Why Physicians Don't Like Big Data

Remote Patient Monitoring: 9 Promising Technologies (click image for larger view)

Dr. Venkat Warren worries in a recent Wall Street Journal article that "some bean-counter will decide what performance is," and fears that the application of big data analytics and the pressure to meet a long list of performance metrics might force clinicians to avoid caring for older and sicker patients who could drag down their performance numbers. "If it isn't cost-cutting, what is it?" says Warren, a cardiologist in California's MemorialCare Health System.

Warren is not alone in his concern. Dr. Thomas Santo, a physician who lives and works in New York City, has come to a similar conclusion: "… for any physician who sees their reimbursement cut by providing 'sub-optimal care,' as deemed by CMS [the Centers for Medicare and Medicaid Services], what is to stop them from refusing to serve their sickest, most chronically ill and frequently hospitalized patients? In so doing, they raise the 'quality' of care they provide, and lower the cost at the same time …"

Many doctors also rightly point out that their scores in a pay-for-performance system are not completely under their control. They might do their utmost to make sure all their Type 2 diabetic patients get regular eye exams to check for retinopathy, for instance, or urge patients to take their oral hypoglycemic drugs. But when patients don't care enough about their own health, or are forced to choose between paying the rent and buying their medication, a doctor's metrics are going to suffer.

[ Time to start a family? Read Big Data Knows When You're Fertile. ]

The problems tied to economic disparity were well documented in a Harvard study that evaluated the performance rankings of primary care physicians. Their conclusion: "Among primary care physicians [PCPs] practicing within the same large academic primary care system, patient panels with greater proportions of underinsured, minority, and non–English-speaking patients were associated with lower quality rankings."

What's even more frustrating for doctors -- as well as the bean counters -- is if all the performance metrics are put in place and the system still doesn't generate savings.

The latest data from the Pioneer ACO program drives home that point. Thirty-two accountable care organizations enrolled in the experiment have been participating in the Medicare program for about a year. And in that time, all 32 have been able to demonstrate that pay for performance does in fact improve the quality of care. Patients have received more cancer screenings and blood pressure control has improved, for example. But only 18 of the ACOs managed to lower costs for their Medicare patients.

Although this research suggests that pay-for-performance metrics can frustrate both clinicians and CFOs, it's an overstatement to conclude that data analytics has no value in healthcare. Such analysis has demonstrated that many tests and treatment protocols have been oversold to the profession and to the public and need to be limited.

Among the tests and procedures that are probably wasting money because there just isn't enough scientific evidence to support their use: screening EKGs in healthy adults, routine annual Pap tests in women between 30 and 65, performing labor induction or cesarean delivery before 39 weeks when there is no medical indication, and giving antibiotics to a child with a sore throat, cough or runny nose. The list goes on and on.

And lest you think the list of unsupported tests and procedures was put together by a group of heartless efficiency experts, it's worth mentioning that it was compiled by the American Board of Internal Medicine Foundation in partnership with nine medical specialty associations and Consumer Reports.

In the final analysis, many physicians don't like big data being applied to their practice's quality and cost performance because it limits their options. In some circumstances, these limitations are dangerous because they impede a clinician's ability to provide good patient care by preventing him or her from ordering valuable tests that aren't on an officially recommended list. In other cases, it prevents needless tests and wasting insurers' and patients' money. Health IT leaders, along with tech vendors and public policy makers, shouldn't shrink from the challenge of putting in place the analytical systems and metrics that can separate the two.