Risk scoring system for prediction of obstetric anal sphincter injury

Abstract

The objective was to begin the process of developing an antenatal risk scoring system, as a first step towards examining whether elective Caesarean section for women at high risk of injury could be an effective and acceptable intervention. Retrospective study. Tertiary maternity unit in the UK. One hundred and twenty-three women who sustained an obstetric anal sphincter injury (OASI) and 123 controls without OASI. Case notes of women with a third or fourth degree tear between 1997 and 1999 were examined for risk factors. Controls matched for age and week of delivery were identified from the maternity record database and case records reviewed for the presence of risk factors. Unweighted and weighted risk scores were produced using odds ratios, and compared between cases and controls. Receiver operating characteristics (ROC) curve analysis of the risk scores was performed to discriminate between cases and controls and to calculate the sensitivity and specificity of each scoring system. Odds ratio (OR) and 95% confidence interval (CI) for each risk factor. Sensitivity and specificity from ROC curves for weighted and unweighted risk score. Among the cases there were more nulliparous women (OR 1.77; CI 1.05-2.99) and a trend towards more women with an episiotomy (OR 1.57; CI 0.99-2.47). Among women with sphincter injury, trends towards more epidurals (OR 1.64; CI 0.97-2.75), and more babies weighing more than 4000 g among (OR 1.45; CI 0.85-2.49) were noted. The median unweighted risk score was 2 for cases and 2 for controls (P= 0.05), while the weighted risk score was 2.1 and 1.37 (P= 0.03), respectively. The ROC curves approximated to a straight line demonstrating very poor discrimination between cases and controls. The predictive test performed poorly, suggesting that the risk factors identified do not exert a large enough effect in a cohort of this size.