The Burden of Proof Risk Function: A new approach to evaluate health risk

Presenter: Dr. Michael Brauer is a Professor in the School of Population and Public Health at The University of British Columbia and a Principal Research Scientist and Affiliate Professor at the Institute for Health Metrics and Evaluation at the University of Washington, where he leads the Environmental, Occupational and Dietary Risk Factors team for the Global Burden of Disease. His research focuses on linkages between the built environment and human health, with specific interest in the global health impacts of air pollution, the relationships between multiple exposures mediated by urban form and population health, and health impacts of a changing climate. Summary: Assessing strength and quantifying risk-outcome relationships is critical for public health prioritization, policy formulation, clinical guidance and to inform personal choices related to modifiable risk factors. While meta-analyses or meta-regressions are often used as an inputs, their use in decision making is often subjective. We introduce the quantitative burden of proof risk function in the context of the Global Burden of Disease comparative risk assessment framework. The burden of proof risk function combines meta-regression mean relationship between exposure and risk with unexplained between-study heterogeneity, adjusted for number of studies. We developed and applied a Bayesian meta regression framework to robustly estimate the mean and burden of proof risk function, allowing for non-linear relationships, and applied this to 197 behavioral, metabolic and environmental risk factor – outcome relationships. Relationships were summarized by a ‘star-rating’ where 1-star risks had a probability of no association after accounting for between-study heterogeneity, and 2, 3, 4, and 5-star risks indicated a 0-15%, 15-50%, 50-85% and 85% increase in risk, respectively, over the 15%-85% percentiles of exposures of included studies. The burden of proof risk function is a cautious interpretation of evidence, incorporating both magnitude and uncertainty in relative risks. Higher risk scores indicate a larger effect and/or a lower probability of results driven by residual confounding or other bias. Risk-outcome pairs with low star ratings indicating substantial between-study heterogeneity may suggest a need for additional studies, especially where exposure and outcome prevalence are high.

BC CDC Presenters

12/13/2022 8:00:00 PM