Protecting privacy while adequately adjusting for a large number of covariates poses methodological challenges for distributed data networks that can enable large-scale epidemiologic studies. Using 2 empirical examples, Li et al. determined that when used in conjunction with confounder summary scores, several combinations of data-sharing approaches and confounding adjustment methods allow researchers to perform multivariable-adjusted analysis using only aggregate-level information from participating sites and produce results identical to or comparable to those from pooled individual-level data analysis which help to protect privacy.
