Jervis et.al report a retrospective comparison between GLP-1 and mixed insulin use in hospitalized patients on hypoglycemia and glycemic control. While the manuscript highlights an important issue of continuing GLP1 RA in inpatients, there are some methodological flaws that should be discussed (see below) and some assumptions which should be updated based on current guidelines and recommendations.
Abstract
page 3, line 8- Glycaemic control improves outcomes for hospitalized patients. This is debatable and the language should be softened. Numerous conflicting publications exist on this topic.
Introduction
Page 4-line 6- first paragraph- the fact that hyperglycemia is a predictor of poor outcomes does not mean that glycemic control changes these outcomes. This is a controversial issue with numerous conflicting reports. The first sentence in the introduction is unproven and should be changed.
Page 4 Line 25- “Insulin risks causing hypoglycemia, which is associated with poor hospital
Outcomes” – this is unclear. Do you mean treatment with insulin increases the risk for hypoglycemia? Furthermore, there is no evidence linking insulin treatment to poor hospital outcomes!
Page 4 Line 30- similar or better- i.e semaglutide vs. glargine.
Methods
It is unclear from the text how patient selection was done. How do you know patients were previously treated with insulin mix or GLP1 and that treatment (especially with insulin) was not initiated in the hospital?
Page 5 line 1- what about novomix 50 or other ratios of premixed insulins? Please explain why only these formulations were used as search terms.
Page 5 line 9 – only patients who had been ordered a dose of GLP-1 RA as inpatients were included- could glp1 be initiated during hospitalization?
Page 5 line 43- “Rescue therapy was defined as the use of 10% or 20% dextrose or glucagon.” What about oral administration of sugar/glucose/ food? This is probably the most common method of treating hypoglycemia in conscious inpatients.

Results
Page 6, line 13 - “regular rapid insulin” – do you mean regular (i.e. human insulin) or rapid-acting (i.e asparte, glulisine, or lispro insulin).

Please use generic names throughout the manuscript (bydureon and Byetta- are weekly exenatide and daily exenatide formulations, etc. )

Page 6 line 55- a lower proportion of days was spent with hyperglycemia in GLP-1 active- hyperglycemia was not defined in the methods section.

Discussion
.
Page 7 line 17- “Our observational study has shown that inpatient use of GLP-1 RA is associated with less hypoglycemia and lower mean glucose concentration than those using twice daily mixed insulin.”

An inherent problem with this statement is that 32 subjects in the GLP1 users' group also used mixed insulin (was this accounted for or continued during hospitalization?). Therefore, this is not a comparison of glp1 to mixed insulin but of GLP-1 on top of ‘ usual care’ including some patients receiving mixed insulin vs. no GLP-1 and mixed insulin.

Currently, the ADA /EASD standards of medical care in diabetes state that “Therefore, premixed insulin regimens are not routinely recommended for in-hospital use.”
This should be discussed.

Table 2-
% days of hyperglycemia are higher in insulin active vs. inactive- this suggests reactive use of insulin and not as currently suggested in the ADA/EASD standards of medical care in diabetes, a proactive approach to insulin use in hospitalized patients. The lack of a standardized insulin protocol in your facility should be discussed and explained in the manuscript.