Table 3.

Statistical Analyses for Confirmatory Endpoints.

Hypoglycemic episodes during 36 weeks of maintenance, on-treatment analysis
Data used for analysis	Preamendment: On-treatment data during maintenance period (36 weeks) Postamendment: On-treatment data during maintenance period 2 (36 weeks)
Statistical model	Negative binomial with treatment, number of pretrial OAD, region, sex, and dosing time as factors, age as covariate, and log of exposure time during analysis period as offset
Handling of missing data	For patients without on-treatment data during the analysis period, values will be imputed through multiple imputation by treatment arm based on data from patients discontinuing the randomized treatment during the analysis period
Hypoglycemic episodes during 36 weeks of maintenance, ITT analysis
Data used for analysis	Preamendment: On- and off-treatment data during maintenance period (36 weeks) Postamendment: On- and off-treatment data during maintenance period 2 (36 weeks)
Statistical model	Negative binomial with treatment, number of pretrial OAD, region, sex, and dosing time as factors, age as covariate, and log of observation time during the analysis period as offset
Handling of missing data	For subjects who withdraw from trial prior to analysis period, values are imputed through multiple imputation by treatment arm based on all off-treatment data from the analysis period
Basal insulin doses at end of trial, on-treatment analysis
Data used for analysis	Preamendment: All available doses of randomized treatment during the trial duration (52 weeks) Postamendment: All available doses of randomized treatment during the trial duration (up to 88 weeks)
Statistical model	MMRM with treatment, number of pretrial OADs, region, sex and dosing time as factors and age and pretrial insulin dose as covariatesa
Handling of missing data	MMRM
Basal insulin dose at end of trial, ITT analysis
Data used for analysis	Preamendment: All available doses of randomized treatment and basal insulin taken after premature discontinuation during the trial duration (52 weeks) Postamendment: All available doses of randomized treatment and basal insulin taken after premature discontinuation during the trial duration (up to 88 weeks)
Statistical model	ANCOVA with treatment, number of pretrial OAD, region, sex and dosing time as factors and age and pretrial insulin dose as covariates
Handling of missing data	Missing doses at last assessment visit imputed through multiple imputation by treatment arm and treatment completion status

^aIf the model does not fit, the last visits in maintenance period 1 might be excluded from analysis as needed.