Dietary glycemic index is associated with overweight and obesity in preschool children: a national cross-sectional study in Lebanon

Abstract

Background

Although dietary glycemic index (GI) and glycemic load (GL) have been proposed to modulate overweight risk, evidence in preschoolers remains scarce and contentious, and lacking in the Eastern Mediterranean Region (EMR). This cross-sectional national survey investigated the association of dietary GI and GL with overweight/obesity in Lebanese preschoolers (n = 518; aged 2–5 years).

Methods

The study is based on a national cross-sectional survey conducted in 2012/2013 “Early Life Nutrition and Health, ELNAHL". Sociodemographic data were obtained using a multi-component questionnaire, and anthropometric characteristics were measured. The 24 h recall approach was used to collect dietary data. The calculations of daily dietary GI and GL were based on available carbohydrates (and repeated using total carbohydrates). Three logistic regression models were performed to investigate the association of dietary GI and GL with overweight/obesity, per unit and per 10-unit increases.

Results

In this study sample, 9.5% of the preschool children were found to be overweight/obese. Average dietary GI was determined at 56.1 ± 0.3 and dietary GL at 99.2 ± 1.8, when using available carbohydrates. The full regression model (model 3) which was adjusted for demographic/socioeconomic variables and energy and macronutrients’ intakes, showed a significant association between dietary GI and overweight/obesity, per unit and per 10 units increases (odds ratio (OR): 1.10; 95% confidence interval (CI): 1.09–1.11 and OR: 2.67; 95% CI: 2.45–2.92, respectively), while only a slight association was observed for GL (OR: 1.01, 95% CI: 1.00, 1.01 and OR: 1.05, 95% CI: 1.04, 1.07 for 1 unit and 10 units increases of GL, respectively). Similar results were obtained when using total carbohydrates for GI and GL calculations.

Conclusions

This study shows that each 10-unit increase in dietary GI was linked with approximately three-fold higher odds of overweight/obesity amongst Lebanese preschoolers, suggesting that dietary GI may be of public health significance in the epidemic of childhood overweight.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12887-025-05860-w.

Introduction

Pediatric overweight and obesity constitute a significant public health concern, with available evidence highlighting secular increases in their prevalence in both developed and developing nations [1–3]. These increases have been observed as early as the preschool years [4, 5]. Over the last two decades, the proportions of overweight among children aged below 5 years increased from 4.8% to 6.1% [6, 7], and according to the World Health Organization (WHO), 39 million children aged under five were classified as overweight or obese in 2020 [8]. If these trajectories continue, 70 million children aged under five are predicted to be overweight or obese by 2025 [2]. Of more concern is the fact that the prevalence of pediatric obesity is still rising in low- and middle-income nations, despite some evidence suggesting a plateauing in high-income countries [9, 10]. The Middle East region harbors a high burden of preschool overweight and obesity, with available estimates showing that this prevalence has increased from 5.7% in 1990 to 9% in 2018 in Western Asia (which includes the Middle-East) [4].

Abnormal adiposity in children may have both short and long term impacts on their health, ranging from physical comorbidities, such as musculoskeletal problems, asthma, obstructive sleep apnea, and psychological sequela [11, 12] to higher risks of chronic diseases, on the long term, such as cardiometabolic complications, some types of cancer, and premature death [13, 14]. In addition children who are overweight or obese by the time they are five years old are four to five times more prone to be overweight/obese in adolescence [15] and to remain obese in adulthood [16].

Recent research indicates that in children with obesity, a large proportion of excess weight is acquired prior to the age of five years [17], highlighting this lifecycle phase as a critical period for obesity prevention [18]. It is a phase that directly precedes the increase in body mass index (BMI), referred to as the adiposity rebound [19], and it is also a period when dietary behaviors and lifestyle habits are changing and developing [20]. High dietary energy consumption, regular intake of fatty/sweet foods and sweetened beverages, coupled with sedentarity have been identified as significant contributors to overweight and obesity in this age group [21, 22]. Lately, carbohydrate quality and quantity as evaluated by the Glycemic Index (GI) and Glycemic Load (GL) have received increasing attention as to their role in the etiology of obesity [23, 24]. The GI and GL concepts were introduced to classify carbohydrates according to their impact on blood glucose levels, since not all carbohydrate-rich foods increase blood glucose levels equally [25]. The consumption of high GI/GL foods was proposed to raise obesity risk by several mechanisms including the reduction of satiety and the increase in carbohydrate oxidation at the detriment of fat oxidation [26, 27]. Limited research has explored the association between dietary GI, GL and overweight/obesity in preschool children yielding conflicting results [28–34]. Furthermore, research on the link between GI, GL and preschool overweight is completely lacking in the Middle East, despite the high prevalence of pediatric overweight and obesity in this region. In this context, this study was conducted with the aim of evaluating the association between dietary GI, GL and overweight/obesity among preschool children in Lebanon. More specifically, the study’s objectives are to: 1) establish the GI and GL values of food items consumed by a national sample of Lebanese preschoolers (2–5 years) based on the recent International Tables of Glycemic Index and Glycemic Load, 2) determine the overall dietary GI and GL, and 3) investigate their association with overweight/obesity.

Methods

Study population

This investigation is based on the “Early Life Nutrition and Health, ELNAHL” (NUT.LN.13) national cross-sectional survey that was carried out among children aged under five and their mothers, in the years 2012–2013. The survey was developed for this study (supplementary file). As described elsewhere [35, 36], the ELNAHL national survey adopted a stratified cluster sampling technique. The six Lebanese governorates were the strata, while the clusters were chosen at the level of the district. The primary sampling units in the various districts were the housing units. Based on an estimated prevalence of 13% of overweight and obesity among children under five [2], a sample size of 1,030 children under five was required to determine the prevalence of overweight and obesity, with a 2% error and a 95% confidence interval.

The recruitment of children in the national survey was based on the following inclusion criteria: lack of inborn errors of metabolism, chronic illness, or physical abnormalities that could influence feeding habits or body composition; gestational age at birth ≥ 37 weeks; and born to Lebanese mothers. For the present investigation, data pertinent to 2–5 year old children was considered (n = 518). Infants and younger children aged below 2 years of age were not considered part of this study given that their diet relies heavily on milk.

Ethics

This study was approved by the Institutional Review Board of the American University of Beirut. Parental written informed consent was also acquired from parents before enrolling the participants in the national survey.

Data collection

Data was collected using a multi-component questionnaire that included socio-demographic and economic characteristics, and feeding history (breastfed, yes/no) in addition to dietary assessment amongst under-five children (24-h recall), and anthropometric measurements of both mother and child.

Anthropometric measurements

Height measurements were collected using a portable stadiometer (Seca, model 213, Hamburg, Germany) to the nearest 0.5 cm with the child being bare footed. Body weight’s measurements were obtained using a standard clinical balance (Seca, model 770, Hamburg, Germany) to the nearest 0.1 kg with the child being bare footed and in light clothes. Accordingly, children were classified into wasted, normal, at risk of overweight, overweight, or with obesity based on the sex and age-specific BMI-for-age z-scores of the WHO [37]. The anthropometric characteristics of the mothers were also measured according to standard protocols, including body weight (kg) and height (m) [38], which were measured using a weighing scale (SECA model 770,) and a portable stadiometer (SECA model 213), respectively. Maternal BMI was calculated as weight (kg)/height (m²).

Dietary intake assessment

Dietary intake of the participating children was assessed using the 24h multiple pass approach, with the mother serving as proxy. During the recall, the interviewers adopted the five steps of the United States Department of Agriculture (USDA) multiple pass 24h recall method [39]. Mothers were asked to report every item that the child consumed in the 24 h prior to the interview and give descriptions of every dish and beverage. The names of the foods/beverages consumed, portion size, cooking techniques, and brand names (when applicable) were noted. Household measures and standard cups were used to help mothers in portion size estimation. Mothers were encouraged to consult with any other caregiver (helper or family member) regarding the child’s food consumption the previous day. The Nutritionist Pro software (version 4.2; Axxya System) was utilized to conduct analysis of the 24h dietary recall, using the USDA database. For composite Lebanese food items that are not featured in the afore-mentioned database, we have added recipes based on a popular local cookbook. The software permitted the determination of energy and macronutrients’ intakes for all participating children.

Calculation of GI & GL

The steps that were adopted for allocating GI and GL values to the various foods were based on the literature [40]. Accordingly, food items with lower than or equal to 5 g of carbohydrates/ 100 g were allocated a GI value of zero. Otherwise, the main data source to identify the GI values of the various foods was the fourth edition of the "International Tables of Glycemic Index and Glycemic Load Values: 2021″. Therefore if a particular food was included in these International Tables, the matching GI value was used [41]. If more than one possible GI value was featured in the database for that particular food, the average of the available GI values was allocated to the food item in question [41]. Foods that did not have a GI value in the International Tables were allocated a GI value based on the food item with the closest nutritional similarity (examples: yellow cake was allocated the GI value of a vanilla cake and éclair custard was allocated the GI value of a pastry puff) [40, 42]. However, for foods that did not have a value in the latest International Tables and did not have a close match in the database, GI values published elsewhere were adopted [43–46], when available. For traditional mixed meals, the GI was calculated as follows:

$${\text{GI}}_{\text{meal}}=\frac{{\sum }_n^{x=1}({\text{GI}}_x\times C_x)}{C_{\text{meal}}}$$

where GI_x represents the GI of the x^th food, C_x the quantity of available carbohydrate (CHO) in the x^th food and C_meal the quantity of available CHO in the meal [47, 48].

After allocating the GI values to each of the consumed food items, the calculation of daily dietary GI was performed as the “sum of the GI of all foods consumed per day, multiplied by the corresponding available CHO content per serving, divided by the available daily CHO consumed” [49, 50]. The calculation of daily dietary GL was performed as “the sum the GI of all foods consumed per day, multiplied by the corresponding CHO content per serving, divided by 100” [50, 51]. These calculations are shown below:

$$\text{Daily dietary GI}={\mathrm{\Sigma }}^{n_{i=1}}\text{GI}i\text{x CH}{\text{O}}_i/{\mathrm{\Sigma }}^{n_{i=1}}{\text{CHO}}_i$$

$$\text{Daily dietary GL}={\mathrm{\Sigma }}^{n_{i=1}}\text{GI}i\text{x CH}{\text{O}}_i/100$$

where GI_i represents the GI for food i, CHO_i represents the available CHO content in food i (grams per day), and n represents the number of foods consumed per day. Available CHO is defined as the carbohydrate that is “digested, absorbed and metabolized”, and was estimated as total carbohydrate in the food minus its fiber content [50, 52].

The calculations described above were repeated using total carbohydrates instead of available carbohydrates. Given that the literature lacks a clear consensus on which approach to employ in the estimation of GI and GL (available or total carbohydrates), it was deemed essential to use both approaches in our calculations [53–55].

Statistical analysis

Children were categorized based on their BMI-for-age z-scores into two categories: 1) children with normal weight/at risk of overweight and 2) children who have overweight/obesity. Sociodemographic and dietary intake characteristics of children were described as means (and standard errors) for continuous variables, and as frequencies (n) and percentages (%) for categorical ones, for the whole study sample as well as by BMI category. Independent student t-tests with equal variances were performed to compare continuous variables, while chi-square and Fisher exact tests were utilized for the comparison of categorical variables between the two BMI categories, as appropriate.

Logistic regression analyses were conducted to examine the associations of each of dietary GI and GL with overweight/obesity, whereby children’s BMI category was the dependent variable (normal/at risk of overweight versus overweight/obesity) and each of dietary GI and GL as independent variables. In addition, dietary GI and GL in the multivariable analyses were rescaled by dividing each by 10, so that results are also interpretable per ten-unit increase of each. Variables that showed statistically significant associations in the simple logistic models and/or were reported as significant confounders in the literature [34, 56–63] were included as independent variables in the multiple logistic regression models. Accordingly, associations were investigated based on the following three models:

Model 1: adjusted for age, gender, type of school attended, mother’s education, father’s education, presence of paid helper.

Model 2: Model 1 + adjustment for energy intake (EI).

Model 3 for GI: Model 2 + adjustment for fat intake (g), protein intake (g) and total dietary fiber intake (g).

Model 3 for GL: Model 1 + adjustment for percentage of energy from protein + percentage of energy from fat and total dietary fiber intake (g).

The logistic regression analysis’s results were shown as odds ratio (OR), and 95% confidence interval (CI). Using the variance inflation factor (VIF), multicollinearity between the independent variables was assessed, whereby a VIF < 10 indicated no collinearity between the variables. Results in this study were also weighted to account for the effect of the cluster sampling technique [64]. Statistical analysis was performed using Stata, version 18. The level of significance was set at p-value < 0.05.

Results

The characteristics of the study sample are presented by BMI category in Table 1. Around 10% of the children had either overweight or obesity (6.8% were overweight and 2.7% had obesity). Children’s mean age was 40 months, with 53.5% being boys and 46.5% girls. More than half of the participating mothers (61.4%) had obtained intermediate to secondary educational level with a significant difference in maternal education level across the BMI categories (p = 0.003). Likewise, 62.9% of the fathers had attained intermediate to secondary educational level and a significant difference was observed in paternal educational level across the two categories (p = 0.004). While almost all the fathers were employed (93.4%), most of the participating mothers were not working (84.9%). As for the household monthly income, 48.1% reported earning between 100,000–2000,000 L.L at the time the survey was conducted (1 USD = 1,500 Lebanese pounds when the survey was carried out). Of the surveyed households, only 16.1% had a paid helper, with a significantly higher proportion among children who had overweight or obesity (p = 0.004). Average maternal BMI was determined at 26.7 ± 0.2 kg/m², and there was no significant differences between the groups. Governorate, marital status, mother’s employment, father’s employment and household income, did not show any significant differences between the two categories.

Table 1.

Sociodemographic and other characteristics among Lebanese preschoolers according to BMI category (n = 518)

	Total sample (n = 518)		BMI categorya				p-value*
			Normal/At risk of overweightb (n = 469)		Overweight/ with obesity (n = 49)
	Mean	SE	Mean	SE	Mean	SE
Child's age in months	40.0	0.4	40.0	0.5	40.3	1.2	0.830
Mother's age in years	32.8	0.3	32.7	0.3	33.9	0.9	0.207
Mother’s BMI (Kg/m2)	26.7	0.2	26.5	0.2	27.9	0.8	0.075
	n	%	n	%	n	%
Sociodemographic characteristics
Child's gender							0.810
Female	241	46.5	219	46.7	22	44.9
Male	277	53.5	250	53.3	27	55.1
Governorate							0.133
Beirut	56	10.8	47	10.0	9	18.4
Mount Lebanon	163	31.5	144	30.7	19	38.8
North Lebanon	172	33.2	158	33.7	14	28.6
South and Nabatiyeh	87	16.8	81	17.3	6	12.2
Beqaa	40	7.7	39	8.3	1	2.0
Mother’s marital status							0.279
Married	507	97.9	460	98.1	47	95.9
Separated/divorced/widowed	11	2.1	9	1.9	2	4.1
Mother's educational level							0.030
Less than intermediatec	98	18.9	95	20.3	3	6.1
Intermediate-secondaryd	318	61.4	286	61.0	32	65.3
College	102	19.7	88	18.8	14	28.6
Father's educational level							0.004
Less than intermediatec	114	22.4	112	24.3	2	4.2
Intermediate-secondaryd	320	62.9	285	61.8	35	72.9
College	75	14.7	64	13.9	11	22.9
Mother's employment status							0.271
Working	78	15.1	68	14.5	10	20.4
Not working	440	84.9	401	85.5	39	79.6
Father's employment status							0.721
Working	484	95.1	439	95.2	45	93.8
Not working	25	4.9	22	4.8	3	6.3
Presence of paid helper in the household							0.004
Yes	83	16.1	68	14.5	15	30.6
No	434	83.9	400	85.5	34	69.4
Household monthly income (Lebanese pounds)e							0.453
< 1,000,000 (~ 667 USD)	167	32.3	155	33.1	12	24.5
1,000,000–2,000,000 (~ 667– 1333 USD)	173	33.4	155	33.1	18	36.7
> 2,000–0000 (~ 1333 USD)	93	18	81	17.3	12	24.5
Do not know/refused to answer	85	16.4	78	16.6	7	14.3
Other characteristics
Ever breastfed
Yes	468	90.3	423	90.2	45	91.8	1.000
No	50	9.7	46	9.8	4	8.2

Abbreviations: BMI Body mass index, SE Standard error

^aAnthropometric measurements of children were categorized based on WHO classification [37]

^bThis category includes one wasted child

^cLess than elementary includes being illiterate, not attending school, being able to read and write only, or attending primary school

^dIntermediate to secondary includes attending intermediate school, high school, or technical diploma

^e1 USD = 1,500 Lebanese pounds at the time of the survey

^*p-values were obtained by conducting independent t-test with equal variances for continuous variables and chi-square test or Fischer exact test for categorical variables, as applicable. P-values in bold font represent significance at p < 0.05

A total of 543 different food items were identified based on the 24-h recalls. Of these, 276 food items were available in the fourth edition of the International Tables of GI values [41]. For 14 foods that did not have a GI value in the international tables, other sources were consulted to identify GI values [43]. A total of 117 food items had less than 5 g of carbohydrates/100 g and were therefore allocated a GI value of zero [44–46]. The GIs for 136 composite food items were calculated based on the respective GI values of the ingredients identified in the recipes (Alef Baa al Tabkh) [65].

Mean caloric and macronutrient intakes as well as the dietary GI and GL estimates (based on the two approaches) are shown in Table 2, by BMI category. Fat intake (expressed as percent of EI) was higher among children with overweight/obesity compared to children with normal BMI/at risk of overweight, while carbohydrate intake (%EI) was significantly lower. No significant differences were observed between the two categories in terms of protein intake. In addition, overweight/children with obesity had higher dietary GI estimates (using both approaches) compared to normal weight participants, whereas children with normal BMI had higher dietary GL estimates, but these differences did not reach statistical significance.

Table 2.

Mean caloric and macronutrients intakes and dietary GI and GL among Lebanese preschool children according to BMI category (n = 518)^a

	Total sample (n = 518)	BMI category
		Normal weight/At risk of overweight (n = 469)	Overweight /with obesity (n = 49)	p-value*
	Mean ± SE
Energy (kcal/day)	1533.2 ± 23.7	1534.1 ± 24.9	1524.5 ± 78.9	0.906
Macronutrients intake
Protein (g/day)	51.6 ± 1.0	51.5 ± 1.1	52.8 ± 3.3	0.717
Protein (% of EI)	13.7 ± 0.2	13.7 ± 0.2	13.9 ± 0.5	0.723
Fat (g/day)	66.8 ± 1.3	66.4 ± 1.4	70.2 ± 4.3	0.398
Fat (% of EI)	38.6 ± 0.3	38.3 ± 0.4	40.9 ± 1.1	0.031
Carbohydrates (g/day)	186.7 ± 3.1	187.9 ± 3.3	174.6 ± 9.8	0.215
Carbohydrates (% of EI)	49.1 ± 0.4	49.4 ± 0.4	46.3 ± 1.3	0.027
Total sugar (g/day)	76.2 ± 1.7	76.9 ± 1.8	69.0 ± 4.6	0.165
Total sugar (% of EI)	20.5 ± 0.4	20.6 ± 0.4	18.9 ± 1.0	0.160
Dietary fiber (g/day)	11.6 ± 0.3	11.7 ± 0.3	10.4 ± 0.8	0.211
Dietary GI and GL
GI-Available CHOb	56.1 ± 0.3	55.9 ± 0.3	57.3 ± 1.1	0.169
GI—Total CHOc	55.7 ± 0.3	55.5 ± 0.3	56.9 ± 1.1	0.197
GL Available CHOb	99.2 ± 1.8	99.6 ± 1.9	96.0 ± 6.2	0.570
GL Total CHOc	104.9 ± 1.9	105.2 ± 2.0	101.1 ± 6.5	0.528

Abbreviations: BMI Body mass index, CHO Carbohydrates, EI energy intake, GI Glycemic index, GL Glycemic load, SE Standard error

^aData summarized in this table as mean ± SE

^bValues based on the use of available carbohydrates to calculate daily GI

^cValues based on the use of total carbohydrates to calculate daily GL

*p-values were obtained by conducting independent t-test with equal variances. P-values in bold font represent significance at p < 0.05

The results of the multivariable logistic regression analyses for the associations of dietary GI and GL with overweight/obesity are shown in Table 3. Based on model 1, which was essentially adjusted for demographic and socioeconomic variables, GI was significantly associated with higher odds of overweight/obesity using both approaches (available and total carbohydrates), per single unit increase as well as per 10-units increases. There was no significant association with GL based on model 1.

Table 3.

Associations of dietary GI and GL with overweight/obesity based on multivariable logistic regression analyses^a

	Adjusted OR (95% CI)
	Model 1b	Model 2c	Model 3d
Glycemic Index, calculated using available CHO
Per 1 unit	1.09 (1.09, 1.11)	1.09 (1.09, 1.11)	1.10 (1.09, 1.11)
Per 10 unitse	2.55 (2.39, 2.73)	2.55 (2.39, 2.73)	2.67 (2.45, 2.92)
Glycemic Index, calculated using total CHO
Per 1 unit	1.09 (1.08, 1.10)	1.09 (1.08, 1.10)	1.10 (1.09, 1.11)
Per 10 unitse	2.46 (2.31, 2.61)	2.46 (2.30, 2.62)	2.65 (2.42, 2.89)
Glycemic Load, calculated using available CHO
Per 1 unit	0.99 (0.99, 1.00)	1.00 (1.00, 1.00)	1.01 (1.00, 1.01)
Per 10 unitse	0.99 (0.98, 1.01)	1.02 (1.01, 1.04)	1.05 (1.04, 1.07)
Glycemic Load, calculated using total CHO
Per 1 unit	0.99 (0.99, 1.00)	1.00 (0.99, 1.00)	1.01 (1.00, 1.01)
Per 10 unitse	0.99 (0.98, 1.01)	1.01 (0.99, 1.03)	1.05 (1.04, 1.07)

Abbreviations: CHO Carbohydrates, CI Confidence interval, GI Glycemic index, GL Glycemic load, OR Odds ratio

^aAnalyses presented are weighted to account for the effect of the cluster sampling

^bModel 1: adjusted for age, gender, type of school attended, mother’s education, father’s education, presence of paid helper

^cModel 2: Model 1 + adjustment for energy intake

^dModel 3 for GI: Model 2 + adjustment for fat intake (g), protein intake (g) and total dietary fiber intake (g). Model 3 for GL: Model 1 + adjustment for percentage of energy from protein + percentage of energy from fat and total dietary fiber intake (g)

^eThe continuous GI and GL variables were rescaled in units of 10

Results in bold format show significance at p < 0.05

The results were similar in model 2 (which was additionally adjusted for energy intake), but with an additional significant association being observed for GL (based on available carbohydrates) (OR: 1.02, 95% CI: 1.01, 1.04 per 10 units increases of GL). Model 3, which included additional adjustments for macronutrients, also yielded significant associations for GI. In specific, every 10 units increase in GI (either of the two approaches) was associated with approximately 2.7 times the odds of having overweight or obesity. There was also a slight 5% increase in the odds of being overweight or with obesity among preschoolers for a 10-unit increase in dietary GL (OR: 1.05, 95% CI: 1.04, 1.07).

Discussion

To the best of our knowledge, this study is the first in the Eastern Mediterranean Region (EMR) to assess dietary GI and GL in preschool-aged children, and to investigate their association with overweight/ obesity in this age group. Using two different calculation approaches, the study showed that average dietary GI ranged between 56.1 ± 6.7 and 55.7 ± 6.8, and dietary GL between 99.2 ± 41.6 and 104.9 ± 43.5 in Lebanese preschoolers. We found that a 10-unit increase in dietary GI was significantly associated with approximately three-fold higher odds of overweight/obesity in the study sample, while a slight association was noted between dietary GL and overweight/obesity. Studies examining these relationships are entirely lacking in the EMR, however previous studies conducted in different areas of the world produced contradictory findings [28–34].

In our study, and for the purpose of estimating the GI and GL values of the various foods, the International GI Tables were the main adopted reference [41]. The calculation of dietary GI was then performed by summing up the GI values of all foods consumed per day, multiplying them by the corresponding available (or total) carbohydrate content/serving, then dividing by the total daily available (or total) carbohydrates consumed. From a scientific stand-point, the use of available carbohydrates is more congruent with the definition of the GI [66]; however, and because the published literature on the topic includes studies that have used either available or total carbohydrates [30, 34, 59, 67, 68], we have opted to use two approaches for GI calculation (one based on available carbohydrates, and the second on total carbohydrates) for the purpose of meaningful comparison. Consequently, the calculated average dietary GI ranged between 56.1 ± 6.7 (based on available CHO) and 55.7 ± 6.8 (using total CHO), which is in line with estimates reported from Portugal 54.8 ± 3.45 [68] and Australia, 54.1 ± 1.6 [69]. In another investigation carried out in Australia [59], age-specific values for dietary GI were reported in children, with these estimates ranging between 53.3 ± 5.4 in 2 -3 year olds, and 55.5 ± 4.8 in 4–8 year olds.

In our study, the overall GL was determined as the product of the GI of the consumed foods and the corresponding available (or total) carbohydrate content/ serving. As such, mean dietary GL for preschool children in Lebanon was determined at 99.2 ± 41.6 using available carbohydrates and 104.9 ± 43.5 using total carbohydrates. These findings are in agreement with those reported by studies conducted in Portugal (103.25 ± 23.04 among 4 year old children) [68], Australia (96.1 ± 32.6 among 2- 3 year old children) [59] and Canada (110.1 ± 30.8 amongst 8–10 year old children) [67]. Our estimates are however higher than those determined by Buken et al., 2008 amongst 2-year-old children from Germany (62.8 ± 14.6), while being lower than those reported by some other studies. For instance, in Italy, Barba et al. [30] divided GL into quartiles which ranged from Q1 138.7 ± 46.6 to Q4 190.9 ± 56.1 amongst children aged 6 -11 years, while Murakami et al. reported a value of 190 ± 19 amongst children aged 4–10 years in Japan [32]. It is reasonable to argue that differences in age groups and in total carbohydrates’ intakes may explain the observed differences in dietary GL between the various studies. The average daily intake of carbohydrates was 144 g in the current study, compared to quartiles ranging from Q1 268.5g/day to Q4 329.6 g/day, respectively in the ARCA project study conducted by Barba et al.,0.2012 [30]. It is also noteworthy to note that most of the available studies have computed GI and GL primarily using an older version of the GI table [46], which included a smaller number of foods with a variety of GI values acquired in both healthy and diabetic participants. Amongst studies conducted on the pediatric population, this study is the first to have used the updated version of the international tables of GI values published in 2021 [41].

There are currently no recommendations on dietary GI and GL, and available research in this context is related to adults. A common fallacy is that a diet that approximates 55 constitutes a low GI dietary pattern based on the fact that food items that have a GI of 55 are classified as low GI [59]. Available evidence suggests that an overall dietary GI of 45 or below and a dietary GL below 92 are thought to be related with superior health outcomes, according to adult cohort studies [70, 71]. If these cut-offs were to be extrapolated to children, and considered for the sample of preschoolers in our study; then according to our findings, only 5% of Lebanese pre-school children are in a healthy range.

Based on logistic regression analyses, our findings showed that higher dietary GI was significantly associated with a three-fold increase in the odds of overweight and obesity amongst Lebanese preschoolers, after the adjustment for potential confounders. Studies examining the link between dietary GI and body weight have yielded inconsistent results amongst children. In line with our results, dietary GI was reported as an independent predictor of adiposity and body fat distribution in a sample of 3734 children aged 6 to 11 years in Italy [30]. In a retrospective cohort study by Spieth et al. (2000) [33], where GI was considered as the independent variable in dietary modifications, a low-GI diet was found to be a potential alternative to standard reduced-fat diets for the management of pediatric obesity. The group on the low-GI diet was found to have a mean decrease of 1.53 in BMI and a mean reduction of 2.03 kg in body weight over a period of four months, as compared to energy-restricted, low-fat diet in children aged 10–11 years. Similarly, Warren et al. examined three meals containing foods with various GIs (55, 75 and 100) and their impact on satiety [31]. Their findings show a lower caloric intake of the meals consumed subsequent to a low GI meal in comparison to a high GI meal in children aged 9 to 12 years [31]. Another study that was conducted over a six-month period in 8 European centers investigated 4 diet-related interventions [72]. The results highlighted that neither protein nor GI had a discernible impact on body composition if consumed independently, but that a diet high in protein and low in GI exerted a protective effect against overweight and obesity [72]. Other studies did not detect a link between GI and body weight in young children. For example, there was no correlation between dietary GI, GL, added sugar intake, and body composition among 2 to 7 year old children in Germany [34].

For GL, the regression analyses performed in the present study showed only a slight significant association between GL and overweight/obesity. Other investigations have also identified a positive association between GL and overweight/obesity, but this was mainly observed amongst older children than those considered in our study. For instance, in Japan, higher dietary GL, but not GI, was linked to an elevated risk of overweight in 6–11 year old children and 12–15 year old boys [62]. These observations are in line with those reported by a cross sectional study amongst British children, where a positive association was found between dietary GL and overweight in children aged 4–10 years and between dietary GL and abdominal obesity in adolescents [32]. In addition, a study carried out among healthy pre-pubertal children, subjective ratings of hunger decreased after a 6-week intervention with a low GL diet (where at least 50% of high GI items were replaced by low GI foods) [73]. In opposition, other investigations did not find any link between GL and pediatric overweight/obesity. For instance, meal GL was not a significant predictor of overweight among 6–7 year old children in Hong Kong, and the authors argued that meal GL might be associated with a higher risk of overweight in children only when coupled with high energy consumption [28]. In younger children (4–6 years old), a high GL breakfast (compared to a low GL breakfast) led to increased hunger ratings before lunch, but this did not translate into significant differences in the quantities of food and energy taken during an ad libitum lunch [29]. The discrepancies between the studies’ findings could be explained by the disparities in age group, dietary practices (cultural differences), lifestyle characteristics, measures of overweight and adiposity, dietary assessment methodologies, GL calculation methods, variations in macronutrients and fiber intakes and the confounding variables taken into account [62].

It is unclear why, in the current investigation, dietary GI’s association with overweight/obesity was of a higher magnitude compared to GL. Considering that dietary GI, unlike GL, may be more reflective of the general aspects of the diet than just carbohydrates, such as the ingested food combinations [51], the impact of dietary GI could, at least in part, be related to the overall food consumption patterns associated with GI [51, 71, 74]. It is possible to argue that the GI is a more accurate indicator of carbohydrate quality than the GL because any given dietary GL value is influenced by a combination of both GI and carbohydrate intake [75]. That is, a high GL diet could be a high carbohydrate diet with an important proportion of refined grains and beverages with added sugar [75]. Alternatively, a high GL diet may consist of a range of whole-grain foods that are abundant in nutrients, phytochemicals, and insoluble fiber but also rich in carbohydrates and slightly increased in GI [75]. In contrast, high GI diets almost always contain a high proportion of nutrient-poor carbohydrates that are highly processed and lack fiber, regardless of whether they are also high in GL. Compared to low-GL diets, low-GI diets may more accurately and realistically reflect a nutrient- and fiber-rich diet [75].

Our study had several strengths. First the updated version of the international tables of GI values was used for the assessment of dietary GI [41]. These published values were based on careful investigations that took into account cooking methods (e.g. fried, boiled, mashed and canned potatoes) to the greatest extent possible, and were based on experiments conducted amongst healthy individuals. We were also able to determine the GI and GL for mixed dishes, which represented 31% of the carbohydrate-containing foods (mixed meals). For this purpose, the GI values were determined as the weighted mean of the GI values of the meals’ constituents utilizing culture-specific recipes (Alef Baa al Tabekh) [65], and a robust software (Nutritionist Pro) for the analysis of nutrients. However, there remain some controversies as to whether summing the GI values of the individual foods in a mixed meal can be accurately used to calculate the GI of the meal [76, 77]. According to some studies, meal GI calculation may be inaccurate when participants consume a mixed meal that include fat, protein, and fiber in addition to carbohydrates [45, 78]. The inaccuracy is suggested to be caused by nutritional interactions, which may modulate the impact of carbohydrates on glycaemia. However, GI calculations appear to be the sole practically available method for epidemiologic studies like ours that have comprehensive data on carbohydrate-containing foods [34]. Another area of strength in our study is the representative sample of children from all over Lebanon, as well as the adjustment for several confounders when investigating the link between GI/GL and overweight/obesity. In fact, since the original survey had collected data on several sociodemographic and dietary characteristics, we were able to perform statistical adjustment for several potential confounders in the regression analyses.

It is crucial to highlight some of the limitations of this study. First, general limitations to allocating appropriate GI values should be recognized. These include multiple entries for the same food item, missing values for some foods, the inclusion of mostly American or Australian food item in the database, the lack of values for mixed meals, and the absence of information on discrepancies between varieties/species, degree of ripeness (e.g. banana), and composition (e.g. higher or lower amounts of fat) [63]. The observation that the same food type may have diverse GI values in different studies further complicates assigning GI values to food items [46]. In our study, we made every effort to identify GI values for food items not comprised within the international database and to use an average GI value when more than one estimate was reported. It is also noteworthy that, in our study, and in agreement with the international GI database and several other studies [28, 34], food items with negligible carbohydrate content were excluded from the dietary GI determination. Such food items included for example meat, eggs, poultry, and green vegetables, which were all assigned a GI value of zero [46]. However, some studies [43, 44] have proposed GI values for some of these foods. Other limitations in our study include the cross-sectional design of the research, and thus the results do not establish any causality between GI/GL and obesity/overweight. Cross-sectional designs are more susceptible to reverse causality and confounding, where other unmeasured factors (e.g., physical activity, genetic predisposition or socioeconomic status) may influence both dietary choices and weight status, potentially affecting the observed associations. In addition, it can be challenging to accurately assess the dietary intakes of young children as it has been proposed that parental reporting may be subject to recall bias, under- or over-reporting [79]. The fact that this study included only one 24-h recall should be also acknowledged as a limitation since it may not be reflective of the participants’ usual diet. The limitations of the 24-h recall also include its potential reliance on memory and the likelihood of social desirability bias [80]. In our study, the nutritionists who performed the recalls were extensively trained before the initiation of data collection to decrease interviewers’ errors. The fact that, in our study, the 24-h recall was performed using the USDA multiple pass approach is also a factor that reduces the errors in dietary assessment. Finally, the survey was conducted more than a decade ago and, although it may be argued that food consumption patterns and overall dietary GI/GL may have changed over time, it is unlikely that such potential changes may modify the observed association between GI/GL and overweight/obesity risk.

Conclusions

This is the first study from the region to characterize dietary GI and GL amongst preschoolers and report on their association with overweight/obesity, using data from a nationally representative survey of Lebanese preschool children. We found that a 10-unit increase in dietary GI was independently associated with approximately three -fold higher odds of overweight and obesity in 2–5 year old children after the adjustment for several confounding factors. A slight significant association was also noted between dietary GL and overweight/obesity in the study sample. Despite the ongoing debate on the usefulness of the GI/GL concepts, the findings of this study infer that dietary GI may be of public health significance, especially if practical guidance is provided to spur positive changes in dietary behavior. Further research is still needed, particularly longitudinal studies and interventional trials to confirm the observed relation between dietary GI/GL and overweight/obesity risk in preschool children. Specifically, further studies in other Middle Eastern countries are needed to further confirm these findings, given that cultural and dietary variations could affect the observed associations.

Abbreviations

BMI

Body mass index

CHO

Carbohydrate

CI

Confidence interval

EI

Energy intake

ELNAHL

Early Life Nutrition and Health

EMR

Eastern Mediterranean Region

GI

Glycemic index

GL

Glycemic load

OR

Odds ratio

SE

Standard error

USDA

United States Department of Agriculture

VIF

Variance inflation factor

WHO

World Health Organization

Authors’ contributions

C.E.: data collection, formal analysis; writing – first draft in partial fulfillment of the requirements for MSc in Nutrition at the Department of Nutrition and Food Sciences; F.A.Z.C.: formal analysis, review and editing. K.O.: review and editing; N.H.: methodology; review and editing; L.N.: conceptualization; methodology; writing—review and editing. All authors read and approved the final manuscript.

Funding

This work was supported by Lebanese National Council for Scientific Research (Beirut, Lebanon) through its support of the Associated Research Unit (ARU) on ‘Nutrition and Noncommunicable Diseases in Lebanon’, and by the University Research Board (American University of Beirut, Lebanon) (Grant number 102724).

Data availability

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

Declarations

Ethics approval and consent to participate

The study was approved by the Institutional Research Board of the American University of Beirut (Protocol number NUT.LN.13). It was conducted according to the guidelines laid down in the Declaration of Helsinki. Written informed consent was obtained from parents prior to the enrollment of the study participants.

Consent for publications

Not applicable.

Competing interests

The authors declare no competing interests.