Table 2.

Primary objective analysis

Model	Estimated intercept (SE)	One-sided 95% Wald-type CI for model intercept (lower bound)	Number of coerced centers	Number of titration successes	Total N	Missing (%)
Linear mixed effect	0.72 (0.08)	0.59a	4	21	29	0 (0.0%)
Generalized linear mixed effects	0.72 (0.09)	0.57	4	21	29	0 (0.0%)

The titration success of study subjects after the SMS-supported basal insulin titration (at latest at visit 4) was the primary endpoint of this study and was evaluated as the proportion of successfully titrated subjects compared to the historic control proportion of 0.4 according to literature search. The decision was based on the comparison of the lower one-sided 95% confidence interval (CI) bound for the fixed intercept to the control value based on the linear mixed effects model (LMM). Significant superiority of the SMS-supported titration over the control was shown if the lower one-sided 95% CI bound is greater than 0.4. Generalized linear mixed effects model (GLMM) result was reported as sensitivity analysis only. LMM/GLMM model with fixed intercept and random intercept per center. The center random intercepts are estimated (close) to zero due to insufficient data amount as premature recruitment stop results in low sample size. The centers with less or equal to three subjects are coerced to one

^aSuperiority of SMS service over the historically reported control is shown

Total N number of subjects who started the service, SE standard error