Variation in Clinical Trial Results for Asthma by Environmental and Socioeconomic Exposures

Abstract

Randomized clinical trials (RCTs) are generally considered the gold standard in medical research; by randomizing participants to treatment and control arms, they maximize internal validity and reduce between-group biases. RCTs, however, do not routinely examine variation by environmental and social exposures (e.g., socioeconomic status), which may impact clinical outcomes, treatment response, and study generalizability. To assess whether variation in socioeconomic positon (SEP) and environmental exposures modify treatment response, we have developed and applied Geographic Information Systems (GIS)-based methods to examine three RCTs performed by the NIH AsthmaNet network, which recruited and implemented RCTs using the same protocols across 17 U.S. cities. In preliminary analyses, we have found that: (1) compared to race-specific U.S. averages, AsthmaNet participants disproportionately live in impoverished census tracts. (2) At baseline, traffic density and SEP predicted variation in lung function, in the hypothesized directions. (3) In one longitudinal trial, we found that near-residence roadway density explained greater variation in asthma symptoms than did corticosteroid use, and that children in higher-poverty areas had significantly shorter times to first corticosteroid use. For this trial, we examined effect modification by near-residence roadway traffic and noise exposures, neighborhood violence, and health care access. Preliminary results suggest - only among higher-SES children - significantly greater negative impacts (shorter times to prednisone use) among participants receiving higher inhaled corticosteroid (ICS) doses (220 ug), relative to those receiving lower ICS doses (44 ug).Results suggest that participants in clinical trials may not be representative of U.S. asthmatics, and participants from different SES backgrounds may differentially respond to intensive corticosteroid interventions. Using spatial analysis and GIS to understand the neighborhoods of RCT participants – better accounting for socioeconomic and environmental conditions – may improve the interpretability and applicability of RCT results, by more clearly identifying subpopulations for whom a given intervention may be most effective.