One of the most common reasons for visits to family physicians is the detection of asymptomatic disease, or screening—especially screening for cancer. In this issue of American Family Physician, Zoorob and colleagues1 provide a useful summary of current guidelines for cancer screening in adults. Two facts stand out from their review. First, compelling evidence and relatively unanimous agreement exist for only three cancer screening recommendations: Papanicolaou (Pap) smears; mammography in women older than 50 years; and fecal occult blood testing (FOBT), flexible sigmoidoscopy or colonoscopy to screen for colon cancer. Second, a plethora of medical organizations, government task forces, specialty societies and advocacy groups are making recommendations that are sometimes contradictory. I believe that in order to make sense of current as well as future screening controversies, three tasks are essential for the family physician.
1. We must understand the nature of evidence that is required to justify screening tests. The value of screening and early detection seems intuitively obvious to us but, in fact, our intuition is often misleading. Specifically, certain types of bias frequently lead to an overestimation of the benefits of screening.2,3
“Lead-time” bias occurs because diagnosis by screening during the asymptomatic period will lead to improvement in such surrogate measures as five-year survival rates and median survival rates, even in the absence of therapy. However, the apparent increase in survival time does not necessarily mean that patients are living longer but, rather, that they find out about their disease earlier. Final mortality may remain unchanged despite improvements in five-year survival rates, as it did in lung cancer screening trials.4
“Length-time” bias occurs because slow-growing, less aggressive tumors by definition have a longer asymptomatic period and are therefore more likely to be detected by screening. Consequently, screening programs preferentially identify a cohort of patients with less aggressive tumors; even in the absence of therapy, these patients will have a better prognosis than patients who present with symptomatic disease.
Finally, “screening” bias refers to the observation that patients who volunteer for screening (or in randomized trials, those who comply with screening recommendations) tend to be healthier than those who do not volunteer or do not comply. Thus, an observed benefit may be the result of self-selection of a cohort of healthier volunteers rather than an observed benefit from screening.
Because of the complex and unpredictable nature of these various biases, the only reliable way to prove the effectiveness of a screening intervention is to demonstrate a lower mortality rate in a randomly assigned screened population, compared with unscreened control patients, using intention-to-treat analysis.2,3 Such studies are long and expensive, but screening interventions that have not met this standard should be considered unproven and experimental.
2. We must recognize the unique ethical context of the screening encounter. In ordinary medical practice, the patient initiates the encounter because of a troubling symptom and is often willing to accept reasonable but unproven therapy in hopes of finding relief. In contrast, the screening encounter is usually initiated by the physician (although sometimes indirectly, through professional or advocacy groups) and involves an asymptomatic and presumably healthy patient. In this situation, there is an “implied promise,” not just that the screening procedure might be worthwhile, but that it is, in fact, of proven benefit, and likely to result in more good than harm being done.2
Even if a screening test has proven benefits, those benefits accrue to only a few individuals. In contrast, everyone who participates in a screening program is at risk of harm.5 Over and above the cost and discomfort of the actual test, screening can harm patients by having false-positive results (e.g., the 50 percent 10-year cumulative false-positive rate for mammography),6 false-negative results (e.g., false reassurance in a smoker with a negative result on chest radiography), and even by a true-positive result if the therapy is unproven and potentially dangerous (e.g., radical prostatectomy for low-grade localized prostate cancer).
3. Finally, for proven screening interventions, we must understand the magnitude of the potential benefit and be able to convey this to our patients in a meaningful way. The efficacy of screening interventions is most often reported in terms of the relative risk reduction (RRR); for example, “mammography reduces breast cancer mortality by 23 percent” rather than the absolute risk reduction (ARR); for example, “annual mammography has a 0.2 percent chance of preventing an individual woman's death from breast cancer over 10 years.”
However, the number needed to screen (analogous to the number needed to treat in therapeutic trials, calculated as the reciprocal of the ARR)9 provides a better method of expressing the effectiveness of a screening intervention.10 Numbers needed to screen (i.e., the number of patients that would have to be enrolled in an ongoing screening program, here normalized over 10 years, to prevent one death from the disease in question) vary from about 540 (for mammography in women older than 50 years) to about 1,100 (for Pap smears), with numbers for FOBT screening falling between those two.
Physicians may become discouraged by the apparent inefficiency of screening for cancer as measured by the relatively high numbers needed to screen. We must keep in mind, however, that by scrupulous adherence to cancer screening guidelines and proven screening interventions, the average family physician can expect to prevent several premature deaths from cancer over the course of a professional career. High numbers needed to screen also remind us that true prevention (e.g., preventing lung cancer by encouraging smoking cessation) will usually be more efficient than early detection through screening. Likewise, these numbers remind us that, for most patients, careful attention to cardiovascular risk factors (smoking, hypertension, hyperlipidemia and physical inactivity) will yield higher dividends than cancer screening.10
One of our most challenging tasks as family physicians is to better convey this kind of prioritizing information to our patients—especially those sedentary, noncompliant, hypertensive patients who smoke, and who come in each year for their prostate-specific antigen screening.