Factors predictive of inappropriateness in requests for parenteral antimicrobials for therapeutic purposes: A study in a small teaching hospital in Brazil

Abstract

Background: The identification of patterns of inappropriate antimicrobial prescriptions in hospitals contributes to the improvement of antimicrobial stewardship programs (ASP). Methods: We conducted a cross-sectional study to identify predictors of inappropriateness in requests for parenteral antimicrobials (RPAs) in a teaching hospital with 285 beds. We reviewed 25% of RPAs for therapeutic purposes from y 2005. Appropriateness was evaluated according to current guidelines for antimicrobial therapy. We assessed predictors of inappropriateness through univariate and multivariate models. RPAs classified as 'appropriate' or 'probably appropriate' were selected as controls. Case groups comprised inappropriate RPAs, either in general or for specific errors. Results: Nine hundred and sixty-three RPAs were evaluated, 34.6% of which were considered inappropriate. In the multivariate analysis, general predictors of inappropriateness were: prescription on week-ends/holidays (odds ratio (OR) 1.67, 95% confidence interval (CI) 1.20-2.28, p = 0.002), patient in the intensive care unit (OR 1.57, 95% CI 1.11-2.23, p = 0.01), peritoneal infection (OR 2.15, 95% CI 1.27-3.65, p = 0.004), urinary tract infection (OR 1.89, 95% CI 1.25 -2.87, p = 0.01), combination therapy with 2 or more antimicrobials (OR 1.72, 95% CI 1.15-2.57, p = 0.008) and prescriptions including penicillins (OR 2.12, 95% CI 1.39-3.25, p = 0.001) or 1(st) generation cephalosporins (OR 1.74, 95% CI 1.01-3.00, p = 0.048). Previous consultation with an infectious diseases (ID) specialist had a protective effect against inappropriate prescription (OR 0.34, 95% CI 0.24-0.50, p < 0.001). Factors independently associated with specific prescription errors varied. However, consultation with an ID specialist was protective against both unnecessary antimicrobial use (OR 0.04, 95% CI 0.01-0.26, p = 0.001) and requests for agents with an insufficient antimicrobial spectrum (OR 0.14, 95% CI 0.03-0.30, p = 0.01). Conclusions: Our results demonstrate the importance of previous consultation with an ID specialist in assuring the quality of prescriptions. Also, they highlight prescription patterns that should be approached by ASP policies.