Glycaemic index and glycaemic load of breakfast predict cognitive function and mood in school children: a randomised controlled trial

The researchers tested four breakfast meals that varied in two dimensions: glycaemic index (GI — how fast carbohydrates raise blood sugar) and glycaemic load (GL — total carbohydrate amount multiplied by GI). The four meals were:

**Low-GI, high-GL** (e.g., porridge/oatmeal with milk)

**High-GI, high-GL** (e.g., cornflakes with milk)

**Low-GI, low-GL** (e.g., bran flakes with milk)

**High-GI, low-GL** (e.g., white bread with margarine and jam)

The primary outcomes were cognitive function (memory, attention, processing speed) and mood (alertness, happiness, nervousness, hunger, thirst). Secondary outcomes included blood glucose levels and salivary cortisol (a stress hormone). The key question was whether the glycaemic potency of breakfast — not just its carbohydrate content — predicts how well children think and feel in the mid-morning (95–140 minutes after eating).

**74 school children** (boys and girls) aged 11–14 years

Recruited from five schools in London, UK

All were in good health, free from learning disabilities, and not underweight or obese (BMI-for-age z-scores between -2 SD and +2 SD)

Excluded: children with anaemia, diabetes, glucose intolerance, colour blindness, mood disorders, or allergies/intolerances to breakfast foods

Children who never ate breakfast were also excluded

Participants were matched in pairs on: same school year, same sex, height (±3–5 cm), age (±6 months), BMI (±1 centile), and same school

**Cognitive function:** Seven tests administered in fixed order:

1. Word generation task (verbal fluency)

2. Immediate word recall (short-term verbal memory)

3. Stroop task (attention and interference control)

4. Matrices (non-verbal reasoning)

5. Number search task (vigilance and processing speed)

6. Serial sevens (working memory and arithmetic)

7. Delayed word recall (long-term verbal memory)

**Mood:** Self-rating questionnaire with 22 items (e.g., alert, happy, nervous, sluggish, hungry, thirsty) on a 0–4 scale ("not at all" to "extremely"), adapted from the Profile of Mood States (POMS-BI) and Activation-Deactivation Adjective Checklist

**Blood glucose:** Capillary finger-prick samples measured with Accu-Chek Aviva BG meter (CV <5%)

**Salivary cortisol:** Saliva collected via Salivette cotton swabs, analysed by ELISA (intra-assay CV <3%, inter-assay CV <7%)

**Anthropometrics:** Height (portable stadiometer) and weight (portable scale)

**Task demand:** Self-reported ratings of how difficult, effortful, and tiring the tests were

**Design:** Randomised, matched-pairs, cross-over trial with a 2×2 factorial design. Children were matched in pairs, then one member of each pair was randomly allocated to a high-GL group and the other to a low-GL group. Within each GL group, each child received both a high-GI and a low-GI breakfast on separate visits (2 weeks apart), in random order. This means each child ate two different breakfasts total (one per visit), but the GL level was fixed for the pair.

**Randomisation:** Pairs were matched on key confounders (age, sex, height, BMI, school, school year), then randomly allocated to GL group. Within each GL group, the order of GI meals was randomised. This matching reduces variability from individual differences.

**Blinding:** The paper does not explicitly state that participants or researchers were blinded to the breakfast condition. This is a significant weakness — children could see what they were eating, and researchers administering tests may have known the condition. Blinding is difficult with food studies but partial blinding (e.g., using similar-looking meals) was not attempted.

**Washout period:** 2 weeks between visits. This is adequate for a single-meal intervention since glycaemic effects last only hours, but it does not account for potential menstrual cycle effects in girls (noted by authors).

**Duration:** Each testing session lasted approximately 56 minutes (SE 7 minutes), starting 90 minutes after breakfast. Cognitive tests were administered 95–140 minutes post-breakfast. This is a single-day, acute effect study — not a chronic dietary intervention.

**Statistical approach:** Repeated-measures ANOVA, with breakfast type (GI and GL) as within-subject factors and GL group as a between-subject factor. Baseline mood, glucose, and cortisol were included as covariates. The analysis tested main effects of GI and GL, and their interaction.

**What this design can prove:** Because it's a randomised cross-over trial, it can establish causal effects of specific breakfast meals on cognitive function and mood in the short term (2–3 hours post-meal). Matching reduces confounding by individual differences. Measuring glucose and cortisol provides mechanistic insight.

**What this design cannot prove:** It cannot tell us about long-term effects of dietary patterns (e.g., eating low-GI breakfasts for weeks or months). It cannot tell us about effects in other age groups (adults, younger children). It cannot tell us whether the effects persist beyond 2.5 hours post-meal. The small sample size (74 children, but only 32–37 per GL group for some analyses) limits statistical power for detecting small effects. The lack of blinding is a major threat to validity — expectation effects could influence both mood ratings and cognitive performance.

**Major methodological weaknesses:**

No blinding of participants or researchers

Only one time point for cognitive testing (95–140 minutes post-meal)

No measure of habitual diet or breakfast habits beyond screening

The GL groups were between-subjects (different children), while GI was within-subjects (same children) — this confounds the GL comparison with individual differences

Mood was self-reported by 11–14 year olds, which may have limited reliability

The cognitive tests were administered in the same order every time, so order effects (fatigue, practice) are confounded with test type

**Primary outcomes (cognitive function):**

**Declarative-verbal memory (delayed word recall):** Low-GI meals predicted better performance compared to high-GI meals (P=0.03). This effect was independent of GL group.

**Vigilance (number search task):** High-GI meals predicted better performance compared to low-GI meals (P<0.03). This effect was also independent of GL group.

**Other cognitive tests (word generation, immediate word recall, Stroop, matrices, serial sevens):** No significant differences between breakfast types.

**Secondary outcomes (mood):**

**Low-GI meals predicted:** Feeling more alert (P<0.05), more happy (P<0.05), less nervous (P<0.05), and less thirsty (P<0.05)

**High-GL meals predicted:** Feeling more confident (P<0.05), less sluggish (P<0.05), less hungry (P<0.05), and less thirsty (P<0.05)

**No significant GI × GL interactions** for mood outcomes — the effects of GI and GL were independent

**Physiological outcomes:**

**Blood glucose at 90 minutes:** High-GL meals increased glucose levels compared to low-GL meals (P<0.001); high-GI meals increased glucose compared to low-GI meals (P=0.05)

**Salivary cortisol:** High-GI meals increased cortisol levels compared to low-GI meals (P<0.01)

Baseline glucose and cortisol levels predicted mood and cognitive performance, and adjusting for them strengthened some findings

**Task demand:** No significant differences between breakfast conditions in how difficult, effortful, or tiring children found the tests.

The paper reports P-values but does not provide effect sizes (e.g., Cohen's d, mean differences with confidence intervals) for the significant cognitive findings. This is a limitation. However, we can translate the findings qualitatively:

**Memory improvement with low-GI:** The effect on delayed word recall was statistically significant (P=0.03), meaning the probability that this difference occurred by chance is about 3%. The actual number of words recalled is not reported in the abstract or available text, so we cannot say "children remembered X more words." However, the effect was robust enough to survive adjustment for baseline mood, glucose, and cortisol.

**Vigilance improvement with high-GI:** The effect on number search task was significant (P<0.03). Again, exact mean differences are not provided in the available text.

**Mood effects:** The mood differences were statistically significant at P<0.05, but the magnitude on a 0–4 scale is not reported. Given the large sample for a cross-over trial (74 children), these are likely small-to-moderate effects.

**Glucose increase:** High-GL meals raised blood glucose "significantly" (P<0.001) — this is expected by design and confirms the meals worked as intended.

**Cortisol increase:** High-GI meals raised cortisol (P<0.01). Cortisol is a stress hormone that also enhances glucose availability. The magnitude is not reported, but this is a novel mechanistic finding.

In plain English: If you eat a low-GI breakfast (like porridge), you might remember slightly more of what you learned 2 hours later, compared to eating a high-GI breakfast (like cornflakes). But you might be slightly less vigilant — slower to spot targets in a repetitive task. The mood effects are subtle: low-GI makes you feel more alert and happy but also more nervous; high-GL makes you feel more confident and less sluggish.

**Acknowledged by authors:**

The study only tested acute effects (single breakfast, single morning)

The sample was limited to healthy adolescents aged 11–14, so results may not generalise to younger children, adults, or clinical populations

Menstrual cycle phase was not controlled for in girls, which could affect cortisol and cognitive performance

The cognitive tests were administered in the same order each time, so practice effects and fatigue are confounded with test type

Only one time point for cognitive testing (95–140 minutes post-meal), so the time course of effects is unknown

**Critical reader observations:**

**No blinding:** This is the most serious limitation. Children knew what they ate, and researchers likely knew too. Expectation effects could explain mood differences and even cognitive performance (if children felt differently about the meals).

**No effect sizes reported:** The paper reports P-values but not mean differences, standard deviations, or confidence intervals for the primary outcomes. This makes it impossible to assess practical significance.

**Small sample for subgroup analyses:** With 74 children split into two GL groups (roughly 37 each), and then each child receiving two GI meals, the actual comparisons for GI effects have reasonable power, but the GL comparison is between different children (not within-subject), reducing power.

**GL confounded with individual differences:** Because GL was a between-subjects factor (different children in high vs. low GL groups), any GL effect could be due to pre-existing differences between the groups despite matching.

**Only one breakfast composition per condition:** Each GI × GL combination was a single meal (not multiple options), so the results may be specific to those particular foods, not to GI/GL in general.

**Industry funding:** The School Food Trust (a UK government body) provided funding to schools, but the authors declare no conflicts of interest. However, the study was designed to inform government policy, which could introduce subtle bias in interpretation.

**Self-reported mood in adolescents:** 11–14 year olds may have difficulty accurately reporting internal states, especially on abstract scales.

**No measure of habitual diet:** Children's usual breakfast habits were not quantified, so we don't know if the effects differ for regular breakfast-eaters vs. occasional skippers.

For someone running their own n=1 experiment:

### What to test

Compare two breakfasts that differ in GI but are matched for total carbohydrate (GL):

**Low-GI breakfast:** Rolled oats/porridge made with milk, topped with berries and nuts (GI ~40–50)

**High-GI breakfast:** Cornflakes or puffed rice cereal with milk (GI ~80–90)

Alternatively, compare high-GL vs. low-GL:

**High-GL breakfast:** Large bowl of cereal with milk + banana + glass of orange juice (total carb ~80–100g)

**Low-GL breakfast:** Same cereal but smaller portion + eggs or Greek yoghurt (total carb ~30–40g)

### Minimum meaningful duration

**Single morning:** The effects in this study appeared 90–140 minutes after eating. Test for at least 3 mornings per condition (to account for day-to-day variability).

**For chronic effects:** If you want to see if adapting to a low-GI breakfast changes your baseline cognition, run each condition for 1–2 weeks.

### What to measure

**Cognitive performance (primary):**

- Verbal memory: Learn a list of 15–20 words, then recall them 30 minutes later (delayed recall)

- Vigilance/attention: Use a reaction time test or continuous performance task (e.g., press a button when you see a target number)

- Processing speed: Simple reaction time or digit-symbol coding test

**Mood (secondary):**

- Rate alertness, happiness, nervousness, sluggishness, hunger, and thirst on a 0–10 scale every 30 minutes for 3 hours post-breakfast

**Physiological (optional but informative):**

- Fingertip blood glucose at baseline, 30, 60, 90, and 120 minutes after eating (using a home glucose meter)

- Subjective energy levels (1–10 scale)

### Key confounds to control for

**Time of waking:** Wake at the same time each test day

**Previous night's sleep:** Aim for same duration and quality (use a sleep diary)

**Dinner the night before:** Eat the same meal at the same time before each test day

**Caffeine and alcohol:** Avoid for 12 hours before testing

**Physical activity:** Avoid strenuous exercise the morning of testing

**Menstrual cycle (if applicable):** Test during the same phase (e.g., follicular phase, days 5–12)

**Hydration:** Drink the same amount of water with each breakfast

**Test timing:** Start cognitive tests exactly 90 minutes after finishing breakfast

**Practice effects:** Do the cognitive tests at least twice before starting the experiment to stabilise performance

**Expectation:** If possible, have someone else prepare the breakfasts and label them "A" and "B" so you don't know which is which

### What a positive result would look like

**Low-GI improves memory:** You recall 2–4 more words on the delayed recall test after the low-GI breakfast compared to the high-GI breakfast, consistently across 3+ test days

**High-GI improves vigilance:** Your reaction time is 50–100 ms faster on the vigilance task after the high-GI breakfast

**Mood effects:** You rate yourself 1–2 points higher on alertness and happiness after low-GI, and 1–2 points lower on sluggishness and hunger after high-GL

**Glucose pattern:** Low-GI breakfast produces a smaller glucose spike (rise of <30 mg/dL) and a slower decline compared to high-GI (rise of >50 mg/dL followed by a rapid drop)

**Practical significance:** The cognitive effects should be noticeable in daily life