The Aftermath of Efficacy

As a member of the US Preventive Services Task Force for 12 years and its chair for 5, I have given countless presentations on Task Force recommendations. During questions and comments, clinicians often voice frustration and even hostility to the recommendations because the Task Force does not advise on exactly what should happen next. My response is that, indeed, the science is the easy part and that effective implementation is hard, particularly in a health care environment that discourages system innovation. Although the process of sifting through the medical literature to determine the quality of evidence that supports (or fails to support) a preventive intervention is time consuming, costly, and, at times, tedious, it is not particularly complicated. If the Task Force has set up the analytic framework correctly, identified the key questions, and carefully followed protocol in completing the evidence review, the conclusions usually follow easily. Screening for colorectal cancer is a good example of where the science showing efficacy of screening is fairly straightforward, but where the paths to effective implementation are not. In its updated recommendation released in October, 2008,¹ the Task Force has judged that screening works, but then what?

Four articles in the current issue of the Annals of Family Medicine address “then what” in different ways. Potter and colleagues show that offering fecal occult blood kits to patients during flu shot clinics increased screening from 57% to 84%.² Using data from the Behavioral Risk Factors Surveillance System, Cardarelli and Thomas show that having a personal health care provider is associated with a 3-times higher likelihood of screening.³ Jimbo and colleagues examined reasons that positive fecal occult blood tests were not followed up, finding that such decisions were at variance from established guidelines or could not be determined in nearly one-half.⁴ Finally, Wilkins’ group conducted a quantitative meta-analysis of the literature on the outcomes of screening colonoscopies performed by primary care physicians, showing that quality, safety, and efficacy are similar to indicators proposed by specialty professional groups.⁵ These new articles are high-quality work, showing imagination and skill on the part of the investigators. My comments briefly address the audiences of physicians, patients, and policy makers.

Physicians eager to implement prevention in practice have long recognized the importance of an established relationship and of using every opportunity to sneak in indicated preventive interventions, the messages from Carderelli and Thomas and Potter et al, respectively (although the Potter et al strategy falters as the list of potential add-on interventions grows). Screening and follow-up according to protocol are essential to realize the full benefits, the message of Jimbo et al. Health care teams seeking to implement prevention in practice, perhaps as part of their work on the medical home, should be encouraged by these new studies to establish relationships, order indicated prevention during unrelated office visits, and adhere to proven protocols. Such resources as the AHRQ’s Put Prevention into Practice program for clinicians already promote these messages.⁶ Although the issue of colonoscopy is clearly of interest to individual clinicians, in my view the obstacles to having primary care physicians perform colonoscopies are principally political, economic, and medicolegal, not the issue of clinical competence, as examined by Wilkins et al. Nonetheless, their new meta-analysis is published at a critical time as the debate evolves with payers, specialty professional groups, and malpractice carriers.

And the patient? First, recall that although the benefits of colorectal screening are huge when the small risk reduction from screening is multiplied across the population, even in ideal circumstances most patients do not directly benefit: that is they do not now and never will have colorectal cancer. They may experience some reassurance from a negative test result, but they otherwise experience only the inconvenience, discomfort, and costs of screening, the follow-up of false-positive results, and, in the case of screening colonoscopy, rare morbidity and mortality from the procedure itself or from removing polyps that would never become cancers. Further, if the large clinical trials are true, even among those who are screened, colorectal cancer mortality is reduced by no more than one-third, probably less, and all-cause mortality perhaps not at all. The number needed to screen to prevent 1 colorectal cancer-associated death is more than 1,000.⁷ Thus it is difficult enough for a patient to benefit under the best circumstances, so that the obstacles addressed in these 4 studies further illustrate the challenge.

Against this sobering backdrop, I marvel at how clinicians and patients maintain enthusiasm for screening. Family physicians who meticulously adhere to screening protocols may prevent only a handful of colorectal cancer deaths during an entire career, and they may not recognize it when it occurs (the problem of proving that something did not happen), so they receive little positive reinforcement for their effort. Most patients benefit only in the reassurance from a negative test result and along the way may have unneeded testing in response to positive screening test results that prove to be false-positives. These issues are fully aired in Gilbert Welch’s highly recommended 2004 book on cancer testing, Should I Be Tested for Cancer: Maybe Not and Here’s Why.⁸

I believe the most important audience for this new research should be policy makers, underscoring issues long neglected in our broken health care system and extending far beyond the particular question of colorectal cancer screening. Here is more evidence that having a personal health care provider matters. Here is more evidence that we need systems to deliver indicated services regardless of reason for visit. Here is more evidence that we need systems to insure adherence to proven clinical protocols. Here is more evidence that we need to find ways around irrational limitations on clinicians who could competently provide indicated services. These findings are potentially useful in the provision of many clinical interventions, not just those related to cancer screening. I hope that the current national environment for health care, with its many voices clamoring for change, will at last make it possible to address these issues as public policy is reshaped.