The primary objective of this study is to estimate the time spent on initial treatment regime before significant treatment change for patients with type 2 diabetes initiating therapy with either insulin or exenatide for the first time.

Initial treatment regime is defined as the treatment regime prescribed when the patient is enrolled in the study.

Significant treatment change for patients initiated on insulin or exenatide is defined as at least one of the following:

Insulin:

• Addition of a new medication for the treatment of type 2 diabetes
• A change in the number of times insulin is administered per day
• Discontinuation of any insulin initiated at baseline
• Substitution of a human insulin for an analogue insulin or vice-versa.
• Switching between brands of the same class/type of insulin is not included in the definition of significant treatment change.

Exenatide:

• Addition of a new medication for the treatment of type 2 diabetes
• Discontinuation of exenatide.

All patients who provided consent to release information and who fulfil the study entry criteria were included in the analyses. Patients were assigned to the exenatide BID or insulin cohort based the injectable treatment started at baseline; analyses were conducted irrespective of later treatment changes.

Higher BMI was one of the Factors evaluated for association with treatment choice at baseline. BMI was calculated as body weight in kilograms (kg) divided by height in meters (m) squared (kg/m^2). The mean BMI at baseline is provided below and the statistical analysis provides the 2 arms odds ratio for BMI=1 kg/m^2 higher. Participants were assigned to the exenatide BID or insulin cohort based the injectable treatment started at baseline; analyses were conducted irrespective of later treatment changes. Baseline was Visit T1 (prior to start of treatment).

All participants who provided consent to release information, fulfilled the study entry criteria, and had a start date provided were included in the analyses.

Higher HbA1c was one of the Factors evaluated for association with treatment choice at baseline.HbA1c was reported in percent of hemoglobin. The mean HbA1c at baseline is provided below and the statistical analysis provides the 2 arms odds ratio for HbA1c=1% higher.

All participants who provided consent to release information, fulfilled the study entry criteria, and had a start date provided were included in the analyses.

Older age (1 year older) was one of the Factors evaluated for association with treatment choice at baseline. The mean age at baseline is provided below and the statistical analysis provides the 2 arms odds ratio for age 1 year older. Participants were assigned to the exenatide BID or insulin cohort based the injectable treatment started at baseline; analyses were conducted irrespective of later treatment changes. Baseline was Visit T1 (prior to start of treatment).

All participants who provided consent to release information, fulfilled the study entry criteria, and had a start date provided were included in the analyses.

Diabetes Health Profile (DHP-18) - consists of 18 items across 3 domains (psychological distress, barriers to activity, and disinhibited eating), with each item standardized score rated from 0-100; 0=no dysfunction, higher numbers=greater dysfunction. The subscale of disinhibited eating was one of the Factors evaluated for association with treatment choice at baseline. The number of participants with disinhibited eating at baseline is provided below and the statistical analysis provides the 2 arms odds ratio for disinhibited eating. Participants were assigned to the exenatide BID or insulin cohort based the injectable treatment started at baseline; analyses were conducted irrespective of later treatment changes. Baseline was Visit T1 (prior to start of treatment).

All participants who provided consent to release information, fulfilled the study entry criteria, provided the specific data (DHP-18 subscale on disinhibited eating) and had a start date provided were included in the analyses.

Random Glucose 1 millimole per liter (mmol/L) higher was one of the Factors evaluated for association with treatment choice at baseline. Random glucose is a glucose within the last 6 months prior to baseline. The mean is provided below and the statistical analysis provides the 2 arms odds ratio for the glucose 1 mmol/L higher. Participants were assigned to the exenatide BID or insulin cohort based the injectable treatment started at baseline; analyses were conducted irrespective of later treatment changes. Baseline was Visit T1 (prior to start of treatment).

Frequent glucose self-testing (1 test/week more) was one of the Factors evaluated for association with treatment choice at baseline. The mean number of self monitoring blood glucose tests per week over the last 4 weeks prior to baseline was determined at baseline and is provided below. The statistical analysis provides the 2 arms odds ratio. Participants were assigned to the exenatide BID or insulin cohort based the injectable treatment started at baseline; analyses were conducted irrespective of later treatment changes. Baseline was Visit T1 (prior to start of treatment).

All participants who provided consent to release information, fulfilled the study entry criteria, and had a start date provided were included in the analyses.

Receipt of diet and exercise advice was one of the Factors evaluated for association with treatment choice at baseline. The number of participants who checked yes or no during the baseline visit for prior receipt of diet/exercise advice in his/her Diabetes management is provided below and the statistical analysis provides the 2 arms odds ratio. Participants were assigned to the exenatide BID or insulin cohort based the injectable treatment started at baseline; analyses were conducted irrespective of later treatment changes. Baseline was Visit T1 (prior to start of treatment).

All participants who provided consent to release information, fulfilled the study entry criteria, and had a start date provided were included in the analyses.

Higher (1 mmol/L higher) LDL cholesterol was one of the Factors evaluated for association with treatment choice at baseline. The mean LDL cholesterol at baseline is provided below and the statistical analysis provides the 2 arms odds ratio for 1 mmol/L higher at baseline. Participants were assigned to the exenatide BID or insulin cohort based the injectable treatment started at baseline; analyses were conducted irrespective of later treatment changes. Baseline was Visit T1 (prior to start of treatment).

All participants who provided consent to release information, fulfilled the study entry criteria, and had a start date provided were included in the analyses.

Changes in HbA1c From Baseline to Month 24

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Percentage of Patients Achieving HbA1c Concentration <7.0% at Month 24. Only patients with baseline HbA1c >= 7.0 % were included in this analysis

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Percentage of Patients Achieving HbA1c Concentration <6.5% at Month 24. Note: Only patients with baseline HbA1c >=6.5% were included in this analysis.

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Changes in Weight From Baseline to Month 24

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Incidence of Gastro Intestinal Symptoms between Baseline and 24 Months

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Incidence of Hypoglycemia between Baseline and 24 Months

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Reasons for Discontinuation of Baseline Regimen

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Hazards ratios from Backward Cox Regression Model for time to significant treatment change in Insulin cohort

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Hazards ratios from Backward Cox Regression Model for time to significant treatment change in Exenatide BID cohort. EQ-5D (Health Questionnaire Copyright @ Euro QoL Group 1998).

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Percentage of Patients Contacting Health Care Providers Between Baseline and 24 Months

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Number of contacts with Health Care Providers Between Baseline and 24 Months

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.

Percentage of Patients Hospitalized Between Baseline and 24 Months

Patients were assigned to the exenatide BID or insulin cohort based on their initial injectable treatment started at baseline, and analyses were conducted irrespective of later treatment changes.

All patients who provided consent to release information and who fulfilled study entry criteria were included in this analysis population.