Economic Grand Rounds: Physician or Payer: Who Determines Length of Stay?

Inpatient psychiatry currently operates in a business environment of both managed and unmanaged payers. The conventional wisdom is that the nature of the payer determines length of stay, with managed care leading to shorter stays. After all, is that not what the insurance revolution of the past decade was supposed to accomplish?

Nevertheless, the clinical reality of patient care may be more complex than suggested by this gross simplification. In settings that have resources to deliver specialized patient care—for example, by diagnostic groups—clinical demands begin to compete more effectively with economic pressures, and variances in length of stay emerge. Length of stay may be predetermined in such instances by "adverse case selection"—that is, the very existence of specialty programs implies referrals of patients for whom conventional treatment may already have failed. However, even within such programs, the question arises as to whether the presence or absence of a managed care payer determines length of stay.

Our health system became interested in this question in the context of trying to understand the impact of managed care on clinical care. Sheppard Pratt Hospital operates about 150 inpatient beds and has about 5,300 admissions a year. The hospital is organized into specialty programs, including geriatric, child and adolescent, trauma disorders, psychotic disorders, substance abuse, and managed care.

Patients are triaged to these programs through an admissions process that essentially weighs the demands of the payer against the patient's clinical need and finds the right fit between the two. Patients are sometimes readmitted—as is currently the case in many hospitals and programs—and it is possible for a patient to be treated in various programs over time, either because of changes in clinical need as perceived by the clinical staff or the managed care reviewer or because of changes specific to the payer, such as when patients exhaust one form of health insurance and move on to another.

We were curious about how the role of the payer might affect length of stay in this setting. Would the length of stay for specialty programs vary by specialty? Would there be a significant difference in length of stay within programs, depending on the payer?

Study design

To approach this problem, we compared two treatment settings in our hospital for patients with a diagnosis of schizophrenia or schizoaffective disorder. We chose this diagnostic group because of its significance, in terms of both the number of patients and the severity of illness, and because it highlights some of the more fundamental tensions in the managed care model. Unlike most members of employer-sponsored health plans who have short-term illnesses that lend themselves to the "crisis" model of managed care, patients who have schizophrenia typically are indigent, chronically ill, and treatment refractory. Nevertheless, such patients are treated in managed care settings, including in programs in our hospital.

Patients with schizophrenia who are between the ages of 18 and 65 years are admitted either to the specialty psychotic disorders unit or to one of two other general adult units, both of which have a very short stay for managed care patients. Patients who are admitted to either of these settings can have managed or unmanaged health care insurance. The managed care unit is a 20-bed program in which the average length of stay is six days. Most patients have commercial managed care payers and are often specifically referred to the unit by managed care case managers, with whom the clinical staff of that program have cultivated a working relationship. The treatment model of the managed care unit is classic crisis stabilization, and the team prides itself on delivering high-quality care within the constraints of sometimes very limited resources.

The substance abuse unit is similar in orientation to the managed care unit but has a more explicit and more fully developed focus on chemical dependency and dual diagnosis. Practically speaking, there is a great deal of overlap between the two populations.

The psychotic disorders unit, on the other hand, was explicitly designed to treat severely disabled and chronically ill patients with schizophrenia, neuropsychiatric disorders, and developmental disorders. This population, once served primarily by the public sector in state and county hospitals, has become increasingly integrated with the mainstream of general hospital psychiatry as treatment and payer systems have evolved. It is often a treatment-refractory population with multiple axis I, II, and III diagnoses and multiple severe social problems (axis IV). The cognitive treatment approach of crisis stabilization typically does not lend itself well to this population because of the initial gross disorganization of the patient and the residual enduring cognitive deficits that are evident even as the patient recovers from the acute psychosis. The psychotic disorders unit treats 39 percent of the patients with schizophrenia who are admitted to our hospital.

We compared the length of stay of patients with schizophrenia or schizoaffective disorder in the hospital as a whole and between the two settings. All inpatients who were admitted with a primary or secondary DSM-IV diagnosis code of 295.xx were identified between April 1999 and April 2000. Patients were categorized as having managed or unmanaged care. Managed care patients were defined as those with commercial insurance that provided care through either a health maintenance organization or a psychiatric carve-out to a managed care company. Patients with unmanaged care were those who were covered by unmanaged commercial insurance, Medicare, or Medicaid (Maryland Medical Assistance) or who paid for their own health care.

In Maryland, Medicaid has been semiprivatized to a commercial managed care group called Maryland Health Partners, which is owned by Magellan, a carve-out behavioral managed care company. Maryland Health Partners manages the Medicaid mental health resources without a fiscal incentive to deny care—that is, the group is paid an administrative fee only. Although Maryland Health Partners reviews cases closely and is fully capable of denying benefits for a particular level of care or a particular treatment plan, as a practical matter it has functioned in a supportive and nonintrusive manner for this very ill population, presumably because it operates with a set of clinical guidelines that is more closely congruent with the clinical realities of severe and persistent mental illness. Also, because state law mandates a 30-day cap on length of stay for Medicaid patients in a freestanding "institution for mental disorders," this managed care group's exposure to extreme loss is somewhat limited. For these reasons, we considered the Medicaid patients to be in the unmanaged care category.

Results

For the hospital as a whole, a total of 623 (12 percent) of all patients admitted during that period had a DSM-IV diagnosis code of 295.xx. The patients' mean±SD age was 40.7±14.6 years, ranging from 8 to 88 years, although only 20 patients were younger than 18 years. A total of 342 patients (55 percent) were male. The average length of stay for this total sample was 13 days.

A total of 126 (20 percent) of these patients received managed care. The mean±SD age of the managed care group was 37±11 years, and their average length of stay was 12 days. The unmanaged group had a mean±SD age of 42±14.5 years and an average length of stay of 14 days, although the difference was not statistically significant.

To better answer the questions we had posed, we analyzed the data for the psychotic disorders unit separately and compared the length of stay for patients who had a diagnosis code of 295.xx. Adjusting for the 18- to 65-year age range in the psychotic disorders unit reduced the sample by 10 percent to 561, effectively eliminating the child and adolescent and geriatric services and limiting the comparison to one between the psychotic disorders unit and the two managed units—the managed care unit and the substance abuse unit.

The psychotic disorders unit had 245 patients with schizophrenia. The mean±SD age of these patients was 39±11 years, 140 (57 percent) were male, and 42 (17 percent) received managed care. The other units treated 315 patients with schizophrenia. Their mean±SD age was 39±11.2 years, 170 (54 percent) were male, and 75 (24 percent) received managed care.

Patients' data were then analyzed according to payer within these overarching treatment programs. Because the psychotic disorders unit has some treatment-refractory patients with atypically long stays, the median length of stay was examined instead of the mean, and nonparametric significance testing was performed. On the psychotic disorders unit, the median length of stay for patients in the unmanaged care group was 13 days, compared with 12 days for patients in the managed care group. The difference was not significant. On the other units, the median length of stay was seven days for patients in both the managed and the unmanaged care group.

These results suggest two striking points. First, having unmanaged benefits does not necessarily lead to overuse of those benefits. Length of stay was comparable within programs regardless of insurance type. Second, clinical necessity can significantly affect length of stay, regardless of whether the patient has managed or unmanaged health care insurance. Patients in the program with adverse case mix selection had longer stays for both the managed and the unmanaged care group, and length of stay under managed care was lower by only one day out of 13 (about 8 percent).

Discussion

What can we glean from these findings? Data of this kind are admittedly highly tentative and represent only a preliminary exploration into this complex area. We are aware of the multiple confounds involved, including lack of outcome measures other than length of stay, nonrandom assignment of patients, a mix of primary and secondary diagnoses, and differences in aftercare options that might directly influence length of stay, to mention only a few.

It would be instructive to look at similar data across other settings, such as general hospital psychiatry units, or to compare public and private hospitals. Nevertheless, the data do have a certain real-world quality and are probably comparable with the data that most managed care organizations and public health agencies use to make decisions about resource allocation. Hence this vignette from a busy institution for mental disorders in the public and private trenches is perhaps revealing.

Managed care has certainly succeeded in its initial task of reducing costs by defining "medically necessary" health care in a very narrow sense to mean short-term crisis stabilization (1). We are all now operating in a more cost-efficient framework, with even specialty units keeping their mean length of stay well below the 30-day ceiling theoretically mandated by Medicaid. Nevertheless, patients' needs continue to be a reality, and as managed care attempts to embrace the entire spectrum of severe psychiatric disorders, not just those typically seen in the commercially covered population, we find that there are clinical limits to reducing length of stay, whether care is managed or unmanaged.

Given these realities, the question arises as to whether we still need the added cost and burden of daily utilization review and case management. This example suggests that physician practice may have been sufficiently influenced by the changes of the past decade and that we can now move on to a more collaborative, patient-focused relationship. The dollars diverted from expensive managed care overheads could be redirected toward more clinically urgent needs. These needs might include expanding coverage for the uninsured, improving the quality of services in rehabilitation and residential settings—which is so critical to preventing relapse and fostering eventual independence, both of which could also reduce costs —or directing additional resources toward that intractable but very real group of patients who remain treatment refractory and who consistently exceed the average length of stay, even for their particular diagnoses. There is still much work to be done.

Dr. Boronow is affiliated with Sheppard Pratt Health System, 6501 North Charles Street, Towson, Maryland 21285 (e-mail, [email protected]). Steven S. Sharfstein, M.D., is editor of this column.