Dimension reduction of 911 Good Samaritan Laws: Drawing inferences from policy surveillance

Abstract

Background911 Good Samaritan Laws (GSLs) extend legal protection to people reporting drug overdoses who may otherwise be in violation of controlled substance laws. Mixed evidence suggests GSLs decrease overdose mortality, but these studies overlook substantial heterogeneity across states. The GSL Inventory exhaustively catalogs features of these laws into four categories: breadth, burden, strength, and exemption. The present study reduces this dataset to reveal patterns in implementation, facilitate future evaluations, and to produce a roadmap for the dimension reduction of further policy surveillance datasets. MethodsWe produced multidimensional scaling plots visualizing the frequency of co-occurring GSL features from the GSL Inventory as well as similarity among state laws. We clustered laws into meaningful groups by shared features; produced a decision tree identifying salient features predicting group membership; scored their relative breadth, burden, strength, and exemption of immunity; and associated groups with state sociopolitical and sociodemographic variables. ResultsIn the feature plot, breadth and strength features segregate from burdens and exemptions. Regions in the state plot differentiate quantity of substances immunized, burden of reporting requirements, and immunity for probationers. State laws may be clustered into five groups distinguished by proximity, salient features, and sociopolitical variables. DiscussionThis study reveals competing attitudes toward harm reduction that underly GSLs across states. These analyses provide a roadmap for the application of dimension reduction methods to policy surveillance datasets, accommodating their binary structure and longitudinal observations. These methods preserve higher-dimensional variance in a form amenable to statistical evaluation.