How to Improve the Decision Making of the USPSTF

Scott Gottlieb points to a significant problem in the US Preventive Services Task Force's decision making: Too frequently, they conclude that the evidence about a preventive service is insufficient to evaluate whether it works. When the USPSTF gives a preventive service an "I" ("Insufficient") grade, insurers may not be willing to cover that form of care. 

So why don't we have enough evidence about the effectiveness of many preventive services? Gottlieb correctly observes that

The USPSTF typically favors large prospective, randomized trials to validate a preventive service. But such research is generally hard to conduct for screening tests. It would require, for example, patients at risk for a particular disease to be randomly selected to either receive a screening test for the ailment or forgo the diagnostic measure. The patients would then need to be followed, sometimes for many years, to see if the tool enabled screened patients to recognize better health outcomes (and lower overall utilization of medical services) than patients who were randomly selected to forgo the screening test. 

That is, it is challenging, expensive, and time consuming to carry out the studies that would really tell us whether preventive services really benefit -- or harm -- the patients who receive it.

So what is Gottlieb's solution to this problem? I find two suggestions: 

1. "Congress also should make sure that recommendations issued by the USPSTF are in sync with sister public health agencies that have far more expertise in the domains in which they operate. These include the CDC, the National Institutes of Health, and the FDA."
2. The USPSTF should be subject to the rules about public comment on regulations that bind other US government regulatory agencies.

I disagree with point 1 but agree with point 2 (and see the Task Force's view here), even though point 2 will slow the speed at which the Task Force issues recommendations. 

Please notice, however, is that points 1 and 2 are just requirements that the Task Force take into account the opinions of experts in other health agencies, and public opinion. But gathering more opinions just increases the amount of opinion. The fundamental problem, as Gottlieb himself noted, is that we lack data on the outcomes of preventive services. So what we need is more data, not more opinion.

The best way to improve USPSTF decision making is to fund more randomized clinical trials on preventive services, and to fund registries that can follow patients who (and do not) receive these services. Think evidence, not opinion. 

Comments

[Kathi]

Great post! I'm going to borrow your last line and use it in my closing signature.

But to the point, I think such RCT's are unlikely to become more prevalent in the current fiscal environment of the scientific world. Registries of persons with intermediate phenotypes of disorders may prove to be a more cost-effective approach.

[Bill Gardner]

Thanks for the kind words, Kathi. Sadly you are probably right about support for research. But maybe we can print up some "Think evidence, not opinion" buttons and start a movement?

[Daniel S. Goldberg]

But Bill, this is simply a well-studied feature of the problematic that is evidence-based public health policy. Namely, that we frequently (usually?) lack evidence of reasonable, let alone high, quality that would be sufficient to guide us with a modicum of confidence on any particular public health policy question.

No one disagrees that the best answer to this problem is to obtain better evidence of sufficient quality, but that does not address the real-time problem that the data simply may not exist at the time a decision needs to be made.

So, while I certainly agree that we need more data, I have to ask what you think we ought to do in the absence of that data when we need to make a decision?

The fact of this problem, its commonality, and its significance is one of the main reasons expert consensus panels are utilized -- it is suboptimal, of course, but in the absence of quality evidence, it is adjudged as the least worst evil. This may be an incorrect judgment, but what would you suggest in the alternative?

[Bill Gardner]

Dan,
Great questions -- see my next post...