RESUMO
BACKGROUND: Retained surgical items (RSI) continue to occur. Large RSI studies are few due to low RSI frequency in single institutions and the medicolegal implications. Consequently, RSI risks are not fully defined, with discrepancies persisting among published studies. The goals of this study were to better define risk factors for RSI, to clarify previously discrepant risk factors, and to evaluate other potential contributors to RSI occurrence, such as trainee presence during an operation. STUDY DESIGN: Multicenter case-match study of RSI risk factors was conducted between January 2003 and December 2009. Cases complicated by RSI were identified at participating centers using clinical quality improvement and adverse event reporting data. Case match controls (non-RSI) were selected from same or similar-type cases performed at each respective institution. Retained surgical item risk factors were evaluated by univariate and multivariate conditional logistic regression. RESULTS: Fifty-nine RSIs and 118 matched controls were analyzed (RSI incidence 1 in 6,975 or 59 in 411,526). Retained surgical items occurred despite use of confirmatory x-rays (13 of 27 instances) and/or radiofrequency tagging (2 of 32 instances). Among previously discrepant results, we confirmed that body mass index, unexpected intraoperative events, and procedure duration were associated with increased RSI risk. The occurrence of any safety variance, and specifically an incorrect count at any time during the procedure, was associated with elevated RSI risk. Trainee presence was associated with 70% lower RSI risk compared with trainee absence. CONCLUSIONS: Longer duration of surgery, safety variances, and incorrect counts during the procedure result in elevated RSI risk. The possible positive influence of trainee presence on RSI risk deserves additional study. Our findings highlight the need for zero tolerance for safety omissions, continued study and development of novel approaches to RSI reduction, and establishing anonymous RSI reporting systems to better track both the incidence and risks associated with this problem, which has yet to be solved.