Effects of eating breakfast and school breakfast programmes on children and adolescents : a systematic review

(1)

Martina Lundqvist, Nicklas Ennab Vogel,

Lars-Åke Levin

CMT Report 2018:2

Center for Medical Technology Assessment

Effects of eating

breakfast and

school breakfast

programmes on

children and

adolescents:

a systematic

review

Center for Medical Technology Assessment (CMT)

Center for Medical Technology Assessment is an

inter-disciplinary research unit based at the Department of

Medical and Health Sciences, Linköping University. The

primary objective of CMT is to develop and sustain a

high-quality research program in the area of health

technology assessment and evaluation studies in

health care.

(2)

Effects of eating breakfast and school breakfast

programmes on children and adolescents:

a systematic review

Martina Lundqvist

Nicklas Ennab Vogel

Lars‐Åke Levin

CMT Report 2018:2

Printed by LiU Print Services Office, Linköping University

LIU CMT RA/1802

ISSN 0283-1228

eISSN 1653-7556

(3)

Address: CMT

Department of Medical and Health Sciences Linköping University

581 83 LINKÖPING Visiting Address: CMT

Building 511-001, Entrance 76, level 13 Campus US

Linköping

Tel.: 013-28 10 00

(4)

ABBREVIATIONS

AD ACL – Activation-Deactivation Adjective Check List ASC – After School Club

BC – Breakfast Consumption BIC – Breakfast in Classroom BMI – Body Mass Index BP – Breakfast Programme BW – Body Weight

CAT – Cognitive Ability Test CDR – Cognitive Drug Research CF – Cognitive Function

CL – Cognitive Load

CPT – Continuous Performance Test

CSHQ – Children’s Sleep Habits Questionnaire CRF – Cardiorespiratory Fitness

CT – Controlled Trial

CVD – Cardiovascular Disease

d2 — Aufmerksamkeits-Belastungs-Test DABS – The Diet and Behaviour Scale DiaBP – Diastolic Blood Pressure EB – Eating Breakfast

ECLS-K – Early Childhood Longitudinal Study-Kindergarten Class ELA – English Language Arts

FRP – Free and Reduced Price HDL – High-Density Lipoprotein

HDL-C – High-Density Lipoprotein-Cholesterol HEI – Healthy Eating Index

HOMA-IR – Homeostasis Model Assessment-Insulin Resistance Index HSCL-10 – Hopkins Symptoms Checklist (10-Item Version)

(7)

2 LDL – Low-Density Lipoprotein

LDL-C – Low-Density Lipoprotein-Cholesterol LGT-3 — Lern- und Gedachtnistest

MetS – Metabolic Syndrome N/A – Not Available

NAEP – National Assessment of Educational Process

NAPLAN – National Assessment Program-Literacy and Numeracy PA – Physical Activity

PBF – Percent Body Fat

PSID-CDS – Panel Study of Income Dynamics-Child Development Supplement

QALY – Quality Adjusted Life Years RCT – Randomized Controlled Trial

RVIP – Rapid Visual Information Processing task SAT – Scholastic Assessment Test

SBP – School Breakfast Programme SysBP – Systolic Blood Pressure SES – Socioeconomic Status TAG – Triacylglycerol TC – Total Cholesterol TG – TriGlyceride

USBP - Universal School Breakfast Programme VAS – Visual Analogue Scale

VGZ – Verfahren zur Erfassung des Gefühlszustandes

WIAT(III) – Wechsler Individual Achievement Test (3-Components) WPQ – Well-being Process Questionnaire

(A+PAAC) – Physical Activity and Academic Achievement Across the Curriculum

(8)

3

ABSTRACT

Background: Breakfast is often described as the most important meal

of the day. The regularity of wholesome, daily meal patterns comprised of healthy foods is important for the physical and mental development of children and adolescents. One way to make sure that children and adolescents eat breakfast on a regular basis is to serve it at school. Several published reviews have examined the effects of eating breakfast and studied effects of school breakfast programmes on children and adolescents. Informed decisions of whether to promote eating breakfast or to introduce a school breakfast programme require a broader perspective.

The aim was to conduct a systematic review of scientific publications that study the effects potentially relevant for economic evaluations of eating breakfast or implementing school breakfast programmes for children and adolescents.

Method: In the systematic literature review, studies were identified by

searching the electronic databases PubMed, CINAHL, Web of Science and PsycINFO from January 2000 through October 2017. The following inclusion criteria guided the selection of the identified studies: published articles from peer-reviewed journals with full text in English, studies collecting primary data, quantitative studies, studies performed in countries comparable to Sweden in terms of access and quality of nutrients, economic and social conditions and publications studying at least one of the topics; academic achievement, quality of life and wellbeing, risk factors/morbidity or cognitive performance.

Results: Twenty-six studies on eating breakfast and eleven studies on

school breakfast programmes fulfilled the inclusion criteria and were judged to be of at least moderate quality were included in the analysis. The results of the review of studies on eating breakfast showed positive and conclusive effects on cognitive performance, academic achievement, quality of life, well-being and risk factors/morbidity. This indicates that eating breakfast is important. The results of the studies on school breakfast varied. Minor positive effects were seen on cognitive performance and academic achievement. Most studies showed no effects and all studies suffered from different methodological weaknesses.

Conclusions: The overall assessment of the studies on eating breakfast

indicated positive effects. The studies on school breakfast programmes showed minor positive effects, but the majority of the studies showed no

(9)

4

effects and all studies suffered from different methodological weaknesses. In order to assess the cost-effectiveness of eating breakfast and of school breakfast programmes we need to know more about how the identified affects influence the societal cost and the individuals gain in quality-adjusted life years both in the short and the long term.

(10)

5

CONCLUSION (IN SWEDISH)

Bakgrund: Frukost beskrivs ofta som dagens viktigaste måltid och flera

studier har visat att kostintag har en positiv effekt på barn och ungdomars såväl fysiska som mentala utveckling. Ett sätt att öka barn och ungdomars frukostintag är att servera frukost på skoltid. Idag finns flera publicerade översiktsatiklar som har granskat effekterna av att äta frukost såväl som effekterna av skolfrukostprogram. För att kunna fatta ett beslut om huruvida man ska lägga resurser på att främja frukost-ätande eller servera frukost på skoltid krävs dock ytterligare information. Syftet med rapporten är att systematiskt granska de vetenskapliga publikationer som studerar effekter av att äta frukost eller effekter av implementering av skolfrukostprogram potentiellt relevanta att inkludera i en ekonomisk utvärdering.

Metod: En systematisk litteratursökning genomfördes. Studierna

identifierades genom att söka de elektroniska databaserna PubMed, CINAHL, Web of Science och PsycINFO från januari 2000 till oktober 2017. De inklusionskriterier som tillämpades var: artiklar på engelska publicerade i granskade tidskrifter, artiklar som presenterar primärdata, kvantitativa studier, studier jämförbara med Sverige gällande näringsintag, ekonomiska och sociala förhållanden samt publikationer som studerar minst ett av ämnena; akademisk prestation, livskvalitet och välbefinnande, riskfaktorer/morbiditet eller kognitiv prestanda.

Resultat: Tjugosex artiklar som studerade effekter av att äta frukost

samt elva studier som studerade effekter av skolfrukostprogram uppfyllde de uppsatta inklusionskriterierna och bedömdes vara av minst måttlig kvalitet inkluderades i analysen. Studierna som studerade effekter av att äta frukost var samstämmiga. De visade att frukostätande hade positiva effekter på kognitiv prestanda, akademisk prestation, livskvalitet och välbefinnande samt på morbida riskfaktorer. Artiklar som studerade effekter av skolfrukostprogram visade på varierande resultat. Ett fåtal studier fann att skolfrukostprogram hade positiva effekter på kognitiv prestanda och akademisk prestation men majoriteten av studierna kunde inte påvisa några effekter. Flertalet av studierna var dock behäftade med metodologiska svagheter.

Slutsats: Äta frukost har överlag positiva effekter på de studerade

utfallsmåtten. Skolfrukostprogram har små till måttliga effekter på de studerade utfallsmåtten. Majoriteten av studierna som undersökte skolfrukostprogram visade dock inga effekter och alla studier var

(11)

6

behäftade med metodologiska svagheter. För att kunna bedöma kostnadseffektiviteten av att äta frukost och skolfrukostprogram, behöver vi mer kunskap om hur de identifierade effekterna påverkar samhällets kostnader och en individs livskvalitet och levnadsår, på såväl kort som lång sikt.

(12)

7

1. BACKGROUND

Breakfast is often described as the most important meal of the day. Overall, eating breakfast on a regular basis is linked to a range of beneficial health outcomes in adults, including improved cognition, physical activity, and lower risk of hypertension [2-4]. Moreover, several studies have focused on examining if breakfast habits have any bearing on school attendance, academic achievement and general health in children and adolescents. Research findings indicate that the regularity of wholesome, daily meal patterns comprised of healthy foods is important for the physical and mental development of children and adolescents [5]. The quality and regularity of breakfast consumption have also shown to be associated with mental performance, academic achievement, physical activity and quality of life [6-12]. Despite this, young people often skip breakfast [13], and the prevalence of children and adolescents skipping breakfast is increasing [14]. In addition, it has been shown that breakfast skipping is particularly common amongst children of lower socioeconomic status [15].

One way to make sure that children and adolescents eat breakfast on a regular basis is to serve it at school. A School Breakfast Programme (SBP) aims to provide a healthy breakfast in school, to improve e.g. nutrition and academic performance [16]. SBP is commonly supervised by staff, offering school children a nutritious breakfast at school before going to class [17]. A variant of SBP is Breakfast in Classroom (BIC). With BIC, students get breakfast in their classroom at the beginning of the school day, rather than in the cafeteria before school start.

In addition to research studying the effects of eating breakfast (EB), there is a growing body of literature that aims to measure the effects of School Breakfast Programmes (SBP). Several published reviews have examined the effects of EB and SBP on children and adolescents [18-21]. Informed decisions of whether to promote eating breakfast or to introduce a school breakfast programme require a broader perspective. Since societal resources are scarce, and the needs are endless, choices between health interventions must be made. Economic evaluation of different interventions seeks to identify and quantify the costs and consequences of each alternative for well-founded and informed decision-making.

(13)

8

Economic evaluations are comparative analyses of alternative courses of action in terms of both costs and consequences. The costs are weighed against the health effects measured in such a way that it corresponds to a value, usually quality adjusted life-years (QALYs). This measure should ideally encapsulate the impact of an intervention on a person’s length of life, and the impact on their health-related quality of life which is recognized as an important indicator of treatment outcomes. To our knowledge there are no published economic evaluations of EB and SBP. In the absence of cost-effectiveness studies, and as a starting point for future economic assessments, this review will focus on studies that measure outcomes that directly or indirectly can be transformed into QALYs that are traditionally used in cost-effectiveness analysis.

The aim was to conduct a systematic review of scientific publications that study the effects potentially relevant for economic evaluations of eating breakfast or implementing school breakfast programmes for children and adolescents.

(14)

9

2. METHOD

Studies were identified by searching the electronic databases PubMed, CINAHL, Web of Science and PsycINFO between January 2000 and October 2017.

To capture all relevant articles published in the field, two different search strategies were constructed. See table 1.

Table 1. Search strategies.

Search strategy 1

Breakfast

AND

Children OR Adolescents OR Teen OR Youth OR students OR kids OR pupils

AND

Behavior OR Mental OR Learning OR Effect OR Cognitive OR Academic OR Achievement OR Performance OR Well-being OR Quality of life OR Health

Search strategy 2

School

AND

Breakfast

AND

Children OR Adolescents OR Teen OR Youth OR students OR kids OR pupils

AND

Behavior OR Mental OR Learning OR Effect OR Cognitive OR Academic OR Achievement OR Performance OR Well-being OR Quality of life OR Health

2.1 Selection of studies

The overall inclusion criteria applied were:

 Published articles from peer-reviewed journals with full text in English.

 Studies collecting primary data.  Quantitative studies.

 Studies performed in countries comparable to Sweden in terms of access and quality of nutrients, economic and social conditions.  Studies using well defined and valid measures of breakfast

consumption.

 Publications studying at least one of the following topics:  Academic achievement

(15)

10  Morbidity risk factors  Cognitive performance

The association between eating breakfast and weight-related outcomes is a thoroughly studied topic with broad coverage in previous literature [22-24]. The established links for utilizing weight-related outcome measurements in cost-effectiveness analysis highlight the redundancy of further examinations of this association. Weight-related outcomes are therefore omitted from the EB section of this review [25, 26]. The review of SBP studies included studies using weight/obesity as outcome.

One of the authors (ML) conducted the search in October 2017. Initially, two of the authors (ML, NEV) read all titles and abstracts of the identified studies to determine the relevance of each article. If title and abstract met with the inclusion criteria, the study proceeded to the next stage of the review process. Studies with insufficient information in title and abstract also proceeded to the next stage of the review process. After this selection, all authors (ML, L-ÅL, NEV) read the remaining studies in full text, in order to confirm inclusion eligibility and to conduct the quality assessment.

2.2 Quality assessment

In pairs, the authors read the articles in full to assess both eligibility and scientific quality. If the two authors who made the initial assessment did not agree, the third author also read the article. The scientific quality assessment was based on the following criteria: adequate control group(s), randomization, sufficient statistical power, control of confounders, sufficient descriptions of experimental design, relevance of outcome measures and consistency between reported results and conclusions [27, 28]. In order not to let the results of the study affect the quality assessment, this was done separately without, as far as possible, reading the study results. The criteria for quality assessment varied with different study types. The quality of the studies was rated as high, moderate or low. A study rated ‘high quality’ had to fulfill all the quality criteria. Studies of moderate quality fulfilled several of the criteria. Finally, low-quality studies either failed to meet several of the criteria or had major shortcomings in certain criteria. Studies rated as being of high or moderate quality were included in the further analysis. The

(16)

low-11

quality studies were excluded. They are presented with a comment in Appendix 1. Tables 2 and 3 contain the following information from the included studies: reference, participant characteristics (number, age, gender), study design, study purpose, outcomes, authors conclusions and the scientific quality assessment.

2.3 Result compilation

A compilation of the study results based on statistical inference is presented in tables 4 and 5. A study was deemed positive if it had at least one statistically significant positive outcome measure; a study was deemed negative if it had at least one statistically significant negative outcome and a study was deemed ‘no effect’ if it showed no statistically significant results. A two-tailed p-value of 0.05 was considered statistically significant.

(17)

12

3. RESULTS

The flow chart in figure 1 illustrates the work process. Database searches identified 5,200 articles. No articles were identified through other types of article sources. After removal of duplicates, 2,958 unique articles remained. Exclusion based on information given in title and abstract resulted in the removal of 2,830 articles. Full-text articles were categorized as “Effect of Eating Breakfast” (EB) or “Effect of School Breakfast Programmes” (SBP). The full-text reading of the EB articles resulted in the additional exclusion of sixteen articles that did not meet the inclusion criteria, and eight were excluded due to low quality. Full-text reading of the SBP articles excluded 63 articles that did not meet the inclusion criteria, and three were excluded due to low quality. Finally, 26 EB articles and 11 SBP articles met the inclusion criteria and remained for further analysis. The 26 EB studies included in the analysis are summarized in table 2, and the 11 SBP studies included in the analysis are presented in table 3.

Records identified through database search (n = 5200) Sc ree n in g Inclu d ed E li gib ili ty Id en ti fi ca ti

on Additional records identified _{through other sources}

(n = 0)

Records after duplicates removed (n = 2958)

Records screened

(n = 2958) Records excluded (n = 2830)

Full-text EB articles assessed for eligibility

(n = 50)

Full-text articles excluded, with reasons

EB n = 16+8 SBP n = 63+3 EB studies included in synthesis (n = 26) Full-text SBP articles assessed for eligibility

(n = 77)

SBP studies included in synthesis

(n = 11)

Figure 1. Flow chart of the work process. PRISMA 2009 Flow

(18)

13 Ta ble 2. Sum m ar y of include d E a ti ng Breakfas t s tudies. F ir st a u thor (year), Co u n tr y Pa rt icipants Study d esi gn Study purpose O u tcomes Author conclus ion Scientifi c qua lity as se ss m ent N Age Ge n d er (f ema le) Morbidity risk fa ct ors Ha lls tr om et a l. (201 3) , SWE, ESP, B E L , DE U , F R A GRC, ITA, AUT [ 29] 2929 14. 7 ye ar s 53% Cr oss-sectional, obser vational T o e xamin e th e associ at io n be tw ee n brea kf as t consump tion and C V D r isk fa ctor s in European ad ol es ce n ts . Ca rd io re sp ir at or y fi tn es s Ph ys ical a cti vi ty Ser um TA G TC HD L -C LDL-C Gluco se F ind in gs re ga rd in g Euro pean adol esce nts confir m prev ious da ta indicatio n s: ad ol es ce nt s w h o consume bre ak fa st re gu la rl y h ave lowe r body fa t conten t than other pe er s. Re sults al so show th at regula r BC is a ss ociate d w ith hig h er cardi o-respir at or y fitne ss an d (e sp ec ia lly in male a d ol esc en ts) wi th a healthi er ca rd iova sc ula r profi le an d ne ga tion of the effect o f exce ss ad ip osity on TC and LDL-C. Mode ra te Ma rl att et al. (201 6) , USA [30] 367 14. 7 ye ar s 49% Obs er vational T o ev al uate the relationship be tween bo th brea kf as t an d fast food consump tion o n sel ecte d biomarker s a n d impor ta n t ca rdi ova scul ar and me

tabolic risk fac

to rs am on g h ea lt h y ad ol es ce n ts , an d fu rt h er ex am in e t h e r el at io n sh ip BM I PBF S-, DBP L-, HDL TG T h e finding suggests tha

t fast food and BC

ar e a ss oc iat ed w ith some meta bo li ca ll y impo rta n t chronic disease r isk fac tors in he al th y a d ol esc ents. Mode ra te

(19)

14 be twe en the se dieta ry be ha vi

ors and the known

risk fa cto r clusteri n g th at occurs with the meta bolic sy ndro me . Gluco se Insuli n e HO M A -IR

MetS cluster scor

e Moschiano et al . (201 2) , I T A [ 31 ] 800 10 ≤ ye ar s 40 .6% Obs er vational T o ass ess the p oss ible asso ciati on b etw ee n h ea d ac he and s p ec ific h abits a n d lifes tyle f acto rs. Hea d ach e Evide n ce of cl ear as sociation b etween he ad ac he a n d ir regula r in tak e of mea ls ( especially ir regula r br ea kf ast) an d sle ep distu rbance with s ignificant differences wh en comparing subjects with and without he ad ac he. Mode ra te Pa poutso u et al . (201 4) , CY P, GRC , D E U, IT A, SWE , EST, B E L , ESP [ 32] 8863 2 < 10 ye ar s 48 .8% Cr oss-sectional Obs er vational T o i n vestig at e t h e

relationship between brea

kf as t ro ut ine a n d C V D risk fa cto rs in a multination al s amp le. B lood g lucose TC LDL-C HD L -C TG PA Daily BC contr ibutes to controlli n g s chool-ag ed childre n 's weig ht an d l ipi d profil e. It prom ote s hi gh er PA. Mode ra te Sese e t al. (201 2) , ESP, GBR , FRA, BE L, DEU, A U T , HUN, G R C [33 ] 826 14. 8 ye ar s 52 % Obs er vat ional T o e xamin e th e asso ciat ions of foo d be ha vi ors and pr ef erences with mark ers of insuli n resistance a n d clu ste re d me

tabolic risk fac

to rs score a fter co n tr olling for po te ntial confo und ers, including bo dy fat in Europea n ad ol esce n ts . TG TC HD L -C B lood g lucose SB P HO M A -IR Th e r es u lt s of th is st u d y in d ic ate th at insulin resista n ce a n d a clustered met aboli c risk fact ors sco re are posit ive ly associa ted wi th fo od be ha vi ors an d preferenc es. Ski ppi ng b reakfa st ex pla in s part of the insulin resista n ce Mode ra te

(20)

15 va ri ance . Smi th et al . (2 01 0), A U S [34] 2184 N/A 53 .3% L on gitu d in al , Obs er vational T o e xamin e lo ng itudinal asso ciati ons of breakfast sk ipp in g in chil dhoo d a n d ad ulthood with ca rdi ometa bo li c r isk fa ctor s in a d ulthoo d. Mean Weight Circ u m fe re n ce Ca rd io m et abo li c R is k fa ct or s Pa rti cip an ts sk ip p ing breakfa st in bo th child h oo d and ad ulthood h ad larg er waist circumference s, hig h er B M Is, and poor er ca rd iometaboli c profiles tha n did

those who repor

ted ea tin g brea kf as t at

both time points.

Mode ra te W alter (2014) , USA [35 ] 13 57 0 11-17 ye ar s 51% Cr oss-sectional T o stud y h ow li festy le be ha vi

ors (skipping meals,

water inta ke, to ba cco use , al cohol use , an d physic al ac

tivity) and ill

n es s-relate d factors (d epression, so ma tic complaints, insomnia , a n d obesity) w ork tog eth er to pr ed ic t h ea d ac h e in a n ad ole scen t popul ati on . R ec u rre n t he ad ac he L ifestyle beha viours an d il lness-rel ate d fact or s are a sso ciat ed with adolesce nt he ad ac he. Ski ppi ng b reakfa st three or mor e time s

was one of the

m . Mode ra te W ennb er g et a l. (201 5), SWE [36] 889 16 y ears follow- up a ge : 18 , 21, 30 43 52 .2 % L on gitu d in al , Obs er vational To a n al ys e w he the r po or brea kf as t ha bit s in ad ol es cenc e pr ed ic t th e me tabol ic s yn d ro me a n d its comp onents in ad u lt h oo d. Meta bo lic s ynd rome C entral O be sity High F asting Glu co se Poor br ea kfast h abits in ado lesce n ce predicted th e meta bo lic s ynd rome in adulthood. Of th e meta bo lic s ynd rome components, poo r breakfa st habit s in ad ol esce n ce predicted cen tr al ob es ity and h igh fa st in g g lu cos e in adulthood. Mode ra te W ennb er g et a l. (201 6) , SWE [37 ] 889 16 y ears follow- up a ge : 18 , 21, 52 .2 % L on gitu d in al , Obs er vational T o i n vestig at e wheth er irr eg u lar ea ti ng of mea ls in ad ol es cenc e pr ed ic ts the me tabol ic s yn d ro me a n d Meta bo lic s ynd rome Ir regul ar eatin g of mea ls i n ado les cen ce predicted th e meta bo lic s ynd rome Mode ra te

(21)

16 30 43 its comp onents in ad ulthoo d, and if an y sp ec if ic meal is of particular impor ta n ce. in adulthood, but not independ entl y o f BM I an d li fe st yl e in ad ol es ce n ce . Po or breakfa st in ad ol es ce nc e w as t h e only sp ec ifi c m eal as sociated with future meta bo lic s ynd rome, ev en a fter ad ju st m ent s. Cognitive performance Coo p er et a l. (201 1) , G B R [3 8] 96 13. 3 ye ar s 62. 5% Ra ndomized cro ssove r de sig n T o e xamin e th e effects of brea kf as t consumptio n o n cogniti ve function, m ood an d bloo d g lucose conce n tration i n ad ole scent sch oolch il dr en .

Modified AD ACL (moo

d q uesti onnair e) VA S – h u ng er , fu llness B lood g lucose co n ce n tration C F tests: Visual se ar ch (focused att ent ion) Str oop (n eg at iv e pr im in g) Sternberg pa radig m (w ork ing memory) BC im pro ved th e accura cy of res p onses on the vis u al se arch and Str oop te st s. BC also im proved respo n se time s on the mo re co mplex l ev els of t h e S ter n ber g parad igm, but did not have co nsistent effects on resp onse time s on the othe r tests conducted. BC wa s pa rt ic u la rl y bene fi cial for th e mo re co gniti ve ly dema n d in g ta sks , w h er ea s t h e s im p le r tasks could be perfor med to a simila r le ve l foll owi n g br ea kfast omis sion. Mo d er at e Defeyte r and Russo (201 3) , GBR [39] 40 14. 2 ye ar s 52.5% Cr ossover desig n T o i n vestig at e t h e eff ect of brea kf as t consumptio n o n cogniti ve perfor ma nce and mood in ado les cen ts, and an y i n te ra ct io n t h at brea kf as t consumptio n B ond-L ad er (m ood sc al e) VA S - thirs t, hunger, sa ti et y O verall , i t appe ared tha t after brea kf as t, parti cipa n ts fel t mor e alert, sa ti at ed, and content. Only in the

recall task did

Mo

d

er

at

(22)

17 mig h t have with cog n itive load. C L tests: De la ye d w ord re ca ll (memor y C h oice rea cti on ti me (attention ) RVIP (sustained atten tion) Str oop (n eg at iv e pr im in g) Ser ia l 3s, 7 s (attention , memory ) perfor man ce appear to be sig n ificantly mo du la ted b y t h e in teractiv e co m bi-nati on of the effect of BC and ta sk dif ficulty, wi th im pr ov ed perfor man ce a t ti me two when the t ask wa s harder. Hjor th et al. (201 6) , DNK [40] 710-828 9. 9 ye ars 49% Cluster- randomiz ed cro ssove r de sig n T o e xamin e th e inde p end en t a ssociations be tw ee n weig ht st atus a n d lifesty le indicators with cogniti ve p erfor ma nce in 8 - to 11-ye ar-old D anish ch il dren. CS H Q CR F d2-test (conce ntr ati on) Sent en ce re ad in g sp

eed and corr

ec tn ess Mathe m atics pr ofi cie n cy No rmal wei ght children h ad h igher cog n itive perfor man ce com pared to

overweight/obese and und

er w eig h t chi ldren. Daily BC was associa ted with hig h er cog n itive perfor man ce i n the d2-te st, mathematics and/or sen te n ce -reading test. Mo d er at e W esnes et al. (2003), GBR [41] 29 12 ye ars 51.7 % Ra ndomized, four -w ay cro ssove r de sig n T o d ete rm in e th e ex te n t t o which breakfa st cereals woul d help to prevent decl in es in cogniti ve fun cti on i n scho ol ch il dren. CD R t es t: w ord presenta ti on, immediat e w ord re call, pi cture p res en ta ti on , s im p le re action time, dig it vig ila nc e, c h oi ce re action time, s p at ia l and nu meri c wor king me mo ry , d ela ye d w ord recall, w ord an d Skippi n g b re akfast im pa irs a tte n tio n an d episo di c memo ry, in cr ea sin g i n magni tude over the mo rning. Ing esting carbohydrates in the for m of breakfa st

cereals reduces atte

ntion defici t by mo re than half and, for some a spects of Mo d er at e

(23)

18 picture r ec ogn it ion (attentio n , wor king me mo ry , e p is od ic se co n d ar y me mo ry ) Bo n d an d L ad er (moo d, alertne ss) memor y (im m ed ia te w

ord recall), prev

ents the d eficit al together. No benefits to atte ntio n or e p is odic memor y w ith t h e gl ucose drin k; in fa ct, gr ea ter ini tial im pa irm en t wi th th e drink th an with no drink or breakf ast. Improveme n ts in alertness a n d contentm ent di d occur fo r 90 m in fol lo w ing the glucose drink, but effec ts fade d th erea fte r, w h er ea s t h e b en ef it s continu ed from the two ce re als . W esnes et al. (201 2) , G B R [42] 13 8 6 10 .5 9 ye ar s 52 % Con tro ll ed tria l T o d ete rm in e th e ex te n t t o which breakfa st cereals woul d help to prevent decl in es in cogniti ve fun cti on i n scho ol ch il dren. Pow er of a ttent ion Response spe ed variability Dig it v igilan ce ta sk C h oice rea cti on ti me task Pictur e r ecog n it io n Power o f Attention, a scor e r eflecting the abil ity to fo cus atte ntio n a n d av oi d dis tra ction, w as slowed by 7% in th os e children w h o d id not have bre ak fa st. T h e abil ity to s u sta in atte ntio n w as a lso com pr omise d, 7 % less targets being detecte d in the d igit vig ila nce task while 23% more false al arms w er e made . The a bi li ty to cor rectly iden ti fy pi ctur es wa s impa ired by 9% a n d sp ee d was slowed by 9 % . F ina lly, the response sp eed va ri abil ity w as 10 % gr ea ter in chil d ren Mo d er at e

(24)

19 wh o d id n ot hav e br ea kf as t. T h es e scor es reflect e very aspect o f co gniti ve perfor man ce assessed, show ing a com pr eh ensiv e dif fe re n ce betw een the tw o gr oups. W ide n h orn -Mülle r ( 2008) , DEU [ 43] 104 17 .2 ye ar s 46% Ra ndomized cro ssove r de sig n T o det er m ine wheth er brea kf as t ha d effects on the cog n it ive perfo rman ce an d mood of hi gh school stud ents. d2-Te st (conce ntr ati on s p ee d and atte n ti on) LG T -3 (l ea rn in g capa ci ty , imm edi at e me mo ry ) VGZ (mo od assessment scale) This cro ssove r tr ial demonstrated pos itive shor t-term effects of br ea kfast on c ognitiv e functioni n g a n d self-repor te d alertness in hig h school students. Mo d er at e Pa ge et al . (2009), USA, SVK, H U N, ROU , CZE [44] 31 21 16. 6 ye ar s 54 .7% Cr oss-sectional T o investig at e self-rate d he al th (S R H ) i n C ent ra l an d E as te rn E u ro p ea n (C E E ) adol es ce nts and de te rm in e it s asso ci at io n with psychosocia l fun cti oni n g and other di mensio ns of adolesce nt he al th . Sel f-Rat ed H eal th R-UCLA Lone li ness Scale Beck Hopeless ness Scale Ch ee k a n d B u ss Shyne ss Sc al e

MacArthur Scale of Subj

ec ti ve Soci al Status – Yo uth Ver sion Self-R at ed Ha ppin ess Self-rel at ed H ealth appe ars to be associa ted with psy chosocia l functioni n g a n d other dim ensi ons of adolesce nt h eal th in CEE y ou th. Eating br ea kfast was one of twelv e si gnifi ca n t p redictor s of SRH. Mo d er at e

Richards and Smi

th (2016) , GBR [45 ] 2307 13. 6 ye ar s 51 .5% L on gitu d in al study with two cro ss-se ctions T o i n vestig at e t h e eff ects of consuming e n ergy dr ink s a n d m is si n g brea kf as t on stress, an xi ety, and de pression in a cohor t o f se co n dary DA B S E xerci se fr eque nc y questi onnair e Sel f-As se ssed Mental T h e c u rr en t s tu d y h as prov ide d e vide n ce to sug gest tha t hig h str ess , anxiet y, an d depressio n leve ls in adolesce nts a re Mo d er at e

(25)

20 school children. Hea lth ( W P Q -i te ms ) associa ted with br ea kfast omis sion. Th e rel ati ons h ip is unli ke ly to be c ausa l in natur e and there ma y be bi-d ir ec tiona l mech an is ms i n volved , with men tal hea lth al so in fl u en cing wh et h er o r n ot br ea kfast is co n su m ed. Smi th (2010), GBR [46] 213 8 .11 _year s 50 .7 % Sepa ra te gr oups desig n T o e xamin e th e effects of consuming di ff er en t brea kf as t cereals on pa

rents' perceptions of the

al er tn es s, c og n it iv e fun cti on and othe r aspects of th e w el l-bei ng of their ch il dren. Qu es ti on na ir e me as u re s o f we ll -bein g ( alertnes s, co gniti ve diffi culties, an xi et y, d ep re ss io n , emo tiona l distress , fa tig u e, soma tic sy m p to ms , po sitive/n eg at iv e m ood, s ympto m s, bo wel p ro blems). Br ea kfast cereal consump ti on b y children i s as sociated with greate r well-being . Mo d er at e Ac a d emic achievement Bo sch loo et a l. (201 2) , NLD [47 ] 605 14. 8 1 ye ar s 56% Cr oss-sectional T o i n vestig at e wheth er ad ole scents wh o habitu al ly skip b reakfast h av e lower en d -o f-ter m g ra d es t h an ad ole scents wh o eat brea kf as t da ily. Bre akfa st C onsump tion Attention Probl em Scale Schoo l perfor m anc e – ar ith m et ic m ean o f su bjects Dutch, ma th em at ic s a n d En gl is h . Study s h ow s th at br ea kfast skippin g and schoo l perfor man ce a re rela ted, p artially mediat ed b y atte ntio n. No causal conc lusi ons drawn. Moderate Bur row s e t al. (201 7) , AU S [48] 4245 11. 33 ye ar s 50 .55% Obse rv ati on al T o conduct seco n d ary an al ys is t o e xa m in e asso ciati ons be twee n a ra n ge o f d iet ar y b eha vi ou s Di et ar y b eha vi or s NAP L AN (re adin g, writ in g, spe ll in g, Th e fi n d in gs

demonstrate the associa

tio n b etween dietary b eh av io rs and Mo d er at e

(26)

21 an d c h il d ren 's a cad em ic ac h iev em en t. gr ammar/ p u nctua ti on, num eracy) h ig h er ac ad em ic ac h iev em en t. Br ea kf

ast was only

significa

n

tly

associa

ted with the

aca d emic dom ain of wr it in g. Faug ht et al . (201 7) , C A N [ 49] 28 60 8 14. 1 ye ar s 50 .9% Obs er vat ional To c h ar ac te ri ze t h e asso ciati ons be twee n h ealth be h av io rs and se lf-reported academic ach iev em en t. Qu es ti on na ir e (a ca d em ic ac h iev em en t, PA, health y eati ng ha bi ts, sl ee p, s cr een t im e, BW-sta tus, SES) Th e present fi n d ings demonstrate th at lifestyle b eh av iors are associa ted with aca d emic ac h iev em en t. Moderate Lie n ( 2007 ), NO R [50] 730 5 15 -16 ye ar s 50 .6% Cr oss-sectional survey T o e xamin e th e relationship between me nt al distr es s, ac ad emi c pe rf or m an ce an d re gu la r brea kf as t consumptio n ac ro ss gende r and immi gratio n status. Average gr ad e for ma thema tics, written N orw eg ian, En gl ish and s ocial sci ence . HSCL-10 (menta l distress) Th e i m p licatio ns of sk ipping br ea kfast on ment al distress and aca d emic perfor man ce a re st ro n ge r f or b oy s t h an gi rls a n d stro n ger for No rweg ia ns than immig ran ts . Moderate Littlecott et a l. (201 6) , G B R [12] 3093 (base line), 3055 (follo w-up) 9-11 ye ar s 50 .8 % (b aseli n e) 49. 5% (f ol lo w -u p ) Obs er vational T o e xamin e th e l ink be tw ee n brea kf as t consump tion i n 9- to 1 1-ye ar -o ld children a n d ed u ca ti on al o u tc om es obta in ed 6-18 mo n ths later . Ed uc at io na l ou tc om es : S A T -sc or es Significa n t positiv e associa tio n b etween self-re p or te d BC a n d educ at io n al outcom es . Moderate Ptome y et al . (201 6) , USA [51] 698 7.5 year s 50 .5 %

Cluster- randomized contro

lled tri al T o det er m ine wheth er brea kf as t consumptio n o r content a ffects aca d emic ac h iev em en t me as u re d by sta n d ar d ized tests. W IAT-III Bo th BC a n d br ea kfast cont ent may be a ssociat ed w ith im pro ve d standardize d te st perfor man ce i n elem ent ar y sch ool students. Moderate

(27)

22 Sampa sa-Kanyinga & Ha milto n (201 7) , C A N [ 52] 9912 15 .2 ye ar s 48 .6% Obse rv at io n al T o in ve sti ga te th e asso ciati on b etw ee n brea kf as t consumptio n an d school co nnect ed ness an d to extend previous rese

arch on the associa

tion be tw ee n brea kf as t consump tion and ac ad em ic ac h ie vem en t. Sc h oo l c on n ec te d n es s (questionnaire) Academic perf or m an ce ( goo d ma rk s: 70 % -10 0% , poor mark s: <7 0% ) Provides supporti n g evid en ce for the associa tio n b etween reg u lar BC an d hig h er sc h oo l c on n ec te d n es s and ac ad em ic perfor man ce. Moderate Stea and Torstveit (20 14), NO R [53] 2432 15 -17 ye ar s 51 % Cr oss-sectional study T o e xamin e th e asso ciati ons be twee n se veral lif esty le habits a n d ac ad em ic ac h ie vem en t in ad ole scent girls and boy s. Sel f-r ep or ting questi onnair e (dietary-,PA-, sm okin and sn uffing ha bits , academi c ac h iev em en t) Regular meal p attern, in take of h ealt h y food items a n d be in g physicall y acti ve wer e all a sso ciated wi th in cr ea se d odds of hig h a ca d em ic ac h iev em en t, w h er ea s the int ake of unh ealth y food a n d bever ag es, smo king

cigarettes and snuffin

g were associa ted with decrea se d odds of hig h a ca d em ic ac h iev em en t. Moderate N/A – No in formatio n Av ai lab le Ta ble 3. Sum m ary of included School Breakfas t Progr a mme s tudies. F ir st a u thor (year), Co u n tr y P artici p an ts Stud y d esig n Study purpose Outcomes Author co n cl u sion Scientific qua lity as se ss m ent N A ge/ G ra d es Ge n d er (f ema le)

Cognitive performance Moor

e et a l. (201 4) , G B R 11 1 schools (n ≈ 45 0 0 ) Ag e r ange 9-11 y ea rs N/A RCT / Pre -, po st study / Cross-se ctiona l de sign To e xamin e im pa ct s of th e Pri m ary Sc hoo l Number of health y and un h ealth y i te m s Universal br ea kfast p rovis ion may reduce so ci o-ec on om ic Mo d er at e

(28)

23 [54] Fr ee Br eak fast In it ia tiv e in Wa le s on in equ ali ti es in children's dietar y beha vi ors and co gniti ve fu n ct io n in g. at breakfast Fr u it s + ve ge tables and un h ealth y i te m s during r es t o f th e day Breakfa st skipping a n d attit u des Ep iso d ic memory (gro u p ) Hy peractivity inequ al iti es in consumptio n o f health y breakfast items an d breakfast s kipping . Shemil t e t al. (2004), GBR [55] 60 42 9.83 50. 2% RC T / obser vational an al ys is To describe t h e econom ics of UK sc h ool br ea kfast clubs, t o e stima te co sts resultin g fr om c lu bs a n d t o in ve sti ga te re lat ionships betw ee n costs an d ou tc om es . L ife, He al th an d School Questionn aire scores The S tr eng th s an d Difficulties Questionn aire scores The Tr ai l Ma ki ng Tes t scores Fam il y Questionn aire

scores School attendance Punc

tu al it y Atta in ment In tention to trea t anal ysis s h ow ed th at pupi ls i n in te rv en ti on gr oup h ad: - impro ved conce n tr at io n at 3 months into st udy - f ew er reports of skippe d cla sses and skippe d da ys at school wi thin last mo nth at 1 year. Hig h er pr opo rti on of pri m ar y-aged b rea kfast club attendee s r eported ea tin

g fruit for brea

kf as t an d had bor d er lin e or abn ormal co nd uct an d

total difficulties scores (primary-a

ge d) a n d pros ocial scor e (secondary-ag ed) compared to no n -Mo d er at e

(29)

24 attend ee s. Ac a d emic achievement Anzman -F raca et a l. (201 5), USA [5 6] 446 s ch ools N/A N/A Observati ona l To estimate t h e im pa ct of a Brea kf as t in the C lassro om (BI C ) pr og ra mme on Scho ol Br ea kfa st Pro gramme pa rtic ip at io n , sc h oo l atten d an ce, and academi c ac h iev em en t.

School level: Breakfa

st p ar ti cip ation Atte ndan ce rate s Gr ad e l eve l: Aca d emic ac hi evem ent BI C -progr am s eems to: - In cr ease school breakfast partici patio n Im pr ov e ov er al l school atten d ance r at es. Mo der at e Co rco ran e t al . (201 6) , USA [57] N = 100 0 -110 0 schools n= 71 3000 -73 00 00 pupil s Grade s 4 8 N/A Observationa l, long itudinal st udy To estimate BICs impa ct on me al s pr og ra mme pa rtic ip at io n , bo dy m ass ind ex (BMI), ach iev em en t, an d atten d ance. School m eals pro gramme p ar ti cip ation BMI a n d obesity Stud ent Ac h ie vem en t Atte ndan ce Offerin g brea kf as t in classroo m ( B IC ) increa se s scho ol brea kf as t pa rt ic ipatio n . Study provides no ev id enc e of ho ped-for gains in ac adem ic per formance , o r of fe ared i n creas es i n B M I an d o be si ty. Mo der at e Klei n m an e t al. (2002) USA [58] 97 Grade r ange 3-5: 82% Grade r ange 6: 18% 59% P re -, stud y To det er m ine wh et h er nu tr ie n t in ta ke a n d academi c a n d psy chosoci al fu n ct io n in g improv e a fter the st ar t o f a uni versal-free sc h ool br ea kfast p rog ra m ( U SB P) . Hu ng er School brea kf as t p ar ti cip ation Ps yc h os oci al funct ioning Aca d emic pe rf orma n ce scores P arti cipa tio n in a sc h oo l brea kf as t programme en h anced da il y nutri ent inta ke. I m pr ov ements in nu tr ie nt in ta ke w er e asso ciat ed with sig n ificant impro vements in stud ent a cad em ic per for mance a n d ps yc hoso cial f u nct ioning while also reducing Mo der at e

(30)

25 hu n ger . Leos-Urbel et al. ( 2013), U SA [59] 72 3843 Grade s 3-8 49. 6% Observationa l, difference-in -differences des ign T o ex am in e th e im pa ct of th e im plement ati on of a uni ve rsal f ree sc h ool br ea kfast pol icy on mea ls pr og ra mme pa rtic ip at io n , atten d an ce, and academi c ac h iev em en t. Mea l pro gramme p ar ti cip ation Stud ent attendance Sta n d ar d ized ma th an d rea d in g te st scores Unive rsa l p rovisi on se rv ice s incre ase pr og ra m pa rtic ip at io n by

social rather than econom

ic pri cing me ch anisms; b y redu cing stig ma. No ev id enc e of pr ogr amm e pa rtic ip at io n in fl u en ce on e d u ca tiona l outcomes. Mo der at e Mhurchu e t al. (201 3) , NZL [60] 424 9.4 ye ars 53 % RCT 1-ye ar ste pped wedge-cluster To in vestig at e the effects of a free sch ool br ea kfast pr og ra mme (SBP) in Ne w Zeala n d on childre n ’s sc h oo l atten d ance. Second ary aim s w er e to ex am in e the impact of t h e br ea kfast pr og ra mme on chi ldren’s academi c ac h iev em en t, se lf-re po rted grades , se n se o f belongin g a t sc h ool, behavio r, sh ort-te rm hunger , bre akfast habits a n d fo od se cu ri ty . School, SB P attendance Aca d emic ac hi evem ent Se lf -reported re ad in g g rad e Se ns e of be longing Behavior Hu ng er Breakfa st habit s Fo od s ec u ri ty A fr ee SBP did not have a s ig n if ic an t ef fe ct o n ch ildren's sc hool atten d an ce or acade m ic ac h iev em en t bu t h ad sig n ificant pos itive effects on c h ildr en 's short-te rm sa tiet y ra tings, i.e . red u ce d hu n ger . H ig h er S B P atten d ance r at es may be required to infl ue n ce school atte nd an ce an d /or academ ic outcomes. Mo der at e Ribar et a l. (201 3) , USA [61] N = 10 sc h ools n = 8 0 78 Gr ad e 1 -2 : n = Grade r ange 1-5 Ra nge 47 .3 – 50 .0 % Pr e-, p os t-c h an ge de si gn To in vestig at e st ud ent outcomes asso cia ted with cha n ges in th e av ai labili ty of School m eal p ar ti cip ation School attendance Switch fro m un iv er sa l free to el igi bil ity-b ased SBP r educ ed b rea kfast pa rtic ip at io n ; l arge st redu cti on occu rri n g Mo der at e

(31)

26 328 1 3-5: n = 47 97 uni versal fr ee br ea kfasts at el em en ta ry sc h oo ls . T est scor e pe rf orma n ce am on g s tu de n ts ine lig ibl e to fr ee or redu ced-price mea ls. Sw it ch to e ligi bil ity -ba

sed provision did

not

h

arm test scor

es o r atten d ance r at es. Weight/Obes it y Gle as on an d Do dd (2009), USA [62] 2228 Grade s 1-12 50 .6 % Observat iona l, cr oss-sectiona l To estimate t h e re la ti on sh ip betw ee n pa rtic ip at io n in sc h oo l m ea l pr og ra mme s an d children's bod y m as s i n d ex (B M I) and their lik elihood of bein g ove rweig ht or obese, testin g the hy pothesis that school m eal pa rtic ip at io n in fl u en ce s st ud en ts ' w eigh t st at u s, as m easured by th eir BM I a n d indicato rs of ov er we ig h t a n d ob es it y. BMI BMI z scor e Obesity Ove rweig ht No evid ence th at SBP p artici p at io n contribute s to r isin g rates of child h oo d obesi ty . Mo d er at e Hofferth an d Cur ti n ( 200 5) , USA [63] 1268 (1997 PS ID -C D S) Ag e r ange 6-12 y ea rs N/A Observati ona l To e xamin e wh et h er in co me is li n ke d to ov erw eight in sc h oo l-ag e children. Seco nd, it e xpl ore s wh et h er fo od BMI Ove rweig ht No evid ence th at children f rom poor famil ies are m ore li ke ly to be ove rweig ht o r th at foo d pr og ra mme s contribute to overwe ight in thes e chi ldre n . Mo d er at e

(32)

27 pr og ra ms such as the Food Sta m p Pro gram, th e Na tio n al S ch oo l Lun ch Pro gram, and the S chool Bre akfa st Pro gram are asso cia ted with ov er we ig h t among ch ildre n in different inco me gr ou ps . Sudha rs ana n et a l. (201 6) , USA [64] 6495 Grade s 5 & 8 49. 7 % Observationa l, difference-in -differences To det er m ine wh et h er re ce iv in g br ea kfast at sc h ool is related to changes in ch il dre n 's we igh t betwe en th e fif th and e ighth gr ad es , an d wh et h er th e re la ti on sh ip betw ee n school br ea kfasts an d ob es it y va ri es fo r

children of different socioeconom

ic backg rounds. Obesity For mo st ch ild ren, receiving scho ol breakfast is unrelate d to ob es ity . F or ch il dre n fr om fa m ilies be low the federa l p over ty li ne, unh ealt h y ob esit y between fifth and e ighth gr ad e could be related to the re ce ipt of s ch ool breakfast in fifth gr ad e. Mo d er at e N/A – No in formatio n Av ai lab le

(33)

28

3.1 Studies of Eating Breakfast

In total, 26 studies were included in the analysis. All studies were published between 2003 and 2017. The countries represented were the UK (seven studies), the US (three studies), Canada, Norway, Australia, Sweden (two studies each), and Denmark, Germany, Italy and the Netherlands (one study each). Four studies were multinational. Most of the studies included were observational. Five were randomized controlled trials, one was a non-randomized trial and one used a separate-group design.

The number of participants included in the studies varied from 29 to 28,608. Based on the predetermined inclusion criteria, they all study children and adolescents. The majority studied adolescents [12, 29-31, 33, 35-39, 41-50, 52, 53]. Only four studied children with a mean age below ten years [32, 34, 40, 51].

The studies included a variety of outcome measures and instruments. Based on the applied criteria, studies were divided into four categories. Studies of morbidity risk factors associated with eating or not eating breakfast were most common [29-37]. In addition, studies of links between breakfast and academic achievement were frequent [12, 47-53]. In six studies, the association between breakfast eating and cognitive disorders was analyzed [38-43]. Finally, in three studies the link between breakfast eating and QoL and/or well-being was studied [44-46]. No study concerning the cost-effectiveness of eating breakfast was found. Table 4 shows a compilation of the EB study results.

3.1.1 Morbidity risk factors

All the studies that had morbidity risk factors as outcomes were observational studies. Most of the studies examined the association between breakfast consumption and the metabolic syndrome. Marlatt et al. found that higher levels of breakfast consumption were significantly associated with lower risk for insulin resistance as well as other risk factors for metabolic syndrome [30]. This is in line with what both Sese et al. and Wennberg et al. concluded in their studies [33, 36, 37]. The two studies conducted by Wennberg et al. were longitudinal with 27 years of follow-up. In both studies, they found an association between poor breakfast habits and future risk of developing metabolic syndrome. Hallstrom et al. studied the association between breakfast consumption

(34)

29

and cardiovascular disease risk factors. Their results indicated that adolescents who regularly consumed breakfast had higher cardiorespiratory fitness and lower total adiposity [29]. However, they did not find any significant effect of breakfast consumption on blood lipid levels, blood pressure or insulin resistance, in females. Two studies examined the association between breakfast skipping and headache [31, 35]. Both found that irregular breakfast eating is associated with headache.

3.1.2 Cognitive performance

A various number of instruments and test were used to measure cognitive performance among the studies included. All of them found that eating breakfast has positive effects on cognitive performance. Wesnes et al. concluded in both their studies that breakfast improved children’s attention [41, 42]. Cooper et al. found that eating breakfast had particularly beneficial effects when study participants faced more cognitively demanding tasks [38]. They also measured self-reported tension and calmness between the breakfast and no breakfast groups, but found no difference. In addition, they found that breakfast consumption improved response time measured with Sternberg paradigm, a memory-scanning task for short term memory testing. This effect was not possible to confirm with other similar tests measuring response time. Wesnes et al. also showed that breakfast affects response speed variability. Participants who had breakfast showed lower response speed variability. Hjorth et al. and Widenhron-Müller used the d2-test to measure selective and sustained attention and visual scanning speed. Hjorth et al. found a positive interaction between breakfast consumption and d2-test results [40]. Widenhorn-Müller showed beneficial effects of breakfast on short-term memory and mood but not on sustained attention [43].

3.1.3 Quality of life and Well‐being

All of the included studies used different measurements to study the association between breakfast consumption and QoL and/or well-being. However, the results are consistent. Page et al. found that eating breakfast was one of twelve significant predictors of self-rated health measures [44]. Richards and Smith provided evidence showing that stress, anxiety and depression were associated with skipping breakfast [45]. Nevertheless, they highlighted the possibility that mental health might influence whether a person consumes breakfast or not. Smith et al. also found that breakfast consumption is linked to increased well-being [46].

(35)

30

3.1.4 Academic achievement

All the studies that investigated breakfast consumption and academic achievement found positive associations between breakfast consumption and academic achievement. In a study conducted by Lien in Norway, the implications of skipping breakfast on mental distress and academic achievement were stronger for boys than girls and were stronger for native Norwegians than immigrants [50]. Burrows et al. concluded that dietary behavior overall was associated with higher academic achievement, but that breakfast was only significantly associated with the domain writing [48].

Table 4. Compilation of results from EB studies.

First author Cognitive

Performance Academic Achievement Morbidity Risk Factors QoL/ Well-being

Hallstrom et al. [29] + Marlatt et al. [30] + Moschiano et al. [31] + Papoutsou et al. [32] + Sese et al. [33] + Smith et al. [34] + Walter [35] + Wennberg et al. [36] + Wennberg et al. [37] + Cooper et al. [38] +

Defeyter and Russo [39] +

Hjorth et al. [40] +

Wesnes et al. [41] +

Wesnes et al. [42] +

Widenhorn-Müller [43] +/-

Page et al. [44] +

Richards and Smith [45] +

Smith [46] + Boschloo et al. [47] + + Burrows et al. [48] + Faught et al. [49] + Lien [50] + + Littlecott et al. [12] + Ptomey et al. [51] +

Sampasa-Kanyinga & Hamilton

[52] +

Stea and Torstveit [53] +

Number of studies indicating

(36)

31

3.2 Studies of School Breakfast Programmes

In total, 11 studies were included in the analysis. The studies were published between 2002 and 2016. The countries that were represented were the US (eight studies), the UK (two studies), and New Zealand (one study). The number of participants included in the studies varied from 97 to 723,843. The age of the participants in the studies varied. Most of them examined children and adolescents in a range of ages using data from different grades. Ribar et al. and Hofferth and Curtin studied children that were a bit younger (between 6-12 years).

Seven studies were observational studies [55-57, 59, 62-64]. Two were randomized controlled trials and the remaining two used pre-post design. Most of the studies compared regular SBP participation with non-participation. However, a few had slightly different approaches. Ribar et al. examined changes from universal free school breakfast programmes to eligibility-based programmes, which is quite different from previous studies which only investigated intervention changes in the opposite direction. The majority of the studies examined the effects of SBP. Corcoran et al. examined BIC. Anzman-Frasca et al. compared BIC with non-BIC schools (where non-BIC schools offered traditional SBP). The identified outcome measures used in the studies included were cognitive performance, academic achievement and weight/obesity. Studies of academic achievement were most common. No studies of the association between QoL and/or well-being and SBP were identified. Nor did any studies examine the association between SBP and different morbidity risk factors. Furthermore, no study of the cost-effectiveness of SBP was identified. Table 5 shows a compilation of the SBP study results.

3.2.1 Cognitive performance

Moore et al. used several different outcomes to study the effect of free SBP compared to non-free SBP [54]. They found that free SBP had no effect on episodic memory but concluded that SPB might reduce socio-economic inequalities in terms of breakfast skipping. Shemilt et al. also studied several outcome measures [55]. They showed that SBP improved

Number of studies indicating

negative effects 1 (14%) 0 (0%) 0 (0%) 0 (0%)

Number of studies indicating no

effects 0 (0%) 0 (0%) 0 (0%) 0 (0%)

(37)

32

concentration and reduced the number of days the adolescents skipped classes and school.

3.2.2 Academic achievement

Studies of the association between SBP and academic achievement showed varied results. Anzman-Frasca et al., Corcoran et al., Leos-Urbel et al. and Mhurchu et al. all concluded that serving breakfast had no effect on educational outcomes. Mhurchu et al. found that SBP had no effects on children’s school attendance rates but had significant positive effects on children’s short-term satiety ratings [60]. More specifically, Leos-Urbel et al. examined meal programme participation, student attendance and standardized mathematics and reading test scores [59]. They found no impact on educational outcomes from meal programme participation. However, results indicated that universal free SBP increased participation among children previously eligible to free school breakfast. Anzman-Frasca et al. also showed that participation rates were greater in BIC schools compared to non-BIC schools, and found an association between BIC participation and improved overall school attendance rates [56]. However, as stated above, they found no difference regarding performance in mathematics and reading in the group that received BIC versus the group that did not. Kleinman et al. differed from the other studies [58]. They found that students participating in SBP enhanced their daily nutritional intake and found an association with significant improvements in academic performance.

Ribar et al. had a slightly different approach [61]. They studied if switching from a free to an eligibility-based SBP reduced participation. They found that it led to substantially reduced participation rates. However, they found no evidence that the reduced participation harmed test scores or attendance.

3.2.3 Weight/Obesity

Three of the studies used weight/obesity as outcome measure. In principle, all three studies agreed. Gleason & Dodd found no evidence that SBP participation contributes to childhood obesity [62]. Hofferth & Curtin also found no evidence that food programmes contribute to overweight in children from poor families [63]. Sudharsanan et al. reported that for most children, receiving school breakfast was unrelated to obesity [64]. However, SBP participation in 5th grade was possibly

(38)

33

associated with weight gain between 5th and 8th grade, for children from families in poverty compared to children of similar socioeconomic status that did not receive breakfast.

The study conducted by Corcoran et al. also used weight/obesity as outcome measure [57]. In line with the other studies, they found no association of excessive weight gains or obesity rates from BIC.

Table 5. Compilation of results from SBP studies.

First Author Cognitive

Performance Academic Achievement Weight/ Obesity

Anzman-Frasca et al. [56] 0

Corcoran et al. [57] 0 0

Gleason and Dodd [62] 0

Hofferth and Curtin [63] 0

Kleinman et al. [58] + + Leos-Urbel et al. [59] 0 Mhurchu et al. [60] 0 Moore et al. [54] 0 + Ribar et al. [61] 0 Shemilt et al. [55] + + Sudharsanan et al. [64] -

Number of studies indicating positive

effects 1 (50%) 2 (29%) 2 (33%)

Number of studies indicating negative

effects 0 (0%) 0 (0%) 1 (14%)

Number of studies indicating no effects 1 (50%) 5 (71%) 3 (50%)

(39)

34

4. DISCUSSION

This review searched for studies examining the effects of eating breakfast (EB) and of school breakfast programmes (SBP) respectively. Throughout the review, the results of the two searches were handled separately. The EB studies included focused mainly on studying the effects of eating breakfast versus not eating breakfast; divided into four categories of study outcomes. Similarly, the range of study outcomes in the SBP studies were divided into three outcome categories. All studies included in the analysis met the inclusion criteria of this review and were judged to be of at least moderate quality. Consequently, the excluded studies either failed to meet the inclusion criteria or had a low study quality rating.

There was a notable variation in study design across the studies. In general, studies were either of observational or experimental design. The strongest study design for drawing causal relationship is the randomized controlled trial. Observational studies are important and common but also vitiated by several limitations. This is important to keep in mind when interpreting and assessing the value of the results presented in observational studies. For instance, it is problematic to assign the effects of eating breakfast to its sole nutritional value. Whether children eat breakfast or not and whether it has any substantial effect on study outcomes might very well be derived from a large number of unobserved influential factors, such as the children’s social environment and parental support.

Variation in the studied type of breakfast intervention also differed. In the SBP category, each intervention had a unique design. The interventions could include Breakfast In Classroom (BIC), universal-free school breakfast programme, extended availability and serving capacity of school breakfast, and the launch of information campaigns to promote breakfast eating to schoolchildren, their parents and staff. It was not possible to draw any definite conclusions regarding the importance of how the breakfast is served. However, possible mechanisms related to the serving of breakfast are discussed below.

The results of the SBP studies varied. A few studies showed that SBP had minor positive effects regarding cognitive performance and academic achievement. Studies examining the association between SBP and weight/obesity showed both positive and negative effects. However, the majority of the studies found no effects at all. In addition, no studies examined the effects of SBP on quality of life, well-being and morbidity

(40)

35

risk factors. Two of eleven SBP studies examined the effects of BIC. Both studies found that BIC improved the overall participation rate compared to SBP [56, 57]. Breakfast access can be facilitated when served in the classroom, reaching students who otherwise are unlikely to participate. However, based on the two studies included in this review it is difficult to draw any firm conclusions regarding this. Moreover, none of these studies showed any effects in any of our outcome topics. Leos-Urbel et al. found that universal-free SBP increased participation rates primarily among those who previously skipped eligibility-based school breakfast. The increased participation rates were driven by social mechanisms, as the stigma of receiving eligibility-based SBP vanished. These studies highlight reasons for introducing universal-free SBP interventions. Eligibility-based SBPs that target individual schoolchildren in need of interventional support may therefore prove unsuccessful. This supports the conclusion of Moore et al., that universal-free SBP might reduce socio-economic inequalities among schoolchildren. Many of the SBP studies included suffered from methodological weaknesses. This could potentially explain the thin set of observable effects. Whether the detected associations were of a causal nature remains unanswered, since few studies were randomized controlled trials. The most common shortcoming among the SBP studies was the lack of confounder control. For example, the recurring lack of control for whether the children had breakfast at home or elsewhere.

The review of EB studies showed positive and conclusive effects on cognitive performance, academic achievement, quality of life, well-being, and morbidity risk factors. This is clear evidence that eating breakfast is important. The consistently positive results may indicate different forms of bias. There is always a risk of publication bias that could contaminate this type of review, which could have overestimated the strength of the evidence [65]. In addition, it is also possible that authors facing non-significant results can tweak the hypothesis to better suit data. This is known as HARKing – hypothesizing after the results are known [66]. Also, the lack of control for confounders in the EB studies may have affected the results. As it is not possible to adjust for all confounders, it is difficult to establish a clear causal link between intervention and outcome [67]. EB studies that examine the association between breakfast and weight/obesity has been thoroughly studied and reviewed before. Therefore, such studies were excluded from this review. However, despite the exclusion of the weight/obesity effects of eating breakfast in this review, it should be noted that it is relevant to include weight/obesity in economic evaluations of breakfast.

(41)

36

To conduct this review, several considerations and restrictions had to be applied. Including only studies from regions comparable to Sweden restricted the study catchment area to Europe and North America, plus Australia and New Zealand. The potential exclusion of relevant studies from other nations is one limitation. In the assessment of study quality, the risk of incorrect classification is contingent. Although, by proactively disregarding the study results in the articles included when performing quality assessment, the authors took measures to limit the risk of study quality misclassification.

Studies of breakfast attract researchers from different discipline. In part, this explain the high diversity in designs of SBP interventions, the multiple definitions and the measurement methods used in studying EB. Often, systematic reviews include a meta-analysis component. Meta-analysis uses statistical methods to synthesis data from the systematic review into a single quantitative outcome. Pooling results from different instruments and outcomes may lead to between-study heterogeneity and biased meta-analysis [68]. Therefore, meta-analysis of this research area is currently unwarranted.

Our interest in finding and valuing the effects that can be attributed to breakfast eating and SBP emanates from the question of the cost-effectiveness of breakfast interventions. Our study cannot answer that question, but shows that breakfast eating and SBP do have effects that could be used in economic evaluations, especially when using decision analytic modeling, calculating the long-term cost and effects of intervention. For instance, improved cognitive performance and academic achievements may influence long-term effects on an individual’s productivity through improved school results, grades, higher education. Further, this gives higher human capital, which can be reflected in higher income, both for the individual and for society. Improvements in quality of life and well-being will have an immediate effect on the child’s/adolescent’s QALY weight but also, if the effects are sustained, on the long-term QALY gain. Finally, the health effects mediated by a reduction of long-term morbidity risk factors such as obesity, will contribute to lower morbidity/mortality and a reduction in health care costs. In order to answer the questions regarding cost-effectiveness of breakfast and interventions promoting breakfast, we need to know more about how the identified effects influence the societal cost and the individual’s QALY gain. There is also a need to develop models that can simulate breakfast interventions long-term costs and effects. In this study, we have started that work by identifying important model parameters.

(42)

37

5. CONCLUSIONS

 The overall assessment of the eating breakfast studies indicated positive effects in cognitive performance, academic achievement, quality of life and well-being, and reduced long-term morbidity risk factors.

 The school breakfast programmes studied showed minor positive effects regarding cognitive performance, academic achievement, with mixed impact on weight/obesity. However, most studies showed no effects, and all studies suffered from different methodological weaknesses.

 Universal-free SBP, such as BIC, may lead to increasing participation rates. Uncertainty remains whether it has any impact on school attendance rates and socio-economic inequalities.

 In order to answer the questions regarding cost-effectiveness of breakfast and breakfast interventions, we need to know more about the long-term effects of the interventions.

(43)

38

CONFLICT OF INTEREST

Financial support for conducting this review has been provided from Arla Foods. The authors have collected data and performed all analyses independently, and without unwarranted involvement from the funder.

(44)

39

REFERENCES

1. Moher, D., et al., Preferred reporting items for systematic reviews and meta-analyses: the PRISMA Statement. 2009, Open Medicine Publications, Inc.: United States, North America.

2. Chowdhury, E.A., et al., The causal role of breakfast in energy balance and health: a randomized controlled trial in obese adults. Am J Clin Nutr, 2016. 103(3): p. 747-56.

3. Galioto, R. and M.B. Spitznagel, The effects of breakfast and breakfast composition on cognition in adults. Advances in Nutrition, 2016. 7(3): p. 576S-589S.

4. Lee, T.S., et al., Habit of Eating Breakfast Is Associated with a Lower Risk of Hypertension. Journal Of Lifestyle Medicine, 2016. 6(2): p. 64-67.

5. Riby, L., M. Smith, and J.K. Foster, Nutrition and mental

performance : a lifespan perspective. 2012, Basingstoke: Palgrave Macmillan. xxi, 347 pages.

6. Dye, L., K. Adolphus, and C. Lawton, Associations between habitual school-day breakfast consumption frequency and academic

achievement in British adolescents. Annals of Nutrition and Metabolism, 2015. 67: p. 372-372.

7. Huang, C.J., et al., Associations of breakfast skipping with obesity and health-related quality of life: evidence from a national survey in Taiwan. International Journal of Obesity, 2010. 34(4): p. 720-725.

8. Adolphus, K., C.L. Lawton, and L. Dye, The effects of breakfast on behavior and academic performance in children and adolescents. Frontiers in human neuroscience, 2013. 7: p. 425.

9. Ahadi, Z., et al., Association between breakfast intake with anthropometric measurements, blood pressure and food

consumption behaviors among Iranian children and adolescents: the CASPIAN-IV study. Public Health, 2015. 129(6): p. 740-7.