S/n | Research question | Endpoint | Required variables | Proposed analysis method |
1. | What is the prevalence and incidence of DM/ hyperglycaemia in patients receiving treatment for tuberculosis in Eswatini? |
|
| Descriptive: Frequency tables with percentages, Mean (SD) and/or median (IQR) Comparative analysis: Pearson χ2 test or Fischer’s exact test for categorical variables. T-test or Mann-Whitney for continuous variables. Statistical analysis: Univariate and multivariate logistic regression and mixed-effects model. Missing data: If ≥10%, imputation of missing data will be done, and the results averaged across all the datasets imputed. Sensitivity analysis: Nested multilevel logistic regression analysis with random effects at two levels (region and individual) to account for clustering at regions. Model Fitness: Hosmer-Lemeshow goodness of fit test |
2. | Does DM or hyperglycaemia affect TB treatment outcome in patients receiving treatment for tuberculosis in Eswatini? |
|
| Descriptive: Frequency tables with percentages, Mean (SD) and median (IQR) Comparative analysis: Pearson χ2 or Fischer’s exact test for categorical variables. T-test or Mann Whitney for continuous variables. Statistical analysis: Univariate and multivariate logistic or linear regression models. Missing data: If ≥10%, imputation of missing data will be done, and the results averaged across all the datasets imputed. Model Fitness: Hosmer-Lemeshow goodness of fit test, residual sum of squares |
3. | What factors limit the effective integration of diabetes care into TB Services provision at the health facilities providing TB care in Eswatini? |
|
| Descriptive: Frequency tables with percentages, Mean (SD) and median (IQR) Qualitative analysis: Analysis of both deductive and inductive codes from healthcare worker’s interviews. |
DM, diabetes mellitus; TB, tuberculosis.