Globally, obesity rates have risen markedly from 1975 to 2014, increasing from 3.2% to 10.8% in males, and from 6.4% to 14.9% in the female population . This constitutes a large financial and health burden on society (; WHO, 2016). In Saudi Arabia, for example, the World Obesity Federation (2015) reported that the prevalence of obesity was 24.1% in males and 33.5% in females in 2013. A global epidemic level has been reached in childhood obesity as well. In 2012, there were 23.8% and 22.6% of boys and girls, respectively, who were overweight or obese in developed countries, while the percentages were 12.9% and 13.4%, respectively, in developing countries . While in Saudi Arabia the 2010 estimated prevalence of overweight and obesity among school-aged children was 19.6% and 7.9%, respectively, higher rates were reported for adolescents, with 26.6% overweight and 10.6% obese . Body weight status during early life – whether overweight or underweight – has a direct impact on the overall health of an individual by its consequences later throughout the life course .
Nutritionally, breakfast is deemed to be the most essential meal of the day . Many studies have shown the proven positive impact of breakfast in appetite regulation  and in reducing the risk of type 2 diabetes . Despite all these facts, breakfast is frequently skipped, especially among the young generation, with almost 74% of female students being found to be skipping or irregularly consuming breakfast in Saudi Arabia .
Many causes of skipping breakfast and changing eating patterns could be considered. For instance, Saudi cultural and economic shifts, such as a recent increase in the number of working mothers together with the increasing economic wealth of the population may have contributed to changing family habits and food choices [54,8,9]. This has been found to have had a negative impact on children's breakfast intake .
Appetite has been found to fluctuate across the course of a day, with its lowest point occurring in the early morning; thus, the relatively early school and work start time in Saudi Arabia may contribute to a reduced appetite [33,28].
Research from other countries have indicated a correlation between skipping breakfast and obesity and overweight in children and adolescents, but the evidence are contradictory between studies. It was found to be significant only for boys in several studies [61,21,62,67], while in others, significant only for girls [52,29,36,66]. On the other hand, some studies take into consideration that apart from just skipping breakfast, there are other family behavior factors and breakfast locations that indirectly influence body weight changes, [43,64], and some studies address nutrient intake along with breakfast as predictors for overweight prevention [66,73,6]. However, surprisingly, Hopkins et al. (2017) demonstrated no significant effect of breakfast intake on body weight in adolescents of urban families living in the USA.
With all this uncertainty and a lack of adequate research examining the hypothesis of an association between breakfast skipping and body weight in Saudi children and young people, this systematic review aims to review the evidence on the impact of skipping breakfast on the body weight of children and young people of Saudi Arabia and the strength of that impact. The objectives of this review are:
Many different approaches and frameworks have been developed as a guide for conducting systematic reviews [42,23,39]. These guidelines are similar in linking to the main concept of a systematic review but have minor differences. The Cochrane online framework includes an easy-to-access handbook, tools for review of studies, and provides an online short training course on the process of conducting a systematic review. Reviewers following the Cochrane method are also able to use tools from different sources according to the suitability [39,26]. Therefore, the Cochrane approach was selected to be used for the current systematic review. Although the Cochrane handbook for systematic reviews of interventions is designed mainly for randomized trials, it can also be used for supporting reviews that include other observational studies . Following Cochrane guidance, a protocol for this systematic review was registered on PROSPERO (2018) with the registration number CRD42018094084.
Inclusion and exclusion criteria for selection of studies
Type of studies
All types of studies, including observational studies, were considered for this systematic review as no experimental studies were identified during a preliminary scan of the literature. Furthermore, the publication period extended over the past 20 years to include studies published from 1998 to 2018 as the initial background search of 10 years yielded only a few studies.
Type of participants
Only children and young people living in Saudi Arabia were included. The subjects' age in primary studies ranged from 5 to 24 years, as they are all students (from primary school to undergraduate university). Children under 5 years were excluded as their breakfast habits may differ from older children, taking into account that they are at home generally and therefore having the access to food differently. Participants of both genders were included and no study with disease-specific participants was eligible.
Types of intervention and comparison
All studies that considered breakfast as an intervention were included, whether considered alone or evaluated with other lifestyle variables. Furthermore, the current review included studies that define breakfast skippers as people who consume breakfast a few times a week, as well as studies that define them as never or rarely consuming breakfast.
Types of outcome measures
Body weight was the main outcome measure identified in the present review, and this was measured in terms of Body Mass Index (BMI). Considering that age ranges varied, all different cut-off references were included with no specifications.
Moreover, two secondary outcome measures were included. First, students' academic achievements as measured and classified, if applicable, by the Saudi grading system into 'Excellent, Very good, Good, Pass, and Fail' . Second, the socioeconomic status (SES) of participants and their families, using, if applicable, the classification of 'low, medium, and high' in terms of their income, educational level, and occupation.
One reviewer selected studies for inclusion in the review, following stages 1 to 5 of the Cochrane process. Stage 1 (Bibliographic database search): as guided by Cochrane ; the following bibliographic databases were located during March 2018:
• Cochrane Library
• EBSCO: AMED
• EBSCO: MEDLINE
• EBSCO: CINAHL
• Web of Science
By using the following PICO question : What is the impact of skipping breakfast on the body weight of children and young people in Saudi Arabia?, Figure 1 shows the retrieved key terms which were used for the systematic search. The `OR' operator was used in each category below, then an `AND' operator was used to combine those terms to identify relevant articles. An example of one electronic search strategy is provided in Appendix 1.
Stage 2 (Hand searching)
Referring to the Cochrane , subject-specific journals, such as BMC Obesity, BMC Nutrition, Nutrition Journal, and Journal of Nutrition Science were searched manually for articles that examine the habit of breakfast intake in Saudi Arabia and its effect on body weight. The searching panels in those journals were not of the same level of advance as in the bibliographic databases, therefore, the keywords used were limited to 'breakfast, skipping, and Saudi Arabia'.
Retrieved studies from stages 1 to 3 were screened in two ways for removing any duplication. Manually by looking at titles and then by the Mendeley Ⓡ Software for reference management (2018). Afterward, both lists were checked and compared.
Following that, studies' titles and abstracts were screened against the inclusion criteria for the purpose of selecting the relevant and excluding any irrelevant articles. In case of being unsure, the full article was reviewed to make the final decision .
Stage 4 – of searching – (Hand searching of the sources)
The reference list of all finally selected studies was screened in order to locate any potentially relevant studies.
Data extraction and study quality assessment
Two independent reviewers performed the process of data extraction and study-quality assessment of the selected studies. For data extraction, it was recommended to design a form suitable for the review, as it can vary from topic to topic . Thus, a specially designed form was used for that purpose (Appendix 2) that included the key items recommended by the Cochrane for data extraction .
The Effective Public Health Practice Project (EPHPP) quality assessment tool for quantitative studies (Appendices 3 and 4), established by Thomas and Ciliska in 1999 , was utilized for the current review, as it is suitable for all quantitative studies and covers the quality assessment domains. It is also simple, easy to use, and showed a fair inter-rater agreement in terms of individual domains, and an excellent agreement of final decision .
Both data extraction and study-quality assessment steps were undertaken first in a pilot phase. It was completed independently by both reviewers using two studies, to check for any modification needed. It was planned that in case of any disagreement between both reviewers, this would be resolved by discussion, consulting a third reviewer, or if needed the author of the primary study will be contacted for any additional information required to aid in solving the disagreement.
Once data was extracted and the quality of each study was assessed, a descriptive analysis of all included studies was undertaken. A narrative synthesis was also performed in combination with the analysis of risk of bias between studies. Moreover, due to the considerable heterogeneity between the included studies in terms of intervention and outcome definitions, a quantitative analysis of the results and meta-analysis was not appropriate.
Details of the search process results are presented in the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) flow diagram (2015) in Figure 2. A total of 57 articles were screened to remove duplications (n = 16). The remaining 41 studies were screened for their titles and abstracts, of which 26 did not meet the inclusion criteria. The final 15 articles needed a full-paper review in order to make the decision, which yields back only 8 eligible articles to be included, while 7 articles were excluded due to reasons listed in Figure 2. Forward and backward citation searches did not result in any potentially relevant studies. Therefore, the final number of included studies was eight.
Descriptive analysis of included studies
The summary of studies characteristics and key findings are presented in Table 1. All included studies used the cross-sectional design in their method. The geographic location varied among those articles. Two studies have covered three major cities (Jeddah, Riyadh, and Alkhubar) in Saudi Arabia [13,14], while one study covered a governorate in the eastern part of the country . The remaining five studies [12,18,15,1,17] have surveyed one city each, resulting in covering different areas around the country. These cities and areas were: Arar 'northern', Riyadh –twice – 'central', Jeddah 'western', and Rass 'central', respectively.
Both genders were involved in three studies [13,14]. Only one considered females alone , while Amin et al. (2008), Alsubaie (2010), Alenazi et al. (2014), and Alrethaiaa et al. (2010) have included only male participants in their surveys. As shown in Table 1, the participant's age groups ranged from 5 to 24 years old, at different school levels.
Definition of intervention/comparison
There were variations in terms of defining 'breakfast skipping' among included articles (Table 1). Five studies have considered the frequency of breakfast intake/week as a method, either by categorizing them into 3–4 groups as in Alsubaie (2010), Aloboudi, (2010), Alhazzaa et al. (2012), and Alrethaiaa et al. (2010); or numerical data of breakfast intake frequency without categorization as found in Alhazzaa et al. (2011). Two studies defined skipping as irregular breakfast intake compared to regular intake Alenazi et al. (2014), and Abalkhail and Shawky (2002). While Amin et al. (2008) have combined two methods of identification. In other words, using 'yes, no' question for asking if breakfast was taken at home on the same day of the interview, as well as the frequency of breakfast intake/previous week. Moreover, most studies have used a self-reported questionnaire for defining the intervention, while only two depended on an interview using a structured questionnaire (Table 1).
Definition of the main outcome
Body weight outcome definitions also varied across the included studies (Table 1). Overweight/ obesity was defined by height and weight measurements to determine BMI, but the cut-off values were referenced differently among the studies. However, that was not clearly defined in two papers Alenazi et al. (2014) and Aloboudi (2010). The cut-off references in the remaining six studies were the IOTF by Cole et al. (2000), the adult reference , the National Institutes for Health , and the National Health and Nutrition Examination Survey (NHANES) by Must et al. (1991).
BMI was not the only indicator, Alrethaiaa et al. (2010) and Alhazzaa et al. (2012) have considered combining BMI with % of body fat (%BF), and visceral fat (VF); as well as waist/height ratio (WHtR), respectively, in terms of defining abdominal obesity.
Studies quality assessment
Both reviewer's feedbacks were similar in terms of assessing the risk of bias, except for two papers, where there was a disagreement in the final scoring and resolved via discussing and reviewing the dictionary attached to the quality assessment tool (Appendix 4). Although both reviewer's scorings in the final stage were the same in terms of the overall rate of each study, there were some minor differences in the scoring rate of each section of the tool (Table 2). The majority of studies scored weak in the current review except for one study Alhazzaa et al. (2011) which had a moderate risk of bias.
Synthesis of key findings
Skipping breakfast and body weight
An association between skipping breakfast and BMI status of participants was found in six papers. Alsubaie (2010) found an independent and positive association between regular breakfast intake and having low BMI with a p value of 0.001 when controlling for confounding factors such as students' age, type of school, and parent's education. The same was found by Alhazzaa et al. (2012) when controlling for the effect of age only, the survey indicates that consuming breakfast on 3 days per week was significantly associated with overweight and obesity (aOR 1.44, 95% CI 1.20–1.71, p = 0.000), it was also associated significantly with abdominal obesity (aOR 1.47, 95% CI 1.22–1.76, p = 0.000). While Alhazzaa et al. (2012) controlled for age concluding that there was a statistically significant relationship between breakfast eating and BMI in both females (r = –0.09, p 0.001) and males (r = –0.13, p 0.001); however, it was noticed that it was statistically significant but not clinically significant as the r value, as shown, is very low.
Aloboudi (2010), and Alenazi et al. (2014) found that frequent breakfast skippers were more likely to have an abnormal weight (overweight/underweight) with a statistical significance of (p = 0.046) and (p 0.001, OR 0.02), respectively. While Amin et al. (2008) showed that significantly more obese versus none obese children were likely to have breakfast only 3–6 times/week (OR 2.6, 95% CI 1.9–3.6, p 0.001) or 2 times/week (OR 1.6, 95% CI 1.2–2.2, p 0.001). Also, taking breakfast at home had an inverse relationship with the development of obesity and overweight (OR 0.54, 95% CI 0.33–0.89, p = 0.018).
On the other hand, the remaining two studies have indicated no significant relationship between breakfast intake and BMI status [17,1] (Table 1). However, the latter did not provide sufficient statistical data to support the findings, as no p value was mentioned.
Half of the studies did not include the measurement of socioeconomic status (SES) as illustrated in Table 1. Abalkhail and Shawky (2002) and Alsubaie (2010) have evaluated the association of socioeconomic status with breakfast intake status and found no difference in regular breakfast intake by SES. While Alsubaie (2010) and Amin et al., (2008) examined the relationship of SES with BMI status, Alsubaie (2010) found no association, but Amin et al. (2008) stated that living in urban areas (OR 1.85, p = 0.011), low maternal education (OR 1.87, p = 0.020), maternal occupation status (OR 1.85, p = 0.014), and being in small family (OR 1.95, p = 0.004) have significantly positive association with being overweight or obese. The same was indicated by Alhazzaa et al. (2012), but the SES was not clearly measured or classified. It was based on an assumption that children in private schools have higher SES and vice versa. The survey indicated that being in private school was significantly associated with being overweight or obese (aOR 1.50, 95% CI 1.26–1.78, p = 0.000).
The breakfast intake and academic performance association was measured only in three studies , , and . Alsubaie (2010) found a negative association between low academic performance and eating breakfast regularly with a p-value of 0.001. Abalkhail and Shawky's (2002) findings indicated that regular breakfast intake has an influence on school performance, 95% CI reporting 62.9–83.1 for fail/pass, 77.7–90.3 for good, 82.5–89.9 for very good, and 83.8–91.8 for excellent in regular breakfast consumers. While Aloboudi (2010) found insignificant association.
In terms of the main outcome, the evidence generally indicates that breakfast skipping has a negative effect on student's body weight. Therefore, students who skip breakfast are more likely to have a higher BMI than those who take breakfast regularly. Gender and age effects are demonstrated as strong confounders, showing that females and older subjects are more likely to skip breakfast. The overall weakness of study quality has limited the strength of evidence, largely a consequence of poor study design and a poor description of the study design in published papers, with uncontrolled confounding factors.
This systematic review indicates that regular breakfast intake reduces the risk of overweight and obesity in children and young people in Saudi Arabia. They are more likely to have a normal BMI levels when compared to their peers who skip breakfast. Similar findings have been reported in the literature outside Saudi Arabia [61,21,62,67][52,29,36,66][43,64]. Similarly, a systematic review and meta-analysis involving 19 observational studies conducted in Asian countries suggested the protective effect of breakfast consumption against developing overweight and obesity . Also, in the USA children who skipped their breakfast are reported as having elevated levels of BMI when compared to consumers of breakfast .
However, in this present systematic review the relationship between breakfast and BMI was not statistically significant in two small studies by Alrethaiaa et al. (2010) with n = 357 subjects and Abalkhail and Shawky (2002) with n = 800 subjects. Both are in agreement with the recently published cross-sectional survey on 239 US adolescents by Hopkins et al.'s (2017), where the effect was also not significant (p 0.10). However, their focus group was only low-income urban families, and mostly African Americans, which cannot be generalized to the whole population .
Intervention and outcome definitions
This present systematic review has identified considerable variation in the method of defining overweight/obesity and measuring breakfast intake in this research field. Van Lippevelde et al. (2013), Tin et al. (2012), Tee et al. (2018), and Hopkins et al. (2017) used the same definition of breakfast as 'any food' consumed after wake up in the morning, while Utter et al. (2007) was focusing on 'at home breakfast' and defines it as 'anything' eaten 'at home' before going to school. A question could be raised here, whether eating `anything' could be considered as enough to reflect breakfast? Only Arora et al. (2012) defined breakfast as the first 'meal' consumed in the morning. Adolphus, Lawton, and Dye (2013) suggest including breakfast frequency and composition in the evaluation, as well as distinguishing between weekends and weekdays. Unfortunately, no specification was indicated among included studies.
Moreover, the four different cut-off references found in the included studies have also created confusion. In the USA three reference values were compared on children and adolescents . They reported similar estimates among three reference values, but they were not identical, especially when observing different age groups and genders. Inconsistency in comparison findings were found  and similar observations were presented by Wang and Wang (2000). Surprisingly, found significant differences between overweight and obesity prevalence – in percentages – in their sample of French children when comparing between four BMI cut-off values. The findings revealed 16.3%, 18.1%, 20.6%, and 23.9% of overweight and obesity using French, Cole et al.'s, CDC, and Must et al.'s values, respectively .
Abrantes, Lamounier and Colosimo (2003) and Flegal et al. (2001) have recommended the use of Cole et al.'s cut-off values for international level comparisons.
Considering other factors
This systematic review has found the same level of effect in both genders, which supports the findings of Nurul-Fadhilah et al., (2013) and Croezen et al. (2009). An age difference effect was also observed, as younger children are more likely to regularly consume their breakfast, as previously reported in earlier studies [21,62,64,6,66].
It is well-known that observational studies do not reflect the cause and effect relationship . However, several lifestyle and dietary factors that underlie the association between skipping breakfast and weight status were found both in the literature and reviewed papers. For instance, regular breakfast intake was strongly associated with less consumption of high-fat food, and soft drinks; and more consumption of milk and dairy [13,18]. BMI levels were also associated with physical activity, consumption of sugar-sweetened beverages, and snacking habits [18,14,17]. Moreover, eating with family negatively  and eating away from home was positively  associated with obesity and overweight.
Socioeconomic status (SES)
Among all indicators of SES measurement, the income level was missing. Also, Abalkhail and Shawky (2002) have considered only mother's education and employment status. In Uganda, mother's educational status was the stronger predictor for children's health and nutrition status when compared to other SES indicators , the same finding was reported in Cambodia , and China  but that was only for rural areas. Al-agha et al. (2017) have explained that generally as children's eating patterns are mainly influenced by their mothers, therefore mother's education and occupation have a strong impact on their health and nutritional status. However, recent Spanish cohort demonstrated a significant direct impact of both parent's SES on the health of their children, in fact, they also found that the effect was more from father's than mother's SES . Fortunately, Alomar, Parslow and Law (2018) have recently formulated two new indices that cover the SES diversities over all geographical regions in Saudi Arabia, as well as a SES classes, established from common perceptions of the population. It is anticipated that these suggested indices will contribute to a significant improvement in research in this field if a unified tool for SES is available in Saudi Arabia.
The analysis of Abalkhail and Shawky (2002), and Alsubaie (2010) were concerned with overall academic performances. However, it could be argued that self-reported academic performances in both papers may have resulted in inaccurate reporting, especially when it comes to a sensitive area for students .
Aloboudi (2010) focused on three subjects to evaluate the academic performance (Math, Science, and Reading). Arguably, there could be a chance of subject grades variability, children may achieve differently in different subjects as found by Ma (2001). It could also be anticipated that for that reason the result showed an association but failed to reach the significant level.
All included studies state that research approvals was obtained from different governmental authorities. In Saudi Arabia, the main ethics committee is The National Committee of Medical and Bioethics (NCBE) . However, the primary studies of this current systematic review did not state clearly the inclusion of that registered committee.
In addition, children are categorized as a 'vulnerable' group, and additional precautions should be taken into account when conducting research involving this category . The Research Ethic Guidebook (2018) clearly emphasizes the importance of children and parents consents and this is also found at the top of almost all ethic guidelines of Arabian countries . However, 5 out of 8 papers in the present systematic review did not state whether any participants or parents informed consent was obtained. One study stated that in Saudi Arabia parent's consents are not considered necessary especially when there is an approval from the Ministry of education .
Strength and limitation
This systematic review has successfully reached its aim of identifying studies evaluating the nutritional habits of children in many different cities around Saudi Arabia, both from published and unpublished sources, to reduce the chance of publication bias. This is the first study, to the best of the author's knowledge, to investigate the correlation between breakfast habit and children's body weight for that particular country. The review has also succeeded in correlating the relationship. However, several unavoidable limitations were observed and should be acknowledged.
A limited number of relevant studies were obtained, most of which failed to focus on breakfast as the main element of investigation. With a shortcoming caused by the study design. Moreover, failure to undertake meta-analysis was due to the clinical heterogeneity of interventions and outcomes . Lack of standardized definitions of breakfast intake/skip, overweight/obesity, and the secondary outcomes were significant among the included studies. Which may contribute to incomparable variations in evidence.
5. Conclusion and Recommendations
Overall, this systematic review suggests that breakfast consumption is associated with reducing the risk of overweight and obesity in children and young people living in Saudi Arabia, by having a direct impact on their BMI. However, the overall quality of included studies was not strong enough to support that evidence. Therefore, ascertainment and causality should not be assumed based on findings of this review, strongly suggesting a need for further large research studies in this field, to answer the question of the mechanism of that impact.
In addition, our findings suggest adopting awareness programs for children and their families, with more focus on adolescents. Such programs should target schools, universities, community centers, and media in highlighting the impact of breakfast intake in the prevention of overweight/obesity. Additionally, when children are not consuming breakfast at home, it could be highly beneficial to have healthy breakfast meal sessions at school before starting their academic lessons, where children are able either to bring their packed breakfast or have a balanced meal from the school, organized by dietitians.
In research, there is a need for a unified definition of breakfast and how to measure its frequency, a unified cut-off values of BMI suitable for the Saudi population, as well as a well-known index of measuring SES. This will all ensure a higher level of evidence reported in Saudi studies.
Conflicts of Interest
This research was conducted as a part of the MSc Advanced Professional Practice program with the University of Plymouth, UK, and the authors declare no conflict of interest or funding for this project.
The authors are grateful to the staff of the University of Plymouth for their support in completing this work, especially the library specialist as well as the statistician for their contribution and valuable help. They also acknowledge their families and friends for their endless support and encouragement.