Epidural Analgesia and the Risk of Obstetric Anal Sphincter Injury (OASI)

Objectives: There are conflicting data regarding the association between epidural analgesia and obstetric anal sphincter injuries (OASI). This study aimed to build a predictive risk model for OASI, identify effect modifiers and confounders, and assess the association between epidural analgesia and OASI in a diverse, urban population.

Methods: This was a secondary analysis of a retrospective case-control study including women who had a term, singleton vaginal delivery at a large, urban hospital. Individuals with the primary outcome of OASI were identified and matched 1:4 with controls by delivery date. Our primary analysis evaluated the association between epidural analgesia and OASI, adjusting for risk factors using multivariable logistic regression analysis. Demographic characteristics and maternal medical history-related risk factors were evaluated for multicollinearity using Chi-square or Fisher’s Exact test. Effect modification was assessed with the Woolf test. Confounders were identified by calculating percent differences between crude and Mantel-Haenszel (MH) odds ratios.

Results: Our sample consisted of 1,326 women. There were 289 women (21.8%) with OASI and 1,037 women (78.2%) without OASI. Women with OASI were more likely to be Asian (9.0% vs 4.2%) or White (58.5% vs 43.8%), have a history of thyroid disease (11.8% vs 5.3%), be nulliparous (70.9% vs 34.5%), have a labor induction (36.0% vs 28.1%), episiotomy (4.2% vs 1.4%), operative delivery (31.1% vs 3.5%), and neonatal birth weight >4000g (13.5% vs 7.4%; all p≤ 0.05).
Among the women with OASI, 218 women (75.4%) had an epidural, and among those without OASI, 620 women (59.8%) had an epidural (p < 0.001). In bivariable analysis, epidural analgesia increased the risk of OASI by 107% (OR 2.07, 95% CI [1.54 to 2.77]). Assessment of multicollinearity demonstrated a significant association with operative delivery and episiotomy to epidurals. Variables that did not demonstrate multicollinearity were evaluated for effect modification and confounding. None of the four risk factors analyzed were identified as effect modifiers and the only confounders were operative delivery and labor induction (Table 1).
Therefore, the final multivariable logistic regression model controlled for operative delivery and labor induction as confounders and neonatal birth weight >4,000 grams and thyroid disease as covariates. When accounting for these additional factors, we found that the association between epidural analgesia and OASI was no longer statistically significant (aOR 1.38, 95% CI [1.0, 1.91]; Figure 1).

Conclusions: We found epidural analgesia to be an independent risk factor for OASI, however, after accounting for additional risk factors and performing extensive statistical analysis to adjust for multicollinearity, confounders, and effect modification, this association was no longer statistically significant.