Abstract Body

Background: The World Health Organization (WHO) recommends tuberculosis (TB) intensified case finding (ICF) among people living with HIV (PLHIV) using an evidence-based clinical screening algorithm including current cough, fever, night sweats, and weight loss. However, limited data exist on the performance of this algorithm in routine program settings. We assessed the performance of the WHO algorithm alone and with chest radiography among PLHIV in Nyanza Province, Kenya.

Methodology: We recruited PLHIV aged 7 years and older newly enrolling in HIV care at 15 randomly-selected HIV clinics. Patients were screened with the WHO algorithm, underwent chest radiography, and provided two or more sputum samples for smear microscopy, liquid culture, and Xpert MTB/Rif (Xpert). TB was defined as any specimen positive for TB by Xpert or culture. We calculated the performance characteristics of clinical screening compared to laboratory diagnostic testing results.

Results: Overall, 779 PLHIV were enrolled in the study and 718 (92.3%) met criteria for inclusion in the analysis. Median age was 29 years (interquartile range: 24-39 years); 4% were younger than 15 years. Eighty-one patients (11.3%) had TB. Preliminary analysis shows that the sensitivity, specificity, and negative predictive value (NPV) of the WHO algorithm were 73.7%%, 50.3%, and 93.8%, respectively (Table 1). Testing asymptomatic persons with chest radiography increased sensitivity and NPV of the screening algorithm to 90.6% and 96.1%, respectively, but decreased specificity to 32.5%.

Conclusions: By limiting TB diagnostic testing to symptomatic patients, the WHO algorithm missed more than 25% of TB patients. Adding a chest radiograph improved case finding and more reliably identified PLHIV who did not have TB and who should receive isoniazid preventive therapy. Additional analyses including screening performance during follow-up visits are needed.