Unexplained variation in hospital caesarean section rates

Objectives: To assess recent hospital caesarean section (CS) rates in New South Wales, adjusted for casemix; to quantify the amount of variation that can be explained by casemix differences; and to examine the potential impact on the overall CS rate of reducing variation in practice. Design and setting: Population-based record linkage study of births in 81 hospitals in New South Wales, 2009-2010, using the Robson classification to categorise births, and multilevel logistic regression to examine variation in hospital CS rates within Robson groups. Main outcome measures: Hospital CS rates. Results: The overall CS rate was 30.9%, ranging from 11.8% to 47.4% (interquartile range, 23.9%-33.1%) among hospitals. The three groups contributing most to the overall CS rate all comprised women with a single cephalic pregnancy who gave birth at term, including: those who had had a previous CS (36.4% of all CSs); nulliparous women with an elective delivery (prelabour CS or labour induction, 23.4%); and nulliparous women with spontaneous labour (11.1%). After adjustment for casemix, marked unexplained variation in hospital CS rates persisted for: nulliparous women at term; women who had had a previous CS; multifetal pregnancies; and preterm births. If variation in practice was reduced for these risk-based groups by achieving the best practice rate, this would lower the overall rate by an absolute reduction of 3.6%, from 30.9% to 27.3%. Conclusion: Understanding hospital heterogeneity in performing CS and implementing evidence-based practices may result in improved maternity care. We have identified five risk-based groups as priority targets for reducing practice variation in CS rates.