Linked data are increasingly being used to inform health policy and practice. Yet the impact of linkage error on these estimates is often poorly understood, with potentially important impacts on the evidence used to inform policy.