Table 1 Description of quantitative and qualitative methods

Refine intervention framework and counseling training

Quantitative: None

Qualitative: We invited clinic staff involved in TB/HIV care to a focus group discussion (FGD) to discuss the barriers to retention in TB care and to obtain their feedback on the intervention. The FGD was facilitated by a trained study team member (BS) in English and isiZulu and followed a semi-structured, open-ended interview guide. We thematically analyzed transcripts and applied the findings accordingly.

Assess the impact of counsellor training

Quantitative: Knowledge, attitudes and practices (KAP) surveys (Additional file 1: Box 3) were administered to TB counsellors pre- and post-training. We compared pre-and post-training responses to the KAP survey knowledge questions (20 questions) using Fisher’s exact test and to the attitude questions (21 questions) using Wilcoxon rank-sum test (α < 0.10). We compared total knowledge scores pre- and post-training using a Wilcoxon rank-sum test.

Qualitative: A trained social worker conducted a FGD with TB counsellors at the end of the intervention (“post-intervention FGD”), following a semi-structured interview guide.

Assess implementation and refine intervention

Quantitative: None

Qualitative: TB counsellors recorded notes and memos after health talks and counselling sessions, which were reviewed by the study coordinator (NS) and a study team member (BS), discussed with the study team, and informed any ongoing changes to the intervention, as needed.

Assess impact on TB testing

Quantitative: We performed an interrupted time series analysis using a quasi-Poisson regression model, including calendar month as a fixed effect to account for the background seasonal trend, to compare the weekly number of TB tests pre- and post-intervention.

Qualitative: Post-intervention FGD

Assess impact on TB treatment initiation

Quantitative: We performed univariate and multivariate (adjusting for age, sex and calendar month, without imputing any missing data on confounders) binomial regressions to compare probabilities of treatment initiation in the two periods, and a Mann-Whitney non-parametric test to compare the median treatment delay (i.e., the number of days from testing to starting TB treatment).

Qualitative: Post-intervention FGD

Assess impact on TB treatment outcomes

Quantitative: We performed univariate and multivariate logistic regression analyses (adjusting for age, sex, smear status, and HIV and ART status, without imputing missing confounders) to compare the probability of TB treatment completion. In the main analysis, we compared the study and historical control periods and included all new TB patients in the study period regardless of whether they enrolled into the study (i.e. an intention-to-treat analysis). In our sensitivity analyses, we compared: patients enrolled in the study period to all other patients (in both the study and historical control period); patients enrolled in the study to patients in the historical control period; and patients enrolled to patients not enrolled in the study during study period (Additional file 1: Table 8).

Qualitative: Post-intervention FGD

Explore counsellors’ and patients’ perspectives on the impact, acceptability and feasibility of the intervention

Quantitative: Descriptive analyses of brief exit surveys (Additional file 1: Box 4) that were administered to all enrolled patients who received both counselling sessions.

Qualitative: In addition to the post-intervention FGD and brief patient exit surveys, a purposive sample of enrolled patients (aiming for maximum variation in patient characteristics based on age, gender, education, and TB/HIV history) were recruited for one-on-one, in-depth interviews with the study coordinator (NS), following a semi-structured interview guide. All of the collected data (survey responses and transcripts) were thematically analyzed using a constant comparative approach.