Study population baseline data.

All data are at individual level and pre-treatment except where indicated.*Univariable regression adjusted for sex. The first p-value column corresponds to comparison of controls (n = 487) with all TB patients regardless of MDR status (n = 876). The second p-value column corresponds to comparison of patients with non-MDR TB (n = 783) versus MDR TB (n = 93).**Household earnings per month during different treatment stages represented as mean Peruvian Soles and, in parentheses, as a proportion of TB patients' mean monthly household earnings throughout entire illness. Confidence intervals are those of mean monthly earnings in Peruvian Soles.***Debt at recruitment represented as mean Peruvian Soles and, in parentheses, as a proportion of TB patients' mean monthly household earnings. Debt at recruitment was used in the final multivariable regression model rather than total debt (sum of debt at recruitment plus debt at 24 wk of treatment) because only 461 patients had 24-wk debt data available.CI, confidence interval; IQR, interquartile range; SD, standard deviation.