Purpose: Obstetrical quality is measured at the national level rather than the local level.  Our study sought to compare the predicative ability of primary cesarean delivery models at the national, regional, and state levels.

Methods: National birth certificate data for 2003 were used to build a standard national risk-adjustment model for primary cesarean delivery. Variables in the model included: maternal age, gestational age, multiple births, nulliparity, complications of pregnancy, maternal medical conditions, and the trimester in which prenatal care began. The same variables were then used to run models and generate beta weights at the regional and state levels.  All models were boot strapped and c statistics and confidence intervals were determined.

Results: There were 3,475,663 births in the data set after cleaning. C statistic were similar for all models and 95% confidence intervals show significant overlap at the national, regional, and state levels.

Figure 1. C statistics and 95% confidence intervals for primary cesarean risk-adjustment models for the US, regional and selected states.

Conclusions: Within the United States, region specific models do not outperform a national model for risk-adjusting primary cesarean delivery rates. These data appear to suggest that regional models are not necessary and that a national risk-adjustment model can safely be used by quality assessment organizations.