Later eating rhythm measured in children at 7 years of age in the ALSPAC cohort

Later eating rhythm (LER) refers to later timing, greater energy intake (EI), and higher frequency of eating occasions (meal/snack) in the evening. The significance of LER in child health is becoming increasingly recognised. However, the lack of consensus regarding definitions of LER make it challenging to fully comprehend its role. This data note describes LER variables derived in the Avon Longitudinal Study of Parents and Children (ALSPAC), an ongoing birth cohort which enrolled 14,541 pregnant women living in Avon, UK, with an expected date of delivery between April 1991 - December 1992. When children were 7 years, parents completed a structured 3-day food diary, recording all foods/drinks consumed over 3 days (preferably 1 weekend day and 2 weekdays). Data was available for 7,285 children (50.1% response rate). A subsample of 4,869 children had exact time of eating occasions added to the existing database, which only included broad indications of eating timing based on 2-7 hour long meal slots. 13 LER variables were derived for the entire week and weekdays/weekend days separately. These comprise: 1) eating around individual bedtime (number days); 2) eating around average bedtime (number days); 3) time of evening main meal (hrs:mins); 4) time of last eating occasion (hrs:mins); 5) EI in the evening (percentage of total daily energy intake, %TDEI); 6) EI within 2hrs before bedtime (%TDEI); 7) EI for evening main meal (%TDEI); 8) EI for evening snacks (%TDEI); 9) Night eating1 (NE1): eating over 30% of total daily energy intake after 18:00 (number days); 10) NE2: eating over 25% of total daily energy intake within 2hrs before bedtime (number days); 11) eating frequency after 17:00 (number of eating occasions); 12) regularity of dinner (number of days); 13) frequency of evening snacks (number days). We describe the derivation, prevalence and inter-corelations between LER variables.


Introduction
Recent studies have emphasized that timing of food intake, known as chrono-nutrition, may play an important role in adiposity development by demonstrating changes in energy regulation through circadian-driven processes, such as transport of lipids, glucose, and dietary proteins in the intestine [1][2][3][4][5][6][7] .Trials in adults have shown that eating late at night results in less-efficient energy metabolism and changes in metabolic markers such as decreased whole-body fat oxidation and reduced glucose tolerance and insulin secretion, thereby increasing the risk of metabolic syndrome [8][9][10] .In addition emerging studies in animals have shown that a wide variety of metabolic markers are affected by later eating such as adipokines, glucocorticoids, and clock genes 11,12 .Experimental trials in rodents suggested that mice restricted to feeding during light cycle developed obesity 13 , whereas mice restricted to feeding during dark cycle were protected from obesity 14 .Psychologically, a bigger evening main meal has been suggested to cause higher hunger and a lower suppression of appetite compared with a bigger breakfast in weight management 11 .Thus, the time of day of energy intake (EI), in particular later in the evening around bedtime, has been highlighted as a particular concern for the development of adiposity.
The concept of later eating rhythms has previously been studied in many different ways.Night Eating (NE) was first proposed by Stunkard et al. 15 in 1955, when they defined it as consumption of more than 25% of total daily energy intake (TDEI) after the evening meal.NE was commonly considered as the main criterion for diagnosing night eating syndrome (NES), a disordered eating behaviour, generally diagnosed as morning anorexia, evening hyperphagia, a depressed mood and insomnia in adults 16 .Stunkard et al. modified these criteria in 1996 to "50% of TDEI or more after 19:00" 17 .Later on, several researchers adapted this into a number of diverse definitions taking into account cultural differences, such as "the largest food intake occurring during a time period after 19:00" or "overeating following the evening meal but before the end of the sleep period".However, these definitions were restricted to adults as NES does not typically appear in childhood, when eating times are largely determined by parents 18,19 .In addition, it is essential to differentiate NES/NE as psychiatric conditions from habitual later eating with and without overt medical problem.Therefore, other terms such as, "late-night snacking", "eating late in the evening" 20,21 , "nocturnal eating" 22 , "late-night overeating" 23 , "evening chronotype" 22 and "night time eating/EI" [24][25][26] have been used to capture one or more specific features of habitual later eating behaviours and distinguish them from NES. Definitions across studies are diverse, differing in the definition of "late", and perspectives such as, timing of food intake, EI, and meal frequency in the evening/night.Therefore, to harmonize the inconsistent use of terms and to cover later eating behaviours comprehensively from all perspectives, a broader term "later eating rhythm" (LER) encompassing NE and eating more in the later part of the day was created to describe later eating behaviours in children and adolescents by our recent systematic review 27,28 .This paper describes the LER variables that have been derived from parent-reported diet diaries over three days collected at age 7 years as part of the Avon Longitudinal Study of Parents and Children (ALSPAC).Given the growing evidence of the importance of LER for obesity-related eating behaviours and obesity in later life, these data can be used by studies focusing on chrono-nutrition, disordered eating, other timerelated eating behaviours, and metabolic health in children.

Sample size
ALSPAC is an ongoing birth cohort study that enrolled 14,541 pregnant women living in Avon, UK, with an expected date of delivery between April 1991 and December 1992.The subsequent children (n = 14,062 at birth; n = 13,988 alive at 1 year) have been followed using questionnaires completed by parents and the children, educational records, and hands-on assessment at dedicated research clinics.An overall total of 913 additional children were recruited at age 7 years (n=456), age 8-18 years (n=262) and age 19-26 years (n=195) to bolster the initial sample, increasing the total child cohort to 14,901 alive at 1 year.By age 7 years, 546 children were permanently lost to follow-up, leaving 13,972 children eligible for follow-up at 7 years.A detailed account of the study methodology can be found elsewhere 29,30 .Ethical approval for the overall study was obtained from the ALSPAC Law and Ethics Committee (ALEC) and the local research ethics committees (IRB00003312).Ethical approval for the 7 year clinic was obtained from the United Bristol Healthcare Trust (E4168), Southmead (North Bristol Trust; Project 084/99) and Frenchay (North Bristol Trust; Project 99/42).Written informed consent for the use of data collected via questionnaires and clinics was obtained from participants following the recommendation of ALEC at the time.
3-day food diaries were collected as part of a face to face clinic at the age of 7 years 31 .Parents were invited by post to complete a structured 3-day food diary and were asked to bring the completed food records to their clinic visit.Of all of those invited to the research clinic (n = 13,146), a total of 7,285 children had food diaries completed, resulting in a response rate of 55.4% (Figure 1).

Diet diary data
In each diet diary, parents were asked to record the exact time, name and size of any drinks, medicines and foods their children consumed over 3 individual days (preferably 1 weekend day and 2 weekdays), both in and out of the home.Parents were instructed to speak to teachers or carers regarding meals consumed away from home.

Amendments from Version 2
We have clarified the non-random selection of the first 1834 diaries that were coded for exact times.We have corrected an error noted by a reviewer where the text referred to Table 3 rather than Table 4. Finally, we have added further discussion regarding the reliabiity and validity of food diaries generally but also those used in ALSPAC more specifically.

Data preparation
Time data entry.The exact time of consumption for every food or drink was not originally entered into the electronic dietary dataset, instead, general meal times were coded according to the rules described in Table 1.For example, when a food or drink was consumed at 8:15, the time of consumption was originally entered as "Breakfast, 0800".However, the exact time of food intake is of particular importance for estimating LER variables and this was therefore keyed in at a later date.
The original paper diet diaries were retrieved from the archive for data entry.The process of time entry in the database was undertaken as follows: 1.The original diary ID number and day number were entered.
2. The foods listed for the first meal slot in the original data file were assessed to identify whether they matched with the foods reported in the original diet diary for the first exact time reported.
3. If all the foods reported matched, then the exact time for that eating occasion was entered next to each food.Time was entered as a four-digit number in 24-hour clock format.

A record was kept when any of the following occasions happened:
• If the person consumed the same food more than once in the same meal time slot and there was only one slot in the electronic diary, then the earliest eating occasion time was coded, and a note was kept about this case.
• If the food in the hard copy diary could not be matched to the food item in the database directly: ○ If there was no time of consumption of a food item in the hard copy diary but the food was found in the database, then '99' was entered under the column of Time.
○ If the food in the database was not found in the diary (which may be due to the wrong input of the food name), the number '88' was entered under the column of Time.
To date, food diaries for a subsample of 4,869 out of 7,285 children have been selected and coded with exact time of each food/drink (Figure 1).Among these, 1,834 were selected based on the availability of other data (eating behaviour asked about in a self-completion questionnaire sent out when participants were 24 years of age) for a previous research project.The remaining 3,035 were selected randomly.Of these, 840 children with less than three days recorded were excluded from the analysis for weekday and weekend days combined.Thus, a total of 4,029 children were included in the analysis for the whole week, with 644 of those having data for three weekdays rather than a combination of week and weekend days.Those children who had at least two weekday food diaries available were included in the analysis for weekdays (N=4,063), and those who had at least one weekend food diary were included in the analysis for weekend days (N=3,757).There were 3,385 children who had food diaries recorded both for the week and weekend, hence they were included in the analysis for week vs weekend comparison.
The characteristics of children included and excluded in this study are presented in Table 2.It can be seen that there are no differences in those with precise timings compared to the remainder of the cohort with diet diaries according to child sex, BMI at 7 years of age and maternal age and education at birth (all p>0.10).
Defining LER in the ALSPAC cohort Eating Occasion (meal/snack) definition.The definition of eating occasion (EO) will significantly influence the definition of LER variables.Therefore, the definitions of EO, meal and  snack in the current study were developed based on our data in line with the approach used in previous work.As shown in Table 3, the current study defined EO as any occasion when food/drink was consumed, which was adopted by most previous studies with exact time of food intake in children [32][33][34] .Since food diaries do not usually identify participant-defined meals or snacks, previous studies tended to approximate breakfast, lunch, and dinner by applying "the biggest EO in the time slots that breakfast, lunch, and dinner usually fall into", respectively 32,35,36 .
It is important to be able to assess whether the time slots we set for usual mealtimes was in line with when distinct large EOs tended to happen in our data.Thus, to gain insight into the usual timing of meals in this population, the mean EI (as a percentage of TDEI) was plotted for every 30-min time interval over a 24-hour period, across all participants (Figure 2).Bedtime and type of day (week or weekend) were also taken into account.
The resulting graph shows the lowest points of EI approaching 0, suggesting that suitable time cut-offs between morning, afternoon, and evening/night would be: 6:00-11:59, 12:00-16:59, and 17:00-23:59.Thus, main meals were determined as the biggest EOs in the morning, afternoon, and evening, similar to previous studies 32,35,36 .All other smaller EOs were defined as snacks.In addition, the graph also displays three clear peaks in EI across the day, which suggests that the usual mealtimes for breakfast, lunch and dinner would be: 6:00-9:59, 12:00-14:59 and 17:00-18:29.Thus, main meal skipping was determined as "no EOs during the corresponding usual mealtimes".
Definitions and calculations for LER variables.Table 4 presents the various definitions of LER that were adopted in this study from three different perspectives: T (timing), E (energy intake), and F (meal frequency).Each domain includes different variables focusing on clock time, evening main meal, and evening snacks.The different definitions of EO (meal/snack) shown in Table 4 were applied.To minimize the impact of cultural differences, most LER variables were left in their original format without applying any categorical criteria (cut-offs).
As suggested by our review 28 , the definition of "late" needs to be comparable when defining T1 (eating at a late time at night), or E1c (higher EI after a late time in the evening/night), and it is worth considering the time criteria relative to bedtime.
In the current study, the parent-reported usual bedtime (hour: min) at around 7 years old (81 months) was used to indicate "late"."Eating after the individual usual bedtime" and "EI within 2 hrs before bedtime" were therefore adopted.In addition, to maximise the sample size, "eating after the average bedtime" was used.
Notably, in the subgroup of E1 (EI after a specified time in the evening), time thresholds used in previous studies, such as 21:00 and 23:00, were so late that almost all UK children had no EI and had gone to bed (Figure 2), hence they were not used in the current study.Others, such as EI after 16:00, 17:00, 18:00 and 19:00, were initially calculated.The practical meaning of these variables varied by country and was set depending on whether the researchers expected the evening main meal to be included 20,32,[37][38][39][40][41][42] .Thus, the meaning of these clock-time related variables may be duplicated with other meal-typerelated variables, and the clock time may mean different things for different countries.In order to reduce the duplication across variables and to increase comparability with previous research, box plots were used to detect the differences by exhibiting the quartiles of EI for different times (i.e., 16:00, 17:00, 18:00, 19:00, evening main meal, evening snacks, and 2 hours before bedtime).As shown in Figure 3, EI after 16:00 was similar to EI after 17:00, and the definition of evening in this study was 17:00-23:59, so the EI after 17:00 was kept referring to "EI in the whole evening".Similarly, the distribution of EI for 19:00-23:59 and for evening snack were similar and EI for conventional meal/snack was used more commonly than the exact time of consumption, so EI for evening snack was kept.
Regarding the definition of NE, the time/energy criteria of NE varied between countries (e.g., consuming >25% or 50% of TDEI after 19:00, 20:00 or 21:00).Despite these variations, studies consistently reported a prevalence of NE ranging from 12.8% to 37.0% 32,[39][40][41][42] .Thus, an appropriate definition of NE aligns with the eating culture of the target population.In the current study, very few children consumed 50% of TDEI in the evening regardless of the time criteria, so 50% of TDEI was excluded as energy criteria.Additionally, 17:00-23:59 was too broad, and 19:00 and later was too restrictive to

Eating occasion (EO)
An individual EO is separated by the unique recorded time
Consequently, a total of 13 LER variables (T1-a/b, T2, T3; E1-a/b/c1/c2, E2, E3; F1, F2, F3) were considered in the current study.The variables were calculated for the whole week, weekdays only, and weekend days only as follows: • Timing: All the recorded times for food intake from the 24-hour clock were transformed to minutes which enabled the calculation of the mean values.The mean timing for the EO was then calculated by dividing the total timing for the EO over n days by the number of recorded days (n)."n" could be the overall number of weekdays (n≥2), weekend days (n≥1), or weekdays and weekend days combined (n=3), depending on the type of day."k" was the specific recorded day (k=1, the first day; k=2, the second day; k=3, the third day).For easier interpretation of the results, the units needed to be transformed from minutes to hours, where appropriate.• EI: The mean EI from the EO of interest as a percentage of total daily energy intake was calculated by dividing the overall energy from the EO over n days by the total energy intake over n days."n" could be the number of weekdays (n≥2), weekend days (n≥1), or weekdays and weekend days combined (n=3), depending on the type of day.• NE: The frequency of being a night eater was calculated as the number of recorded days an individual met the criteria of NE.For weekdays and weekend days combined, four categories were created: 0 days, 1 day, 2 days, and 3 days.For weekdays, three categories were created: 0 days, 1 day, ≥ 2 days.For weekend days, two categories were created: 0 days, ≥ 1 day.
• MF: (1) The frequency of EOs, whether for the whole day or for the evening/night, as count variables, was calculated as the median number of EOs over the recorded days.
(2) The frequency of evening main meals and evening snacking were calculated as the number of recorded days consuming an evening main meal and evening snacking, respectively.Four categories were created for week/weekend: 0 days, 1 day, 2 days, and 3 days; three categories for weekdays: 0 days, 1 day, ≥ 2 days; and two categories for weekend days: 0 days, ≥ 1 day.Statistical analysis.LER variables were generated using Stata v.17.0.Histograms were used to examine the distribution of continuous variables.Continuous variables (T2, T3, E1-a/b, E2, E3) were summarised using means and standard deviations (SDs) if they were normally distributed.The count variables (F1) and non-normally distributed continuous variables were summarised using medians and interquartile ranges (IQR 25th, 75th percentile).Ordinal/binary variables (T1-a/b, E1-c1/c2, F2, F3) were reported as percentages (%).The prevalence of LER was reported by the type of day (weekdays, weekend days, and the whole week).Paired t-tests for continuous variables, Wilcoxon matched-pairs signed-rank tests for count variables, and Pearson's chi-square tests for categorical variables were used to estimate any differences between the week and the weekend.Spearman's correlations were calculated to explore inter-correlations between the 13 LER variables.

Key results
A total of 4,029 children (males 46%, females 54%) aged around 7.5 years (mean 90.4 SD 4.1 months) were included in this study.The average bedtime was 20:24, (20:00 in the week and 20:42 at the weekend).The median daily number of EOs was 6.0, which was similar to the number of conventional meals a day (breakfast, morning snack, lunch, afternoon snack, dinner, evening snack).
The distribution of continuous LER variables is shown in Figure 4, three of them are skewed (T2, time of the biggest EO; E1b, EI within 2 hours before bedtime; E3, EI for evening snacks).Thus, they were reported as medians (IQR).Table 5 describes the prevalence of LER by sex, and type of day.From the perspective of timing, more than half of children did not eat around bedtime on any of the days recorded, regardless of which bedtime (individual/average) was used.
Children were more likely to eat before going to bed on weekdays than on weekends, (38.4% and 11.4%, respectively).In addition, the time of the greatest EO in the evening was around 18:00 (6 mins later on weekends).The last EO happened at 19:12 on average (3 mins later on weekends).Sex differences were observed in timing of the last EO only, with around 6 mins later in males than females.No substantial sex differences were found in other time-related LER variables.
In terms of EI, on average, children consumed 35.8% of TDEI in the evening (after 17:00), with 27.4% of TDEI from the evening meal, and 5.5% from evening snacks.Notably, there was strong evidence that EI for the evening main meal and evening snack were greater for the week than the weekend.Additionally, more than half of children met the NE criteria on at least one recorded day, with 58.2% consuming more than 25% of TDEI within 2 hours of bedtime (NE1) and 61% consuming more than 30% of TDEI after 18:00 (NE2).However, substantially fewer children were recurrent night eaters, which were classified as maintaining NE for at least two recorded days, with prevalences of 29.2% and 26.2%, respectively, for NE1 and NE2.Sex differences were found in EI for evening snacks only, with greater EI consumed by males than females.No substantial sex differences were found in other EI-related LER variables.
The median count of daily EOs in the evening over three recorded days was 2.More than half of children consumed dinner regularly, with 59.9% consuming dinner on every recorded day.Consuming evening snacks tended to be prevalent, with 65.8% consuming evening snacks for at least two recorded days.The frequency was greater on weekdays than on weekends, for both dinner and evening snacks.Sex differences were found in the consumption of evening snacks only, which was more common in males than females.

Correlations between LER variables
Figure 5 presents the correlations between the LER variables for week/weekend.In general, most correlations were positive.The red frame highlights the correlations between variables within the same domain.Most variables were positively correlated with other variables in the same group.The variables in the timing group were all positively correlated with each other.The timing of the last EO had a strong positive correlation with eating around bedtime.It was also positively correlated with the consumption of evening snacks (both EI and frequency).Very few negative correlations were observed; most negative ones involved the time of the evening main meal (the biggest EO in the evening) which was the only variable that had an equivalent number of positive and negative correlations with other variables.The time of evening main meal was negatively correlated with the frequency of having EOs between 17:00 and 18:29 (usual dinner period), and EI in the evening.In terms of EI, EI in the evening (E1a) were positively correlated with all variables in EI and meal frequency groups, but not timing, and the correlations with EI for evening main meal/evening snacks were stronger.Conversely, EI within 2 hrs before bedtime was mainly correlated with timing-related variables.In terms of meal frequency, more frequent EOs in the evening was strongly correlated with eating at a later time, such as eating around bedtime, later last EO, and more evening snacks (EI for evening snacks and frequency of evening snacks).
The correlation matrix for LER variables for the week and the weekend are presented in Figure 6 and Figure 7, respectively.Overall, the inter-correlations between the 13 LER variables did not differ substantially between weekday and weekend.Despite the slight differences in the strength of the correlations, they always pointed in the consistent direction.
In summary, most LER variables were positively correlated with each other within the same main group (timing, EI, and meal frequency).Stronger and more consistent intercorrelations were observed in the groups of timing and meal frequency.On the other hand, some of the variables showed strong correlations across groups.Stronger positive associations were found between timing of last EO (T3), EI/frequency for evening snack (E3, F3), and meal frequency in the evening (F1).The time of the evening main meal was involved in most of the negative associations, especially with LER variables in the EI group.For example, a later time of the evening main meal was found to be associated with smaller EI for evening main meal and a lower frequency of evening snacks.

Strengths and limitations of the data
There are a number of strengths including the use of 3-day food diaries for collecting dietary data, ensuring accurate and detailed recording of food and beverage consumption.Notably, as emphasized in the current 2020 Dietary Guidelines Advisory Committee Report 43 , to capture habitual/customary eating behaviours, it is vital to collect dietary data for multiple days.Dietary records, also known as food diaries or food records, are often regarded as the "gold standard" against which other dietary methods can be assessed 44,45 .Participants are asked to prospectively record the exact time of foods and beverages, and the amounts of each consumed over one or multiple days.In addition, it enabled the type of day (week; weekend) to be recorded, which is important information as eating habits are very likely to be different between the week and weekend [46][47][48] .Since food records are completed while foods are consumed, information on food consumption could be captured more accurately and less likely to be omitted than other methods that are not.However, according to the recent systematic review investigating the association between LER and adiposity in children, very few studies (5/47) used food diaries for at least three days 28 , due to relatively large respondent burden and high personnel costs 49 .Despite this, a large number of participants were included in the current study.To address the inconsistent definition of LER used in previous literature, this study applied a large set of comparable and operative   * n=3303 on week/weekend days; n=2795 on weekdays; n=2790 on weekend days due to missing data in bedtime.
† Detailed definitions of LER variables can be found in Table 4.  Key limitations of the study include the use of a subsample of children who had completed food diaries, however those coded for exact times were randomly selected and no biases were introduced.The definitions of LER are contingent upon the availability of dietary and sleep-related data.In addition, bedtime was self-reported by parents as the usual bedtime and was not recorded on the same days as the food diaries.This may lead to two problems: 1) the observed bedtime-related variables, such as eating around individual bedtime and eating within 2 hours before bedtime, may not be as accurate as their names implied, and 2) we assumed that children were unlikely to get up and eat after their bedtime, however the reported usual bedtime was earlier than the time of last EO in a substantial group of children (23.7%).Another limitation is that the prevalence of dinner skipping may be overestimated due to the inaccuracy of usual dinner time (17:00-18:29).Children who consumed dinner after 18:29 were therefore counted as dinner skippers.However, this usual dinner time was determined based on the peaks of the distribution of EI during the evening (see Figure 2), thus it is believed that this would be the most appropriate way, similar to previous studies 50 , to approximate "dinner skipping" when it was unlikely to capture the participants' perception of dinner using food diaries.
Any method of dietary data collection is subject to a variety of biases, affecting both reliability and validity.The ideal method of measuring dietary intake is via direct observation with all foods weighed and measured before and after consumption.However, this is intrusive for the participant and both expensive and resource intensive for the researcher.Food diaries are therefore often regarded as the "gold standard", however, they can lead to relatively large respondent burden and high personnel costs 49 .In addition, they can be biased by the selection of the study sample and by the completion of the number/type of days recorded 51 .For example, participants who have low literacy and motivation are not likely to cooperate in completing multiple food records 52 .Thus, it is highly recommended that the number of recorded days does not exceed four, as the proportion of incomplete data is believed to increase considerably as additional days of records are kept, due to respondent fatigue 45 .Additionally, participants who do comply may differ systematically from those who do not.
According to a study which validated seven day diaries, the validity of the information collected was often seen to decrease in the latter two days of the recording period, compared to the data collected in the earlier days 53 .The best way to assess validity of a food diary is to simultaneously observe food intake -this was not feasible in ALSPAC given the size of the cohort.However, validity of the three day food diaries used in ALSPAC has been examined by comparing the estimated nutrient intakes with those observed in the UK's National Diet and Nutrition Survey (NDNS); when comparing nutrient intakes from ALSPAC with the 7-10 year old children from the 1997 NDNS survey, very similar results were reported 54 .The same dietary diaries were also used at younger ages in the ALSPAC cohort.Estimated nutrient intakes have been shown to correlate highly with relevant blood markers (e.g.omega-3 fatty acids) which provide objective measures for comparison 55 , suggesting that the ALSPAC dietary diaires provide valid estimates of general dietary intake.
To enhance the accuracy of measuring meal-related variable and comparability with other dietary measures, we would recommend that future studies using food diaries include bedtime and participant-identified meal types for each EO.Specifically, food diaries could be improved by asking participants to label each eating occasion (EO) with a self-defined meal type (e.g., breakfast, lunch, dinner or snack).

Ethical approval and consent
Ethical approval for the study was obtained from the ALSPAC Ethics and Law Committee and the Local Research Ethics Committees.Informed consent for the use of data collected via questionnaires and clinics was obtained from participants following the recommendations of the ALSPAC Ethics and Law Committee at the time.Study participants have the right to withdraw their consent for elements of the study or from the study entirely at any time.Full details of the ALSPAC consent procedures are available on the study website (http://www.bristol.ac.uk/alspac/researchers/research-ethics/).

Data availability
Underlying data ALSPAC data access is through a system of managed open access.The steps below highlight how to apply for access to the data included in this data note and all other ALSPAC data: 1. Please read the ALSPAC access policy (www.bristol.ac.uk/ media-library/sites/alspac/documents/researchers/data-access/ ALSPAC_Access_Policy.pdf) which describes the process I confirm that I have read this and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Yangbo Sun
Preventive Medicine, The University of Tennessee Health Science Center, Memphis, Tennessee, USA Thanks for providing the revised manuscript.It is much approved and address most of the questions.One minor comment: it is still unclear about the description/comment/discussion on the reliability and validation on the measures of food dairies.some other data was used as the inclusion criteria for 1,834 diaries, data were these?Why? Was the number of diaries chosen randomly steered by power calculations?More details about the selection of the subsample would be appreciated.Table 3: I am a bit confused with Table 3, which is referred to define eating occasions but also LER variables.Should the latter refer to Table 4? 2.
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Nutritional epidemiology, public health nutrition, nutrition and wellbeing in families I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Yangbo Sun
Preventive Medicine, The University of Tennessee Health Science Center, Memphis, Tennessee, USA Thanks for the opportunity to review this interesting paper.The authors used ALSPAC dataset to describe the late eating patterns among children.I have a few suggestions for the authors to consider.
It is unclear about the validity of the food diaries collection in the study 1.
It will be more informative if the authors can compare the children included in this study and those excluded 2.
the authors did not compare their findings with other studies, which will provide more information 3.

Are the datasets clearly presented in a and accessible format? Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: nutritional epidemiology I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Henna Vepsäläinen
Department of Food and Nutrition, University of Helsinki, Helsinki, Finland Thank you for the opportunity to review this carefully written data note, which describes the formation of variables used to assess later eating rhythm among children.However, I hope that the authors will be willing to clarify some parts of the paper to make it easier to follow, which will in turn help other researchers in using the same types of variables, where applicable.I also wish for a few sharp recommendations for future research.Please find more detailed questions and comments below.
Figure 1 + related text: Figure includes information that is not explained in the text -what is the justification for selecting only a share of the food diaries for investigating the timing of consumption?I understand that the procedure was burdensome and this would be a perfectly acceptable reason for not including all diaries, but maybe you performed a sample size calculation to estimate how many diaries would be needed to explore specific research questions and used that as a justification for including ~4000 diaries or something like that?
In addition, it is somewhat unclear what some of the Ns in the figure represent (for example N=4029; 4063; 3757) since the text explaining this comes later.Maybe the figure could be moved later? 1.
Later along in the text, it is mentioned that "Among these, 1,834 were selected based on the availability of other data, while 3,035 were selected randomly."Based on this, I would remove the word "randomly" from Figure 1, as this was not the case for all of the 4869 diaries.

2.
Table 2: There probably is no universal definition for a breakfast (is it a main meal or more like a snack), but using the definition in Table 2, is it possible that for example breakfast and lunch are within the same time slot (between 6:00 and 11:59), or lunch and dinner (between 12:00 and 16:59)?This would then result in some of the "main meals" to be coded as snacks 3.
-how would that effect the results and is there any solution that would prevent this from happening?Table 3: This table is nicely compiled.However, some of the are a bit unclear, for example T1a: is the final variable a continuous or categorical variable, if it is categorical, which are the categories?From the abstract it can be deducted that this is a binary variable, maybe a column indicating the type of the variable could be added?

4.
Night eating (NE) is mentioned in the text and Table 3, but not in the abstract.For the sake of clarity, the variable names should remain the same throughout the text, although I am not sure whether night eating is an appropriate term when the participants are children and the authors have also touched upon this topic in the introduction.Later also recurrent night eating is mentioned -should this be included in the definitions of Table 3? 5.
Table 4: Even though the variables have been explained in Table 3, it should also readable on its own, and thus, I recommend that the variables are briefly defined in this table as well (for example, what does "Eating around individual bedtime" mean).

6.
Tables 5, 6 and 7: I am not sure if the color coding (and the red frame referred to in the footnote) is working as designed, for me it seems like there is normal table with correlation coefficients.

7.
Referring to question #3, would you recommend researchers to use participant-defined meals or to categorize meals according to time slots?Why?Some kinds of recommendations could be incorporated into the Strengths and limitations of the data to steer future research.

8.
The authors state that, due to the randomization, it is unlikely that bias was introduced.This could be easily checked by comparing those included (randomized into the current study) and excluded, if appropriate data is available (for example sex, parental education or income etc.).

9.
It is mentioned that dietary records are regarded as the "gold standard" and that to capture habitual/customary eating behaviours, it is vital to collect dietary data for multiple days.For some food groups that are no that often consumed, such as fish, three days may not be enough, and some other dietary assessment methods may be better in assessing those.However, this study focused solely on the timing of eating.It would have been nice to read how well the authors think a 3-day diary is able to capture the timing of eating?Is three days enough and if so, why?How many days you recommend should be recorded?10.

Is the rationale for creating the dataset(s) clearly described? Yes
Are the protocols appropriate and is the work technically sound?Yes

Are sufficient details of methods and materials provided to allow replication by others? Partly
Are the datasets clearly presented in a useable and accessible format?Partly Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Nutritional epidemiology, public health nutrition, nutrition and wellbeing in families I confirm that I have read this submission and believe that I have an appropriate level of expertise to state that I do not consider it be of an acceptable scientific standard, for reasons outlined above.

Figure 1 .
Figure 1.The sample size flow diagram for investigating later eating rhythm by the type of day.

29 Figure 2 .
Figure 2. Summary of the mean energy intake for every 30-min time interval across the day and the distribution of bedtime by the type of day in children at age 7 years in ALSPAC.*Abbreviations: ALSPAC, Avon Longitudinal Study of Parents and Children; TDEI, total daily energy intake; EI, energy intake.
from a meal or time frame kcal Mean EI from an EO or time frame TDEI kcal

Figure 3 .
Figure 3. Percentage of TDEI for different meals/ time periods by weekdays or weekend days in the evening/night in children at age 7 years in ALSPAC.*Abbreviations: ALSPAC, Avon Longitudinal Study of Parents and Children; TDEI, total daily energy intake; EI, energy intake.

Figure 4 .
Figure 4.The distribution of continuous LER variables for week/weekend in ALSPAC.*Abbreviations: ALSPAC, Avon Longitudinal Study of Parents and Children; LER, Later eating rhythm.

Figure 5 .
Figure 5. Correlations between variables describing later eating rhythm in the whole week in ALSPAC.

Figure 6 .
Figure 6.Correlations between variables describing later eating rhythm on weekdays in ALSPAC.

Figure 7 .
Figure 7. Correlations between variables describing later eating rhythm on weekend days in ALSPAC.

Version 2 Reviewer
Report 15 August 2024 https://doi.org/10.21956/wellcomeopenres.25185.r91323© 2024 Sun Y.This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Version 1 Reviewer
Report 08 May 2024 https://doi.org/10.21956/wellcomeopenres.22808.r75404© 2024 Sun Y.This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reviewer Report 17
April 2024 https://doi.org/10.21956/wellcomeopenres.22808.r76725© 2024 Vepsäläinen H.This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Table 4 . Definitions of potential later eating rhythm variables in ALSPAC.
Abbreviations: EO, eating occasion; EI, energy intake; TDEI, total daily energy intake; NE, night eating.T1: Eating at a late time at night (i.e., a. individual usual bedtime; b. average bedtime).E1: EI from a time frame in the evening/night (i.e., a. specific clock time; b. within 2 hours before bedtime; c. categorical variable derived from continuous variables in E1 to define *NE, including: c1-NE1 and c2-NE2).NE: Night eating was initially defined as eating more after a late time by previous studies, our study adapts the criteria for UK children in ALSPAC.

b Male n=1849 Female n=2170 Total n=4019 Difference in means/ proportions (95%CI) p-value a Male n=1570 Female n=1815 Total n=3385
Wilcoxon matched-pairs signed-rank test; and b3, paired t-test.c: very few children (<4%) ate around bedtime every day (3 days), thus children eating around bedtime for 2 days and 3 days were grouped together.