I've been thinking about Joe Bugajski's recent blog post about his health ordeal and resulting skepticism about electronic medical records. OK, skepticism is perhaps a bit mild compared to what Joe actually feels.
Joe had a life-threatening health emergency in early 2009 that was exacerbated by innacuracies and user experience problems in the electronic medical records system. He calls out weak data modeling as the culprit preventing effective insight into his medical history, medications, and courses of action.
As I read deeper into Joe's personal blog, some other things emerged that I want to call attention to. Health care culture and style of work, especially in fast-paced emergency settings, may resist the intrusion of technology beyond that which attaches to the patient's body. To ask a doctor or nurse to divert their attention to data entry or retrieval takes them away from their primary goal, even though that data may contribute to the overall care of the patient. In effect, the technology becomes a distracting third-party to the provider-patient relationship. It is not a passive technology, and as a result may actually increase the risk of medical error.
To be effective, medical records technology must be passive and comprehensive. It must become transparent through its use, so that the provider-patient relationship is strengthened, not weakened. As Joe suggests, data modeling is an essential component of ensuring that the correct data is surfaced at the correct time to the correct device/person. The data model is a logical representation of the patient drawn from various sources— some historical and some real-time. Some data is more critical than others, and that criticality can shift in the midst of care. It is not a replacement for human insight. Automation of patient data will not improve care on its own. Automation of correctly modeled data should facilitate and accelerate human insight.
Electronic medical record technology should fit the various ways that medical professionals work. To the extent possible, EMR approaches should be woven into the ergonomics of care. But at this early stage, EMR processes are bolted on to hospital procedures. Perhaps the early demand for EMR is driven by liability protection vs. improved care. The paradox is that unless EMR entry and retrieval is integrated into overall work practice, risk and liability increases.
The German philosopher Martin Heidegger wrote about "throwness", "breakdown", and "readiness-to-hand", and they apply in this and many other human computer interface (HCI) scenarios.
Throwness is the ability to react intuitively in highly fluid situations.Throwness is the result of learning and assimilating actions so that they can be performed without active analysis or thought. When it comes to golf, for example, Tiger Woods exhibits throwness.
Breakdown occurs when intuition is interrupted, and the objects present in the scenario stop functioning as a cohesive set. Breakdown occurs, for example, when your car blows a tire at high speed, and your attention is suddenly drawn to the control of the car. In effect, those actions that were suppressed into your mid-consciousness come to the fore. Breakdown occurs in HCI when your intuitive actions are distracted by system issues (like crashing programs, annoying data entry).
It is difficult to design sophisticated tools that allow for human "throwness", avoid breakdown (or handle it gracefully), and reach the stage of readiness-to-hand. This is especially true in a highly human-dependent work environment that moves at an accelerated and interrupt-driven pace. Unless the technology you place into that environment is "ready-to-hand", it will inhibit effective work rather than supporting it. I'm afraid that's what happened to Joe.
So, healthcare will not be improved with EMR alone. Good data modeling, coupled with significant user experience design will work together to make humans more effective and healthier, too.