A prospective cluster trial to increase antibiotic prescription quality in seven non-ICU wards.

Aim: To evaluate general shortcomings and faculty-specific pitfalls as well as to improve antibiotic prescription quality (ABQ) in non-ICU wards, we performed a prospective cluster trial.

Methods: An infectious-disease (ID) consulting service performed a prospective investigation consisting of three 12-week phases with point prevalence evaluation conducted once per week (=36 evaluations in total) at seven non-ICU wards, followed by assessment of sustainability (weeks 37-48). Baseline evaluation (phase 1) defined multifaceted interventions by identifying the main shortcomings. Then, to distinguish intervention from time effects, the interventions were performed in four wards, and the 3 remaining wards served as controls; after assessing effects (phase 2), the same interventions were performed in the remaining wards to test the generalizability of the interventions (phase 3). The prolonged responses after all interventions were then analyzed in phase 4. ABQ was evaluated by at least two ID specialists who assessed the indication for therapy, the adherence to the hospital guidelines for empirical therapy, and the overall antibiotic prescription quality.

Results: In phase 1, 406 of 659 (62%) patients cases were adequately treated with antibiotics; the main reason for inappropriate prescription was the lack of an indication (107/253; 42%). The antibiotic prescription quality (ABQ) significantly increased, reaching 86% in all wards after the focused interventions (502/584; nDf=3, ddf=1,697, F=6.9, p=0.0001). In phase 2 the effect was only seen in wards that already participated in interventions (248/347; 71%). No improvement was seen in wards that received interventions only after phase 2 (189/295; 64%). A given indication significantly increased from about 80% to more than 90% (p<.0001). No carryover effects were observed.

Discussion: ABQ can be improved significantly by intervention bundles with apparent sustainable effects.