This work is licensed under a Creative Commons Attribution 4.0 International License.
June 11, 2015
The Health Datapalooza used to be the most exciting event in health IT, but this year a torpor hung over it. I confirmed this feeling with numerous colleagues, and traced it to a few different trends, some within control of the conference organizers and some outside it.
The essential problem, many of my friends agreed, is that the apps market and other signs of progress in health IT are taking too long to launch. Many excellent building blocks are falling into place in both technology (DirectTrust and FHIR) and policy (recent announcements by the Center for Medicare & Medicaid Services as well as private insurers that they will greatly increase risk sharing and pay-for-value). But these are slow to produce change.
In former centuries, religious fanatics would sell off all their possessions in expectation of the Messiah’s imminent arrival. These believers’ disappointment must have resembled that of current health entrepreneurs who sink their life savings into a company promoting some innovative health care reform and who then wait for the market to come around to their solutions.
Essentially, I believe the optimists have stretched their arguments a bit too far. They hope that payment reform will convert data-hoarding hospitals and clinics into champions of data sharing. These providers would then supposedly transport their records en masse to EHR systems that support frictionless data exchange.
In the meantime, FHIR adoption would generate a third-party marketplace of dynamic applications making it easy to filter records and view the critical patient information needed to make decisions, along with running analytics to find new cures.
This vision is just fine, and I share it to a great extent. But when laid out as starkly as I have here, it reveals some of the huge barriers that still stand in the way. That is why health care reform will continue to take several more years—and why the Health Datapalooza is in danger of stagnating and losing its appeal.
Some of the barriers are:
Organizational workflows that don’t yet include data transfer, along with the personal prejudices of clinicians who insist on relying on their own judgement.
For example, specialists routinely re-order tests that were performed by the doctors who made the original referrals. The specialists claim that they can’t trust the tests ordered by other doctors. Funny, I’ve never heard a specialist say, “I don’t trust that the referring doctor is competent to make the referral, so I refuse to see the patient.” Their lack of trust extends only to things that help them generate more costs in the health care system.
EHR vendors who have repeatedly been shown incapable of supporting standards and interoperability. A new crop of EHRs may emerge that solve these problems, but providers will still be held back by existing records trapped in proprietary formats.
App developers who still need help understanding their markets. Most are out of touch with the multiple stakeholders that they need to appeal to in order to make a sale (doctors, patients, medical administrators, insurers).
The reticence of most individuals to use sensors or apps offering guidance until they are recommended by clinicians and covered by insurance. Meanwhile, the clinicians won’t recommend the technologies or insurers pay for them until their efficacy is proven. One interesting breakout session at the Datapalooza heard from the experiences of developers who had achieved the rare feat of getting their device or app through the FDA approval process. Many other innovators are looking for lower-cost means for winning the trust of clinicans and insurers—a subject I will cover in a future article.
The relatively young stage of data mining, even though impressive results have been reported in using analytics for low-hanging fruit such as predicting when a high-risk patient may require hospitalization. At the Datapalooza, former National Coordinator Farzad Mostashari complained that predictive analytics rate everybody along one simplistic five-step scale, and is mostly braced on crude claims data.
Data—especially genomic data—so fraught with privacy risks that researchers probably won’t get access to it until a new generation of patient consent tools emerge. This was the topic of two sessions at a Health Privacy Summit that followed the Datapalooza, where some innovative experiments were proposed.
Like the evolution of key body parts, the health care field needs to advance on several fronts at once. We need more engaged patients who have control over their own data, more sophisticated applications that reach deep into both clinical workflows and data analysis, a better understanding of real health care costs (in order to make pay-for-value fair), and much more flexible electronic records. Other experts could probably go on for many pages adding to this list of requirements.
As these reforms emerge, the Datapalooza may regain its excitement. But it’s also critical for the organizers to restore this conference to a cutting-edge scene, which means purging it of some indulgent practices that marked it as early as 2011.
The main stage has always been a platform for HHS and other government agencies to strut their recent advances, which is acceptable, but the leaders who appear there should be admonished to present some uniquely useful perspectives at the same time. When such luminaries as Mostashari, Ceci Connolly, Bob Kocher, and Ed Park come together on the stage, they don’t need to waste half their time extolling a change to a fee-for-value payment structure, something well understood by the audience.
Keynotes also need to focus on the new and ground-breaking. There is no justification for handing over the plenaries to a fifteen-minute ad for Box’s secure file sharing service, something matched by a dozen other technology options. Nothing was said in this session (or on the web page, except for a vague promise of “granular permission”) about authentication and trust, which are the real issues in health data sharing and are addressed by such technologies as DirectTrust and OpenID. Let’s respect the audience’s expertise and provide them with something new to chew on.
This work is licensed under a Creative Commons Attribution 4.0 International License.