Use of the relationship between absolute lymphocyte count and CD4 count to improve earlier consideration of pneumocystis pneumonia in HIV-positive emergency department patients with pneumonia

Background: The ability to accurately assess the level of immunosuppression in HIV+ patients in the emergency department (ED) is often limited and can affect management of these patients. Objective: To evaluate the relationship between the absolute lymphocyte count (ALC) and CD4 count in HIV patients admitted through the ED with pneumonia and how utilization of this relationship may affect early consideration and evaluation of Pneumocystis jiroveci pneumonia (PCP). Methods: Retrospective multicenter 5-year study of HIV+ patients with an ICD-9 diagnosis of pneumonia. Included patients had an ALC measured on ED presentation and a CD4 count measured in < 24 h. A receiver operator curve (ROC), decision plot analysis, and McNemar test of proportions were used to characterize the relationship between study variables. Results: Six hundred eighty six patients were enrolled, 23.2% (95% confidence interval [CI] 20.2-26.1) were diagnosed with PCP. The geometric mean CD4 count and ALC were 81 and 1089, respectively. The correlation between ALC and CD4 was r = 0.60 (95% CI 0.55-65, p < 0.01). The ROC was 0.78 (0.75-0.82). An ALC < 1700 cells/mm³ had a sensitivity of 84% (95% CI 80-87) and specificity of 55% (95% CI 48-70) for a CD4 < 200 cells/mm³. An ALC threshold of 1700 cells/mm³ would have identified 86% of patients with PCP but falsely identified 2.5 patients without PCP for every one accurately identified. Conclusion: The ALC threshold of 1700 cells/mm³ retains significant discriminatory value and would moderately improve identification of patients with a CD4 < 200 cells/mm³ but is not likely to be reliable as the sole method of early recognition and evaluation of PCP.