Framework for Using Risk Stratification to Improve Clinical Preventive Service Guidelines

Abstract

People should only receive a preventive service if the potential benefits of the service outweigh the potential harms. Both benefits and risks may vary for different populations. Thus, it is clinically important to understand when and how guidelines for preventive services should be stratified according to the underlying risk of the population. For example, preventive services may be risk stratified with specific clinical recommendations based on age, sex, race/ethnicity, family history, genotype, behavior risks, or comorbidities. This paper articulates the conceptual approach and practical tools that were developed for consideration by the U.S. Preventive Services Task Force to determine if and how risk stratification should be incorporated into clinical guidelines. This approach is described in an algorithm with six sequential questions: (1) Are there clinically relevant subpopulations? (2) Are there credible subgroup analyses for these subpopulations? (3) Do subgroup analyses show clinically important differences? (4) Do these differences result in variation of net benefit, or does the evidence only exist in persons with a narrow spectrum of risk? (5) Can the subpopulations be easily identified? and (6) Does a well-validated multivariate risk tool improve identification of clinically relevant subpopulations compared with a simpler approach? This framework allows for a systematic approach to determine if and how to incorporate evidence for specific populations, a consistent application of critical thinking about this evidence, and transparent communication about the derivation of risk-stratified recommendations or evidence gaps.