756. Use Risk Factors to Early Screen Pneumocystis Pneumonia in Hospitalized Patients

Background The diagnosis of Pneumocystis pneumonia (PCP) may be delayed or missed if the underlying risk factors are not recognized, particularly in HIV-uninfected patients. Methods The adult hospitalized patients (≥18 years) with pneumonia were selected from the Nationwide Inpatient Sample database (2005-2014). The in-hospital outcomes of PCP patients including mortality, length of hospital stay (LOS) and charge were analyzed. The risk factors for PCP were evaluated by Logistic regression. A risk-adjusted model to screen PCP in HIV-uninfected patients was developed by discriminant analysis. Results 24,025,696 hospitalized patients with pneumonia were identified, including 135,749 PCP patients. The incidences of PCP in pneumonia patients were 0.12% in HIV-uninfected group and 30.5% in HIV-infected group respectively. Comparing with other pneumonia patients, those with PCP had higher mortality (11.8% vs. 8.2%), longer LOS (median 8 vs. 5 days) and increased hospital charges (median $40,082 vs. $26,980). HIV infection was the major risk factor for PCP (OR=270.2, 95% CI 264.5-276) in all patients with pneumonia. In HIV-uninfected patients, the comorbidities including lymphoma (OR=10.7, 95% CI 10.2-11.2), CMV infection (OR=8.1, 95% CI 7.6-8.7), leukemia (OR=6.8, 95% CI 6.4-7.1), metastatic cancer (OR=5.3, 95% CI 4.6-6.0), immune thrombocytopenic purpura (OR=5.0, 95% CI 4.5-5.5), chronic steroid use (OR=4.1, 95% CI 3.9-4.3), solid organ transplant (OR=3.5, 95% CI 3.3-3.8), inflammatory bowel disease (OR=2.6, 95% CI 2.4-2.8), connective tissue disease (OR=2.4, 95% CI 2.3-2.6) and non-metastatic solid tumor (OR=2.3, 95% CI 2.1-2.4) were associated with increased risk for PCP. A risk-adjusted model composed of risk factors above could help to screen PCP with the sensitivity 42.9%, specificity 94.4% and accurate rate 94.3% (Table 1). Table 1 Conclusion PCP should be considered as one of the differential diagnoses in patients with pneumonia if they have underlying risk factors other than HIV infection. The risk-adjusted model can help to early screen PCP for HIV-uninfected patients before the pathogen test. Disclosures All Authors: No reported disclosures