• Predictive value of C-reactive protein for tuberculosis, bloodstream infection or death among HIV-infected individuals with chronic, non-specific symptoms and negative sputum smear microscopy.

      Bedell, RA; van Lettow, M; Meaney, C; Corbett, EL; Chan, AK; Heyderman, RS; Anderson, ST; Akesson, A; Kumwenda, M; Zachariah, R; et al. (Wiley-Blackwell, 2018-01-01)
      BACKGROUND: C-reactive protein (CRP) is an inflammatory biomarker that may identify patients at risk of infections or death. Mortality among HIV-infected persons commencing antiretroviral therapy (ART) is often attributed to tuberculosis (TB) or bloodstream infections (BSI). METHODS: In two district hospitals in southern Malawi, we recruited HIV-infected adults with one or more unexplained symptoms present for at least one month (weight loss, fever or diarrhoea) and negative expectorated sputum microscopy for TB. CRP determination for 452 of 469 (96%) participants at study enrolment was analysed for associations with TB, BSI or death to 120 days post-enrolment. RESULTS: Baseline CRP was significantly elevated among patients with confirmed or probable TB (52), BSI (50) or death (60) compared to those with no identified infection who survived at least 120 days (269). A CRP value of >10 mg/L was associated with confirmed or probable TB (adjusted odds ratio 5.7; 95% CI 2.6, 14.3; 87% sensitivity) or death by 30 days (adjusted odds ratio 9.2; 95% CI 2.2, 55.1; 88% sensitivity). CRP was independently associated with TB, BSI or death, but the prediction of these endpoints was enhanced by including haemoglobin (all outcomes), CD4 count (BSI, death) and whether ART was started (death) in logistic regression models. CONCLUSION: High CRP at the time of ART initiation is associated with TB, BSI and early mortality and so has potential utility for stratifying patients for intensified clinical and laboratory investigation and follow-up. They may also be considered for empirical treatment of opportunistic infections including TB.