Adjusting for confounding in population administrative data when confounders are only measured in a linked cohort | Synapse