Attributable, checkable public claims about autism, ADHD and “AuDHD” — the unofficial community term for having both autism and ADHD.
Review window26 September 2025–26 September 2026
Claims assessed52
VerdictsFour evidence categories
What this record does
Claims, not identities
Each verdict attaches to the central factual proposition, not to a person's diagnosis, worth or lived experience. Opinions, predictions, anonymous assertions and personal medical advice are outside scope.
AuDHD is not a separate formal diagnosis. Autism and ADHD are recognized conditions that can be diagnosed together. Community language is assessed only when it contains a specific factual proposition.
Snapshot
Claim trends in this review window
Counts reflect the claims found and assessed, not the prevalence of a belief or condition.
Autism24claims
ADHD22claims
AuDHD6claims
Supported21
False4
Misleading14
Unresolved13
Evidence record
Public claims assessed
Every card links to the direct claim record and to the evidence used for the finding.
SupportedADHD
Among adults with ADHD whose type 2 diabetes was diagnosed before age 45, ADHD medication use was associated with a higher rate of major adverse cardiac events in a multinational cohort.
Among adults with ADHD whose type 2 diabetes was diagnosed before age 45, medication-use periods were associated with more major adverse cardiac events, defined as myocardial infarction, stroke or cardiovascular death (hazard ratio 1.58, 95% confidence interval 1.18–2.10). The association remained after the reported adjustments. This supports the narrow subgroup association, not a claim that medication caused the events or that the result applies to all people taking ADHD medication. A related multinational hypertension cohort found no overall medication–event association, and a wider observational meta-analysis found no statistically significant overall cardiovascular association.
Speaker
Zihan Dong, Yiling Zhou and colleagues
Context
The authors reported a seven-setting, population-based cohort analysis of adults with type 2 diabetes diagnosed during 2010–2020, comparing periods with and without ADHD medication use among those with ADHD.
Publisher
BMC Medicine · Peer-reviewed, early accepted version of a multinational observational cohort study
This is an observational subgroup finding from an early accepted version, with potential residual confounding and differences in patients, medication types and baseline cardiac risk. The available early article presents the abstract but not full methods, country estimates or absolute subgroup event rates, limiting independent appraisal. The hypertension cohort studied a different population and does not directly refute this subgroup result. No individual should start, stop or change ADHD medication because of this finding; cardiovascular assessment, monitoring and treatment decisions require a qualified prescriber.
MisleadingADHD
NHS bodies in England have imposed a two-year minimum wait for ADHD and autism assessments.
West Yorkshire's NHS commissioning body directly confirms a two-year minimum from referral for routine autism and ADHD assessments with independent providers. Other boards do impose restrictions, but the North East and North Cumbria body specifies approximately 78 weeks, and Greater Manchester expressly says it has no fixed minimum. The headline's general two-year framing therefore overextends a real West Yorkshire policy.
Speaker
Anna Bawden and Denis Campbell
Context
The Guardian's headline described NHS bodies imposing a two-year minimum, while its report identified several integrated care boards with different waiting-time policies.
Publisher
The Guardian · Established reporting of regional NHS commissioning policies
The two-year rule is a local NHS-funded independent-provider policy, not a nationwide minimum or a rule for every NHS pathway. West Yorkshire describes an urgent-assessment exception for people meeting significant-risk criteria, with an aim of assessment within 12 weeks. Commissioning arrangements and individual waits can change; people seeking assessment should check current local NHS guidance with their provider or qualified clinician. These access figures do not measure autism, ADHD or AuDHD prevalence or diagnostic validity.
SupportedAutism
The observed association between prenatal paracetamol exposure and autism or ADHD weakens substantially when higher-quality evidence is emphasized; current evidence does not establish causation.
Unrestricted analyses reported small associations with autism (risk ratio 1.17) and ADHD (1.29). Estimates attenuated substantially, and in several models approached no association, after the authors restricted analyses to higher-quality reviews and lower-risk primary studies. An earlier BMJ umbrella review and a large Swedish sibling-control cohort also found that the available evidence does not establish a causal link. This supports the specific evidence-quality proposition, not a categorical claim that every possible effect has been ruled out.
Speaker
H. Tayebi-Hillali, M. Blanco-Suárez and colleagues
Context
The authors published a registered umbrella review and meta-meta-analysis of seven meta-analyses, covering more than three million participants, on 23 September 2026.
Publisher
Developmental Medicine & Child Neurology · Peer-reviewed umbrella review and meta-meta-analysis of observational evidence
The included studies are observational, share many primary cohorts, and differ in exposure measurement, confounding control and outcome ascertainment. Quality restriction reduces bias concerns but cannot prove the absence of a small effect. An FDA notice has highlighted possible associations, while acknowledging that causality is unestablished. This population evidence cannot determine an individual's pregnancy care; pain, fever and medicine decisions require qualified clinical guidance.
SupportedADHD
Among U.S. children aged 3 to 17, parent-reported diagnosed ADHD rose from 9.5% to 12.0% and autism from 3.1% to 4.5% between 2019–2020 and 2023–2024.
The study pooled 2019–2020 and 2023–2024 National Health Interview Survey data and found statistically significant increases in both parent-reported conditions, including after adjustment for selected demographic and socioeconomic characteristics. ADHD increased from 9.5% to 12.0% and autism from 3.1% to 4.5%. The cautious, survey-specific trend proposition is therefore supported.
Speaker
Benjamin Zablotsky, Lindsey Black, Amanda Ng and Stephen Blumberg
Context
The authors reported trends from the National Health Interview Survey in a peer-reviewed Pediatrics article published on 22 September 2026.
Publisher
Pediatrics · Peer-reviewed repeated cross-sectional analysis of a nationally representative household survey
These are weighted prevalence estimates for U.S. civilian noninstitutionalized children whose parent reported being told of a diagnosis by a physician or other health professional; the study did not independently reassess each child. The categories can overlap, but the paper does not estimate autism-and-ADHD co-occurrence, and the combined results must not be described as AuDHD prevalence. A rise in recorded prevalence is not incidence and does not by itself identify a cause; the authors note that earlier detection and awareness may contribute. The estimates do not diagnose an individual child.
SupportedADHD
From 2019 to 2025, Personal Independence Payment receipt for autism or ADHD among 16- and 17-year-olds in England grew fastest in the least deprived fifth of neighbourhoods.
The analysis reports that 51,148 16- and 17-year-olds, 3.4% of the age group, received PIP for autism or ADHD in 2025, compared with 16,045, or 1.3%, in 2019. In the least deprived fifth of neighbourhoods the recipient rate more than tripled from 0.8% to 2.5%; in the most deprived fifth it rose from 2.1% to 5.1%. Rates remained higher in more deprived areas, but proportional growth was fastest in the least deprived group, supporting the precise administrative trend.
Speaker
Nuffield Trust
Context
The independent health think tank published a new analysis of Department for Work and Pensions Stat-Xplore records linked to the 2025 English Indices of Deprivation.
Publisher
Nuffield Trust · Independent policy analysis of official administrative benefit records and area-level deprivation data
PIP receipt measures successful benefit claims involving functional impairment, not autism or ADHD prevalence, incidence, diagnostic accuracy or severity. The source combines autism and ADHD and does not report co-occurrence, so the result is not an AuDHD estimate. Deprivation is assigned from a neighbourhood-level employment index, and diagnosis access, awareness, eligibility evidence, application support, migration and PIP decision-making can all affect recorded rates. The descriptive analysis cannot establish that private diagnosis caused the pattern or that recipients in less deprived areas have lower individual needs.
SupportedAutism
In a Swedish nationwide cohort, registered congenital TORCH infections were associated with about three times the subsequent hazard of an autism diagnosis.
Registered congenital cytomegalovirus, toxoplasmosis, rubella or herpes infections were associated with autism in the population analysis (hazard ratio 3.10, 95% CI 2.55 to 3.76) and in a sibling comparison (hazard ratio 3.19, 95% CI 1.97 to 5.15). The association was stronger for autism with co-occurring intellectual disability. That supports the precise association claim; notably, the study's initial ADHD association attenuated in sibling analysis.
Speaker
Hugo Sjöqvist, Christina Dalman, David Mataix-Cols, Reneé Gardner and Håkan Karlsson
Context
A peer-reviewed JAMA Pediatrics cohort study linked Swedish national registers for 3,666,002 people born from 1987 through 2021, including 975 with a registered congenital TORCH infection.
This is an observational register study of rare, clinically recognized congenital infections, not evidence that common maternal infections or vaccines cause autism. The TORCH grouping combines different pathogens, diagnostic practices changed across the long birth period, and the smaller sibling sample leaves wider uncertainty. A relative hazard ratio is not an individual's absolute probability and cannot identify causation in a particular case. Infection prevention and pregnancy care should follow qualified public-health and clinical guidance rather than be inferred from this study alone.
SupportedADHD
ADHD medication dispensations to people aged 24 and younger in Manitoba rose substantially from 2015 to 2023, especially among ages 19 to 24 and females.
From 2015 to 2023, stimulant-dispensation prevalence rose from 183.64 to 337.99 per 10,000 and incidence from 12.76 to 23.95 per 10,000, both increases of more than 80%. Non-stimulant prevalence rose from 7.32 to 26.52 per 10,000 and incidence from 1.17 to 3.17 per 10,000. The largest stimulant-incidence increases were among people aged 19 to 24 and females, directly supporting the administrative trend proposition.
A dispensed prescription is not a direct measure of ADHD prevalence, diagnostic validity, medication adherence, clinical benefit, harm or whether treatment was appropriate. The data describe one Canadian province, stop in 2023 and may reflect changes in recognition, access, clinical practice, drug availability and population composition. The study does not justify starting, stopping or changing medication; individual treatment and monitoring require a qualified prescriber.
FalseADHD
Adult ADHD is overdiagnosed in the United Kingdom.
The survey directly supports the narrower statement that 583 of its 829 GP respondents believed adult ADHD was overdiagnosed. It does not establish that the diagnoses were incorrect: Pulse advertised the survey to its own readers, offered a prize-draw incentive, did not weight the sample and expressly described it as a non-scientific snapshot. The 81.3% figure was a midpoint estimate of GPs' recollections about patients who had already passed referral filters, not an audit of diagnostic criteria or false-positive diagnoses. A 2026 peer-reviewed UK expert review found no evidence that ADHD is overdiagnosed nationally and said available data instead point toward underdiagnosis. The government's independent interim review likewise found recorded ADHD prevalence below epidemiological benchmarks in most age groups and that almost all adults reporting a professional diagnosis screened positive on the ASRS, while cautioning that current adult prevalence data are limited. The categorical nationwide proposition is therefore contradicted by the best available UK evidence, even though concerns about individual misdiagnosis and assessment quality are legitimate.
Speaker
The Telegraph, reporting a Pulse survey of 829 GPs
Context
The newspaper headline stated that ADHD is over-diagnosed after 583 respondents to an unweighted Pulse reader survey selected either "slightly overdiagnosed" or "very overdiagnosed" for adult ADHD. The article also highlighted GPs' estimates that 81.3% of patients they referred received a diagnosis.
Publisher
The Telegraph · Established news report based on a self-selected clinician opinion survey
Overdiagnosis is difficult to measure, and a population can contain both missed cases and some incorrect diagnoses. The government review says robust recent adult prevalence data and direct diagnostic-quality audits remain limited, so the evidence cannot rule out problems in particular providers, pathways or subgroups. A high diagnosis rate among carefully selected referrals is not proof of overdiagnosis, and neither a screening scale nor a national prevalence comparison validates every individual assessment. Diagnosis should follow a specialist evaluation of childhood onset, impairment across settings, developmental and mental-health history, and alternative explanations; personal assessment or treatment decisions belong with a qualified clinician.
SupportedAutism
A complete census in Coxilha, Brazil, found autism prevalence of 3.36%—1 in 30—among children aged 2.5 to 12 years.
All 475 eligible children participated. Sixteen met diagnostic criteria after Mini-TEA screening and assessment by a neurologist experienced in autism, giving 16 divided by 475, or 3.36%—approximately 1 in 30. All 16 had also previously been diagnosed by other professionals, and one earlier diagnosis was rejected during the study assessment. The reported count and narrow prevalence proposition are therefore directly supported by the census record.
Speaker
Cassiano Forcelini, Ana Luisa Carregosa, Eduarda Leitzke, Gilmar da Silva, Regina Ampese, Itamara de Moura and Helena de Melo
Context
The authors of a peer-reviewed brief communication screened every child in the target age range in the small municipality of Coxilha, Rio Grande do Sul, from March through December 2024 and clinically assessed children above the screening threshold.
Publisher
Cadernos de Saúde Pública · Peer-reviewed complete-municipality screening census with clinical assessment
This is a census of one small, predominantly agricultural municipality, not a representative estimate for Rio Grande do Sul or Brazil. With only 16 cases, an approximate 95% Wilson interval spans about 2.1% to 5.4%; the 15-boy and one-girl split is especially imprecise and should not be generalized. The age range and direct screening-and-assessment method also differ from Brazil's 2022 census question about an existing professional diagnosis and from U.S. record-based surveillance of eight-year-olds. The result describes identified prevalence in Coxilha in 2024, not incidence, a national rate, a trend or a cause.
UnresolvedADHD
Six months of methylphenidate treatment causes subclinical systolic and diastolic cardiac dysfunction in children with ADHD.
After six months, the treated group had statistically significant increases in heart rate and systolic and diastolic blood pressure, lower tissue-Doppler measures of ventricular diastolic function and lower speckle-tracking measures of left- and right-ventricular systolic function. No child developed an arrhythmia, and conventional echocardiography detected no change. The study therefore identifies a subclinical imaging signal in its small cohort. Wider randomized-trial evidence supports small average increases in blood pressure or pulse with ADHD medicines, while a 3.9-million-person observational meta-analysis found no statistically significant association with cardiovascular disease in children and adolescents. The new study does not by itself establish clinically important cardiac dysfunction or overturn the wider safety evidence.
Speaker
Ahmed O. Abd Rab El Rasool, Amr Badreldeen, Doaa El Amrousy and Ola Younes
Context
The authors of a newly published study used tissue Doppler and two-dimensional speckle-tracking echocardiography before and after six months of methylphenidate treatment in 40 children with newly diagnosed ADHD, alongside 40 age- and sex-matched healthy controls.
Publisher
European Journal of Pediatrics · Peer-reviewed prospective observational echocardiography study
This was a non-randomized study of 40 treated children with only six months of follow-up; the abstract does not establish that the healthy controls underwent an equivalent longitudinal exposure comparison. Multiple sensitive imaging measures, a small sample and the absence of symptoms or conventional-echocardiography changes make clinical meaning and generalizability uncertain. Rare or long-term events cannot be assessed in a cohort this size. Current guidance recommends cardiovascular history and examination plus heart-rate and blood-pressure monitoring, with further evaluation for defined risk factors or concerning findings. No one should start, stop or change methylphenidate because of this card; individual risks and monitoring belong with a qualified prescriber.
MisleadingADHD
Noradrenaline reuptake inhibitors may have little to no effect on adult ADHD symptoms or daily functioning, while atomoxetine more than doubles serious adverse events compared with placebo.
The review reports more treatment discontinuation and non-serious adverse events with noradrenaline reuptake inhibitors and an atomoxetine serious-adverse-event risk ratio of 2.31. Those safety findings require attention. The broad benefit claim is not consistent with the wider synthesis, however. A 2025 umbrella review found at least moderate-certainty evidence that atomoxetine improves core adult ADHD symptoms, and a network meta-analysis found improved patient-reported response and quality of life without a statistically significant increase in serious adverse events. NICE therefore retains atomoxetine as an option for adults who cannot tolerate or do not respond to stimulants, with individualized titration and monitoring. Presenting the new review as showing little or no benefit and generally more harm overstates a disputed synthesis while omitting established evidence of symptom benefit.
The authors of a newly published systematic review used 21 randomized trials involving 5,124 adults to make a combined benefit-and-harm claim about noradrenaline reuptake inhibitors, principally atomoxetine, for adult ADHD.
Publisher
International Journal of Risk & Safety in Medicine · Peer-reviewed systematic review and meta-analysis of randomized trials
The new review's abstract does not provide the absolute number or nature of serious events, and relative risks can appear large when events are uncommon. Its certainty ratings range from very low for several outcomes to moderate for the atomoxetine serious-event estimate; trial duration, selective reporting, blinding and sparse events can materially affect both benefit and harm estimates. Atomoxetine can cause important adverse effects and is not appropriate for everyone, but this review alone does not establish that it is ineffective or that it causes twice as many serious events in routine care. People taking ADHD medication should not start, stop or change it because of this card and should discuss benefits, adverse effects and alternatives with a qualified prescriber.
SupportedADHD
ADHD pharmacotherapy is associated with reduced appetite and transient reductions in weight, body-mass index and growth velocity, while long-term nutritional and bone effects remain uncertain.
Across 27 studies, mostly in children or mixed-age samples, stimulant treatment was repeatedly associated with reduced appetite, lower energy intake during active treatment and temporary reductions in weight, body-mass index and growth velocity. An earlier meta-analysis of 18 studies and 4,868 participants found small reductions in height and weight with long-term methylphenidate, most prominent for weight in the first year and height in the first two to two-and-a-half years. NICE guidance independently recognizes decreased appetite and weight change as clinically important and recommends regular height and weight monitoring. The narrow evidence summary is therefore supported.
Speaker
Jens-Mirco Engbrink, Janina Dapprich, Tobias Fischer and Anja Markant
Context
The authors of a peer-reviewed scoping review mapped nutrition, eating, growth and bone outcomes reported for people with ADHD receiving medication.
Publisher
Nutrients · Peer-reviewed scoping review of 27 mostly observational studies
A scoping review maps evidence rather than producing one pooled causal estimate. Medication type, dose, treatment duration, age, baseline ADHD-related eating patterns and study design varied, and most evidence was observational. The review found limited and heterogeneous evidence for adults, nutritional biomarkers, metabolic outcomes and bone health; reduced bone density does not automatically mean more fractures, and some observational research reports lower injury-related fracture risk during stimulant treatment. The findings do not show that every patient will lose weight or have impaired growth and do not justify an individual medication or diet change without clinical review.
UnresolvedAutism
A model using three-minute smartphone videos of caregiver-child interaction offers a scalable, low-cost way to broaden early autism screening in toddlers.
Autism-CLIP reported an area under the receiver-operating-characteristic curve of 0.903 in an internal cohort of 103 toddlers and 0.873 in an external cohort of 102, outperforming the conventional machine-learning comparators tested by the authors. External evaluation and use of short naturalistic videos make this stronger than an internal-only proof of concept. Those results establish promising discrimination in the two reported cohorts, but they do not yet establish that the system is accurate, equitable, low-cost or beneficial when deployed for population screening.
Speaker
Ruikang Deng, Huishi Huang, Shaoli Lv, Yu Xing, Yijie Li, Cong You, Fei Wu, Hongzhu Deng and Zhengxing Huang
Context
The authors of a peer-reviewed multicentre study presented Autism-CLIP, a pretrained vision-language model for screening toddlers aged 15 to 24 months from naturalistic interaction videos without manual behavioral coding.
Publisher
Nature Communications · Peer-reviewed multicentre diagnostic-model development and external-evaluation study
The study includes only 205 toddlers, and an area-under-the-curve value does not tell families the chance that a positive or negative result is correct at a chosen threshold; predictive values change with prevalence and referral setting. Independent prospective replication is still needed across languages, cultures, devices, recording conditions, co-occurring developmental differences and community samples, with calibration, false-positive and false-negative consequences, privacy safeguards and workflow outcomes. Screening is not diagnosis: current guidance requires developmental history, clinical observation and qualified assessment. No individual should infer an autism diagnosis or delay a clinical evaluation from a video model result.
MisleadingAutism
A placenta study identified the earliest signs of autism risk and shows that maternal immune activation can be a gateway to autism in some pregnancies.
The underlying experiment used the viral mimic poly(I:C) to activate the immune system of pregnant mice at midgestation. Within 24 hours, about 30% of embryos had acute abnormalities ranging from lower fetal weight to absent external sensory-organ development; the affected embryos were male, and their placentas showed inflammatory and structural changes. Those findings support a mechanism inside this mouse model. Human studies separately report a modest association between maternal infection or fever and later autism diagnosis, but they do not establish that this mouse placental signature identifies human autism risk, explains the male diagnostic difference or provides an intervention target.
Speaker
Cold Spring Harbor Laboratory and Lucas Cheadle
Context
An institutional news release used a human-facing autism-risk headline and takeaway for a mouse maternal-immune-activation experiment, while later text disclosed that the reported abnormalities and autism-like traits were observed in male mice.
Publisher
Cold Spring Harbor Laboratory · Institutional research news release materially republishing a mouse study
The experiment did not study pregnant people, human placentas or children and did not observe an autism diagnosis. A synthetic viral mimic, mouse strain and dose cannot reproduce the varied infections, immune responses, genetics, environments or diagnostic pathways in people, and gross embryonic abnormalities are not equivalent to autism. Reviews describe maternal-immune-activation models as useful but difficult to translate because results vary with immune challenge, timing, strain and behavioral methods. The release's intervention language is therefore speculative. Pregnant people with infection or inflammation concerns should use established prenatal and clinical guidance rather than infer an autism test, prevention strategy or treatment from this animal study.
SupportedADHD
After adjustment for maternal psychiatric severity and family factors, prenatal exposure to benzodiazepines or Z-hypnotics was not significantly associated with diagnosed ADHD or autism in offspring.
Taiwanese insurance and birth-registry data identified 25,159 exposed offspring and 50,318 sex- and birth-date-matched unexposed controls whose mothers had insomnia, depression or anxiety. Crude diagnosis rates were slightly higher after exposure, but multivariable adjustment attenuated the associations: adjusted hazard ratio 1.08 for ADHD (95% confidence interval 0.99 to 1.18) and 1.08 for autism (0.88 to 1.32). The confidence intervals include no increase, so the stated narrow finding of no statistically significant adjusted association is supported. Earlier large cohorts and a systematic review likewise found that estimates weakened when designs better addressed indication and family confounding.
The authors of a peer-reviewed nationwide cohort study reported adjusted neurodevelopmental outcomes after restricting the comparison to mothers with diagnosed insomnia, depression or anxiety.
Publisher
Archives of Women's Mental Health · Peer-reviewed nationwide population-based retrospective birth cohort
A non-significant observational result is not proof that every medicine, dose, timing or individual pregnancy is risk-free. Prescription records may not measure ingestion, residual psychiatric severity and family factors can remain, diagnoses depend on health-care records and follow-up, and the autism estimate is compatible with a modest decrease or increase. The result concerns benzodiazepines and Z-hypnotics in this Taiwanese live-birth cohort and should not be generalized to other sedatives or outcomes. Medication for anxiety or insomnia during pregnancy should not be started, stopped or changed because of this study alone; personal benefits, risks and untreated illness require qualified prenatal and prescribing guidance.
SupportedAutism
Specific daily living skills measured in autistic children and adolescents predict employment, residential status and social relationships in adulthood.
The study followed 232 autistic participants from early childhood into their early thirties. Daily living skills had substantial predictive utility within the reported models for employment at age nine (R² 0.557), social relationships at age 14 (ROC AUC 0.839) and residential status at age 18 (ROC AUC 0.906); skills measured at age five explained about half the variance in adult vocational outcomes (R² 0.507). Community skills such as using money and a telephone repeatedly ranked among the stronger predictors. The data support the narrow longitudinal proposition that measured skills predicted these cohort outcomes.
Speaker
Elaine Clarke, Hannah Singer and Catherine Lord
Context
The authors of a peer-reviewed longitudinal study used machine-learning models to identify which item-level daily living skills across childhood and adolescence predicted defined outcomes in early adulthood.
Publisher
Development and Psychopathology · Peer-reviewed longitudinal cohort study
Prediction is not causation: the study does not show that training one skill will produce employment, independent living or a particular relationship outcome. Sample size differed by childhood wave, the cohort was 69% White, attrition and model selection can affect estimates, and the models need external validation in more representative groups and service systems. Daily living skills had only a weak relationship with subjective well-being, and adult outcomes also depend on preferences, opportunity, accessibility, discrimination, support and economic conditions. Findings should inform individualized support goals rather than define success or predict an individual's future.
MisleadingAutism
Applied behavior analysis is an unproven, inherently harmful autism therapy that crowds out more effective and humane treatment.
The reporting raises legitimate concerns about financial incentives, service quality and the goals and delivery of some ABA programs. The clinical proposition is nevertheless too broad. ABA is an umbrella for heterogeneous behavioral practices rather than one standardized treatment. A Cochrane review found weak, cautious evidence that early intensive behavioral intervention may improve adaptive behavior, language and cognitive outcomes for some young children, while the American Academy of Pediatrics says behavioral interventions can support specific skill development. These records do not establish that every ABA program is effective, appropriate or humane, but they contradict the blanket claim that the entire approach is unproven and inherently harmful.
Speaker
Beth Hawkins and Nicholas Perez
Context
An investigative article about private-equity investment and Medicaid spending described applied behavior analysis, or ABA, as unproven and even harmful and said its market dominance crowds out better treatment.
The evidence base has important weaknesses: small and non-randomized studies, variable programs and goals, limited long-term and quality-of-life outcomes, and inadequate monitoring of harms. A review of 150 early-autism intervention reports found that only 11 mentioned adverse events, so absence of reported harm is not proof of safety. Treatment quality also depends on individualized goals, consent or assent, practitioner conduct and whether the work supports the autistic person's priorities rather than normalization. This verdict does not endorse a particular provider or program; families should discuss benefits, burdens, alternatives and safeguarding with qualified clinicians and the autistic person where possible.
SupportedAutism
Across 12 South Korean birth cohorts born from 1998 through 2010, standardized screening at age seven found no significant increase or decrease in autism cumulative incidence or prevalence.
The study screened 62,081 children born from 1998 through 2010 in one South Korean city. Model-based seven-year cumulative incidence estimates ranged from 1.9% to 3.2% and prevalence estimates from 2.0% to 3.4%, with no statistically significant trend across cohorts. Diagnoses among children already known to services rose modestly while estimated previously unidentified cases declined, consistent with improved detection redistributing rather than clearly increasing the measured total. The same-day result supports the narrowly framed finding for these cohorts and this standardized age-seven design.
Speaker
Young Shin Kim, Xiao Liu, Joshua Chang and colleagues
Context
The authors of a peer-reviewed population study used the same community screening and diagnostic framework across birth cohorts to examine whether age-seven autism occurrence changed over time.
Publisher
JAMA Pediatrics · Peer-reviewed repeated population screening study
Only 881 of 2,077 screen-positive children invited for a full assessment participated, so estimates for nonparticipants required statistical imputation using limited observed information. Screening participation also varied, subgroup estimates were imprecise, and the research covers one South Korean setting, older birth cohorts and assessment only through age seven. It cannot establish a worldwide or current biological incidence trend, explain changes in administrative diagnosis counts elsewhere or predict an individual's diagnosis. Diagnosis and support remain based on an individual's developmental history and needs.
SupportedADHD
NHS community prescribing in England for central nervous system stimulants and ADHD drugs rose about 12% in April to June 2026 compared with the previous quarter.
The official NHSBSA tables record 1,309,903 dispensed items in the relevant British National Formulary section in April to June 2026, up 12.1% from 1,168,980 in January to March. The number of identified patients rose 11.6%, from 354,479 to 395,455. The article's rounded quarter-on-quarter claim therefore accurately reflects the published England community-dispensing data.
Speaker
Corrinne Burns
Context
A pharmacy news report used newly released NHS Business Services Authority statistics to describe continued growth in prescribing and differences in recorded ADHD diagnosis rates by age and sex.
Publisher
The Pharmaceutical Journal · Established pharmacy reporting materially republishing official statistics
A dispensed item is not the same as one prescription, one patient, a new diagnosis or a measure of treatment benefit. The BNF section includes several medicines that can have indications beyond ADHD, and the statistics exclude secondary care, prisons and private prescribers. A single quarterly comparison can reflect seasonality, stock or dispensing practice and cannot establish why use changed. The figures are a population-service measure, not guidance to begin, stop or alter an individual's medication.
SupportedAutism
Among California singleton births in 2015–2019, a 10°F higher pregnancy-average heat index was associated with 14% higher odds of an autism diagnosis in developmental-services records.
The study analyzed 2,013,431 singleton births, including 37,083 children with an autism record. A 10°F higher pregnancy-average heat index was associated with 14% higher adjusted odds of autism (odds ratio 1.14, 95% confidence interval 1.12 to 1.17). Minimum temperature, used as a proxy for nighttime heat, showed similar positive associations, while maximum temperature showed little evidence of association. Models adjusted for sociodemographic factors and fine-particle pollution, and results were robust to additional ozone and seasonal adjustments. An earlier independent Southern California healthcare cohort of 294,937 births also found higher autism risk estimates for extreme nighttime heat during early and late pregnancy. Together, those records support the stated population-level association, not a causal effect.
Speaker
Karl O’Sharkey, Bibiana Martinez and colleagues
Context
The authors of a peer-reviewed statewide linked-record birth-cohort study reported adjusted associations between prenatal heat exposure and autism recorded through California's Department of Developmental Services, with stronger signals for sustained and nighttime heat.
Publisher
Environmental Health Perspectives · Peer-reviewed population-based retrospective birth cohort
Both studies were retrospective and observational. Outdoor weather estimates assigned to residential locations cannot measure personal heat exposure, indoor temperature, cooling access or physiological response, and residual confounding, residential mobility, diagnostic access and selection into developmental-services records may affect the estimates. A 2026 systematic review found the wider human evidence on early-life heat and later development limited and methodologically heterogeneous. The result cannot predict an individual child's diagnosis or show that reducing heat exposure prevents autism. Pregnant people should follow public-health heat guidance and seek qualified clinical advice about personal risks rather than infer a care change from this study.
UnresolvedAuDHD
Children with AuDHD have a characteristic strengths profile combining creativity and zest associated with ADHD with attention to detail and honesty associated with autism.
Parents in a Dutch registry described diverse strengths among 143 children aged five to 16 with diagnosed or suspected ADHD, autism or AuDHD. Creativity and zest were prominent themes in the ADHD descriptions, while attention to detail and honesty were prominent in the autism descriptions; the authors said AuDHD descriptions reflected themes from both. That is a useful descriptive account of this sample, but it does not establish a stable, group-specific strengths profile.
Speaker
Lessa Madelief Schippers, Amy De Roubaix, Sander Begeer, Corina Greven and Martine Hoogman
Context
A non-peer-reviewed preprint reported a qualitative analysis of parents' free-text descriptions of strengths in children with diagnosed or suspected ADHD, autism or both and described the AuDHD themes as a combination of the other groups.
This is a small qualitative preprint based on parent-written descriptions, not a validated comparison of abilities. Some children had suspected rather than confirmed diagnoses, subgroup sizes were limited, there was no non-neurodivergent comparator or objective strengths measure, and coding and cultural context can shape which themes appear. The result requires peer review and replication with child or self-report and clearer diagnostic groups. It should not be used as a diagnostic shortcut or to assume an individual's strengths, support needs or experience.
UnresolvedAutism
A four-marker blood panel has potential to detect autism after achieving 97.6% sensitivity and 100% specificity.
The study compared plasma measurements from 54 autistic children with 37 age-matched healthy controls. MPGES-1, 8-isoprostane and cPLA2 were elevated while MRCC-I was not; a logistic-regression combination reported an area under the ROC curve of 0.976, 97.6% sensitivity and 100% specificity in this sample. The panel may be useful for further research, but these apparent results do not establish a clinically accurate blood test.
Speaker
Afaf El-Ansary, Hanan A. Alfawaz and colleagues
Context
The authors of a peer-reviewed case-control biomarker study reported near-perfect discrimination from a combined panel of MRCC-I, MPGES-1, 8-isoprostane and cPLA2 and proposed the panel as a potential autism-detection tool.
Publisher
Scientific Reports · Peer-reviewed case-control biomarker study
This was a small, selected case-control comparison rather than prospective testing in children referred for diagnostic assessment. The accessible report does not describe independent external validation, calibration or proof that the panel distinguishes autism from other developmental or medical conditions. Recent meta-analysis finds group-level inflammatory blood differences but says individual diagnosis and disease specificity remain unestablished. Current diagnosis relies on developmental history and behavior, not a blood test; no individual assessment or care decision should rely on this research result.
MisleadingADHD
Undiagnosed ADHD in a child impacts parenting stress, coparenting quality and family quality of life.
Among 253 Canadian and American parents, families whose child had no ADHD diagnosis but crossed the study's symptom-scale cutoff reported more parenting stress and poorer coparenting and family-quality-of-life outcomes than the no-diagnosis, below-cutoff group. That supports an association between parent-reported elevated symptoms and family strain, consistent with wider evidence linking diagnosed childhood ADHD and symptom severity to parenting stress. It does not establish that those children had ADHD or that ADHD caused the family outcomes.
Speaker
Lara Penner-Goeke and colleagues
Context
The conclusion of a peer-reviewed online parent survey described children without an ADHD diagnosis who crossed a symptom-scale threshold as having undiagnosed ADHD and said the condition impacts several family processes.
A rating-scale cutoff is not a diagnosis, and the cross-sectional survey cannot determine direction of effect. The same parent reported the child's symptoms and family outcomes, so shared-reporter effects, parental mental health, co-occurring child difficulties and other family circumstances may contribute to the associations. ADHD diagnosis requires a full clinical and psychosocial assessment, developmental history, impairment and evidence across settings. Families concerned about a child's symptoms or family strain should seek qualified assessment and support rather than infer a diagnosis from this study.
UnresolvedAutism
Prenatal exposure to the plasticiser DEHP may contribute to autism- and ADHD-related symptoms in early childhood through changes in DNA methylation.
In the Barwon Infant Study, prenatal DEHP estimates were available for 847 children and the mediation sample included at most 589. Each doubling of estimated exposure was significantly associated with one of six symptom outcomes, while directions were weaker but broadly consistent for the others; the top 2% exposure group, only 12 children in the mediation sample, showed larger associations with three outcomes. The authors found that a methylation-profile score and a 531-gene co-methylation network statistically mediated some associations with parent-reported autism- and ADHD-related symptoms at ages two and four. Related methylation patterns were tested in small independent blood and postmortem-brain datasets, but those datasets did not reproduce the full exposure-to-methylation-to-symptoms chain. The study therefore supplies an association and a plausible candidate mechanism, not a settled causal effect.
Speaker
Samuel Tanner and colleagues
Context
A Florey Institute news release, republished with independent expert reaction, said that higher prenatal exposure to DEHP may contribute to autism and ADHD and proposed cord-blood DNA methylation as a biological pathway.
Publisher
Australian Science Media Centre · Institutional research release with independent expert reaction
This was one observational birth cohort, DEHP exposure was estimated from a single urine sample at 36 weeks, the outcomes were symptom questionnaires rather than diagnoses, and the high-exposure analysis rested on a very small subgroup. Causal mediation requires strong no-unmeasured-confounding assumptions, and indirect-effect tests across six related outcomes were not adjusted for multiple comparisons. External checks were small, covered autism but not ADHD diagnosis, and did not test the complete proposed pathway; several sex-stratified effects appeared only in males. An independent expert described the effects as modest, noted opposite blood-versus-brain directions, and stressed that no single exposure is likely to materially determine an individual child's risk. The result does not justify predicting a child's diagnosis or changing prenatal care without qualified clinical advice.
MisleadingAutism
People with autism in the United States have a life expectancy 14 years shorter than the general population.
The linked study analyzed 2,048,046 Medicaid beneficiaries with an autism diagnosis across all 50 states and Washington, DC, from 2000 through 2020. It estimated life expectancy at birth at 64.9 years (95% confidence interval 64.6 to 65.1), 13.8 years below the U.S. general population but 5.6 years below the overall Medicaid population. The 14-year comparison is accurate for this Medicaid study population, and wider reviews support elevated mortality among autistic people, but the headline generalizes a selected insurance population to every autistic person in the United States.
Speaker
Columbia University Mailman School of Public Health
Context
A university news release said autistic people in the United States face a 14-year life-expectancy deficit, based on a newly published study of Medicaid beneficiaries with an autism diagnosis.
Publisher
Columbia University Mailman School of Public Health · Institutional research news release
Medicaid is means-tested and also covers many people with disabilities, so autistic beneficiaries may differ from autistic people with other or no insurance in socioeconomic circumstances, disability, co-occurring conditions and access to care. The sample was disproportionately young and male, and period life expectancy is a population estimate built from age-specific death rates, not a prediction of an individual's lifespan or an average age at death. Comparisons with a general population also cannot isolate autism from poverty, disability, health-care access or co-occurring conditions. Personal health concerns require qualified clinical advice.
SupportedADHD
From 2012 to 2022, Danish children diagnosed with ADHD or autism became more similar to comparison peers across measured prediagnostic characteristics, especially for ADHD.
Among 2,194,951 Danish children and adolescents, 71,317 people diagnosed with ADHD or autism from 2012 through 2022 were matched to 713,170 controls. Odds ratios for all 19 prediagnostic characteristics moved toward the null over time; for example, the low-birth-weight association fell from 1.54 in 2012–2013 to 1.17 in 2020–2022. Attenuation was greater for ADHD than autism and for diagnoses at ages 10–17. Population studies in Sweden and Australia have also found that diagnosis growth outpaced change in measured ADHD traits, so the study's narrowly stated temporal finding is supported.
Speaker
Magnus Elias Tarp and colleagues
Context
A peer-reviewed nationwide registry study reported that associations between 19 family, socioeconomic, birth, perinatal and health-care characteristics and a later ADHD or autism diagnosis weakened across diagnosis years, and suggested that changing diagnostic practice or health-care capacity may contribute to rising diagnosis rates.
Publisher
JAMA Psychiatry · Peer-reviewed nationwide matched registry study
The registry measured changing associations with selected proxies, not core symptom severity, functional need or a biological incidence rate. Coding, referral, ascertainment, family help-seeking and access to services may all change over time, and the observational design cannot identify how much each explanation contributed or rule out changes in underlying risk. The result applies to Danish under-18 diagnoses and should not be generalized automatically to adults or other health systems. It does not show that ADHD or autism is less impairing for an individual or that any diagnosis is invalid; assessment and support remain clinical decisions based on the person's history and needs.
MisleadingAutism
More green-space contact during mid-to-late pregnancy protects against autism and should be recommended by prenatal-care providers.
Among 207 autistic children and 621 age- and sex-matched controls recruited through one maternal and child health hospital in central China, each interquartile-range increase in satellite-measured residential greenness across the identified window was associated with 46.6% lower odds of autism (odds ratio 0.534, 95% confidence interval 0.415 to 0.688). Other observational cohorts also report inverse associations, but their effect sizes, spatial buffers and suggested timing differ. The new study supports an association in its sample; it does not establish that green-space contact prevented autism or that a prenatal clinical recommendation follows.
Speaker
Jun Li and colleagues
Context
A peer-reviewed matched case-control study called residential greenness protective, identified months six to one before birth as a critical window and suggested that prenatal-care providers advise pregnant women to increase green-space contact.
Publisher
Environmental Pollution · Peer-reviewed observational environmental-health study
The retrospective design estimated greenness around a recorded home address rather than a person's actual contact with green space and cannot eliminate selection, mobility, socioeconomic, health-care or other residual confounding. Matching only age and sex does not make the exposure causal, and the data-selected window needs prospective replication. Green space may support general well-being, but this result is not a way to predict or prevent an individual's autism and should not create blame or alter prenatal care without qualified clinical guidance.
UnresolvedAutism
Electroconvulsive therapy substantially improves catatonia, aggression and self-injury in autistic patients, with uncommon adverse events.
The study reports substantial improvement in both groups: all autistic patients with a recorded global-improvement outcome responded, 76.9% had at least a 50% reduction on the Bush–Francis Catatonia Rating Scale, and self-injury among autistic patients fell from 42.9% to 17.1%. Autistic patients received a median of 35 sessions over 342 days, compared with 14 sessions over 97 days for non-autistic patients; adjusted analysis attributed that difference to longer maintenance rather than a higher treatment rate. Those results strengthen an existing clinical signal but cannot by themselves establish how much improvement ECT caused or how uncommon harms are in broader practice.
Speaker
Joshua Ryan Smith and colleagues
Context
A peer-reviewed single-center observational study compared 43 autistic and 67 non-autistic patients treated with electroconvulsive therapy for catatonia and argued that the autistic group needs sustained access because its treatment courses were longer.
Publisher
Autism Research · Peer-reviewed observational treatment study
Every participant received ECT at one specialist center, there was no untreated or alternative-treatment comparison, and group differences in age, sex and intellectual disability required statistical adjustment. Treatment selection, concurrent care, incomplete outcome records and retrospective adverse-event documentation can bias both benefit and safety estimates. Consensus guidance already considers benzodiazepines and/or ECT for moderate to severe catatonia in autistic people, but the autism-specific evidence base remains mostly observational and earlier systematic review evidence was low quality. Catatonia can be medically serious; diagnosis, consent, ECT and maintenance decisions require specialist clinical assessment and cannot be inferred from this study for an individual.
SupportedADHD
U.S. counties with 1 µg/m³ higher average fine-particle pollution over 2005–2015 had childhood ADHD prevalence about 0.31 percentage points higher.
Across 3,106 U.S. counties, the study's spatial model estimated a 0.312 percentage-point increase in childhood ADHD prevalence for each additional 1 µg/m³ in mean PM2.5 exposure over 2005–2015 (standard error 0.058; p<0.001). Alternative exposure periods and adjusted models produced a consistent county-level pattern. Recent meta-analyses also report associations between particulate pollution and ADHD risk, so the narrowly stated ecological association is supported.
Speaker
R. Alexander Bentley and Larisa Ozeryansky
Context
A peer-reviewed ecological study linked long-term county-level PM2.5 estimates with small-area childhood ADHD prevalence estimates and described the association as robust after socioeconomic, demographic, state and spatial adjustments.
Publisher
Scientific Reports · Peer-reviewed ecological observational study
This does not show that pollution caused ADHD in an individual child. The outcome is a modeled county estimate based on about 70,000 children aged 5–17 in the 2016–2018 National Survey of Children's Health, not linked person-level diagnoses and exposures. County averages can conceal within-county differences, migration and unmeasured factors such as health-care access, and the design cannot separate prenatal from postnatal exposure windows. The authors explicitly describe the result as non-causal; it is population evidence, not a diagnostic or treatment tool.
MisleadingAutism
A new, largest-ever autism genetics study identified 65 risk genes, including 28 with very high confidence.
The underlying 2015 Neuron analysis did report 65 genes at a false-discovery-rate threshold of 0.1, including 28 at the stricter 0.01 threshold, plus six copy-number-variant regions for 71 loci. The same UCSF story on its legacy site is dated 28 September 2015. Later studies analyzed far larger cohorts: a 2020 exome study examined 35,584 samples and identified 102 associated genes, and a 2022 analysis included 42,607 autism cases. The genetics result is real, but a 2026 date paired with “new” and “largest-ever” makes an eleven-year-old benchmark look current.
Speaker
UCSF Department of Psychiatry and Behavioral Sciences
Context
A university news page dated 7 September 2026 presents the work as a new and current largest-ever genomic analysis, even though its own body says the study was published in September 2015 and an older UCSF copy carries the original 2015 date.
Publisher
University of California, San Francisco · Institutional research news page with a newly displayed date
The public page may reflect a content-migration or metadata error rather than a deliberate republication. Gene counts and cohort sizes are not directly interchangeable across variant classes, statistical thresholds and study designs, so later work does not invalidate the 2015 findings. The verdict concerns the current dating and present-tense framing, not whether every reported locus remains scientifically relevant. Population genetic associations cannot diagnose autism or predict an individual's outcome.
UnresolvedADHD
Existing ADHD diagnostic frameworks appear unable to distinguish neurodevelopmental ADHD from attentional difficulties caused by the modern information environment.
The editorial identifies a real evidence gap: ADHD has no diagnostic biomarker, adult assessment often depends on retrospective childhood history, and ordinary clinical frameworks do not measure digital-exposure intensity in detail. But it reports no diagnostic-validation study showing that current assessments systematically confuse the proposed environmental presentation with ADHD. Formal guidance specifically tries to separate ADHD from temporary or alternative explanations by requiring childhood onset, persistent impairment in multiple settings, a full developmental and psychiatric history, observer evidence where possible, and assessment of other conditions and circumstances.
Speaker
David Rhys Alchin
Context
A British Journal of Psychiatry guest editorial proposed “environmentally mediated attentional dysregulation” as a distinct explanation for some adults' disabling attention difficulties and argued that current symptom-based assessment cannot reliably separate it from childhood-onset ADHD.
Publisher
The British Journal of Psychiatry · Guest editorial in a peer-reviewed medical journal
“Environmentally mediated attentional dysregulation” is a proposed formulation, not a recognized diagnosis with validated criteria. Experiments showing that reduced mobile-internet access can improve sustained attention do not test whether people were misdiagnosed with ADHD, and a systematic review found the evidence on apparent adult-onset ADHD too weak to distinguish true late onset, missed childhood symptoms and mimics. The absence of a biomarker limits causal certainty but does not by itself show that careful clinical differential assessment is unable to distinguish presentations. Diagnosis and treatment decisions require a qualified clinician and should not be inferred from digital exposure or a response to medication.
UnresolvedAutism
NeuroMimicNet achieves higher accuracy and faster responses than conventional deep-learning models and has clinical relevance for early autism screening in children.
The paper reports 98.92% accuracy with a 70:30 split and 99.3% with an 80:20 split, plus lower inference time than four comparison models. Those figures cannot presently establish the public claim. The paper describes its only dataset as about 3,000 audio and facial-image samples in two diagnostic categories, while its classifier and confusion matrices use four severity classes without explaining how those labels were created. The letter also calculates different implied dataset totals from the two confusion matrices and notes that the abstract says EEG datasets were used even though the methods omit EEG and the conclusion lists it as future work.
Speaker
Poornima S and Elakya R
Context
A peer-reviewed computational paper published on 14 August reported near-perfect performance for a multimodal audio-and-facial-image classifier. A peer-reviewed methodological letter published on 6 September materially resurfaced the claim and requested clarification or correction of internal contradictions that affect reproducibility and interpretation.
Publisher
Journal of Multidisciplinary Healthcare · Peer-reviewed computational study with subsequent methodological letter
The concerns are documented in a scholarly letter, not yet an editorial correction, retraction or author response, so they do not prove that the numerical results are false. The paper also alternates between ResNet50 and ResNet15 and does not provide independent external or prospective clinical validation. Systematic reviews find promising AI classification results but recurring risks from unclear patient selection, inconsistent methods, limited external validation and weak generalizability. A screening model does not diagnose autism, and no individual assessment or care decision should rely on this research result; current clinical guidance requires developmental history, professional observation and a thorough evaluation after a positive screen.
MisleadingADHD
Regular monthly research-team follow-up maintains comprehensive treatment gains better than follow-up as needed in the community for children with ADHD.
The study randomized 186 children aged 6 to 13 only after all had completed the initial intensive program. Improvements were maintained over two years in both follow-up groups. The community group had worse parent-rated hyperactivity or impulsivity on one Conners scale and worse results in some teacher reports, but the abstract does not report broad superiority for monthly research-team follow-up across the other assessed domains. Those limited differences do not support a general claim that monthly specialist follow-up maintained the overall gains better than follow-up as needed.
Speaker
Lily Hechtman and colleagues
Context
A newly published prospective study compared monthly research-team follow-up with usual community follow-up after every child first completed the same three-month program of optimized long-acting stimulant medication, weekly parent training, and child social-skills and academic training.
Publisher
Journal of Attention Disorders · Peer-reviewed randomized prospective follow-up study
The randomized phase can compare the two follow-up arrangements, but it cannot isolate the effect of the initial three-month program because every participant received it. The accessible record does not report absolute effect sizes, confidence intervals, attrition, adherence, medication exposure during follow-up or adverse events, and community care varied with local availability. The sample used DSM-IV combined-type ADHD and does not establish a fixed monthly schedule for other presentations, ages or settings. Current guidance recommends follow-up according to clinical need and structured monitoring of benefits and adverse effects, not one universal interval. Medication and follow-up decisions require an individual plan with a qualified clinician.
UnresolvedAutism
An AI analysis of blood metabolites can support autism diagnosis in young children referred for developmental concerns with more than 80% accuracy.
The study enrolled 140 children aged 18 to 60 months from diagnostic waitlists at two developmental clinics: 114 received an autism diagnosis and 26 had other developmental delays. In internal validation, the best five-feature model reported 87.3% leave-one-out balanced accuracy and 80.0% mean balanced accuracy across repeated train-test splits. That shows potentially useful discrimination within this dataset, but it does not establish how accurately the model will perform in new clinics or populations.
Speaker
Halil Arici and colleagues
Context
The authors of a peer-reviewed prospective diagnostic-accuracy study described a metabolite panel and machine-learning classifier as a clinically relevant step toward a blood-based test that could support autism diagnosis.
Publisher
Annals of Biomedical Engineering · Peer-reviewed prospective observational diagnostic-accuracy study
This is one small, highly imbalanced study with only 26 non-autistic comparison participants. Feature construction and model selection were performed on the same dataset, and there was no independent external or prospective clinical validation, predefined deployment threshold, calibration analysis or evidence that using the model improves care. The registered study planned 200 participants, remains listed with unknown status and has no posted results. Several authors work for the assay developer or hold related intellectual property. Current clinical guidance says autism is diagnosed from developmental history and behavior, not a blood test; this research result is not a diagnostic service or a basis for individual medical decisions.
MisleadingAutism
Pre- and probiotic interventions are generally safe and, in most studies, effective for gastrointestinal symptoms in autistic people, although their effects on core autism symptoms are unclear.
The review found gastrointestinal improvement in many included studies and inconsistent behavioral effects. A 2024 meta-analysis of six randomized trials likewise found a modest reduction in gastrointestinal symptom scores but no significant improvement across autism-related behavioral measures. However, a broader 2026 meta-analysis found no statistically significant overall gastrointestinal or behavioral benefit. That conflicting pooled evidence supports possible gastrointestinal benefit in some settings, not the general claim that these interventions are effective in most cases.
Speaker
Marcela França Dias and colleagues
Context
The authors of a peer-reviewed systematic review synthesized 15 studies involving 1,015 participants and presented microbiota-targeted interventions as a potentially useful adjunct for gastrointestinal difficulties associated with autism.
Studies vary substantially in probiotic strains, prebiotic products, treatment duration, diet, outcome measures and design, and most participants were children. Limited adverse-event reporting cannot establish broad safety; U.S. clinical guidance notes rare serious infections in people with major underlying medical problems. The evidence does not show that these products treat autism itself or identify a reliable product, dose or subgroup. Gastrointestinal symptoms and supplement use should be discussed with a qualified clinician, particularly when a person is medically vulnerable.
UnresolvedADHD
White-matter structural-connectivity deviations can serve as a developmental biomarker for ADHD in young people and help predict atomoxetine treatment response.
Across a large longitudinal developmental dataset and an independent cohort, the study found ADHD-associated deviations concentrated in higher-order association connections. Within-person decreases over two years tracked symptom improvement, and baseline deviations predicted response after 12 weeks of atomoxetine but not methylphenidate. These are substantial, prospectively relevant findings, but they establish performance inside the reported research datasets rather than a clinically validated biomarker.
Speaker
Xiaoyu Xu and colleagues
Context
The authors of a peer-reviewed imaging study described individualized deviations from age-related brain-connectivity patterns as a biomarker with potential relevance for tracking symptoms and selecting treatment in young people with ADHD.
Publisher
Nature Biomedical Engineering · Peer-reviewed primary neuroimaging study
This is one primary study. The independent cohort replicated developmental patterns, but the treatment-prediction result still needs external prospective replication, predefined clinical thresholds, calibration and evidence that it improves care beyond standard assessment. Earlier diffusion-imaging evidence is heterogeneous, with many low-quality studies and no robust pediatric case-control result in high-quality sensitivity analyses. MRI is not currently a diagnostic test for ADHD or a basis for choosing an individual's medication; assessment and treatment require qualified clinical guidance.
UnresolvedADHD
About 4% of South African adults — up to two million people — have ADHD.
A 2023 umbrella review estimated adult ADHD prevalence at 3.10% worldwide, while a 2021 global meta-analysis estimated 2.58% for persistent adult ADHD and 6.76% for symptomatic adult ADHD. Four percent is therefore plausible as an extrapolation, and applying it to South Africa's current adult population produces an order of magnitude near the stated total. But South African clinical guidance says the country's specific adult ADHD prevalence is unknown; its earlier estimate of 3% to 5% among people aged 20 to 50 was itself extrapolated from international evidence rather than measured in a representative national survey.
Speaker
Stellenbosch Business School
Context
A university news release previewing the Southern African Multidisciplinary ADHD Congress paired this country-level estimate with the statement that 60% to 70% of children with ADHD continue to experience symptoms in adulthood.
Publisher
Stellenbosch Business School · University institutional news release
The release does not identify a South African prevalence study, define “adult,” specify whether 4% means persistent diagnosed ADHD or current symptoms, or show the denominator used for “up to two million.” International pooled rates vary by definition, age and study method and cannot establish a country-specific rate. This population estimate is not a diagnosis or a measure of treatment need; individual assessment requires a qualified clinician.
SupportedAutism
Research on autistic women's experiences of menopause remains sparse, and emerging evidence suggests the transition may intensify sensory, emotional, sleep and executive-function difficulties for some.
A prospectively registered mixed-methods systematic review found only eight peer-reviewed studies and seven relevant grey-literature sources. It reported limited quantitative evidence, no evaluated interventions and recurring accounts of increased sensory sensitivity, sleep disruption, mental-health strain, fatigue, cognitive difficulty and reduced capacity to mask or maintain usual activities. A 2026 clinical review likewise says there is no empirical best-practice evidence specific to pharmaceutical or psychological management for autistic people or ADHDers in menopause.
Speaker
Dr Claire Agathou
Context
The GP with a specialist interest in women's health told HuffPost UK that autism and menopause are two historically neglected research areas and described the possible additional difficulties cautiously, stressing that menopause does not affect every autistic person in the same way.
Publisher
HuffPost UK via Yahoo News New Zealand · Established health reporting with identifiable medical commentary
The available studies are small and heterogeneous, largely involve online-recruited autistic people assigned female at birth, include few non-autistic comparison groups and have limited ethnic and intellectual-disability representation. They support the cautious words “may” and “for some,” not a universal effect or a distinct menopause syndrome. Personal symptoms and treatment, including hormone therapy, require individualized advice from a qualified clinician using current menopause guidance.
SupportedAutism
Behavioral sleep interventions are generally effective for improving at least some sleep outcomes in autistic children.
The review reported improvement in at least some sleep outcomes in 95% of included studies. Its cautious core proposition is supported by randomized-trial syntheses: a 2019 meta-analysis found improvements in sleep duration, sleep onset, sleep efficiency and parent-rated sleep problems, and a later meta-analysis of 11 randomized trials found significant improvements across objective and parent-reported sleep measures. American Academy of Neurology guidance also recommends sleep-habit and behavioral strategies as a first-line approach after assessing medicines and coexisting conditions that may contribute to the sleep problem.
Speaker
Maureen O. Mazurek and colleagues
Context
The authors of a peer-reviewed systematic review of 39 quantitative studies concluded that behavioral sleep interventions have consistent overall benefit for autistic children, while emphasizing weaknesses in much of the evidence base.
The finding does not mean every intervention works for every child or sleep problem. Fifty-one percent of studies in the new review were rated weak quality and only 23% strong; interventions, delivery methods, samples and outcomes varied, and some evidence excludes children with intellectual disability. Behavioral sleep support does not treat autism itself. Persistent sleep problems and any medicine or supplement decision require individualized assessment and qualified clinical guidance.
SupportedAuDHD
About 40% of autistic people have ADHD.
The cited 63-study meta-analysis estimated current ADHD prevalence among autistic people at 38.5% (95% CI 34.0% to 43.2%) and lifetime prevalence at 40.2% (34.9% to 45.7%), supporting the rounded proposition. A larger 2023 systematic review estimated ADHD point prevalence at 37% (28% to 46%). An earlier review requiring confirmed clinical diagnoses produced a lower pooled estimate of 28% (25% to 32%), showing that the exact rate depends on methods and populations.
Speaker
Rebecca Ready, PhD, ABPP
Context
The University of Massachusetts Amherst professor and clinical neuropsychologist presented the figure as a key point in an explainer about the recognized co-occurrence of autism and ADHD and the unofficial community term “AuDHD.”
This is a group-level pooled estimate, not a universal probability for every autistic person and not the prevalence of a separate formal “AuDHD” diagnosis. Estimates vary with age, intellectual disability, country, recruitment setting, diagnostic criteria and whether studies measure current or lifetime diagnoses; screening traits are not themselves a diagnosis. Individual assessment requires a qualified clinician.
UnresolvedADHD
One English integrated care board's monthly spending on ADHD and autism rose from £146,000 to £1.5 million in one year, an increase of about tenfold.
The reported endpoints imply a 10.27-fold increase, so the arithmetic matches the rounded proposition. But the statement does not identify the integrated care board, publish its invoices or accounts, define whether the two periods and service categories are like for like, or provide a denominator. Established health-finance reporting repeats the figure but does not independently verify it.
Speaker
NHS Alliance
Context
The health-system membership body used the unnamed board's figures to argue that assessment and support costs were becoming unsustainable and that England needs a national tariff and service specification.
Publisher
NHS Alliance · Health-system representative body statement
This is an attributable but presently uncheckable example supplied by an unnamed board. It is spending, not prevalence, incidence, clinical need, diagnostic quality or proof that demand alone caused the increase; provider mix, prices, service scope and accounting changes could also affect the comparison.
MisleadingAutism
Partner-assisted text and letter-based communication approaches used by nonspeaking autistic people have meaningful benefit, and the lack of evidence should not be treated as evidence that the methods are invalid.
Independent typing and evidence-based augmentative and alternative communication can be valid and beneficial. The plan's broad wording, however, does not separate those methods from facilitator-dependent supported typing, letterboarding or Rapid Prompting Method. Controlled authorship research and professional reviews find facilitator influence in facilitated communication and insufficient evidence that Rapid Prompting Method messages are authored independently. Treating that record as mere absence of evidence materially understates the contrary evidence and the risk of misattributing a partner's words.
Speaker
Interagency Autism Coordinating Committee strategic plan
Context
Medical reporting on the committee's 27 August vote materially republished an official plan that groups text and letter methods involving a trained communication partner, urges continued access while research develops, and says absence of evidence is not proof of invalidity.
Publisher
STAT · Established medical reporting of an adopted federal advisory plan
The plan does not name or define every method, calls for more research, and says communication access does not require federal endorsement of a specific technique. The verdict therefore does not apply to independent typing or validated individualized AAC, and it does not assess any nonspeaking person's abilities. Communication assessment should use qualified clinical guidance and methods that establish independent authorship.
SupportedADHD
Complaints to England's health ombudsman about ADHD and autism services rose from 410 to 1,257, more than tripling in five years.
The Ombudsman's report records 410 complaints relating to ADHD and autism services in 2021–22 and 1,257 in 2025–26. The increase is about 206.6%, so the count is just over three times the starting level and supports the normalized proposition.
Speaker
The BMJ, reporting Parliamentary and Health Service Ombudsman figures
Context
A BMJ news report materially republished the Ombudsman's new casework figures while covering waits, fragmented care and private-treatment costs.
Publisher
The BMJ · Established medical reporting of an official ombudsman report
These are complaints received by the Ombudsman, not diagnoses, prevalence, all NHS incidents or a rate adjusted for the number of people using services. The combined category cannot show how the rise divides between ADHD and autism, and the figures alone do not establish why complaints increased or the quality of care for every patient.
MisleadingAuDHD
Neurodivergent people are 4.3 times more likely to have a structural connective-tissue disorder and 3.4 times more likely to have a hypermobility disorder.
The source analysis did find higher recorded connective-tissue diagnoses in all three neurodevelopmental groups, strongest among patients with both autism and ADHD. But the article turns increases of 438% and 340% into 4.3-fold and 3.4-fold estimates rather than about 5.38-fold and 4.4-fold, applies those peak co-occurrence estimates to a generic “neurodivergent individual,” and transposes several subgroup percentages in its table.
Speaker
ADDitude
Context
A health-news article generalized the largest estimates from an electronic-health-record analysis of people aged 15 and older with autism, ADHD or both.
The source is a descriptive, non-peer-reviewed analysis of diagnosis codes, not a test of cause or direction. The disorders remained uncommon in absolute terms, residual differences in health-care contact may inflate associations, and the results cannot determine whether an individual should be assessed or treated.
FalseAutism
Dividing routine childhood vaccinations into five smaller appointments will substantially lower autism rates.
High-quality population evidence and systematic reviews do not support vaccines as a cause of autism, including evidence on combined MMR and on greater immunologic exposure from several vaccines given together. The American Academy of Pediatrics says there is no medical reason to delay or space routine vaccines and that doing so prolongs the period without protection.
Speaker
President Donald Trump
Context
At a White House education event, he told families to seek five “small little shots” rather than one “big, massive shot” and predicted a very large difference in autism rates.
Publisher
The White House · Presidential remarks (official video)
No study can test every imagined alternative schedule, but there is no credible evidence that splitting recommended doses reduces autism. Alternative schedules may leave children susceptible to preventable infection; vaccination decisions should follow current qualified clinical guidance.
SupportedAutism
One in 50 children in Ontario is diagnosed with autism.
The Public Health Agency of Canada's 2019 survey estimated diagnosed autism at 2.1% among Ontario children and youth aged 1 to 17, or about 1 in 48. That is consistent with the event page's rounded 1-in-50 proposition.
Speaker
Hamilton Ride4Autism
Context
The Dundas Lions Club and Autism Ontario used the figure on the public page for their 23 August 2026 Hamilton motorcycle fundraiser.
Publisher
Hamilton Ride4Autism · Community organization event page
This is a 2019 household-survey estimate, not a current census or an incidence rate. Its 95% confidence interval was 1.9% to 2.4%, and the survey excluded children and youth on First Nations reserves, in foster homes and in institutions.
SupportedAuDHD
AuDHD is community shorthand, not a formal diagnosis; a person can be diagnosed with both autism and ADHD.
AuDHD is not a standalone diagnostic category. Current diagnostic practice permits autism and ADHD to be diagnosed together, so the term can accurately describe that co-occurrence in ordinary language.
Speaker
AuDHD Australia
Context
The public explainer defines the community term and distinguishes it from two co-occurring clinical diagnoses.
Using the term for oneself is not a clinical assessment. Definitions and assessment standards come from the two recognized diagnoses, not from a separate AuDHD test.
UnresolvedAuDHD
The autism-and-ADHD combination behaves differently from either condition alone.
Co-occurrence and interacting traits are well recognized, and exploratory research reports group differences. But “behaves differently” is broad: there is no separate AuDHD diagnostic entity or universally established profile that applies to everyone with both diagnoses.
Speaker
AuDHD Australia
Context
The explainer says the community needed a name because the combination behaves differently from either diagnosis alone.
Early studies use different samples and measures. More replicated research is needed before a distinct, general AuDHD profile can be treated as established.
MisleadingAuDHD
ADHD medication works the same way for people who are also autistic.
The same ADHD medicines may be considered and can reduce ADHD symptoms in autistic people. However, the evidence base is smaller, response and tolerability can differ, and guidelines call for slower titration and closer monitoring when neurodevelopmental conditions coexist.
Speaker
AuDHD Australia
Context
The explainer says co-occurrence changes strategy more than medicine and that ADHD medication “works the same way.”
Individual response varies substantially. Medication choice, dosing and monitoring require a prescriber who can consider the person's full clinical picture.
FalseADHD
ADHD is not a medical condition and may instead be a social construct.
ADHD is a clinically defined neurodevelopmental disorder in international diagnostic systems. Diagnosis depends on a persistent pattern of symptoms, impairment, developmental history and differential assessment; it does not require a single scan or laboratory biomarker.
Speaker
Dr Iona Heath, former President of the Royal College of General Practitioners
Context
Channel 4 promoted The Great ADHD Myth? with Dr Heath's statement that ADHD is “certainly not a medical condition” and framed the programme around whether ADHD is genuine.
Diagnostic practice and service quality can be debated, and other conditions or environments can produce similar difficulties. Those issues do not make ADHD itself fictitious.
MisleadingADHD
NHS data show a 200% increase in referrals for an ADHD diagnosis between 2020 and 2025.
The linked NHS publication supports a substantial recent rise, including a 20.7% year-on-year increase in new referrals in September 2025. It does not present the quoted 2020–2025 comparison, and it warns that data from February 2025 are not directly comparable with earlier months because new providers increased recorded open referrals.
Speaker
Channel 4 Press
Context
The promotional release for The Great ADHD Myth? attributed the figure to NHS England's November 2025 management information.
A consistent longer-run series could support a different percentage, but the cited record does not let readers verify this exact 200% figure on a like-for-like basis.
FalseAutism
The statement “vaccines do not cause autism” is not evidence-based because infant vaccines have not been ruled out.
High-quality population evidence and systematic reviews do not support a causal link between vaccines and autism. Requiring research to eliminate every hypothetical possibility reverses the burden of proof and does not make the evidence-based conclusion unsupported.
Speaker
U.S. Centers for Disease Control and Prevention
Context
The CDC's Autism and Vaccines page used the absence of absolute proof against every possible vaccine mechanism to reject the established conclusion.
Science rarely proves a universal negative with absolute certainty; the verdict concerns what the best available causal evidence supports, not whether every imaginable mechanism has been tested.
Important limit
This is evidence review, not medical advice
The record assesses published propositions. It cannot diagnose autism or ADHD, determine whether “AuDHD” fits an individual, or recommend treatment. Speak with a qualified health professional about personal assessment, medication, pregnancy or vaccination decisions.
