Key points: (max 100 words)
1. Distributed networks are rapidly becoming a cornerstone of
pharmacoepidemiologic research.
2. Given their use of multiple real-world study populations, external
validity is an essential but under-investigated part of how these
networks generate data on drug safety and effectiveness.
3. How Sentinel, CNODES, PCORnet, and other networks approach external
validity is a function of their official purposes, the study questions
they answer, and the data partners they include.
4. Formal statistical methods to improve external validity of estimates
are rarely implemented within these networks for a variety of
reasons.