We thank Dr. Hutcheon et al for their interest in our study on Oregon's hard-stop policy.1 We agree with the need for a strong evidence base supporting obstetric practice and policy for maximal maternal–neonatal health.
Our approach of modeling time using a single binary main term is a logical starting place to assess the presence and magnitude of an adjusted association in a multivariable modeling framework but does not address known secular trends in practice, policy climate, and outcomes, which we acknowledged.1 Therefore, it is also logical to consider approaches such as difference-in-differences and interrupted time series, as well as more advanced methods including data-adaptive splines accounting for spatial as well as temporal variability.2 More fine-grained analytical treatments of time come at a methodologic cost: each requires the investigator to incur additional assumptions (eg, variations of the untestable parallel paths assumption for interrupted time-series and difference-in-differences).3,4 More restrictive assumptions have less chance of being satisfied, with implications for study validity. Nonetheless, we are now publishing temporally fine-grained analyses of Oregon's hard-stop policy.
Designing, implementing, and evaluating real-world policies is a complex process that often does not conform to rigid assumptions required in statistical modeling. We agree with Dr. Hutcheon et al that rigorous and thoughtful deployment of various analytical techniques best informs researchers, policymakers, and practitioners on the true effects of policies.