Fig 3.

Percentage of patients experiencing an adverse tuberculosis (TB) outcome analysed by poverty, education level, symptom duration, previous TB, catastrophic costs and resistance profile. Error bars represent 95% confidence intervals; p-values correspond to the association of each variable with adverse TB outcome in univariable logistic regression except for poverty and symptom duration that were analysed as continuous variables. In multivariable regression analysis, the following variables remained independently associated with adverse TB outcome: time too unwell to work (p=0.02), catastrophic costs (p=0.003), having had a previous episode of TB (p=0.004) and currently having multidrug resistant (MDR) TB (p<0.0001). Reproduced with permission from Wingfield et al.¹¹