An epidemiologist's viewpoint on screening.

Abstract

Prostate cancer is an important disease causing a considerable number of new cases, deaths and premature loss of life each year. Three screening tests have been suggested: DRE, TRUS and PSA measurement in venous blood. From aggregated data from studies examining the performance of these tests, DRE or PSA measurement look the most promising with respect to their estimated detection rates and false positive rates (detection rate [DR] = 68%, false positive rate [FPR] = 5% for DRE and DR = 79%, FPR = 8% for PSA with a cut off of > 4 micrograms/l). However, these results are tentative-more research is needed on the performance of these screening tests. In particular, the prevalence of prostate cancer in the studies was considerably higher than that expected in the general population and so the OAPR estimates derived are more favourable than predicted. A randomized controlled trial of prostate cancer screening is needed to know whether screening reduces mortality from this disease, although this would take about 10 years to perform. Current stage distribution and 5 year survival data suggest that if screening could increase the proportion of cancers diagnosed at the localized clinically inapparent stage from 23% (currently diagnosed) to 100%, this would represent an equivalent reduction in mortality of 48%-about 1500 lives saved per year if screening were offered to 55-74 year old men. When more substantive information is available on the potential screening tests, it may be appropriate to conduct a randomized controlled trial. In the meantime, any ad hoc screening should be strongly discouraged.