Algorithm compiles misleading readmissions information

As readmission rates are increasingly used to measure quality at hospitals, methods for determining those rates will come under increasing scrutiny. But the numbers are not always accurate, some say.

For example, University of California, San Francisco physicians have found flaws in an algorithm that abstracts readmissions data from the Centers for Medicare & Medicaid Services. The algorithm pulls "all cause" information, meaning it lumps scheduled readmissions and unscheduled readmissions together, overestimating unplanned readmissions by 25 percent.

The researchers--who presented their findings at last week's American Association of Neurological Surgeons in Miami, according to an article on the UCSF website--based their conclusions on spinal encounter information pulled out of the University Health-System Consortium's clinical database/resource manager from 2007 to 2011.

Senior author Praveen Mummaneni, M.D., called statistics determined by the algorithm "unrealistic,"  and added that "substantial drill down" is needed for the algorithm to truly be useful.

Such statistics potentially could be harmful to hospitals treating patients in high-poverty areas, as starting in 2013, Medicare will penalize facilities with high readmissions rates.  Last September, the American Hospital Association called readmissions are an unfair method for gauging hospital quality, as many disagree what constitutes a "preventable" readmission.

To learn more:
- read the UCSF article
- here's the study's abstract (.pdf)