Treatment differences between baseline and 6 months and baseline and 12 months will be computed by analysis of covariance, with age, sex, relation to patient, and baseline scores on the co-morbidity, MMSE scores, GDS scores, and outcome variable at baseline as covariates.
To adjust for multiple baseline characteristics, standardized outcome ratios will be computed using the 'indirect standardization' approach.
For all of these covariables, the correlation coefficients for their changes from baseline vs. the HbA1c parameters will be computed, too.
Descriptive statistics for change in endpoints (mean, median, and standard deviation) between baseline and 12-week measurements will be computed separately for each program format (face-to-face or hybrid mobile technology-supported program).
Descriptive statistics will be computed for baseline demographic characteristics and for all outcome measures, at each time point.
Descriptive statistics of the total population and of the two treatment groups separately will be computed to provide baseline characteristics of the patients in both treatment arms.
Effect sizes and their confidence intervals (adjusted for baseline values of outcomes and nested study design) will be computed.
The difference of differences of the results from the intervention and control sites (obtained from baseline and end line surveys and from pre and post intervention facility data) will be computed to ascertain the influence of the intervention.
Estimates and CIs for outcomes will be computed and compared across the intervention and comparison groups at baseline and follow-up.
The main analyses will be computed using analyses of covariance in view of statistical studies suggesting that controlling for the baseline score is superior to simple pre-post comparisons and usually leads to an increase in power.
