Validation of a comorbidity index for use in obstetric patients: A nationwide cohort study.

Abstract

A previously developed Obstetric Comorbidity Index has been validated in highly selected cohorts. Validation of the index in an unselected population as well as in other health registers is, however, of high importance to determine external validity. Using nationwide registers, we formed a nationwide cohort including completed pregnancies (both live- and stillborn) in Denmark from 2000 through 2014. Maternal age and 20 comorbid conditions were assessed and weighted. Outcomes were maternal end-organ injury or death within 30days postpartum. The index's predictive and discriminative ability was estimated by Brier score and the area under the receiver operating characteristic curve (AUC), respectively. Logistic regression analysis was used to estimate odds ratios (OR) with 95% confidence interval (CI). In 876496 completed pregnancies by 527079 women, 1.40% (n=12314) experienced an outcome. The majority of women (64.1%) did not have any record of a condition included in the index and only 0.3% (n=3044) had a score >6. The incidence of an outcome increased with increasing comorbidity score from 0.9% (95% CI 0.8-0.9) in women scoring 0% to 10.4% (95% CI 7.6-13.9) in women scoring 9-10. Compared with women scoring 0, a score of 1-2 yielded an OR of 2.34 (95% CI 2.25-2.44), 3-4 an OR of 5.16 (95% CI 4.81-5.54), 5-6 an OR of 4.84 (95% CI 4.31-5.44), and 8-9 an OR of 7.97 (95% CI 6.54-9.72) for experiencing the outcome. The index had a Brier score of 0.01 and an AUC of 0.64. Despite potential weaknesses in the outcome definition, the Obstetric Comorbidity Index showed a moderate ability to discriminate and predict end-organ injury and death in a nationwide cohort in Denmark, in accordance with previous findings. These results suggest that the index may be a useful tool to control for confounding in health research and clinically to identify women at high risk for adverse maternal outcomes.