After-hours surgery: concerns in confounders and conclusion – a reply

Abstract

We thank Sakurai and Takeda [1] for their interest in our study [2]. It is inevitable there will be confounding variables present in a retrospective, large-scale population-based, cohort study. Ultimately, we assert that effective reflection on and control of confounding variables enhances the reliability of the study findings.

Initially, it was necessary to incorporate the emergency surgery group to guarantee a sufficiently large after-hours cohort; this clearly represents a constraint. We tried to enhance the reliability of our findings by accounting for the complexity and risk associated with the surgical procedures, referred to as ‘total relative value unit’. We trust this addresses the concerns regarding the absence of information on surgical procedures.

Regarding the absolute standardised mean difference calculation, we have verified that 0.071 is accurate. The calculation of absolute standardised mean difference clearly considers the substantial numbers of patients (very large in our study), the 1:5 matching ratio and the 0.25 matching calliper. We consulted our statistician regarding the varying outcomes from the formula by Sakurai and Takeda [1], and they indicated that it is improbable that all factors were considered.

In summary, we acknowledge that there are constraints to our findings. There appears to be significant potential for additional investigation.