Wealth and air pollution emerge as top predictors of state autism rates

A recent study published in Psychological Reports suggests that a state’s socioeconomic status and its levels of air pollution are the strongest predictors of its Autism Spectrum Disorder prevalence. The research provides evidence that higher average wealth and education, combined with higher microscopic particle pollution, tend to align with higher rates of autism diagnoses across the United States. These findings offer a new perspective on how broad environmental factors interact with regional health trends.

Autism Spectrum Disorder, commonly referred to as ASD, is a developmental condition that affects how people communicate, learn, and interact with the world. Over the past twenty years, the rate of autism diagnoses has increased significantly across the country. However, the prevalence of these diagnoses is not uniform, and some U.S. states report noticeably higher numbers of autism cases than others.

Stewart J. H. McCann, professor emeritus in the department of psychology at Cape Breton University, conducted the new study to explore what might be driving these regional differences. He wanted to see how various social, economic, and environmental factors interact to predict state-level autism rates. The overarching framework for the project relies on geographical psychology, which is a scientific field that maps how psychological traits, health outcomes, and behaviors organize themselves across physical spaces.

McCann has spent decades mapping these types of psychological and physical differences across the country. “Most of my research in the past 20 years has focused on the implications of the geographical dispersion of personality and other individual differences across the U.S. states,” McCann said. He noted that his past studies have covered a wide variety of topics, including emotional health, obesity, voting habits, labor force participation, and Alzheimer’s disease.

He decided to turn his attention to the rising rates of autism development to see if geographic patterns could offer new insights. “The contemporary interest shown by the federal government and others in the rather alarming increases in the reported prevalence of ASD over recent years spurred my curiosity regarding the potential factors that might be involved in the differences in the U.S. state ASD rates,” McCann explained.

Free daily newsletter

Past research on individuals suggests that autism diagnoses are associated with a wide array of variables. These include maternal age, local healthcare availability, urban living, and racial demographics. Interestingly, many of these same variables also share a strong relationship with socioeconomic status, which is a combined measure of a person’s income and educational background.

Socioeconomic status, often abbreviated as SES, tends to be a heavy influence in all areas of sociological and health research. McCann noticed that because SES is so deeply intertwined with other factors like healthcare access and city living, it might be the underlying reason why those other factors seem to predict autism rates. He set out to test fifteen different variables at the state level to see which ones actually maintain their predictive power when SES is factored into the mathematical equation.

To conduct the study, McCann gathered data from all fifty U.S. states. He focused primarily on information from the year 2017. He utilized a statistical model to analyze state-wide autism prevalence rates for adults aged 18 to 84, and he then collected state-level data for fifteen potential predictors to see how they aligned with the autism numbers.

These predictors included SES, racial demographics, average intelligence scores, urbanization percentages, and the concentration of microscopic air pollution particles. The author also factored in the number of local mental health and pediatric providers, physician shortages, education spending per student, and the percentage of uninsured residents.

In addition, the study included data on maternal age, obesity rates before pregnancy, low birth weight percentages, and enrollment in government health insurance programs like Medicaid. McCann also measured early childhood policy strategies, known as prenatal-to-3 policies, which evaluate how well a state supports equitable early childhood care. Finally, the study looked at a state’s average personality profile, utilizing the Big Five personality traits, which measure openness, conscientiousness, extraversion, agreeableness, and neuroticism.

McCann first looked at how each of these variables related to autism rates on a basic level. The initial analysis showed that autism rates correlated significantly with nearly all of the variables. States with higher autism rates tended to have higher SES, higher intelligence scores, more urban populations, older mothers, and greater per-pupil educational spending.

However, basic correlations can be misleading when researchers fail to account for overlapping influences. McCann used a technique called statistical control, which is a mathematical method to freeze one variable in place so scientists can see the true, independent effect of another. He ran a series of calculations that controlled for SES to see if the other fourteen variables still predicted autism rates.

When SES was held constant, most of the other variables lost their predictive power. Only a state’s racial makeup, average personality profile, urban population percent, air pollution, prenatal-to-3 policies, and maternal age continued to show a significant relationship with autism rates. This suggests that for variables like healthcare provider availability or education spending, the state’s wealth and education level were doing the actual predictive work.

In a final step, McCann placed SES and the six surviving variables into a single statistical model together. When they were all tested at once, only SES and air pollution emerged as significant, independent predictors of state autism rates. The other variables dropped out of the picture when forced to compete directly with SES and air pollution.

“I was surprised that the correlations of 14 of the 15 potential predictors with ASD, largely gleaned from previous research with individuals as the units of analysis, also replicated with states as the analytic units,” McCann told PsyPost. “I also was surprised by the fact that only SES and air pollution ultimately showed any independent relation to state ASD prevalence rates.”

Together, SES and air pollution accounted for 55.7 percent of the variance in autism rates across the fifty states. Variance refers to how much the numbers fluctuate from the average. This means that more than half of the differences in autism rates from state to state can be explained just by looking at their socioeconomic standing and air quality.

These results provide evidence that wealth and pollution metrics are vital pieces of the puzzle. “These findings suggest that any future analyses using either states or individuals as the analytic units could benefit from paying careful attention to SES and air pollution as potentially important statistical controls when evaluating other explanatory factors regarding the development and diagnosis of ASD,” McCann said.

The author also ran a supplementary analysis looking at Attention-Deficit/Hyperactivity Disorder, commonly known as ADHD. He wanted to ensure that the patterns he found were specific to autism and not just a general trend for all neurodevelopmental conditions. The data showed that state ADHD rates had the exact opposite relationship with the fifteen predictors, providing evidence that the autism findings were unique to that specific diagnosis.

While the findings offer important insights, readers should avoid drawing strict causal conclusions from this research. The study uses a cross-sectional design, meaning it takes a snapshot of data at a single point in time rather than tracking changes over many years. Because of this, it is impossible to state definitively that air pollution or high SES directly cause autism.

McCann emphasized this point when discussing the boundaries of the data. “The present study has two important limitations,” McCann noted. “First, the cross-sectional nature of the research means that readers cannot draw inferences regarding cause and effect and that the results cannot serve as empirical support for any causal speculations that might pertain to the findings.”

It is quite possible that higher SES simply leads to better access to diagnostic services. Wealthier states tend to have better healthcare infrastructure, which makes it easier for parents and adults to seek out and receive official psychological evaluations. The study cannot entirely separate a true biological rise in autism development from a simple rise in diagnostic awareness and medical access.

Another potential misinterpretation involves the ecological fallacy, which happens when people assume that broad geographic data perfectly applies to individual people. “The findings apply to U.S. states, and whether similar relations also apply to individuals cannot be known without replicative research, although speculation suggests that the state-level relations do stem from parallel individual-level relations,” McCann said.

Moving forward, researchers suggest that scientists will need to investigate the role of maternal and paternal age more deeply. Older parenthood is strongly tied to higher SES, as wealthier individuals often delay childbirth to pursue education and careers. Continued research in these areas may help public health officials craft better policies to understand and support developing children.

The study, “An Exploratory Study of Socioeconomic Status, Air Pollution, and 13 Other Variables as Predictors of U.S. State Autism Spectrum Disorder Rates,” was authored by Stewart J. H. McCann.