Some of the problems associated with models of screening for colorectal cancer are related to the multitude of factors and their interactions, which can potentially influence the outcome. Dr. Chelmowski is correct in suggesting that a markedly reduced cost of flexible sigmoidoscopy could render this procedure more cost-effective. Assuming perfect adherence and a sigmoidoscopy cost of less than $162, the incremental cost-effectiveness ratio of flexible sigmoidoscopy (compared with no screening) falls below that of the fecal occult blood test or colonoscopy. Assuming adherence rates of 75% for both repeated sigmoidoscopy every 5 years and colonoscopy after positive results on sigmoidoscopy, this threshold drops to $113.

Dr. Das suggests varying another set of variables in the screening model. However, shaving off the costs of two blood tests from the overall cost of colonoscopy does little to affect its average and incremental cost-effectiveness ratios. Adjustments for the quality of life associated with various types of screening or surgery would add a slew of poorly defined variables to the model that are difficult to measure and do not improve its overall precision. If reliable adjustments were known for the quality of life saved through various procedures, they could be applied to the life-years shown in our Table 2. Varying the discount rate from 0% to 7% changes the absolute values calculated for the average and incremental cost-effectiveness ratios but does not affect the proportionate relations among the strategies. As we discussed, changes in the surveillance interval after polypectomy exert only small influences on all three screening programs.

The primary aim of the analysis was to compare no screening with three common screening strategies for colorectal cancer, that is, fecal occult blood test, flexible sigmoidoscopy, and colonoscopy. Considering what is known about the effectiveness, performance characteristics, and relative costs of the three screening options, the analysis led us to conclude that colonoscopy would compare quite well with the others. Several other issues were not addressed by our analysis. For example, should society embark on a general screening program for colorectal cancer? How much money should be made available for such a program, and what is its urgency compared with other pressing societal and medical needs? In addition, as Drs. Budenholzer and Welch point out, there is little doubt about the primacy of prospectively collected data for assessment of the effectiveness of screening in preventing cancer and in saving lives.