Objective

To evaluate the performance of the modified Keith Edwards TB clinical diagnostic tool among HIV infected children.

Methods

Cross sectional study of 252 HIV infected children < 15 years old suspected with TB at four HIV/AIDS Care Clinics in Dodoma, Tanzania from November 2018 - March 2019. The modified Keith Edwards TB clinical diagnostic tool was compared to gastric aspirates, lymphnode aspirates or sputum gene x-pert MTB/RIF and TB culture. Sensitivity, specificity, negative and positive predictive value of the clinical tool were determined. Data was analyzed using SPSS version 25.

Results

Out of 252 children evaluated, 13.5% (34/252) had TB using the clinical diagnostic tool and 5.2% (13/252) had culture positive TB. The sensitivity of the clinical tool was 76.9%, specificity of 90%. Culture positive TB predictors were lymphadenopathy (AOR 13.74, 95%CI (3.86 - 48.86) p value < 0.001), weight loss (AOR 3.19,95%CI (1.38 - 7.36) p value 0.007), and difficulty breathing (AOR 7.25, 95%CI (1.54 - 34.16) p value 0.012).

Conclusion

The utility of the modified Keith Edwards clinical diagnostic tool for Tuberculosis diagnosis among HIV infected children is limited, calling for further validation. HIV infected children with lymphadenopathy, failure to thrive and difficulty in breathing are at high risk of Tuberculosis.