Severe Community-Acquired Pneumonia: Impact of HIV on Clinical Presentation, Microbiological and Laboratory Findings, and Outcome

Severe community-acquired pneumonia (SCAP) is associated with significant morbidity and mortality, but there is a paucity of data regarding these infections in sub-Saharan Africa, especially among people living with HIV (PLWH). This study investigated the impact of HIV on clinical presentation, microbial aetiology, laboratory findings, and outcome of SCAP. This was additional analysis of data from a large, single-centre, retrospective, observational study conducted among consecutive adult patients (≥18 years) admitted to the multidisciplinary ICU at the Charlotte Maxeke Johannesburg Academic Hospital, between 1 July 2007 and 31 May 2019, with SCAP. The current study describes 718 PLWH and 131 HIV-negative cases extracted from the initial cohort. The median age was 37 [IQR 30-46] years with PLWH significantly younger than their HIV-negative counterparts (36 years [IQR 29-44] years vs 52 years [IQR 34-65] years; P < .001). PLWH were more commonly female (P = .053), while more of the HIV-negative patients were male. The median CD 4 count of the PLWH was 42 [IQR 14-108] cells/mm 3 and only 15.5% were on anti-retroviral therapy (ART) prior to hospitalisation. Differences were noted in clinical, laboratory and radiological features between the groups. Overall, Mycobacterium tuberculosis was the most common microbial aetiology in both groups, followed by Streptococcus pneumoniae, which was associated with a significantly lower mortality, whereas mortality with Pneumocystis jirovecii infection, which occurred only in PLWH, was high. Overall ICU mortality was high (48.9%), and while HIV was an independent risk factor for mortality (OR 0.58, 95% CI 0.37-0.92; p = .02) on univariate analysis, this finding was not true when HIV considered within the multivariable analysis. This study describes one of the largest cohorts of PLWH with SCAP and compares their findings with HIV-negative cases. HIV was not a significant predictor of mortality when considered in the context of other covariables on multivariable analysis.