Why clinical practice guidelines shift over time : a dynamic model with application to prostate cancer screening

Abstract

Essay 1: A Dynamic Model for Understanding Long-Term Trends in Prostate Cancer Screening Cancer remains the second leading cause of death in the U.S. after heart disease. After 35 years of routine cancer screening, we still have only a limited understanding of screening dynamics. There is evidence of over-screening and resulting overtreatment in certain cases, and significant provider variation and fluctuations over time in screening criteria. Here I present empirical data for fluctuations in official screening guidelines and in actual practice for the use of the prostate-specific antigen (PSA) test. I explore how these dynamics are affected by the main guideline-issuing organizations in the U.S. and by clinicians, patient groups, and the media. Essay 2: Our Walk to the End of Cancer? Understanding Long-Term Trends in Medical Screening In this study we develop the first integrated, broad boundary feedback theory and formal model to explain the dynamics of medical screening. The theory includes a decision-theoretic core around harms and benefits including the fundamental tradeoff between sensitivity and specificity; and feedbacks that condition guidelines and actual practice. To provide context we use the case of PSA screening for prostate cancer as a motivating example, but our model is generic and applicable to other contexts. We present a behaviorally realistic, boundedly-rational model of detection and selection for health screening that creates oscillations in policy recommendation thresholds of formal guidelines. This core model, entailing only the evidence generation and translation processes, demonstrates how oscillations are natural to this category of problems due to inherent delays in evidence-based screening. These fluctuations lead to long periods during which screening guidelines are suboptimal. Essay 3: A Dynamic Model for Understanding Long-Term Trends in Prostate Cancer Screening Whereas guidelines for routine screening should be based on medical evidence, evidence often has relatively little impact on practice. This situation has led to ongoing controversy and conflict over appropriate guidelines among scientists, clinicians, and patient advocacy groups. There are significant variations in clinical practice, including evidence of over-screening for some diseases, and under-screening for others. To explain the patterns of over-screening, fluctuations, low adherence to guidelines, and conflict, I develop the first explicit broad boundary feedback theory of the dynamics of medical screening, tested in a formal mathematical model. The model presents an extended case study specific to PSA screening for prostate cancer, including realistic presentations for the fundamental tradeoff between test sensitivity and specificity, the natural progression of the disease, and respective changes in population size and composition.