Introduction Research on natural experiments (NEs) allows investigation into topics that are important for policy-makers, including many social determinants of health. NEs can be conceptualised as differing from other research in three important ways. First, unlike trials, investigators cannot allocate the exposure of interest which is instead externally decided. Second, researchers are unable to influence the exposure's characteristics which results in uncertainty of the fidelity, dose and whether the NE will occur at all. Third, and in contrast to many observational studies, the timing of the exposure and hence research itself, cannot be chosen by researchers.

Methods Using three case studies of research on H1N1 (two quantitative sero-epidemiology studies, one qualitative study) we identified key factors that allowed us to successfully conduct our investigation of a NE. We compared these factors with the wider NE literature to identify common barriers and facilitators to research.

Results We identified key external factors that influence capacity to successfully take advantage of NEs. As a result of the uncertainty of a NE occurring and the rapid response required, flexibility is needed by researchers, policymakers, practitioners, ethics committees and funders. This is achievable when research is perceived as immediately important for health (as in H1N1) but may otherwise be difficult.

Discussion We suggest researchers investigating NEs face additional challenges to traditional observational studies. While some barriers are insurmountable, actions such as rapid response funding, fast-track ethics procedures and improvements in routine data can create a more conducive environment allowing policy-relevant evaluation.