Skip to main content

Table 4 Joint summary of main quantitative and qualitative findings

From: Acceptability, feasibility, and impact of a pilot tuberculosis literacy and treatment counselling intervention: a mixed methods study

 

Quantitative analyses

Qualitative themes

Study implementation

TB counsellors’ median pre-training and post-training TB knowledge score were 50% (IQR 7.5%) and 65% (IQR 17.5%). The median change in score was 12.5% (95%CI = 5.0 to 20.0%)a

Enrolled 51.5% (84 out of 163) of all TB patients who started treatment during the study period.

Median duration of health talks (n = 58) was 94.1 (± 42.4) minutes

• Clinic engagement and limitations

• Counsellors’ knowledge gaps

• Proficiency and tedium of health talks

TB testing & treatment initiation*

Comparing the study period to the historical control period:

- Number of diagnostic tests increased by 1.36 times (95%CI 1.23 to 1.58) (see Fig. 2 for observed and expected TB tests performed pre-and post-intervention)b

- Probability of treatment initiation increased from 7.8 to 19.0%; with an estimated increase of 10.1% (95%CI 1.5 to 21.3%) after adjusting for potential confoundersc

- Median treatment delay decreased from 7.0 days to 4.5 days, a change of 2.5 days (95%CI 2.0 to 3.0 days)d

• Perceived patient engagement

• Difficult patient queries

Treatment retention*

Probability of treatment completion was similar during the study period (45%) and the historical control period (46%). There was an estimated increase during the study period of 4.4% (95%CI −7.3 to 16.0%) after adjusting for baseline confounderse

Among those enrolled, 26 (31%) received only the first of two study counselling sessions.

• Improved treatment self-efficacy

• Alleviation of anxiety, fears and perceived stigma

• Barriers to treatment and counselling

  1. aEstimated using a Wilcoxon rank-sum test
  2. bRate ratio comparing number of TB tests performed during the study period to the historical control period estimated via a quasi-Poisson regression model, including a dummy indicator variable for the intervention and a fixed effect for calendar month to account for background seasonal trend.
  3. cEstimate adjusted for age, sex and calendar month using a multivariate binomial regression model with an identity link.
  4. dConfidence interval of the difference in medians estimated using the adjusted bootstrap percentile (BCa) method.
  5. eBased on an intention-to-treat analysis comparing all patients who started treatment during the study period versus the historical control period, using a binomial regression with an identity link, and adjusted for the following baseline characteristics: age, sex, smear status, and HIV and ART status