JAMA Pediatrics

Objective: To compare healthcare access among U.S. children with attention-deficit/hyperactivity disorder (ADHD) only, autism only, and co-occurring ADHD-autism, and to identify social inequities associated with unmet care using recent nationally representative data. Methods: We analyzed combined 2023-2024 National Survey of Childrens Health data for 13,576 children aged 3-17 years with current ADHD and/or autism. Six healthcare-access outcomes were examined: forgone healthcare, unmet mental health care, no medical home, coverage gap or inadequate insurance, unmet specialist care, and unmet care coordination. Survey-weighted modified Poisson regression estimated adjusted prevalence ratios (aPRs) and 95 percent confidence intervals (CIs). Six multiply imputed poverty measures were combined using Rubins rules. Sensitivity models additionally adjusted for functional difficulty. Results: Compared with ADHD only, co-occurring ADHD-autism was associated with higher prevalence of forgone healthcare (aPR, 1.73; 95 percent CI, 1.38-2.16), no medical home (1.11; 1.02-1.20), inadequate or discontinuous insurance (1.14; 1.03-1.28), unmet specialist care (1.49; 1.14-1.94), and unmet care coordination (1.15; 1.04-1.28). Autism only was associated with unmet mental health care (1.49; 1.13-1.97) and care coordination (1.13; 1.02-1.25). Food insufficiency, housing instability, and non-supportive neighborhood conditions were consistently associated with poorer access. After adjustment for functional difficulty, associations remained for co-occurring ADHD-autism with forgone and specialist care and for autism only with unmet mental health care. Conclusions: Healthcare access inequities differed by neurodevelopmental diagnostic group and social context. Pediatric systems should strengthen care coordination, specialty and mental health access, insurance continuity, family navigation, and responses to food and housing hardship. Keywords: attention-deficit/hyperactivity disorder; autism spectrum disorder; healthcare access; health disparities; social determinants of health

ObjectivesTo determine whether single-layer evaluation characterises clinical risk in the final note of a multilingual ambient AI scribe, and where serious errors arise. DesignTwo-arm evaluation on one common clinical-risk scale, using a frozen, reference-aligned synthetic corpus. SettingScripted consultations spanning 77 languages and five complexity levels, from simple general practice to expert multidisciplinary-team handover. Main outcome measuresPer-session incidence of at least one serious (HIGH or CRITICAL) note-layer error in intrinsic (reference script to note, n=385) and end-to-end (automatic speech recognition (ASR) transcript to note, n=2,302) generation; three LLM raters classified discrepancies using a four-mode taxonomy. ResultsIntrinsic note generation had a 3.4% serious-error rate, with no detectable gradient across complexity (1% to 6%; Cochran-Armitage p=0.42) or language resource (high 2%, medium 5%, low 4%). End-to-end serious errors rose steeply with complexity (L1 1% to L4/L5 24%) and language scarcity (high 7%, medium 10%, low 15%; OR 1.54 per tier, 95% CI 1.15-2.07; p=0.004, GEE clustered on language). Of serious in-note errors, 87% were ASR-derived and 7% note-originated; amplification exceeded correction threefold (fate entropy 1.19 bits). Repeated generation showed moderate reproducibility (Fleiss kappa 0.42); word error rate explained 24% of cross-language variance versus [~]1% for transcription risk density. ConclusionsLow serious-error rates in individual layers did not preclude higher end-to-end note risk. Evaluation should therefore include the clinician-facing note, because component-level metrics cannot fully capture risk created or transformed at the transcription-to-generation interface. What is already known on this topicO_LIAmbient AI scribes are being deployed globally, and safety evaluation has characterised the transcription layer in isolation, most commonly with word error rate C_LIO_LIFrequency-based transcription metrics such as word error rate do not directly encode clinical consequence, and component-level performance may not represent the risk of the final generated note C_LIO_LILarge language models produce clinically relevant output stochastically, so identical prompts can yield materially different content C_LI What this study addsO_LINote generation from a perfect transcript had a low observed serious-error rate and no detectable complexity or language-resource gradient, whereas the end-to-end system showed steep gradients on both axes. C_LIO_LISerious note-layer errors were predominantly ASR-derived, but note generation transformed upstream errors variably, amplifying serious errors about three times more often than it corrected them C_LIO_LITranscription-layer metrics had limited predictive value for note-layer risk after accounting for complexity and language-level dependence; word error rate explained approximately one-quarter of cross-language variance C_LI How this study might affect research, practice or policyO_LIEvaluation and procurement of ambient scribes should include end-to-end assessment of the clinician-facing note, with results stratified by consultation complexity and language where relevant C_LIO_LIA three-surface root-cause alignment using a four-mode root-cause taxonomy (note-originated, propagated-ASR, amplified-ASR and silent-correction) can attribute documentation errors to their point of origin or transformation C_LIO_LILanguage-related performance should remain a surveillance target because disparities can emerge after components are composed, even when corresponding gradients are not detectable in an individual layer C_LI

Background Population-level administrative data cohorts hold huge potential for generating evidence to improve child health. Understanding the quality of these data is critical to enabling us to fully realise this potential. We aimed to evaluate linkage between children's health, education and social care administrative records (ECHILD) to identify possible sources of bias. Methods We created four cohorts to understand the drivers of inclusion/exclusion within ECHILD. For each cohort, we calculated the linkage rate by counting the total number of eligible individuals who appeared in the ECHILD linkage spine and any of the corresponding hospital/school component datasets. We then assessed whether linkage rates varied according to sociodemographic characteristics (year of birth, sex, ethnicity, region). Results ECHILD currently captures 34.2 million patients and 25.3 million pupils born between September 1984 and March 2023, aged 0-38 years. Over 90% of children with hospital birth records linked to a school record (rising to 93% for those born from 2014 onwards); 88% of children attending school in 2001/02 linked to a patient record (rising to 96% for those born in 2019/20), resulting in a total of 21 million linked individuals. Substantial variation existed between sociodemographic groups, with those living in London, residing in more deprived areas, or with recorded ethnicity other than White being less likely to be included. Conclusions Population-level datasets such as ECHILD often under-represent specific groups. Irrespective of the mechanisms by which individuals are excluded (e.g. opt outs, linkage errors or not interacting with services), those in minoritised ethnic groups and those living in more deprived areas are most affected. Linkage evaluations are critical for understanding who is and who is not included in analyses of these data so that researchers can account for potential sources of bias within analyses.

Objective Large language models (LLMs) are increasingly used to generate, summarize, and evaluate clinical documentation, yet standardized methods for evaluating nuanced, context-dependent clinical language remain limited. Existing approaches rely primarily on annotated clinical corpora or keyword-based methods, restricting reproducibility, public dissemination, and systematic comparison across models. We present ClinStigma, a clinician-guided framework for constructing, annotating, and evaluating synthetic benchmarks of contextual stigmatizing language, demonstrated through ClinStigma-SUD, its inaugural implementation for substance use disorder documentation. Materials and Methods Candidate clinical sentences were generated using a frontier language model and iteratively reviewed according to a clinician-developed annotation rubric refined through expert review in stigmatizing language. The resulting benchmark intentionally incorporates explicit stigmatizing, neutral, and contextually ambiguous ("gray-zone") documentation requiring clinical reasoning rather than keyword recognition. Three locally deployed open-weight LLMs were evaluated using zero-shot inference, followed by external evaluation using independently annotated clinical documentation from MIMIC-IV. Results ClinStigma-SUD consists of 500 clinician-adjudicated synthetic clinical documentation sentences. Inter-rater agreement was 98.8% across the full benchmark (Cohen's {kappa} = 0.976) and 94.0% within the intentionally ambiguous subset ({kappa} = 0.880). Under zero-shot prompting, binary accuracy ranged from 84.8% to 91.8% across evaluated models, while F1 for the ambiguous (gray) class ranged from 0.038 to 0.423. Zero-shot external evaluation on 3,716 annotated examples from the independently annotated Harrigian MIMIC-IV corpus yielded 84.4% accuracy (84.4% balanced accuracy; macro F1 0.824), against a 69.9% majority-class baseline. Stigmatizing-class F1 was 0.879 in the substance-related subgroup and 0.883 in the full external corpus. Discussion and Conclusion ClinStigma establishes a reproducible clinician-guided framework for developing synthetic evaluation benchmarks for contextual clinical language. By combining structured clinical expertise, explicit annotation guidance, synthetic benchmark generation, and external evaluation, the framework enables open, privacy-preserving evaluation of language models while providing a scalable foundation for benchmark development across additional clinical specialties, including obstetrics, emergency medicine, and mental health.

Individuals with autism spectrum disorder (ASD) and co-occurring mood or psychotic disorders face substantially elevated suicide-attempt risk, but prior research is largely limited to population-level incidence and cross-sectional risk-factor studies, leaving unclear whether individual-level, prospective prediction is possible in this population. We conducted a retrospective prediction model study using Epic Cosmos electronic health record (EHR) data from 322,879 individuals with ASD and co-occurring mood or psychotic disorders to predict suicide attempts at the next clinical encounter using only information available at or before the preceding encounter. Five machine-learning algorithms were compared, with performance evaluated across discrimination, calibration, and clinical utility. All algorithms achieved strong discrimination using the full predictor set (AUROC 0.87-0.91). A parsimonious feature set (22 predictors) matched or exceeded full-model performance, with extreme gradient boosting (XGBoost) achieving the highest discrimination and standardized net benefit post-ablation (AUROC = 0.90; AUPRC = 0.34), though net benefit was comparable for elastic net and logistic regression. A prior suicide attempt was the strongest individual predictor. Decision-curve analysis indicated that model-guided decisions provided greater net benefit than treating all or no individuals across a broad range of clinically plausible thresholds. These findings demonstrate that near-term suicide-attempt risk can be predicted with high decision utility in this high-risk population using data already captured in routine care, suggesting a feasible basis for future EHR-based clinical decision-support tools.

Objective: Emotion dysregulation and irritability occur across pediatric disorders, complicating distinctions between general symptom burden and specific clinical presentations. Person-centered analyses can look at profiles of scores to identify groups with similar presentations. We examined whether modeling shared and domain-specific symptom variance identified differentiated profiles in youth. Method: Caregivers of 2,187 US youth ages 6-18 completed 87 items assessing emotion dysregulation, depression, mania-related symptoms, sleep-related impairment, and attention and behavioral problems. An ordinal bifactor model estimated one general and 13 specific factors. Latent profile analyses used standardized scores on all 14 factors, with an expanded raw-score analysis as a comparator. We examined caregiver-reported diagnoses and functioning and approximately two-week classification agreement among 166 recontacted caregivers. Results: The general factor accounted for 62% of common variance ({omega}H=.94), although model fit was mixed (CFI=.884; RMSEA=.058; SRMR=.092). A provisional four-class solution distinguished population-modal (67.4%), internalizing (11.8%), attention/hyperactivity (11.0%), and unusual-thought/elevated-mood (9.8%) profiles. Entropy was .83, but model enumeration and initialization checks did not establish a unique solution. Profiles showed overlapping diagnostic and functional correlates. Recontact agreement was 78.9% ({kappa}=.60). Irritability-specific reliability was limited, constraining interpretation of its weak profile differentiation. Conclusions: General symptom burden coexisted with differentiated residual symptom patterns. These findings support evaluating severity alongside symptom configuration, without equating profiles with diagnoses or weak residual irritability signals with limited clinical relevance. Measurement refinement, model-selection verification, and confirmation of classification-error-adjusted associations are needed before clinical application, but profiles could simplify clinical interpretation of sets of scales.

Purpose: This study examined whether word-based and non-word natural language sampling (NLS) measures complement standardized language scores in characterizing expressive communication among preschool children with autism with language impairment (AUT-LI), Down syndrome (DS), and fragile X syndrome (FXS). Method: Participants included 82 preschool children with AUT-LI (n = 32), DS (n = 34), or FXS (n = 16). Expressive communication was characterized using the Preschool Language Scales and 10-minute natural language samples collected during a standardized play-based protocol. NLS measures included traditional word-based metrics, such as lexical diversity, intelligibility, and word approximations, as well as tiered non-word vocalization measures, including non-speech, single-phoneme, and multi-phoneme vocalizations. Analyses examined age associations, diagnostic group differences, between group differences in variability, and exploratory communication profiles derived using principal component analysis and partitioning around medoids clustering. Results: Diagnostic groups did not significantly differ on standardized expressive language scores. Children with FXS showed higher performance than the AUT-LI and DS groups on several word-based NLS measures, whereas non-word vocalization measures primarily revealed differences in within-group variability. Word-based NLS measures were consistently associated with standardized expressive language skills, while associations for non-word vocalizations varied by vocalizations type. Cluster-validity indices factored a two-cluster solution. A hypothesis-informed three cluster solution that included non-word vocalizations suggested Higher Language, Lower Output, and High Output-Low Structure profiles. Conclusions: Standardized and NLS measures provide complementary descriptions of expressive communication in children with neurodevelopmental disorders. Non-word vocalization coding may be especially useful for describing clinical meaningful variation among children with limited expressive language. Keywords: Natural language sampling; expressive language; neurodevelopmental disorders; autism spectrum disorder; Down syndrome; Fragile X syndrome

As internet and mobile phone access increases, digital health interventions are increasingly used to support remote and low-resource populations. In many settings, the health impact of digital interventions is constrained by limited or short-lived engagement. We tested if conditional mobile data credit incentives increase digital intervention engagement within a digital intervention trial. This nested cluster randomized controlled trial was conducted with caregivers of children 3-9 months at baseline. Of 72 neighborhood clusters from the treatment arm of the parent trial (evaluating a digital early childhood development intervention), 20 were randomly assigned to treatment and 52 as controls. Eligible participants had a phone provider through which mobile data credits could be most affordably administered. Treated participants received around 250 MB mobile data credits every two weeks, conditional upon using the digital intervention at least once in the same period. Credits were digitally administered via a third-party service. Digital intervention engagement (primary outcome) was monitored continuously until endline data collection approximately 2 years after assignment. Secondary outcomes included caregivers views, behaviors, and early childhood development. We used intent-to-treat analyses to assess all outcomes. This trial was registered as NCT05202106 at clinicaltrials.gov and was completed in 2025. Participants were recruited between September 2021 and March 2023 (N Treatment = 147; N Control = 333). Conditional mobile data credits extended engagement with the digital early childhood development intervention by 6.50 weeks on average (treatment mean = 26.19 weeks, control mean = 19.69; p=0.011). No significant impacts were identified on secondary outcomes. As such, mobile data credits may be a cost-effective way for increasing caregiver engagement with digital health intervention in low resource settings.

Background: The first thousand days are a sensitive window for child health and social health inequalities. When routine care falls short, parents of young children turn to online sources and, recently, to generative artificial intelligence (AI). Evidence on their concerns, child-health literacy, and AI use is scarce. Objective: To describe health informational needs and concerns, child-health literacy, and use of AI for child-health questions, including differences by parental role and parity, among parents of young children. Methods: In this cross-sectional study, two anonymous self-administered online questionnaires were disseminated through a digital health application between May 12 and June 11, 2026, to French-speaking parents of at least one child aged 0 to 47 months in France. Questionnaire 1 covered unmet information needs, concerns across 16 topics, perceived intra-couple concordance, and confidence in routine care. Questionnaire 2 covered a four-item child-health literacy quiz and AI use (frequency, topics, 12-month trend, barriers, trust conditions, perceived effect on consultations, and disclosure). Results: In total, 12,634 parents responded (Questionnaire 1, n=5,410; Questionnaire 2, n=7,224), mostly mothers (94.6% and 92.9%). Most (61.8%) had searched for information without finding a clear, reliable answer, selecting a median of 4 (Interquartile range [IQR] 3-7) over 16 topics. Leading concerns were sleep (51.3%), nutrition (48.1%), and education (i.e., upbringing and behavior) (47.8%). Mothers were nearly twice as likely as fathers to select accident prevention (26.5% vs 16.3%) and parental mental health (38.7% vs 28.6%); first-time parents selected most topics more often than experienced parents, most markedly dental hygiene (43.8% vs 27.6%). The median child-health literacy score was 3 (IQR 3-4) on a 0-to-4 quiz, with 19.9% of parents scoring 2 or below; tooth-brushing age was the weakest item (68.4% correct). Mothers outscored fathers on every item. More than half of parents (54.2%) had used AI for a child-health question, 11.2% weekly; among those reporting a trend, 68.3% had increased AI use for child-health questions over the year and 6.2% decreased it. The main barriers to AI use were a preference for professionals (55.9%) and reliability concerns (48.7%); trust depended most on medical oversight of the tool (76.6%). AI use appeared additive, with 66.8% reporting no change in healthcare visit frequency. One third (35.5%) had told an AI something shared with neither a professional nor their network, rising to 64.1% among regular users. Conclusions: Parental informational needs were widespread and frequently unmet. AI has become a common and fast-rising source for child-health questions, at times a confidential one, used alongside rather than instead of professionals. Child-health literacy was generally high but uneven, with oral health the clearest gap. Strengths include the sample size and early-generative-AI timing; limitations include the cross-sectional design, self-report, digital-recruitment selection bias, and the small proportion of fathers.

Infected individuals drive infectious disease transmission, yet empirical data on social interactions during acute infection remain limited. We conducted a prospective longitudinal study of 1,000 medically attended acute gastroenteritis and acute respiratory infection cases and 709 household members in the northwestern United States from 2024 to 2025 to quantify illness-associated changes in social contacts and their implications for transmission modeling. Cases substantially reduced social contacts during peak illness; weighted mean number of contacts increased 2.0-fold (95% confidence interval, 1.8-2.2) over the two-week follow-up. In transmission modeling, the simulated outbreak trajectory and estimated effectiveness of interventions differed substantially between models that did and did not account for temporal reductions in social contacts during acute infection. These findings show that behavioral responses during acute infection could substantially influence transmission dynamics and intervention evaluation, and provide an empirical framework for incorporating illness-associated behavioral change into infectious disease models.

Background Autism spectrum disorder (ASD) shows wide phenotypic variation, from adults without intellectual disability and with low support needs to minimally verbal children with intellectual disability and severe behavioral symptoms. Genetic studies of this heterogeneity have focused mainly on high-impact de novo variants and polygenic scores; here we investigate, at the molecular level, the role of common variants. Materials and methods We analyzed genotypic and phenotypic data from 9,332 individuals with ASD in the SPARK cohort. We considered 31 traits: six latent factors derived from the Social Communication Questionnaire and the Repetitive Behavior Scale-Revised, 22 co-occurring diagnoses, parent-reported IQ range, age at diagnosis, and age at first walking. We performed within-case genome-wide association studies (GWAS) and, for the 10 traits with evidence of SNP heritability, association analyses with genetically predicted gene and protein expression, splicing, and enhancer activity in brain and blood (xWAS). Nominated genes were tested for enrichment in biological processes, brain cell types, and developmental stages. xWAS-derived transcriptional signatures were compared with chemical perturbation signatures from CMap/LINCS L1000 to prioritize candidate drugs. Results GWAS identified seven genome-wide significant loci for five traits, including loci near GABRG3 and TENM2 associated with co-occurring sleep, feeding or toileting problems. xWAS identified seven gene-trait associations involving three phenotypes (FDR < 0.1), including CACNA1G with communication skills and NEGR1 with learning disability. GWAS and xWAS nominated different genes which however converged on synaptic transmission and gated channel activity, with preferential expression in neurons during later stages of brain development. PARP and p38 MAPK inhibitors were predicted to reverse the signatures of both social-communication and repetitive-behavior phenotypes, whereas several compound classes, including histone deacetylase inhibitors and mTOR inhibitors, were predicted to reverse the signatures of some phenotypes while mimicking those of others. Conclusion Combining within-case GWAS with xWAS nominates candidate genes mediating common-variant effects on ASD heterogeneity, provides biological insight, and suggests candidates for phenotypically stratified drug repositioning.

Background: Valid and feasible measures of preschoolers' 24-hour movement behaviours are needed for population-level research. We developed a Japanese version of the Movement Behaviour Questionnaire-Child (MBQ-C) and evaluated its measurement properties and practicality. Methods: After translation and cross-cultural adaptation following the COSMIN procedure, including cognitive interviews, we evaluated the questionnaire in a cross-sectional validation study of Japanese children aged 3-6 years. Concurrent validity was assessed against hip-worn accelerometry and a 7-day diary; we also examined test-retest reliability, measurement error, agreement between response formats and modes, and practicality (completion time, burden, and clarity). Analyses used Spearman correlations, intraclass correlation coefficients, the SEM, smallest detectable change (SDC), and Wilcoxon signed-rank tests. Results: Of 208 children, 170 provided valid accelerometer and diary data for the concurrent-validity analysis, and 187 completed the MBQ-C for the questionnaire-based analyses. Concurrent validity was weak for active and energetic play (Spearman {rho} = 0.12-0.26), strong for total screen time (0.78-0.85), and moderate for total sleep (0.48-0.58). Test-retest reliability was moderate to good (ICC, 0.66-0.81), and within-child variability was lowest for sleep (CV 3%) and highest for the physical activity items (21-33%). The open- and closed-ended versions showed no significant differences; paper-electronic agreement was good for screen time and sleep but poor for active play. Completion took approximately 6-9 minutes, with low reported burden. Findings were robust in sensitivity analyses. At the group level, SDC for a mean was SDCind/{surd}N: 119.8/{surd}N, 74.0/{surd}N, 112.6/{surd}N, and 92.5/{surd}N min/day for active play, energetic play, total screen time, and total sleep, respectively (N = number of children surveyed). The systematic differences between the two test-retest occasions were +15.7, +6.5, +4.0, and -5.4 min/day, respectively. Conclusions: The Japanese MBQ-C is a practical option for assessing preschoolers' active play, screen time, and sleep at the group level. By quantifying its group-level measurement error, this study provides benchmarks for interpreting group-level differences and change, supporting the use of the MBQ-C in surveillance and evaluation of programmes delivered at scale.

Abstract Autistic young adults experience substantial gaps in services supporting community living and participation. The Community Autism Peer Support (CAPS) program is a community-based intervention in which trained autistic peer specialists provide individualized support to autistic young adults pursuing self-determined goals related to community participation and independent living. We conducted a single-site, two-arm randomized controlled trial to evaluate the feasibility of implementing the CAPS program. Forty autistic adults aged 18 to 30 years were randomized to CAPS (n = 20) or services as usual alone (n = 20) and assessed at baseline, 3 months, and 12 months. Study findings showed that CAPS attracted autistic young adults with diverse demographic and diagnostic backgrounds, and most intervention participants remained engaged in services for extended periods. While the study was underpowered to detect statistical differences in outcomes because it was not a main focus of the research, a consistent pattern of results favored the CAPS condition across all outcomes. The largest positive signals were observed for overall and social quality of life, with additional favorable changes in internalized stigma, unmet needs, social functioning, loneliness, and positive autistic identity. Findings provide support for the feasibility and promise of an autistic-delivered peer specialist model focused on community living and participation. The consistent pattern of favorable changes supports continued development of CAPS as a person-centered approach to addressing outcomes important to autistic adults.

Background: Gender-affirming medical care (GAMC) for children and adolescents in Australia has become increasingly politicized, despite care being provided in accordance with latest evidence-based international guidelines. Critics of GAMC claim that current assessment and deliberation processes are insufficient, and that too many young people are accessing medical interventions, no longer identifying as trans, and expressing treatment regret. These claims have led to widespread bans on the provision of paediatric GAMC, despite growing evidence regarding harms caused by such bans. Aim: This study sought to describe the care journeys of the young people attending Australia's largest paediatric gender service, with a focus on reporting the rates at which young people commenced medical interventions, re-identified with their birth-assigned sex, and expressed regret about their treatment. Methods: The Royal Children's Hospital Gender Service (RCHGS) is a statewide service that provides individualised multidisciplinary care to trans and gender diverse young people in Victoria, Australia. An initial extended triage appointment explores the needs and goals of the young person and their family. Some proceed to receive individualised multidisciplinary care (MDC) at RCHGS which can include psychosocial assessment and management, family support, and medical interventions including pubertal suppression (PS) and gender-affirming hormonal therapy (GAHT). We undertook a retrospective file audit of all patients who had attended at least one RCHGS appointment and been discharged between 1/1/2017-31/12/2024. For each person, key data were extracted from their electronic medical record and recorded in REDCap. Results: 1,391 patients were discharged over the 8-year study period and many different care trajectories were observed. 814 (58.5%) did not access ongoing MDC. Of the remaining 577, 353 (25.4% of total cohort) started a medical intervention ((PS: n=64/1391 (4.6%); estradiol: n=97/1391 (7.0%); testosterone: n=244/1391 (17.5%)). For those who started these medical interventions, the median duration and number of sessions from first appointment until commencement of medications was 530 (IQR: 371-693) days and 9 (IQR: 7-13) sessions. 346 of these patients (98.0%) had data on gender identity at discharge and subsequent care pathways. 7 (2.0%) re-identified with their birth-registered sex and stopped treatment: 4 did so after commencing PS (4/61=6.6% of those starting PS), 2 did so after commencing testosterone, and 1 did so after commencing estradiol (3/341=0.9% of those starting GAHT). Just one of these patients, who had been on estradiol for 1 month, expressed treatment regret given concerns about initial breast budding and future fertility. Conclusion: This study highlights varied trajectories of those referred to paediatric gender services, directly challenges claims used to justify restrictions on paediatric GAMC and underscores the importance of a person-centred approach to care.

Objectives: To assess the feasibility and early performance of an AI-assisted screening workflow that reviews inpatient discharge summaries to identify potential treatment injuries for clinician review and, where appropriate, Accident Compensation Corporation (ACC) treatment injury claim consideration in New Zealand. Design: Single-site, six-month pilot reported as a quality-improvement/service-evaluation activity. A large language model (LLM)-enabled system acted as an index screening test over discharge summaries to generate candidate treatment injury signals for structured clinician adjudication and escalation. Setting: A New Zealand district hospital (Wairarapa). Screening operated on routinely produced inpatient discharge summaries. Participants: All inpatient discharge summaries produced during the pilot period were screened (n = 2,642). The unit of analysis was one discharge summary per inpatient episode. Intervention: An LLM-based screening workflow that outputs structured candidate harm signals with supporting excerpts and a certainty label. Outputs were reviewed by clinicians using a structured adjudication template. Suitable cases could be escalated to senior clinicians and followed through the local ACC treatment injury process. Main outcome measures: Primary outcomes were yield, positive predictive value (PPV) of AI- flagged signals for clinician-adjudicated iatrogenic harm, and the rate of unsupported/hallucinated outputs. Secondary outcomes included overlap with incident reporting and administrative coding, downstream ACC claim outcomes, and clinician review workload. Results: Across 2,642 inpatient discharge summaries screened, the AI-assisted workflow generated 237 candidate signals (yield 9.0%; 95% CI 7.9% to 10.1%). Clinician adjudication classified 102 as confirmed iatrogenic harm, 43 as possible harm requiring additional context, 81 as no harm, and 11 as unsupported/hallucinated outputs. PPV for confirmed harm was 43.0% (95% CI 36.8% to 49.4%). PPV for confirmed-or-possible harm was 61.2% (95% CI 54.9% to 67.1%). Seven ACC treatment injury claims were lodged as a direct result of AI-assisted surfacing; three were accepted, two declined, and two were held/pending at analysis. Organisational ACC data identified 68 treatment injury claims lodged during the same period, of which 17 overlapped with AI-identified cases. Conclusions: AI-assisted screening of discharge summaries can feasibly increase detection of potential treatment injuries for structured clinician review and may support more systematic ACC consideration and safety learning. Safe scaling requires transparent reporting of model and prompt details, robust human adjudication and escalation pathways, and governance controls to mitigate false positives and unsupported outputs. A next-stage evaluation should estimate diagnostic accuracy against a chart-review reference standard and examine implementation outcomes including workload, acceptability, fidelity and sustainability.

Background Short-term weather conditions can trigger paediatric hospital admissions, but evidence remains limited on the health relevance of intraday temperature variability, humidity, and precipitation, and generally on which subgroups are most susceptible. Methods We conducted a nationwide case time-series study of Swedish children and adolescents aged 6-18 years using linked register data. Associations between daily mean temperature, intraday temperature variability, relative humidity, and precipitation and inpatient hospitalisations were estimated with Poisson regression (individual fixed effects) combined with distributed lag non-linear models over lags 0-21 days. Outcomes included all-cause hospitalisation and cause-specific admissions. Effect modification was examined by sex, age group, socioeconomic indicators, region, and calendar periods. Findings Among 4,638,247 children and adolescents in the cohort, 1,070,702 (23.1%) experienced at least one inpatient hospitalisation, contributing 1,736,169 hospitalisation events. All-cause hospitalisation increased with higher mean temperature, with risk highest at hot extremes. Intraday temperature variability showed a U-shaped association, with higher rates at both lowest and highest exposure levels. Relative humidity showed no association with hospitalisation risk. Higher precipitation was associated with increased hospitalisation risk, but estimates at high rainfall levels were imprecise. Stratified analyses indicated modest differences by sex and socioeconomic indicators, but clearer heterogeneity by age and region, with the highest risks among younger children and in East Sweden; associations varied across calendar periods and were imprecise for most cause-specific outcomes. Sensitivity analyses mostly showed that the main curve shapes were robust. Interpretation Short-term exposure to hot temperatures, both low and high intraday temperature variability, and higher levels of precipitation were associated with an increased risk of all-cause hospitalisation among children in Sweden, with clearer variation by age group and region than by sex or socioeconomic position. These findings support targeting paediatric climate-health preparedness and early-warning systems toward younger children and higher-risk regions such as eastern Sweden and extending such systems beyond mean temperature to include intraday temperature variability, a distinct, currently overlooked driver of risk that heat-focused systems are not designed to capture. Fundings This research was funded by Formas the Swedish Research Council for Sustainable Development (grant no. 2023-01774 and 2022-01845) and the Swedish Research Council (2025-02804 and 2022-06599).

Objective: To compare the accuracy, cost, and speed of a low-cost, non-generative structured-judgment model and four general-purpose large language models (LLMs) for extracting smoking status, pack-years, and quit date from clinical notes to support lung cancer screening (LCS) clinical decision support (CDS). Materials and Methods: We built a synthetic, shareable benchmark of 3,000 outpatient notes in three conditions: 1,000 template-generated notes, 1,000 realistic, ``messy'' notes written by an LLM from structured facts, and 1,000 ``messy'' notes that required complex arithmetic to determine pack-years and quit dates. Ground truth reference labels were programmatically generated before each note was created. We compared TypeSafe Jev 1.13 with Claude Haiku 4.5, Claude Sonnet 5, GPT-6 Luna and GPT-6 Sol, each using an identical structured output schema. We determined United States Preventive Services Task Force (USPSTF) 2021 and American Cancer Society (ACS) 2023 eligibility from each system's output in code. Results: Jev made the correct eligibility decision for 99.1%, 99.9%, and 94.4% of notes in the three conditions, compared with 97.8%, 98.4%, and 98.1% for Haiku, 100.0%, 99.6%, and 99.8% for Sonnet, 99.7%, 98.8%, and 98.5% for Luna, and 100.0%, 99.8%, and 99.6% for Sol. In the complex condition, Jev produced 27 false positive and 14 false negative screening flags per 1,000 notes, the most of any system. Luna cost $0.12 to $0.21 per 1,000 notes and Jev $0.61 to $0.64, compared with $2.44 to $4.22 for Sol, $3.37 to $4.43 for Haiku, and $3.76 to $6.61 for Sonnet. Jev was the fastest system (median 0.45 to 1.21 seconds per note, compared with 1.71 to 3.50 seconds for the others). Discussion: All models tested had high accuracy. The structured-judgment model had the lowest latency, which could allow for synchronous use in clinical decision support systems rather than batched, cached invocation. The cost of the structured-judgment model was low, but the newest cost-efficient LLM, Luna, was less expensive and more accurate on the complex notes, though its latency was higher. Accuracy, latency, and cost are all moving quickly as new models are released. Conclusion: Fast, low-cost language models could make real-time, note-based CDS practical, and current models are highly accurate. Prices and capabilities are changing rapidly, so continuous benchmarking and system updates are essential.

Multi-agent prompting of large language models has produced contradictory diagnostic results, and it remains unclear whether any benefit comes from specialist personas or from perspective independence. We compared a single direct call, five personas in one context, and the same five roles as isolated agents integrated by a moderator, with five repeat runs per case, on 87 CPC cases, 406 MedCaseReasoning cases, and 364 emergency department encounters, under an LLM judge validated against clinicians. On the external benchmark the team beat the single call on both pre-specified recall endpoints (top-3 +3.0 points, p = 0.0079; top-5 +3.9, p = 3.8 x 10^-5;); a factorial attributes the gain to independent generation plus moderated synthesis, not the specialist roles. On real emergency presentations the benefit reversed (top-1 40.1% versus 34.3%, p < 0.0001), carried by the specialist role lists and surviving added objective results. Deployment should key on the question and the input at hand.

Importance: Conditions that are hard to diagnose, several more common in women, may be missing from electronic health records (EHRs), hiding patients from research cohorts and case finding. Objective: To measure how often the EHR contains a diagnosis code for chronic conditions that participants report, and to compare this with national survey benchmarks. Design, Setting, and Participants: This cross-sectional study used the All of Us Research Program Registered Tier Dataset v9. Participants were adults with EHR visits in both 12-month periods from July 2021 to June 2023 who completed a health history survey. We analyzed the data in October 2026. Exposures: Self-reported clinician diagnosis of 10 conditions with ongoing visits or treatment, as well as sex at birth, race and ethnicity, age, and years of EHR records. Main Outcomes and Measures: Share of participants reporting current care who had a qualifying diagnosis code, alone and relative to hypertension and type 2 diabetes, and the ratio of age-standardized coded shares to national survey prevalence. Results: Of 161,480 respondents, 114,610 (73,870 female [64.5%], median [IQR] age 63 [50-72] years) had at least 5 years of EHR records before the survey. Among those reporting current care, 84.6% (95% CI, 84.2%-85.0%) with hypertension and 87.0% (95% CI, 86.5%-87.6%) with type 2 diabetes had a code, compared with 41.3% (95% CI, 38.9%-43.8%) for endometriosis, 39.4% (95% CI, 34.3%-44.8%) for Lyme disease, and 24.4% (95% CI, 23.2%-25.6%) for chronic fatigue. Counting ICD-9-CM and other mapped codes, 55.3%, 51.5%, and 28.3%, respectively, had any qualifying record. Women had lower coded shares than men for hypertension, type 2 diabetes, obesity, depression, and anxiety (crude ratios 0.93-0.96, within-health-system ratios 0.98-1.02) and higher shares for migraine and fibromyalgia (within health systems, 1.11 [95% CI, 1.07-1.15] and 1.41 [95% CI, 1.29-1.57]). Age-standardized coded shares among active patients were 1.1 to 2.4 times national prevalence of the same self-reported diagnoses. Conclusions and Relevance: In this cross-sectional study, fewer than half of participants reporting ongoing care for chronic fatigue, Lyme disease, or endometriosis had a qualifying ICD-10-CM code. Linking each participant's report to their own record measured this gap directly.

Attention Deficit Hyperactivity Disorder (ADHD) frequently co-occurs with distinct personality profiles, but whether ADHD shapes personality (scar model) or personality influences the development of ADHD (vulnerability model) remains unclear. Leveraging genome-wide association study (GWAS) data, we examined directional associations between ADHD and the Big Five personality traits (Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness). We estimated genetic correlations and conducted bidirectional two-sample Mendelian Randomization (MR) analyses between ADHD and the Big Five. Robustness was evaluated using multiple sensitivity tests, including MR-CAUSE, which accounts for correlated and uncorrelated pleiotropy, and colocalization analyses on fine-mapped variants. Higher genetic liability to Neuroticism and Extraversion, alongside lower liability to Conscientiousness, increased ADHD risk, consistent with the vulnerability model. Conversely, ADHD genetic liability was associated with lower Agreeableness, supporting the scar model. Associations were generally robust across sensitivity analyses, with MR-CAUSE supporting the Neuroticism and Agreeableness associations and colocalization providing variant-level support for associations between ADHD and Neuroticism, Conscientiousness, and Agreeableness. These findings provide early, population-level insight into personality-related vulnerability to ADHD and motivate further work on the role of personality traits in ADHD risk.