This manuscript details patterns of use of oral glucose lowering drugs (OGLD) among adults with T2D in the multi-national JADE registry, revealing SUs are commonly used OGLDs from 2007 - 2019.

SUs were the most common OGLD; of SU-based OGLD regimens, SU + metformin was most common; of SUs, gliclazide use was associated with having HbA1c < 7% at JADE registration and lower self-reported hypoglycemia 3 months prior to registration.

Strengths include the diversity of data from 11 Asian countries/regions including over 50,000 people with T2D + OGLDs.

However, several major limitations exist --

[1] The introduction doesn’t clearly articulate the “why” motivating the study, leaving the reader missing context for several key study design choices. Recommend enhancing introduction to provide the context needed to support rationale for the study’s:

(i) emphasis on SU’s over other OGLDs, (ii) decision to exclude insulins/GLP-1 RAs, (iii) decision to highlight gliclazide but pool “other SUs” in section headers, results, and regression analyses, (iv) emphasis on “no drugs” as comparison group of interest, and (v) decision to pool utilization across over a decade (see Point [3] below).

Introduction largely skips the broader SU safety debates occurring the study inclusion periods, and would benefit from contextualizing the impact of recent landmark study results on this topic, like CAROLINA (PMID: 31536101). Lines 312 – 323 in the discussion may also be useful introductory context.

[2] As written, the abstract, methods, and results don't specify important temporal relationships between an individual’s study index date, initiation of OGLD therapy, and variables of interest (HbA1c < 7%, hypoglycemia events, demographics) that influence interpretation of results.

(i) For example, lines 51 – 52 and 276 – 277 are unclear (and/or may imply at first read) that gliclazide precedes HbA1c goal attainment or hypo events. Table 3 title clearly explains this is an association with HbA1c and hypo history at or ahead of registry entry, but this unclear throughout key sections of the manuscript.

(ii) Does entry to cohort equate to baseline evidence of SU use at registration? Or may evidence of SU use occur at any time an individual is enrolled in JADE, i.e. variable time between registration and first SU use across the cohorts?

(iii) Similarly, missing data on treatment duration / exposure length to OGLD isn’t introduced until line 346 in discussion.

[3] From 2007 to 2019, paradigm shifts in T2D treatment landscape and care guidelines have occurred. How much of the present results are driven by contemporary prescription patterns (e.g. the last ~3 or 4 years) verses the years prior to the accumulation of SGLT-2i, GLP-1RA, and DPP-4i cardiovascular outcomes trial results (~2015) and changes to care guidelines (2018+)?

Recommend discussion on how the study period impacts the interpretation and applicability of the findings, and authors may consider additional analyses to compare earlier study periods to more recent.

The revisions provided by the authors have elevated the impact of the descriptive content evaluating multi-national utilization patterns, particularly with revisions to the introduction, datasource description, and discussion.

The revised description of JADE provides excellent details to aid the reader in interpretation of this work, including collection methods and known missing data elements.

The major limitations of the manuscript in it's current form are (1) the interpretation of the cross-sectional observational design as effectiveness, tolerability, and safety evaluation of SU's, given the multiple temporal confounders between variables, absence of pre- and post-intervention glucose control or A1c risk, and prevalent user design and (2) the missing background details to strengthen the rationale for study design choices.

(1) Recommend (i) being explicit and consistent around the associative nature between drugs and variables of interest and (ii) softening language around "effectiveness" when interpreting results.

There are several statements throughout the manuscript that overstate the results of the regression models. Given the cross-sectional design, what was evaluated was the characteristics of prevalent users rather than the effectiveness of a treatment for which no start date, duration, or adherence was known or accounted for.

Similarly, the authors discuss the “pragmatic” nature of this study design in relationship to RCTs. However, this is a real-world cross-sectional design, not an emulated RCT with efforts to assess pre- and post- intervention changes in A1c or hypo risk.

I Recommend modifying language accordingly. For example, the use of the word "achieve" appears in several places, in lieu of "associated with" or "enter the cohort with" or "have history of" or other language appropriate for the methods employed.

Example, Line 106: “…we evaluated the effectiveness, tolerability, and safety of SU-based therapy in a real-world setting.”

Example, Line 272: “After adjustment for predefined covariables, gliclazide users were more likely to achieve HbA1c <7% […] than users of other SU.”

Example, Line 306: “Amongst the SU group, 50% were treated with gliclazide who had increased odds of achieving HbA1c < 7% and lower risk of hypoglycemia than other users of SU […]”

(2) I Recommend explicitly clarifying core components of the study design and rationale, as previously recommended. The authors have revised the introduction to address (i) the emphasis on SU's over other OGLDs, which provides additional context to (ii) the exclusion of insulins/GLP-1RAs. Still missing is (iii) the explicit discussion of the rationale to highlight gliclazide while pooling "other SUs" in discussion and analyses (iv) why "no drugs" was chosen as a comparison group (v) why utilization was pooled across a decade.

The authors have addressed previously raised points to the best of their ability given the limitations of the present study. Compliments again on the excellent description of the JADE register included in Revision 1.