Elective primary caesarean delivery: accuracy of administrative data

Publisher Website
Google Scholar
Abstract
The caesarean delivery rate has become a commonly used measure intended to reflect the quality of obstetric care. At least 25% of all primary caesarean deliveries occur electively, i.e. to women who are not in labour. This study is intended to validate a previously published model designed to use ICD‐9‐CM codes to identify and categorise cases of elective primary caesarean delivery by their indication. ICD‐9‐CM codes were compared with diagnoses written in the medical record for all women without a prior caesarean who delivered in the same month in a single hospital to examine the accuracy of the codes for 12 potential elective primary caesarean indications derived by the published model: malpresentation; bleeding; genital herpes; severe hypertension; uterine scar; multiple gestation; macrosomia; unengaged fetus; maternal soft tissue conditions; other hypertensive conditions; prematurity; and chromosomal anomalies. Of 440 eligible women, a total of 26 (5.9%) had an elective primary caesarean by medical record review vs. [27] (6.1%) by administrative data. Using medical record data as the gold standard, the sensitivity, specificity, and accuracy of administrative data for the identification of elective primary caesarean delivery were 73.1%, 98.1%, and 96.6%, respectively. Administrative coding for all of the 12 conditions was highly specific, although wide variability existed in its sensitivity; its accuracy ranged between 83.9% and 100%. These results suggest that, despite widespread use of caesarean delivery rates obtained through administrative data, more experience is needed to determine which obstetric codes may be sufficiently specific, sensitive, or prevalent to serve a monitoring or surveillance function reflecting the quality of obstetrical care. The results support continued efforts to use administrative data to monitor elective primary caesarean delivery.