Search the Health Library
Get the facts on diseases, conditions, tests and procedures.
I Want To...
I Want To...
Find Research Faculty
Enter the last name, specialty or keyword for your search below.
Inside Tract - Using Billing Data as a Predictor for Readmissions
Inside Tract Winter 2014
Using Billing Data as a Predictor for Readmissions
Date: January 1, 2014
Susan Huftless and Danning He co-authored a groundbreaking study simplifying readmissions data.
Among the biggest changes spurred by the Affordable Care Act is the push to improve care and save money at the same time by reducing costly hospital readmissions.
Under the new law, the Centers for Medicare and Medicaid Services will no longer reimburse care for patients admitted to the hospital a second time in 30 days for the same problem, a change that sent health care economists and other researchers into a scramble to predict which patients are at greatest risk for readmission.
"It's a major challenge, and hospitals across the country are working on it," says Susan Hutfless, a Johns Hopkins University School of Medicine instructor who applies advanced epidemiological techniques to gastroenterology research. "We felt we needed a prediction tool that was fast, simple and accurate."
Using data collected from billing records, Hutfless and her colleagues built a model for predicting readmission that has outperformed many more complex models that use clinical data and are based on whole populations of admissions.
"Most people who have studied this problem have looked at whole admission populations of surgical patients or of medical patients," says Hutfless. "They haven't considered the underlying condition the patient had at readmission. We thought we might get better predictors if we got more specific and added one more data element."
Gastroenterology Director Anthony Kalloo noticed that patients with chronic pancreatitis seemed to return to The Johns Hopkins Hospital more often than other patients. Hutfless and a team of colleagues went to work examining the billing data for pancreatitis readmissions.
"The nice part about billing data is that it's easy to get," Hutfless says. "We decided to apply the billing code for pancreatitis to this and see how it worked. It's working very well."
Hutfless and her colleagues published their findings this fall in the Journal of the American Medical Informatics Association.
"This is just our approach with gastroenterology," Hutfless says. "Every field of medicine has its 'pancreatitis,' so to speak."
Hutfless also stresses the predictor model's portability. "Any hospital in America can use this formula to predict readmissions for any field of care."
The next step is what to do with that information. "Some people think it could be as simple as scheduling a follow-up appointment before patients are discharged or maybe having nurses call them," says Hutfless.
But she adds that the key to reducing or eliminating hospital readmissions may be easier than anyone expected.
"The answer could be as simple as the information itself," she says. "If doctors know that a patient is likely to be readmitted, it may affect their approach to initial treatment."