The article by Dr. Schapira and colleagues [1] and the accompanying editorial by Dr. Pauker [2] reach a correct conclusion by a misleading analysis. Screening for ovarian cancer is indeed ineffective, but not because the disease is rare. Screening is ineffective because we do not know the cause of the disease and do not have effective therapy.

Something is wrong with their application of decision analysis when it leads to such a nihilistic conclusion (that very limited benefit can be obtained through screening with a perfect test with a riskless work-up for a uniformly fatal disease because—when prevalence is low—the maximum benefit to each person screened is only 2 to 18 days). The problem is their use of a global statistical end point (the number of days of extended life provided, on average, to each individual screened). Such an argument would prove, for example, that we should not screen for phenylketonuria—another rare disease—because the average benefit to each child screened is less than a month. Yet, for an individual child with phenylketonuria, the life-extension benefit is measured in years, not days, and the psychosocial benefit may be boundless. Isn't that what medicine is supposed to be about?

Life is finite, and we can do little to extend life beyond its maximum limit, but we can do a great deal to improve the quality of life during the interval from birth to death.