Data discrepancies and audit effects in the umbilical artery catheter trial

This paper examines the results of a quality control audit of a randomized trial of umbilical artery catheter placement in very low birth weight infants. A single auditor traveled to 12 neonatal intensive care units (NICUs) which participated in the trial to audit a random sample of 10% of the charts of the first 600 infants enrolled. Demographic, medical history, and clinical information were abstracted from the medical records. The greatest discrepancies between the audit results and the original data occurred in the recording of heparin intake. Selected results of the audit were mailed to the participating NICUs. This attempt to point out the discrepancies between the auditor's recording of heparin intake and the original heparin intake recorded by the NICU staff may have resulted in a significant increase in the recorded mean heparin intake of the last 300 infants enrolled in the study. The mean ± s.d. heparin intake among the preaudit cohort was 321 ± 235 units/kg/day compared to 366 ± 237 among the postaudit cohort (p = 0.01). These observations reinforce the suggestion that quality control audits may affect subsequent data collection and that the timing of the audit during the course of a trial appears critical to assuring the collection of high quality data.