The Plan to Find the Causes of Autism Is Doomed to Fail

Health and Human Services Secretary Kennedy’s recently unveiled plan to identify the causes of autism relies heavily on chart reviews and retrospective data. While framed as a step toward addressing a public health issue, the plan’s foundation is scientifically weak. Without a rigorous longitudinal, population-based design and an understanding of the genetic and biopsychosocial complexities of autism, the plan is likely to mislead, misallocate resources, and ultimately fail.

The Flaws of Chart Reviews and Retrospective Data

Chart reviews depend on existing medical records, which vary in quality and often lack standardized diagnosis or symptom severity measures. While retrospective studies may identify correlations, they cannot establish causality. These studies are particularly limited in developmental disorders like autism, where the timing of symptom onset, environmental exposures, and early-life variables are essential. Any analysis utilizing this data is built on shaky ground.

Retrospective designs are also susceptible to recall bias, selection bias, and missing data—all of which undermine validity. They frequently overrepresent children who have already been diagnosed and treated within specific healthcare systems, thereby excluding marginalized or undiagnosed populations. This results in an incomplete picture of autism's diverse presentations and origins.

Autism Is Not One Thing—It’s Many

Autism spectrum disorder (ASD) is highly heterogeneous. It is not caused by a single gene or exposure but likely by the complex interaction of hundreds of genes and environmental factors. Current genomic research shows that autism involves both rare de novo mutations and common polygenic risk variants (Sandin et al., 2014). No two individuals with autism have the same genetic profile.

For instance, mutations in CHD8, SHANK3, and SCN2A genes have been associated with autism, as well as with intellectual disability, epilepsy, or even a lack of clinical symptoms, depending on the individual’s genetic background and environment (Bourgeron, 2015). Chart reviews cannot account for this level of nuance.

Multifinality and Equifinality: The Genetic Maze

Multifinality refers to the phenomenon where the same genetic variation can lead to multiple outcomes, while equifinality indicates that different pathways can result in the same condition. These concepts are essential for understanding neurodevelopmental disorders. A pseudoscientific approach that disregards this complexity is misguided and risks perpetuating simplistic narratives that may stigmatize families or generate false hopes.

Genetic risk is probabilistic, not deterministic. For instance, having a sibling with autism increases the risk, but it is not predictive. Environmental factors such as prenatal exposures, parental age, and early-life stress dynamically interact with genetic susceptibilities. A thorough investigation must account for these layers.

The Need for a Longitudinal, Biopsychosocial Approach

Only a longitudinal, population-based study can begin to answer the question of autism's origins. Such studies recruit a large cohort of babies at birth (or even prenatally) and follow them over time, collecting biological, psychological, social, and environmental data. This enables researchers to track development and identify risk and resilience factors.

Examples of rigorous longitudinal studies include:

1. The Norwegian Mother and Child Cohort Study (MoBa) has tracked more than 100,000 pregnancies since 1999 and identified associations between prenatal factors and later neurodevelopmental outcomes (Magnus et al., 2006).
2. The Autism Birth Cohort Study (ABC), nested within MoBa, has provided valuable data linking genetic variants with early behavioral markers of ASD (Stoltenberg et al., 2010).
3. The Early Autism Risk Longitudinal Investigation (EARLI) focuses on infants at high risk for autism and examines environmental exposures, epigenetic markers, and early signs of ASD (Fallin et al., 2013).
4. Generation R Study in the Netherlands, tracks more than 9,000 children from fetal life onward and has provided insights into early brain development and psychiatric risk (Jaddoe et al., 2012).
5. SEED (Study to Explore Early Development) is a Centers for Disease Control and Prevention-led project that examines risk factors in diverse U.S. populations (Schendel et al., 2012).
6. Avon Longitudinal Study of Parents and Children (ALSPAC) is a U.K.-based study that identifies prenatal and familial influences on various developmental outcomes (Boyd et al., 2013).

Autism Essential Reads

A Brief History of the Science of Autism Spectrum Disorder

We Already Know What Causes Autism

These studies have unequivocally found that identifying causal pathways requires integrating genetics, neurodevelopmental markers, environment, and family context over time.

What Secretary Kennedy’s Plan Misses

Kennedy’s plan reduces autism research to a simplistic search for a “cause,” as though autism were a singular condition with a singular explanation. It fails to incorporate the foundational principle of developmental psychopathology: that behavior emerges from complex, evolving interactions among genes, brains, and environments. The plan lacks scientific rigor and risks diverting resources from practical, evidence-based strategies by relying on retrospective chart reviews and neglecting longitudinal evidence.