Delivery of Dose Adjustment For Normal Eating (DAFNE) in the UK and Ireland 2019-2022

Overall: 
This work is a welcome addition to the literature relating to DAFNE in the UK and Ireland and covers an obvious gap in the literature that has now been addressed. Namely, they review completion rates of DAFNE courses over a 3 year period, with assoicated descriptions of the attendees, and analysis of factors potentially assocaited with non-completion. 
The authors have produced a report that is generally well-written and deserving of getting indexed. 
In my review I have outlined a number of minor comments that should be considered by the authors as suggestions that may potentially improve clarity through minor ammendments. 
My primary concerns relate to  the presentation of data and their associated reporting of statistical information. These aspects require direct improvement prior to being indexed.
Thank you to the authors for this contribution to the literature!  


Justification of answers:
1) "If applicable, is the statistical analysis and its interpretation appropriate?"
• I have selected "partly" due to comments regarding statistical test reporting and need for clarification regarding certain aspects (e.g. fraction denominators, and the handling of missing data). I suspect the answer will become "Yes" if these aspects are addressed. 

2) "Are all the source data underlying the results available to ensure full reproducibility?"
I have selected "No" as the "underlying data are not publicly available for this study." Nethertheless, the author's comments on this are clearly outlined and likely fully justified (given the stipulations of DAFNE Central). 

Abstract:
These is a well-written and easy-to-read summary of the work. 
Minor comment: "during Covid-19"
• This presently lacks phrasing that is not directly implied but is required relating to time period
e.g. during the COVID-19 pandemic.
• While covid may often appear in lower case in lay  writting, my understanding is the correct version is uppercase COVID-19 given that it is shortened version of Coronavirus disease 2019. 
• These suggestions should be address elsewhere whereever they apply, if accepted by authors. 

"hypo"
• While I very much appreciate  the author's avoidance of any technical language the shortening of "hypoglycaemic event" seems to be used inconsistently in the text. I also suspect  that leaving this in full i.e. "hypoglycaemic event" does not reduce readability in any way, and hence keeping in full would be the preferrable  option of the two.  

Keywords: 
• Should it be permitted by this journal you may wish to use the maximum number permitted by this journal in the interests of improved indexing, ease of searching for readers, and potentially improved citation. 
• From my personal review of keywords and MeSH terms for DAFNE-related publications, you may wish to consider addition of some the below:
diabetes mellitus;
diabetes mellitus, type  1; 
glycemic control;
diet;
DAFNE.

Introduction: 
This is very well-written and clear. No comments/concerns. 

Methods: 
This is generally well-written and clear; however,  on review of results and tables a number of points have come up that may also need addressing in the methods section. (I have not duplicated these comments). 

" on an annual basis to an anonymised central clinical database."
• Recommendation: list the name of the database here i.e. an anonymised central clinical database (DAFNE Central). I have  taken the name from supplementary files. 

"(i.e., age (dob), "
• suggest remove the comma after i.e. 
• suggest remove "(dob)" given that  you only ever received "year of birth", and not date of birth.   

"As data were provided at the level of the patient, to determine unique courses within the dataset for each centre, course delivery date and course format were used: standard DAFNE, 5X1 [Face to Face (F2F)], remote DAFNE [fully remote], remote DAFNE [blended with F2F support], pump DAFNE [F2F], remote pump DAFNE [fully remote]). We relied on the assumption that more than one course of the same format would not be delivered by the same centre on the same date. For example, if a standard DAFNE course was delivered at Centre 1 on 08-09-21 we assumed that represented a single course"
• This is well enough explained that this could be replicated. However, assuming this may be tedious, with a plausible chance of error, you may wish to consider recommending that DAFNE Central should consider indexing their courses in their dataset with a unique ID (note this is optional, and only for your consideration). Ultimately it is users of these dataset whose feedback is most important in making improvements.  


Results: 
The results section is generally easy-to-follow with appropriate provision of in text and in table data. 
My main concerns that require improvement relate to the reporting of statistical information so as to make it clear - what has been tests, and what the results of the tests are. Practically this should be easily addressed by ammendments to tables. 
I suspect that providing table/figure standard legends, that provides enough information about the table/figure result for it to be understood (without much need for  reference elsewhere) would be preferrable than the use of various asterixes, which differ table-to-table. This point may be optional, but the authors should consider whether legends could be improved and the use of asterixes minimised, in the interests of improved readability.

Missingness
• "however there were high levels of missingness post course" consider adding "however there were high levels of missingness post course for certain variables" as this does not apply to all variables (at least from what I can see in the associated tables). 

Average
• when using the term "average" I would recommend explicitly stating the type of average used (i.e. mean, median, mode etc). I suspect where the word "average" was used this was referring to the mean - but this should be explicit and consistent. 

Punctuation
"In the UK a greater proportion of people did not complete DAFNE 5x1 courses compared to Ireland (21% (n = 127) vs. 12.5% (n = 5))" missing fullstop at end. 

Brackets
"did not complete (mean [sd] 39.0 (14.0) vs. 42.1 [14.9])." ammend brackets to either square brackets or curved for consistency. 

Missingness
• I recommend the authors provide the definition of this used in the study (given that the practicalities of the definition used will often vary study to study). 
• see comments elsewhere regarding censorship which may apply to this. 


Discussion: 
The discussion is well-written, and  appropriately addresses the main study findings, limitations,  and relevant literature. I have hence only left minor comments. 

"We analysed data on DAFNE courses delivered between 2019 and 2022, during which a median of 17 courses, ranging from 2 to 74 courses across centres, were delivered." 
• I presume the intended meaning is a "median of 17 courses" per centre,  if so please include per centre. If it is a different aspect e.g. per year, please include.

"exception that Irish graduates" 
• given that "Irish" when used as an adjective for a person, most commonly pertains to a nationality, as opposed to attending an DAFNE centre in Ireland,  consider remove words pertaining to nationality in preference for more accurate desriptor. e.g. "with the exception that attendes in Ireland were" or "exception that graduates of Irish DAFNE centres..."

"suggest early offering of courses"
• consider change "early" to "earlier" which may better address the point made (i.e. a relative difference). 

"had dropped from pre to post DAFNE" 
• consider changing to "decreased following DAFNE." 

"The few graduates of ethnic minority in Ireland likely reflects the different population profile, but this may change over time; 12% of the Irish population in 2022 are non-Irish, representing an increase from previous years^21"
• given this point is referring to ethnicity, but the census data is referring to nationality, consider locating and using the census data on ethnicity instead. 
• if your preference is to retain original, consider adding the word citizens "non-Irish citizens" as based on my review that is what this reference is regarding (as opposed to non-Irish ethnicity). 

"However, data on those offered, but who do not take up, a DAFNE course, are not collected or submitted centrally, although centres are advised to keep these data locally. Without knowing the profile of people who do not accept, or are not offered, a course, it is difficult to fully understand the reach of the programme and how it can be enhanced." 
• This is well written and an important point. 
• The later reference [24] suggests it is possible to collect national data on other structured education - which raises the question of why not for DAFNE.
• Consider whether the strength of these statements may merit a recommendation to the effect that this data should be collated centrally if it we are to achieve the aspects of understanding "the reach of the programme and how it can be enhanced." Note this is an optional point, for the authors consideration only but as stated for a different point: ultimately it is users of these dataset whose feedback is most important in making improvements.   

Drop out data
"suggesting people who dropped out were engaged throughout most of the course" 
• I am not sure that attending 3 out of 5 days of a 5 day course provides any information on whether they were "engaged throughout most" i.e. it only provides attendance data, not engagement data.  With that in mind you should consider ammending or deleting "suggesting people who dropped out were engaged throughout most of the course."

"dropped out only on day/week four of their course .... non-completion in later stages, for example... expectation that the final session will not provide new material" 
• given that people are dropping out on day/week 4 as opposed to the the final session, consider changing "final session" to "latter sessions"
• It seems inappropriate to comment on what days people are non-attending without presenting any evidence regarding this. Given that you have this information and deemed it worthy of comment, and that it should be easy to tabulate, I would recommend that you present this information in the supplement. As I understand it at present the authors present the total number of days attended in Table 3, not which days were attended. If the authors preference is not to report this data then it may be best practice to remove their comment regarding this; however, I do think this data is interesting and valuable if substantiated! 

Missing data 
"Missing data  Overall missingness was low, with the exception of five variables where missingness was ≥ 5%"
• I would have expected that  the variables below would be included in this list; however, they are not presently (see Table legend) "
DKA: 200 (5.3%) missing pre-DAFNE, 1258 (33.4%) missing post-course; Hypos: 39 (1.0%) missing pre-course, 1139 (30.2%) missing post-course." the reasons for their omission are not clear to me. 

Conclusions 
"We found wide variation in course delivery across UK centre in particular."
• Add s to centres (plural)

"A higher proportion of people not completing remotely delivered courses than in-person courses,"
• Phrasing/wording. I suspect this would be easier read in the format:  A higher proportion did not complete remotely delivered as compared to in-person courses. 

Ethics and consent
• This section does not outline, whether the study/publication itself had ethics approval (from a named ethics committee/board), ethics waiver, or whether ethical approval deemed not required (with rationale) - however, I think it should at least provide a direct comment regarding this. 
• Note: I have no concerns regarding the ethical conduct of the study. 

Table/Figures 
Table 1: 
• Recommend listing all abbreviations in full in the legend (this has not presently been done here). 
• Recommend adding a column to list the p value result for each row (otherwise it is not clear whether the few reported values were prespecified analyses that also happened to have very low p values, or in fact all rows had a statistical test performed but there is selective reporting...). Note: this comment applies to ever tables where statistical testing has been performed. 


Table 2
• has the header of Mean (SD) however this does not apply to several of the variables e.g. "DKA requiring admission in past 12 months, N (%),    Severe hypo in last year, N (%)". in the interests of clarity I would suggest either Mean (SD) is removed as a subheading within the table and instead associated with the rows it applies to OR N (%) becomes an additinoal  subheading within the table. 
• I suspect that depending on data availibility/missingness there is considerable variability in the demoninator of fractions used to calculate the associated percentages. It is not presently clear for any of the tables whether the percentages supplied represent (presence/presence+absence) or (presence/presence+absence+missing). The most transparent way of doing this may be to include the denominator as a fraction within tables or have the denominators as a separate column for each row. 
• It is not completely clear to be whether the denominators on the left (before) and right (after) handside of the table are identical. I presume that they are but this should be explicit. i.e. state explicitly whether only individuals with both pre and post course data for an individual variable are included in the data row.   
• This is mentioned elsewhere in my review, but it should be stated whether this before/after linked data is inclusive of 2021 data. 

Table 4 - p value of < 0.0001 is used.
• If the authors preference is to exclusively report very small p values and nil others it would be helpful for the authors to denote why such a strigent value is used here for determining statistical significance (this could be in every table legend where this applies, or in the methods). Assuming p values are used to denote "no significant differences" vs "factors independently associated with non-completion", it would be helpful to know what threshold is being used by the authors. Additionally, it would be helpful for the authors to double check/confirm that that all of these values with "*" are in fact "< 0.0001".
• I would recommend that the authors consider simply adding a column to every table for which a statistical test was conducted and report each row's respective p value so that the reader can easily access this information without need to review the meaning of asterixes that vary table to table. The can then be enabled to make their own judgement about statistical significance e.g. if a reader felt that <0.05 or <0.01 or < 0.0001 was meaningful. 
• It is not clear to me why the demograpic variable Ethnic minority was not included in regression analyses. I suspect a reasonable agruement could be that it should be included in the univariable regression analysis at least. Please could the authors offer an explanation. If they feel the it does not merit inclusion, and the explaination does not merit inclusion in the manuscript - please list this in the response to review. 
• Suggestion: for ease of readers who are less familiar with reviewing results of regression models, I suspect it is helpful to add "univariable" and "multivariable" as part of the column header to label their respective odds ratios. In addition, I believe it is often useful to include "reference" inserted in the cells neigbouring 1) Gender (Female), 2) Course format (Standard DAFNE) and 3) Country (UK) to aid reader's understanding that the stastical test compares the reference to the test variable. 
• Please provide the relevant details of the model parameters in the legend. I am not a formally trained statistician but I beleive that these would include: total number of individuals, Akaike information criterion (AIC), c-statistic, Chi-squared value, and overall p value. 
• In terms of selected variables it is not clear to me why the variables "DKA in last 12 months requiring admission" and "Hypo in the last year unable to treat themselves" are being used in preference to DKA in the 12months before DAFNE course, or Hypo in the 12months before DAFNE course. A reason with this reasoning would be helpful. 


Supplementary 
Supplemental File 1: Overview of Dose Adjustment for Normal Easting (DAFN
• The inclusion of this is a welcome addition, and unfortunately often omitted from publications on DAFNE.

SupplementalFile 2: Data items requested from the extract
• It would be of help if it could be stated whether the definitions listed here are those use by DAFNE Central dataset, or your own definitions specific to your study. 
• It is not clear to me how the data requested in this extract was adequate to produce the results for post course data. In the interests of replicability I do think this is an important point that is addressed in the main text and potentially in this supplementary file if  applicable. 
For example the authors report: "Of those who had experienced a DKA requiring admission in the 12 months pre-course (n = 297), 4% (n = 12) experienced a DKA in the 12 months post-course. " 
However according to their data extract the only DKA data they requested was: "1) DKA total: Total number of DKA episodes requiring admission experienced by the patient since diagnosis DKA year*, 2) Number of DKA episodes requiring admission experienced by the patient in the 12 months before the course"
Neither of these numbers would provide the data presented - this  requires  clarification (e.g. perhaps the provided definition is incorrect and the data extract would include the dates of DKA, or alternatively were their additinoal datare quests that are not outlined here, or other explaination). 
Note this equally applies to the data on hypoglycaemia. 
It is possible that I have missed something obvious here, which the authors may identify, but despite a few reads I have not found adequate clarification. 

Supplementary Table 1Completion rates(MDI or pump) by course type (N = 6586)
• The legend includes the following: "Chi2 P = 0.000"
However, to my knowledge, this has omitted several aspects of the standard reporting of this test listed here in bold i.e. Χ2 (degrees of freedom, N = sample size) = chi-square value, p = p value).

STROBE reporting checklist for cross sectional studies.
• The inclusion of this is a welcome addition. 
• Regarding "follow-up": In the methods it is clear you are looking at "courses delivered between 01.01.2019 and 01.01.2022". However for a substantial proportion, 12 months will not have passed to enable them to be eligible for post-course events at 12 months, given that  " At the time of the extract [Q1 2022]..." is very close to the 2021 data.  Please include in methods the approached used to censoring these individuals (i.e. who were not at risk of 12 month post course events) and/or if you did not censor these individuals provide an associated explaination. Note clarifying this aspect, may directly relate to misssingness data (see related comments).