Associations of autism diagnosis, traits, and genetic liability with subsequent night-time sleep duration trajectories from infancy to adolescence

Using data from a large longitudinal cohort, this study identifies four distinct night-time sleep duration trajectories from infancy to adolescence and finds that an autism diagnosis and specific autistic traits (repetitive behavior, speech coherence, and social communication), but not the autism polygenic score or sociability, are associated with shorter sleep trajectories, underscoring the need for early, sustained, and trait-specific sleep interventions.

The Gist

Imagine your child's sleep pattern as a journey on a highway that starts when they are a baby and continues all the way through their teenage years. Most kids travel on a wide, smooth, middle lane where they get a "normal" amount of sleep every night. But for some children, the road looks different. They might be stuck in a lane that is consistently shorter (less sleep) or take a winding path that changes over time.

This research paper is like a traffic report that tries to figure out why some drivers (children) end up in the "short sleep" lanes, specifically looking at the connection to Autism.

Here is the breakdown of what the researchers found, using simple analogies:

1. The Map: Four Different Sleep Lanes

The researchers looked at data from nearly 14,000 children over 15 years. Instead of just looking at one night of sleep, they mapped out the whole journey. They discovered that children generally fall into four distinct "sleep lanes":

• The Long Lane: Kids who consistently get a lot of sleep.
• The Short Lane: Kids who consistently get very little sleep.
• The Middle-Short Lane: Kids who get a decent amount, but slightly less than average.
• The Middle-Long Lane (The "Normal" Lane): The biggest group. These kids get a standard, healthy amount of sleep. This was used as the baseline for comparison.

The Big Discovery: As children grew up, the differences between these lanes got smaller (like lanes merging together), but the "Short Lane" group stayed consistently behind the rest.

2. The Driver's License: Does Autism Put You in the Short Lane?

The team asked: "If a child has an Autism diagnosis, are they more likely to be driving in the Short Lane?"

The Answer: Yes, absolutely.
Having an Autism diagnosis was like having a specific type of car that was statistically much more likely to end up in the Short or Middle-Short lanes. Children with autism were significantly more likely to have less sleep than their peers, and this pattern lasted from infancy all the way through adolescence.

3. The Engine Parts: Which Specific Traits Cause the Trouble?

Autism is a bit like a complex engine with many different parts. The researchers wanted to know: Is the whole engine the problem, or are specific parts causing the sleep issues?

They tested four specific "parts" (traits):

• 🗣️ The Social Communication Engine: (Difficulty understanding social cues or talking back and forth).
  • Result: Broken. Kids with higher difficulties here were much more likely to be in the Short Lane. It's like if the radio is static, it's hard to relax and drift off.
• 🗣️ The Speech Coherence Engine: (How clear and logical speech is).
  • Result: Broken. Kids with lower speech coherence (harder to understand or less clear speech) were also more likely to be in the Short Lane.
• 🔄 The Repetitive Behavior Engine: (Rocking, repeating actions, or needing strict routines).
  • Result: Broken. Kids with more repetitive behaviors were more likely to get less sleep. Imagine a car that keeps revving its engine at a red light; that extra energy makes it hard to stop and sleep.
• 🤝 The Sociability Engine: (Do they prefer being alone or with people?).
  • Result: Fine. Interestingly, whether a child preferred being alone or with others did not predict their sleep problems. Just because a child likes to be alone doesn't mean they will sleep poorly.

4. The Blueprint: Is it in the Genes?

The researchers also looked at the genetic blueprint (the DNA instructions) for autism. They asked: "If a child has a high genetic load for autism, will they automatically have less sleep?"

The Answer: No.
Surprisingly, the Autism Polygenic Score (a measure of genetic risk) was not linked to the sleep lanes.

• The Metaphor: Think of genetics as the blueprint for a house. You might have the blueprint for a house with a specific roof (Autism), but that doesn't automatically mean the plumbing (Sleep) will be broken. The sleep issues seem to be caused more by the daily behaviors (the traits mentioned above) than by the raw genetic code itself.

The Takeaway: What Does This Mean for Families?

1. It's a Long-Term Issue: Sleep problems in autistic children aren't just a "baby phase." They tend to follow the child from infancy through their teenage years.
2. It's About Specific Behaviors: Not all autistic traits affect sleep. The biggest culprits seem to be social communication struggles, speech clarity, and repetitive behaviors.
3. Genetics Isn't Destiny: Just because a child has a genetic risk for autism doesn't mean they are doomed to have sleep issues. The way they behave and interact (their traits) plays a bigger role in their sleep than their DNA does.

In short: If you are a parent or caregiver, this study suggests that helping a child with communication skills and managing repetitive routines might be the key to unlocking better sleep, rather than just worrying about the genetic diagnosis itself. It's about fixing the specific "engine parts" that are causing the car to run hot at night.

Technical Summary

1. Problem Statement

Autistic individuals experience high rates of sleep problems (40–83%), which are linked to adverse mental and physical health outcomes. However, existing research is largely cross-sectional, focusing on average group differences rather than individual variation over time. This limits the understanding of how sleep patterns evolve from infancy to adolescence in autistic populations. Furthermore, the mechanisms driving these sleep issues remain unclear: it is unknown whether they are driven by a formal autism diagnosis, specific behavioral traits (e.g., repetitive behaviors vs. social communication), or shared genetic liability (polygenic scores). There is a need to identify heterogeneous sleep trajectories and determine which specific autism-related factors predict membership in these trajectories.

2. Methodology

Study Design and Cohort

• Data Source: The Avon Longitudinal Study of Parents and Children (ALSPAC), a UK-based population birth cohort.
• Sample Size: $N = 13,886$ participants (150 autistic, 13,886 non-autistic).
• Timeframe: Longitudinal data collected from infancy (0.5 years) to late adolescence (15.5 years).

Variables

• Outcome: Night-time sleep duration measured on 10 occasions (6 months to 15.5 years).
  • Measurement: Parent-reported (ages 0.5–12) and self-reported (age 15.5). Derived from bedtimes and wake times.
• Independent Variables (Exposures):
  1. Autism Diagnosis: Identified via multi-source data (parent report at 9.5y, NHS health records, and education records).
  2. Autism Polygenic Score (PGS): Derived from GWAS summary statistics (Grove et al., 2019), standardized to mean 0/SD 1.
  3. Autistic Traits: Four parent-reported domains measured at specific ages:
     • Social Communication (SCDC, age 7.7y).
     • Speech Coherence (CCC, age 9.7y).
     • Repetitive Behaviour (ALSPAC questionnaire, age 5.8y).
     • Sociability Temperament (EAS scale, age 3.2y).
• Covariates: Sex, maternal postnatal depression, maternal anxiety, epilepsy diagnosis, and socioeconomic status (education, occupation, financial problems).

Statistical Analysis

• Trajectory Identification: Latent Class Growth Analysis (LCGA) was used to identify distinct subgroups of sleep duration trajectories. Models with 2–5 classes were tested using Bayesian Information Criterion (BIC), Vuong–Lo–Mendell–Rubin (VLMR) test, and entropy.
• Association Testing: Multinomial logistic regression was used to regress the identified trajectory classes onto the independent variables.
  • The largest class (Intermediate-longer) served as the reference group.
  • Analyses were conducted for both continuous trait scores and binary "high trait" groups.
  • Missing data were handled via Full Information Maximum Likelihood (FIML) for outcomes and Multiple Imputation (50 datasets) for covariates.

3. Key Contributions

• Longitudinal Scope: This is one of the first studies to map sleep trajectories from infancy (6 months) through late adolescence (15.5 years) in relation to autism, moving beyond narrow developmental windows.
• Trait-Specific Analysis: Unlike previous studies that treated autism as a monolith, this study disentangles the effects of a formal diagnosis from specific behavioral domains (repetitive behavior, speech, social communication, sociability) and genetic liability.
• Genetic vs. Behavioral: It explicitly tests whether the genetic liability for autism (PGS) predicts sleep trajectories, distinguishing between biological predisposition and behavioral manifestations.

4. Results

Sleep Trajectory Classes
The LCGA identified four distinct night-time sleep duration trajectories:

1. Shorter: 4.1% ($n=512$).
2. Longer: 13.1% ($n=1,654$).
3. Intermediate-shorter: 28.8% ($n=3,630$).
4. Intermediate-longer (Reference): 54.1% ($n=6,825$).
   All trajectories showed a decreasing trend from childhood to adolescence, with differences most pronounced in early childhood.

Associations with Autism Factors

• Autism Diagnosis: Strongly associated with Shorter (OR = 6.64) and Intermediate-shorter (OR = 1.89) trajectories compared to the reference class.
• Specific Autistic Traits:
  • Social Communication Difficulties: High scores associated with Shorter sleep (OR = 2.55).
  • Repetitive Behaviour: High scores associated with Shorter sleep (OR = 1.81).
  • Speech Coherence: Low scores (poor coherence) associated with Shorter sleep (OR = 2.28).
  • Sociability: No significant association with any sleep trajectory.
• Genetic Liability (PGS): The autism Polygenic Score showed no significant association with any of the sleep duration trajectories.

Summary of Associations:

Factor	Shorter Sleep	Intermediate-Shorter Sleep	Longer Sleep
Autism Diagnosis	Yes (Strong)	Yes (Moderate)	No
Social Comm. Diff.	Yes	No (Marginal)	No
Repetitive Beh.	Yes	No	No
Low Speech Coherence	Yes	No	No
Sociability	No	No	No
Autism PGS	No	No	No

5. Significance and Implications

• Trait-Specific Mechanisms: The findings suggest that sleep problems in autism are not uniform. Specific traits (repetitive behaviors, social communication deficits, speech coherence) drive shorter sleep trajectories, while others (sociability) and genetic liability do not. This implies that interventions should be trait-specific rather than one-size-fits-all.
• Developmental Scalability: The association between autistic traits and sleep persists from infancy through adolescence, suggesting that mechanisms (e.g., anxiety related to social interaction, cognitive rumination related to repetitive behaviors) are active across developmental stages, not just in early childhood.
• Genetic vs. Environmental: The lack of association between the Autism PGS and sleep trajectories suggests that the link between autism and sleep is likely mediated by behavioral phenotypes and environmental interactions rather than a direct shared genetic architecture.
• Clinical Recommendations: The study highlights the need for early, sustained sleep interventions that target specific autistic traits (e.g., addressing bedtime routines for repetitive behaviors, or communication strategies for speech coherence issues) rather than treating sleep issues solely as a symptom of the diagnosis.

Limitations:

• Reliance on parent/self-report for sleep (subjective) rather than objective measures (actigraphy).
• General population sample may not fully capture the severity of traits in clinical autism populations.
• Cross-sectional nature of trait measurement (traits measured at one time point vs. sleep over time).