University at Buffalo - The State University of New York
Skip to Content
Official UB news and information for the media

Shortening In-Hospital Rehabilitation Doesn't Diminish Treatment Effectiveness, But Patients Die Sooner, Study Finds

Causes of increased mortality after discharge remain unknown

By Lois Baker

Release Date: October 12, 2004

BUFFALO, N.Y. -- A reduction by health insurance carriers in the number of treatment days covered in medical rehabilitation hospitals by nearly 40 percent during the past decade didn't diminish treatment effectiveness, a new study has found.

However, the analysis revealed a disturbing trend: Patients were not living as long after discharge from the hospitals.

Researchers from the University at Buffalo and University of Texas Medical Branch analyzed records of 148,897 patients treated in 744 inpatient medical rehabilitation hospitals in the U.S. between 1994 and 2001.

They found that deaths between discharge and a 6-month follow-up increased from less than 1 percent in 1994 to 4.7 percent in 2001. Results of the study appear in the October 13 issue of JAMA.

Carl V. Granger, M.D., professor in the Department of Rehabilitation Medicine in the UB School of Medicine and Biomedical Sciences and senior author on the study, said the increase in mortality over the study period was unexpected and needs to be monitored.

He said researchers have no concrete explanations for the earlier deaths and cautioned about misinterpreting the study's results.

"We don't have enough data on other diseases, medical events and treatments that occurred either in the acute care hospital or in the rehabilitation hospital," Granger said. "Patients weren't randomly selected, so the data are subject to selection bias. In addition, this is the first time these data have been brought together and analyzed to this degree. Further analysis may produce different results.

"There have been dramatic changes throughout health care over the past decade, and all the factors affected by these changes are not included in the databases we examined," continued Granger. "Plus, we don't know the settings in which patients died or how soon they died after rehabilitation hospitalization."

He emphasized that this analysis includes data from rehabilitation hospitals only, because they routinely perform follow up assessments after discharge. The outcomes for patients who received rehabilitation in other settings, such as home care, nursing homes, or outpatient clinics are unknown because they aren't monitored systematically.

Kenneth J. Ottenbacher, Ph.D., formerly of UB and now at the Sealy Center on Aging, University of Texas Medical Branch in Galveston, Texas, is first author on the study.

Data for the study were drawn from the Uniform Data System for Medical Rehabilitation at the University at Buffalo. UDSMR, which was established in 1983, is the largest national registry of standardized information on medical rehabilitation in the U.S. Granger is one of its founders.

The study's goal was to find out how changes in reimbursement, which reduced the number of days insurers would pay for services in medical rehabilitation hospitals, have affected patients' functional status, living setting and mortality after discharge.

The analysis covered length of stay for five major impairment groups – stroke, brain dysfunction, other neurological conditions, spinal cord dysfunction and orthopedic conditions – from 1994 through 2001. The study also assessed changes in rehabilitation effectiveness, treatment efficiency (how quickly patients reached a certain functional status), living setting, and mortality between discharge and follow-up.

Functional status was based on the patient's Functional Independence Measure, or FIM™ score, a standardized instrument for quantifying everyday living skills, which was developed at UB by Granger and colleagues. A FIM™ score was obtained at admission, at discharge and at follow-up, which took place between 80 and 180 days after discharge. Living setting, also identified at both discharge and follow-up, was defined as "home" or "not at home."

Results showed that the length of stay for all impairment groups combined had decreased from a median of 20 days in 1994 to 12 days in 2001. Effectiveness, the difference between admission FIM™ and discharge FIM™, decreased slightly, but Granger said the change would not be clinically significant.

FIM™ scores at admission remained stable over the seven years, results showed, while treatment efficiency improved significantly for all categories of impairment. There was no substantial difference from 1994 to 2001 in the percentage of patients who were discharged to their homes or in the percentage who were at home at follow-up.

Most significantly, the mean number of patients who died between discharge and follow-up increased in all impairment groups, results showed. Mean mortality increased from less than 1 percent to 4.7 percent overall. The largest increases in mortality occurred in patients admitted for brain and spinal cord dysfunctions, while deaths in orthopedic patients remained fairly stable.

There was no significant change in age at admission (patients admitted in 2001 were actually 2 years younger, on average, than in 1994), or the number or type of additional health problems.

Granger said the researchers don't have an explanation for the increase in death. "Our examination for the possibilities that older age and/or more severe disability were a factor showed these variables to be rather stable, so they would not account for increased mortality. In future research we will try to identify other variables that might help to explain these findings."

Researchers on the study, in addition to Granger and Ottenbacher, were Sandra B. Illig and Pam M. Smith, D.S.N., who conducted the follow-up interviews while with National Follow-up Services in Buffalo; Glenn V. Ostir, Ph.D., of the Sealy Center on Aging; and Richard T. Linn, Ph.D., of UDSMR and the UB Department of Rehabilitation Sciences.

The study was supported by grants from the National Institutes of Health to Ottenbacher.