The models were estimated using a weighted linear regression with clustered SEs to account for multiple observations per patient. Weights were set as the inverse of the number of the patient’s observations so that weights for each patient summed to 1 and each patient’s data contributed equally to the analysis. SEs are presented in parentheses. Ref, reference. *** P < 0.01, **P < 0.05, *P < 0.1.

^aUse of advanced diabetes medications was defined as equal to 1 if the patient was treated with any monotherapy or combination of type 2 diabetes drugs beyond metformin monotherapy and defined as equal to 0 if the patient was treated with metformin monotherapy during the baseline year.

^bAny diabetes complications was defined as 1 if patients the DCSI score was ≥1 (22).

^cCovariates measured during follow-up were measured between the index HbA_1c measurement and the subsequent measured used to calculate change in HbA_1c levels and could vary within patients with multiple change in HbA_1c estimates. For example, a patient may be adherent at the first change in HbA_1c reading but may be classified as poorly adherent at a later change in HbA_1c reading.

^dAddition of another type 2 diabetes drugs after initiation was defined as 1 if additional drug classes were added after initiation of the index drug.

^eAdherence to index drug was defined as the percentage of days covered by GLP-1 RA or DPP-4 ≥80%.