). No difference in the in-hospital mortality rate was seen among the insurance groups (P=.37). The readmission rate was slightly higher in managed care patients (9.1%) than in commercial insurance patients (6.8%) and Medicare patients (7.5%). The differences, however, were not statistically significant after adjusting for the confounding factors (P=.59).
Our results suggest no association between managed care and poor short-term outcomes of hospitalization in patients with congestive heart failure. Attention, however, needs to be paid to the increased use of emergency departments by managed care patients.
CONGESTIVE heart failure (CHF) has remained the leading cause of hospitalization among people aged 65 years or older,1- 3 and consequently has imposed a large financial burden on patients, their families, and health care systems. To improve the quality of care for heart failure while containing the cost of care, it is important to understand factors associated with accessibility of care and short-term outcomes of CHF in hospital settings. Previous studies4,5 have reported that patients with different types of health insurance but similar medical problems may receive different treatment strategies and intensity of treatment, which may result in differences in short-term and long-term outcomes. Specifically, concerns have been raised regarding the quality of medical care received by managed care patients.6- 9 Although a number of studies10,11 have suggested that managed care patients experienced shorter hospital stays, whether managed care patients received substandard care that resulted in poorer outcomes has remained controversial. Additionally, to our knowledge no prior study has examined the impact of the health maintenance organization (HMO)–managed care on patients hospitalized for CHF. Understanding this issue is important with regards to the increasing prevalence of CHF among the elderly population as well as the growing enrollment of Medicare beneficiaries in managed care.12- 14
The objective of this study, therefore, was to examine the impact of HMO–managed care on the outcomes of hospitalization among patients with CHF by analyzing the 1995 Oregon hospital discharge data set. We compared the HMO–managed care patients with other health insurance coverage groups regarding their type of admission, length of hospital stay, in-hospital mortality rate, and readmission rate within 3 months after discharge.
We conducted a retrospective cohort study among patients who were 65 years or older and discharged from a hospital in 1995 with a primary diagnosis of CHF. Patients were identified by the International Classification of Diseases, Ninth Revision (codes 428.0-428.9)15 from the Oregon State hospital discharge index data set maintained by the Oregon Association of Hospitals and Health Systems (OAHH). The OAHH collects data on inpatients discharged from all hospitals in Oregon. The OAHH requires hospitals to submit the paper UB-92s (Uniform Billing form modified in 1992) monthly and submit computer media (magnetic tape or PC diskette) quarterly. After submission, the hospitals receive periodic reports that list those patient records and/or data elements that did not meet the criteria. On receipt of the report, hospitals can make corrections by contacting the OAHH within 10 days.
The hospital discharge database contains information on patients' age; sex; admission and discharge dates; type of admission (ie, emergency department, urgent, elective, newborn, and other); patients' discharge status (routine, transfer to another short-term hospital, transfer to skilled nursing facility, transfer to intermediate care facility, transfer to home health care services, and dead); length of hospital stay; diagnosis related group; principal discharge diagnosis (the condition that was considered to be responsible for the patient's admission); up to 8 secondary diagnoses; up to 6 procedures; payer (Medicare, Medicaid, commercial insurance, HMO–managed care, self-pay, workers compensation, and other), and hospital charges.
Based on the patients' primary payers at the time of admission, patients were classified into 6 health insurance coverage groups. These groups included managed care (HMO–managed care, including Medicare managed care); traditional Medicare (use Medicare throughout this article); Medicaid (HMO and non–HMO health plan, payment covered by governmental source and provided to medically indigent patients); commercial or private insurance (payment for hospitalization was provided by nongovernmental sources and covered under traditional indemnity plans or a commercial carrier); self-pay (payment for hospitalizations was expected from the spouse, family, or next of kin), and an "other" group consisting of miscellaneous payers such as workers compensation and division of health services. Since our study subjects are patients aged 65 years or older, we assume the majority of managed care patients are Medicare beneficiaries.
Outcome variables include type of admission, length of hospital stay, in-hospital mortality rate, and hospital readmission for CHF within 3 months of discharge.
Since the hospital discharge data do not track individuals, readmission to the hospital of the same person is counted as a separate event. Because of confidentiality, the key identifiers such as name and Social Security number were erased from the data we obtained. Therefore, we matched the admission records by each patient's age, sex, and ZIP code to identify patients who were readmitted to a hospital within 3 months after discharge.
For a patient having more than 1 discharge record in the database because of a hospital transfer, we selected his/her first admission for CHF to compare the type of admission between different health insurance coverage groups. The last discharge record, however, was used as an index admission when estimating readmission rate.
In-hospital mortality rate was calculated by dividing the number of individuals who died in hospitals by the total number of discharged patients.
Statistical analysis was performed using the SPSS (SPSS Inc, Chicago, Ill) for Windows software. A 2-sided P≤.05 was required for statistical significance. Based on the statistical power calculation, we had enough sample sizes to detect a difference between the comparison groups for the variables of interest. We used multiple linear regression analysis to assess the association between Medicare managed care and length of hospital stay while controlling confounding factors such as age (defined as 65-69 years, 70-74 years, 75-79 years, 80-84 years, and ≥85 years), sex, and comorbidity index. We developed a modified Charlson Comorbidity Index using the method suggested by Ghali et al,16 which assigns study-specific weights to the original Charlson comorbidity variables. To derive the study-specific weights for the Charlson comorbidity, we developed a stepwise logistic regression model based on the 1995 hospital discharge data with in-hospital death as a dependent variable and comorbidities as independent variables. The analysis was restricted to patients who survived the index hospitalization. Comorbidities remained in the model if P<.10. The resulting odds ratios (ORs) were used to generate integer weights. That is, an OR of less than 1.2 was discarded, an OR of 1.2 to 1.5 received a weight of 1, an OR of 1.5 to 2.5 received a weight of 2, an OR of 2.5 to 3.5 received a weight of 3, and so forth. This method was shown to better predict in-hospital mortality rate than the original Charlson Index.16
In addition, the multiple logistic regression method was used to examine the association of Medicare managed care with type of admission (via emergency department or not), in-hospital mortality rate, and readmission for CHF. In this analysis, the managed care group was used as a reference for the health insurance variable; the outcomes of the managed care patients were compared with patients with other health insurance coverage. Confounding factors considered in the regression analysis included age, sex, and comorbidity index. We calculated ORs to measure the magnitude of the association. Ninety-five percent confidence intervals for the adjusted OR were based on the Cornfield method.17 The goodness-of-fit of the model was tested by the likelihood ratio test.
In total, 5821 patients were hospitalized for CHF in Oregon during 1995. Table 1 shows the demographic characteristics and coexisting illnesses of these patients. The majority of patients (72%) were 75 years or older and more than half (58.7%) were women. Of all discharges, approximately one quarter of the patients had diabetes, the most common comorbidity among the patients with CHF. Other comorbidities included chronic pulmonary diseases (10.4%), renal disease (9%), vascular disease (7%), myocardial infarction (3.9%), and cancer (1.0%).
Table 2 compares prior characteristics by insurance coverage groups. The managed care patients were older than the Medicaid (P=.03) and commercial or private insurance patients (P=.02), but were comparable with other insurance coverage groups (P=.25). No difference was found in sex distribution among different health insurance coverage groups (P=.56). The percentage of patients having comorbidity index of 1 or higher was slightly lower in the managed care patients than the Medicare patients (9.8% vs 13.5%; P=.01), but was similar to the patients with private or commercial insurance (9.8% vs 10.7%; P=.56).
Table 3 shows that approximately half of the patients with CHF were admitted to a hospital via emergency departments. Compared with the other health insurance coverage groups, Medicare managed care patients were much more likely to enter a hospital via emergency departments (P<.01).
Separate analyses were performed in all patients as well as in those discharged alive to compare the length of hospital stays between the health insurance groups. In all patients, the average length of hospital stay was 3.6 days for the managed care patients, 4.0 days for other Medicare patients, 3.4 days for Medicaid patients, 3.6 days for commercial or private insurance coverage, 4.3 days for self-pay, and 4.6 days for others. Restricting the analysis to patients discharged alive did not significantly change the average length of hospital stay.
Medicare managed care patients experienced a significantly shorter length of hospital stay than other Medicare patients and self-pay patients after controlling for age, sex, and comorbidity index by the multiple linear regression analysis (Table 4). The analysis restricted to the patients discharged alive generated consistent results.
Overall, 287 patients with CHF (4.9%) did not survive hospitalization. Of those who died, 62 patients (21.6%) were younger than 74 years and 53 patients (18.5%) had renal disease. Approximately half of the deaths occurred within 2 days after admission to a hospital.
The unadjusted in-hospital mortalities were comparable among all health insurance coverage groups except for self-pay and other insurance groups (Table 3). The results from multiple logistic regression analyses revealed no statistical differences in the risk of dying during the hospital phase among the 6 health insurance coverage groups after controlling for confounding factors such as age, sex, and comorbidity index (Table 5).
For the purpose of this analysis we excluded the 278 patients who died in a hospital. In total, 446 patients were readmitted to a hospital because of CHF within 3 months after discharge, generating a readmission rate of 8.0%. The unadjusted health insurance-specific readmission rates were similar (Table 3). Adjusting for the potential confounding factors using multiple logistic regression analysis did not alter the results.
The most significant finding from our study is the increased hospital admission via emergency departments among patients with CHF with managed care insurance coverage. As our study population is 65 years or older, our results indicate a possible lower accessibility of care among patients with CHF in Medicare managed care (HMO Medicare) than those in traditional or fee-for-service Medicare (Medicare) and other types of insurance or payment mechanisms. To our knowledge, this association was not previously reported.
The reasons for such a high percentage of hospital admissions via the emergency department are unclear, but may be partly attributable to fiscal incentives to clinicians of HMO Medicare health care systems. Patients with CHF, like those with other chronic medical illnesses, may require more frequent physician visits for optimal management. Reimbursement to HMO Medicare is capitated on a per-patient per-month basis. Although the fiscal incentives of such a capitated HMO Medicare health plan are clearly to improve the quality of outpatient care and thereby decrease hospital and emergency department utilization, the physicians and other health care providers of the plan may have little individual fiscal incentive to achieve this end. For example, providers may not receive reimbursement directly related to the frequency of outpatient care or the severity or comorbidity of chronic illnesses such as heart failure. As a result, patients with CHF in HMO Medicare may be less likely to receive frequent and preemptive outpatient visits compared with fee-for-service Medicare where provider reimbursements are directly related to the frequency of outpatient visits. Hence, it is possible that patients and/or providers in this article may have substituted an emergency department visit for an office visit even though the use of emergency departments will increase the overall cost of care.
Alternatively, a different fiscal managed care incentive may have lead to increased HMO Medicare emergency department utilization. Managed care exerts strong pressure to reduce hospitalization rates, one of the most costly components of health care delivery. Indeed, recent evidence10 has suggested a lower hospitalization rate among managed care patients. Heart failure is a chronic progressive illness with considerable mortality and morbidity.18,19 Even if the frequency of outpatient visits was appropriate for the HMO Medicare beneficiaries described in this article, in an effort to curtail hospital utilization it is possible that these patients with CHF may not have received timely, preemptive hospital admission from the outpatient clinic, and therefore may have appropriately sought urgent hospital admission via the emergency department.
Another possible explanation is that Medicare managed care patients were less likely to be able to access to their primary care physicians' office because of financial or other social economic factors. This explanation is reasonable only if Medicare beneficiaries with lower social economic status are more likely to choose a managed care plan. We have searched the literature and found no data to support this contention.
Other more complex issues may also have affected the provision of care to patients with CHF in managed care. We surmise that these issues relate to the most appropriate methods to manage chronic disease in HMO Medicare or any other managed health care delivery system. Even though we in our Oregon Heart Failure Project in Portland as well as other pilot projects in managed care entities use a comprehensive outpatient management strategy for our patients with CHF centered around a team of specially trained nurses and physicians, we speculate that the difficulty observed with increased hospitalization of patients with CHF via the emergency department reflects an emerging agenda of managed care—that of chronic disease management. Even though the fiscal incentives of an HMO Medicare plan are to provide comprehensive outpatient management of a chronic disease such as CHF, the systems and infrastructure for providing this care, much less the economic incentives, may not yet be present to achieve this end.
Since 1980, the Health Care Financing Administration has encouraged the enrollment of Medicare beneficiaries in managed care as an alternative to fee-for-service care. As of January 1, 1997, more than 4.9 million Medicare beneficiaries were enrolled in managed care plans, accounting for 13% of the total Medicare population.20 The majority of beneficiaries enrolled in HMO plans live in Oregon, California, Florida, New York, Arizona, and Hawaii. The enrollment of Medicare managed care in Oregon dramatically increased from 12% in 1991 to 23% in 1995.21 Despite this increasing trend, a recent report22 showed a disenrollment rate as high as 42% in some health care plans in Medicare HMO markets. Only 1 study23 evaluated the quality of care received by patients with CHF in a Medicare HMO, but did not find differences in the routine management of CHF between Medicare HMO patients and fee-for-service patients. In Oregon, no data are available regarding the quality of care and satisfaction of care among Medicare HMO beneficiaries.
Like previous studies,24- 26 we found that patients enrolled in managed care plans had the shortest length of hospital stay among all health insurance coverage groups except for the Medicaid patients. One possible explanation is that differences in the average lengths of stay observed may be related to the pressures placed on the health care system by insurers for early discharge. For example, utilization review by managed care organizations or nurse reviewers may have helped to facilitate movement of patients to alternative, less costly settings where additional services could be rendered. However, we did not find a difference in discharge status between Medicare managed care and other health insurance coverage plans. It was believed that the managed care companies have attracted healthier patients than the other health care plans. Since we controlled for a number of factors that may affect the outcomes of hospitalizations due to CHF, the association between the managed care and length of hospital stay is unlikely to be confounded by a difference in health status between health insurance plans.
The rate (7.3%) of hospital readmission for patients with CHF estimated in our study is slightly lower than the estimation (7.8%) reported by Krumholz et al3 who examined hospital readmission rates by analyzing the 1991 to 1994 Medicare administrative data in Connecticut. The difference may be attributable to the variation in the follow-up period (3 months vs 6 months) after hospital discharge or a possible difference in the distribution of patients' age and severity of CHF between these 2 states. Their study also observed that heart failure only accounted for one quarter of the discharge diagnoses and many diagnoses were responsible for the repetitive readmissions among patients with CHF.3 Applying this ratio to our study, the overall readmission rate among patients with CHF in Oregon would be 29.0%.
Several limitations of the study warrant discussion. First, a percentage of Medicare managed care patients might be misclassified into traditional Medicare coverage when a patient's insurance status was not clearly stated. According to a report from the Health Care Financing Administration21 the percentage of Medicare managed care enrollees was 23% in Oregon by the end of 1995, which is higher than an estimate based on our data (651  of 5048). Nevertheless, whether the overall enrollment of Medicare managed care represents the enrollment among patients with CHF is questionable. On the other hand, this misclassification, if it exists, would only result in an underestimation of the effect of managed care on outcomes of hospitalization among patients with CHF.
Second, we were unable to differentiate between different types of managed care organizations involved in the treatment of these patients. The managed care plans serve Medicare beneficiaries through 2 types of contracts: risk and cost plans. The risk plans receive a monthly payment (per patient per month) from the Health Care Financing Administration that is used to provide the necessary services to its Medicare members. If health care expenditures exceed payments, the plan must absorb these excess costs. Approximately two thirds of Medicare beneficiaries in HMO–managed care are in risk plans.21 The Medicare cost plan beneficiaries are not limited to receive care from the plan. However, the member must receive or coordinate his/her care through the plan's primary care physician, otherwise traditional fee-for-service Medicare will be billed. However, we cannot provide data on differences in the outcomes of hospitalization of patients with CHF between these 2 types of plans.
Another limitation resulted from the lack of a patient identifier in the hospital discharge data set we obtained for the reason of confidentiality. By linking the data with 3 variables such as age, sex, and ZIP code to identify readmissions, we may miss the patient who relocated after his/her index admission. Additionally, we did not account for readmissions outside the state. Therefore, we may underestimate the hospital readmission rate for patients with CHF.
The quality of hospital discharge data has been questioned. A recent study conducted by Green and Winfeld27 in California found that "urgent admission" had been miscoded in 45.5% of records because of problems distinguishing between urgent and emergent admission. Assuming the similar problem existed in Oregon, the percentage of emergent admission in our study may be overestimated. This misclassification, however, would be nondifferential because there is no evidence to believe that the type of health insurance coverage may have affected the admission classification.
In conclusion, Medicare managed care patients have a high risk of emergent admission and a shorter stay in hospitals. Yet, no difference was found in in-hospital mortality rate and hospital readmission rate within 3 months of discharge. Because of the increasing prevalence of CHF in the elderly population as well as the growing enrollment of Medicare managed care, further studies are needed to clarify the impact of Medicare managed care on the quality of CHF care.
Accepted for publication November 11, 1997.
Correspondimg author: Hanyu Ni, PhD, MPH, Oregon Heart Failure Project, Heart Failure Treatment Program, Division of Cardiology, UHN62, Oregon Health Sciences University, 3181 SW Sam Jackson Park Rd, Portland, OR 97201-3098.