What is the Role of the Physician in a Data Rich World?

Last week’s Project HealthDesign workshop, held at the Vanderbilt Center for Better Health in Nashville, focused on how clinicians could use “observations of daily living,” or ODLs – data associated everyday experiences such as diet, exercise, sleep and pain – to provide better care to people with chronic diseases.  The five Project HealthDesign teams are refining their plans to integrate ODLS into the treatment of premature infants and their parents, obese teens at risk for depression, adults with Crohn’s disease and its complications, adults with asthma and depression or anxiety disorder, and elders with mild cognitive impairment.

A presentation by Kevin Johnson, vice chair of the Vanderbilt University Medical Center Department of Biomedical informatics and the project director of a previous Project HealthDesign grant, raised interesting questions about how to present information captured through ODLs and who should interpret the information.  Johnson showed this graph, which represented a self-report of medication usage over a month’s period as compared with a schedule. 

Clearly, any of us can tell at a glance that the patient adheres to this medication schedule extremely well. The presentation conveys this information quite clearly and, frankly, it doesn’t take a great deal of clinical training to interpret the information. But consider two other cases.  First, if this chart showed a much poorer rate of adherence, the challenge would be to look for patterns in the missed, late and on-time doses to see if there were behavioral triggers or environmental factors that explained the results (e.g. a change in work schedule means the noontime dose is problematic, a Thursday evening softball game makes it unlikely to remember). In the second case, one could overlay on the medication chart other data, such as pain level, mood or even clinical signs like blood pressure and look for patterns that might lead to inferences about correlations and interrelationships.

In each of these three cases, one could ask what skills and training are needed to review, interpret and act upon the information (one can even take it a step further and ask which of the cases requires human vs. algorithmic interpretation). Of course, at some level, the answer is “it depends,” but thinking about the question gives some insight into the broader question we’ve been asking of late in the Pioneer Portfolio:  “What is the role of the physician in a data rich world?” I’m not a clinician, so I’m on shaky ground here, but it seems to me that only the third case requires clinical training – because it’s a case that requires integrating clinical knowledge into a pattern sensing activity. It’s a form of clinical problem solving. The second case doesn’t seem to require clinical knowledge so much as an understanding of some concepts from consumer behavior, design or even behavioral economics and an ability to motivate – to work with the patient to find a solution.  And the first case seems to lend itself to automated processing to determine if the behavior is within some predefined range.

The question gets even more interesting when one factors in the patient’s own engagement in these cases. One would hope that the patient is looking at the same data and developing her own questions and hypotheses. With whom should she discuss them?

What do these scenarios and questions imply for the way we currently primary care and the health professions that make up that enterprise? How will those professions need to evolve?  Will we need new professions? And are the right skills being taught to the right students today?

Comments

[Elaine Schattner, M.D.]

I think the doctor's input is useful at multiple levels before, during and after intervention with ODL documentation. The most crucial involvement might be early on - in establishing the right diagnosis, in deciding whether treatment is needed and, if so, in helping the patient choose among treatment options.