rethinkTHC Search
Menu
Study breakdown

A US Epidemiological Study Found Cannabis Use Was Associated with Rising Autism Rates, Including a Dose-Response Pattern

EpidemiologicalModerate evidence
The takeaway

Using national education and drug survey data on 40 million 8-year-olds, daily cannabis use and first-trimester pregnancy exposure were both significantly associated with autism rates, with THC and cannabigerol showing exponential dose-response relationships at the state level.

Autism researchers, epidemiologists, reproductive health specialists, drug policy researchers.

Daily cannabis use was significantly associated with autism rates across US states

What the researchers found

National-level analysis found daily cannabis use significantly related to autism rates (beta=4.37, P<10^-16) and first-trimester exposure (beta=0.12, P=1.7x10^-12). At state level, THC (beta=1.96, P=4x10^-4) and cannabigerol (beta=-13.77, P=1.8x10^-6) were significant. Geospatial modeling showed exponential relationships. Legalization status was linked to elevated autism rates.

Why it matters

If the association between cannabis use and autism rates is causal, it would have major implications for prenatal guidelines and cannabis policy. However, this type of ecological analysis has significant limitations in establishing causation.

The numbers in context

N=40,119,464 eight-year-olds (1994-2011). 266,950 autistic. Daily cannabis beta=4.37 (P<10^-16). First-trimester exposure beta=0.12 (P=1.7x10^-12). THC exponential coefficient=7.053 (6.39-7.71). CBG coefficient=185.334 (167.88-202.79).

How the study worked

Longitudinal epidemiological study using IDEA autism census data (1991-2011), NSDUH drug exposure data, and US Census data for 40,119,464 8-year-olds across US states. Two-way fixed-effects and geospatial modeling.

What this study cannot tell us

Ecological study design (state-level correlations cannot prove individual-level causation). Autism diagnostic criteria changed significantly during the study period. Many confounders could explain both rising cannabis use and rising autism diagnoses. Prior work by these authors has drawn methodological criticism. E-values do not prove causation.

How to read the evidence

Large ecological analysis with significant associations, but ecological design limits causal inference. Diagnostic changes over the study period are a major confounder.

When this study was published

Published in 2023 using data from 1991-2011.

The bigger picture

Autism rates have risen dramatically in the US over recent decades, and the causes remain debated. This analysis from Reece and Hulse argues cannabis is a contributing factor, but many researchers attribute the rise primarily to expanded diagnostic criteria and awareness. Ecological studies like this cannot distinguish between these explanations.

Questions still open

  • Can individual-level prospective studies confirm a link between prenatal cannabis exposure and autism?
  • How much of the autism rate increase is due to diagnostic expansion vs. true incidence changes?

Common questions

Is there a link between cannabis and autism?
This ecological study found a statistical association between cannabis use and autism rates at the state level, but ecological correlations cannot prove that cannabis causes autism in individuals.
Does prenatal cannabis use cause autism?
First-trimester cannabis exposure was associated with autism rates in this study, but the ecological design, changing diagnostic criteria, and potential confounders make it impossible to confirm a causal link from this data alone.

Read the original research

Impact of converging sociocultural and substance-related trends on US autism rates: combined geospatiotemporal and causal inferential analysis.

European archives of psychiatry and clinical neuroscience, 273(3), 699-717

Citation

Reece, Albert Stuart; Hulse, Gary Kenneth. (2023). Impact of converging sociocultural and substance-related trends on US autism rates: combined geospatiotemporal and causal inferential analysis.. European archives of psychiatry and clinical neuroscience, 273(3), 699-717. https://doi.org/10.1007/s00406-022-01446-0

Explore the wider topic