Prevalence of subclinical pulmonary tuberculosis and its association with HIV in household contacts of index tuberculosis patients in two South African provinces: a secondary, cross-sectional analysis of a cluster-randomised trial

Table 2 Association between HIV status and TB state assessed by univariable and multivariable logistic regression

Analysis	HIV status	Number/sample	Unadjusted OR (95% CI) [p value]	Adjusted^* OR (95% CI) [p value]
Subclinical TB versus no TB^a	Negative	33/1686	Ref	Ref
Subclinical TB versus no TB^a	Positive	15/367	2.15 (1.15–4.01) [0.016]	2.00 (0.99–4.01) [0.052]
Symptomatic TB versus no TB^b	Negative	10/1663	Ref	Ref
Symptomatic TB versus no TB^b	Positive	10/362	4.61 (1.94–11.13) [< 0.001]	5.05 (2.22–11.59) [< 0.001]

CI confidence interval, OR odds ratio, Ref reference category. Unadjusted ORs calculated using simple logistic regression via the glm() function in R, with province included as fixed term effect to adjust for clustering. Adjusted ORs were calculated using multiple logistic regression via the glm() function in R. Robust-clustered CIs were calculated based on clustering at a household level, using the coeftest() function in R. ^*Adjusted for the province, household contact sex, household contact age and index case HIV status. Missing data dealt with listwise: ^a1955/2057 included in the adjusted analysis, ^b1929/2029 included in the adjusted analysis

ISSN: 2731-913X