Detecting adverse events in Thai hospitals using medical record reviews: agreement among reviewers.

BACKGROUND Attention to the problem of adverse events increases worldwide. The present study aimed to evaluate inter-rater reliability of medical record reviews of adverse events in the Thai context. MATERIAL AND METHOD A total of 279 inpatient records were reviewed by 23 clinical auditors. Each record was examined independently by 3 auditors using a standardised review form. Agreements on the occurrence of AEs among auditors were assessed using Kappa statistic (kappa). RESULTS Agreement of the auditors in detecting at least 1 medical condition potentially related to AEs was moderate (kappa = 0.54, 95% confidence interval (CI) 0.42-0.65). Agreement on each item of the listed conditions ranged from slight to substantial (kappa = 0.08 to 0.79). There was a positive trend of the correlation between the rate of the conditions and the kappa statistic (Spearman rank correlation = 0.65, p-value = 0.058). Agreement on determining AEs occurrence was fair (weighted kappa = 0.34, CI = 0.22-0.45). Agreement on determining AEs preventability was fair (weighted kappa = 0.27, CI = 0.16-0.39). CONCLUSION The reliability of medical record review to detect AEs is influenced by the prevalence rate of AEs, as well as the variability among reviewers. The use of chart review to detect AE is reasonable if the rate of the event is sufficiently high.