Intrapartum prediction system for third-degree perineal trauma

Objective The incidence of anal sphincter trauma has seen a dramatic rise in recent years, which is most likely due to improved clinical practice during postdelivery examination. The risk factors for sustaining such an injury have been investigated thoroughly. In this study, the authors assessed the feasibility of developing an antenatal prediction system for anal sphincter trauma based on the presence of one or more of these risk factors. Methods This is a retrospective study using data from all women delivering vaginally at Aarhus university hospital, Denmark, from 1989 to 2006 (n=71 469). The cohort was randomly divided into two equal cohorts. A univariate and multivariate analyses were carried out on the first cohort. The final logistic regression model was adapted through the backward stepwise selection function and then validated against the second cohort of data. Coefficients from the final model were then utilised to produce a software interface for the prediction of a women's risk. Results The final multiple logistic regression model used maternal age at delivery, parity, gestational age, macrosomia, fetal position, length of the pushing phase, instrumental delivery and mediolateral episiotomy as the statistically significant predictor variables. This model showed good discrimination capability among the first cohort and validation dataset, with an area under the receiver operating curve of 0.79 and 0.78, respectively. Conclusion This study has seen the development of a predelivery prediction system for use by clinicians in the identification and subsequent counselling of women at high risk for this type of injury.