Study identifies patients at high risk for hospital readmissions

How often patients land in the hospital—and how long they stay—were better indicators of which patients would return to the hospital unnecessarily than types of illnesses, number of prescriptions and other factors, a study found.

The results, published by JAMA Internal Medicine, are among the latest in a growing body of research that seeks to identify patients at high risk for avoidable hospital stays by shifting through patient data in search of flags that predict who will make a repeat hospital visit.

Policymakers have targeted hospital readmissions as a source of potential waste. Last October, Medicare began to cut pay to hospitals with higher-than-expected readmissions within 30 days for heart attack, heart failure or pneumonia patients.

Hospitals, too, have targeted readmissions in response to Medicare's push and in a bid to lower healthcare costs under new insurance contracts, such as accountable care, that include incentives to slow health spending.

Dr. Jacques Donze, a research associate with the Brigham & Women's department of medicine who contributed to the study, said researchers sought to identify factors that could be tracked before patients leave the hospital so that clinicians might intervene with support that could prevent a repeat visit. A study already under way will use the score to identify patients to test potential interventions, such as individual coaching, home visits and pharmacist oversight of patients' medication, he said.

The analysis used data for roughly 9,200 patients who stayed at Brigham and Women's Hospital for at least 24 hours between July 1, 2009, and June 30, 2010. Patients either did not return within 30 days to any of three Partners HealthCare hospitals, including Brigham and Women's, or were readmitted within a month for what was identified as an avoidable visit.

Researchers with Brigham & Women's Hospital and the Bern University Hospital in Switzerland combed through two dozen patient characteristics that could be culled from patient records accessible during a hospital visit to look for potential risk factors.

Of those characteristics—which included factors such as age, whether patients had a caregiver upon leaving the hospital and certain laboratory results—the study identified eight that best predicted which patients would return to the hospital within a month.

Ultimately, researchers used seven of the eight factors to create a risk score for potentially avoidable readmissions. The seven factors are hemoglobin at discharge; discharge from oncology; sodium level at discharge; number of procedures during first admission; non-elective versus elective admission; number of admissions within prior year and length of the hospital stay.

Six medical conditions were examined as possible risk factors, and only congestive heart failure appeared to have any predictive value for readmissions. Congestive heart failure was eliminated from the risk score, however, because it was the weakest indicator and because the diagnosis in some cases doesn't materialize until billing data is generated after discharge.

The score will be tested in an international study, which includes seven U.S. locations, said Dr. Jacques Donze, a research associate with the Brigham & Women's department of medicine who contributed to the study.

Researchers said they believe that no prior studies have identified procedures during admission and sodium level at discharge as risk factors for readmissions. The number of hospital visits and length of stay, identified as important predictors, could be surrogates for other indicators that reflect the severity of patients' illness, the authors wrote.