Screening Sigmoidoscopy— How Often and How Good?

Flexible sigmoidoscopy is a member in good standing in the family of screening tests for colorectal cancer. Consistent evidence shows that even a single screening sigmoidoscopy reduces mortality from colorectal cancer by two thirds in the area examined, about half of the large bowel.^{1 - 4} All major screening guidelines recommend sigmoidoscopy, often coupled with fecal occult blood testing.^{5 - 7} In the United States, more than half of patients appropriately screened have had sigmoidoscopy.⁸ The cost-effectiveness of screening sigmoidoscopy compares favorably with other recommended screening tests for colorectal cancer⁹ and with screening for other cancers.

But several questions remain. How often should sigmoidoscopy be repeated in the course of a screening program? Is sigmoidoscopy a good enough option relative to other available tests, such as colonoscopy? How can sigmoidoscopy be made available to all adults who want to be screened in this way?

In this issue of THE JOURNAL, Schoen et al¹⁰ describe the yield of adenomas and cancers in the distal colon when sigmoidoscopy was repeated 3 years after a negative screening sigmoidoscopy. On the repeat examination, nearly 1% of patients (n = 72) had an advanced adenoma, the kind most likely to become malignant because these lesions are large or contain villous features or severe dysplasia. Six of the 9317 patients reexamined had cancer. The commonly recommended screening interval is 5 years, not 3, but even after 3 years, potentially dangerous neoplasms were found. It is not clear, however, whether finding advanced adenomas translates into the real goal, preventing colorectal cancer deaths, because it is not known how often advanced adenomas progress to symptomatic, incurable cancers.

Did these adenomas and cancers arise in the interval between examinations, or might they have been present before and missed on the earlier examination? The investigators reviewed medical records of the 72 patients with newly discovered advanced distal adenomas and found that only 14 (19.4%) could have been found because of increased depth of insertion or better bowel preparation on the follow-up examination. Four of the 6 newly discovered cancers in the distal colon also could not be attributed to inadequate examination. The implication is that about 80% of the advanced adenomas and a few cancers had arisen since the last examination 3 years earlier.

It is important to get the screening interval right. Mathematical models¹¹ show that as the interval between colorectal cancer screening tests, including sigmoidoscopy, is shortened, the cost of screening programs rise sharply, while effectiveness (colorectal cancer deaths prevented) increases very little. One reason for this is that the yield of subsequent screening is substantially less than that of the initial examination. In the study by Schoen et al, the ratio of cancers detected at the initial and follow-up examinations was 4.2:1.0. Even if cost were not an issue, more frequent examinations have other disadvantages. Screening sigmoidoscopy causes 1 perforation per 25 000 to 50 000 procedures performed in centers of excellence¹² ; the rate is much higher for all sigmoidoscopies in the general population.¹³ More frequent screening also increases exposure to the discomfort and inconvenience of the procedure. Because individuals in screening programs are asymptomatic in the first place, the overarching responsibility is to be sure that screening does more good than harm.

Even if the screening interval were 3 years, is sigmoidoscopy good enough, given the other screening options such as colonoscopy? No one disputes that screening sigmoidoscopy reduces mortality from colorectal cancer and probably also reduces incidence.¹⁴ But the procedure has some intuitively unattractive features. Only about half of the large bowel can be examined using sigmoidoscopy, and, since adenomas and cancers occur more or less evenly throughout the large bowel, only about half of them can be directly visualized. A policy of following up abnormal sigmoidoscopy with full colonoscopy detects an additional 20% of adenomas and cancers, but about 1% to 3% of adults have advanced proximal lesions that would not have been detected by sigmoidoscopy alone.¹⁵

This is a familiar dilemma. All screening tests, sigmoidoscopy included, are imperfect. They can improve patients' chances of a good outcome but cannot guarantee it. Screening tests are chosen because they are easier, safer, and less expensive than definitive tests, but at the cost of missing some cancers and sounding many false alarms. The greater complexity, risk, and expense of diagnostic tests are justified by the need to be very accurate because of the consequences of a false-positive or false-negative diagnosis of colorectal cancer. Perhaps clinicians and patients tend to expect colorectal cancer screening to be more effective than it really is because some of the same tests (sigmoidoscopy and colonoscopy) are used for both screening and diagnosis.

Clinicians and patients alike must wrestle with a value judgment: how effective is effective enough? Clinicians must be prepared to miss some cancers because many other factors—complications, inconvenience, discomfort, cost, cost-effectiveness, and the workforce needed to perform the procedures—are also in the balance when deciding which screening policies make the most sense.⁵ Although it would be ideal if screening sigmoidoscopy performed better, current procedures are at least as effective in reducing cancer-specific mortality as other screening tests in common use, such as mammography and clinical breast examination in women aged 50 years or older,¹⁶ and are better supported by evidence of effectiveness than prostate-specific antigen screening for prostate cancer¹⁷ and computed tomography screening for lung cancer.¹⁸

Another issue is how sigmoidoscopy can be made readily available to all who want to be screened in this way. Although screening for colorectal cancer saves lives and, based on strong evidence, has been recommended by expert groups for several years, only 30% to 40% of Americans (depending on age and race) are screened appropriately.¹⁹ Concerns about a few missed cancers in screened patients should not detract from the fact that most preventable cases of colorectal cancer occur in patients who were never screened.

Logistic difficulties impede efforts to implement screening sigmoidoscopy. Screening decisions take place in the offices of primary care physicians who usually have just 2 options, either to perform the examination themselves or to refer the patient to an endoscopist. Neither option works well enough. It is difficult to find time for routine sigmoidoscopies in a fast-paced primary care practice²⁰ and it is also difficult to maintain skills for procedures that are done infrequently. Office sigmoidoscopies may be money-losing if they are not done frequently.²¹ In contrast, referral physicians (mostly gastroenterologists) who perform endoscopies are hard-pressed to keep up with the recent increased demand for screening colonoscopy and diagnostic colonoscopy following other positive screening tests, in addition to their other work.

How can the medical community organize to make screening sigmoidoscopy available to all adults who choose it? Physician assistants and clinical nurse specialists can perform sigmoidoscopy as capably as physicians, provided they are well trained and appropriately supervised.²² Some large group practices have had successful sigmoidoscopy screening centers staffed by nonphysicians for years. Perhaps it is time for communities of physicians to form local centers for screening sigmoidoscopies, where the procedures are performed by local clinician specialists who are not necessarily gastroenterologists. With such centers, primary care physicians could initiate screening by a simple referral, just as they do now for mammography.

Most colorectal cancer deaths could be prevented. Screening with current tests at currently recommended intervals would contribute substantially toward reaching this goal if applied to most of the adult population. Better health-related lifestyle choices related to diet, physical activity, smoking, and obesity could reduce colorectal cancer incidence by half and would have many other health benefits too.²³ Evidence for the effectiveness of chemoprevention, with drugs as safe as folic acid or as simple as aspirin, is increasing and might play a greater role in the future.²⁴ New screening tests, notably virtual colonoscopy²⁵ and stool tests for DNA,²⁶ are under development and are likely to add to the screening armamentarium. Meanwhile, clinicians should make the best of available screening tests, and sigmoidoscopy is a mainstay.

In the quest for better ways to screen, clinicians should not overreact, pressing sigmoidoscopy beyond what it can reasonably do. While the results of the study by Schoen et al, as sound as they are, are not sufficient to prompt change in the currently recommended screening interval of 5 years, the findings do reveal much more about the consequences of this screening interval for cancer prevention than was known before.