The WHO recommends pre-antiretroviral treatment (ART) CD4-targeted cryptococcal antigen (CRAG) screening in sub-Saharan Africa. Implementing this strategy only in outpatient settings may underestimate the true CRAG prevalence and decrease its impact.
In October 2013, lab-reflex CRAG screening was implemented at the St. Francis Referral Hospital, Ifakara, Tanzania for all HIV+ hospitalized patients and outpatients with CD4 ≤150/μL. The impact on CRAG detection and outcome was assessed. Cox regression identified predictors of death/loss to follow-up (LFU) at 6 months.
Of 1976 persons registered from 10/2013 to 07/2015, 500 (25%) ART-naive had CD4 ≤150/μL and were CRAG screened, contributing 2965 persons-month follow-up. Median age was 39 years (IQR 33-46), median CD4 count was 58 cells/μL (IQR 23-100), and 12% (59/500) had tuberculosis. CRAG prevalence was 6.4% (32/500) and 7.7% (30/376) with CD4 counts ≤150 and ≤100 cells/μL respectively, 1.7-fold higher than the 2008-2012 outpatient prevalence in the same cohort (3.7% ≤150cells/μL, p=0.021). Inpatients (n=82) had a CRAG prevalence of 12% vs. 5.3% in outpatients (p=0.02), and accounted for 31% of all CRAG+. Median time from HIV to CRAG testing was 1 day (IQR 0-6). A lumbar puncture was done on the same day of CRAG testing in 97% (31/32) CRAG+, and 39% (12/31) had cryptococcal meningitis (CM), 17% of whom (2/12) without neurologic symptoms. Fluconazole tailored for CM presence was started in 81% (26/32) of CRAG+ and ART in 72% CRAG+ (23/32) and 76% (382/500) overall. Known 6-month mortality for those recruited before 02/2015 (n=361) did not differ between CRAG-negative and CRAG+ without CM (9% vs.7%, p=0.9), yet was 86% (6/7) among CM patients (p<0.001). LFU was 31% (104/340), 29% (4/14), and 14% (1/7) respectively. Independent predictors of death/LFU at 6 months were CRAG+ (adjusted hazard ratio (aHR) 3.2, 95% CI 1.2-8.2), CM (aHR 5.5, 95% CI 1.7-18), no ART initiation (aHR 2.2, 95% CI 1.5-3.4), tuberculosis (aHR 1.8, 95% CI 1.03-3.2), and hemoglobin (aHR 1.2 per 1 g/dL decrease, 95% CI 1.1-1.3) (Fig.1).
Implementation of lab-reflex CRAG screening resulted in an increased and rapid detection of CRAG and CM. Mortality was highest for CM patients but did not vary between CRAG+ without CM treated with pre-emptive fluconazole and CRAG-negative patients. These results support the urgent adoption of the WHO guidelines for CRAG screening in Africa and its implementation both in inpatient and outpatient settings.