Limitations
Confounders may exist in any non-blinded retrospective data set. However, the use of similar durations of time before and after adoption at sites that did not add new operators, with a large effect size seen across three distinct hospital sites, suggests that the influence of unmeasured confounders may be limited. Although retrospective, these data reflect the objective endpoint of actual procedure volumes. In the months prior to this analysis, regulations surrounding COVID-1923,24 impacted procedural volumes and caseloads across a variety of surgical specialties; however, procedure volumes at the study sites had generally returned to normal by the beginning of the study, consistent with trends seen at the national level.25,26 We did not acquire demographic or comorbidity data for this analysis; however, it is unlikely that there were significant demographic differences between each group. Study sites remained consistent through the analysis period, no major demographic transformations occurred at the sites, and in general, a trend of increasing age and comorbidity burden among patients undergoing AF ablation exists, which would be expected to bias against the post-adoption cohort.27,28
Operator experience can affect procedural efficiency, but physicians in this analysis were generally experienced or even senior operators with several years out of practice completed. As such, marginal gains in efficiency over time are likely to be minimal and of limited contribution to the findings reported.