356. Risk Assessment of Pneumocystis jirovecii Pneumonia among Hospitalized Patients with Hypoxic Respiratory Failure - A Proposed Multivariable Calculator Based on Previous Prednisone Equivalent Dose

Abstract

Abstract Background Corticosteroids increase the risk of Pneumocystis jirovecii pneumonia (PJP). It is unknown how much corticosteroid dose exposure would modify the risk of PJP in different populations. We aim to develop a PJP risk calculator based on the previous dose of corticosteroids and modulated by additional clinical factors. Methods A multicenter retrospective case-control study was performed on patients tested for PJP within the UCHealth system from 2000 to 2021. We developed a model for estimating PJP risk based on previous prednisone equivalent daily dose (PEDD) and adjustable for additional clinical variables. PJP was fit to a generalized additive model (GAM), with a spline for prednisone dose and additive covariates for demographics and risk factors. We used a multicenter federated network to calibrate the model to estimate the PJP prevalence among hospitalized patients with hypoxic respiratory failure. Results We had a complete sample of 199 patients, 104 cases with PJP, and 95 controls. Patients with HIV (OR: 19 CI: 6.3-60.8, p< 0.0001), diabetes (OR: 4.1 CI: 1.2-14.8, p< 0.0288), and autoimmune disease (OR: 5.2 CI: 1.4-19.2, p< 0.0139) were more likely to have PJP. Patients with preexisting lung disease (OR: 0.3 CI: 0.1-0.6, p< 0.0041) and on PJP prophylaxis (OR: 0.06 CI: 0.02-0.2, p< 0.0001) were less likely to have PJP. 36.8% of controls and 49% of cases were on steroids with a mean PEDD of 15 mg/day and 20.4 mg/day, respectively. We found a prevalence of PJP of 0.126% among hospitalized patients with hypoxic respiratory failure. The developed model can estimate the PJP risk based on a previous PEDD in 32 different clinical combinations: e.g., a PEDD of 20 mg/day would give a calculated annual PJP risk of approximately 1.74% (95% CI: [0.39, 7.42]) if a patient has autoimmune disease only but 6.29% (95% CI: [1.34, 24.91]) if a patient has HIV only (Figure 1). Figure 1.Predicted probability of PJP based on previous prednisone dose among hospitalized patients with hypoxic respiratory failure for three different clinical scenarios Conclusion Previous corticosteroid dose alone is inadequate to inform of an increased risk of PJP. A multivariable calculator incorporating the absence or presence of additional traditional risk factors could optimally stratify the PJP risk. Future directions include validating the findings in external cohorts and modeling PJP risk in the ambulatory setting to inform the need for PJP prophylaxis. Disclosures All Authors: No reported disclosures.