Bi-variable analysis
We performed bi-variable conditional logistic regression to assess the crude odds ratio. In the bi-variable analysis eight variables give a P-value < 0.25. The variables were previous tuberculosis infection history, baseline functional status, baseline WHO clinical stage, baseline Hgb value, baseline BMI, IPT intake, baseline CD4 value, and months of ART intake. Based on the crude data, all these variables except baseline BMI had a statistically significant association with TB infection among HIV patients who were taking ART. The crude odds ratio and the confidence interval for the variables is as follows, previous tuberculosis infection history (cOR; 2.744, 95%CI;1.454-5.180), baseline functional status (ambulatory, cOR; 3.243, 95%CI;1.269-8.288, bedridden, cOR; 4.00, 95%CI; 0.363-44.133), baseline WHO clinical stage (cOR; 8.293, 95%CI;1.77-38.693), baseline Hgb value (cOR; 1.857, 95%CI;1.125-3.064), IPT intake (cOR; 3.415, 95%CI; 1.437-8.118), baseline CD4 value(cOR; 1.796, 95%CI; 1.273-2.54) and months of ART intake (<12 Months, cOR; 28.391, 95%CI;7.705-104.6, 13-36 Months, cOR; 3.876, 95%CI; 1.416-10.609) (Table 4) .