Nutritional Contribution and Quality of Lunches Consumed During School Lunch Periods in Canadian Elementary Schools: A Plate Waste Analysis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript addresses a topic of significance in public health nutrition, particularly in the context of the advancement of Canada’s National School Food Program. The use of a digital photography–assisted plate waste method to assess the actual intake of elementary students during school lunch periods has a certain degree of innovation and policy value. However, the current version still has relatively evident problems in the definition of the study population, the classification of lunch provenance, the calculation of the NRF 9.3 Index, the statistical analysis strategy, the boundaries of result interpretation, and table presentation. Some of these issues may affect the credibility of the main conclusions.
1. The manuscript repeatedly uses “school lunches,” but this term sometimes refers to “lunches consumed at school” and at other times may be easily understood as “school-provided lunches.” It is recommended that the manuscript consistently distinguish between “lunches consumed during school lunch periods” and “school-provided lunches” throughout the text; otherwise, readers’ understanding of the comparative results for home-packed, school-provided, and mixed lunches may be affected.
2. The background in the Introduction section is relatively brief and does not yet sufficiently establish the research gap or the necessity of the study. The authors state that Canadian evidence describing what children actually consume during the school day remains limited and emphasize the lack of objective dietary assessment methods, which is reasonable. However, the manuscript currently lacks a systematic comparison of existing studies on Canadian school meal, home-packed lunch, and school-hour dietary intake, and it does not clearly explain the incremental contribution of the present study in terms of methods, region, sample, or indicators compared with previous research. It is recommended that the final paragraph of the Introduction clearly state the primary and secondary objectives of the study, for example, by specifying that the study aims to assess nutrient content, nutrient density, and contributions to age- and sex-specific dietary recommendations during school lunch periods, and to compare differences by lunch provenance, age group, and sex/gender. The current objective statement is too broad and does not prospectively explain the analytical logic underlying the large number of subgroup comparisons presented later.
3. The study design and sample selection require greater transparency. The manuscript states that the four schools were purposively selected, were all located in lower socioeconomic neighbourhoods, and served ethnically diverse student populations, including higher proportions of Indigenous and newcomer children. However, it does not provide the total number of eligible students, the number of opt-out students, the number actually observed, the number excluded, reasons for missing data, or a flow diagram showing the final analytic sample. Because the study used passive informed consent and verbal assent, and because plate waste data were collected only on specific observation days, both selection bias and attendance bias may exist. It is recommended that the authors add a STROBE-style participant flow diagram and report the number of included participants and missing data by school, grade, and lunch provenance. If individual-level socioeconomic or ethnicity data are unavailable, the authors should avoid making overly strong inferences regarding ethnically diverse populations or equity implications in the Results and Discussion.
4. The operational details of the digital photography–assisted plate waste method are insufficient, which limits reproducibility. The authors state that foods and beverages were weighed and photographed before and after consumption, but they do not specify the model, precision, or calibration procedures of the weighing equipment; the photography angle; whether a standardized reference object was used; how packaging weight was deducted; how liquids and semi-solid foods were handled; how mixed dishes were disaggregated; how homemade foods without recipes were estimated; how uneaten food taken away by students was handled; or the actual frequency of food sharing, spillage, and discarding. It is recommended that the Methods section provide a complete protocol, explain how field notes were used to revise intake estimates, and report consistency indicators such as inter-rater reliability or independent verification, rather than merely stating that discrepancies were resolved through discussion. Only with these details can the authors’ methodological claims regarding objective estimates and reduced reporting bias be supported.
5. The classification of lunch provenance is one of the issues that most requires clarification in the current manuscript. The authors define school-provided, home-packed, and mixed lunches, and also state that for students with two observed lunch days, nutrient values from both observations were averaged at the participant level prior to analysis. However, if the same student had different lunch provenance across the two days, for example, home-packed on one day and school-provided on another, the authors do not explain how the participant-level lunch type was determined. In addition, Table 1 reports Type of Lunch as n (%) at the level of 379 students, whereas the main text states that school-provided lunches accounted for 38.0% of observations, creating confusion regarding the unit of analysis. The authors should clarify whether the analytic unit is the lunch observation or the participant. A more appropriate approach would be to retain the observation-level data and use mixed-effects models or generalized estimating equations to account for repeated measures, while also considering student and school as hierarchical levels. If the authors insist on participant-level averaging, they must explain the assignment rule for lunch provenance and conduct sensitivity analyses for participants with inconsistent provenance across days.
6. The data sources and matching process for the nutritional analysis require more detail. The manuscript states that Food Processor Nutrition and Fitness Software, the Canadian Nutrient File (CNF), and the USDA database were used, but it does not specify the exact version dates of the CNF and USDA database, how branded foods and school recipes were matched, how missing nutrient values were handled, whether school recipes were adjusted for cooked yield or edible portion, or whether nutrition labels were used for common processed foods in home-packed foods. Because indicators such as vitamin D, total sugar, and sodium are highly sensitive to database selection and brand-level differences, it is recommended that the authors supplement the food coding hierarchy and report the error range from random-sample verification or double coding. For school-provided meals, the authors should explain whether recipes provided by school staff included ingredient weight, serving size, cooking method, and fortification information.
7. The calculation and interpretation of DRIs contribution need to be more rigorous. The authors state that EAR, AI, and the CDRR threshold were used, and they emphasize that these comparisons do not represent overall dietary adequacy, which is appropriate. However, Table 4 also presents sodium under “Contribution to daily nutrient recommendations (% of DRI),” which may lead readers to mistakenly assume that a higher sodium contribution is better. In fact, sodium was compared with the Chronic Disease Risk Reduction (CDRR) threshold and should be separately labeled as “% of CDRR threshold” or “% of recommended maximum.” The Results and Discussion should clearly state that a single lunch reaching approximately 49%–55% of the CDRR threshold for sodium may indicate a relatively high sodium burden, rather than a favorable nutritional contribution. In addition, Table 4 should specify whether EAR or AI was used for calcium, potassium, vitamin D, and other nutrients. It is recommended that the supplementary materials list the corresponding reference values for each age and sex/gender group.
8. The interpretation of NRF 9.3 per 2000 kcal standardization also requires caution. The mean energy of school-provided lunches was clearly lower than that of home-packed and mixed lunches, whereas the NRF 9.3 is a nutrient density indicator. A lower-energy meal may obtain a higher score after energy standardization if it contains certain micronutrients. Therefore, the higher NRF 9.3 scores of school-provided lunches do not necessarily mean that their absolute nutrient contribution is more adequate. In fact, Table 3 shows that school-provided lunches were lower in folate, iron, and vitamin C than home-packed or mixed lunches, and Table 4 also shows lower folate, iron, and vitamin C contribution for school-provided lunches. The Discussion should clearly distinguish between “nutrient density” and “absolute nutrient adequacy” and avoid directly interpreting higher NRF 9.3 scores as indicating better overall nutritional provision.
9. The current statistical analysis is overly simple and is insufficient to support strong comparative conclusions. The authors used independent-sample t-tests, one-way ANOVA, Mann–Whitney U, and Kruskal–Wallis tests, but the data structure includes at least repeated observations within students and students nested within schools. Even if repeated observations were averaged by the authors, school-level clustering may still substantially affect standard errors and p-values, particularly because school-provided lunches may be determined by school menus and therefore may have an obvious cluster effect. It is recommended that the authors use multivariable regression or mixed-effects models, treating school as a random effect or fixed effect, and adjust for age group, sex/gender, number of lunch observations, and lunch provenance. If the number of schools is only four and a random effect is unstable, school fixed effects or cluster-robust standard errors should at least be used, and the inability to adequately account for clustering should be acknowledged as a limitation.
10. The issue of multiple comparisons needs to be addressed. The manuscript conducts a large number of significance tests involving 18 nutrients, energy, NRF 9.3, sex, age group, age-sex subgroup, and lunch type. However, except for some post-hoc Dunn’s test with Bonferroni correction, it does not explain how the overall multiple testing issue was controlled. If each item is judged using p < 0.05, the risk of type I error is high, particularly for the sex differences in total fiber and magnesium in Table 2 and the age differences in multiple nutrient contributions in Supplemental Table 1. It is recommended that these analyses be clearly defined as exploratory analyses and that false discovery rate or Bonferroni-Holm correction be applied, or that adjusted p-values at least be reported in the Results. Otherwise, the emphasis on marginally significant findings should be reduced.
11. The presentation of post-hoc pairwise comparisons in the Results section is insufficient. The text repeatedly states that “school-provided lunches contributed significantly less energy than both home-packed and mixed lunches” or that “school-provided and mixed lunches generally provided higher amounts,” but Table 3 only provides the overall Kruskal–Wallis p-value and does not show specific pairwise adjusted p-values or letter annotations. It is recommended that post-hoc comparison columns be added to Table 3 and Table 4, or that superscript letters be used to indicate the specific differences among school-provided, home-packed, and mixed lunches. This would prevent readers from being unable to determine which between-group comparisons actually produced statistically significant differences.
12. The overall direction of the Discussion is reasonable, but the causal and policy implications are overstated. This study is a descriptive quantitative analysis of baseline data; the schools were purposively selected, and the school-provided lunches came from small, self-managed lunch programs. Therefore, it cannot be inferred that school food programs themselves caused higher nutrient density, nor can the findings directly represent the potential effects of Canada’s National School Food Program. It is recommended that the statement “highlighting the potential of school food programs to improve diet quality” be revised to a more cautious expression, such as “suggesting that, in this sample, school-provided lunches were associated with higher nutrient density.” At the same time, the authors should emphasize that although school-provided lunches had higher NRF 9.3 scores, they were lower in energy, carbohydrates, folate, iron, and vitamin C, suggesting that improvements may be needed in portion size and the provision of fruit/vegetable/whole grain/iron-rich food.
13. The interpretation regarding older students requires further quantification. The authors state that older children showed lower proportional contributions relative to their higher nutritional requirements, which is reasonable. However, this has not yet been analyzed in relation to actual lunch amount, portion size, or energy intake. If the absolute intake of older students was similar to that of younger students, this would suggest that meal planning failed to adjust for age-related requirements; if older students consumed more but DRIs increased even more, the interpretation would differ. It is recommended that the authors add a table of absolute energy and nutrient intake by age group, or provide both absolute intake and % DRI in the Supplemental Table, to support the recommendation for “age-responsive meal planning.”
14. The impact of the COVID-19 pandemic on the data requires more in-depth discussion. Data were collected in May and June 2021, a period during which school meal provision, household food access, student attendance, restrictions on food sharing, use of packaged foods, and school hygiene policies may all have been abnormal. The authors only briefly mention this in the limitations, which is clearly insufficient. It is recommended that the authors provide additional information on whether pandemic-related restrictions were in place in the schools at that time, whether school lunch programs were affected by supply chain or donation changes, whether students were restricted from sharing food or using communal utensils, and how these factors may have influenced plate waste and lunch provenance. Because the study is used to support policy discussions after the 2024 National School Food Program, the temporal context must be handled with particular caution.
15. The description of Sex and Gender-based Analysis needs to be more accurate. The authors acknowledge that DRIs are based on biological sex, but the available data were self-identified gender as reported by the schools, which was used as a proxy for sex. This is a reasonable but sensitive approach. It is recommended that “Sex” in all tables be revised to “reported gender used as proxy for sex-specific DRIs,” or that this be clearly stated in table notes. If only Female/Male categories were available, the authors should explain whether there were any non-binary, missing, or undisclosed categories, and how these students were handled. Otherwise, directly presenting “Sex Females/Males” may be inconsistent with the methodological explanation.
16. There are obvious errors in the references, and they need to be checked one by one. In the Methods section, the manuscript states that the analysis used pre-intervention baseline data from a larger population health intervention research project and cites [9] in that sentence, but reference [9] is Willett’s Nutritional Epidemiology, whereas the protocol paper is [10]. This citation mismatch affects the rigor of the manuscript and must be corrected. The authors should comprehensively check the correspondence between in-text citations and the References, especially [9], [10], the evidence supporting modification of the NRF 9.3, and the references for DRIs and Health Canada Daily Values.
17. In terms of language and formatting, the manuscript is generally readable but still requires professional polishing.

Author Response

We sincerely thank Reviewer 1 for the thorough and thoughtful evaluation of our manuscript. We appreciate the detailed comments regarding study design, methodological transparency, statistical analyses, interpretation of findings, and presentation of results. The suggestions provided were extremely valuable and have helped us improve the clarity, rigor, and overall quality of the manuscript. We have carefully considered each comment and revised the manuscript accordingly. Detailed responses to each point are provided below.

 

1. The manuscript repeatedly uses “school lunches,” but this term sometimes refers to “lunches consumed at school” and at other times may be easily understood as “school-provided lunches.” It is recommended that the manuscript consistently distinguish between “lunches consumed during school lunch periods” and “school-provided lunches” throughout the text; otherwise, readers’ understanding of the comparative results for home-packed, school-provided, and mixed lunches may be affected.

R: Thank you for this observation. We have revised the manuscript to consistently distinguish between lunches consumed during school lunch periods and school-provided lunches. Terminology has been clarified throughout the manuscript to improve readability and avoid ambiguity when comparing home-packed, school-provided, and mixed lunches.

2. The background in the Introduction section is relatively brief and does not yet sufficiently establish the research gap or the necessity of the study. The authors state that Canadian evidence describing what children actually consume during the school day remains limited and emphasize the lack of objective dietary assessment methods, which is reasonable. However, the manuscript currently lacks a systematic comparison of existing studies on Canadian school meal, home-packed lunch, and school-hour dietary intake, and it does not clearly explain the incremental contribution of the present study in terms of methods, region, sample, or indicators compared with previous research. It is recommended that the final paragraph of the Introduction clearly state the primary and secondary objectives of the study, for example, by specifying that the study aims to assess nutrient content, nutrient density, and contributions to age- and sex-specific dietary recommendations during school lunch periods, and to compare differences by lunch provenance, age group, and sex/gender. The current objective statement is too broad and does not prospectively explain the analytical logic underlying the large number of subgroup comparisons presented later.

R: Thank you for this suggestion. We have revised the Introduction to better articulate the research gap and the contribution of the present study within the Canadian context. We have also expanded the study objectives in the final paragraph of the Introduction to explicitly state that the study aimed to assess nutrient content, nutrient density, and contributions to age- and sex-specific dietary recommendations among lunches consumed during school lunch periods, and to examine differences by lunch provenance, age group, and reported gender.

3.The study design and sample selection require greater transparency. The manuscript states that the four schools were purposively selected, were all located in lower socioeconomic neighbourhoods, and served ethnically diverse student populations, including higher proportions of Indigenous and newcomer children. However, it does not provide the total number of eligible students, the number of opt-out students, the number actually observed, the number excluded, reasons for missing data, or a flow diagram showing the final analytic sample. Because the study used passive informed consent and verbal assent, and because plate waste data were collected only on specific observation days, both selection bias and attendance bias may exist. It is recommended that the authors add a STROBE-style participant flow diagram and report the number of included participants and missing data by school, grade, and lunch provenance. If individual-level socioeconomic or ethnicity data are unavailable, the authors should avoid making overly strong inferences regarding ethnically diverse populations or equity implications in the Results and Discussion.

R: Thank you for this observation. We have clarified participant inclusion procedures in the Methods section and acknowledged the potential for selection and attendance bias in the limitations section. Individual-level socioeconomic and ethnicity data were not available; therefore, references to lower-income and ethnically diverse populations have been framed as characteristics of the participating school communities rather than individual participants.

4. The operational details of the digital photography–assisted plate waste method are insufficient, which limits reproducibility. The authors state that foods and beverages were weighed and photographed before and after consumption, but they do not specify the model, precision, or calibration procedures of the weighing equipment; the photography angle; whether a standardized reference object was used; how packaging weight was deducted; how liquids and semi-solid foods were handled; how mixed dishes were disaggregated; how homemade foods without recipes were estimated; how uneaten food taken away by students was handled; or the actual frequency of food sharing, spillage, and discarding. It is recommended that the Methods section provide a complete protocol, explain how field notes were used to revise intake estimates, and report consistency indicators such as inter-rater reliability or independent verification, rather than merely stating that discrepancies were resolved through discussion. Only with these details can the authors’ methodological claims regarding objective estimates and reduced reporting bias be supported.

R: Thank you for this thoughtful observation. We have expanded the Methods section to provide additional detail regarding the digital photography–assisted plate waste protocol. Specifically, we now describe the use of digital kitchen scales, standardized photographic procedures, the collection of pre- and post-consumption weights, the use of participant identification labels, the recording of field notes, and procedures used to account for food containers and packaging. We have also clarified how field notes were used to assist with food identification, portion estimation, and interpretation of photographs during data entry and cleaning. All dietary entries were independently verified by a second researcher, and discrepancies were resolved through discussion. Formal inter-rater reliability statistics were not calculated and are therefore not available for reporting.

5. The classification of lunch provenance is one of the issues that most requires clarification in the current manuscript. The authors define school-provided, home-packed, and mixed lunches, and also state that for students with two observed lunch days, nutrient values from both observations were averaged at the participant level prior to analysis. However, if the same student had different lunch provenance across the two days, for example, home-packed on one day and school-provided on another, the authors do not explain how the participant-level lunch type was determined. In addition, Table 1 reports Type of Lunch as n (%) at the level of 379 students, whereas the main text states that school-provided lunches accounted for 38.0% of observations, creating confusion regarding the unit of analysis. The authors should clarify whether the analytic unit is the lunch observation or the participant. A more appropriate approach would be to retain the observation-level data and use mixed-effects models or generalized estimating equations to account for repeated measures, while also considering student and school as hierarchical levels. If the authors insist on participant-level averaging, they must explain the assignment rule for lunch provenance and conduct sensitivity analyses for participants with inconsistent provenance across days.

R: Thank you for this important observation. We have clarified the lunch classification procedure in the Methods section. Nutrient values from two observed lunch days were averaged at the participant level prior to analysis, and lunch provenance was assigned based on the source(s) contributing to the averaged intake. Students whose lunches originated from different sources across observation days were classified as mixed. We have also revised the Results section and Table 1 to clarify that lunch provenance categories are reported at the participant level.

The participant was selected as the unit of analysis because the primary objective was to characterize usual lunch intake patterns while avoiding non-independence arising from repeated observations of the same student. We acknowledge that alternative analytical approaches, such as mixed-effects models or generalized estimating equations, could be used when retaining observation-level data. However, given the descriptive objectives of the study and the limited number of repeated observations per student, participant-level averaging was considered an appropriate and parsimonious approach. The manuscript has been revised to improve transparency regarding these analytical decisions.

6. The data sources and matching process for the nutritional analysis require more detail. The manuscript states that Food Processor Nutrition and Fitness Software, the Canadian Nutrient File (CNF), and the USDA database were used, but it does not specify the exact version dates of the CNF and USDA database, how branded foods and school recipes were matched, how missing nutrient values were handled, whether school recipes were adjusted for cooked yield or edible portion, or whether nutrition labels were used for common processed foods in home-packed foods. Because indicators such as vitamin D, total sugar, and sodium are highly sensitive to database selection and brand-level differences, it is recommended that the authors supplement the food coding hierarchy and report the error range from random-sample verification or double coding. For school-provided meals, the authors should explain whether recipes provided by school staff included ingredient weight, serving size, cooking method, and fortification information.

R: Thank you for this suggestion. We have expanded the Methods section to provide additional information regarding nutrient analysis procedures and food coding. Nutrient analysis was conducted using Food Processor Nutrition and Fitness Software (version 11.6.522). Food items were matched using photographs, field notes, observed ingredients, and available recipe information. The Canadian Nutrient File was used as the primary source for food matching, while USDA food composition database entries were used when more appropriate matches were available, particularly for fruits and vegetables. Brand-specific products were not routinely coded; instead, nutritionally comparable generic food items were selected. School-provided foods were coded using recipes and ingredient information supplied by school food service staff when available, while home-prepared foods were coded using observed ingredients and standard recipes available within the software. Formal double-coding procedures and coding error estimates were not conducted and are therefore not available for reporting.

 

7. The calculation and interpretation of DRIs contribution need to be more rigorous. The authors state that EAR, AI, and the CDRR threshold were used, and they emphasize that these comparisons do not represent overall dietary adequacy, which is appropriate. However, Table 4 also presents sodium under “Contribution to daily nutrient recommendations (% of DRI),” which may lead readers to mistakenly assume that a higher sodium contribution is better. In fact, sodium was compared with the Chronic Disease Risk Reduction (CDRR) threshold and should be separately labeled as “% of CDRR threshold” or “% of recommended maximum.” The Results and Discussion should clearly state that a single lunch reaching approximately 49%–55% of the CDRR threshold for sodium may indicate a relatively high sodium burden, rather than a favorable nutritional contribution. In addition, Table 4 should specify whether EAR or AI was used for calcium, potassium, vitamin D, and other nutrients. It is recommended that the supplementary materials list the corresponding reference values for each age and sex/gender group.

R: Thank you for this helpful observation. We agree that sodium differs conceptually from nutrients evaluated against EARs or AIs because it was compared with the Chronic Disease Risk Reduction (CDRR) threshold rather than a recommended intake target. We have revised Table 4 to identify sodium as a percentage of the CDRR threshold and added explanatory text in the Results and Discussion clarifying that higher sodium values indicate a greater contribution toward the recommended maximum intake rather than a favourable nutritional contribution. We have also expanded the table notes and supplementary materials to specify the reference values and reference types (EAR, AI, or CDRR) used for each nutrient and age/sex group.

8. The interpretation of NRF 9.3 per 2000 kcal standardization also requires caution. The mean energy of school-provided lunches was clearly lower than that of home-packed and mixed lunches, whereas the NRF 9.3 is a nutrient density indicator. A lower-energy meal may obtain a higher score after energy standardization if it contains certain micronutrients. Therefore, the higher NRF 9.3 scores of school-provided lunches do not necessarily mean that their absolute nutrient contribution is more adequate. In fact, Table 3 shows that school-provided lunches were lower in folate, iron, and vitamin C than home-packed or mixed lunches, and Table 4 also shows lower folate, iron, and vitamin C contribution for school-provided lunches. The Discussion should clearly distinguish between “nutrient density” and “absolute nutrient adequacy” and avoid directly interpreting higher NRF 9.3 scores as indicating better overall nutritional provision.

R: Thank you for this important observation. We agree that the NRF 9.3 Index is a measure of nutrient density rather than absolute nutrient content or adequacy. We have revised the Discussion to clarify that the higher NRF 9.3 scores observed for school-provided lunches reflect a more favourable nutrient profile relative to energy content and should not be interpreted as indicating greater absolute nutrient provision. We also acknowledge that school-provided lunches contributed lower amounts of some nutrients, including folate, iron, and vitamin C, compared with home-packed and mixed lunches, despite achieving higher nutrient density scores.

9. The current statistical analysis is overly simple and is insufficient to support strong comparative conclusions. The authors used independent-sample t-tests, one-way ANOVA, Mann–Whitney U, and Kruskal–Wallis tests, but the data structure includes at least repeated observations within students and students nested within schools. Even if repeated observations were averaged by the authors, school-level clustering may still substantially affect standard errors and p-values, particularly because school-provided lunches may be determined by school menus and therefore may have an obvious cluster effect. It is recommended that the authors use multivariable regression or mixed-effects models, treating school as a random effect or fixed effect, and adjust for age group, sex/gender, number of lunch observations, and lunch provenance. If the number of schools is only four and a random effect is unstable, school fixed effects or cluster-robust standard errors should at least be used, and the inability to adequately account for clustering should be acknowledged as a limitation.

R: Thank you for this important comment. We acknowledge the potential for school-level clustering because students were nested within four participating schools. As a sensitivity analysis, we fitted linear regression models for NRF 9.3 scores that included lunch provenance, age group, reported gender, and school fixed effects. The association between lunch provenance and NRF 9.3 remained statistically significant after adjustment for school, and the direction and magnitude of the observed differences were unchanged. These findings suggest that school-level clustering did not materially alter the primary conclusions regarding nutrient density. Given the descriptive objectives of the study, the participant-level averaging of repeated observations, and the small number of participating schools (n = 4), we retained the primary descriptive analyses and have added text acknowledging the potential influence of clustering as a study limitation.

 

10. The issue of multiple comparisons needs to be addressed. The manuscript conducts a large number of significance tests involving 18 nutrients, energy, NRF 9.3, sex, age group, age-sex subgroup, and lunch type. However, except for some post-hoc Dunn’s test with Bonferroni correction, it does not explain how the overall multiple testing issue was controlled. If each item is judged using p < 0.05, the risk of type I error is high, particularly for the sex differences in total fiber and magnesium in Table 2 and the age differences in multiple nutrient contributions in Supplemental Table 1. It is recommended that these analyses be clearly defined as exploratory analyses and that false discovery rate or Bonferroni-Holm correction be applied, or that adjusted p-values at least be reported in the Results. Otherwise, the emphasis on marginally significant findings should be reduced.

R: Thank you for this important observation. We acknowledge that multiple statistical comparisons were conducted and that this increases the possibility of Type I error. The primary purpose of this study was descriptive and exploratory rather than hypothesis-testing. We have clarified this in the Statistical Analysis section and have interpreted statistically significant findings cautiously, placing greater emphasis on overall patterns rather than isolated marginal associations. In addition, Bonferroni-adjusted post-hoc comparisons were used where appropriate following significant overall tests.

11. The presentation of post-hoc pairwise comparisons in the Results section is insufficient. The text repeatedly states that “school-provided lunches contributed significantly less energy than both home-packed and mixed lunches” or that “school-provided and mixed lunches generally provided higher amounts,” but Table 3 only provides the overall Kruskal–Wallis p-value and does not show specific pairwise adjusted p-values or letter annotations. It is recommended that post-hoc comparison columns be added to Table 3 and Table 4, or that superscript letters be used to indicate the specific differences among school-provided, home-packed, and mixed lunches. This would prevent readers from being unable to determine which between-group comparisons actually produced statistically significant differences.

R: Thank you for this suggestion. We agree that reporting only the overall Kruskal–Wallis test limits interpretation of between-group differences. To improve clarity, we have added superscript letter annotations to Tables 3 and 4 to indicate statistically significant pairwise differences between lunch provenance groups based on Dunn’s post-hoc tests with Bonferroni correction.

 

12. The overall direction of the Discussion is reasonable, but the causal and policy implications are overstated. This study is a descriptive quantitative analysis of baseline data; the schools were purposively selected, and the school-provided lunches came from small, self-managed lunch programs. Therefore, it cannot be inferred that school food programs themselves caused higher nutrient density, nor can the findings directly represent the potential effects of Canada’s National School Food Program. It is recommended that the statement “highlighting the potential of school food programs to improve diet quality” be revised to a more cautious expression, such as “suggesting that, in this sample, school-provided lunches were associated with higher nutrient density.” At the same time, the authors should emphasize that although school-provided lunches had higher NRF 9.3 scores, they were lower in energy, carbohydrates, folate, iron, and vitamin C, suggesting that improvements may be needed in portion size and the provision of fruit/vegetable/whole grain/iron-rich food.

R: Thank you for this important observation. We agree that the descriptive design of this study does not permit causal inference regarding the effects of school food programs, nor do the findings directly represent the potential impact of Canada's National School Food Program. We have revised the manuscript to use more cautious language and to emphasize that school-provided lunches were associated with higher nutrient density within this sample rather than concluding that school food programs caused improved diet quality. We have also expanded the Discussion to acknowledge that, despite higher NRF 9.3 scores, school-provided lunches provided lower amounts of some nutrients, including folate, iron, and vitamin C, as well as lower energy content, suggesting opportunities for continued improvement in meal composition and portion size.

13. The interpretation regarding older students requires further quantification. The authors state that older children showed lower proportional contributions relative to their higher nutritional requirements, which is reasonable. However, this has not yet been analyzed in relation to actual lunch amount, portion size, or energy intake. If the absolute intake of older students was similar to that of younger students, this would suggest that meal planning failed to adjust for age-related requirements; if older students consumed more but DRIs increased even more, the interpretation would differ. It is recommended that the authors add a table of absolute energy and nutrient intake by age group, or provide both absolute intake and % DRI in the Supplemental Table, to support the recommendation for “age-responsive meal planning.”

R: Thank you for this helpful observation. We agree that lower proportional nutrient contributions among older students should be interpreted in the context of both absolute intake and age-specific dietary requirements. We have added a supplementary table presenting absolute energy and nutrient intakes by age group. Older students consumed greater amounts of energy and several nutrients than younger students; however, these increases were modest relative to the substantially higher dietary reference values applicable to older children. For some nutrients, including calcium and vitamin D, absolute intakes were similar across age groups despite higher requirements among older students. We have revised the Discussion accordingly.

14. The impact of the COVID-19 pandemic on the data requires more in-depth discussion. Data were collected in May and June 2021, a period during which school meal provision, household food access, student attendance, restrictions on food sharing, use of packaged foods, and school hygiene policies may all have been abnormal. The authors only briefly mention this in the limitations, which is clearly insufficient. It is recommended that the authors provide additional information on whether pandemic-related restrictions were in place in the schools at that time, whether school lunch programs were affected by supply chain or donation changes, whether students were restricted from sharing food or using communal utensils, and how these factors may have influenced plate waste and lunch provenance. Because the study is used to support policy discussions after the 2024 National School Food Program, the temporal context must be handled with particular caution.

 

R: Thank you for this important observation. We have expanded the limitations section to provide additional context regarding the COVID-19 public health measures in place during data collection. These measures included restrictions on food sharing and modifications to school food service operations that often favoured individually packaged foods and simplified meal service models. We now acknowledge that these conditions may have influenced both school-provided and home-packed lunches, as well as the comparability of findings with non-pandemic school environments. We have also clarified that the results should be interpreted within the context of pandemic-era school food practices and may not fully reflect current school food environments.

15. The description of Sex and Gender-based Analysis needs to be more accurate. The authors acknowledge that DRIs are based on biological sex, but the available data were self-identified gender as reported by the schools, which was used as a proxy for sex. This is a reasonable but sensitive approach. It is recommended that “Sex” in all tables be revised to “reported gender used as proxy for sex-specific DRIs,” or that this be clearly stated in table notes. If only Female/Male categories were available, the authors should explain whether there were any non-binary, missing, or undisclosed categories, and how these students were handled. Otherwise, directly presenting “Sex Females/Males” may be inconsistent with the methodological explanation.

R: We have expanded the Sex and Gender-Based Analysis section to clarify that reported gender information provided by schools was used to assign sex-specific DRI reference values. Two participants had missing gender information and were classified using available demographic information for the purposes of DRI assignment. Given that this represented less than 1% of the sample, it is unlikely to have materially influenced the findings.

16. There are obvious errors in the references, and they need to be checked one by one. In the Methods section, the manuscript states that the analysis used pre-intervention baseline data from a larger population health intervention research project and cites [9] in that sentence, but reference [9] is Willett’s Nutritional Epidemiology, whereas the protocol paper is [10]. This citation mismatch affects the rigor of the manuscript and must be corrected. The authors should comprehensively check the correspondence between in-text citations and the References, especially [9], [10], the evidence supporting modification of the NRF 9.3, and the references for DRIs and Health Canada Daily Values.

R: Thank you for identifying this error. The incorrect citation has been corrected, and we have reviewed all in-text citations and references to ensure consistency and accuracy throughout the manuscript.

17. In terms of language and formatting, the manuscript is generally readable but still requires professional polishing.

R: Thank you for this comment. The manuscript has been carefully reviewed and edited to improve clarity, consistency, and readability. We have also standardized terminology, formatting, and reporting throughout the manuscript.

 

 

Reviewer 2 Report

Comments and Suggestions for Authors

The study addresses a critical gap in Canadian public health: the actual quality of what children consume at school. The primary contribution is the use of an objective method (plate waste analysis) as opposed to 24-hour recalls, demonstrating that school-provided lunches are nutritionally superior to those brought from home (NRF 9.3 of 378.8 vs. 228.7). This is a timely piece of work to inform the implementation of the new National School Food Program.

The identified points for improvement are:

• Representativeness and Selection Bias: The study was conducted in schools within low-income neighborhoods with high ethnic diversity. The authors must discuss more explicitly how this limits the generalizability of the results to other socioeconomic contexts in Canada.

• Macronutrient Analysis: It is mentioned that protein contributed 14% of calories across all groups, but there is no discussion on whether this percentage aligns with the Acceptable Macronutrient Distribution Ranges (AMDR) for this age group.

• Impact of COVID-19: Data collection took place in May-June 2021. A deeper discussion is required regarding how pandemic restrictions specifically affected the types of food allowed in schools at that time (e.g., pre-packaged vs. cooked foods).

• Clarification of the NRF 9.3: Although the substitution of Vitamin E with Vitamin D is justified, it would be beneficial to include a brief mention in the main text on how this modification might bias comparisons with international studies using the standard index.

• Abbreviations: Ensure that terms such as "CDRR" (Chronic Disease Risk Reduction) are clearly defined at their first mention in the text.

• Suggestion: It is suggested to add a supplementary table or paragraph comparing the results with Canadian AMDRs for fats and carbohydrates, not just the EAR/AI for micronutrients.

• Limitation Extension: It is suggested to expand the limitations section by acknowledging that the data represents a "snapshot in time" (two days) and not necessarily long-term habitual intake.

Author Response

We sincerely thank Reviewer 2 for the careful review of our manuscript and for the constructive comments and suggestions. We appreciate the positive assessment of the study and its potential contribution to informing school food policy in Canada. The feedback provided helped strengthen the manuscript, particularly with respect to the interpretation of findings, contextualization of results, and discussion of study limitations. Detailed responses to each comment are provided below.

1. Representativeness and Selection Bias: The study was conducted in schools within low-income neighborhoods with high ethnic diversity. The authors must discuss more explicitly how this limits the generalizability of the results to other socioeconomic contexts in Canada.

R: Thank you for this important observation. We have expanded the Discussion and Limitations sections to clarify that the participating schools were purposively selected from lower-income, ethnically diverse neighbourhoods and therefore may not be representative of all Canadian elementary schools. Findings should be interpreted within this context, as lunch patterns, access to school food programs, and nutritional intake may differ in higher-income or less diverse settings.

2. Macronutrient Analysis: It is mentioned that protein contributed 14% of calories across all groups, but there is no discussion on whether this percentage aligns with the Acceptable Macronutrient Distribution Ranges (AMDR) for this age group.

R: We appreciate this suggestion. We have added a brief discussion noting that protein contributed approximately 14% of lunch energy, which falls within the AMDR for children aged 4–18 years (10–30% of total energy intake).

 3. Impact of COVID-19: Data collection took place in May-June 2021. A deeper discussion is required regarding how pandemic restrictions specifically affected the types of food allowed in schools at that time (e.g., pre-packaged vs. cooked foods).

R: Thank you. We have expanded the limitations section to describe how COVID-19 public health measures operating during data collection may have influenced lunch practices. Measures designed to reduce food handling and sharing often favoured individually packaged foods and modified food service operations, which may have affected both the nutritional composition of lunches and the functioning of school-based meal programs.

 

4. Clarification of the NRF 9.3: Although the substitution of Vitamin E with Vitamin D is justified, it would be beneficial to include a brief mention in the main text on how this modification might bias comparisons with international studies using the standard index.

R: Thank you for this suggestion. We have clarified in the Methods section that the substitution of vitamin D for vitamin E may limit direct comparability with studies using the original NRF 9.3 formulation and that comparisons of NRF 9.3 scores across studies should therefore be interpreted with caution.

 5. Abbreviations: Ensure that terms such as "CDRR" (Chronic Disease Risk Reduction) are clearly defined at their first mention in the text.

R: Thank you. The term Chronic Disease Risk Reduction (CDRR) has now been defined at first mention in the text and included in the Abbreviations section.

6. Suggestion: It is suggested to add a supplementary table or paragraph comparing the results with Canadian AMDRs for fats and carbohydrates, not just the EAR/AI for micronutrients.

R: We agree and have added a brief comparison of macronutrient energy contributions with the AMDRs for children and adolescents in the Discussion section.

7. Limitation Extension: It is suggested to expand the limitations section by acknowledging that the data represents a "snapshot in time" (two days) and not necessarily long-term habitual intake.

R: Thank you. We have expanded the limitations section to clarify that lunch observations were based on one or two school days and therefore represent a snapshot of intake that may not reflect habitual dietary intake.

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Overall, the response letter is positive in tone, and the authors have indeed made several textual, explanatory, and tabular revisions. However, there are still some issues that need to be corrected.
1. Regarding terminology, the authors have now used “foods and beverages consumed during school lunch periods” or “lunches consumed during school lunch periods” more frequently in the Abstract, Methods, and Results, while reserving “school-provided lunches” to refer specifically to lunches provided by the school. This largely addresses the reviewer’s concern. However, the title remains “Nutritional Contribution and Quality of School Lunches in Canadian Elementary Schools,” which may still lead readers to interpret “school lunches” as “school-provided lunches” rather than all foods and beverages consumed during the school lunch period. In addition, the Abstract conclusion still states that “School-provided lunches had higher nutrient density than home-packed lunches, suggesting they may support healthier dietary intake during the school day.” Although this wording is more cautious than before, it still carries a mild policy implication. It is recommended that the title be revised to “Nutritional Contribution and Quality of Lunches Consumed During School Lunch Periods in Canadian Elementary Schools: A Plate Waste Analysis.” The Abstract conclusion should also be revised to “in this sample, school-provided lunches were associated with higher nutrient density, although their absolute contributions to several nutrients remained limited.”
2. The revisions to the Introduction represent clear progress. The revised manuscript now explicitly states that the study aimed to assess energy and nutrient content, evaluate the NRF 9.3 Index, examine contributions to age- and sex-specific dietary recommendations, and compare differences by lunch provenance, age group, and reported gender. This is consistent with the response letter. However, the discussion of the research gap remains relatively general and still does not systematically compare existing studies on Canadian school meal, home-packed lunch, and school-hour dietary intake in terms of region, sample, dietary assessment methods, and indicators. In the Discussion, the authors state that “To our knowledge, this is the first Canadian school-based study to apply this method,” but this incremental contribution should preferably be moved forward to the Introduction. The authors should clearly explain that, compared with previous studies such as those by Tugault-Lafleur et al. based on survey or recall data, the uniqueness of the present study lies in the use of a digital photography–assisted plate waste method, a Saskatchewan lower-income neighbourhood school sample, comparisons by lunch provenance, and the combined indicator framework of DRIs and the NRF 9.3 Index.
3. Sample selection and participant flow remain among the most important unresolved issues. The response letter states that participant inclusion procedures have been clarified and that selection or attendance bias has been acknowledged in the limitations. However, the revised manuscript still only explains that the four schools were purposively selected, that all students from kindergarten to grade six were eligible, that parents could opt out, and that students provided verbal assent, before directly reporting that plate waste data were collected from 379 students. The manuscript still does not report the total number of eligible students, the number of opt-out students, the number of students actually observed, excluded observations, reasons for missing data, or the distribution of included and missing data by school, grade, and lunch provenance. A STROBE-style participant flow diagram should be added. If historical data on these numbers are truly unavailable, the Methods section should explicitly state that “the total number of eligible students and opt-out students was not available,” rather than merely stating in the limitations that “some data were missing.”
4. The classification of lunch provenance still involves conceptual ambiguity. The authors now state that, for students with two observed lunch days, nutrient values were averaged at the participant level, and that students whose lunches originated from different sources across observation days were classified as mixed. Although this rule improves transparency, it conflates two different situations: a “mixed” lunch in which the same meal contained both home and school sources, and a participant whose two observation days consisted of one home-packed lunch and one school-provided lunch. This makes the nutritional interpretation of the mixed lunch category less clear. More importantly, the Results still state that “38.0% of lunches were classified as school-provided,” whereas Table 1 actually reports Lunch Provenance at the participant level among 379 students. It is recommended that all relevant wording be revised to “participants’ averaged lunch intake was classified as…,” and that the observation-level distribution of provenance be reported in the Supplementary Materials. A sensitivity analysis should also be added, at minimum comparing the results after excluding students with inconsistent provenance across days and, preferably, conducting an observation-level analysis accounting for student and school clustering.
5. The NRF 9.3 Index section still requires further standardization. The authors now explain that vitamin D was substituted for vitamin E and that total sugar was substituted for added sugar, and they acknowledge that the modified NRF 9.3 Index limits direct comparability with the original NRF 9.3 formulation. This revision is reasonable in direction. However, there remains an internal inconsistency in the Methods. The manuscript first defines the NRF 9.3 Index as the sum of the percentage of daily values minus the percentage of maximum recommended values, but later states that a binary scoring approach was applied for nutrients to limit, whereby values at or above the recommended maximum were assigned a score of 1 and values below the threshold were assigned a score of 0. This is no longer the standard NRF 9.3 calculation method. It is recommended that the manuscript consistently refer to the index as a “modified NRF 9.3 Index,” provide the exact formula, and explain how binary scoring is combined with per 2000 kcal standardization. Otherwise, the scores reported in Table 5 cannot be reproduced by external researchers.
6. The revised Sex and Gender-based Analysis still presents ethical and methodological risks. The authors state that the available data reflected reported gender as provided by the schools, and that two participants had missing gender information and were “classified using available demographic information.” However, Table 1 directly reports Females 182 and Males 197, summing to 379, without showing missing, undisclosed, or non-binary categories, and without explaining how the two participants with missing gender were classified. Using “available demographic information” to infer gender or assign sex-specific DRI values is not sufficiently transparent and may not fully align with current sex and gender reporting standards. It is recommended that these two participants be excluded from gender-stratified analyses or listed as missing/unknown. The table notes for Table 1 and all relevant tables should state that “reported gender was used as a proxy for sex-specific DRIs; two participants with missing reported gender were excluded from gender-stratified comparisons but retained in overall analyses.”
7. The Discussion and Conclusion are generally more cautious than in the previous version, but some overextension remains. The revised manuscript acknowledges purposive selection, only four schools, residual school-level clustering, the modified NRF 9.3 Index, and limitations in generalizability, which is appropriate. However, the Discussion still states that small-scale school food initiatives “may play an important role in supporting equitable access,” even though the study did not collect individual-level socioeconomic or ethnicity data and did not evaluate access or equity outcomes. The Conclusion also states that “structural approaches beyond individual or household-level solutions may help support children’s nutritional intake” and that the findings can “inform the development of coordinated and nutrition-focused school food systems,” which still tends toward policy advocacy. It is recommended that these statements be revised to “these descriptive findings identify nutritional gaps during school lunch periods and may provide context for future evaluations of school food systems,” in order to avoid using cross-sectional baseline data to overstate causal or policy implications.

Author Response

Dear Reviewer,

We sincerely thank you for the time and effort invested in reviewing our manuscript and for the thoughtful and constructive feedback provided. We appreciate the detailed comments, which have helped us improve the clarity, transparency, and methodological reporting of the study. In response to the recommendations, we have carefully revised the manuscript and provided point-by-point responses below. We hope these revisions adequately address the concerns raised and improve the overall quality and clarity of the manuscript.

1. Regarding terminology, the authors have now used “foods and beverages consumed during school lunch periods” or “lunches consumed during school lunch periods” more frequently in the Abstract, Methods, and Results, while reserving “school-provided lunches” to refer specifically to lunches provided by the school. This largely addresses the reviewer’s concern. However, the title remains “Nutritional Contribution and Quality of School Lunches in Canadian Elementary Schools,” which may still lead readers to interpret “school lunches” as “school-provided lunches” rather than all foods and beverages consumed during the school lunch period. In addition, the Abstract conclusion still states that “School-provided lunches had higher nutrient density than home-packed lunches, suggesting they may support healthier dietary intake during the school day.” Although this wording is more cautious than before, it still carries a mild policy implication. It is recommended that the title be revised to “Nutritional Contribution and Quality of Lunches Consumed During School Lunch Periods in Canadian Elementary Schools: A Plate Waste Analysis.” The Abstract conclusion should also be revised to “in this sample, school-provided lunches were associated with higher nutrient density, although their absolute contributions to several nutrients remained limited.”

Response: Thank you for this suggestion. We agree that the term “school lunches” in the title could be interpreted as referring specifically to school-provided meals rather than all foods and beverages consumed during the school lunch period. To improve clarity and maintain consistency with the terminology used throughout the manuscript, we revised the title to: “Nutritional Contribution and Quality of Lunches Consumed During School Lunch Periods in Canadian Elementary Schools: A Plate Waste Analysis.”

We also revised the Abstract conclusion to better reflect the descriptive nature of the study. The revised wording emphasizes that, in this sample, school-provided lunches were associated with higher nutrient density than home-packed lunches while acknowledging that their absolute contributions to several nutrients remained limited. In addition, we removed language that could be interpreted as implying policy effects and reframed the concluding statement to indicate that the findings provide baseline evidence that may inform future evaluations of school food systems and policies.

2. The revisions to the Introduction represent clear progress. The revised manuscript now explicitly states that the study aimed to assess energy and nutrient content, evaluate the NRF 9.3 Index, examine contributions to age- and sex-specific dietary recommendations, and compare differences by lunch provenance, age group, and reported gender. This is consistent with the response letter. However, the discussion of the research gap remains relatively general and still does not systematically compare existing studies on Canadian school meal, home-packed lunch, and school-hour dietary intake in terms of region, sample, dietary assessment methods, and indicators. In the Discussion, the authors state that “To our knowledge, this is the first Canadian school-based study to apply this method,” but this incremental contribution should preferably be moved forward to the Introduction. The authors should clearly explain that, compared with previous studies such as those by Tugault-Lafleur et al. based on survey or recall data, the uniqueness of the present study lies in the use of a digital photography–assisted plate waste method, a Saskatchewan lower-income neighbourhood school sample, comparisons by lunch provenance, and the combined indicator framework of DRIs and the NRF 9.3 Index.

Response: Thank you for this observation. We agree that the research gap and the specific contribution of the study could be more clearly articulated. In response, we revised the Introduction to provide a more explicit comparison with previous Canadian studies examining school meal consumption, home-packed lunches, and dietary intake during school hours.

We now highlight that, unlike prior Canadian studies that have primarily relied on dietary recalls, surveys, or self-reported intake measures, the present study used a digital photography-assisted plate waste methodology to directly assess foods and beverages consumed during school lunch periods. We also emphasize the unique context of the study, which was conducted in elementary schools located in lower-income neighbourhoods in Saskatchewan, and the study's examination of differences by lunch provenance. Finally, we clarify that the study combines assessments of nutrient contributions relative to Dietary Reference Intakes with evaluation of overall dietary quality using the modified NRF 9.3 Index. These revisions were made to better position the study within the existing literature and clarify its novel contribution.

3. Sample selection and participant flow remain among the most important unresolved issues. The response letter states that participant inclusion procedures have been clarified and that selection or attendance bias has been acknowledged in the limitations. However, the revised manuscript still only explains that the four schools were purposively selected, that all students from kindergarten to grade six were eligible, that parents could opt out, and that students provided verbal assent, before directly reporting that plate waste data were collected from 379 students. The manuscript still does not report the total number of eligible students, the number of opt-out students, the number of students actually observed, excluded observations, reasons for missing data, or the distribution of included and missing data by school, grade, and lunch provenance. A STROBE-style participant flow diagram should be added. If historical data on these numbers are truly unavailable, the Methods section should explicitly state that “the total number of eligible students and opt-out students was not available,” rather than merely stating in the limitations that “some data were missing.”

Response: Thank you for this observation. We agree that greater transparency regarding participant inclusion and study flow is important. Additional study records were reviewed to better characterize participant inclusion. Approximately 620 students in kindergarten through grade 6 were enrolled across the four participating schools, and lunch intake data were obtained from 379 students (approximately 61% of enrollment).

However, data collection occurred across multiple observation days, and although study records documented information such as absences, parental opt-outs, students who went home for lunch, and observations with missing or incomplete data, attendance and participation varied across observation days. Consequently, complete participation information was not available for every student on every observation day, making it difficult to accurately reconstruct a comprehensive STROBE-style participant flow diagram.

To improve transparency, we revised the Methods and Results sections to describe the available participant information and participation procedures in greater detail. We also expanded the Limitations section to acknowledge that incomplete participation records across observation days limited our ability to fully assess participant flow and potential selection or attendance bias.

4. The classification of lunch provenance still involves conceptual ambiguity. The authors now state that, for students with two observed lunch days, nutrient values were averaged at the participant level, and that students whose lunches originated from different sources across observation days were classified as mixed. Although this rule improves transparency, it conflates two different situations: a “mixed” lunch in which the same meal contained both home and school sources, and a participant whose two observation days consisted of one home-packed lunch and one school-provided lunch. This makes the nutritional interpretation of the mixed lunch category less clear. More importantly, the Results still state that “38.0% of lunches were classified as school-provided,” whereas Table 1 actually reports Lunch Provenance at the participant level among 379 students. It is recommended that all relevant wording be revised to “participants’ averaged lunch intake was classified as…,” and that the observation-level distribution of provenance be reported in the Supplementary Materials. A sensitivity analysis should also be added, at minimum comparing the results after excluding students with inconsistent provenance across days and, preferably, conducting an observation-level analysis accounting for student and school clustering.

Response: Thank you for this observation. We agree that participant-level lunch provenance classification and observation-level lunch provenance represent distinct concepts and that greater clarity was needed.

In response, we revised the Methods section to explicitly describe how lunch provenance was assigned. Nutrient intakes were averaged at the participant level across available observation days, and lunch provenance classifications were subsequently assigned based on the source(s) represented across the observations contributing to the averaged intake. We also clarified that participants whose lunch provenance differed across observation days were classified as mixed and acknowledged the resulting heterogeneity of this category.

To improve transparency, we revised the Results section to consistently refer to participant-level classifications rather than lunches and added observation-level lunch provenance distributions to the Supplementary Materials (Table S2).

We also conducted sensitivity analyses to assess the potential influence of lunch provenance classification. First, we repeated the NRF 9.3 analyses after excluding participants whose lunch provenance differed across observation days. Findings were materially similar to those observed in the primary analysis. Second, we fitted a regression model including school fixed effects to assess the potential influence of school-level clustering, and results remained unchanged. These analyses have been added to the Methods, Results, and Supplementary Materials.

We acknowledge that the mixed category may include both lunches containing foods from multiple sources within a single observation day and participants whose lunch provenance varied across observation days. This limitation has been further acknowledged in the Discussion and Limitations sections.

5. The NRF 9.3 Index section still requires further standardization. The authors now explain that vitamin D was substituted for vitamin E and that total sugar was substituted for added sugar, and they acknowledge that the modified NRF 9.3 Index limits direct comparability with the original NRF 9.3 formulation. This revision is reasonable in direction. However, there remains an internal inconsistency in the Methods. The manuscript first defines the NRF 9.3 Index as the sum of the percentage of daily values minus the percentage of maximum recommended values, but later states that a binary scoring approach was applied for nutrients to limit, whereby values at or above the recommended maximum were assigned a score of 1 and values below the threshold were assigned a score of 0. This is no longer the standard NRF 9.3 calculation method. It is recommended that the manuscript consistently refer to the index as a “modified NRF 9.3 Index,” provide the exact formula, and explain how binary scoring is combined with per 2000 kcal standardization. Otherwise, the scores reported in Table 5 cannot be reproduced by external researchers.

Response: Thank you for this important observation. We agree that the previous description of the NRF 9.3 calculation created ambiguity and did not provide sufficient detail to ensure reproducibility.

In response, we recalculated NRF 9.3 scores using the conventional continuous scoring approach rather than binary scoring for nutrients to limit. All references to binary scoring have been removed. The Methods section has been revised to provide the complete calculation approach, including the exact formula, per-2000 kcal standardization, and truncation of individual nutrient contributions at 100%, consistent with established NRF methodology.

We also expanded the description of the nutrients included in the index and the reference values used for scoring. In addition, we clarified the substitutions made due to data availability and alignment with the Canadian context, including the use of vitamin D in place of vitamin E and total sugar in place of added sugar. These revisions improve 

6. The revised Sex and Gender-based Analysis still presents ethical and methodological risks. The authors state that the available data reflected reported gender as provided by the schools, and that two participants had missing gender information and were “classified using available demographic information.” However, Table 1 directly reports Females 182 and Males 197, summing to 379, without showing missing, undisclosed, or non-binary categories, and without explaining how the two participants with missing gender were classified. Using “available demographic information” to infer gender or assign sex-specific DRI values is not sufficiently transparent and may not fully align with current sex and gender reporting standards. It is recommended that these two participants be excluded from gender-stratified analyses or listed as missing/unknown. The table notes for Table 1 and all relevant tables should state that “reported gender was used as a proxy for sex-specific DRIs; two participants with missing reported gender were excluded from gender-stratified comparisons but retained in overall analyses.”

Response: Thank you for this observation. We agree that the previous handling of participants with missing reported gender information was not sufficiently transparent.

In response, the two participants with missing reported gender information have been classified as missing rather than assigned to a gender category. They were retained in overall analyses but excluded from all gender-stratified analyses and age–gender subgroup comparisons.

We also revised the Sex and Gender-Based Analysis section, Table 1, and the relevant table notes to clarify that reported gender was used as a proxy for sex when applying sex-specific Dietary Reference Intake (DRI) values. These revisions improve transparency and better align the reporting of sex- and gender-related variables with the available data.

7. The Discussion and Conclusion are generally more cautious than in the previous version, but some overextension remains. The revised manuscript acknowledges purposive selection, only four schools, residual school-level clustering, the modified NRF 9.3 Index, and limitations in generalizability, which is appropriate. However, the Discussion still states that small-scale school food initiatives “may play an important role in supporting equitable access,” even though the study did not collect individual-level socioeconomic or ethnicity data and did not evaluate access or equity outcomes. The Conclusion also states that “structural approaches beyond individual or household-level solutions may help support children’s nutritional intake” and that the findings can “inform the development of coordinated and nutrition-focused school food systems,” which still tends toward policy advocacy. It is recommended that these statements be revised to “these descriptive findings identify nutritional gaps during school lunch periods and may provide context for future evaluations of school food systems,” in order to avoid using cross-sectional baseline data to overstate causal or policy implications.

Response: Thank you for this observation. We agree that some of the previous wording could be interpreted as extending beyond the descriptive scope of the study.

In response, we revised the Discussion and Conclusion to remove language implying effects on equity, access, or policy outcomes, which were not directly assessed. Statements suggesting that school food initiatives support equitable access or that structural approaches are needed have been replaced with more neutral wording.

The revised text now emphasizes that these descriptive findings identify nutritional gaps during school lunch periods and differences in nutrient density by lunch provenance. We also clarified that, given the purposive selection of four schools and the cross-sectional nature of the study, findings should not be generalized beyond similar school settings. The Conclusion now states that the results provide baseline information on lunchtime dietary intake and may provide context for future evaluations of school food systems and school-based nutrition initiatives.