#90-S sensitivity and specificity of symptomology to screen for eligibility for anti-retroviral therapy among pregnant women

Abstract

PURPOSE: Maternal-To-Child-Transmission Plus programs have proposed to identify ART eligible women through antenatal care. Screening using AIDS-related symptoms is feasible in resource-poor settings. However among pregnant women attending antenatal care clinics, it is not known how well symptomology is correlated to CD4+ count and HIV viral load (VL) which are traditionally used (Table 1) Table 1 Cut-off ⩾ 1 symptom Cut-off ⩾ 3 symptoms CD4+ (cps/ml) VL (cps/ml) CD4+ (cps/ml) VL (cps/ml) Sens (%) 100.0 97.9 25.0 28.6 Spec (%) 11.0 10.1 80.0 75.3 . METHODS: Using a population of HIV+ pregnant women from Rakai District, Uganda (n = 325), we calculated the sensitivity, specificity, PPV, and NPV of identifying women for treatment using symptomology versus using the DHHS VL (>55,000 cps/ml) criterion for initiation of treatment. For a sub-set of the pregnant women (n = 86) we also analyzed the correlation between symptoms and CD4+ count (< 350 cells/ml). Symptoms were self-reported and selected from 1993 WHO staging criteria. RESULTS: There was no association between WHO major or minor criteria and VL or CD4+ cell count. Total number of symptoms was used to calculate the validity. Using the cutoffs of ⩾1 and ⩾3: It is clear that pregnant women with zero symptoms have low VLs and high CD4+ count. Treatment based solely on number of symptoms will over treat because the majority of women with symptoms would not have been eligible for treatment according to DHHS guidelines. CONCLUSION: CD4+ or VL measurements are needed to identify HIV-positive pregnant women eligible for ART. Low-cost CD4+ testing as a screen for treatment eligibility is urgently needed.