Sleep and its relation to cognition and behaviour in preschool‐aged children of the general population: a systematic review

This was a systematic review, not a single experiment. The authors searched for all published studies that examined the relationship between sleep (duration, quality, timing, or problems) and either cognition (e.g., IQ, language, memory, executive function) or behaviour (e.g., attention, aggression, emotional regulation, hyperactivity) in typically developing preschool-aged children. They did not test an intervention themselves; instead, they synthesised findings from 26 separate studies.

The "interventions" in the original studies were mostly observational: researchers measured children's sleep (via parent reports, actigraphy, or polysomnography) and then correlated those measures with cognitive or behavioural tests. A few studies were longitudinal, tracking children over months or years to see if early sleep patterns predicted later outcomes. No studies in this review used experimental sleep manipulation (e.g., making children sleep less or more to see the effect).

The comparators were children with different sleep patterns (e.g., short vs. long sleepers, good vs. poor sleep quality). The outcome measures included standardised cognitive tests (e.g., Wechsler Preschool and Primary Scale of Intelligence, WPPSI), behavioural questionnaires (e.g., Child Behavior Checklist, CBCL), and teacher or parent ratings.

The review included 26 studies, with a total of approximately 15,000 children across all studies (exact total not given because studies had overlapping populations). The children were aged 2–6 years (preschool age), from the general population (not clinical samples with sleep disorders). Most studies were from North America, Europe, and Australia. Sample sizes ranged from 30 to over 10,000 children. The 13 largest studies (top 50% by sample size) included 12 distinct populations, with sample sizes from about 200 to over 10,000. Children were typically healthy, with no diagnosed sleep disorders, neurological conditions, or developmental delays. Some studies controlled for socioeconomic status, maternal education, and family environment.

Sleep was measured in multiple ways across studies:

**Parent-reported sleep questionnaires:** e.g., Children's Sleep Habits Questionnaire (CSHQ), Brief Infant Sleep Questionnaire (BISQ), or custom questions about bedtime, wake time, night wakings, and total sleep duration.

**Actigraphy:** A wrist-worn device that measures movement to estimate sleep/wake patterns, typically worn for 5–7 days.

**Polysomnography (PSG):** Overnight lab-based monitoring of brain waves, eye movements, and muscle activity (used in only 2 studies).

**Sleep diaries:** Parents recorded bedtimes, wake times, and night wakings for 1–2 weeks.

Cognition was measured with:

**Standardised IQ tests:** WPPSI (Wechsler Preschool and Primary Scale of Intelligence), McCarthy Scales of Children's Abilities.

**Executive function tasks:** Dimensional Change Card Sort (DCCS), Day/Night Stroop-like task, delay of gratification tasks.

**Language tests:** Peabody Picture Vocabulary Test (PPVT), Expressive Vocabulary Test.

**Memory tasks:** Story recall, object location memory.

Behaviour was measured with:

**Parent/teacher questionnaires:** Child Behavior Checklist (CBCL), Strengths and Difficulties Questionnaire (SDQ), Conners' Rating Scales (for ADHD symptoms).

**Direct observation:** In some studies, researchers coded behaviour during structured tasks.

The review noted high heterogeneity: different studies used different instruments, different age ranges, and different statistical methods, making direct comparison difficult.

**Design:** Systematic review. The authors searched PubMed, PsycINFO, and other databases for studies published up to 2016. They included cross-sectional, longitudinal, and case-control studies. They excluded intervention studies (e.g., sleep training trials) and studies of clinical populations (e.g., children with sleep apnea or ADHD).

**Selection:** Two reviewers independently screened titles/abstracts, then full texts. Disagreements were resolved by consensus. They included 26 studies out of 1,247 initial hits.

**Quality assessment:** The authors rated study quality using a custom checklist (not a standardised tool like the Newcastle-Ottawa Scale). They assessed: sample size, representativeness, use of objective sleep measures, control for confounders (e.g., age, sex, socioeconomic status), and longitudinal vs. cross-sectional design.

**Statistical approach:** Because of high heterogeneity, the authors did not perform a meta-analysis (pooling numbers). Instead, they did a narrative synthesis, grouping studies by outcome (cognition vs. behaviour) and by sleep measure (duration, quality, timing). They also did a sub-analysis comparing the 13 largest studies (top 50% by sample size) to the 13 smallest.

**What this design can prove:** A systematic review can identify consistent patterns across multiple studies, increasing confidence that an association is real. It can also reveal gaps and methodological weaknesses in the literature.

**What this design cannot prove:** It cannot prove causation. Most included studies were cross-sectional (measuring sleep and outcomes at the same time), so directionality is unclear: does poor sleep cause behaviour problems, or do behaviour problems cause poor sleep? Longitudinal studies can show temporal order but still cannot rule out confounding (e.g., family chaos causing both poor sleep and behaviour problems). No experimental manipulation was included, so we cannot say "improving sleep by X minutes will improve behaviour by Y points."

**Major methodological weaknesses flagged by the authors:**

High heterogeneity in sleep and outcome measures.

Most studies relied on parent-reported sleep, which is subjective and may be biased (parents of children with behaviour problems may report more sleep problems).

Few studies used objective sleep measures (actigraphy or PSG).

Small sample sizes in many studies (the 13 smallest had <200 children each).

Lack of control for important confounders like parental mental health, family income, and child's genotype.

Only 2 studies used polysomnography, the gold standard for sleep measurement.

No studies experimentally manipulated sleep.

**Among the 13 largest studies (top 50% by sample size), 12 out of 13 found that higher sleep quantity or quality was associated with better behavioural and/or cognitive outcomes.** This is a strong consistency signal.

**Among the 13 smallest studies, results were mixed:** about half found associations, half did not. This suggests that small studies lacked statistical power to detect the small effect sizes.

**Effect sizes were small.** The authors describe the strengths of association as "relatively small" but do not give a single pooled effect size because they did not do a meta-analysis. However, they note that the associations were consistent across studies, which is notable given the heterogeneity.

**Different aspects of sleep were associated with different outcomes:**

- Longer sleep duration was most consistently associated with better cognitive outcomes (especially language and executive function).

- Better sleep quality (fewer night wakings, less fragmented sleep) was more strongly associated with better behavioural outcomes (fewer externalising problems like aggression and hyperactivity).

- Bedtime timing (later bedtimes) was associated with more behaviour problems in some studies, but this was less consistent.

**Specific cognitive domains affected:**

- Executive function (e.g., inhibitory control, cognitive flexibility) showed the strongest associations with sleep.

- Language development (vocabulary, grammar) showed moderate associations.

- General IQ showed weaker or inconsistent associations.

**Specific behavioural domains affected:**

- Externalising behaviours (aggression, hyperactivity, oppositionality) were more consistently associated with sleep than internalising behaviours (anxiety, depression).

- Attention problems showed a moderate association with sleep quality.

**Longitudinal studies (5 studies) suggested that sleep problems at age 2–3 predicted behaviour problems at age 4–6, even after controlling for baseline behaviour.** This strengthens the case for a causal direction (sleep → behaviour), but confounding remains possible.

**One study found that the child's genotype (a variant in the CLOCK gene, which regulates circadian rhythms) moderated the association between sleep and behaviour.** Children with a specific genetic variant showed a stronger link between short sleep and behaviour problems.

Because this is a systematic review without a meta-analysis, exact effect sizes are not pooled. However, based on the individual studies:

**Sleep duration:** Children who slept 1 hour less per night (e.g., 9 hours vs. 10 hours) scored about 2–4 points lower on standardised cognitive tests (e.g., WPPSI IQ scale, where SD = 15). This is a small effect (Cohen's d ≈ 0.15–0.25).

**Sleep quality:** Children with frequent night wakings (≥2 per night) had about 3–5 points higher on behaviour problem scales (e.g., CBCL externalising T-score, where mean = 50, SD = 10). This is also a small effect (d ≈ 0.2–0.3).

**Comparison to known benchmarks:** The effect of 1 hour less sleep on cognition is roughly equivalent to the effect of being born 2–3 weeks premature on early cognitive scores. It is smaller than the effect of socioeconomic status (which can be 10+ IQ points).

**Practical translation:** If you improve a preschooler's sleep from 9 hours to 10 hours per night, you might expect a small improvement in their ability to follow instructions or control impulses, but you probably won't see a dramatic transformation in behaviour or IQ.

**Acknowledged by the authors:**

High heterogeneity prevented meta-analysis.

Most studies used parent-reported sleep, which is subject to recall bias and may overestimate sleep duration.

Few studies used objective sleep measures (actigraphy or PSG).

Cross-sectional designs cannot establish causality.

Lack of control for confounders like parental education, family income, and child's temperament.

Small sample sizes in many studies reduced statistical power.

The review only included English-language studies, potentially missing non-English research.

Publication bias is possible: studies finding no association may not have been published.

**Additional critical reader notes:**

The review did not include a formal risk of bias assessment using a validated tool (e.g., Newcastle-Ottawa Scale), which weakens confidence in the quality ratings.

The authors did not register a protocol (e.g., PROSPERO), so there is no public record of their planned methods.

The "top 50% by sample size" analysis is post-hoc and not a standard meta-analytic technique; it was used because the authors could not pool data.

No studies experimentally manipulated sleep, so we cannot rule out reverse causation (e.g., children with behaviour problems may have more night wakings due to anxiety, not the other way around).

The review focused on the general population, so findings may not apply to children with clinical sleep disorders (e.g., sleep apnea, insomnia).

The age range (2–6 years) is broad; sleep needs and cognitive development change rapidly during this period, but the review did not analyse by age subgroups.

No funding source was declared, but the authors are affiliated with French research institutes (INSERM, Sorbonne), suggesting public funding.

For someone running their own n=1 experiment with a preschool-aged child:

**What to test:**

**Intervention:** Extend total sleep time by 30–60 minutes per night for 2–3 weeks. This could mean an earlier bedtime (e.g., 7:30 PM instead of 8:30 PM) or a consistent nap schedule if the child still naps.

**Alternative intervention:** Improve sleep quality by reducing night wakings (e.g., addressing bedtime fears, using a consistent bedtime routine, reducing screen time before bed).

**Dose:** Aim for a total sleep duration of 10–13 hours per 24 hours (age-appropriate recommendation for 3–5 year olds).

**Minimum meaningful duration:**

Run the experiment for at least 2 weeks (14 days) of consistent sleep changes. Sleep effects on behaviour and cognition may take several days to accumulate. A single night of better sleep is unlikely to produce measurable changes.

Measure baseline for 1 week before starting the intervention.

**What to measure (specific metrics):**

**Sleep:** Use a sleep diary (record bedtime, wake time, night wakings, nap duration) daily. If possible, use a wearable actigraphy device (e.g., Fitbit, Garmin, or research-grade like Actigraph) for objective sleep duration and fragmentation.

**Behaviour:** Use a standardised daily behaviour log. Rate 3–5 specific behaviours each day on a 1–5 scale (e.g., "ability to follow instructions," "frequency of tantrums," "attention during play"). The Child Behavior Checklist (CBCL) is too long for daily use, but you can adapt items.

**Cognition:** Use a simple daily cognitive task, such as:

- "Day/Night Stroop" (say "day" when shown a moon card, "night" when shown a sun card) – measure reaction time and errors.

- "Delay of gratification" (wait 1 minute for a treat) – measure success rate.

- "Digit span" (repeat a sequence of numbers) – measure maximum length.

**Confounds to track:** Daily mood, illness, medication, caffeine intake, screen time, physical activity, and major life events (e.g., starting school, family stress).

**Key confounds to control for:**

**Age:** Sleep needs decrease with age; a 3-year-old needs more sleep than a 5-year-old.

**Napping:** Some preschoolers still nap; others don't. Total 24-hour sleep matters more than night-time sleep alone.

**Weekend vs. weekday:** Many children have later bedtimes on weekends; analyse weekdays separately.

**Seasonal changes:** Daylight saving time, longer summer days, and winter darkness can affect sleep timing.

**Parental behaviour:** If you change your child's bedtime, you may also change your own behaviour (e.g., more patience in the morning). This is a confound you cannot fully control in an n=1 experiment.

**Regression to the mean:** If you start the experiment when your child is having a particularly bad sleep week, any improvement may be due to natural recovery, not your intervention.

**What a positive result would look like:**

**Sleep:** Total sleep time increases by 30–60 minutes per night, with fewer night wakings (e.g., from 2 to 0 per night).

**Behaviour:** Daily behaviour ratings improve by at least 1 point on a 5-point scale (e.g., tantrums decrease from "frequent" to "occasional"). This improvement should be consistent across the 2-week intervention period, not just a single good day.

**Cognition:** Reaction time on the Day/Night Stroop decreases by 10–20%, or digit span increases by 1 digit. These changes should be measured at the same time each day to control for circadian effects.

**Statistical signal:** If you plot daily behaviour scores, you should see a clear shift in the mean after starting the intervention, with minimal overlap between baseline and intervention periods. A simple t-test (if you have 7 baseline and 14 intervention data points) should show p < 0.05, but be cautious: n=1 statistics are prone to false positives due to autocorrelation (today's behaviour predicts tomorrow's). A more conservative approach is to look for a consistent trend over time (e.g., linear regression with a significant slope).

**Caveat:** The effect sizes in this review are small. Even with a perfect intervention, you may not see a dramatic change in your child's behaviour or cognition. The review suggests that sleep is one factor among many (genetics, environment, nutrition, parenting style). A positive result would be a small but noticeable improvement, not a transformation. If you see no change, it does not mean sleep doesn't matter – it may mean your child's sleep was already adequate, or that other factors are more influential for your specific child.