Measuring maternal morbidity in routinely collected health data: development and validation of a maternal morbidity outcome indicator.

BACKGROUND: As maternal deaths become rare in many countries, severe maternal morbidity has been suggested as a better indicator of quality of care.
OBJECTIVE: To develop and validate an indicator for measuring major maternal morbidity in routinely collected population health datasets (PHDS).
METHODS: First, diagnoses and procedures that might indicate major maternal morbidity were compiled and used to sample possible cases in PHDS; second, a validation study of indicated cases was undertaken by review of birth admission medical records using a nested case-control study approach with 400 possible cases and 800 controls; finally "true" morbidity from the validation study was used to define a maternal morbidity outcome indicator (MMOI) with a high positive predictive value (PPV). Sensitivity, specificity, PPV, negative predictive value (NPV), and exact 95% confidence intervals (95% CI) were weighted by the sampling probabilities.
RESULTS: There were 1184 records available for review. Of 393 possible cases only 188 were confirmed as suffering major morbidity (weighted PPV 47.3%, sensitivity 72.9%) and of the 791 initial noncases, 787 were confirmed as noncases (weighted NPV 99.5%, specificity 98.5%). Revision of the initial indicator with exclusion of noncontributing International Classification of Disease (ICD) codes provided a MMOI with population-weighted rate of 1.5%, PPV 94.6% (95% CI: 72.3-99.9), sensitivity 78.4% (95% CI: 55.2-93.1), specificity 99.9% (95% CI: 99.5-99.9), and 99.5% agreement with "true" morbidity (kappa 0.86).
CONCLUSIONS: PHDS can be used reliably to identify women who suffer a major adverse outcome during the birth admission and have potential for monitoring the quality of obstetric care in a uniform and cost-effective way.