Skip to main content

Table 2 Pooled effect sizes for different sub-categories of studies.

From: Alcohol use as a risk factor for tuberculosis – a systematic review

Study category No of studies Hetero-geneity test
Cochrane's Q p-value (I2)
Pooled, fixed effect assumption
(95% confidence interval)
Pooled, random effect assumption
(95% confidence interval)
Level of exposure     
High exposure 11 < 0.01 (0.82) 2.90 (2.39–3.51) 3.50 (2.01–5.93)
Low exposure 4 0.46 (0.00) 1.08 (0.82–1.40) 1.08 (0.82–1.40)
High-exposure studies     
Controlled* for HIV status 7 0.03 (0.57) 2.93 (2.37–3.61) 3.26 (2.26–4.70)
Controlled* age, sex, SES, smoking 5 0.04 (0.61) 3.27 (2.38–4.50) 3.49 (2.06–5.90)
Controlled* HIV, age, sex, SES, smoking 4 0.07 (0.42) 3.92 (2.70–5.71) 4.08 (2.49–6.68)
Controlled* infection, age, sex, SES 4 0.23 (0.30) 4.11 (2.84–5.94) 4.21 (2.73–6.48)
Excluding three smallest studies 8 0.03 (0.59) 2.75 (2.19–3.46) 2.94 (1.89–4.59)
Excluding three smallest and Brown I and Kim 6 0.32 (0.15) 2.76 (2.34–3.81) 2.96 (2.28–3.85)
Pulmonary TB cases only** 2 0.49 (0.00) 3.67 (2.58–5.22) 3.67 (2.58–5.22)
All types of TB** 6 < 0.01 (0.83) 2.52 (1.98–3.19) 2.87 (1.47–5.58)
  1. *Controlled for respective covariates, either by design (e.g. through inclusion/exclusion criteria) or in the analysis (stratification or multivariate analysis)
  2. **Excluding three smallest studies