Bedtime snacking and glycemic deterioration in young children with Type 1 diabetes on multiple daily injections: a randomized controlled crossover trial

Abstract

Objective

To determine if a bedtime snack in young children with type 1 diabetes (T1D) prevents nocturnal hypoglycemia, and the impact on glycemia overnight.

Methods

In this randomized controlled crossover trial, 10 grams of carbohydrate (milk, yoghurt, and kefir) was given 150–180 minutes after dinner over three nights to 5–8-year-old children with T1D using multiple daily injection therapy. Continuous glucose monitoring (CGM) data were collected for 6 hours following the snacks on one control and three snack nights. Time in 70–180 mg/dL (3.9–10 mmol/L) range (TIR), time below 70 mg/dL (3.9 mmol/L) (TBR), and other metrics were analyzed according to international CGM consensus. Trial day was terminated if blood glucose exceeded 300 mg/dL (16.7 mmol/L) or fell below 70 mg/dL (3.9 mmol/L).

Results

Of 28 children (13 female, mean age 6.6 ± 0.8 years, HbA1c 7.0 ± 0.5% (53 mmol/mol)), mean glucose values before the test snacks were 137.8 ± 14.5 mg/dL (7.7 ± 0.8 mmol/L) for milk, 141.9 ± 16.9 mg/dL (7.9 ± 0.9 mmol/L) for yoghurt, 136 ± 19.1 mg/dL (7.6 ± 1.1 mmol/L) for kefir, and 140.8 ± 17.0 mg/dL (7.8 ± 0.9 mmol/L) for control without significant difference (p = 0.548). TIR was 34.7% for milk, 38.7%. for yoghurt, 45.9% for kefir, and 75.5% for control during the 6-hour post snack period, with TIR on the control day significantly higher than the three snack days (p < 0.001). TBR did not differ by group (p > 0.05). Of 112 trial days, 13 days were terminated due to hyperglycemia (>300 mg/dL) (16.7 mmol/L) (8 milk, 4 yoghurt, 1 kefir), and 3 trial days due to hypoglycemia (<70 mg/dL) (3.9 mmol/L) (1 yoghurt, 2 control).

Conclusion

Bedtime snacking in young children with T1D impairs nocturnal glycemia and reduces TIR, without decreasing the frequency of hypoglycemia.

Introduction

Bedtime snacking has been practiced to prevent nighttime hypoglycemia in the management of type 1 diabetes (T1D), especially in young children [1]. While nocturnal hypoglycemia was a significant problem with intermediate-acting insulin (NPH) therapy due to its peak effect, it has been reduced by the introduction of long-acting basal insulin such as insulin detemir and insulin glargine [2]. This shift in the treatment has led to questioning of this regular practice regarding the need for a bedtime snack [3]. While the benefits of bedtime snacking to prevent nocturnal hypoglycemia remain controversial in multiple daily injection therapy [4, 5], the International Society for Pediatric and Adolescent Diabetes (ISPAD) does not mandate bedtime snacking and recommends individual tailoring [6]. Similarly, the American Diabetes Association (ADA) suggests individualized targets and nutritional interventions for young children due to their vulnerability to nocturnal hypoglycemia [7].

The controversial use of a prescribed bedtime snack to avoid hypoglycemia in young children using multiple daily injection therapy entails an assessment of its impact on nocturnal glycemic outcomes. To our knowledge, no study has prospectively investigated the change in nocturnal glycemia after prescribed bedtime snacks in children with T1D. Therefore, there is a knowledge gap in the literature regarding this common clinical practice, which is frequently implemented by health professionals with no evidence. This controlled prospective study was conducted to investigate the change in nocturnal glycemia within 6 hours after administration of milk, yoghurt, and kefir.

Methods

Participants

This randomized crossover study was conducted on children aged 5–8 years with T1D who were attending Koç University, Department of Pediatric Endocrinology and Diabetes, Istanbul, Turkey. Inclusion criteria were diagnosis of T1D, receiving multiple daily injections (MDI) (≥4 injections/day) treatment for at least 1 year, age of 5-8 years inclusive, last HbA1c level < 8% (64 mmol/mol), and using the same CGM system (Freestyle Libre, Abbott, USA). Children who had any chronic disease other than T1D (celiac disease, kidney disease, cystic fibrosis, food allergy, eating behavior disorder, etc.), any of the chronic complications of diabetes, or a daily insulin dose of ≤0.5 IU/kg/day (children in the honeymoon period) were excluded from the study. An overview of the flow of participants through the study is shown in Fig. 1. Since there were 133 children aged 5–7 years with Type I diabetes mellitus followed up in our department and only 40 of these children were accessible, the sample size of the study was calculated as 28, with a Type 1 error α = 0.05 and an acceptable error rate d = 0.10. Ethical approval for the study was received from the Koç University Clinical Research Ethics Committee (2020.326.IRB1.116). Detailed information about the study was given in the clinic to children and their families who met the study criteria and agreed to participate in the study. Written informed consent from the caregivers and verbal consent from the children were obtained via email. All methods were performed in accordance with the relevant guidelines and regulations.

Fig. 1.

An overview of the flow of participants through the stages.

Study design

The study was conducted at home settings by the caregivers under the supervision of the researchers via video-phone calls and messages throughout the trial.

Age, diabetes duration, body weight, height, body mass index (BMI), amount of carbohydrate taken for each meal, carbohydrate-insulin ratio, insulin correction factor, daily basal and bolus insulin doses, and basal-bolus insulin ratios were collected from the caregivers via a questionnaire. The responses were verified from electronic medical records.

One week before the first trial day, basal and bolus insulin doses were adjusted by the researchers to achieve dinner fasting blood glucose values between 70–145 mg/dL (3.9–8.1 mmol/L) and 2-hour postprandial glucose values between 90-180 mg/dL (5-10 mmol/L) in accordance with ISPAD glycemic targets [8, 9]. During the study days, children had 4 meals (breakfast, lunch, afternoon snack, dinner) with calculated rapid-acting bolus insulin administered before meals.

Given that milk is a frequently consumed snack among children with T1D before bedtime, milk and its products were preferred as the test snacks in this study. Four separate trial days were planned, with at least one day off between each trial day. Children with T1D were given three test snacks measured by precise weighing (200 mL full-fat organic milk, 140 g full-fat yoghurt, and 200 mL full-fat plain kefir) containing the same amount of carbohydrates (10 grams) on different days in randomized order. No test meal was given on one night as a control day. The snacks were provided to the families by the researchers, and on each occasion, prior to administration, were confirmed verbally and visually by the researchers before being given to the child. The order in which the test snacks were given was determined by computerized random selection.

To ensure that evening meal fasting glucose levels were within the target range (70–145 mg/dL (3.9–8.1 mmol/L)), the children’s glucose levels were measured at 4.00 p.m. and corrected before dinner if needed. The dinner menu was planned by the researchers. The dinner portions were adjusted by a researcher, and macronutrients were standardized for each child on trial days. The macronutrient distribution of dinner was planned as 40-50% carbohydrates, 15–25% protein, and <35% fat, in line with the nutritional recommendations for children and adolescents [10]. The dinner was given to the children by the caregivers and was consumed within 20–30 minutes. The bolus dose was calculated by the researcher, based on the children’s insulin sensitivity factor, insulin carbohydrate ratio, and the amount of carbohydrates and other nutrients for dinner. The bolus doses were administered 10–15 minutes before the start of dinner by the caregivers. Test snacks were consumed at home under the supervision of the caregivers within 150-180 minutes after dinner. No insulin administered with snacks.

The trial day was postponed to next day if severe hypoglycemia (glucose value < 54 mg/dL (3 mmol/L)) or ketoacidosis occurred in the last 24 hours, caffeine (tea, coffee, cola) was taken in the last 24 hours, there was administration of correction dose before and/or after dinner, a post-dinner hypoglycemia (glucose value < 70 mg/dL (3.9 mmol/L)) that required taking carbohydrates to treat hypoglycemia or children had unusual more intense activity 24 hours before or during the trial day. Any snacks other than the test meal were not consumed on trial days.

During the trial, capillary blood glucose measurements were performed by the caregivers if CGM glucose values fell below 70 mg/dL (3.9 mmol/L) or reached 300 mg/dL (16.7 mmol/L), these values were recorded simultaneously by a researcher. The trial day was terminated 6 hours after the administration of the test snack or if hypoglycemia developed after administration of the test snack (capillary blood glucose <70 mg/dL (3.9 mmol/L)) or sensor glucose and capillary blood glucose reached 300 mg/dL (16.7 mmol/L). Confirmed hypoglycemia and hyperglycemia were treated with 0.3 gram/kg of carbs, and administering the calculated correction dose of rapid-acting insulin, respectively.

Measurements

No trial was planned on the first sensor day due to the risk of inaccurate measurements in the first 24 hours of the CGM use [11]. CGM glucose values starting from the test snack consumption until the termination of the trial day (up to 6 hours after the snack consumption or hypoglycemia/hyperglycemia termination events) were collected from raw data via LibreView. To evaluate the morning glucose levels, we also collected 3 hours of CGM data after trial termination as an extended period (6–9 h). Percentages of time spent in the 70–180 mg/dL (3.9–10 mmol/L) range (TIR), time in 54–69 mg/dL (3–3.8 mmol/L) (TBR1), time below <54 mg/dL (3 mmol/L) (TBR2), time in 181–250 mg/dL (10.1–13.9 mmol/L) (TAR1), time above >250 mg/dL (13.9 mmol/L) (TAR2), mean glucose, and coefficient of variation (CV) were calculated from raw data according to the international consensus [12].

The primary endpoint of the study was the difference in mean glucose between milk, yoghurt, kefir, and control days. Secondary endpoints were differences in other CGM metrics (TAR1, TAR2, TBR1, TBR2, TIR, and CV) between the groups. Termination due to severe hyperglycemia and hypoglycemia was also noted for the groups.

Statistical analysis

Categorical parameters are presented as a number (n) and a percentage (%), and numerical data are presented as a mean and standard deviation (SD). CGM data of 28 children who completed all four trial days (112 test days in total) were included in the analysis. The Shapiro-Wilks test was used to test the normality of the distribution of data. Repeated measures ANOVA test was used to compare the variables of the 4 trial days. To determine which snack caused the difference between the four separate test meal conditions, the pairwise comparisons test was applied as a post hoc analysis and evaluated by applying the Bonferroni correction. P-values < 0.05 were considered statistically significant. IBM’s SPSS (version 23.0) and Microsoft Excel software were used for the analysis.

Results

Of 28 children, characteristics, insulin intake, and meal management at the baseline are shown in Table 1. In brief, the mean age of 28 children (13 female (46%)) was 6.6 ± 0.8 years with diabetes duration of 2.7 ± 1.3 years, and HbA1c of 7.0 ± 0.5% (53 mmol/mol). While all children were using insulin glargine (100 IU/mL) and insulin lispro (100 IU/mL), the mean total daily dose/kg/day was 0.8 ± 0.2 IU, basal/total daily insulin ratio was 42.8 ± 8.9%. According to WHO references [13], there were no children with a BMI Z score below -1SD, 60.7% of children were classified as normal (≥-1SD -<1 SD), 32.1% as overweight (≥1 SD–<2 SD) and 7% as obese (≥2 SD). There was no significant difference by sex for all variables in Table 1 (p = ns for all).

Table 1.

Baseline characteristics of 28 children.

	All children (n = 28) Mean ± SD
Age, years	6.6 ± 0.8
Sex, female, n (%)	13 (46)
Age at diabetes diagnosis, years	3.9 ± 1.3
Diabetes duration, years	2.7 ± 1.3
HbA1c, % (mmol/mol)	7.0 ± 0.5 (53)
Body weight, kg	24.6 ± 3.7
Body height, cm	120.9 ± 6.7
BMI, kg/m²	16.8 ± 1.6
BMI z-score, SDS	0.8 ± 0.9
Insulin treatment
Total daily insulin, IU/kg/day	0.8 ± 0.2
Basal/total insulin ratio, %	42.8 ± 8.9
Basal insulin, IU/day	8.8 ± 3.0
Total daily bolus insulin, IU/day	11.3 ± 3.7
Insulin correction factor, mg/dL (mmol/L)/IU	102 ± 30 (5.7 ± 1.7)

BMI body mass index, Carb carbohydrates.

Data were collected from parents via a questionnaire and verified from electronic medical records.

The dinner meals were similar on all trial days regarding total energy, carbohydrate, protein, fat, and fiber (p = ns for all). The macronutrients and energy values of the dinner on all trial days are shown in Supplementary Table 1S.

On 112 trial days, 62% had the test snack or control between 09 and 10.00 p.m., 24% had between 10 and 11.00 p.m., 12% had between 08-09.00 p.m., and 2% had between 11.00 p.m. and 12.00 a.m. Mean glucose values just before the test snack consumptions were 137.8 ± 14.5 mg/dL (7.7 ± 0.8 mmol/L) for milk, 141.9 ± 16.9 mg/dL (7.9 ± 0.9 mmol/L) for yoghurt, 136 ± 19.1 mg/dL (7.6–1.1 mmol/L) for kefir, and 140.8 ± 17.0 mg/dL (7.8 ± 0.9 mmol/L) for control without a significant difference (p = 0.548). The 6-hour test period was divided into 3 2-hour periods 0–2 h, 2-4 h, and 4–6 h, additionally, 0–2 h was considered early and 2–6 h was the late period. As shown in Table 2, glucose values on control days were significantly lower than on milk, yoghurt, and kefir days in all time periods (p < 0.001). Glucose values were similar in all periods between milk, yoghurt, and kefir days. The fastest increase was within the first hour and the peak was reached in the second hour after test snacks (Fig. 2).

Table 2.

Mean glucose values before and after control and test snacks (n = 28).

CGM glucose values (mg/dL) (mmol/L)
Time	Milk Χ±SD	Yoghurt Χ±SD	Kefir Χ±SD	Control Χ±SD	P-value*
Before the test snack	137.8 ± 14.5 (7.7 ± 0.8)	141.9 ± 16.9 (7.9 ± 0.9)	136.0 ± 19.1 (7.6 ± 1.1)	140.8 ± 17.0 (7.8 ± 0.9)	0.548
0-2 hours after	186.5 ± 31.5a (10.4 ± 1.8)	182.3 ± 28.2a (10.1 ± 1.6)	176.7 ± 30.6a (9.8 ± 1.7)	151.6 ± 18.0b (8.4 ± 1)	<0.001
2-4 hours after	239.3 ± 12.6a (13.3 ± 0.7)	212.5 ± 10.7a (11.8 ± 0.6)	200.2 ± 12.1a (11.1 ± 0.7)	136.4 ± 9.8b (7.6 ± 0.5)	<0.001
4-6 hours after	205.6 ± 14.0a (11.4 ± 0.8)	192.2 ± 13.7a (10.7 ± 0.8)	174.5 ± 13.7a (9.7 ± 0.8)	122.8 ± 11.2b (6.8 ± 0.6)	<0.001
All period (0-6 h)	203.1 ± 35.6a (11.3 ± 2.0)	193.9 ± 32.6a (10.8 ± 1.8)	188.9 ± 36.8a (10.5 ± 2.0)	141.3 ± 26.6b (7.9 ± 1.5)	<0.001

^*Repeated measures Anova (Pairwise comparisons). Bonferroni adjustment applied.

Different superscript letters in the same row indicate that the difference between the values is significant.

Fig. 2.

Continuous glucose monitoring glucose values by hours following a bedtime snack (n = 28).

CGM data in test periods were analyzed according to International consensus CGM metrics (Table 3). During all trial period (0–6 h), TIR was 34.7% for milk, 38.7%. for yoghurt, 45.9% for kefir, and 75.5% for control. TIR on the control day was significantly higher than the milk (p < 0.001), yoghurt (p < 0.001), and kefir days (p = 0.004). TBR1 was not observed on the milk days. TBR1 differences between the control, yoghurt, and kefir days were similar (p = 0.054). TAR1 was significantly lower on control days than on milk (p = 0.002) and yoghurt days (p < 0.001). Similarly, TAR2 on the control day was significantly lower than milk (p < 0.001), yoghurt (p = 0.007), and kefir (p = 0.012) days. CV was <36% in all days and similar in pairwise comparison (p = 0.565).

Table 3.

Continuous Glucose Monitoring metrics in milk, yoghurt, kefir and control days.

	Early period (0-2 h)					Late Period (2-6 h)					All period (0-6 h)
	Milk	Yoghurt	Kefir	Control	P-value*	Milk	Yoghurt	Kefir	Control	P-value*	Milk	Yoghurt	Kefir	Control	P-value*
TIR (70–180 mg/dL) (3.9–10 mmol/L), %	47.28a	49.11a	58.75ab	74.78b	0.002	22.77a	30.51a	41.76a	75.28b	<0.001	34.66a	38.70a	45.91a	75.47b	<0.001
TBR1 (54–69 mg/dL) (3–3.8 mmol/L), %	-	-	0.45	2.30	0.159	-	0.78a	1.04a	7.34a	0.104	-	0.45a	0.77a	5.38a	0.054
TBR2 (<54 mg/dL) (3 mmol/L), %	-	-	-	-		-	-	-	0.17		-	-	-	0.10
TAR1 (181–250 mg/dL) (10.1–13.9 mmol/L), %	39.87ab	43.90a	32.74ab	22.92b	0.028	34.03ab	47.92a	35.51ab	14.26b	0.004	39.57a	47.03a	36.28ab	17.36b	<0.001
TAR2 (>250 mg/dL) (13.9 mmol/L), %	12.85a	7.21ab	8.99a	0.00b	<0.001	43.58a	21.22ab	21.88ab	3.44b	<0.001	25.93a	14.17a	17.53a	1.98b	<0.001
Mean glucose, mg/dL (mmol/L)	186.54a (10.4)	182.28a (10.1)	176.72ab (9.8)	151.61b (8.4)	<0.001	224.50a (12.5)	202.95a (11.3)	193.18a (10.7)	132.84b (7.4)	<0.001	203.07a (11.3)	193.93a (10.8)	188.88a (10.5)	141.29b (7.9)	<0.001
SD	31.46a	28.17a	30.57a	18.00b	0.007	17.61a	18.88a	25.40a	22.28a	0.210	35.64a	32.55a	36.79a	26.61a	0.075
CV, %	16a	15a	17a	12a	0.121	9a	10a	15ab	18b	<0.001	17a	17a	20a	20a	0.565

^*Repeated measures Anova (Pairwise comparisons). Bonferroni adjustment applied.

Different superscript letters indicate statistically significant difference within the column of the each time period.

Except one instance in a control day, TBR2 was not observed in trial days since the hypoglycemia was treated if blood glucose levels were <70 mg/dL (3.9 mmol/L). In kefir days, CGM glucose values < 70 mg/dL (3.9 mmol/L) could not confirmed as hypoglycemia since capillary blood glucose values were >70 mg/dL (3.9 mmol/L).

In the early period (0-2 hours), TIR was 47.3% for the milk, 49.1% for the yoghurt, and 58.8% for the kefir, 74.8% for the control day. TIR on the control day was significantly higher than the milk (p = 0.002) and yoghurt (p = 0.014). TAR1 was lower on the control day than the yoghurt meal (p = 0.031). TAR2 on the control day was significantly lower than milk (p = 0.009) and kefir days (p = 0.012). TBR2 was not observed, TBR1 and CV were similar between groups in the early period.

In the late period (2–6 hours), TIR was 22.8% for the milk, 30.5% for the yoghurt, 41.8% for the kefir, and 75.3% for the control day. The TIR on control days was significantly higher than the milk (p < 0.001), yoghurt (p = 0.001), and kefir days (p = 0.012). TAR1 in the late period was significantly higher in yoghurt than the control (p = 0.004). Milk day was found to be higher than the control for TAR2 (p < 0.001). CV was less on the milk and yoghurt days compared to the control day (p = 0.001 and p < 0.001, respectively).

When the early (0–2 h) and late (2–6 h) period’s TIR were examined, the TIR value decreased significantly on the trial days of the milk (47.3% to 22.8%, p < 0.001), yoghurt (49.1% to 30.5%, p = 0.006), and kefir (58.8% to 41.8%, p = 0.01). Early and late TIR values remained similar on the control day (p = 0.839) (Table 3).

Of 112 trial days, 13 trial days were terminated early due to hyperglycemia (>300 mg/dL (16.7 mmol/L)) and 3 trial days due to hypoglycemia (<70 mg/dL (3.9 mmol/L)) that required treatment. The timing of hyperglycemia and hypoglycemia events by test groups is shown in Table 4.

Table 4.

Termination events by test snacks and control (n = 112).

Termination Events, n	Milk	Yoghurt	Kefir	Control	Total
Total events in all period (0–6 h)	8	5	1	2	16
Hyperglycemia (≥300 mg/dL) (16.7 mmol/L) in all period (0–6 h)	8	4	1	-	13
Early period (0–2 h)	1	1	-	-	2
Late period (2–6 h)	7	3	1	-	11
Hypoglycemia (<70 mg/dL) (3.9 mmol/L) in all period (0–6 h)	-	1	-	2	3
Early period (0–2 h)	-	-	-	-	-
Late period (2–6 h)	-	1	-	2	3

The mean glucose values at the end of 6 hours for milk, yoghurt, and kefir meals were similar (milk-yoghurt p = 0.342, milk-kefir p = 0.187, yoghurt-kefir p = 0.542), and the mean glucose values on the control day were lower than the milk day (p < 0.001), yoghurt day (p = 0.002) and kefir day (p < 0.001).

After trial termination at 6 h, CGM data were collected during extension phase (6-9 h). Mean glucose values at 6^th hour, at 9th hour, and between 6-9 hours after the test meals are shown in Table 5. The mean glucose values at the end of 9 hours (at 9th h) were similar for milk, yoghurt and kefir meals (milk-yoghurt p = 0.193, milk-kefir p = 0.574, yoghurt-kefir p = 0.681) and the mean glucose values on the control day were lower than the milk day (p = 0.014) and kefir day (p = 0.026).

Table 5.

Mean glucose values at 6th hour, at 9th hour and between 6-9 hours after the test meals (mg/dL) (mmol/L).

	Milk Mean ± SD	Yoghurt Mean ± SD	Kefir Mean ± SD	Control Mean ± SD	p-values*
At 6th h	198.0 ± 13.4a (11.0 ± 0.7)	183.5 ± 14.2a (10.2 ± 0.9)	173.2 ± 14.9a (9.6 ± 0.8)	122.4 ± 11.8b (6.8 ± 0.7)	<0.001
At 9th h	168.7 ± 15.4a (9.4 ± 0.9)	146.4 ± 13.6ab (8.1 ± 0.8)	154.8 ± 19.5a (8.6 ± 1.1)	109.9 ± 13.9b (6.1 ± 0.8)	0.059
6-9 h	180.5 ± 14.1a (10.0 ± 0.8)	162.3 ± 18.7a (9.0 ± 1.0)	151.9 ± 12.7a (8.4 ± 0.7)	115.4 ± 12.6b (6.4 ± 0.7)	0.017

^*Repeated measures Anova (Pairwise comparisons).

Different superscript letters indicate statistically significant differences within the column of each time period.

The mean glucose values during extension period (6–9 h) were also similar for milk, yoghurt and kefir meals (milk-yoghurt p = 0.429, milk-kefir p = 0.090, yoghurt-kefir p = 0.546) and the mean glucose values on the control day were lower than the milk day (p = 0.004), yoghurt day (p = 0.012) and kefir day (p = 0.042).

Discussion

This study was conducted to examine the effects of different snack options taken before sleep on nocturnal glycemia in 5–8 years old children with T1D. While control days resulted in higher TIR and more optimal mean glucose values with no increase in TBR (<54 mg/dL (3 mmol/L)) during the trial period, all snacks resulted in lower TIR and higher mean glucose values starting from the first hour of the snack administration and lasting until the trial termination at the 6th hour and sustained in the extension period (6–9 h).

Before giving the snacks, glycemia was adjusted for the trial. As high fat and protein in meals result in sustained hyperglycemia, children consumed similar dinner meals with similar amounts and distribution of macronutrients according to ISPAD nutritional management guidelines [10] to achieve optimal trial conditions. All children consumed the dinner within 20–30 minutes, and bolus timing was 10–15 minutes before the dinner to avoid postprandial hyperglycemia [3] and sustained effects of meals and insulin. Thus, the post-prandial 90–180 mg/dL (5–10 mmol/L) target was achieved at 2 hours after dinner without any bolus insulin on board, and children started the trial with similar glucose values in all trial days.

The study terminated three times due to nocturnal hypoglycemia, and two of them were on control days. Thirteen terminations were due to hyperglycemia, which was more frequent than hypoglycemia and solely seen on snack days, especially with milk. Moreover, children spent most of their time in above target range on snack days, while children met the goal of TIR > 70% on days with no snacks. When comparing the risk of hyperglycemia and the prevention of hypoglycemia after bedtime snacking, snacks in our study did not provide a good benefit. Since we could not find a significant difference in time below range between snack days and control, bedtime snacking is compromising glycemic control while not providing significant prevention from nocturnal hypoglycemia. Barton et al. investigated nocturnal hypoglycemia events with and without 15-g carbohydrate bedtime snacking in children and showed no difference in hypoglycemia events [3]. Other previous studies had contradictory results regarding protection from nocturnal hypoglycemia with bedtime snacking, with no studies providing strong evidence for children [4].

Our results have an important clinical implication, especially in cases of fear of hypoglycemia. Fear of hypoglycemia in parents frequently leads to bedtime snacking to avoid nocturnal hypoglycemia [14–16]. In this regard, encouragement of parents to avoid bedtime snacking may lower nocturnal hyperglycemia and help to achieve glycemic targets.

Previous clinical studies investigated the changing insulin needs of young children at different times of the day. Insulin requirement decreases in the morning (between 03.00–06.00 a.m.) and increases in the first half of the night (between 09.00 p.m.–12.00 a.m.), especially in our study age group. MDI therapy does not allow flexibility for basal insulin and leads parents to give bedtime snacking to children to avoid nocturnal hypoglycemia. However, our results showed that 10 grams of carbohydrates before bedtime caused nocturnal hyperglycemia even if children started their night in the target range. The timing of night snacking in our study is concurrent with the reverse dawn phenomenon timing, which may have exaggerated the hyperglycemia. Conversely, the peak of intermediate-acting insulin was concurrent with the timing of reduced insulin need between 3.00–6.00 a.m., which made people more susceptible to nocturnal hypoglycemia. The transition from intermediate-acting insulin to long-acting insulin decreased this susceptibility. Ultra-long-acting insulins have less peak effect, which may reduce the risk even further compared to insulin glargine and detemir [17]. Thus, decreased susceptibility for hypoglycemia with newer insulins also reduced the need for prophylactic bedtime snacking. Two hypoglycemia events in the late period of control days may be due to this susceptibility; however, hypoglycemia events were infrequent, and snacking resulted in worse glycemic outcomes. Future studies should investigate the effects of bedtime snacks in children using second-generation basal insulin to clarify this issue.

The strengths of the study are precise adjustment of dinner macronutrients and amounts, glycemic stability before the test snacks, a strict trial plan excluding any glucose-altering event, incorporation of CGM data for nocturnal glycemic evaluation, and use of the same CGM model to avoid measurement bias. The crossover trial design allowed for the comparison of different interventions on the same patient, reducing variability and improving the reliability of the results. This study had some limitations, such as the use of an old-generation CGM, which has a high variability, especially in low glucose values [18]. This increased variability usually overestimates hypoglycemia and time below 70 mg/dL (3.9 mmol/L) [19]. We tried to overcome this limitation with capillary blood glucose measurements to confirm hypoglycemia events. However, this study did not evaluate the option of giving a bedtime snack with insulin. Instead of routinely recommending a bedtime snack to all children with diabetes, it may be more appropriate to individualize this recommendation by taking into account personal differences, such as the type of basal insulin used. Another limitation is the use of dairy products as snacks, so our results cannot be extended to non-dairy snacks. Furthermore, future studies with larger cohorts are warranted to validate these findings and to explore potential subgroup effects.

Automated insulin delivery systems (AID) can adjust basal insulin doses according to CGM input and can overcome the abovementioned problems of MDI therapy. Since fewer factors can affect glucose at nighttime, AID achieves higher TIR during nighttime [20]. Moreover, AID can handle unannounced meals up to 20 grams of carbohydrate in older children, although results in higher glucose spikes than announced meals up to 20-g carbs [21]. Thus, AID may help alleviate nocturnal hyperglycemia after bedtime snacking; however, higher mean glucose and disruption in glycemia would be expected.

Conclusion

To our knowledge, this was the first study that investigated the change in CGM glycemic metrics after bedtime snacking in young children. Bedtime snacking in young children decreased TIR, increased mean glucose, and caused sustained hyperglycemia without a significant difference in TBR. Routine dairy product consumption at bedtime may preclude achieving glycemic targets; thus, going to bed within the target glucose range without bedtime snacking can achieve better overnight glycemic management. As an alternative conclusion, further studies could be conducted to show whether insulin is needed with a bedtime snack to prevent hyperglycemia when the child requires food for nutritional purposes.

Author contributions

Ş.H., C.S., H.G.Ö. designed the study. TG recruited study participants, was the principal investigator responsible for ensuring that the entire study process was conducted in accordance with the study protocol. E.C. collected C.G.M. raw data. TG performed all analyses, performed statistical analysis, K.E.K. wrote the manuscript. Ş.H., C.S., G.Y.M., E.E., H.G.Ö., S.M. and T.G. contributed to editing the manuscript. All co-authors read and approved the final manuscript.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests

TG, KEK, GYM, SM, EE, EC, CS, ŞH and HGÖ declare that there is no duality of interest associated with this manuscript.

Supplementary information

The online version contains supplementary material available at 10.1038/s41387-025-00392-9.