Methodology for sampling women at high maternal risk in administrative data.

BACKGROUND:In population level studies, the conventional practice of categorizing women into low and high maternal risk samples relies upon ascertaining the presence of various comorbid conditions in administrative data. Two problems with the conventional method include variability in the recommended comorbidities to consider and inability to distinguish between maternal and fetal risks. High maternal risk sample selection may be improved by using the Obstetric Comorbidity Index (OCI), a system of risk scoring based on weighting comorbidities associated with maternal end organ damage. The purpose of this study was to compare the net benefit of using OCI risk scoring vs the conventional risk identification method to identify a sample of women at high maternal risk in administrative data. METHODS:This was a net benefit analysis using linked delivery hospitalization discharge and vital records data for women experiencing singleton births in Georgia from 2008 to 2012. We compared the value identifying a sample of women at high maternal risk using the OCI score to the conventional method of dichotomous identification of any comorbidities. Value was measured by the ability to select a sample of women designated as high maternal risk who experienced severe maternal morbidity or mortality. RESULTS:The high maternal risk sample created with the OCI had a small but positive net benefit (+ 0.6), while the conventionally derived sample had a negative net benefit indicating the sample selection performed worse than identifying no woman as high maternal risk. CONCLUSIONS:The OCI can be used to select women at high maternal risk in administrative data. The OCI provides a consistent method of identification for women at risk of maternal morbidity and mortality and avoids confounding all obstetric risk factors with specific maternal risk factors. Using the OCI may help reduce misclassification as high maternal risk and improve the consistency in identifying women at high maternal risk in administrative data.