October 18, 2011

Predicting hospital readmission is risky business: study

Each year millions of Americans return to the hospital within 30 days of their previous discharge. Although many readmissions could be preventable, most statistical models for predicting them “perform poorly,” according to researchers at Vanderbilt and the Oregon Health and Science University and their affiliated VA medical centers.

Operating Room
(iStock)

Each year millions of Americans return to the hospital within 30 days of their previous discharge.

Many of these readmissions are potentially preventable. Avoiding them not only would be better for patients, but could save the nation’s health care system billions of dollars a year.

Yet most current statistical models for predicting readmission risk “perform poorly,” according to a study published in the Oct. 19 Journal of the American Medical Association.

The study, by researchers at Vanderbilt University Medical Center, Oregon Health and Science University and their affiliated Veterans Affairs (VA) medical centers, analyzed 26 unique models for predicting readmission risk in a wide variety of settings and patient populations.

Most models had “poor predictive ability,” the researchers concluded, raising concerns about the ability of the federal government to “fairly compare hospital performance.”

Next year the U.S. Centers for Medicare and Medicaid Services will begin reimbursing hospitals based on readmission rates, said Sunil Kripalani, associate professor of Medicine at Vanderbilt who helped lead the study.

A key step in this process is to adjust (or standardize) the hospitals’ readmission rates, using statistical models that attempt to take into account the characteristics and illnesses of the patients at each hospital, so hospitals can be compared to each other more fairly.

“Until risk prediction and risk adjustment become more accurate, it seems inappropriate to compare hospitals in this way and reimburse (or penalize) them on the basis of risk-standardized readmission rates,” the researchers stated. “[rquote]Although in certain settings such models may prove useful, better approaches are needed to assess hospital performance in discharging patients.”[/rquote]

“Understanding the data is the first step to improving systems,” added Nancy Brown, chair of the Department of Medicine at Vanderbilt who was not involved in the study.

Lead author was Devan Kansagara, assistant professor of medicine at Oregon Health and Science University and staff physician at the Portland VA Medical Center.

Kripalani, the paper’s senior author, is chief of the Section of Hospital Medicine and associate director of the Effective Health Communication Program in the Department of Medicine at Vanderbilt.

Other Vanderbilt co-authors were Amanda Salanitro, instructor in medicine, and Cecelia Theobald, chief resident in ambulatory medicine. Salanitro also is affiliated with the Geriatric Research, Education and Clinical Center at the Nashville VA Medical Center.

The research was supported in part by the VA and the National Institutes of Health.