Association between hospital-level cesarean delivery rates and severe maternal morbidity and unexpected newborn complications

Abstract

Although there are many indications for a cesarean delivery, the "optimal" cesarean delivery rate is unknown. Neonatal and maternal morbidity have largely not been considered in the generation of hospital-level cesarean delivery rate targets. We sought to examine if the widely adopted and reported markers of maternal and neonatal morbidity were associated with hospital cesarean delivery rates to provide context for potential comparison and consideration for defining cesarean delivery rate targets. We hypothesized that hospitals with higher cesarean delivery rates would have increased rates of severe maternal morbidity, though we were less certain of the associations of the cesarean delivery rates with unexpected newborn complications. This is a cross-sectional, ecological study using data from the 2016 Nationwide Readmission Database of hospitals with at least 100 deliveries per year. The exposure of interest was hospital cesarean delivery rate. The outcomes were (1) severe maternal morbidity with and without transfusion-in accordance with the Centers for Disease Control and Prevention's definition, and (2) neonatal morbidity-defined using The Joint Commission's Perinatal Quality metric of moderate and severe unexpected newborn complications among term, singleton, and nonanomalous neonates. Before assuming a single linear relationship to model the associations between morbidity and cesarean delivery rates, the Joinpoint Regression Analysis program was used to examine for potential splines in the relationships with both severe maternal morbidity (with and without transfusion) and severe and moderate unexpected newborn complications. Poisson regression model was then used to determine the association between morbidity and cesarean delivery rates. The analysis included 831,111 deliveries from 621 hospitals. The mean cesarean delivery rate was 30.5%. The median severe maternal morbidity rate was 1.40 per 100 deliveries (interquartile range, 0.71-2.21 per 1000 deliveries). Excluding transfusion, the median severe maternal morbidity rate was 0.47 per 100 deliveries (interquartile range, 0.22-0.73 per 100 deliveries). The median rate of severe and moderate unexpected newborn complications was 1.01 per 100 low-risk newborns (interquartile range, 0.64-1.69 per 100 low-risk newborns) and 1.79 per 1000 low-risk newborns (interquartile range, 0.94-2.93 per 100 low-risk newborns), respectively. In the unadjusted analysis, every percentage point increase in a hospital's cesarean delivery rate was associated with a 3.4% (95% confidence interval, 2.3%-4.4%) and a 2.3% (95% confidence interval, 1.0%-3.5%) increase in severe maternal morbidity including and excluding transfusion, respectively. After adjustment for the case mix and hospital factors, only the relationship with severe maternal morbidity including transfusion remained significant: 3.3% (95% confidence interval, 1.7%-4.9%) increase in severe maternal morbidity per 1 percentage point increase in the cesarean delivery rate. There was no observed association between cesarean delivery rates and unexpected newborn complications CONCLUSION: Severe maternal morbidity and unexpected newborn complications occur in fewer than 5 in 100 births. Findings from this analysis of hospitals with cesarean delivery rates ranging from 6.8%-56.3% suggest that those with lower cesarean delivery rates have lower severe maternal morbidity (which includes transfusion) and similar unexpected newborn complications compared with hospitals with higher cesarean delivery rates. This work may provide a helpful context to providers, hospitals, and policymakers who are measuring and reporting outcomes. Regarding neonatal morbidity in particular, the Joint Commission manual notes that the unexpected newborn complication metric was specifically designed to be compared against maternal-focused metrics such as cesarean delivery rates. More work is needed to define and identify appropriate measures of maternal and neonatal morbidity for these types of comparisons.