See corresponding editorial on page 503.

The effect of breakfast composition and energy contribution on

cognitive and academic performance: a systematic review1–3
Valeria Edefonti, Valentina Rosato, Maria Parpinel, Gabriella Nebbia, Lorenzo Fiorica, Emilio Fossali, Monica Ferraroni,
Adriano Decarli, and Carlo Agostoni

ABSTRACT Cognitive and academic performance are affected by several

Background: Most studies that assess the effects of breakfast on determinants, including indicators of the quality of the school
subsequent mental abilities compared performance in subjects who (such as facilities, teaching quality, and allocated teaching time),
had or had not consumed this meal. However, characteristics of of family characteristics [such as socioeconomic status (SES)4,
breakfast itself may induce metabolic and hormonal alterations of parents’ educational level, and attitudes toward school], and
the gastrointestinal tract and potentially modify cognitive perfor- of individual characteristics (such as aptitude, motivation, and
mance. Moreover, as far as the evidence on the positive effects of behavior)—all of which have known interdependent effects (3).
having breakfast is becoming more robust, interest may shift to the Within this context, subject’s health and nutritional status are
specific characteristics of an adequate breakfast. extra determinants of interest and indicators of the performance
Objective: The objective was to summarize existing evidence on themselves (4). Research has been focused on disentangling the
the role of nutrient composition or energy intake at breakfast on the role of these factors (5). In particular, experiments conducted
accomplishment of school-related tasks and cognition. between 1930 and 1980 offered preliminary evidence that not
Design: We conducted a systematic review of the literature through having breakfast could negatively influence school performance
the PubMed database. (6). Although these early experiments had many weaknesses,
Results: From the literature search, we identified 102 articles, 15 of results of more recent studies have supported the hypothesis that
which met the inclusion criteria. Of these, 3 studies provided infor- breakfast intake can modulate school-related and cognitive tasks,
mation on the relation between cognitive and academic performance in general (7–9).
and energy intake at breakfast, 11 provided the same information for Most studies that have assessed the effects of breakfast on
the macronutrient composition of breakfast, and 1 investigated both
subsequent mental abilities and mood have simply compared
the aspects. Eleven studies considered breakfast meals differing in
performance in subjects who have or have not consumed this
glycemic index/load. Selected studies were generally carried out in
meal, although characteristics of breakfast itself (nutrient quality/
well-nourished children and adults of both sexes from general educa-
composition, size and amount of energy intake provided, and
tion. They were mostly experimental studies of short duration and had
time of consumption) may induce metabolic and hormonal
a limited number of subjects. Cognitive and academic performance
was investigated by looking at multiple domains, including mem-
1
ory, attention, reasoning, learning, and verbal and math abilities, From the Section of Medical Statistics and Biometry, Department of
with a variety of test batteries scheduled at different time points Clinical Sciences and Community Health, Università degli Studi di Milano,
in the morning. Breakfast options differed in terms of included Milan, Italy (VE, VR, MF, and AD); the Institute of Hygiene and Epidemi-
ology, Department of Medical and Biological Sciences, Università degli
foods and place and time of administration.
Studi di Udine, Udine, Italy (MP); the Pediatric Clinic (GN and LF) and
Conclusions: There is insufficient quantity and consistency among the Pediatric Emergency Unit (EF), IRCCS Ospedale Maggiore Policlinico,
studies to draw firm conclusions. However, whereas the hypothesis Milan, Italy; the Unit of Medical Statistics, Biometry and Bioinformatics,
of a better and more sustained performance with a breakfast pro- Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy (AD); and the
viding .20% daily energy intake still needs substantiation, there Pediatric Clinic, Department of Clinical Sciences and Community Health,
does appear to be emerging, but still equivocal, evidence that a Università degli Studi di Milano, IRCCS Ospedale Maggiore Policlinico,
lower postprandial glycemic response is beneficial to cognitive Milan, Italy (CA).
2
performance. Am J Clin Nutr 2014;100:626–56. Supported by contributions from the Italian Association for Cancer Re-
search (AIRC grant no. IG10415), the Italian Ministry of Health (IRCCS
INTRODUCTION grant), and the Italian Ministry of University, Education and Research (PRIN
A good deal of research has investigated the importance of 2009 X8YCBN).
3
breakfast consumption for cognitive performance. In the past, Address reprint requests and correspondence to V Edefonti, Dipartimen-
to di Scienze Cliniche e di Comunità, Università degli Studi di Milano, Via
much of this research has been undertaken in healthy young adults.
G. Venezian 1, 20133 Milano, Italy. E-mail: [email protected].
More recently, interest increases in the understanding of the role of 4
Abbreviations used: GI, glycemic index; GL, glycemic load; SES, socio-
breakfast in groups who may be more vulnerable to nutritional economic status.
deficiencies or cognitive impairment, including children and ad- Received January 16, 2013. Accepted for publication March 27, 2014.
olescents (1, 2). First published online May 7, 2014; doi: 10.3945/ajcn.114.083683.

626 Am J Clin Nutr 2014;100:626–56. Printed in USA. Ó 2014 American Society for Nutrition
 BREAKFAST AND COGNITIVE PERFORMANCE 627
alterations of the gastrointestinal tract and potentially modify the varied but generally considered breakfast as the first food or meal
neurohormonal milieu and, therefore, cognitive and academic consumed through the day, although some interventions did not
performance (7, 10–12). In particular, the macronutrient com- provide explicit control for previous intakes. Studies providing
position of the meal per se may be important with respect to a comparison of different breakfast types were included. Studies
both performance efficiency and mood: specific foods and nu- that compared breakfast and “no breakfast” options were ex-
trient combinations (eg, carbohydrate and protein) may influence cluded, unless when several breakfast treatments with specified
blood glucose and insulin concentrations, acting on brain neu- energy or composition were simultaneously available and were
rotransmitter synthesis (13). Similarly, the energy load of the compared one with the other. Studies were included regardless
breakfast meal alone is also likely to play a moderating role in of the content of the meal (eg, drinks or cereal bars). However,
the disposal of energy and neurotransmitter bioavailability, es- we excluded those studies in which breakfast interventions dif-
pecially in short-term mental processes (7). fered by the presence or absence of coffee only. Studies that
The aim of the current systematic review was to collect existing assessed the effects of glucose-based or emulsion-based ma-
evidence on the role of nutrient composition or energy intake at nipulations (including foam-like vanilla creams, gelatins re-
breakfast, in absolute or relative terms, on the accomplishment of sembling milkshakes in consistency, and spoonable creams) were
school-related tasks and cognition. Although there is still debate on excluded. Studies that considered intakes at other mealtimes were
the role of breakfast as a determinant or as a short-term indicator of excluded, except when there was a clear indication of the separate
cognitive performance, these issues may have important fallouts in effect of different breakfast options during the morning.
the definition of public health guidelines, maximizing the potential
benefits of breakfast consumption for the overall population, and in Outcome measures
the assessment of the nutritional, educational, and economic value
Studies referring to any standardized outcome measure of
of school breakfast programs already carried out in many de-
cognitive, academic (school grades and standardized achieve-
veloping and developed countries (14).
ment tests), and school (enrollment, attendance, achievement, in-
class behavior and behavior at school, and school dropout)
MATERIALS AND METHODS performance in general were included. Studies based on teachers’
subjective ratings of performance or that relied only on qualitative
Literature search strategy measures of cognitive performance were excluded. Studies that
We carried out a systematic search through MEDLINE via examined fatigue or physical endurance only were excluded.
PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) to identify all Acute (= performance assessed within 12 h of breakfast con-
the articles on the relation between cognitive, academic, and sumption) and habitual effects of breakfast manipulations (typ-
school performance and breakfast composition and energy intake ically, through school breakfast programs) were included. When
published in English up to 22 November 2013 based on the available, we included the name of the performed test, together
following string (breakfast OR “breakfast composition” OR with the corresponding psychological construct assessed. Oth-
“daily meal distribution”) & (“energy intake” OR “energy erwise, we just indicated the specific neurocognitive construct.
contribution” OR “energy expenditure” OR quality OR energy
OR skipping OR “glycemic index”) & (“intellectual perfor- Exposure measures
mance” OR “neuro-performance” OR “mental performance”
Energy intake at breakfast. Studies providing quantitative es-
OR “cognitive performance” OR “academic performance” OR
timates of total energy intake at breakfast for different breakfast
“school performance” OR performance),” following the
treatments, including absolute intakes of total energy or percent-
guidelines from the Preferred Reporting Items for Systematic
ages of daily energy intake provided by breakfast, were included.
Reviews and Meta-Analyses (PRISMA) group (15). Two authors
Studies based on standardized breakfast options with a fixed
(VR and VE) independently selected the articles and retrieved
quantitative estimate of energy intake, were excluded. Survey
and assessed the potentially relevant ones. The reference lists of
studies showing specific dietary patterns of subpopulations of
the identified articles and of other systematic reviews focusing
interest (eg, subjects of a defined ethnic origin), were not included.
on similar topics were also scanned. Discrepancies in article
Information on energy intake at breakfast was consistently ex-
selection were resolved by involving a third researcher (CA).
pressed in kilocalories throughout the article.
Breakfast composition. Studies that provided quantitative es-
Inclusion and exclusion criteria timates of the macronutrient composition of different breakfast
Articles were included or excluded according to the following options were included. This included those breakfast meals stan-
criteria. dardized for energy supply. When the authors declared in the ar-
ticles that the available breakfast options were isoenergetic or
Participants similar in energy content, we reported this information in the
tables. When the difference in energy content between treatments
Studies of children and adolescents or adults of either sex were was .10% and there was the possibility to distinguish the energy
included. We excluded studies based on subjects with acquired content associated with each effect in the statistical models, we
metabolic disorders (such as hyperlipidemia or type 2 diabetes). included the corresponding article in the analysis concerning en-
ergy intake at breakfast. We excluded from the review all surveys
Breakfast definition describing the macronutrient composition of breakfast for selected
Breakfast was defined according to the descriptions of the subpopulations of interest. We also excluded those studies that
meals or foods consumed provided in the articles reviewed. These assessed the relation between cognition and the interaction
628 EDEFONTI ET AL

between macronutrient composition of the breakfast and glucose and $2 different breakfast meals, but with no available com-
tolerance, because there was no possibility to assess the separate parisons between alternative breakfast meals; other meals pro-
effect of breakfast composition. vided together with breakfast (eg, midmorning snack, evening
meal, lunch, or a combination of meals), with no possibility to
Association between exposure and outcome measures assess the individual contribution of energy or composition of
Studies providing information on any form of relation between each meal; breakfast options based on a combination of re-
cognitive and academic performance and breakfast composition quirements on energy intake and number of “healthy” nutrients
or energy intake at breakfast were included. This included results or foods consumed. Six additional articles were identified from
derived from different statistical approaches, including simpler manual searches of reference lists of selected original and re-
tests and CIs, correlation analysis, multiple regression models, and view articles. Thus, 15 articles, providing information on 15
multivariate repeated-measures ANOVA. We excluded studies different studies were included in our systematic review. Of
providing only histogram-like representations, with means and these, 3 studies provided information on the relation between
SDs (or SEs) superimposed, to distinguish the differential effect cognitive and academic performance and energy intake at
of the available breakfast meals. Finally, we chose not to exclude breakfast, 11 provided the same information for the macronu-
studies on the basis of their quality, because of their limited trient composition of breakfast, and 1 provided both aspects of
number and the huge variation in the adequacy of descriptions the problem (Figure 1).
provided. Experimental and observational studies on the effects of
breakfast on various performance variables are presented in
Tables 1 (16–19) and 2 (19–30). The main characteristics of the
Data extraction 4 selected studies describing the relation between energy intake
Quantitative and qualitative data were extracted from each of at breakfast and cognitive and academic performance are shown in
the studies selected for in-depth review by 2 independent re- Table 1. The corresponding articles were published between 1990
searchers (VR and VE); any discrepancies were resolved after and 2007; the studies were carried out in the United States (16) and
consultation with a third author (CA) to maintain consistency. in Europe, including Sweden and Denmark (17), Spain (18), and the
Information extracted included the following: 1) general char- United Kingdom (19). All 4 studies recruited children and ad-
acteristics of the studies (first author and year of publication of olescents of either sex from local schools, including suburban
the article, country, sponsorship, number and age of the partic- (16, 17) and urban (18) schools. Two articles (16, 19) provided
ipants, percentage of males and females, inclusion and exclusion details on the mean SES of the families attending those schools;
criteria, and study location); 2) design and characteristics of the 1 article (16) chose to limit variability by using a ninth grade
intervention and presence of any school program in support of it class from a school with a middle-class background, and the
(type of design, randomization, counterbalancing and crossover other (19) chose to include in the study schools from a range of
details, and when available the number of days of observation socioeconomic areas. Moreover, 2 articles defined strict in-
and schedule, information on the dinner the night before and on clusion criteria on age range and on the presence of chronic
explicitly stated overnight fast, and schedule of the breakfast); 3)
definition of breakfast: list of the available breakfast options and
corresponding details on absolute and relative values of energy
intake and on the macronutrient composition of the different
treatment options; 4) definition of the outcome according to the
different standardized tests used in the article and information
on how and when performance was assessed with respect to
breakfast (in the standardized time unit of minutes); and 5) main
results on the association between breakfast characteristics and
cognitive and academic performance (corresponding to those
statistical models adjusted for all the available confounders, if
models were fitted).

RESULTS
From the literature search of PubMed database, we identified
102 articles, of which 83 remained when we limited our search to
humans and to the English language. Their full texts were re-
trieved for detailed evaluation. After the exclusion of 12 review
articles, 62 original research articles were also excluded because
they met the exclusion criteria indicated previously. In detail, the
most frequent reasons for exclusion were as follows: absence of
information about cognitive and academic performance and/or
breakfast; information on consumption or frequency of breakfast,
without extra details on energy intake or macronutrient com-
position of the available breakfast options; a fixed estimate of FIGURE 1. Flowchart of the study selection process for the systematic
energy intake; several breakfast options, including no breakfast review. PubMed, http://www.ncbi.nlm.nih.gov/pubmed/.
TABLE 1
Studies that evaluated the effect of energy intake at breakfast on cognitive/academic performance in different settings1
Reference Design and
(sponsorship: intervention (school Definition of the Measure of the
yes or no) Study subjects program: yes or no) Definition of breakfast outcome outcome Results Comments

Cromer et al, 1990 n = 34 Independent groups 424 kcal (government 1) Rey auditory- Tested at 60 and 240 No significant Preliminary Peabody
(16) school BF, n = 18) verbal learning test min after BF differences of BF picture vocabulary
(verbal learning options on the test revised to
and immediate measures exclude subjects
memory) considered (no with IQ ,85
Bonferroni
correction)
USA Mean: 14 6 0.4 y Overnight fast 12 kcal (low-calorie 2) Matching familiar Assessment of Tanner
BF, n = 16) figures test pubertal staging, of
(impulsivity) Hollingshead
measure of SES:
groups were fairly
similar according to
these aspects
No 41% M, 59% F BF at 0700 3) Continuous Quantification of
performance test usual BF in terms of
(sustained attention number of days per
and impulsivity) week and content
No chronic diseases, Yes Statistical analysis:
nonintellectual chi-square test of
compromised independence,
correlation analysis,
repeated-measures
ANOVA
Ninth-grade class,
suburban middle-
class background
school
Metabolic ward in
BREAKFAST AND COGNITIVE PERFORMANCE

research laboratory
Wyon et al, 1997 (17) n = 195 Independent groups: .20% daily EI or 1) Addition (math) Tested in two 40-min Reduced EI at BF was Statistical analysis:
2 standardized BF 5362/4342 kcal for lessons before significantly and chi-square test of
options with high- M and F, lunch negatively independence,
(A) or low- (B) respectively (A) associated with correlation analysis,
energy within each creativity thinking Mann-Whitney test,
sex group for 4 d C-score among M no statistical
from Tuesday to but not among F models, and no
Friday (ABAB or adjustment
AABA) variables
Sweden and 10 y Return of uneaten BF ,10% daily EI or 2) Multiplication Other tests were not No information on
Denmark food for each day 1702/1212 kcal for (math) significant habitual BF
M and F,
respectively (B)
(Continued)
629
 630
TABLE 1 (Continued )

Reference Design and

(sponsorship: intervention (school Definition of the Measure of the
yes or no) Study subjects program: yes or no) Definition of breakfast outcome outcome Results Comments

Yes: in part by % of M, F not Actual EI at BF 3) Grammatical Alternative estimates

Kellogg provided calculated from reasoning (logical of actual EI at BF
Company standardized BF thinking) on the basis of a 24-h
of Great Britain and returned food recall, with good
agreement between
the different
estimates
No diabetes or dietary No 4) Number checking
problems (math)
10 parallel classes at 5 5) Creative thinking
suburban schools (associative
thinking)
BF at home, cognitive
tests in a school
setting
López-Sobaler et al, n = 180 Cross-sectional study: $20% daily EI Spanish Scholastic Tested once, time not Chi-squared test of Self-administered
2003 (18) 2 groups based on (adequate) BF Aptitude Test (SAT- provided independence: of 91 food diary with
the % of EI at a self- 1) assessing verbal, children with a final check from
selected BF, as reasoning, and reasonable test a nutritionist
derived from a 7-d calculation abilities scores, 65.9% took
food diary (Sunday and providing both ,20% daily EI
EDEFONTI ET AL

included) single domain and

an overall score
Spain 9–13 y No ,20% daily EI For each of them Of the children with Statistical analysis:
(inadequate) BF a direct score, high scores, 49.4% chi-square test of
a centile score, and took ,20% daily independence,
an IQ score were EI (significant P Student’s t test,
provided (verbal value) Mann-Whitney test,
reasoning, logical multiple regression
thinking, and math, models, and
respectively) ANOVA with
adjustment for age,
sex, and school
Yes: Danone Mean: 11.5 y ANOVA: reasoning No quantification of
Espana SPA and overall test habitual BF
scores were
significantly higher
for subjects
consuming $20%
daily EI
(Continued)
TABLE 1 (Continued )

Reference Design and

(sponsorship: intervention (school Definition of the Measure of the
yes or no) Study subjects program: yes or no) Definition of breakfast outcome outcome Results Comments

57% M, 43% F Correlations:

reasoning and
overall test scores
correlated
positively and
significantly with %
of EI at BF
Habitual BF eaters
from a nutritional
study
Schools from Madrid
School setting
Ingwersen et al, n = 64 Repeated-measures High-GI BF: Coco Cognitive Drug Tested in 2 sessions at The decline over Statistical analysis:
2007 (19) counterbalanced, 3 Pops (35 g) (133 Research 230 min before BF morning time can 3-factor mixed
assessments on kcal and 77 GI; plus computerized and 10, 70, and 130 be significantly ANOVA
consecutive days 125 mL assessment battery min after BF reduced after the (assessment time 3
(familiarization semiskimmed milk) assessing speed of intake of a low-GI treatment 3 sex)
session + 2 testing attention, speed of BF, as compared with repeated
sessions) memory, accuracy with a high-GI BF, measures on the
of attention, for accuracy of first 2 factors;
secondary memory, attention and Bonferroni
and working secondary memory adjustment for
memory on 9 tasks pairwise
(sustained attention comparisons
and memory)
BREAKFAST AND COGNITIVE PERFORMANCE

United Kingdom 6–11 y (3 age groups) Overnight fast Low-GI BF: All Bran Tests for the other No information on
(35 g) (98 kcal and domains were not habitual BF
42 GI; plus 125 mL significant
semiskimmed milk)
No Mean: 9 y BF at 0930
41% M, 59% F No
Schools from different
SES areas
School setting
1
Coco Pops and All Bran manufactured by Kellogg’s. BF, breakfast; EI, energy intake; GI, glycemic index; IQ, intelligence quotient; SES, socioeconomic status.
2
Mean intakes for each treatment option and sex.
631
 632
TABLE 2
Studies evaluating the effect of breakfast composition on cognitive and academic performance in different settings1
Reference Design and
(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

Lloyd et al, 1996 (20) n = 16 Crossover with Low-fat, high-CHO 4 computer-based tasks: Tested in 4 sessions at No significant Similar protein
counterbalancing, (600 kcal, 98.7 g 230 min before BF difference in content provided by
same day on CHO, 15.2 g P, and 30, 90, and 150 performance among any BF treatment
4 consecutive 18.4 g fat) min after BF BF conditions,
weeks, plus excluding no BF
a familiarization
session
United Kingdom Mean: 26 6 0.4 y Self-selected similar Medium-fat, medium- 1) Bakan task (visual No significant Statistical analysis:
dinner on each CHO (601 kcal, information interaction effect of repeated-measures
occasion 74.8 g CHO, 13.8 g processing) time and BF ANOVA
P, 29.3 g fat) conditions (assessment time 3
treatment 3
order of BF
consumption)
(No) 12% M, 88% F Overnight fast High-fat, low-CHO 2) 2-finger tapping (motor Low-fat, high-CHO
(608 kcal, 56.2 g speed) BF similar in
CHO, 14.5 g P, macronutrient
38.5 g fat) composition to the
isoenergetic BF habitual BF of the
subjects, but
definitely higher in
energy content
EDEFONTI ET AL

Healthy, habitual BF BF at 0830 No BF 3) Free word recall

eaters, same (memory)
workplace
Research laboratory 100% compliance in 4) Simple reaction time
consuming all food
on plates
(No)
Benton et al, n = 106 Independent groups 50 g high-SAG (diet 1) Free (abstract and Tested at 30, 90, 150, A high-SAG BF The BF meals had
2003 (21) (230 kcal, 34.3 g concrete) word recall and 210 min after significantly similar EI and
CHO, 3.3 g P, 8.8 g (immediate and delayed BF improved on the macronutrient
fat, 42.3 GI, 7.9 memory) number of words composition
SAG, 19.75 RAG) recalled at 150 and
210 min after BF, as
compared with the
high-RAG BF, but
not earlier
(significant
treatment 3 time
interaction term)
(Continued)
TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

United Kingdom Mean: 21 y Overnight fast 50 g high-RAG (diet A high-SAG BF Statistical analysis:
2) (219.5 kcal, 31.3 significantly 3-factor ANOVA
g CHO, 3.65 g P, improved on the (treatment 3
8.85 g fat, 65.9 GI, number of abstract assessment time 3
0.05 SAG, 21.15 words recalled immediate/delayed
RAG) throughout the recall) with
morning repeated measures
on the last 2 factors
(No) 100% F BF at 0900 Because there was no
interaction with
immediate/delayed
recall factor,
immediate and
delayed scores were
summed to get
a global score
80% habitual BF (No) No quantification of
eaters habitual BF
Research laboratory
Benton and Nabb, n = 323 Independent groups 50 g high-SAG (230 1) Free word recall Tested at 30, 90, 150, A high-SAG BF Partly overlap with
2004 (22) kcal, 34.0 g CHO, (immediate and 210, 270, 330, and significantly Benton et al 2003
3.3 g P, 8.8 g fat, 42 delayed memory) 390 min after BF improved on the (21)
GI, 7.9 SAG, 18.8 number of words
RAG) recalled, as
compared with the
high-RAG BF, at
210 min after BF
(significant
BREAKFAST AND COGNITIVE PERFORMANCE

treatment 3
assessment time
interaction)
United Kingdom 3 studies combined Overnight fast 50 g high-RAG 2) Simple and choice No significant The 2 BF meals
(Choco Krispies: reaction times differences between offered an identical
219 kcal, 34.0 g SAG and RAG BF weight of CHO, but
CHO, 4.9 g P, 7.3 g for the other tasks different amounts
fat, 66 GI, 0.05 of RAG and SAG
SAG, 21.1 RAG; or
Coco Pops: 220
kcal, 34.0 g CHO,
3.9 g P, 8.3 g fat,
0.4 SAG, 21.6
RAG) BF
(Continued)
633
TABLE 2 (Continued ) 634

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

(Yes: in part by Mean: 20 y BF time not provided No BF 3) Sustained attention The high-RAG BF
Danone Vitapole) was obtained with
Coco Pops in study
2 and Choco
Krispies in study 3,
but the RAG and
SAG contents were
very similar
100% F 100% compliance in The primary aim of
consuming all food the article was to
on plates study the
interaction between
different BF
treatments and
alcohol consumed
with the previous
dinner and lunch;
here, we reported
results on the effect
of BF meals during
the morning for the
no-alcohol-intake
EDEFONTI ET AL

group
Healthy, no history of (No) Statistical analysis: 4-
gastrointestinal factor ANOVA
problems, BMI of (treatment 3
20 to 25 kg/m2 alcohol at dinner 3
assessment time 3
immediate/delayed
recall) with
repeated measures
on the last 2
factors; because
there was no
interaction with
immediate/delayed
recall factor,
immediate and
delayed scores
were summed to
get a global score
Recruitment with No information on
response to a poster habitual BF
Research laboratory
(Continued)
TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

Mahoney et al, 2005 n = 60 Counterbalanced High-GI: ready-to-eat 6 constructs: Tested 60 min after BF Experiment 1: The 2 BFs provided
(23)2 randomized cereal (36 g) and in weeks 1, 2, and 3 approximately
crossover, same day 125 mL skim milk and 215 min equal amounts of
on 4 consecutive (200 kcal, 36 g before BF in weeks fat, sugar, and EI
weeks CHO, 5 g P, 1.5 g 2, 3, and 4 for the but different
fat, 1 g fiber, long-term memory amounts of fiber
22 g sugars) assessment and protein
USA Experiment 1: n = 30 Overnight fast Low-GI: oatmeal (43 1) Spatial memory A low-GI BF Statistical analysis:
g) and 125 mL skim significantly repeated measures
milk (200 kcal, 38 g improved short- ANOVA including
CHO, 8 g P, 2 g fat, term memory, in F treatment as
3 g fiber, 19 g only, in the a repeated measure
sugars) isoenergetic backward digit span and sex
BF task (significant
treatment 3 sex
interaction), as
compared with the
high-GI BF
(No) 9–11 y BF at 0815–0830 No BF 2) Backward and forward No difference for Quantification of
digit span tasks (short- forward digit span usual BF by using
term memory) task short surveys on
how often children
consumed BF
before school,
quality of their
normal BF, and
their opinions on
the BF meals used
BREAKFAST AND COGNITIVE PERFORMANCE

in the study
Experiment 2: n = 30 (No) 3) Rey complex figure No significant
copy and recall test at differences between
20 and 50 min high- and low-GI
(immediate and BFs for the other
delayed visual tasks, regardless of
perception) the different
dependent
measures and
indexes included
4) Visual sustained Experiment 2:
attention
(Continued)
635
 636

TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

6–8 y 5) Auditory sustained A low-GI BF

attention significantly
improved on short-
term memory,
among F only, in the
backward digit span
task (significant
treatment 3 sex
interaction), as
compared with the
high-GI BF
50% M, 50% F 6) Verbal memory No significant
differences for
forward digit span
task
Healthy subjects, Long-term memory: A low-GI BF
$52% habitual BF recalling material significantly
eaters, no learning learned the previous improved the
disabilities and week during the spatial number of hits and
dietary restrictions, learning, visual the number of
not taking perception, and verbal misses for the
EDEFONTI ET AL

medication memory tasks auditory sustained

attention test
Private school with No significant
middle-class differences for the
background number of false
alarms or the
reaction times to
hits or false alarms
of the same task
School setting No significant
differences between
high- and low-GI
BFs for the other
tasks, regardless of
the different
dependent
measures and
indexes included
(Continued)
TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

Ingwersen et al, 2007 n = 64 Repeated measures (A) High-GI: Coco Cognitive drug research Tested in 2 sessions at The decrease over No control for GL,
(19) counterbalanced Pops (35 g) (133 computerized 230 min before BF morning time can different
kcal and 77 GI; assessment battery and 10, 70, and 130 be significantly macronutrient
plus 125 mL assessing 5 constructs: min after BF reduced after the composition of the
semiskimmed milk) intake of a low-GI BF treatments
BF, as compared
with a high-GI BF,
for accuracy of
attention and
secondary memory
United Kingdom 6–11 y (3 age groups) 3 assessments on (B) Low-GI: All Bran 1) Speed of attention Tests for the other Statistical analysis: 3-
consecutive days (35 g) (98 kcal and domains were not factor mixed
(familiarization 42 GI; plus 125 mL significant ANOVA
session + 2 testing semiskimmed milk) (assessment time 3
sessions) treatment 3 sex)
with repeated
measures on the
first 2 factors,
Bonferroni
adjustment for
pairwise
comparisons
(No) Mean: 9 y Overnight fast 2) Speed of memory No information on
habitual BF
41% M, 59% F BF at 0930 3) Accuracy of attention
Schools from different (No) 4) Secondary memory
SES areas
School setting 5) Working memory on 9
BREAKFAST AND COGNITIVE PERFORMANCE

tasks (immediate/
delayed word
recognition,
immediate/delayed
word recall, simple/
choice reaction time,
digit vigilance, spatial/
numeric working
memory) (sustained
attention and memory)
(Continued)
637
 638

TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

Benton et al, 2007 n = 19 Unbalanced repeated High-GL (196 kcal, 1) Classroom behavior in Tested in 2 sessions at ANOVA: significant Assessment of SES
(24) measures: 3 BF 33.9 g CHO, 4.7 g P, 3 time blocks of 10 min 180 min after BF interaction effect of through Welsh
meals designed to 1.7 g fat, 17.9 GL) treatment 3 time index of multiple
be isoenergetic for head down deprivation: groups
working; in the first were fairly similar
10 min spent doing according to SES
their schoolwork,
eating the low-GL
meal significantly
increased time
spent on task as
compared with the
medium- or high-
GL meal
United Kingdom 5–7 y Each day in a 4-wk Medium-GL (168 2) Recall of objects test of No significant effects No control for GI,
program; 1 BF kcal, 21.7 g CHO, the British ability scale on immediate/ different
provided at random 8.9 g P, 5.2 g fat, for memory (verbal/ delayed memory or macronutrient
to all children 12.1 GL) spatial, immediate/ attention composition of the
delayed memory in BF treatments
their combinations,
EDEFONTI ET AL

with a card of pictures)

(Yes: in part by Mean: w6 y Tests on the second Low-GL (157 kcal, 3) Reaction to frustration Correlations: better Statistical analysis:
the British day took place after 5.7 g CHO, 10.8 g P, in 10 sessions of 15 immediate (but not multiple regression
Broadcasting randomly eating 10.2 g fat, 2.9 GL) trials delayed) verbal models, repeated-
Corporation) one of the meals BF3 memory was measures ANOVA
that was not significantly
consumed associated with
previously a lower GL
47% M, 53% F Recording of % of 4) Paradigm of Shakow When a 12-s delay For classroom
uneaten food for ability to sustain was tested for behavior: 3-factor
attention in 4 blocks of a second time, those repeated-measures
6 trials (auditory who had consumed ANOVA (order of
warning 3 or 12 s prior) a lower-GL BF BF consumption 3
were less likely to time 3 treatment)
display lapses in (time = 3 blocks of
attention 10 min)
(Continued)
TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

1 class from a lower- BF at 0815–0845 For memory: 3-factor

SES school repeated-measures
ANOVA (order of
BF consumption 3
treatment 3
immediate/delayed
recall)
School setting (Yes) For attention/reaction
to frustration: 3-
factor repeated
measures ANOVA
(order of BF
consumption 3
treatment 3 trial)
Order of BF
consumption
relevant for reaction
to frustration only
No information on
habitual BF
Smith and Foster, n = 38 Independent groups High-GI: Cornflakes Modified California Tested at 20 A high-GI BF No significant
2008 (25)4 (30 g, 77 GI) verbal learning test (immediate free significantly differences between
and 125 mL (immediate, short- recall), 60 (short- improved the the treatment
semiskimmed milk delay and long-delay delay free and cued number of items groups in terms of
(172 kcal, 31.2 g memory) with recall), and 100 remembered at the age, BMI, and
CHO, 6.5 g P, 2.0 g a secondary motor task (long-delay free and long delay, as average days per
fat) of equal importance cued recall) min compared with the week BF skipped
BREAKFAST AND COGNITIVE PERFORMANCE

(memory under after BF short delay, for the

conditions of divided remembering/
attention at encoding) forgetting indexes
(significant
treatment 3
assessment time/
delay interaction
effect), whereas the
number of items
recalled reduced
progressively at
each recall phase in
the low-GI BF
(Continued)
639
TABLE 2 (Continued )
640

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

Australia 14–17 y Overnight fast Low-GI: All Bran A high-GI BF Statistical analysis:
(30 g, 30 GI) significantly Student’s t test,
and 125 mL improved the ANOVA with
semiskimmed milk number of items assessment time/
(158 kcal, remembered, as delay, treatment
20.5 g CHO, 8.8 g P, compared with the and treatment 3
2.8 g fat) (almost) low-GI BF, at the assessment time/
iso-energetic BF5 long delay only, for delay interaction;
the remembering/ details were not
forgetting indexes provided, post hoc t
(significant tests
treatment 3
assessment time/
delay interaction
effect)
(No) Mean: w16 y BF at 0830 No significant Quantification of
difference between usual BF in terms of
high- and low-GI average number of
BFs in the raw days per week BF
memory retention skipped
scores for free or
cued recall phases
EDEFONTI ET AL

at the short or long

delay
50% M, 50% F 100% compliance in There was a trend
consuming all food of greater
on plates remembering at the
long delay as
compared with the
short delay
(significant time/
delay effect)
Healthy subjects (No)
Independent and
government schools
School setting
Micha et al, 2010 (26) n = 60 Cross-sectional study: High-GL, low-GI 1) Word-generation task Tested at 90–120 min A high-GI BF Pilot study of Micha
4 groups based on (502.2 kcal, 81.2 g (verbal fluency) after BF significantly et al, 2011 (27)
median GI and GL CHO, 17.9 g P, improved
content of self- 14.0 g fat) immediate word
selected BF recall (significant
GI effect)
(Continued)
TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

United Kingdom 11–14 y Overnight fast High-GL, high-GI 2) Immediate word recall A low-GI and a high- 2-step BF: anything
(378.7 kcal, 65.2 g task GL BF significantly eaten or drunk on
CHO, 12.6 g P, 9.3 g improved speed of the way to school or
fat) information since arriving at
processing school (“snack”)
(significant GI and was considered if it
GL effects) and contributed .10 g
Serial Sevens task CHO; however, the
(significant GI and covariate “having
GL main effect and a snack” was
interaction) unrelated to the
cognitive domains
(No) Mean: w13 y BF at 0815 Low-GL, low-GI (272 3) Stroop selective A high-GL BF The BF meals differed
kcal, 34.5 g CHO, attention task significantly in energy content,
10.2 g P, 11.3 g fat) improved the % energy from fat
matrixes task and CHO but not
(significant GL from protein
effect)
40% M, 60% F (No) Low-GL, high-GI 4) Matrixes (inductive Tests for the other However, the 2 low-
(240.3 kcal, 35.4 g reasoning) domains were not GL BFs had similar
CHO, 9.7 g P, 7.7 g significant energy contents,
fat) BF and the 32%
difference in energy
between the 2 high-
GL BFs was not
significant
Habitual BF eaters 5) Number search task Statistical analysis:
BREAKFAST AND COGNITIVE PERFORMANCE

(speed of information 2-factor ANOVA

processing and including GI and
selective attention) GL as main effects
and adjusted for
age, sex, SES
group, height,
weight, BMI, mood,
hemoglobin and
glucose
concentrations, and
time between BF
and the first test
(Continued)
641
TABLE 2 (Continued ) 642

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

Healthy subjects, no 6) Serial Sevens task Several potential

chronic diseases (working memory) confounders were
and learning explored by
disabilities multiple regression
analysis; the
significant ones
were initially
included in the
ANOVA but were
removed when not
significant
2 schools in London 7) Delayed word recall The GI and GL of the
(verbal fluency, speed self-selected dinner
of information the night before was
processing, attention, unrelated to
inductive reasoning, cognitive
and memory) performance
BF at home, cognitive Significant interaction
tests in a school GI 3 SES and GI
setting 3 GL 3 SES for
Serial Sevens
task, with better
EDEFONTI ET AL

performance in
lower-SES groups;
significant GL 3
sex and GI 3 sex
interactions in
matrixes and speed
of information
processing,
respectively
Micha et al, 2011 (27) n = 64 (32 matched Balanced, High-GL, low-GI 1) Word-generation task Tested in 2 sessions at No. of words of the Pilot study to assess
pairs) randomized, (470 kcal, 86.6 g (verbal fluency) 90 min after BF by word-generation the validity of the
crossover CHO, 13.9 g P, using 2 versions of task increased cognitive test on
7.1 g fat, 41 GL, the same test significantly after a comparable
48 GI) battery the intake of a low- population
GI BF (significant
GI effect)
(Continued)
TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

United Kingdom 11–14 y Each pair given by High-GL, high-GI 2) Immediate word recall Number search task Within the same GL
a high-GL and low- (469 kcal, 90.4 g task and Serial Sevens group, high- and
GL group child CHO, 14.0 g P, 5.3 g task showed low-GI BFs had
fat, 55 GL, 61 GI) a significantly similar energy and
better performance macronutrient
in the high-GI BF compositions,
group (significant separating any GI
GI effect) effects from
potential EI and
macronutrient
content differences
(Yes: in part by Mean: w13 y Within each GL Low-GL, low-GI (281 3) Stroop selective Time of completion in Statistical analysis:
School Food group, children kcal, 43.2 g CHO, attention task Stroop task was repeated-measures
Trust) were given a high- 12.5 g P, 6.4 g fat, significantly lower ANOVA including
GI and a low-GI BF 21 GL, 48 GI) in the high-GL, GI and GL as main
high-GI BF effects and
(significant GI and adjustment for age,
GL interaction sex, order of BF
effect) consumption,
height, weight,
BMI, mood, and
glucose and cortisol
concentrations at
baseline
% of M and F not 3 assessments Low-GL, high-GI 4) Matrixes (inductive Tests for the other No quantification of
provided (screening + 2 (276 kcal, 45.2 g reasoning) domains were not habitual BF
testing sessions CHO, 12.0 g P, 5.1 g significant
2 wk apart) fat, 28 GL, 61 GI)
BF
BREAKFAST AND COGNITIVE PERFORMANCE

Habitual BF eaters Self-selected similar 5) Number search task

dinner on each (speed of information
occasion processing and selective
attention)
No chronic diseases or Overnight fast 6) Serial Sevens task
learning (working memory)
disabilities, no
underweight or
obese subjects
5 schools in London BF at 0800 7) Delayed word recall
(verbal fluency, speed of
information processing,
attention, inductive
reasoning, and memory)
(Continued)
643
 644

TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

School setting 100% compliance in

consuming all food
on plates
(No)
Brindal et al, 2012 n = 39 Randomized, High-GL, no dairy 6 constructs: Tested in 3 sessions No significant effect Statistical analysis:
(28) crossover (313 kcal, 50 g before BF and 60, of BF conditions on mixed-effects
CHO, 7 g P, 9 g fat, 120, and 180 min changes in models with a 2-
33 GL) after BF cognitive factor (assessment
performance for time 3 treatment)
any of the cognitive interaction and
domains adjustment for age,
sex, BMI, day of
testing (first,
second, or third),
baseline scores, and
sex 3 treatment
Australia 10–12 y 4 consecutive morning Medium-GL, 1) Speed of processing No information on
assessments medium-dairy habitual BF
(familiarization (312 kcal, 45 g
EDEFONTI ET AL

session + 3 testing CHO, 14 g P, 9 g fat,

sessions) 24 GL)
(Yes: in part by the Mean: w12 y Overnight fast Low-GL, high-dairy 2) Short-term memory
Dairy Health and (315 kcal, 38 g
Nutrition CHO, 18 g P, 10 g
Consortium) fat, 18 GL)
isoenergetic BF
67% M, 33% F BF at 0830 3) Working memory
No chronic diseases or 100% compliance in 4) Perceptual speed
learning consuming all food
disabilities, on plates
attention deficit, not
on diet
Recruitment with (No) 5) Attention switching
public
advertisement or
from an existing
database
(Continued)
TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

Research laboratory 6) Inspection time,

assessed by using
several individual tasks,
including Rey auditory-
verbal learning test for
short-term memory,
backward digit span
task of Weschler
Intelligence Scale for
children for working
memory, finding “A”
task for perceptual
speed (memory and
attention)
Cooper et al, 2012 n = 52 (41) Balanced, High-GI (for a 50-kg 1) Stroop selective Tested in 3 sessions at A low-GI BF Each test was
(29)2 randomized, participant: 422 kcal, attention task (baseline 30 and 120 min significantly preceded by
controlled, 75.0 g CHO, 14.3 g P, and color-interference after BF improved accuracy practice stimuli
crossover 7.2 g fat, 54 GL, levels) (executive in the Sternberg and with a feedback to
72 GI) function and selective Flanker tests across allow the
attention) the morning, on the participants to
more complex test refamiliarize
levels only, as themselves
compared with a
high-GI BF
(significant treatment
3 assessment time
3 test level
interaction effect)
BREAKFAST AND COGNITIVE PERFORMANCE

United Kingdom 12–14 y 4 separate morning Low-GI (for a 50-kg 2) Sternberg paradigm for A low-GI BF The high- and low-GI
assessments participant: 420 kcal, working memory (3 significantly BFs contained 1.5 g
7 d apart 75.0 g CHO, 15.5 g P, levels with different improved accuracy available CHO/kg
(familiarization 6.4 g fat, 36 GL, memory loads) in the Stroop body mass and
session + 3 testing 48 GI) isoenergetic selective attention were matched for
sessions) task, as compared EI, protein, and fat
with a high-GI BF, contents
overall (significant
treatment effect)
and across the
morning
(significant
treatment 3
assessment time
interaction effect)
645

(Continued)
 646

TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

(No) Mean: w13 y Self-selected similar No BF 3) Flanker selective A low-GI BF The 2 BF meals
dinner on each attention task significantly provided a fixed
occasion (congruent and improved on available CHO
incongruent levels) response times in content
Sternberg paradigm
of working memory
across the morning,
as compared with
a high-GI BF
(significant
treatment 3
assessment time
interaction effect)
44% M, 56% F Overnight fast Response times on the Statistical analysis:
Stroop test were linear mixed-effects
quicker overall with models, with a
EDEFONTI ET AL

the high-GI BF than 3-factor treatment 3

with the low-GI BF assessment time 3
(significant test level interaction
treatment effect)
Healthy subjects BF time not provided When the 3-factor
interaction was not
significant, a
2-factor treatment
3 assessment time
interaction was
conducted
2 urban schools in (No) No information on
Loughborough habitual BF
School setting
(Continued)
TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

Nilsson et al, 2012 n = 40 Balanced randomized High-GI or WWB 2 constructs: Tested in 2 sessions at G-WWB significantly Statistical analysis: 2-
(30)2 crossover, 3 (124 g white-wheat 90, 135, 180, and improved selective factor repeated-
assessments flour-based bread, 225 min after BF attention (correct measures ANOVAs
(familiarization 100 GI) for working responses) at the different
session + 2 testing memory (2 separate compared with assessment times
sessions) sessions of 4 WWB BF at 120 (order of BF
subsets given in 2 min for the overall consumption 3
different orders) test (significant treatment),
and at 75, 120, 165, time 3 treatment correlation analysis
and 210 min after interaction) and,
BF for selective over the entire test
attention period (75–225
min), when only
the last half of the
test (most
demanding part)
was considered
(significant
treatment effect)
Sweden 49–71 y Standardized dinner Low-GI or G-WWB 1) Working memory with A significant No information on
on each occasion BF (179 g white- 8 different tests, of improvement in habitual BF
wheat flour-based which 4 performed on selective attention
bread plus guar each day (correct responses)
gum, 45 GI) was observed along
BREAKFAST AND COGNITIVE PERFORMANCE

the test period

(significant time
effect) and during
the second day for
G-WWB BF
(significant order of
consumption 3
treatment
interaction)
(Continued)
647
 648

TABLE 2 (Continued )

Reference Design and

(sponsorship: intervention (school Measurements of the
yes or no) Study subjects program: yes or no) Definition of breakfast Definition of outcome outcome Results Comments

(No) Mean: w63 y Overnight fast 2) Selective attention No significant

divided in 2 parts differences
(selective attention and depending on
memory) treatment or time
were observed in
reaction time for
selective attention
or in the working
memory test
30% M, 70% F BF at 0800
Healthy subjects, BMI (No)
of 20–29 kg/m2,
EDEFONTI ET AL

normal fasting
blood glucose
concentrations
Research laboratory
1
Coco Pops, All Bran, Choco Krispies, and Cornflakes manufactured by Kellogg’s. BF, breakfast; CHO, carbohydrate; EI, energy intake; G-WWB, guar gum–enriched white-wheat bread; GI, glycemic
index; GL, glycemic load; P, protein; RAG, rapidly available glucose; SAG, slowly available glucose; SES, socioeconomic status; WWB, white-wheat bread.
2
The tasks were expressed in terms of both response or reaction times and accuracy (proportion, number of correct responses, or response rates).
3
Actual mean intakes for each treatment option.
4
The tasks were expressed in terms of both raw memory retention scores and derived remembering/forgetting indexes, calculated to control for individual differences in the total items recalled in the
previous recall phase.
5
The GI values provided in the article refer to the cereal intake only, whereas the other information refers to the complete BF meal.
 BREAKFAST AND COGNITIVE PERFORMANCE 649
diseases or intellectual difficulties and dietary issues for the Eight studies were conducted among children (19, 23, 24, 26–
included subjects (16, 17), whereas the other 2 (18, 19) did not. 28) and adolescents (25, 29) recruited in schools or through
Three studies were based on an experimental design; two of public advertisement, whereas the remaining 4 included adults
them (16, 17) showed an independent group design (subjects of different mean ages [w20 (21, 22), 26 (20), and 63 (30) y,
were assigned at random to one breakfast option only) and the respectively], which were examined in research laboratories.
other provided repeated measures (subjects received each The size of the study populations was limited, and studies were
breakfast option) (19). One study (17) was designed to alter the of short duration (,1 mo). Selected studies examined pre-
child’s breakfast regimen at home. dominantly healthy, mixed-sex populations from general edu-
Absolute values of breakfast energy intake varied from 12 cation. Four studies reported information on SES: some of them
(low-calorie breakfast) (16) to 536 kcal (mean value representing recruited children from low- (24) or middle- (23) SES families
.20% of the recommended daily energy intake, for boys only) and another one recruited children from schools of different SES
(17). The ratio of breakfast to the overall daily energy intake was (19). One study adjusted the statistical models for SES (26).
tested as ,10% compared with .20% (17), or as ,20% com- Most studies were based on repeated-measures designs, and 6
pared with $20% (18), with values of w20% a priori indicating studies explicitly stated that a crossover design was used (20, 23,
adequate breakfasts. 27–30), of which 4 specified randomization of subjects to the
As for performance measures, one article assessed the effect of different treatment options. Seven of the selected studies con-
breakfast on academic performance using the Spanish version of sidered isoenergetic breakfast options and provided a control
a standard achievement test, the Scholastic Aptitude Test, and over actual breakfast consumption, either checking that all food
provided details on the overall test score and on single domains of on the plates was consumed or recording percentages of uneaten
verbal, reasoning, and calculation abilities (18). The remaining food (20, 22–25, 27, 28). Two studies were designed to assess
articles considered different aspects of cognitive performance, the effects of isoenergetic breakfast options but did not explicitly
referring to different tests. Overall, 2 studies assessed the effect of provide information on actual breakfast consumption of the
breakfast on attention and memory (16, 19) and the other 2 subjects (21, 29). The remaining studies did not provide iso-
studies (17, 18) focused on math and reasoning abilities; a third energetic breakfast options of different composition (19, 26) or
pair of studies was on verbal learning (16, 18). Two articles did not specify the energy content of the different breakfast
focused also on creativity (17) and impulsivity (16). treatments (30). By design, available breakfast options included
A higher value of breakfast energy intake significantly im- meals with a different fat and carbohydrate content (20), with
proved creativity among boys only (mean: 536 kcal compared a different GI (19, 21, 25, 30) or glycemic load (GL) (24), with
with 170 kcal, within this subgroup) (17). A lower value of w100 a different GI and a fixed available carbohydrate content (22, 23,
kcal (compared with 133 kcal) allowed a limit of the natural 29), with different combinations of GI and GL (26, 27), or with
decline of cognitive performance over time for accuracy (but not different combinations of GL and dairy products (28). [Carbo-
speed) of attention and secondary (but not speed of) memory, hydrate foods that are consumed in isoglucidic amounts produce
although this effect may have been mainly driven by the dif- different glycemic and insulinemic responses depending on the
ferent values of glycemic index (GI) provided by the treatment nature of the food (ie, ratio of amylose to amylopectin) and
options under comparison (19). Moreover, in a different article type and extent of food processing. The GI is a component-
from the United States, no significant differences in verbal referenced index able to capture the qualitative difference in the
learning/immediate memory, impulsivity, and sustained atten- carbohydrate content of foods. The GL is derived from GI as
tion were found between 2 breakfast options of w10 and 400 GI 3 carbohydrate intake; therefore, it is a measure of both the
kcal, respectively (16). Concerning percentages of daily energy quality and quantity of the carbohydrates and represents a mea-
intake, a breakfast providing .20% of daily energy intake im- sure of the dietary insulin demand (31).]
proved creativity significantly, but not math and logical and As for cognitive and academic performance, domains under
associate thinking, in a subgroup of boys (17). Similarly, in the investigation included short-term (primary), long-term (sec-
Scholastic Aptitude Test, the overall test score (measuring ver- ondary), immediate/delayed, verbal/spatial, and working mem-
bal, reasoning and calculation abilities) and logical reasoning ory (19–30); visual/auditory selective and sustained attention
scores were significantly higher for subjects consuming $20% (19, 22–24, 26–30); related reaction time, speed of processing,
of daily energy intake, after adjustment for age, sex, and school visual information processing, perceptual speed, inspection
(18). Finally, 2 of the selected studies were sponsored in whole time, and motor speed (20, 22, 26–28); verbal fluency (26, 27);
(18) or in part (17) by industry. and inductive reasoning (26, 27). Corresponding tests varied
The main characteristics of the 12 selected studies describing greatly across studies, and number and schedule of test sessions
the relation between macronutrient composition of breakfast and varied also, with studies testing subjects either before and after
cognitive and academic performance are shown in Table 2. The breakfast (19, 20, 23, 28) or after breakfast only; the number of
articles were published between 1996 and 2012, and the studies sessions per day ranged from 1 to 7 and had a maximum time
were carried out in Europe (19–22, 24, 26, 27, 29, 30)—mainly interval between breakfast and the last test performed that
in the United Kingdom (19–22, 24, 26, 27, 29)—the United ranged from 60 to 390 min.
States (23) and Australia (25, 28). A few studies were carried Memory ability was significantly sensitive, to various degrees,
out by the same groups of researchers (21, 22, 24, 26, 27); one to different breakfast compositions in 9 studies (19, 21–27, 29).
study (21) is indicated as study 1 in a more recent publication In detail, in 2 articles from the United Kingdom, a high slowly
with an extended data collection and a different research ques- available glucose or low-GI breakfast enhanced the number of
tion (22) and another study (26) was a pilot cross-sectional study words recalled of a combined score of immediate and delayed
of a more recent randomized controlled trial (27). memory in young female adults later in the morning, as compared
650 EDEFONTI ET AL

with a high rapidly available glucose (or high-GI) meal (21, 22). In emerged for the number of false alarms or the reaction times for
another article from the same research group, no immediate/de- the same task or for visual sustained attention (23). Moreover,
layed or verbal/spatial combined memory effect emerged for a study from UK showed that children who had consumed a low-
breakfast treatments with a different GL from an ANOVA analy- GL breakfast (as compared with a medium- or high-GL break-
sis, but an extra correlation analysis showed a significant im- fast) had significantly increased time spent on task in the first 10
provement effect of a lower-GL breakfast in children for min of school work and were less likely to display lapses in
immediate (but not delayed) verbal memory (24). Moreover, as attention for the ability to sustain attention when a 12-s delay
compared with a high-GI breakfast, a low-GI breakfast signifi- was tested for a second time (24). In 2 studies from the same
cantly improved short-term memory, among girls only, in the research group (26, 27), the time of completion in the Stroop
backward digit span task, but no corresponding difference was selective attention task was significantly lower in the high-GL/
found in the forward digit span task, or for spatial memory and high-GI breakfast for the latter article (27), but was not different
immediate/delayed visual perception (23). for the former one (26); the number search task for speed of
In the pair of studies from the United Kingdom investigating information processing/selective attention was better with a low-
the effect of both GI and GL, a high-GI breakfast significantly GI or a high-GL breakfast in the former article (26), but im-
improved immediate (but not delayed) word recall (26), although proved for a high-GI breakfast only in the latter article (27).
results from the immediate and delayed word recall tasks were Finally, 2 articles investigated explicitly selective attention in
nonsignificant in the more recent study (27). In the Serial Sevens terms of both the number/proportion of correct responses and
task used in both articles, working memory was significantly reaction or response times (29, 30). The article from Sweden
better in the high-GL/low-GI breakfast option, with significant GI (30) showed that a low-GI breakfast significantly improved ac-
and GL main effects and interaction (26), or in the high-GI curacy, compared with a high-GI breakfast, over the 75–225-min
breakfast option only (27). In another article from the United test period, when only the most demanding part of the test was
Kingdom, both accuracy and response times to the Sternberg test considered, or during the second day of testing. However, no
assessing working memory were improved across morning after significant difference depending on breakfast was observed in
a low-GI breakfast, although for accuracy this was only evident reaction times for selective attention (30). Similarly, the article
on the more complex test level (29). from the United Kingdom showed that accuracy in the Stroop
Concerning memory under conditions of divided attention, and Flanker tests improved across the morning after a low-GI
although no significant differences emerged between the high- breakfast, as compared with a high-GI breakfast. For the Flanker
and low-GI breakfast meals in the raw memory retention scores test, this was only evident on the more complex test levels;
for free or cued recall phases at the short or long delay, a high-GI however, for the Stroop test, this was also evident overall (29).
breakfast significantly improved the number of items re- Response times were nonsignificant for the Flanker test, but
membered, at the long delay, for an extra analysis based on the improved overall with a high-GI breakfast in the Stroop test
remembering/forgetting indexes, which took into account in- (29).
dividual differences in the total items recalled in the previous Moreover, the number of words of the word-generation task for
recall phase (25). assessing verbal fluency increased significantly after the intake of
Concerning long-term memory, in an article from Sweden and a low-GI breakfast in 1 (27) of the 2 articles assessing this do-
Denmark, the natural decline in secondary memory (as measured main, but not in the other (26).
by a composite score) of children over time was significantly Overall, data were less supportive for the effects of breakfast
reduced after the intake of a low-GI breakfast, as compared with composition on other cognitive variables, such as inductive
a high-GI breakfast (19), although the same was not true in an reasoning, speed of processing, visual information processing,
article from the United States for the long-term recall of material perceptual speed, motor speed, and reaction time (here intended
previously learned during the spatial memory, visual perception, as a direct measure and not as a proxy for a more complex
and verbal memory tasks (23). cognitive function such as attention). Finally, 4 selected studies
On the opposite side, no significant effect was detected for free were sponsored in part (22, 24, 27, 28) by industry.
word recall and all breakfast options differing in fat and car- A summary of the findings on the relation between the amount
bohydrate contents (20), for speed of memory and working of energy intake at breakfast, breakfast composition, and cog-
memory and breakfast options with a different GI (19, 30), and nitive and academic performance—grouped by broad cognitive
for short-term or working memory and different breakfast modality (row) and research question (column)—is shown in
combinations of GL and dairy products (28). Table 3. Specific domains of interest are reported in separate
Attention was influenced by different breakfast compositions rows, including also for reference the original terms provided in
in 7 articles (19, 23, 24, 26, 27, 29, 30), whereas only 2 showed the corresponding articles. Significant and nonsignificant results
nonsignificant results concerning sustained attention and dif- are reported in separate columns. In the following, we comment
ferent slowly available glucose or rapidly available glucose only on the constructs that were investigated in terms of both
breakfast meals (22), and attention switching and different amount of energy intake and composition in the selected arti-
breakfast combinations of GL and dairy products (28). In detail, cles. For the remaining constructs, we refer the reader to the
a low-GI breakfast could reduce the natural decrease in accuracy comments to Tables 1 and 2. With the exception of the combined
of sustained attention of children over time, although the same score of immediate and delayed memory (21, 22), this construct
was not true for speed of attention (19). Similarly, a low-GI was only partially affected by different amounts of energy intake
breakfast significantly improved the number of hits or misses of at breakfast or breakfast composition. Nonsignificant results
an auditory sustained attention test in 6–8-y-old children (but not were equally distributed between long-term (secondary) (23)
in 9–11 y old children), although no significant differences and working memory (19, 28, 30), whereas they were the
TABLE 3
Summary of findings on the relation between amount of energy intake at breakfast/breakfast composition and cognitive and academic performance, grouped by cognitive domain (row) and research question
(column)1
Amount of energy intake at breakfast Breakfast composition
Cognitive and
academic modality Significant effect Nonsignificant effect Significant effect Nonsignificant effect

Memory
Short-term and working memory
Verbal short-term memory [short- 424- compared with 12-kcal Better with low-GL BF under Different fat/CHO BF and no
term (primary) memory, BF (16) certain conditions (24) BF (20)
immediate verbal memory,
immediate word recall,
(immediate) free word recall]2
Better with low-GI BF under certain Different GL/GI BF (27)
conditions (23)
Better with high-GI BF, but Different GL/dairy products
nonsignificant GI and GL BF (28)
interaction (26)
Different GI BF and no BF (23)
Verbal short-term memory with Different GI BF (25)
secondary motor task3
Visuospatial short-term memory Different GL BF (24)
(immediate spatial memory)
Different GI BF and no BF (23)4
Working memory 100- compared with 133-kcal Better with high-GI BF, but Different GI BF (19)
BF (19) nonsignificant GI and GL
interaction (27)
Better with low-GI BF under Different GI BF (30)5
certain conditions (29)5
Better with low-GI, with high-GL BF, Different GL/dairy products BF (28)
and with high-GL/low-GI BF (26)
BREAKFAST AND COGNITIVE PERFORMANCE

Delayed memory
Verbal delayed memory Different GL BF (24)
(delayed word recall)
Different GL/GI BF (27)
Better with high-GI BF with Different GI BF with raw memory
remembering/forgetting indexes retention scores (25)3
under certain conditions (25)3
Different GL/GI BF (26)
Visuospatial delayed memory Different GL BF (24)
(delayed spatial memory)
Different GI BF and no BF (23)4
(Continued)
651
 652

TABLE 3 (Continued )

Amount of energy intake at breakfast Breakfast composition

Cognitive and
academic modality Significant effect Nonsignificant effect Significant effect Nonsignificant effect

Verbal memory: immediate Better with high-SAG (low-GI) BF

and delayed combined under certain conditions (21)
Better with high-SAG (low-GI) BF
under certain conditions (22)
Long-term (secondary) memory Better with 100- compared with Better with low-GI BF (19)6 Different GI BF and no BF (23)4
133-kcal BF (19)6
Speed of memory 100- compared with 133-kcal BF (19)6 Different GI BF (19)6
Attention/concentration
Sustained attention
Visual sustained attention— Better with 100- compared with 133- 424- compared with 12-kcal BF (16) Better with low-GI BF (19) Different GI BF and no BF (23)
number of correct responses, kcal BF (19)
accuracy, number of lapses in
attention, and rates of hits,
misses, and false alarms Better with low-GL BF under Different RAG or SAG (GI) BF
certain conditions (24) and no BF (22)
Visual sustained attention—time 100- compared with 133-kcal BF (19) Better with low-GL BF under Different GI BF (19)
on task, speed of sustained certain conditions (24)
attention, and reaction times to
hits and false alarms
Different GI BF and no BF (23)
Auditory sustained attention— Better with low-GI BF under
EDEFONTI ET AL

rates of hits, misses, and false certain conditions (23)

alarms
Auditory sustained attention— Different GI BF and no BF (23)
reaction times to hits and false
alarms
Selective attention
Selective attention—number of Better with high-GI BF, but
correct responses7,8 nonsignificant GI and GL
interaction (27)
Better with low-GI BF, under certain
conditions (30)
Better with low-GI BF, overall and
under certain conditions (29)
Selective attention—reaction Better with high-GL/high-GI BF (27) Different GI BF (30)
time/completion time8
Better with high-GI as with the Different GI BF as with the
Stroop test (29) Flanker test (29)
Different GL/GI BF (26)
(Continued)
TABLE 3 (Continued )

Amount of energy intake at breakfast Breakfast composition

Cognitive and
academic modality Significant effect Nonsignificant effect Significant effect Nonsignificant effect

Attention switching Different GL/dairy products BF (28)

Information processing and
psychomotor speed
Simple and choice reaction Different fat/CHO BF and no
times9 BF (20)
Different RAG or SAG (GI)
BF and no BF (22)
Speed of information Better with high-GI BF, but Different GL/dairy products BF (28)
processing, speed of processing7 nonsignificant GI and GL
interaction (27)
Better with low-GI BF and with
high-GL BF, but nonsignificant
GI and GL interaction (26)
Visual information processing Different fat/CHO BF and no BF (20)
Perceptual speed Different GL/dairy products BF (28)
Inspection time Different GL/dairy products BF (28)
Psychomotor skill (motor speed) Different fat/CHO BF and no BF (20)
Impulsivity 424- compared with 12-kcal BF (16)
Reasoning
Grammatical .20% (53610 and 43410 kcal for M
and F, respectively) compared with
,10% (17010 and 1210 kcal for M
and F, respectively) EI BF (17)
Logical Better with $20% compared with
,20% EI BF (18)11
Inductive Better with high-GL BF, but Different GL/GI BF (27)
nonsignificant GI and GL
interaction (26)
BREAKFAST AND COGNITIVE PERFORMANCE

Creativity Better with .20% compared with

,10% EI BF among M only (53610
compared with 17010) (17)
Learning 424- compared with 12-kcal (16) Different GL/dairy products BF (28)
Math .20% (53610 and 43410 kcal for M
and F, respectively) compared with
,10% (17010 and 12110 kcal for M
and F, respectively) EI BF (17)
$20% compared with ,20% EI BF
(18)11
Verbal abilities
Command of language $20% compared with ,20% EI
BF (18)11
Fluency Better with low-GI BF, but Different GL/GI BF (26)
nonsignificant GI and GL
interaction (27)
653
 1
BF, breakfast; CHO, carbohydrate; EI, energy intake; GI, glycemic index; GL, glycemic load; RAG, rapidly available glucose; SAG, slowly available glucose.
2
654
The Rey auditory-verbal learning test assessed both verbal learning and immediate verbal memory. Therefore, we reported its results in the corresponding 2 rows of the table, as indicated by Rampersaud et al (1).
3
In the article by Smith and Foster (25), verbal memory was assessed with a simultaneous secondary motor task, so under conditions of divided attention. Therefore, we presented its results in a separate
row in the verbal memory section. However, the article did not provide the corresponding results for the motor task. Moreover, the tasks were expressed in terms of both raw memory-retention scores and derived
remembering/forgetting indexes, calculated to control for individual differences in the total items recalled in the previous recall phase. Although the article did not provide any detail, the statistical analyses for
the 2 outcome measures were probably different, with the latter analysis including one time or delay effect allowing for comparisons between short- and long-delay results. Therefore, we separated the
corresponding results on verbal memory in the short-term (immediate and short-delay) and delayed sections. Because results from retention scores and remembering/forgetting indexes were inconsistent at the
long delay, we summarized them in 2 separate cells.
4
In the article by Mahoney et al (23), the so-called spatial memory and visual perception domains both belonged to the visuospatial memory category. Both of them were assessed at the level of short- and
long-term memory and then contributed to both immediate/short-term and long-term memory categories. Results for short-term memory were consistent across these domains; therefore, we included them in the
same cell. Results on long-term memory for visual perception and spatial learning were also consistent with those of a verbal memory test. Therefore, we included the information on long-term memory from the
3 tests in a single cell. Visual perception was also assessed as a delayed recall and then contributed to the visuospatial delayed memory category too.
5
The tasks were expressed in terms of both response or reaction times and accuracy (proportion or number of correct responses), but results were consistent and we therefore summarized them in one cell
only.
6
The article by Ingwersen et al (19) assessed cognitive performance, referring to the selection of tests from the Cognitive Drug Research computerized assessment system and adopted its cognitive
assessment factor scores. These are combined scores obtained from individual measures derived from simpler tests. Whenever they did not seem to us to be comparable with the simpler test measures provided
by the other articles, we put them in a separate row.
7
In the article by Micha et al (27), the number search task assessed both speed of information processing and selective attention; therefore, we reported its results in the corresponding 2 rows of the table. In
the article by Brindal et al (28), speed of processing was a composite measure based on the responses to different reaction time tasks.
8
In the article by Cooper et al (29), selective attention was assessed by using both the Stroop and the Flanker tests. When results were consistent, we reported them in 1 cell; otherwise, we filled in 2 cells
and indicated from which test the result came.
9
Although reaction time is commonly used as a dependent variable in many tests of cognitive function, some studies include direct measures of reaction time. In this case, reaction time is not a proxy
measure for a more complex function.
10
Mean intakes for each treatment option and sex.
11
The article by López-Sobaler et al (18) considered both an overall score measuring verbal, reasoning, and calculation abilities and its single domains. Here, we presented results for the single domains only.
EDEFONTI ET AL

DISCUSSION

meals that differ in GL, GI, or both (10).

classes over the entire course (19, 25, 27, 29).

provides on 2 crucial but definitely difficult issues.
breakfast on reasoning and verbal abilities (17, 18, 26, 27).

The predominance of studies on children and adolescents

heterogeneity and inconsistency in research designs, interventions,

liminary attempt to understand which information the literature

nitive and academic performance. Although nearly 100 articles
measured by the Rey auditory-verbal learning test in 2 articles
strong (19, 22, 23). Moreover, 2 breakfast options of extremely

effects, including an improved nutrient balance and distribution,

we acknowledge that the current review suffered from severe
specific conditions (19, 23, 24), but the evidence was not so

(and selected nutrients) to modulate the short-term metabolic

compared with adults (32, 33). Moreover, their higher sleep
and higher average cerebral blood flow and oxygen utilization as
vulnerable to the nutritional effects of breakfast on brain activity
parallel classes (16, 17, 23, 24) and the other studies recruited
emerged about the age of the participants. Four studies recruited
Although all the included studies were carried out in well-
although we established strict inclusion criteria and reported
These numbers prevent any definite conclusions. Moreover,
by Lloyd et al (20), and indicated an increasing interest for breakfast
although with a few exceptions. Attention was generally im-
When a significant relation was identified, this was generally

can deplete glycogen stores overnight (34). Breakfast con-

adolescents have a higher metabolic rate of glucose utilization
reflects the evidence that these categories may be particularly
detailed ancillary information on the selected studies, in general,
recent evidence on the role of the amount of breakfast energy
intake and breakfast composition on various measures of cog-
To our knowledge, the current review was the first to collect
(16, 28). Finally, data were inconclusive for the effects of
of energy intake nor different combinations of GL and dairy
proved by adequate breakfast meals. Sustained attention might

responses to fasting conditions. It may also provide long-term

demand corresponds to longer overnight fasting periods, which
adolescents; the studies included one class of the school only/
nourished subjects from developed countries, some heterogeneity
related to low glycemic response treatments (19, 21–24, 26, 29),
28) and dominated in the delayed-memory construct (23–27).

and associated cognitive and academic outcomes. Children and

(20–22, 30). The remaining studies recruited schoolchildren and
adults, with a strong difference in mean ages between them
position were definitely more recent, except for the pioneering work
were identified from the search strategy, only 15 were included in
products had a significant effect on learning, as consistently

timal amount of energy intake at breakfast were published at

the systematic review. Of these, the articles concerning the op-
ferent in terms of performance (16). Neither different amounts
different amounts of energy intake were not significantly dif-
improve consistently with low-GI or low-GL interventions under
majority in the short-term memory construct (16, 20, 23–25, 27,

sumption may therefore provide a continuous supply of energy

interest for this issue over time. The articles on breakfast com-
a constant rate from 1990 onward, showing a constant but limited

and statistical analyses and should therefore be intended as a pre-

 BREAKFAST AND COGNITIVE PERFORMANCE 655
which may positively affect cognitive processes (7). Moreover, servation period (37). In particular, there is well-established
other mechanisms, such as hunger alleviation (35, 36) and evidence that SES is a central determinant of academic perfor-
changes in neurotransmitter concentrations (7), may play a role. mance and cognitive ability (38). At the same time, there is
A few studies selected habitual breakfast eaters (18, 20, 26, consistent evidence that SES is associated with breakfast eating
27), whereas most did not. Even in the former case, the studies do habits and consumption of “healthy” items at breakfast (eg, fruit,
not generally provide details on how regular breakfast con- bread, cereals, and milk), and children from higher-SES back-
sumption is defined and do not use this information in the def- grounds were more likely to regularly eat breakfast and to eat
inition of the interventions or in the statistical analysis. healthy items (39). Similarly, limited attention was generally
Heterogeneity and little consistency between studies domi- devoted to fix and report inclusion and exclusion criteria, in-
nate the definition of the cognitive and academic performance cluding intellectual and learning disabilities, chronic diseases,
measures under comparison. Two studies reported results from overweight and obesity, and dietetic regimens. Together with an
a single standardized test assessing several constructs simul- accurate selection of the sample of interest, solutions to deal
taneously (18, 19), with one of them (18) providing an overall with this issue include preliminary checks of the distribution of
score. The remaining studies considered separate tests for the the relevant factors within the breakfast options under compar-
constructs under examination, but sometimes the same con- ison and/or adjustment in multiple regression and ANOVA
struct was measured according to different indexes (16, 23, 24, models. However, only 2 studies referred to quantitative criteria
29, 30) or tests (16, 23, 26, 27, 29, 30). Moreover, only one of for assessing intellectual difficulties and/or SES and checked
the included studies (27) carried out a pilot study to assess the that breakfast groups were fairly similar according to these as-
validity of the cognitive test on a comparable population. The pects (16, 24). Four studies adjusted the statistical models for
heterogeneity in the age of the selected subjects, together with variables other than age and sex (18, 26–28), but only one of
the different experimental settings, may be responsible, in part, them adjusted the models for a measure of SES (26).
for the heterogeneity detected in the tests used. To improve the Although most experiments have examined cognitive behavior
comparability of the studies, an accurate reclassification of the at only one time point after breakfast (6), information is now
cognitive and academic measures used is shown in Table 3. generally available on the time course of the effects of breakfast
Although most of the included studies examined breakfast on behavioral measures. In the current review, most selected
interventions and were of short duration, some heterogeneity studies were based on multiple testing sessions for each cognitive
exists in the type of interventions offered. Differences exist in test and treatment option, even allowing for testing subjects more
terms of foods included in the standard treatment options. than once after breakfast on the same day. This provides an
Considering breakfast composition, one study developed a low- indication of a higher quality of the more recent studies. How-
GI bread product that resulted in a specific course of glycemic ever, the different time intervals between breakfast and testing
response (30), whereas the others considered complex breakfast sessions, even for the same domains under investigations, still
meals including more than one food. Of the latter studies, 5 prevents a fair comparison between studies. This may reflect
examined cereal breakfast meals differing in GI content. Two a difference in the specific research questions of the articles or
studies simply compared the same amount of different types of a still limited knowledge of the mechanisms linking breakfast to
cereals and semiskimmed milk (19, 25), whereas a third study cognitive performance in the early and the later postprandial
kept the amount of carbohydrates fixed in the interventions and phases (28, 30, 40). Moreover, the short duration of the included
varied the amount of cereal (23). Similarly, 2 studies compared studies was another limitation that prevented the possibility to
biscuits and different types of cereal bars with a fixed carbo- control for different rates of emotional and neurological matu-
hydrate content (21, 22). A study from the United Kingdom just ration and longer-term feasibility and benefits.
added muesli and apple juice to cornflakes and milk to get a 2- In conclusion, there was insufficient quantity and consistency
fold difference in GL between the high- and the low-GL meals among the studies to draw firm conclusions on the relation be-
and a 1.3-fold difference between the low- and the high-GI meals tween the amount of energy intake at breakfast and breakfast
(27). To have meals differing in GL, a first study from the United composition and cognitive and academic performance. The
Kingdom combined several foods, including milk, cornflakes, working hypothesis of a better and more sustained mental per-
eggs, jam, yogurt, ham, cheese, and bread (24). Similarly, formance with breakfasts providing $20% of daily energy intake
a second study from the United Kingdom adds to milk combi- still needs to be substantiated. Despite a few inconsistencies,
nations of various foods differing in GI, to obtain meals with a fixed some evidence suggests that a lower postprandial glycemic re-
carbohydrate content (29), whereas a third study from New Zea- sponse is beneficial to cognitive performance. However, it re-
land replaced high-GI carbohydrate foods with dairy-protein foods mains unclear whether this effect is specifically due to GI or GL
(28). Other interventions consisting of bread rolls with margarine solely, or to both (the recommended approach), or to other ef-
and jam and a milkshake showed that the fat and carbohydrate fects unrelated to glycemic response. Moreover, even after
contents of the meals varied from 27% to 56% of energy from fat control one for the other, it remains unclear whether GI- or GL-
and 62% to 34% of energy from carbohydrate (20). Moreover, based breakfast meals selectively facilitate different cognitive
differences existed between studies that provided isoenergetic domains.
breakfast options (20, 23, 24, 27–29) and studies that did not. Future research should generally benefit from carefully
Only a few studies acknowledged the importance of potentially designed studies based on selected populations of interest and
relevant confounding factors such as SES (16, 18, 23, 24, 26), that adopt standardized tests of cognitive performance scheduled
anthropometric measures (16, 22, 25–27, 30), and physical ac- at different time points and that accurately control for con-
tivity levels (26), although these factors may have a substantial founding factors. For studies that assess the effect of breakfast
effect on individual performance beyond the terms of the ob- composition, we recommend the comparison of isoenergetic
656 EDEFONTI ET AL

breakfast interventions that differ in a single nutrient component, 18. López-Sobaler AM, Ortega RM, Quintas ME, Navia B, Requejo AM.
when possible, to provide more effective and sound messages of Relationship between habitual breakfast and intellectual performance
(logical reasoning) in well-nourished schoolchildren of Madrid
public health. (Spain). Eur J Clin Nutr 2003;57(suppl 1):S49–53.
19. Ingwersen J, Defeyter MA, Kennedy DO, Wesnes KA, Scholey AB. A
The authors’ responsibilities were as follows—VE: wrote the entire man-
low glycaemic index breakfast cereal preferentially prevents children’s
uscript, refined Tables 1 and 2, and prepared Table 3; VE and VR: collected
cognitive performance from declining throughout the morning. Appe-
the existing literature and selected the included articles with the supervision tite 2007;49:240–4.
of CA; VR: prepared the first draft of Tables 1 and 2; MF and AD: provided 20. Lloyd HM, Rogers PJ, Hedderley DI, Walker AF. Acute effects on
useful suggestions to summarize details on design and statistical analysis of mood and cognitive performance of breakfasts differing in fat and
the included studies; MP and GN: provided useful suggestions to summarize carbohydrate content. Appetite 1996;27:151–64.
details on the available breakfast options and contributed to the Discussion 21. Benton D, Ruffin MP, Lassel T, Nabb S, Messaoudi M, Vinoy S, Desor
section; LF and EF: provided useful suggestions to summarize details on the D, Lang V. The delivery rate of dietary carbohydrates affects cognitive
study design; and CA: conceived the systematic review. All authors read and performance in both rats and humans. Psychopharmacology (Berl)
approved the final version of the manuscript. The authors declared that they had 2003;166:86–90.
no conflicts of interest. The Italian Association for Cancer Research had no role 22. Benton D, Nabb S. Breakfasts that release glucose at different speeds
interact with previous alcohol intake to influence cognition and mood
in the design, implementation, analysis, and interpretation of the data.
before and after lunch. Behav Neurosci 2004;118:936–43.
23. Mahoney CR, Taylor HA, Kanarek RB, Samuel P. Effect of breakfast
composition on cognitive processes in elementary school children.
Physiol Behav 2005;85:635–45.
REFERENCES 24. Benton D, Maconie A, Williams C. The influence of the glycaemic
1. Rampersaud GC, Pereira MA, Girard BL, Adams J, Metzl JD. load of breakfast on the behaviour of children in school. Physiol Behav
Breakfast habits, nutritional status, body weight, and academic per- 2007;92:717–24.
formance in children and adolescents. J Am Diet Assoc 2005;105:743– 25. Smith MA, Foster JK. The impact of a high versus a low glycaemic
60, quiz 761–2. index breakfast cereal meal on verbal episodic memory in healthy
2. Benton D. The influence of dietary status on the cognitive performance adolescents. Nutr Neurosci 2008;11:219–27.
of children. Mol Nutr Food Res 2010;54:457–70. 26. Micha R, Rogers PJ, Nelson M. The glycaemic potency of breakfast
3. Wachs TD. Necessary but not sufficient. The respective roles of single and cognitive function in school children. Eur J Clin Nutr 2010;64:
and multiple influences on individual development. Washington, DC: 948–57.
American Psychological Association, 2000. 27. Micha R, Rogers PJ, Nelson M. Glycaemic index and glycaemic load
4. Pollitt E. Malnutrition and infection in the classroom. Paris, France: of breakfast predict cognitive function and mood in school children:
UNESCO, 1990. a randomised controlled trial. Br J Nutr 2011;106:1552–61.
5. Grantham-McGregor S, Baker-Henningham H. Review of the evidence 28. Brindal E, Baird D, Danthiir V, Wilson C, Bowen J, Slater A, Noakes
linking protein and energy to mental development. Public Health Nutr M. Ingesting breakfast meals of different glycaemic load does not alter
2005;8:1191–201. cognition and satiety in children. Eur J Clin Nutr 2012;66:1166–71.
6. Kanarek R. Psychological effects of snacks and altered meal frequency. 29. Cooper SB, Bandelow S, Nute ML, Morris JG, Nevill ME. Breakfast
Br J Nutr 1997;77(suppl 1):S105–18; discussion S118–20. glycaemic index and cognitive function in adolescent school children.
7. Pollitt E, Mathews R. Breakfast and cognition: an integrative summary. Br J Nutr 2012;107:1823–32.
Am J Clin Nutr 1998;67:804S–13S. 30. Nilsson A, Radeborg K, Bjorck I. Effects on cognitive performance of
8. Hoyland A, Dye L, Lawton CL. A systematic review of the effect of modulating the postprandial blood glucose profile at breakfast. Eur J
breakfast on the cognitive performance of children and adolescents. Clin Nutr 2012;66:1039–43.
Nutr Res Rev 2009;22:220–43. 31. Monro JA, Shaw M. Glycemic impact, glycemic glucose equivalents,
9. Adolphus K, Lawton CL, Dye L. The effects of breakfast on behavior glycemic index, and glycemic load: definitions, distinctions, and im-
and academic performance in children and adolescents. Front Hum plications. Am J Clin Nutr 2008;87:237S–43S.
Neurosci 2013;7:425. 32. Kennedy C, Sokoloff L. An adaptation of the nitrous oxide method to
10. Gilsenan MB, de Bruin EA, Dye L. The influence of carbohydrate on the study of the cerebral circulation in children; normal values for
cognitive performance: a critical evaluation from the perspective of cerebral blood flow and cerebral metabolic rate in childhood. J Clin
glycaemic load. Br J Nutr 2009;101:941–9. Invest 1957;36:1130–7.
11. Maffeis C, Fornari E, Surano MG, Comencini E, Corradi M, Tommasi 33. Chiron C, Raynaud C, Maziere B, Zilbovicius M, Laflamme L, Masure
M, Fasan I, Cortese S. Breakfast skipping in prepubertal obese chil- MC, Dulac O, Bourguignon M, Syrota A. Changes in regional cerebral
dren: hormonal, metabolic and cognitive consequences. Eur J Clin Nutr blood flow during brain maturation in children and adolescents. J Nucl
2012;66:314–21. Med 1992;33:696–703.
12. Sedlackova D, Kopeckova J, Papezova H, Hainer V, Kvasnickova H, 34. Thorleifsdottir B, Bjornsson JK, Benediktsdottir B, Gislason T,
Hill M, Nedvidkova J. Comparison of a high-carbohydrate and high- Kristbjarnarson H. Sleep and sleep habits from childhood to young
protein breakfast effect on plasma ghrelin, obestatin, NPY and PYY adulthood over a 10-year period. J Psychosom Res 2002;53:529–37.
levels in women with anorexia and bulimia nervosa. Nutr Metab 35. Kleinman RE, Murphy JM, Little M, Pagano M, Wehler CA, Regal K,
(Lond) 2012;9:52. Jellinek MS. Hunger in children in the United States: potential be-
13. Messier C. Glucose improvement of memory: a review. Eur J Phar- havioral and emotional correlates. Pediatrics 1998;101:E3.
macol 2004;490:33–57. 36. Alaimo K, Olson CM, Frongillo EA Jr. Food insufficiency and
14. Ells LJ, Hillier FC, Shucksmith J, Crawley H, Harbige L, Shield J, American school-aged children’s cognitive, academic, and psychoso-
Wiggins A, Summerbell CD. A systematic review of the effect of di- cial development. Pediatrics 2001;108:44–53.
etary exposure that could be achieved through normal dietary intake on 37. Timlin MT, Pereira MA, Story M, Neumark-Sztainer D. Breakfast
learning and performance of school-aged children of relevance to UK eating and weight change in a 5-year prospective analysis of adolescents:
schools. Br J Nutr 2008;100:927–36. Project EAT (Eating Among Teens). Pediatrics 2008;121:e638–45.
15. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items 38. McCulloch A, Joshi HE. Neighbourhood and family influences on the
for systematic reviews and meta-analyses: the PRISMA statement. cognitive ability of children in the British National Child Development
BMJ 2009;339:b2535. Study. Soc Sci Med 2001;53:579–91.
16. Cromer BA, Tarnowski KJ, Stein AM, Harton P, Thornton DJ. The 39. Moore GF, Tapper K, Murphy S, Lynch R, Raisanen L, Pimm C,
school breakfast program and cognition in adolescents. J Dev Behav Moore L. Associations between deprivation, attitudes towards eating
Pediatr 1990;11:295–300. breakfast and breakfast eating behaviours in 9-11-year-olds. Public
17. Wyon DP, Abrahamsson L, Jartelius M, Fletcher RJ. An experimental study Health Nutr 2007;10:582–9.
of the effects of energy intake at breakfast on the test performance of 10- 40. le Coutre J, Schmitt JA. Food ingredients and cognitive performance.
year-old children in school. Int J Food Sci Nutr 1997;48:5–12. Curr Opin Clin Nutr Metab Care 2008;11:706–10.