2203. Patient-Specific Risk Stratification to Identify Patients at High and Low Risk for P. aeruginosa in Community-Acquired Pneumonia

BACKGROUND: Pseudomonas aeruginosa (PsA) is an infrequent pathogen associated with poor outcomes in community-acquired pneumonia (CAP). Identifying patients at high and low-risk for PsA in CAP is necessary to reduce inappropriate and overly broad-spectrum antibiotic use. We evaluated the distribution of risk-factors in hospitalized CAP patients with and without PsA infection. METHODS: Design: retrospective, single-center, case–control study. Inclusion: hospitalized CAP patients admitted to the general medicine wards between January 1, 2014 and May 29, 2018. Exclusion: cystic fibrosis, ≥ 3 admissions within 30 days, CAP requiring ICU admission, and death within 48 hours of admission. Case patients had PsA in respiratory or blood cultures during the index CAP admission. Controls were randomly selected targeting a 3:1 ratio. Comorbidities, pneumonia severity index, and m-APACHE II were assessed. Gram-negative risk factors defined by Shindo et al. 2013 (PMID: 23855620) and validated by Kobayashi et al. (2018; PMID: 30349327) were scored for each patient. Stepwise logistic regression was used to identify covariates that distinguished cases from controls at a P < 0.2; these were then used to generate propensity weights (i.e., inverse-probability conditioned on covariates). Unadjusted and adjusted odds ratios for case status were estimated using logistic regression according to: the total number of risk factors present and threshold values, respectively. All analyses were conducted using IC Stata (v.14.2). RESULTS: 54 cases and 152 controls were included. The distribution of the patient-specific sum of risk factors for PsA is shown in Figure 1. The univariate OR for case status was 4.29 (95% CI:1.55–11.9) at n = 3 risk factors, which was similar after propensity weight adjustment [aOR = 4.64 (95% CI: 1.32–16.3)]. The univariate OR of case status was 2.98 among patients with ≥ 3 risk factors (95% CI: 1.34–6.62), which was similar after propensity weight adjustment [aOR = 2.8 (95% CI: 1.02–7.72)], and correct classification was 73.8%. CONCLUSION: At a threshold of ≥ 3 PsA risk factors, cases and controls were well classified, even after adjusting for propensity weights. The impact of patient-specific PsA risk-stratification on CAP outcomes and appropriate antibiotic use should be evaluated. [Image: see text] DISCLOSURES: All authors: No reported disclosures.