| The concept of screening for chronic disease in apparently healthy people has its roots in the belief that the function of medicine is not only to cure the ill but to prevent the onset of disease and thus preserve the health of people without overt disease. As a public health measure, screening has been implemented widely for cervical cancer, breast cancer, hypertension, prostate cancer and lung cancer, to name a few diseases. Before undertaking such measures, society must consider the costs and benefits. To this end, clinical trials of mass screening programs have been undertaken during the past two decades. The intent of these trials is to determine the effect of screening on the health and longevity of those screened and to extrapolate from the results to the populace at large.; After preliminary discussions in Chapter I of the cross-sectional model of screening, including the concepts of sensitivity, specificity, prevalence and predictivity, more sophisticated models are reviewed in Chapter II for their applicability to the analysis of clinical trials. The models reviewed are classified as to whether they are simulation models or data analytical models. Chapter III summarizes the issues which arise in the design and analysis of screening trials, including lead time bias and length bias in estimating case-fatality rates and the multitude of outcomes considered.; Chapters IV and V address the design issues of randomization, stratification and replication, and in so doing suggest designs and methods of sample size determination. Chapter VI proposes models for estimating some of the screening test parameters of interest, especially when tests are repeated or when there are multiple tests. Chapter VII proposes a new model for the estimation of lead time, a robust survivorship model applicable to screening studies and a general approach to cost/benefit analysis in screening. |
