Analysis of Nonrandom Distributions of Cesarean Sections as a Quality Improvement Indicator

Managed and nonmanaged care organizations are being evaluated on the basis of HEDIS and other quality indicators used in report cards. Traditionally, two such indicators have been rates of Cesarean sections and rates of vaginal birth after Cesarean, but these Cesarean rate indicators do not use all information available regarding utilization patterns. From a strictly clinical perspective, most Cesarean sections should be randomly distributed by weekday. We suggest a practical statistical test, using standard Poisson and chi-square techniques, to analyze the nonrandom occurrence of two, three, or more Cesareans in a day. We studied a sample of 1,500 Cesarean sections of all types performed by 19 physicians in seven California hospitals during 1995. Poisson and chi-square tests can act as an indicator to reveal nonrandom frequencies of Cesarean sections, suggesting possible nonclinical reasons for these events. We discuss how this methodology can help to focus efforts to improve quality of care.