October 18, 2014

Two stories shared by entrepreneurs at a recent health conference highlight some of the maddening ways our current system fends off precisely the sorts of data-driven innovation we so desperately need. Their experiences also allow you to imagine the advances we could achieve if we were more willing to consider and thoughtfully evaluate new approaches, rather than inclined to reflexively resist them.

The first story was told by an entrepreneur leading a molecular diagnostics company, who described her team’s efforts to introduce a novel test that would enable endocrinologists to diagnose thyroid nodules more accurately. The problem they were trying to solve was the need for a repeat needle biopsy after the initial needle biopsy report comes back as ambiguous or indeterminate — a relatively frequent occurrence.

While the performance characteristics of the new diagnostic test were closely scrutinized, as you’d expect (and hope), the so-called gold standard in the field against which they were compared turned out to be anything but. The team discovered huge issues of inter- and intra-observer test reliability when nodule biopsies were evaluated by trained pathologists — although the pathologists apparently agreed to participate in the evaluation exercise only with the explicit understanding the results would not be used for publication.

Such a lack of reproducibility is hardly unique to pathologists reading biopsies of thyroid nodules. As Vinod Khosla notes in a critical review of the topic, “doctors frequently disagree on diagnoses.” He points out that psychiatrists using DSM V, for example, have “dangerously low diagnostic agreement,” notably for conditions such as anxiety disorder and major depressive disorder (referencing this Scientific American commentary by Ferris Jabr).

A similar phenomenon was seen in an oncology second-opinion study conducted at Partners HealthCare, reports Joanne Wojik:
“According to a recent analysis of 330 oncology cases by Partners, 12 percent of the cases reviewed resulted in changes in diagnosis, while 90 percent resulted in either a new plan or a significant change in prior treatment plan, [Dr. Jospeh Kvedar of MGH] said.”

Displacing a flawed but familiar diagnostic test wasn’t the only challenge faced by the entrepreneur developing a thyroid biopsy test. Even after the company accumulated enough evidence to prove the test worked reliably, the scientists encountered a second problem: many of the endocrinologists didn’t seem to mind performing a repeat thyroid biopsy – presumably because they got paid for it (one of the very few opportunities most endocrinologists have to bill for a procedure). The newer diagnostic, if adopted, would likely cut into physician income.

Thus, a seemingly elegant, molecular solution to a specific and costly clinical problem – indeterminate thyroid nodule biopsies – struggled for adoption, at least in part, because it evidently ran smack into a slew of entrenched incumbents who benefited from the status quo.

The second example of the struggles healthcare entrepreneurs can face involves a company that had developed a molecular diagnostic approach for prostate biopsies, requiring the tissue sample to be sent to an external lab for evaluation.

According to the test developer, urologists were perfectly happy to order the new test – from their perspective, little changed mechanically, and they could wind up with a more accurate diagnosis. However, in order to get hospital labs to send out the test, the diagnostic company discovered it had to pay a disproportionate “tax” to the hospital lab. Worse, at least one hospital further penalized the requesting surgeons (for taking a revenue opportunity away from the hospital) by making it harder for them to get the early operating times they wanted; the surgeons responded to this aversive conditioning very quickly, and stopped requesting the send-out assay; their preferred operating times were quickly restored.

Obviously, these stories are from the perspective the test developers, and naturally include their biases; however, the experiences resonate because, well, this is how many doctors and hospitals really do behave. Incentives are important, and while value-based-care is beloved by policy makers and journalists, activity-based care remains dominant – and deeply destructive, as Jonathan Bush points out, to innovative entrepreneurs who might offer better solutions.
It also raises the important question of how can we get to a better system, where existing “gold standards” are systematically questioned and evaluated, and innovative, data-driven solutions are appropriately embraced?

A popular, optimistic answer is that, as consumers become more empowered – and on the hook for a greater fraction of their healthcare costs – they will become more discriminating and demanding, and will insist on knowing more about the cost and reliability of the tests for which they are paying. The idea is that we’ll slowly evolve to this more enlightened state.

Perhaps. But I wonder whether to really get this right, you need to need to build an entirely new health care system, or clinic, from the ground up (an approach to which I alluded in my last piece, about getting to an Integrated Medical Record).

I can envision a data-loving “Health Clinic Of The Future,” where all participants – doctors, patients, staff – would agree from the outset with a set of common principles, including in particular a shared belief in the value of data (including cost data), data that the clinic would aim to gather and utilize as often and as transparently as possible.

This clinic would ideally attract curious, inquisitive physicians, the sort of doctors who would embrace test/re-test explorations, and maintain an eager and open mind about new approaches. It would be an environment where “First, Do No Harm” is understood as a need to routinely challenge the status quo, rather than instinctively defer to it. Patients and doctors would be partners in discovery — discovery abetted by routine whole genome sequencing (disclosure: I work at DNAnexus, a cloud-based genomic data management company), and rich phenotypic data including dynamic parameters collected by both traditional medical devices like pacemakers, and emerging digital health platforms like and Propeller Health.

Importantly, the clinic wouldn’t seek to replace humanism with cold rationalism, but instead, to combine empathic care with robust, data-driven knowledge – and critically, an inquisitive data-driven mindset — on the part of patients as well as as doctors.
Without question, the Health Clinic Of The Future would represent an ambitious experiment – but it would also provide an exceptional opportunity for patients and doctors who believe medicine should be practiced differently to step up and play an active, highly engaged role in creating the future they envision.

You might even say the clinic would look like an earlier vision of the academic medical center, before their present-day focus on RVUs and throughput. A second critical difference from the traditional AMC model is that the Clinic of the Future recognizes patients as vital partners and participants, and is structured to collect and share data in a way that reflects this restored balance.

Best of all, I suspect that if you built such a clinic, not only would the patients and doctors come, but you would quickly find yourself sitting atop a rich, unique, and exceptionally valuable dataset that could be developed and leveraged to economically support the clinic — all while radically improving patient care and redefining the practice of medicine.


No comments

Be the first one to leave a comment.

Post a Comment