Background: Pneumocystis pneumonia (PCP) is an important cause of acute respiratory failure (ARF) in immunocompromised patients, yet no actual clinical tool suitably identifies patients at risk. Recently, a multivariable prediction model has been proposed for haematology patients with ARF requiring intensive care unit (ICU) admission to assess the risk of PCP (PCP score). However, it has not yet been validated externally.
Methods: To validate the PCP score, a retrospective cohort study was conducted in two large designated haematology centres in Korea. One-hundred and forty haematology patients with ARF were admitted to ICU. They underwent aetiologic evaluations between July 2016 and June 2019. The predictive ability of the score was assessed with the receiver operating characteristic (ROC) curve analysis for both the discrimination and calibration of the score.
Results: Among the 141 patients, 13 (9.2%) were finally diagnosed of PCP. Although the median of PCP score in PCP group was higher than in non-PCP group (3.0 [interquartile range 0.0-4.0] vs. 2.0 [0.5-4.0]), the difference was not statistically significant (P = 0.679). The area under the ROC curve of the PCP score in our cohort was 0.535 (95% CI, 0.449-0.620), indicating no discriminatory ability. When using a cut-off of 3.0 the score, the result was 38.5% (95% CI, 13.9-68.4) sensitive and 7.03% (95% CI, 61.6-78.1) specific. The negative predictive value was 58.8% and positive predictive value was 59.8% for a 10% prevalence of PCP.
Conclusions: In this study, the PCP score was not useful to predict the risk of PCP in haematology patients with ARF. Further prospective validation studies are needed to validate the score's use in routine clinical practice for the early diagnosis of PCP in haematology patients.
Keywords: Immunocompromised; Invasive fungal infection; Pneumocystis pneumonia; Respiratory failure.