With readmissions a target under the healthcare law, hospitals are increasingly using predictive modeling tools to pinpoint patients who might be at risk for coming back to the hospital.
Utah's Intermountain Medical Center, for instance, developed a computerized algorithm to identify heart attack patients at risk for readmissions, presented yesterday at the American Heart Association Scientific Sessions 2012 in Los Angeles.
"Right now, no one has a way to accurately measure the information that these risk factors tell us about readmission and mortality risk," lead study researcher Benjamin Horne, the Intermountain Medical Center Heart Institute director of cardiovascular and genetic epidemiology, said in a statement. "Our tool gives physicians a way to measure their patients' risk and possibly manage their care differently."
Researchers looked at 51 factors, such as age, gender, blood test information, diagnosis history and body-mass index, to calculate an at-risk score. Based on the data, researchers found the greatest risk factors for readmission were age, how many medications a patient was prescribed, the length of stay, depression and atrial fibrillation.
Female patients receive a score, ranging from 0 to 14. A woman with a score of 14 is about three times more likely to return than a woman who scored zero. For each additional point on the scale, women have a 14 percent greater risk of readmission.
Men, on the other hand, receive scores ranging from 0 to 13. If a man receives a score of 13, he has a higher chance of readmitting, at 3.6 times more likely to return to the hospital than a man with a score of zero. For every additional point, men have a 20 percent greater risk of readmission.
Researchers noted they computed the clinical decision tool from inexpensive lab tests. Coupled with the Medicare penalties associated with readmission rates, study authors suggested the predictive model could be used for clinical application with high predictability, as well as cost savings.
Other providers across the county also have experimented with predictive modeling for possible rehospitalization of heart patients, including New York's Bassett Medical Center with its probability calculator that looks at congestive heart failure, chronic lung disease, amount of time on cardiopulmonary bypass and body-mass index over 40, as well as Dallas' Parkland Health and Hospital System with its algorithm, based on physiologic, laboratory, demographic and utilization data from electronic health records.
For more information:
- see the Intermountain research announcement
- here are the presentation abstract and data
Online nurse support improves vascular disease risk factors
Tool predicts survival odds for cardiac-arrest patients
Hospitals cut readmissions with predictive modeling
New IT tools can predict at-risk heart patients
Hospitals target high-risk patients by predicting readmissions