Abstract
Introduction Use of HCAP criteria for predicting drug-resistance in community-onset pneumonia results in overuse of broad-spectrum antibiotics. We derived and validated an alternative prediction tool (DRIP) (Webb AAC 2016) and integrated it into an electronic decision support tool used in 4 U.S. hospitals.
Aims Compare DRIP to HCAP for prediction of drug-resistance and evaluate antibiotic usage and outcomes in a prospective implementation study.
Methods Physician use of the electronic tool and calculation of DRIP was optional. For DRIP ≥4 anti-pseudomonal, vancomycin and azithromycin therapy was recommended. We identified two concurrent cohorts from 11/2014 to 10/2015: 1) cases where DRIP was calculated and 2) usual care. We compared observed rates of antibiotic use between groups and used logistic regression to severity adjust outcomes.
Results DRIP and usual care comprised 894 and 324 inpatients. Drug resistance incidence was 2.4% and 4%. Severity was higher for usual care. Compared to HCAP, DRIP demonstrated equivalent sensitivity but better specificity. Inadequate therapy was <1% in both groups. Relative reduction in unnecessary broad spectrum use (25.9% p=0.008) was observed in the DRIP group. DRIP was associated with decreased length of stay (LOS) (coeff – 0.147; upper 95% CI – 0.137; p < 0.001). Odds of in-hospital mortality did not reach statistical significance (OR 0.643; upper 95% CI 1.04; p = 0.063).
DRIP % | Usual Care % | p= | |
Inadequate Spectrum | 0.67 | 0.93 | NS |
Overtreatment | 20.6 | 27.8 | 0.008 |
Appropriate Spectrum | 78.6 | 71.0 | 0.005 |
Conclusion DRIP use was associated with reduced unnecessary broad-spectrum antibiotic use and LOS, without increased inadequate therapy or mortality.
- Copyright ©the authors 2016