RESUMEN
Background: Mortality from tuberculosis (TB) sepsis is common among patients living with human immunodeficiency virus (PLHIV). We aimed to detect M. tuberculosis (MTB) and additional sepsis etiologies, and mortality determinants in PLHIV. Methods: This prospective cohort study consented and followed-up PLHIV for 28 days in northern Tanzania. From May through December 2021, patients provided urine and sputum for TB testing in lateral-flow lipoarabinomannan (LF-LAM) and Xpert® MTB/RIF. Bacterial blood culture, cryptococcal antigen, malaria rapid diagnostic, C-reactive-protein (CRP), and international normalized ratio (INR) tests were also performed. Sepsis severity was clinically measured by Karnofsky and modified early warning signs (MEWS) scores. Anti-TB, broad-spectrum antibiotics, and antimalarial and antifungal agents were prescribed in accordance with Tanzania treatment guideline. An independent t-test and Chi-square or Fisher's exact tests compared means and proportions, respectively. P < 0.05 was statistically significant. Results: Among 98 patients, 59 (60.2%) were female. Their mean (standard deviation) age was 44 (12.9) years. TB detection increased from 24 (24.5%) by Xpert® MTB/RIF to 36 (36.7%) when LF-LAM was added. In total, 23 (23.5%) patients had other than TB etiologies of sepsis, including Staphylococcus aureus, Streptococcus pneumoniae, Cryptococcus spp., and Plasmodium spp. Twenty-four (94.4%) of 36 patients with TB had higher CRP (≥10 mg/l) compared to 25 (40.3%) non-TB patients (P < 0.001). Nine (9.2%) patients died and almost all had INR ≥1.8 (n = 8), Karnofsky score <50% (n = 9), MEWS score >6 (n = 8), and malnutrition (n = 9). Conclusions: MTB and other microbes contributed to sepsis in PLHIV. Adding non-TB tests informed clinical decisions. Mortality was predicted by conventional sepsis and severity scoring, malnutrition, and elevated INR.