140: Is risk adjusted cesarean delivery rate in primparous women a good measure of quality?


      Since reduction of unnecessary cesarean deliveries (CD) is a public health goal, groups like Health Grades are publishing risk adjusted cesarean delivery (RACD) rates as a measure of obstetric (OB) quality of care. Yet, evidence supporting this is limited. We sought to validate RACD rate as a measure of OB quality through its association with other quality measures.

      Study Design

      Linked birth certificate and hospital admission records for mother and infant were collected on all deliveries in Missouri from 1993-2003. An extensive risk adjustment model was created using maternal and fetal co-morbidities identified by ICD-9 codes. CD for previa, herpes, malpresentation & cord prolapse were excluded. Expected RACD rates in primiparous women were determined and a c-statistic was calculated. After determining hospital-level differences between observed and expected RACD rates, we performed a correlation analysis with maternal outcomes (infection, hemorrhage, prolonged length of stay (LOS)), neonatal outcomes (death rate, asphyxia, prolonged LOS), and patient safety indicators from AHRQ (birth trauma, injury with instrumented vaginal delivery (VD), non-instrumented VD, or CD).


      236,902 patients were evaluated from 108 hospitals. The c statistic for our RACD model was 0.64. RACD rate did not correlate with any maternal or neonatal outcomes. The correlation coefficient (CC) for RACD rate and maternal infection, hemorrhage and LOS were −0.23, −0.16, −0.06. The CC for RACD rate and neonatal death, asphyxia, and LOS were 0.19, −0.01, −0.03. The correlation between RACD and the 4 patient safety indicators were similarly poor (-.20, 0.12, −0.16, −0.19).


      Despite risk adjustment, the poor c-statistic of our model suggests that RACD may be a quality measure. However, variations in RACD rates are not associated with important maternal or neonatal outcomes. Measuring OB quality is critical but cannot be achieved without validated measures. RACD should not be used and validated measures of OB quality need to be developed.