Now though, the threat that healthcare money may be taken away, specifically through Medicare financial penalties under the ACA, has become a key driver in a recent wave of real-time predictive analytics projects using Big Data.
Medicare penalties are attached to 30-day readmission rates for three core measures—heart failure, acute myocardial infarction and pneumonia—but more conditions subject to penalties will be added in 2015.
Are such penalties the best way to incentivize more providers to move to predictive analytics? Maybe not.
While acknowledging the three targeted conditions account for 10% of all readmissions in the Medicare population, researchers Dr. Karen Joynt and Dr. Ashish Jha, working in the department of health policy and management at the Harvard School of Public Health, noted that two studies have indicated that fewer than 20% of readmissions are preventable.
Writing in the New England Journal of Medicine in 2012, when the ACA penalty system was about to be launched, Joynt and Jha complained that the metrics coerce hospitals into devoting limited resources to a “misguided” policy mandate “since much of what drives hospital readmission rates are patient- and community-level factors that are well outside the hospital's control,” they said.
In subsequent writings, Joynt and Jha pointed out that the penalties fell disproportionately on larger, more urban and safety-net hospitals. They noted in a March 2013 NEJM article that the program “opts out of accounting for socio-economic status altogether, leaving hospitals that disproportionately care for the sickest and poorest patients at particular risk for penalties.”
Still, they advocated tweaking rather than abandoning the penalty program. They recommended that the feds add an adjustment factor for patients' socio-economic status. They also advise weighing penalties according to the timing of the readmission, with full penalties for readmissions a few days after discharge (which are more likely the result of poor planning, they said, and therefore more under the hospital's control), and lesser or no penalties for readmissions occurring later after the discharge date (which are more likely a consequence of the severity of a patient's illness, they said).
In June, a group of U.S. Senators introduced the Hospital Readmission Accuracy and Accountability Act that would require the CMS to account for patients' socio-economic status when calculating risk-adjusted readmissions penalties.
A recent study published in the policy journal Health Affairs noted that some safety-net hospitals with lower than expected mortality rates are also susceptible to Medicare penalties for higher readmission rates.
Meanwhile, Jha said his own thinking on readmissions has “evolved” since 2012.
“Readmission rates have started coming down and that's a good thing,” he said, “but safety-net hospitals are twice as likely to get the penalties. In my mind, that doesn't make sense. It's just a different ball game taking care of patients that are homeless. Why not address that?”
And, Jha acknowledged, even if predictive analytics are initially introduced to reduce targeted hospital readmissions and avoid penalties, they could be further employed for other quality improvement and cost-reduction processes.
“You have a couple hundred hospitals participating in bundled-payment programs and a couple hundred participating in ACOs,” Jha said. “There is a broad realization that you have to start doing predictive analytics, and to the extent that the readmissions measures have played a role in that, I like it. We should just fix the negatives.”
Follow Joseph Conn on Twitter: @MHJConn
