Apparent historical increases in autism are exaggerated because older cohorts are undercounted: many were never diagnosed in childhood, and higher mortality among severely affected autistics removes cases before adult surveys. Comparing today’s well‑ascertained children to yesterday’s sparsely diagnosed, partially deceased adults produces a misleading slope.
— This cautions policymakers and media against reading long‑run autism graphs as causal evidence and pushes for bias‑aware trend methods before funding or regulatory shifts.
2026.01.05
82% relevant
The article highlights historical shifts in diagnostic constructs and boundaries — exactly the mechanism that creates cohort and ascertainment bias in prevalence series; its note that DSM‑5 criteria have not been prospectively compared supports the existing idea that observed increases can be partly an artefact of changing case definitions and surveillance.
2026.01.05
86% relevant
The piece quotes the argument that older cohorts were under‑diagnosed and experienced higher mortality for severe cases, which is the cohort‑bias explanation for apparent increases; Kling’s synthesis recapitulates the exact measurement concerns named in that existing idea.
Josh Zlatkus
2025.12.29
90% relevant
The article flags that diagnostic boundaries have widened and prevalence rises partly from definitional change — the same mechanism that the 'cohort bias inflates autism trends' item warns about (diagnostic drift, undercounted older cohorts); it cites autism prevalence and the need to decompose real liability from diagnostic expansion.
Colin Wright
2025.12.03
72% relevant
The article uses the same logic applied in autism debates — that diagnostic practice, cohort ascertainment, and social visibility can create an apparent 'epidemic' — and applies it to transgender ID spikes (referring to rapid‑onset gender dysphoria and changes over 2008–2018), connecting to the idea that trends can be driven by measurement and drift rather than true incidence.
Cremieux
2025.10.15
70% relevant
The author lays out a liability–threshold decomposition using Swedish CATSS/NPR data to show increases in diagnosed autism can be explained by shifting diagnostic thresholds rather than true prevalence growth, consistent with the claim that observed trends reflect measurement/ascertainment effects.
2025.10.07
100% relevant
The article flags California Department of Developmental Services data and notes adult‑age ascertainment and shorter lifespans bias early cohorts downward.
2024.10.30
72% relevant
The article is a counterpoint to the cohort‑bias critique: Escher acknowledges measurement effects but argues they cannot explain the magnitude or exponential shape shown in administrative series—so it directly tests and disputes the cohort‑bias explanation documented in the existing idea.
2018.09.04
78% relevant
The meta‑analysis’s focus on preterm cohorts and its reported heterogeneity underscores why aggregate diagnosis trends must be decomposed (ascertainment, cohort effects); the article’s limitations section (heterogeneity, lack of low‑/middle‑income data) connects to concerns about cohort and measurement bias in prevalence series.
2017.01.04
78% relevant
Modabbernia et al. flag that methodological limitations and changing detection may explain apparent rises in ASD, which supports the existing idea that cohort and ascertainment biases exaggerate long‑run prevalence increases and must be accounted for in public discourse and policy.
2015.01.04
90% relevant
The JAMA Pediatrics paper provides a quantified case study showing that changes in registry/diagnostic practice account for ~60% of the observed rise in autism in a national cohort, which is exactly the type of cohort‑bias explanation the existing idea warns about (start‑year/ascertainment bias skewing trends).
2002.06.04
95% relevant
Croen et al. (2002) is an early, direct empirical example of the claim: it shows autism prevalence rose while prevalence of mental retardation without autism fell by a nearly equal absolute amount (9.1 vs 9.3 per 10,000), supporting the diagnostic‑substitution/cohort‑bias explanation.