Glucose Metrics During School Hours in Primary School‐Aged Children With Type 1 Diabetes: A Population‐Based Cohort Study Authors: Stephen Paull, Adele Connor, Grant Smith, Caleb Jones, Keren Provost, Katrina Ellis, Elizabeth A. Davis, Mary B. Abraham, Craig E. Taplin Published: 2026 ABSTRACT ABSTRACTAimSchool time represents a significant component of overall glycaemia for children with type 1 diabetes (T1D), and glucose levels during instructional time may be important for optimising academic progress. There is, however, limited literature regarding glycaemia during school hours. This study aimed to evaluate glucose levels during school in primary school‐aged children with T1D in Western Australia (WA) and to compare these with non‐school days.MethodsA 12‐week cross‐sectional cohort study was performed comprising a WA school term and subsequent school holidays, utilising demographic, clinical and continuous glucose monitor data from the WA Children's Diabetes Database for primary school children with > 1 year of diagnosis of T1D.ResultsA total of 235 children, mean age 9.54 ± 1.67 years and duration of T1D of 4.08 ± 2.31 years met inclusion criteria. Children had higher mean ± standard deviation time in range (3.9–10.0 mmol/L) during school (58.0% ± 15.8%) than equivalent hours during school holidays (53.5% ± 18.8%, p < 0.001) and weekends (56.6% ± 17.3%, p = 0.006). Children had lower mean sensor glucose, SD sensor glucose and time above range (> 10.0 and > 13.9 mmol/L) during school compared to school holidays and weekends. Time below 3.9 mmol/L was higher during school time compared with weekdays during school holidays (p = 0.024), but lower than on school term weekends (p = 0.009), and < 4% in all periods. There was no significant difference in time below 3.0 mmol/L.ConclusionsPrimary school‐aged children in WA had modestly improved glycaemia during school compared with weekends, with more pronounced improvements compared with school holidays. Introduction Type 1 diabetes (T1D) management at school is important for safety and optimal education outcomes. Furthermore, school time represents a significant component of overall glycaemia; Australian school children spend 40 weeks of the year at school for approximately 6 h per day. Primary school children (generally 5–12 years of age) are especially reliant on school staff for diabetes management support, and thus glycaemia at school. The scientific literature regarding the association between school instructional time and glucose levels in children with T1D suggests there may be a difference between school and non‐school contexts, but findings are mixed and not directly comparable [1, 2, 3, 4, 5, 6, 7]. In a continuous glucose monitor (CGM)‐based study of children aged 7–12 years, time in range (TIR) was overall higher across the same daytime hours during weekends and COVID pandemic‐related virtual schooling compared to in‐person schooling, though glycaemia improved as the school day progressed [1]. In contrast, an earlier study using intermittent finger‐prick testing in children and adolescents 7–19 years showed higher mean blood glucose levels during school holidays, compared with school days [2]. Other school‐related studies have evaluated less direct methods, with one reporting higher glycated haemoglobin (HbA1c) levels in elementary (primary) school children during school term compared to holidays, although this could reflect glucose control outside of school hours [3]; and another finding that online school participation may be associated with poorer glycaemic control and lower pump use [4]. Broader temporal studies also suggest glycaemia varies across routines and time periods, with a mixed‐age CGM study showing poorer glycaemic control around holidays and during winter months [5]; while two Italian CGM studies conducted during COVID‐19 lockdowns demonstrated overall improved glycaemia in paediatric (5–19 years) and mixed‐age cohorts with T1D, compared with pre‐ and post‐lockdown periods [6, 7]. Overall, the literature is heterogeneous because studies differ in age group, design, outcome measure, time periods studied and with some during pandemic lockdowns. There remains a need for CGM‐based data examining routine in‐person school hours in a non‐lockdown setting, compared with equivalent non‐school time. In the state of Western Australia (WA), school staff are provided with education to support students with T1D at school via the national Diabetes in Schools program [8], co‐developed by Diabetes Australia with multiple stakeholders [9], which provides nationally consistent information and training. All school children in WA are provided with personalised school diabetes management plans, which summarise the child's insulin regimen, glucose testing, target glucose range, hypoglycaemia and hyperglycaemia management, physical activity management and the child's level of independence and support for specific diabetes management tasks. All WA schools may apply for individualised practical training to facilitate the plan authorised by the medical team. Hospital schoolteachers based at Perth Children's Hospital maintain links for communication and planning between health care and school settings. School nurses are not consistently present in WA schools, with the Diabetes in Schools program building staff capacity and student inclusion. The primary aim of this study was to evaluate glycaemia in primary school‐aged children with T1D in WA during routine in‐person school hours and to compare glycaemia during school hours with the equivalent hours on weekends during the school term, and on weekdays during school holidays. Methods An observational study with a cross‐sectional cohort design was conducted comprising a WA school term and the subsequent school holidays. All children with T1D in WA are managed by a single statewide diabetes centre based at Perth Children's Hospital, with clinical data collected prospectively and recorded on the Western Australian Children's Diabetes Database (WACDD). Coordination of the Diabetes in Schools program for WA students is embedded within this service. Children with T1D with diabetes duration greater than 1 year, born between 1 July 2010 and 30 June 2016 (WA Department of Education birth dates corresponding to first—sixth school grade at the time of the study), with CGM data available during the predefined timeframe were included in the study. CGM data of ≥ 70% during the study period was required to be included for analysis [10]. All demographic and clinical data were obtained from the WACDD. School hours were defined as 0900 to 1500 from Monday to Friday. The study period was from 18 July to 9 October 2022. The school term was defined as the 10 weeks from 18 July to 23 September 2022 (term three of that WA school year), excluding a national public holiday on 22 September 2022. School holidays were defined as the 2 weeks from 24 September to 9 October 2022. Variables analysed included sex, age, duration of T1D, metropolitan/rural postcode, socioeconomic status using the Australian Bureau of Statistics Socio‐Economic Indexes for Areas (SEIFA) [11], CGM data, insulin regimen (injections vs. continuous subcutaneous insulin infusion [CSII] or hybrid closed‐loop [HCL] therapy), individualised diabetes training for school staff (prior to or during study period), HbA1c (earliest recorded from 25 April to 9 October 2022) and number of episodes of severe hypoglycaemia (requiring third party assistance, such as seizures or altered conscious state). Metrics analysed were TIR (3.9–10.0 mmol/L), time below range (TBR) (< 3.9 and < 3.0 mmol/L), time above range (TAR) (> 10.0 and > 13.9 mmol/L), mean sensor glucose and standard deviation (SD) sensor glucose [10, 12]. For each participant, CGM metrics were calculated for school hours and outside of school hours on weekdays and weekends during both the school term and school holidays. Means with SDs or medians with interquartile range are presented depending on the distribution of the metric. To compare CGM metrics over time periods of interest, heteroscedastic linear mixed models including terms for weekday, school term and their interaction and a random intercept for participant were conducted. All models adjusted for available covariates known or suspected to influence glycaemic control (age, duration of diabetes, sex, insulin regimen, CGM model, sociodemographic quintile of postcode of residence and urban/non‐urban residence) and separate models were run for school‐hour observations and non‐school‐hour observations. Estimated marginal means were calculated for levels of weekday × school term and planned pairwise comparisons across time periods of interest were conducted. Where model assumptions were not met, Wilcoxon signed rank tests comparing CGM metrics across the pair of time periods were conducted. Data were cleaned and analysed using R Statistical Software (v4.3.2) [13] and RStudio (version 2023.12.1.402) [14]. A p value of < 0.05 was considered statistically significant; no adjustments were made for multiple comparisons. Results A total of 332 children with > 1 year of T1D were identified; 97 were excluded (78 with no available CGM data due to issues with access, syncing or sharing of data; 19 with insufficient CGM data). Hence, the study cohort included 235 children (53% males, mean ± SD age 9.5 ± 1.7 years and diabetes duration 4.1 ± 2.3 years). HbA1c was available for 158 children (67.2%), with a mean ± SD of 7.4% ± 1.0% (57 ± 11 mmol/mol). Demographic characteristics of the entire primary school‐aged WA population with T1D and those included in this analysis are described in Table 1. Comparison between the 235 included children and the 97 excluded children indicated that the groups were broadly comparable; however, compared to the excluded students, a lower HbA1c among the included participants was noted (8.0% vs. 7.4% [64 vs. 57 mmol/mol], respectively, p < 0.001). TABLE 1Demographic characteristics of primary school‐aged children with > 1 year of diagnosis of type 1 diabetes and those meeting inclusion criteria.CharacteristicOverall cohort (n = 332)Study cohort (n = 235) p Male, n (%)177 (53.3)130 (55.3)0.308Age (years), mean (SD)9.54 (1.71)9.54 (1.67)0.923Duration of T1D (years), mean (SD)4.24 (2.33)4.08 (2.31)0.052Location, n (%)0.818Metropolitan258 (77.7)184 (78.3)Rural/remote71 (21.4)49 (20.9)Missing data3 (0.9)2 (0.9)SEIFA index, n (%)0.3461 (most disadvantaged)11 (3.3)7 (3.0)262 (18.7)41 (17.4)378 (23.5)47 (20.0)491 (27.4)66 (28.1)5 (most advantaged)87 (26.2)72 (30.6)Missing data3 (0.9)2 (0.9)Regimen, n (%)HCL59 (17.8)53 (22.6)CSII (non‐HCL)142 (42.8)103 (43.8)Injections129 (38.9)78 (33.2)Missing data2 (0.6)1 (0.4)Individualised face‐to‐face school training n (%)203 (61.1)148 (63.0)0.346HbA1c< 0.001%, mean (SD)7.5 (1.2)7.4 (1.0)mmol/mol, mean (SD)59 (13)57 (11)Missing data, n (%)118 (35.5)77 (32.8)Episodes of severe hypoglycaemia n (%)0 (0)0 (0)Abbreviations: CGM, continuous glucose monitor; CSII, continuous subcutaneous insulin infusion; HbA1c, glycated haemoglobin; HCL, hybrid closed‐loop; SD, standard deviation; SEIFA, socio‐economic indexes for areas; T1D, type 1 diabetes. Demographic characteristics of primary school‐aged children with > 1 year of diagnosis of type 1 diabetes and those meeting inclusion criteria. Abbreviations: CGM, continuous glucose monitor; CSII, continuous subcutaneous insulin infusion; HbA1c, glycated haemoglobin; HCL, hybrid closed‐loop; SD, standard deviation; SEIFA, socio‐economic indexes for areas; T1D, type 1 diabetes. Glucose metrics from 0900 to 1500 h on school days and weekends during the school term, and weekdays during the school holidays are summarised in Table 2. TABLE 2Glycaemia of included participants from 9 AM to 3 PM by type of day.Mean (SD) or median (Q25, Q75) p (compared to school term, weekdays)% TIR: 3.9–10.0 mmol/L a Term, weekdays58.0 (15.8)ReferenceTerm, weekend56.6 (17.3)0.006Holidays, weekdays53.5 (18.8)< 0.001% TBR: < 3.9 mmol/L b Term, weekdays2.34 (1.04, 4.24)ReferenceTerm, weekend2.63 (1.09, 4.75)0.009Holidays, weekdays1.66 (0.49, 4.15)0.024% TBR: < 3.0 mmol/L b Term, weekdays0.34 (0.09, 0.79)ReferenceTerm, weekend0.45 (0.08, 0.98)0.040Holidays, weekdays0.17 (0.00, 0.70)0.072% TAR: > 10.0 mmol/L a Term, weekdays39.0 (17.0)ReferenceTerm, weekend39.9 (19.0)0.039Holidays, weekdays43.5 (20.7)< 0.001% TAR: > 13.9 mmol/L b Term, weekdays11.0 (5.5, 20.5)ReferenceTerm, weekend12.0 (6.5, 22.0)< 0.001Holidays, weekdays15.0 (5.6, 27.0)< 0.001Mean sensor glucose (mmol/L) a Term, weekdays9.55 (1.68)ReferenceTerm, weekend9.71 (2.07)0.001Holidays, weekdays10.01 (2.20)< 0.001SD sensor glucose (mmol/L) b Term, weekdays3.50 (3.06, 4.07)ReferenceTerm, weekend3.58 (3.03, 4.21)< 0.001Holidays, weekdays3.53 (2.99, 4.28)0.013Abbreviations: Q25, 25th quartile; Q75, 75th quartile; SD, standard deviation; TAR, time above range; TBR, time below range; TIR, time in range. a Mean (SD) presented, p values calculated based on linear mixed model. b Median (25th quartile, 75th quartile) presented, p values based on Wilcoxon signed rank tests. Glycaemia of included participants from 9 AM to 3 PM by type of day. Abbreviations: Q25, 25th quartile; Q75, 75th quartile; SD, standard deviation; TAR, time above range; TBR, time below range; TIR, time in range. Mean (SD) presented, p values calculated based on linear mixed model. Median (25th quartile, 75th quartile) presented, p values based on Wilcoxon signed rank tests. During the school term, weekday control was improved compared with weekends. During school days, participants had higher mean TIR (58.0% ± 15.8% vs. 56.6% ± 17.3%, p = 0.006), with lower mean sensor glucose (9.55 ± 1.68 vs. 9.71 ± 2.07 mmol/L, p = 0.001), lower median time in hypoglycaemia TBR 3.0 (0.34% [Q25 0.09, Q75 0.79] vs. 0.45% [Q25 0.08, Q75 0.98], p = 0.040) and TBR 3.9 (2.34% [Q25 1.04, Q75 4.24] vs. 2.63% [Q25 1.09, Q75 4.75], p = 0.009) and lower time in hyperglycaemia TAR 10.0 (mean 39.0% ± 17.0% vs. 39.9% ± 19.0%, p = 0.039) and TAR 13.9 (median 11.0% [Q25 5.5, Q75 20.5] vs. 12.0% [Q25 6.5, Q75 22.0], p < 0.001) compared with weekends during the school term. When evaluating just weekdays, glycaemia during the school term was improved compared with school holidays. During school days, participants had higher mean TIR (58.0% ± 15.8% vs. 53.5% ± 18.8%, p < 0.001) with lower mean sensor glucose (9.55 ± 1.68 vs. 10.01 ± 2.20 mmol/L, p < 0.001) and lower time in hyperglycaemia TAR 10.0 (mean 39.0% ± 17.0% vs. 43.5% ± 20.7%, p < 0.001) and TAR 13.9 (median 11.0% [Q25 5.5, Q75 20.5] vs. 15.0% [Q25 5.6, Q75 27.0], p < 0.001) compared with weekdays during the school holidays. While there was a small increase in median TBR 3.9 (2.34% [Q25 1.04, Q75 4.24] vs. 1.66% [Q25 0.49, Q75 4.15], p = 0.024), there was no evidence for a difference in median TBR 3.0 between the two periods (0.34% [Q25 0.09, Q75 0.79] vs. 0.17% [Q25 0.00, Q75 0.70], p = 0.072). Figure S1 shows a comparison of TIR, TBR 3.9 and TAR 10.0 on weekdays during 0900 to 1500 between school term and school holidays. Table S1 describes the overall glucose metrics across all 24 h and Table S2 describes glycaemia outside of school hours from 3 PM to 9 AM on school days, weekends during school term and weekdays during school holidays. There was a small but statistically significant higher mean TIR and lower TAR across the entire 24‐h period on school days. These also demonstrate higher time in hypoglycaemia during the school term, more noticeable over 24 h than the difference in TBR during school hours only. Table S3 shows glycaemia on school days across time periods within the 24‐h day. Table S4 shows the comparison of glycaemia during school hours, stratified by insulin regimen, demonstrating higher TIR and lower TAR, mean sensor glucose and SD sensor glucose in children with HCL or CSII compared with those on subcutaneous injections. Stratification of the model effects by regimen showed no evidence that the differences between school and non‐school periods differed based on regimen, as shown in Table S5. Discussion The primary finding of this study was that primary school‐aged children in WA had modestly improved glycaemia during school hours of 9 AM to 3 PM compared with the equivalent hours on weekends and a more pronounced improvement compared with school holidays. These findings somewhat contrast with those of March et al., who found that school‐aged children in Pittsburgh, USA, had improved glycaemia during weekends and virtual school compared to in‐person school [1], but are similar to those of Belsare et al., who found that children and adults with predominantly T1D have poorer glycaemic control during holiday periods [5]. Our findings might be explained by several factors including the structure and routine of the school day, adult supervision of diabetes care at school and/or increased physical activity [15, 16, 17]. The availability of personalised school diabetes management plans, face‐to‐face training for individual staff members involved in a child's personal care, and consistency of student support by school staff may all enable the conditions for constructive routines and diabetes management during school in WA, and may be different to the support provided in other states and countries. Differences in home and weekend routines across countries and cultures may also vary. Additionally, the impact of the COVID pandemic was felt differently across the world but is likely important to consider. It is notable that in Pittsburgh, students started with higher TAR (8–10 AM and likely reflective at least partially of management before school commencement) but glycaemia then improved significantly as the school day progressed independent of insulin regimen, age, duration or HbA1c. Importantly, this pattern was not seen for home school or weekends: the authors suggest in their commentary that this indeed may be due to school routines, supports and structures that are possibly beneficial for children with T1D. The current study also demonstrated higher glucose levels during weekends and holidays, which might be associated with variable lifestyle patterns related to nutrition, physical activity and/or sleep routines. While participants had slightly higher TBR 3.9 during school compared with equivalent time periods during school holidays, TBR at school was below 4% [10, 12], the difference equates to 2.5 min per school day below 3.9 mmol/L, and no episodes of severe hypoglycaemia were reported during this period. CGM analysis focusing on school‐aged children and the effect of school‐time provides important information on glucose levels in school and outside school, the potential value of targeted education for families when planning for holidays and suggests the case for consistent routines during weekends and non‐school weeks which may benefit children with T1D to maintain optimal glycaemia. Strengths of this study include that it is the first Australian study to evaluate glycaemia in school‐aged children and to demonstrate a difference in glycaemic outcomes between school and non‐school time. CGM data provides an in‐depth overview of glucose metrics, and the study population is from a population‐based, state‐wide sample. Limitations include the number of children with no available (n = 78) or insufficient (n = 19) CGM data, and the period studied which, while representing an entire school term, consists of just one quarter of the entire school year. The school holiday weeks studied may not be representative of other holiday periods, such as an extended summer break, while the school term studied may not be generalisable to other times of year, such as the return from summer holidays where new staff may not be yet familiar with diabetes management, or other potentially different variables such as weather or routines. Absence from school was not able to be accounted for, such as from illness or pupil‐free days. Data on home schooling, diet and physical activity were also not available. This study also occurred during the later stages of the global COVID‐19 pandemic. WA was less impacted by lockdowns than other states in Australia, and children were still able to attend school during 2022 [18], although the impact of COVID‐19 on weekend activities cannot be deciphered. Finally, by excluding participants and time periods with < 70% wear time and participants with no available CGM data, it is possible that those with less optimal glycaemia have been excluded [19, 20]. Further research in this area should include a longer period of analysis, include adolescents and those in secondary school, and the relationship between individualised school training and glycaemia. Nonetheless, these results are likely generalisable to other Australian schools, and potentially similar‐aged children internationally. Conclusions In conclusion, primary school‐aged children in WA on CGM have modestly improved glycaemic control during school hours when compared with the equivalent time periods on weekends, and more pronounced improvements compared with weekdays during school holidays; specifically, higher TIR, lower TAR and lower mean glucose while still achieving low levels for time in hypoglycaemia at school. Author Contributions Stephen Paull: conceptualisation, project administration, methodology, investigation, formal analysis, validation, writing – original draft. Adele Connor: data curation, investigation, writing – review and editing. Grant Smith: software, data curation, investigation, formal analysis, validation, visualisation, writing – review and editing. Caleb Jones: validation, writing – review and editing. Keren Provost: validation, writing – review and editing. Katrina Ellis: conceptualisation, data curation, writing – review and editing. Elizabeth A. Davis: conceptualisation, writing – review and editing. Mary B. Abraham: conceptualisation, validation, supervision, writing – review and editing. Craig E. Taplin: conceptualisation, validation, supervision, writing – review and editing. Funding This work was partially supported by Breakthrough T1D Australia (grant number 5‐SRA‐2021‐1088‐M‐X). Ethics Statement This study protocol was reviewed and approved by Child and Adolescent Health Service Human Research Ethics Committee, approval number RGS0000006408. Consent This study was granted an exemption from requiring written informed consent by Child and Adolescent Health Service Human Research Ethics Committee. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Figure S1: Box and whisker plots with median and interquartile range for time in range 3.9–10.0 mmol/L, time below range < 3.9 mmol/L and time above range 10.0 mmol/L of included participants, 0900–1500 on weekdays (school term vs. school holidays). Table S1: Glycaemia of included participants for 24 h by type of day. Table S2: Glycaemia of included participants during 1500–0900 (outside of school hours) by type of day. Table S3: Glycaemia of included participants on school days (weekdays during school term) by period of day. Table S4: Glycaemia of included participants during school hours, stratified by insulin regimen. Table S5: Stratification of model effects by regimen.