A prediction score integrated into an electronic clinical decision support tool reduced initial empiric broad-spectrum antibiotic use without increasing inadequate empiric antibiotic therapy or mortality in patients presenting to the ED with pneumonia, a recent study found.
Researchers conducted a quasi-experimental, pre-post implementation study of the Drug Resistance in Pneumonia (DRIP) score, a previously validated prediction tool based on host risk factors. The decision support tool was programmed to calculate DRIP scores in real time through an automated query of the electronic health record for 10 criteria, which included risk factors such as antibiotic use in the previous 60 days and residence in a long-term care facility.
At four U.S. hospitals, the researchers compared effects of DRIP, which was integrated into the tool in 2015, with that of health care-associated pneumonia (HCAP) logic, used by the tool in 2012, for ED patients with pneumonia. The researchers estimated the score's effect on broad-spectrum antibiotic use within 12 hours of ED presentation (the primary outcome), mortality, hospital stay, and cost, adjusting for patient-level confounders. In addition, a recommendation was added to the tool advising clinicians to order a nasal swab for methicillin-resistant Staphylococcus aureus (MRSA) polymerase chain reaction testing in patients with high DRIP scores to aid in de-escalation of empiric vancomycin. Results were published online on May 8 by CHEST.
A total of 2,169 adult admissions were analyzed, 1,122 in 2012 and 1,047 in 2015. A drug-resistant pathogen was recovered in 3.2% of patients in 2012 and 2.8% in 2015. Broad-spectrum antibiotic use was about 10-fold higher than the incidence of drug-resistant pathogens. A broad-spectrum antibiotic was administered in 40.1% of admissions in 2012 compared to 33.0% in 2015 (P<0.001). Days of vancomycin therapy per 1,000 patient-days in 2012 were 287.3 compared to 238.8 in 2015 (P<0.001). Inadequate initial empiric antibiotics were prescribed in 1.1% of patients in 2012, compared to 0.5% in 2015 (P=0.12). Patients treated using the DRIP had a significant reduction in broad-spectrum antibiotic use (treatment effect odds ratio, 0.62 [95% CI, 0.39 to 0.98]; P=0.039). However, the average effects of the score on mortality, length of stay, and cost were not significant.
The study authors predicted that empiric broad-spectrum antibiotic use would be 18.9% after the stewardship intervention, which was lower than the actual results, suggesting that “the reduction would be greater had clinicians followed the tool recommendations explicitly,” they wrote. Confounding by indication was the primary limitation of the study, and further evaluation of the score in other EDs with varying patient demographics and resistance patterns is warranted, they noted.
“Compared to HCAP criteria, the DRIP score is a more effective tool to assist clinicians in accurately identifying the risk of drug-resistant pathogens in pneumonia,” the authors concluded. “Nevertheless, significant opportunities for improvement remain to reduce unnecessary antibiotic use in pneumonia.” They plan to reassess broad-spectrum antibiotic use three years after DRIP deployment, with the hypothesis that they will detect further reductions after clinicians become more familiar with the score and MRSA nasal swab strategies.