Created:INDEX.md (18 canonical pages), wiki/overview.md (draft, 26 verified PMIDs), OPEN-QUESTIONS.md (7 seed questions), and this log.
Searches run (live PubMed E-utilities esearch/esummary): child/adult prevalence and persistence; historical diagnosis/nosology reviews; AAP guideline; genetics/GWAS and neurobiology; medication network meta-analyses, guanfacine, MTA and adult CBT; cardiovascular safety, growth and mortality; girls/women and adult underdiagnosis; environmental/peripheral biomarkers; STARS digital therapeutic.
Verification: every cited PMID was retrieved live from PubMed in this session. The required ADD/1987 clarification is sourced to the retrieved lifespan diagnosis review rather than memory alone.
Next: full build queued per CONDITIONS-ROADMAP.md; preserve all-presentation and lifespan scope.
Built: all 18 canonical wiki pages as draft; each is 150–158 lines with 32 live-PubMed-verified records, a quantitative synthesis, evidence map, open questions, related canonical links and full references.
Literature layer: 454-paper deduplicated BIBLIOGRAPHY.md; six landmark notes (Demontis GWAS, MTA, Cortese medication network, Kessler adult prevalence, Kollins digital trial, Hinshaw female lifespan review); worldwide guideline registry with supersession chains and watch list; 84-row statistics sheet; patient-voice method, seven verified organizations, 12 multi-source themes and annotated sources.
Frontier: replaced seed questions with 24 stable tiered questions and 14 “Dots not yet connected” junctions.
Index/roadmap: added five reading paths, real artifact counts and page statuses; condition status changed from seeded to built in CONDITIONS-ROADMAP.md.
PubMed searches: 144 live E-utilities esearch queries (eight per page): the original six topic searches plus two gap-oriented searches for each domain, spanning nosology, assessment, prevalence/course, genetics, neurobiology, cognition, pediatric and adult care, girls/women/reproductive stages, comorbidity, medications, psychosocial/digital care, long-term outcomes, guidelines, biomarkers, trials, patient experience and safety. Live efetch retrieved 817 records, including a fresh fetch attempt for all 26 seed PMIDs; exact esummary rechecks resolved six cross-layer records not represented as PubmedArticle objects in the XML corpus.
Citation integrity check: 454 unique PMIDs occur anywhere in the condition and exactly the same 454 appear in the bibliography. Every one was returned by live PubMed E-utilities in this session. Every wiki body PMID is present in that page’s References section.
Official web sources verified: NICE NG87, AADPA/NHMRC Australian guideline, CADDRA 4.1, AWMF 028-045, CHADD, ADDA, ADHD Europe, CADDAC, ADHD Ireland, ADHD Australia and ADHD Foundation UK.
Could not verify cleanly: ADHD Foundation UK’s site remains live and describes services but also lists a 2025 closure notice; the organization is retained with an explicit status caveat. No clearly patient-led national ADHD organization with an accessible official site was verified for Africa, Asia or Latin America during this build; this is recorded as a coverage gap, not an absence claim.
Next independent audit: re-fetch and check all 454 unique PMIDs and 576 page reference uses against claims; scrutinize 2025–2026 papers, evidence-map relevance, quantitative transcription, guideline currency, the ADHD Foundation UK status caveat, and the negative organization-coverage finding. Promote pages to curated only after that separate pass.
Scope completed: audited all 18 wiki pages and all 13 literature artifacts (31 Markdown files), including quantitative tables, evidence maps, landmark notes, the bibliography, guideline registry, statistics sheet, patient-voice materials and open-question register.
PubMed verification: re-fetched all 454 inherited unique PMIDs through live PubMed E-utilities and resolved every record. Gap searches identified one additional relevant 2026 multicentre diagnostic-classifier study (PMID 42448769), which was fetched live, added to the biomarker synthesis and bibliography, and used to narrow the corresponding evidence-gap statement. The final condition corpus contains 455 unique PMIDs, all resolved live in this audit session; there are 2,050 PMID occurrences overall and 1,363 in the wiki layer.
Claim-to-citation audit: checked each wiki claim/evidence-map/reference pairing and the literature-layer uses against the live records. Corrected 29 discrete metadata or content fields: 18 author or collective-author attributions, three incomplete citation titles, two incorrect DOI fields, two growth-effect direction labels, and four stale or overbroad guideline/biomarker/organization-status statements. This included two references attributed to the wrong lead author. No cited PMID remains unresolved, no wiki body PMID is absent from its page reference list, and bibliography metadata now agrees with the live PubMed records.
Quantitative audit: checked the principal prevalence, diagnostic-performance, medication-effect, growth, cardiovascular, mortality, injury, suicide, misuse and digital-treatment estimates against the retrieved records. The two growth standardized mean differences are now explicitly labelled as magnitudes of reduction; no other material numerical transcription error remained.
ClinicalTrials.gov verification: individually re-queried all 11 cited NCT records through the live v2 API. All resolved and their current recruitment statuses matched the condition text; the STARS study record also matched the stated completed status, actual enrolment of 348 and two-arm design. The corpus contains 37 NCT occurrences.
Evidence-gap searches: re-ran live PubMed searches for current diagnostic/treatment biomarkers, prospective response prediction and a US adult guideline, plus 14 junction-specific searches behind “Dots not yet connected.” Replaced absolute non-existence language with dated, database-bounded positive evidence-gap statements. The current ocular-classifier evidence is acknowledged while its lack of prospective decision-impact validation is stated explicitly.
Guideline and patient-voice currency: verified the current NICE, CADDRA, German S3 and US adult-guideline positions from live official sources. Updated the German S3 entry to its 2026 version 2.0, recorded that the US adult national guideline remains in development, classified ADHD Foundation UK as historical after its 30 August 2025 closure, and added verified current organizations in Peru and Taiwan. The Africa finding remains a dated search-coverage gap, not an absence claim.
Validation: all 18 pages meet the required structure and density (150–158 lines; 32–33 references), all are now status: curated, all internal Markdown links resolve, and there are zero [unverified] flags, zero unresolved citation identifiers and zero unresolved substantive audit issues.
Result: the ADHD condition passed the independent audit and is eligible for roadmap status audited.