October 16, 2013
You may be weary of pundits bearing buzzwords about "data" and "big data" and proclaiming the geeky good news that the solutions to your problems are merely a set of numbers and a good algorithm away. But the proposition is being taken seriously by the health care industry. Providers are inspired by the convergence of two separate trends: the proliferation of ever-cheaper sensors and the Obamacare-inspired need to make more-informed decisions in order to deliver good care for less money.
Many people carry or own devices that can act as sensors. Smartphones have accelerometers that can measure physical activities and advanced cameras that can provide evidence for interpretation. These devices are becoming more sophisticated; Microsoft's Kinect camera, for instance, can estimate blood pressure based on how flushed a user appears.
Consumers' willingness to carry sensors everywhere becomes important as providers are incentivized by the Affordable Care Act to keep costs down while improving outcomes. Acting on this newly available wealth of data—perhaps by intervening earlier, more cheaply and effectively—might benefit providers that are increasingly paid for quality, rather than volume, of care.
As consumer sensors become more sophisticated, professional-grade sensors are becoming easier to use. Joseph Kvedar directs the Center of Connected Health at Partners HealthCare, a Boston-area health system. A decade ago, the hospital's old sensors were "dumb," Kvedar notes in a recent blog post. To extract the data, the sensors had to be tethered to a central hub, limiting their use in hospitals and remotely, as dealing with wires and extracting the data was annoying and time-intensive. But sensors are increasingly wireless, and hubs like Qualcomm's 2net automatically sync with a variety of devices and scrape their data, conveying it directly to patients' electronic health records.
More convenient gathering of data might mean better diagnoses and care. One study, published last month in the Annals of Thoracic Surgery, used common fitness monitors to assess post-surgery recovery in elderly patients. "Speciﬁc patient mobility data are typically found in nursing notes and are not usually part of the workﬂow of the surgical team," the authors write. "Such data may not be obtained in all patients and are intermittent.... With wireless technology, data are objective, acquired, and displayed nearly continuously." That could help doctors better target their care, bringing down costs.
Propeller Health, a Wisconsin-based start-up, uses patient smartphones and a sensor attached to an inhaler to track the location and time of its use for its patients with asthma and chronic obstructive pulmonary disease. Patients had previously used journals to record inhaler use, subjecting their records to the uncertainties of memory. With the new data, they and their care team can better determine what triggers attacks and how to prevent them, avoiding potential emergency-room visits. The start-up has partnered with insurers and hospital systems hoping to reduce costs by providing better care.
It's also possible that objective data can provide a better diagnosis. Many patients, probably due to nerves, are afflicted by "white-coat hypertension," only demonstrating symptoms of the condition when their blood pressure is measured in the clinic. Monitoring devices can make the distinction between white-coat hypertensive patients and full-time sufferers. And scientists were surprised at the level of variation in glucose levels when continuous monitoring of diabetic patients was introduced, says Wendy Nilsen, a health scientist administrator in the National Institute of Health's Office of Behavioral and Social Sciences Research division. Health is more complex than old tools were able to capture.
Doctors are still learning how much and what kind of data to collect. Listening to a sea of data rolling in may introduce too much noise. A meta-analysis, or study of studies, published by the Journal of Medical Informaticsthis summer shows the level of tweaking necessary to determine which are the right sensors and the right data to monitor, and then to translate the findings into action. The study examines different sensors that assess elderly patients' risk of falling and hurting themselves. Performance testing proved that wearable sensors generally performed better than their non-wearable counterparts.
Despite evidence of the sensors' efficacy, the study consistently found resistance from nurses and doctors to incorporating them into their daily work. A possible reason: Hospitals are cacophonous places and "alarm fatigue" is real. One of the studies in the meta-analysis found a false-alarm rate of 16 percent; after 10 percent, the authors say, health care providers are desensitized. Sensors risk crying wolf and bothering patients and providers alike.
"You don't want to disrupt people," Nilsen says. "If [a patient's] blood pressure spikes, it might be for something pleasurable—you don't necessarily want to call them up on a Saturday night and say, 'By the way, your blood pressure spiked.' We might want them to do those activities."
Patients might not love the sense of being in the panopticon. Penn State recently ended its wellness program, which sought sensitive lifestyle data from employees (asking men, for example, whether they conducted regular testicular self-exams) and imposed a $100 monthly penalty for noncompliance. Employees organized and protested what were perceived as intrusive questions, causing the program to end.
The health care system, then, needs to commit to using data responsibly—to be sensitive to patient preferences and to make sure the information is being put to good use. Otherwise, patients might stop monitoring before it ever really begins.
October 16, 2013