3. Results
Written informed parental consent and adolescent assent was obtained from 101 volunteers. However, 18 participants failed to complete main trials after informed consent and screening, primarily for reasons of time, thus leaving 83 individuals who were randomized to the four groups. One participant was excluded from the data analysis because they failed to complete several self-report measures (feeling, motivation) which were included as between-individual (Level 2) variables in the data analysis; HLM software does not allow missing values at higher levels. Thus, 82 participants (64 female) who had no missing values at Level 2 were included. Some descriptive characteristics of these individuals are provided in Table 1.
3.2. Exercise
Heart rate during the 8 min of treadmill running differed between the F-E and B-E groups (t = −2.239, p = 0.033) and was 150 (17) bpm and 161 (11) bpm, respectively. However, RPE at the end of the running was similar (t = −0.521, p = 0.605) at 13.2 (1.5) and 13.4 (1.5) (“somewhat hard”), respectively. During coordinative exercise, the heart rate for F-E and B-E groups again differed (t = −2.661, p = 0.012) and were 125 (17) bpm and 139 (13) bpm, respectively, whilst RPE at the end of these coordinative exercises was similar (t = −0.672, p = 0.506), rated as 10.9 (2.2) and 11.4 (2.1) (approximately “light”), respectively.
3.3. ICCs
ICCs of the outcome variables ranged from 15.1% (blood glucose) to 75.9% (mental effort), showing that variance of the outcome variables occurred at the between-individual level. When the ICC is greater than 10% of the total variance in the outcome variable, the multilevel structure of data should be considered [40]. These high ICCs supported the rationale for using linear mixed modeling as an appropriate form of statistical analysis for the present data.
3.4. Blood Glucose
Fingertip blood glucose was obtained from 57 participants (F-S: n = 16; B-S: n = 16; F-E: n = 13; B-E: n = 12). The remaining 25 individuals expressed discomfort at the time of the first sample procedure and so no samples were taken. In the fasted state, mean blood glucose concentrations ranged between 4.6 and 4.8 mmol∙L−1 in all groups and the concentrations in the F-S, B-S and F-E groups were similar to the B-E group (all p > 0.05). Blood glucose increased in the B-E group and the B-S group over the morning, and the change in the B-S group (β = −0.14, p = 0.558) was similar to B-E (Figure 2). Compared with the B-E group, blood glucose fell over the morning in the F-S and F-E groups (F-S: β11 = −1.45, p < 0.001; F-E: β13 = −1.28, p < 0.001). Mean blood glucose was ≥ 1.19 mmol∙L−1 greater in both breakfast groups at the end of the morning than the two fasted groups.
3.5. Academic Tests
Mathematics test scores in F-S and B-S groups were similar to the B-E group for the baseline measure at the start of the morning (both p > 0.05) but were higher in the F-E than B-E group (p = 0.044). Mean test score increased in all groups over the course of the morning (Figure 3a). The increase was greater in the B-E than F-S group (β11 = −3.30, p = 0.014) but B-E did not differ from the B-S (β12 = −1.68, p = 0.253) or F-E (β13 = −2.81, p = 0.078) groups. Computational time (mathematics speed) for B-E was similar to the other three groups at the start of the morning (all p > 0.05). Computational time improved over the morning in all groups; in B-E to a greater extent than F-S (β11 = −0.070, p = 0.004) and F-E (β13 = −0.057, p = 0.041) but not B-S (β12 = −0.042, p = 0.101) (Figure 3b). Percentage changes in both mathematics test score (F-S: 7.2%; B-S: 11.0%; F-E: 10.4%; B-E: 15.2%; Figure 4a) and computational time (F-S: 7.4%; B-S: 11.1%; F-E: 11.0%; B-E: 16.7%; Figure 4b) over the morning are shown in Figure 4.
For oral word fluency, there were no differences in the total number of valid words retrieved or speed (time taken) of word retrieval (data not shown) in the F-S, B-S and F-E groups at the start of the morning (all p > 0.05) or the change over the morning (all p > 0.05) compared with the B-E group.
3.6. Cognitive Tests
Outcomes from the Stroop Color-Word Test are in Table 2. There were no differences in reaction time or response accuracy between the B-E and other groups at the start of the morning for congruent, incongruent or control trials. However, the improvement in reaction time for congruent, incongruent and control trials in the B-E group from the start to middle of the morning was greater than that of the F-S group (congruent: β11 = 115.67, p = 0.024; incongruent: β11 = 158.75, p = 0.046; control: β11 = 104.20, p = 0.036) but not different from F-E (congruent: β13 = 24.39, p = 0.664; incongruent: β13 = −69.52, p = 0.473; control: β13 = 73.00, p = 0.220) or B-S (congruent: β12 = 111.81, p = 0.072; incongruent: β12 = 69.10, p = 0.404; control: β12 = 111.30, p = 0.104) groups. The improvement in reaction time in B-E did not result in improved response accuracy from the start to middle of the morning compared with F-S (congruent: β11 = −2.08, p = 0.284; incongruent: β11 = −0.11, p = 0.976; control: β11 = 2.01, p = 0.367), B-S (congruent: β12 = −1.92, p = 0.414; incongruent: β12 = 3.89, p = 0.257; control: β12 = −0.80, p = 0.684) or F-E (congruent: β13 = −2.58, p = 0.201; incongruent: β13 = 0.33, p = 0.931; control: β13 = 0.20, p = 0.912) (Table 2). From the middle to end of the morning the improvement in reaction time was greater for the F-S (congruent: β21 = −48.01, p = 0.033) and B-S (congruent: β22 = −55.54, p = 0.029) than B-E group for the congruent trial only (incongruent and control trials all p > 0.05). There was no difference between F-E and B-E in reaction time for any of the trials over the same time period (all p > 0.05). There were also no differences in response accuracy among groups for any of the trials between the middle and end of the morning (all p > 0.05). Interference scores (i.e., incongruent − congruent, control, or (congruent + control)/2) were also examined for reaction time and response accuracy. However, F-S, B-S and F-E groups’ interference scores were not significantly different compared with the B-E group over the morning (all p > 0.05).
Outcomes from the Digit-Span and Go/No-Go Task are presented in Table 3. For the Digit-Span Test, no differences were present between F-S, B-S and F-E with the B-E group at the start of the morning (all p > 0.05) and the maximum number of digits memorized in the forward and backwards direction was similar with B-E in all groups across the morning (all p > 0.05). In the Go/No-Go Task no differences existed in the overall mean for correct reaction times or the overall error rate between B-E and the other groups at the start of the morning (all p > 0.05). Reaction time improved to a greater extent in B-E from the start to mid-morning compared with F-E (β13 = 33.99, p = 0.019) but this finding for F-E was reversed from the middle to end of the morning (β23 = −15.29, p = 0.021). There was no difference in reaction time improvements between B-E and the other two groups across the morning (all p > 0.05). There were no differences in Go/No-Go Task error rates among groups over the morning (all p > 0.05).
3.7. Hunger, Satiety, Fullness and Appetite
The two fasting groups (F-S, F-E) reported greater measures of hunger, and less satiety and fullness at the start of the morning compared with B-E (all p < 0.05), and for the F-S group a greater appetite (p < 0.05) as well (Appendix A). There was no difference in these measures between B-E and B-S (all p > 0.05). Hunger and appetite increased to a greater extent in the two fasted groups over the morning compared with B-E (all p < 0.05) but with no difference between B-E and B-S (all p > 0.05).
3.8. Motivation, Arousal, Feeling and Mental Effort
Motivation and arousal were similar at the start of the morning and across the morning in B-E compared with the other three groups (Appendix B). Mental effort in B-E was also similar at the start of the morning compared with the other three groups. Mental effort perceived by the B-E group significantly dropped from the start to middle morning compared with the F-E group (β13 = 16.63, p = 0.042) but the change was thereafter similar until the end of morning among all groups (all p > 0.05). Whilst feeling did not differ in B-E with the other groups at the start of the morning, feeling (good/pleasure) improved in B-E from the start to mid-morning compared with the other three groups (B-S: β12 = −0.59, p = 0.040; F-S: β11 = −0.59, p = 0.038; F-E: β13 = −0.64, p = 0.036) but were again similar until morning end (all p > 0.05).
CC BY 4.0. "Breakfast and Exercise Improve Academic and Cognitive Performance in Adolescents" by Kawabata M, Lee K, Choo HC et al., Nutrients (2021). https://doi.org/10.3390/nu13041278. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Peer-reviewed and open access; full text from PubMed Central (PMC8068805). Headings and layout adapted for reading; figures and tables are in the original.