Importance: The definition of autism has evolved significantly, reflected in multiple revisions of diagnostic criteria (DSM-III to DSM-5), shifting from a categorical to a spectrum-based view. This evolution and broadened definition have coincided with a notable increase in autism prevalence. These definitional changes, potentially introducing greater heterogeneity within the autism population, may influence the outcomes of autism research, particularly by possibly reducing effect sizes in studies comparing groups.
Objective: This study aimed to investigate the relationship between publication year and effect size in autism versus control group comparisons across various domains of neurocognitive research in autism. Specifically, it explores if the magnitude of differences observed between individuals with autism and control groups has changed over time. This implicitly involves “Comparing Sizes” of group differences as research progresses and diagnostic criteria evolve.
Data Sources: A meta-analysis was conducted, examining 11 existing meta-analyses. These were identified through a systematic PubMed search for meta-analyses published between January 1, 1966, and January 27, 2019, using the search terms “autism AND (meta-analysis OR meta-analytic)”. The literature search concluded on January 27, 2019.
Study Selection: Included meta-analyses were those that assessed the significance of differences between autism and control groups on neurocognitive constructs. To be included, meta-analyses had to report a significant group difference and encompass data spanning at least 15 years, allowing for temporal trend analysis.
Data Extraction and Synthesis: Data were extracted and analyzed following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines. Fixed-effects models were utilized for data synthesis.
Main Outcomes and Measures: The primary outcome measure was the estimated slope of the correlation between publication year and effect size. This analysis controlled for variations in research methodologies, sample sizes, and study quality to isolate the temporal trend effect on effect sizes.
Results: The 11 meta-analyses incorporated data from 27,723 participants. Complete demographic information, such as sex and age, was not universally available. The analysis spanned seven distinct psychological and neurological constructs. Across all constructs, a downward trend in effect size over time was observed (slopes ranging from -0.067 to -0.003). This temporal decline was statistically significant for five of the seven constructs: emotion recognition (slope: -0.028 [95% CI, -0.048 to -0.007]), theory of mind (-0.045 [95% CI, -0.066 to -0.024]), planning (-0.067 [95% CI, -0.125 to -0.009]), P3b amplitude (-0.048 [95% CI, -0.093 to -0.004]), and brain size (-0.047 [95% CI, -0.077 to -0.016]). Notably, when comparing brain sizes, the study found that the difference in brain size between individuals with autism and control groups has diminished in more recent studies compared to older studies.
In contrast, three analogous constructs examined in schizophrenia, a condition also characterized by heterogeneity but without a reported prevalence increase, did not exhibit a similar temporal trend of decreasing effect sizes. This comparison highlights the specificity of the findings to autism and its evolving diagnostic landscape.
Conclusions and Relevance: The findings indicate that the differences between individuals diagnosed with autism and those without the diagnosis have lessened over time. This suggests that shifts in the conceptualization of autism, moving from a narrowly defined, homogenous group to a more inclusive and heterogeneous population, may be impacting our ability to develop robust mechanistic models of autism. The decreasing “sizes” of observed effects over time may reflect the inclusion of a broader spectrum of individuals under the autism diagnosis, which is a crucial consideration for future research and diagnostic approaches.
