Predictive analytics in healthcare: Are they safe?

While hospitals--such as Boston Children's Hospital and El Camino Hospital in Mountain View, California--increasingly are turning to predictive analytics to improve patient care and safety efforts, some worry about the legal and ethical implications of using such technology.

For instance, researchers from Harvard Law School, the University of California, San Francisco and the University of Texas Southwestern Medical Center say in an article published this month in Health Affairs that predictive analytics models may make care recommendations that suggest withholding potentially effective treatments from patients based on probability statistics. What's more, they say, docs who rely on such models could face an increased risk of liability.

"Predictive analytics promises to make dramatic changes in the way healthcare is practiced and delivered," the authors say. "That said, this terrain is changing rapidly and it is not easy to predict where the technology will lead and how physicians and patients will react to it."

Additionally, the authors say, because massive amounts of data are necessary for such models to be effective---as they are meant to represent whole populations--concerns over potential abuse of that information linger.

"[T]he data could be used to identify vulnerable high-risk, high-cost patients and exclude them from care," the authors say.

Still, the authors say they believe if used properly and with appropriate caution, predictive analytics ultimately can help to strengthen healthcare efforts, particularly as providers become more accountable for patient outcomes.

One of the lessons learned from the Pioneer Accountable Care Organizations is that data is key to making everything work, health executives have noted. According to Elliot Fisher, director and professor of medicine at Dartmouth Institute for Health Policy & Clinical, good data systems enable hospitals to identify patients that need special attention.

And according to Greg Sheff, executive vice president of clinical services at Seaton Healthcare Family in Austin, Texas, use of predictive analytics to define high-risk patients has been one of the 11-hospital system's biggest successes.

To learn more:
- here's the Health Affairs abstract