Methodological considerations when studying ethnic disparities in epidural access: a reply

We thank Lyons [1] for his interest in our study exploring ethnic disparities in the utilisation of epidural analgesia in Aotearoa New Zealand [2]. Lyons summarises the core questions behind the investigation of which our study is just the first step. There are many possible reasons why people from different communities may have different attitudes to epidurals, including the role and influence of the lead maternity carer. Understanding these issues may help improve equity of access to epidural analgesia in labour. Before setting out to explore these factors, we needed to understand if there are in fact differences in epidural provision. Our results help provide a baseline and context for ongoing research. The decision not to include caesarean sections from the analysis came after considerable discussion in our group. We are interested primarily in the information and attitudes within a community that may influence the choice to use epidural analgesia. The occurrence of caesarean section adds a medical overlay to this process and would introduce a different set of confounders. However, we agree with Lyons that there are differences in caesarean section rates which deserve further investigation. The interaction between lead maternity carers and the use of epidural analgesia in their patients is complex and cannot be explored fully from the dataset used here. However, an ancillary analysis based on our dataset found that among lead maternity carers who had managed > 10 patients overall, there were 145 lead maternity carers who had managed at least one Māori and one European primiparous mother. Of these, 75 (52%) had treated a higher proportion of European mothers with epidurals than they had Māori mothers, which was not significantly more than a hypothesised 50% split (p = 0.68). The corresponding proportion for European vs. Pacific Peoples was 77 (70%, p < 0.01). Put another way, Māori and European primiparous patients of an individual lead maternity carer appear just as likely to receive an epidural, but the same lead maternity carer is less likely to be involved with an epidural for a mother from Pacific Peoples. Regarding validation of the modelling, areas under the curve (AUC) for each variable in training and test can be seen in our online Supporting Information Table S2 [2]. To add further detail, the values for the AUC for the overall models are as follows. For primiparous women, the AUC were: univariable (train 0.53 vs. test 0.53); lead maternity carer (train 0.65 vs. test 0.59); and full multivariable (train 0.68 vs. test 0.62). For multiparous women, the corresponding AUC were univariable (train 0.56 vs. test 0.52); lead maternity carer (train 0.71 vs. test 0.65); and full multivariable (train 0.75 vs. test 0.69). As Lyons suggests, understanding more about the various factors influencing the utilisation of epidural analgesia among different groups within our city is an important step in improving equity of access. Our results to date provide a basis for our ongoing investigations in this area [2].