Figure 3.

Association between epidemiological and clinical characteristics and losses in the latent tuberculosis infection (LTBI) cascade of care. (A) Time (per week) distribution among losses in the LTBI cascade of care and those who completed the LTBI cascade. Data were compared using the Mann-Whitney U test. (B) Frequency of race/ethnicity (black and pardo), income (see definition below), secondary smoking, comorbidities (see definition below) and education (illiterate) between TB contacts stratified based on losses in the LTBI cascade. Data were compared using Fisher’s exact test. (C) Generalised estimating equations analysis to evaluate association between epidemiological and clinical characteristics and losses in the LTBI cascade of care. The study population was stratified according to complete TB preventive therapy (TPT) in the LTBI cascade (complete LTBI cascade of care vs losses in the LTBI cascade of care, see online supplemental table 1 for detailed univariate comparisons). A multivariable analysis (see the Methods section for details) was employed with each variable individually (unadjusted) and variables (panels A and C) were included in a multivariable model (adjusted). In all the comparisons, significant p values are shown in bold-type font. Comorbidities: at least one comorbidity (diabetes, hypertension, cancer, chronic obstructive pulmonary/emphysema, kidney disease, heart disease, liver disease and depression). Low income: without income/equal or less than a minimum wage (reference: more than a minimum wage) ace (Black/Pardo) reference: white, Asian, Indian. Ethnicity was self-reported. Time (V1_TB-V1_C), time (in weeks) difference between the visit 1 of the tuberculosis case and the visit 1 of the contact.