Review Reports - Early Life Added Sugars and Associated Appetite, Satiety, Growth and Adiposity in the First 2 Years of Life

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study makes a valuable contribution by examining added sugar intake in children under two years of age and linking it to appetite, satiety, growth, and adiposity. The longitudinal design and careful use of validated instruments are clear strengths. That said, several points deserve further reflection:

1.The categorization of added sugar intake into tertiles is straightforward, but the chosen cut-offs may limit sensitivity. Would alternative approaches, such as quartiles or continuous modeling, yield additional nuance?

2.Dietary data were collected through 24-hour recalls, which are widely used but prone to recall bias, especially in infant feeding. It would be helpful to acknowledge how this limitation might have influenced intake estimates.

3.Infant formula emerged as a major contributor to added sugar intake. Clarification is needed on whether sugars in formula were classified as “added” under international definitions, as this has implications for interpretation and policy relevance.

4.The associations with emotional overeating and food fussiness are intriguing. A deeper discussion of potential mechanisms—whether neurobiological or behavioral—would strengthen the interpretation.

5.Breastfeeding duration was inversely associated with sugar intake, which is an important finding. However, residual confounding by maternal feeding practices or cultural norms may remain. How were such unmeasured factors considered?

6.The observed effects on growth and adiposity, while statistically significant, appear modest. A clearer discussion of their clinical relevance would help readers understand the practical implications.

7.Given the study’s Mexican cohort, the generalizability of findings to other settings may be limited. A brief reflection on how cultural dietary practices and socioeconomic factors might shape sugar exposure elsewhere would be useful.

8.The policy implications could be more explicitly drawn out. How might these findings inform infant feeding guidelines, regulation of commercial baby foods, or breastfeeding promotion strategies?

Author Response

Thank you for your insightful comments and suggestions. Please find the point-by-point answers below.

R= Thank you for your comment. We initially examined associations with added sugar as a continuous variable. However, statistical tests indicated that the association was not linear. The final tertile categorization was first used only as exploratory but we found it useful: it allowed us to distinguish between no added sugar, very small amounts (less than ~1 teaspoon), and around 1 teaspoon—cutpoints that are meaningful for early life dietary guidance. Quartiles and finer splits were considered, but they produced sparse categories at some ages. We included this information in lines 138-142

R= We agree that 24hour dietary recalls are vulnerable to recall and reporting bias. We have now added text acknowledging this limitation and noting that such bias could have led mainly to underestimation of added sugar intake (lines 407-413). We also highlight that using trained interviewers and the 5-step multiple-pass method helps mitigate (but not eliminate) this concern. In addition, 24‑hour recalls remain the most widely recommended and commonly used method for dietary assessment in epidemiologic studies because they provide detailed intake data with relatively low respondent burden. At preschool age, parents are typically highly involved in and attentive to their children’s eating patterns, which likely enhances the accuracy of proxy reporting compared with later childhood.

R= Thank you for your comment. We agree this is important to clarify. When the label explicitly reported the amount of added sugars, we used that value directly. When added sugars were not reported, we estimated them by subtracting the amount of lactose from the total carbohydrate content. We have now clarified this in the Discussion (lines 418-422) acknowledging that definitions of added sugars in infant formula vary across regulatory frameworks, which may influence comparability and policy interpretation.

R= We agree with your suggestion. We have now expanded the Discussion (lines 352-360) to outline both neurobiological and behavioural pathways that may underlie the associations observed with emotional overeating and food fussiness. Specifically, we describe how emotional overeating may relate to stress regulation processes involving the hypothalamic–pituitary–adrenal (HPA) axis, whereby consumption of sweet foods can provide short term stress relief and reinforce preferences for palatable foods. We also note that food fussiness has been linked to sensory sensitivity and lower reward responsiveness to food, which may prompt parents to offer sweeter options to encourage intake. In addition, having a higher preference for sweet tastes can make children more selective toward other foods and increase reliance on sweet items to ensure adequate intake. Because foods are often used as rewards in early childhood, this can further reinforce emotional overconsumption of sweet foods and perpetuate a cycle of selectivity and preference for sweetness.

R= We appreciate the opportunity to clarify this point. While unmeasured maternal or cultural factors may contribute to infant feeding patterns, specific cultural norms that would plausibly influence both breastfeeding duration and early added sugar intake are not well characterised, and their role as confounders is therefore uncertain. To minimise potential confounding, we adjusted for several upstream maternal and household characteristics (e.g., maternal education and socioeconomic indicators) that are known determinants of both breastfeeding behaviours and later dietary exposures. We also noted that the covariate section was accidentally missing in the Methods and have now added this information for clarity (apologies, lines 194-204).

R= Thank you for your comment. We consider that differences of this magnitude are meaningful in children under two years of age (+2 cm in abdominal circumference and −0.17 Z in length for‑age). For context, a previous study showed that a one SD increase in abdominal circumference (approximately 3 cm at 36 months) was associated with measurable differences in blood pressure[1]. Even a −0.17 Z score difference in length-for‑age may be relevant given that the first 1,000 days represent a critical window for growth, during which most linear growth faltering occurs in low and ‑ and middle-income countries[2]. This is particularly important in our cohort, where the population mean for length for-age was consistently below zero (−0.80 Z at 6 months, −1.15 Z at 12 months, and −1.01 Z at 24 months), indicating a general vulnerability to impaired linear growth. We have expanded the Discussion (lines 368-379) to provide this context and to clarify the potential significance of these findings.

Nowson, C.A.; Crozier, S.R.; Robinson, S.M.; Godfrey, K.M.; Lawrence, W.T.; Law, C.M.; Cooper, C.; Inskip, H.M. Association of Early Childhood Abdominal Circumference and Weight Gain with Blood Pressure at 36 Months of Age: Secondary Analysis of Data from a Prospective Cohort Study. BMJ Open 2014, 4, e005412, doi:10.1136/bmjopen-2014-005412.
Kumala Putri, D.S.; Widodo, Y.; Gunardi, H.; Kusharisupeni; Besral; Djahari, A.B.; Syafiq, A.; Achadi, E.L.; Bhutta, Z.A. When Did the Substantial Loss of Child Linear Growth Occur? PLoS One 2023, 18, e0291176, doi:10.1371/journal.pone.0291176.

R= We agree that cultural dietary practices and socioeconomic conditions can influence the levels and patterns of added sugar exposure across settings. However, we do not expect the underlying biological mechanisms to differ meaningfully between populations. For this reason, we believe the generalizability of our findings is not inherently limited by the Mexican context. To address your suggestion, we have added a brief reflection in the Discussion (lines 437-440) noting that our results are likely relevant to other settings where young children are exposed to added sugars, even if the specific sources or consumption patterns differ.

8.The policy implications could be more explicitly drawn out. How might these findings inform infant feeding guidelines, regulation of commercial baby foods, or breastfeeding promotion strategies?

R= Thank you for this helpful comment. We have expanded the Discussion to more explicitly articulate the policy implications of our findings (lines 473-478). In particular, we now describe how early exposure to added sugars can inform infant feeding guidelines, including reinforcing recommendations to avoid added sugars during the first two years of life. We also highlight the relevance of our results for strengthening regulation of commercial baby foods and infant formulas—especially regarding added sugar limits and transparent labelling—and for supporting breastfeeding promotion strategies, given the protective association observed between breastfeeding duration and lower added sugar intake.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript addresses a highly relevant and timely public health topic, namely added sugar exposure during the first two years of life and its associations with eating behaviour, growth, and adiposity. The focus on early infancy, a critical developmental window, and the use of data from a prospective birth cohort constitute notable strengths.

I have made some suggestions on how you could improve your work. This doesn't mean that you have to agree with them or rewrite your work in the same way. They are just suggestions to help you see things from a different perspective.

1) In the "Introduction" section, please, briefly acknowledge differences between commonly used definitions (e.g., added sugars vs. free sugars as defined by WHO). This would improve conceptual clarity and facilitate comparison with international literature.

2) While neurobiological mechanisms related to appetite and reward are well described, the link between AS intake and linear growth impairment and central adiposity could be strengthened with one or two sentences referencing potential mechanisms (e.g., nutrient displacement, insulin and IGF-1 signalling, altered gut microbiota, or early metabolic programming).

3) The paragraph presenting ENSANUT data is valuable. To strengthen it, you may explicitly note why Mexico represents a particularly relevant setting (e.g., high sugar availability, early introduction of sweetened beverages, or rising childhood obesity rates).

4) It would be useful to explicitly state that appetite- and satiety-related behaviours in relation to AS intake during infancy have been insufficiently explored, underscoring the originality of this aspect of the study.

5) In the "Materials and Methods" section, it would be helpful to more explicitly state the inclusion and exclusion criteria for mother-child pairs (e.g., preterm birth, low birth weight, congenital conditions, incomplete dietary data), as well as the timing and frequency of follow-up assessments.

6) Please, clarify whether the 24-hour dietary recalls were conducted on weekdays and/or weekends and how many recalls per child were included at each age point.

7) Describe the method used to estimate portion sizes (e.g., food models, photographs, household measures).

8) Please, indicate how breast milk intake was handled or estimated, if applicable.

9) Please, provide a brief description of how added sugars were identified and quantified within mixed or processed foods (e.g., use of food composition tables, ingredient lists).

10) The rationale for categorising AS intake into tertiles, particularly defining the lowest tertile as 0 g/day, should be justified. Please, mention whether alternative categorisations or continuous analyses were explored or tested in sensitivity analyses.

11) Please, specify which version of the Child Eating Behaviour Questionnaire (CEBQ) was used and whether it was culturally adapted or validated for the study population.

12) Indicate the age(s) at which the questionnaire was administered and who completed it (e.g., mother, primary caregiver).

13) Please, provide additional details on measurement protocols (e.g., training of personnel, number of repeated measurements, intra- and inter-observer reliability).

14) Clarify the reference standards used to calculate Z-scores (e.g., WHO growth standards).

15) A clearer description of all covariates included in the models (e.g., maternal education, socioeconomic status, maternal BMI, smoking during pregnancy, total energy intake) and the rationale for their selection would strengthen the analytical transparency.

16) Please, indicate how missing data were handled. Clarify whether assumptions of linear mixed-effects models were checked and specify the software and version used for statistical analyses.

17) In the "Results" section, please, add a brief description (or a supplementary flow diagram) indicating how many observations contributed to each age point and each model. This would help readers assess attrition and statistical power across follow-ups.

18) Providing a supplementary table with key maternal and child characteristics stratified by AS intake tertiles would help contextualize the observed associations and assess potential residual confounding.

19) Given known sex differences in growth trajectories and eating behaviours, it may be useful to briefly state whether sex interactions were tested or whether results were consistent across boys and girls.

20) A short reminder in the text (or table footnotes) indicating the directionality of CEBQ subscales (i.e., higher scores reflecting more pronounced behaviours) would improve clarity for readers less familiar with the instrument.

21) While interpretation should remain minimal in the "Results" section, one brief sentence contextualising the magnitude of key findings (e.g., waist circumference difference) could aid reader comprehension.

22) In the "Discussion" section, the association between higher AS intake and lower linear growth is a novel and important finding. Expanding the mechanistic discussion would strengthen the section. Potential pathways to briefly mention include nutrient displacement and reduced micronutrient density of the diet, effects of high glycaemic load on insulin–IGF-1 signalling and growth regulation, possible alterations in gut microbiota affecting nutrient absorption and growth.

23) Although the prospective design is a strength, it may be helpful to explicitly reiterate that causal inference remains limited and that reverse causation (particularly for eating behaviours) cannot be entirely excluded.

24) While the Mexico-specific regulatory discussion is valuable, a short statement linking these findings to global policy frameworks (e.g., WHO complementary feeding guidelines) would increase international applicability.

25) Please, add a concise paragraph outlining priorities for future research, such as longer-term follow-up into preschool or school age, objective biomarkers of sugar intake, experimental or intervention studies reducing AS exposure in infancy.

26) The “Conclusion” section is not included in the article. Is it correct?

While this manuscript presents valuable findings, improving the clarity and depth of the discussion, and increasing the level of methodological detail, would further enhance its impact. I recommend accepting the manuscript, provided that some revisions are made.

Author Response

Thank you for your insightful comments and suggestions. Please find the point-by-point response below:

R= Thank you for your suggestion, it does improve it. We included this in the Introduction (lines 55-57).

R= We included reference to nutrient displacement in the introduction (lines 73-76)

R= Thank you for the suggestion. We included it in lines 83-87.

R= We agree and included this in lines 90-91.

R= Inclusion and exclusion criteria were stated in lines 109-115. Timing and frequency of data collections are also stated in lines 123, 150, 161, 184. It is organized by data type (dietary intake, CEBQ, breastfeeding, and anthropometric measurements).

6) Please, clarify whether the 24-hour dietary recalls were conducted on weekdays and/or weekends and how many recalls per child were included at each age point.

R= The 24‑hour dietary recalls were conducted on both weekdays and weekends (approximately 20% on weekends). Although two recalls were collected for around 10% of participants at each age point, only one recall per child was used for the present analysis. We included this in lines 124-126.

7) Describe the method used to estimate portion sizes (e.g., food models, photographs, household measures).

R= We specified that the dietary recalls were interviewer-led, with household measures for portion size estimation (lines 129).

8) Please, indicate how breast milk intake was handled or estimated, if applicable.

R= Thank you for this comment. Breastmilk intake in the dietary recalls was estimated indirectly. When expressed breast milk was provided, we used the reported volume to calculate grams consumed. For direct latch, we applied age‑specific estimates of average intake per feeding episode to derive breastmilk intake. We have now clarified this in lines 129-132.

9) Please, provide a brief description of how added sugars were identified and quantified within mixed or processed foods (e.g., use of food composition tables, ingredient lists).

R= Thank you for your suggestion. We do have a brief description in lines 133-136. We can’t specify more details unfortunately because it’s part of another publication in progress.

R= We agree. We included the rationale in lines 138-142. We did initially explore associations with added sugars as a continuous variable. However, associations were not linear. The final tertile categorization was first used only as exploratory but we found it useful: it allowed us to distinguish between no added sugar, very small amounts, and around 1 teaspoon—cutpoints that are meaningful for early life dietary guidance. Quartiles and finer splits were considered, but they produced sparse categories at some ages.

11) Please, specify which version of the Child Eating Behaviour Questionnaire (CEBQ) was used and whether it was culturally adapted or validated for the study population.

R= We used the original CEBQ and applied the full version in our study. As noted in lines 178-182, the instrument was translated and culturally adapted and validated for use in this study population. We have clarified this point in the manuscript to ensure it is explicit.

12) Indicate the age(s) at which the questionnaire was administered and who completed it (e.g., mother, primary caregiver).

R= As indicated in lines 161–162, the CEBQ was administered at 6, 9, 12, 18, and 24 months, and it was completed by the child’s mother or primary caregiver.

13) Please, provide additional details on measurement protocols (e.g., training of personnel, number of repeated measurements, intra- and inter-observer reliability).

R= Thank you for this comment. Anthropometric personnel were trained following the Habicht standardization protocol. All measurements were taken in duplicate, except for skinfolds, which were measured in triplicate. Unfortunately, intra‑ and inter‑observer reliability statistics were not available due to logistical constraints and budget limitations during data collection. We have clarified these details in the manuscript in lines 146-147.

Habicht, J.-P. Estandarización de Métodos Epidemiológicos Cuantitativos Sobre El Terreno. Oficina Sanitaria Panamericana 1974, LXXV, 375–384.

14) Clarify the reference standards used to calculate Z-scores (e.g., WHO growth standards).

R= We used the World Health Organization growth standards for children aged 0–5 years (indicated in lines 157-158).

R= Apologies about this, we agree this is necessary. It seems we accidentally omitted this before. We included the covariate section now in lines 194-204.

16) Please, indicate how missing data were handled. Clarify whether assumptions of linear mixed-effects models were checked and specify the software and version used for statistical analyses.

R= Missing data were handled under the missing‑at‑random assumption, using all available observations. Model assumptions were evaluated using standard residual diagnostics and no major violations were detected. All statistical analyses were conducted using Stata version 19 (StataCorp, College Station, TX). This information was clarified in lines 224-227.

R= Thank you for this suggestion. We now report the total number of observations contributing to the longitudinal analyses (n = 823) in the Methods section (line 223). Because the models use all available data across time points, and the number of observations varies slightly by outcome, we opted to present the overall total to maintain clarity and avoid over‑complicating the narrative.

R= Thank you for this thoughtful suggestion. In our models, children can shift between AS‑intake tertiles across follow‑up visits, making a single stratified table difficult to interpret and potentially misleading. For this reason, we relied on a priori causal assumptions, guided by directed acyclic graphs, to define the minimally sufficient adjustment set rather than on exploratory comparisons across tertiles.

R= Thank you for this comment. We examined whether associations differed by sex by including an interaction term between AS intake and child sex in the models. The interaction was not statistically significant, indicating that results were consistent for boys and girls. We added this in lines 203-204.

R= Thank you we agree, we included this in lines 164-165.

R= We agree. This was included under Discussion in lines 368-377.

R= Thank you for highlighting this point. We have expanded the Discussion to briefly outline potential biological pathways in lines 378-379.

R= We included this statement under lines 440-442.

R= Thank you for your suggestion. We included this in lines 453-454.

R= Thank you for this suggestion. We included the need for studies designed to evaluate strategies for reducing added sugar exposure during infancy, and additional research examining the potential effects on linear growth trajectories into later childhood (lines 481-483).

26) The “Conclusion” section is not included in the article. Is it correct?

R= Correct. It is the final part of the discussion. From line 458 onwards.