In 2013, our group reported on the derivation and validation of an obstetric comorbidity index that predicts severe obstetric morbidity and maternal intensive care unit admission (Bateman BT et al. Obstet Gynecol 2013;122:957–65). The index includes maternal age and 20 maternal conditions identifiable at the time of admission for delivery, weighted to reflect the strength of their association with maternal morbidity. We showed that for each point increase in the score, the risk of severe maternal morbidity increased by 37% and that the index provided moderate discrimination (c-statistic of 0.66) for severe maternal morbidity. We further showed that the index performed significantly better than existing comorbidity scores, including the Charlson comorbidity index, in predicting maternal morbidity. However, our analysis was based on a single data source comprising US Medicaid beneficiaries and on conditions recorded using International Classification of Diseases (ICD), 9th edition codes, which are no longer widely used outside of the USA. The study by Metcalfe et al., which was extremely well designed and executed, accomplishes three important aims that extend the impact and relevance of the Obstetric Comorbidity Index: (1) it shows that the index is valid in a population (Alberta, Canada) that is distinct from that in which it was originally derived; (2) it demonstrates that the index provides reasonable discrimination even when information is derived solely from conditions recorded at the time of the delivery admission; and (3) it adapts the index for use with ICD-10 codes. The discrimination of the Obstetric Comorbidity Index for predicting maternal end-organ damage based on all healthcare contacts in the Alberta population was 0.70, which is on a par with other prediction rules that are widely used in other clinical domains. There are a number of applications for this type of comorbidity score. It can provide a summary estimate of patients’ burden of comorbid illness that can be tracked across providers, institutions, regions, and even countries. It can also be used as a data reduction approach to aid in confounding control in epidemiologic and health services research. Clinically, the index may have utility as a screening tool for triaging high-risk patients to specialised practices and institutions that are well equipped to provide care for complex patients at the time of delivery. Designated levels of maternity care have been suggested as an approach to decrease the occurrence of maternal morbidity and mortality in the USA (Hankins GD et al. Obstet Gynecol 2012;120:929–34), and a high comorbidity score may be able to be used to flag women in need of high level care. Such clinical applications of the Obstetric Comorbidity Index will need to be validated in future prospective clinical studies. However, the relatively high discrimination exhibited by the index, now validated in multiple populations, suggests that such clinical uses for the Index are promising.
B. Bateman, J. Gagne
BJOG: An International Journal of Obstetrics & Gynaecology