As an ObGyn resident @uoftobgyn and a master’s student in Clin Epi @ihpmeuoft performing research using population-based health admin data, here are two quick thoughts on this paper which has generated significant discussion (not meant to be a critical appraisal) 1/10. https://twitter.com/JAMAPediatrics/status/1315683665536528384

2/11 Epidural analgesia is a vital tool in safe labour/delivery management. This paper risks generating panic, being misinterpreted and misquoted, and impacting a patient’s ability to obtain adequate analgesia in labour (which comes with long-term consequences).

3/10 I see obvious risk of residual confounding. While the authors have controlled for some important potential confounders (including using propensity methods), they are limited by those variables available to them in the database.

4/11 Those who practice obstetrics in settings similar to the Kaiser Permanente Southern California hospitals recognize that the group of individuals who choose to labour without an epidural tend to differ from your average labouring patient.

5/11 They, on average, tend to possess unique cultural, behavioural, lifestyle, and personality traits, as well as unique obstetrical and labour/delivery circumstances.

6/11 These factors (applicable both ante- and post-natally) could not be controlled for in the analysis and may act as significant residual confounders explaining the observed association, especially in the absence of a clear mechanism of action.

7/11 I think there’s also potential for information bias in the form of differential outcome misclassification. Outcome assessment relies on the presence of diagnostic codes within with the database. These codes have been previously validated to have a PPV of 88%.

8/11 An example of where differential outcome misclassification could occur would be if those who defer epidurals are less likely to seek care for or medicalize milder forms of Autism, or more likely to seek alternative forms of care outside of Kaiser Permentante (and vice versa)

9/11 I believe that these alternative care seeking patters are plausible and would result in bias away from the null.

10/11 So overall, while the authors have identified an association which may justify further investigation, I do not believe it to be a causal one, and believe that epidural analgesia acts more likely as a marker for other factors, or simply an association identified by chance.

11/11 Thanks to these organizations (and likely others) who have spoken out for the safety of epidural analgesia: @MySMFM @ASALifeline @acog @PediAnesthesia