Methods

This study was a single-center, quasi-experimental study in an urban, 732-bed, teaching hospital. The hospital has a long-standing AS program, staffed by a full-time AS pharmacist and supported by infectious diseases physicians when needed. After MALDI-TOF MS was introduced, no new AS resources or programs were added. We included adult inpatients with the following criteria: a diagnosis of pneumonia in their electronic medical record; a chest x-ray with infiltrates; a positive respiratory culture with pathogenic bacteria; and who received antibiotics. Data were collected retrospectively, and patients were analyzed in 2 cohorts: August 2016–February 2017 (before MALDI-TOF MS) and August 2017–February 2018 (after MALDI-TOF MS). We applied the following exclusion criteria: diagnosis of concomitant bacterial or fungal infections; receipt of <48 hours of effective therapy; expiration prior to receiving 48 hours of effective antimicrobial therapy; initiation of palliative care during the admission; transfer to another facility; and/or discharge against medical advice within 48 hours of a positive respiratory culture.

During the period before MALDI-TOF MS implementation, organism identification was performed primarily on the BD Phoenix instrument (Becton Dickinson, Sparks, MD). Starting February 9, 2017, the first-line method of organism identification was switched to the Bruker Biotyper CA system (Bruker Daltonics, Billerica, MA) an FDA-cleared MALDI-TOF MS system. Antimicrobial susceptibility testing was performed using the BD Phoenix Etest (bioMérieux, Durham, NC) or disk diffusion throughout and results were interpreted using the most current Clinical and Laboratory Standards Institute (CLSI) recommendations. ⁸

The primary outcome was time to optimal antimicrobial therapy, defined as the time interval between specimen collection and initiation of appropriate antimicrobial therapy administration which was, in turn, defined as the narrowest antibiotic for the isolated pathogen that is recommended as definitive therapy for pneumonia. Opportunities for intervention included escalation of therapy, de-escalation of therapy, or discontinuation of inappropriate or duplicative antimicrobial therapy. Secondary outcomes included clinical cure at 7 days, inpatient duration of antibiotic use for respiratory infections, infection-related length of stay (LOS), and overall LOS. Clinical cure was defined as the resolution of baseline signs and symptoms with white blood cell count <10,000 cells/mm³ and absence of oral temperature >38°C for 24 consecutive hours or discharge from the hospital prior to 7 days. Duration of antibiotic therapy was measured in hours and calculated as the time interval between the first dose of antibiotic and antibiotic discontinuation or discharge, whichever was sooner. Infection-related LOS was calculated for all patients as the time interval between the date of the culture collection and antibiotic discontinuation or discharge date, whichever was sooner.

A sample size of 175 subjects per group was required to detect a minimum difference in time to intervention of 12 hours with 80% power at a 5% level of significance. Descriptive statistics were used to summarize continuous and categorical data. The Student t test was used to analyze continuous variables; the χ² test was used for proportions; and the Mann-Whitney U test was used for non-parametric analyses.

A linear regression model was created using JMP Pro 13 statistical software to examine the relationship between time to optimization of antimicrobials (outcome variable) and MALDI-TOF MS implementation (key predictor variable). The final model included the following covariates: MALDI-TOF MS use, age, immunosuppression, hospital acquired infection, ventilated patient, gram-negative organism isolated, intensive care unit (ICU) admission, infectious diseases consultation for the infection, narrowing of antimicrobials, escalation of antimicrobials, and lack of opportunity to optimize antimicrobials.