Via Alex Tabarrok at Marginal Revolution, Michael Mandel has produced an excellent example of a point that is a repeated theme around here: artificial scarcity in healthcare. In his report Dr. Mandel considers the case of a medical device, Melafind, and its failure to receive approval from the FDA.
Melafind is (or was) a handheld computer vision device and expert system designed to detect and identify skin melanomas, the most serious form of skin cancer. Melanoma is common and, sadly, on the increase.
In order to get approval, Mela Sciences, the company that created MelaFind, did a multi-year study of the accuracy of the device compared to a panel of dermatologists. The company claims that it passed the test that the FDA had agreed to.2 Indeed, on some dimensions of the study the device did better than the panel of dermatologists. Nevertheless, the FDA staff deemed the device “not approvable,” saying that MelaFind “puts the health of the public at risk.”3 Despite the strong negative response from the FDA, the company requested that the device be assessed by a panel of dermatologists, statisticians, and other medical experts. The advisory panel met in November 2010 and voted narrowly to recommend approving MelaFind.4 Nevertheless, the FDA has not yet approved the device.
Why has the FDA withheld its approval:
- The device did not do better than the experienced dermatologists in the study (“the FDA review team does not believe this is a clinically significant difference between MelaFind and the examining dermatologist”)
- The device was tested on lesions identified by experienced dermatologists, not on the broader set of lesions that might be identified by “physicians less experienced than these dermatologists.”
- The device did not find every melanoma in the sample (“Since the device is not 100% sensitive, if use based on the device’s diagnostic performance reduces the number of biopsies taken, harm could ensue in the form of missed melanomas.”)
- The device was not demonstrated to make inexperienced physicians the equal of experienced dermatologists (“Currently, formal training is offered to physicians to become board certified dermatologist and thus be able to diagnose clinically atypical lesions. The FDA review team would have to compare this board certification training to that offered by the sponsor to those physicians operating MelaFind to determine if it is found adequate.”)
No first generation device could meet the standard that’s being set out here and by withholding its approval the FDA is ensuring there will be no second generation. It is inevitable that investment will dry up for approaches and devices that can’t get approved.
If Melafind’s case is as its being reported the FDA would appear to me to have exceeded its statutory authority and, again assuming that the facts are as outlined, if the FDA’s actions has acted within its authority, its authority should be revised.
From my point of view there’s a larger lesson here. The FDA should be determining effectiveness and safety not pickiing winners and losers. A standard of “must do better than experienced dermatologists” and “must be 100% sensitive” is telling (are all board certiied dermatologists 100% sensitive?).
The FDA is voting to preserve the status quo. Accordiing to the BLS the median dermatologist earns $215,000 a year, the median GP $180,000, and the median PA half that. Not every lesion is biopsied and it would be needlessly expensive to do so. If the FDA is trying to establish medical policy it should be considering whether many more inspections, aided by Melafind, would be more effective not to mention less expensive than an infrequent or nonexistent examination by a dermatologist.
The essential question is not just one of effectiveness and safety but of how healthcare is supplied. The FDA apparently has a vision of a future in which a relatively small number of highly trained physicians have incomes rising far faster than the average persons and far faster than GDP, something we very clearly cannot afford. My vision is that in every drugstore there will be descendants of Melafind that can inspect, identify, and painlessly remove malignant melanomas of the skin so inexpensively that it will encourage their frequent use. The future should be one in which healthcare is abundant, not artificially scarce.