MOSAIC participants answered the question: "How often did you miss your insulin shots during the last 7 days?". At each visit, participants were defined as visit adherent if the participant answered "I did not miss any shots" or "I missed some of my shots". They were defined as not visit adherent if he/she answered any of the following: "I missed about half of my shots", "I missed most of my shots", or "I missed all of my shots". A participant is defined to be study adherent if at least: 4 out of 5 visit adherent.

As diabetes management practices vary across the countries represented in MOSAIC, a more apt measure of reaching goal is the personalized target that was set for each patient at the beginning of the study.

Baseline HbA1c > 7.80 (reference:HbA1c≤7.80 median)

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy.

Baseline HbA1c missing (reference:HbA1c≤7.80 median)

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy.

Diabetes duration > 11 years

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy.

Age (per year increase)

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy.

Baseline insulin therapy: combination

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy.

Baseline insulin therapy: prandial only

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy.

Baseline insulin therapy: pre-mixed only

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy.

Discrimination domain of the IPC

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy. IPC ranges from 1-5 with higher scores indicating more discrimination.

Diabetes Distress Scale total score > 2

Multivariable logistic regression model examined baseline participant-, physician-, and healthcare system-related factors associated with the occurrence of any change in insulin therapy. Diabetes Distress Scale ranges from 1-6 with higher scores indicating more distress.