Created: four seed files, 18 canonical pages, overview with 25 PMIDs, and 7 seed questions.
Searches run (live PubMed E-utilities esearch/esummary): 2010 criteria and validation; GBD; genetics and immune pathogenesis; TICORA/FIN-RACo; EULAR/ACR guidance and remission; methotrexate; ATTRACT, upadacitinib/abatacept, ORAL Surveillance; difficult-to-treat RA; R4RA; RA-ILD and mortality; patient-reported outcomes; ARIAA/APIPPRA prevention and prevention review.
Verification: every PMID cited was returned live by PubMed. Three recalled treat-to-target/BeSt identifiers returned unrelated non-RA papers and were excluded.
Next: full build queued per CONDITIONS-ROADMAP.md.
Built: all 18 canonical wiki pages as draft (2,800 lines, 30–31 references per page); 122-record master bibliography; six landmark notes; 15-document guideline registry; 56-row statistics sheet; patient-voice method, organization, theme and source files; 29 tiered open questions and 14 “Dots not yet connected” junctions.
PubMed live searches: re-fetched all 25 seed PMIDs; 2010 classification and validation; unclassified-arthritis imaging; GBD 2017/2021 burden and mortality; preclinical RA, ACPA and prevention; HLA/genetics; smoking, oral, lung and gut mucosa; synovial fibroblast/macrophage heterogeneity and single-cell studies; treat-to-target, remission, imaging targets and tapering; methotrexate monotherapy/combination, SWEFOT and glucocorticoids; ATTRACT, PREMIER, TEMPO, ADACTA, AMBITION, REFLEX and comparative biologics; SELECT-COMPARE, SELECT-CHOICE, ORAL Strategy and ORAL Surveillance; JAK malignancy/infection/ILD; difficult-to-treat definitions, phenotypes and rescue therapy; cardiovascular disease, serious infection, fracture, eye disease, vasculitis, pregnancy, cervical spine and septic arthritis; RA-ILD screening, prediction, treatment and guideline appraisal; fatigue, residual symptoms, work and patient-reported outcomes; EULAR, ACR, French, Mexican, Japanese and physical-therapy guidelines; biomarkers, plasma proteomics and R4RA.
ClinicalTrials.gov v2 live queries: RA general landscape; RA prevention; synovial biopsy; RA-ILD; direct terms for ARIAA, StopRA, ORAL Surveillance and other named programs. Thirty unique NCT identifiers now appear in the condition tree; each came from a returned API record in this session.
Public-source verification: fetched NRAS, Arthritis Foundation, Arthritis Society Canada, Arthritis Australia/MyRA, Arthritis Ireland, and EULAR PARE pages on 2026-08-28; recorded only aggregate themes and public service descriptions.
Deliberate choices: retained every page at draft because the playbook requires a separate-session audit before curated; used UNKNOWN exactly as ClinicalTrials.gov returned it and did not infer trial outcome; did not create an NCT identifier for APIPPRA where today’s API search returned none.
Could not verify or complete: no clearly authoritative RA-specific South African national patient-organization page was found in the targeted web search; several trial records had stale UNKNOWN statuses; RA-ILD prevalence was not reduced to one number because screening definitions conflict; no molecular treatment selector is ready for routine use.
Next session: independent audit of every claim→PMID pair, exact citation metadata, numeric transcription and registered-trial status; promote only fully checked pages to curated and then update roadmap to audited.
Deepened all 18 canonical wiki pages in place: preserved the existing verified text and added page-specific quantitative synthesis, including classification sensitivity/specificity and likelihood ratios; global-prevalence heterogeneity; HLA–smoking, silica, periodontal and microbiome estimates; single-cell and spatial synovial biology; biomarker decision thresholds; methotrexate optimization, RACAT, TEAR, CareRA and bridging data; NORD-STAR and biologic safety; FINCH and JAK safety conflicts; CAMERA, BeSt, ARCTIC and TaSER strategy effects; D2T prevalence and inflammatory/non-inflammatory decomposition; cardiovascular, mental-health, pregnancy and malignancy burden; RA-ILD prevalence, prognosis, screening, methotrexate controversy and INBUILD effects; infection, TB, VTE and vaccine-response estimates; prevention successes and null trials; current trial mechanisms; guideline disagreements; and patient preference, sleep, sexual-health and remote-monitoring outcomes.
Density after deepening: pages are 170–185 lines and contain 35–41 unique PMIDs each; clinical-trials-landscape.md contains 29 NCT identifiers. The wiki corpus now contains 245 unique PMIDs, all represented in the master bibliography.
Live PubMed work: ran more than 150 topic-specific E-utilities esearch queries across the 18 page scopes, followed by live esummary metadata retrieval and targeted efetch abstract retrieval for numerical transcription. Re-resolved the 126 pre-existing unique PMIDs in two live esummary batches and resolved every newly used PMID live. Two recalled imaging PMIDs returned unrelated ovarian-cancer and phytoremediation papers; both were rejected, and ARCTIC/TaSER were re-identified by title search as PMIDs 27530741 and 27026689.
Live ClinicalTrials.gov work: re-resolved all 21 pre-existing unique NCT records through API v2 and queried the current RA interventional landscape. Added eight individually re-fetched records: NCT04862117, NCT07137598, NCT07412704, NCT06570798, NCT06687551, NCT06733935, NCT05869955 and NCT07201558. Broad-condition false positives were excluded; multi-disease protocols are labelled as such, and registry status/enrollment are dated 2026-08-30.
Literature layer: expanded BIBLIOGRAPHY.md from 126 to 249 unique PMIDs with cited-by mappings (245 used across wiki pages plus four historical/current guideline-registry records); expanded STATISTICS.md with diagnostic, epidemiologic, strategy-trial, comorbidity, safety, D2T and RA-ILD tables; expanded guidelines/REGISTRY.md with EULAR 2025 supersession, vaccination, reproductive, cardiovascular, lifestyle/work and direct ACR/CHEST screening/treatment documents.
Open questions: sharpened OQ-10 through OQ-12 with newly quantified uncertainty; added OQ-30 through OQ-34 on outcome-changing RA-ILD screening, combined antifibrotic/immunomodulatory treatment, equitable remote monitoring, durable vagus-nerve modulation and preference-concordant care; added four cross-domain junctions D-15 through D-18.
Editorial choices: retained status: draft because the playbook requires a separate-session audit before promotion; reported confidence intervals and denominators where available; treated observational protective associations (especially methotrexate and RA-ILD/mortality) as non-causal; treated pooled null JAK comparisons as unresolved rather than as rebuttals of ORAL Surveillance; and updated the guideline page/registry to recognize EULAR 2025 as superseding EULAR 2022 while preserving the 2022 evidence discussion.
Next check: independent claim-to-abstract audit of all 245 PMIDs and exact journal metadata; verify whether multi-disease cellular-therapy trials have opened RA-specific cohorts; monitor EULAR 2025 implementation, RESET-RA durability and long-term APIPPRA/ARIAA outcomes.
Scope checked: all 18 wiki pages and every literature artifact. The final corpus contains 3,169 wiki lines, 459 substantive claim→PMID pairs outside evidence maps, 789 wiki-body PMID occurrences, 73 numeric/effect-size claim lines, 154 literature-layer PMID occurrences outside the bibliography, 257 unique PMIDs and 29 unique NCT identifiers.
Live verification: re-fetched the original 249 PMIDs in four PubMed efetch batches; retrieved eight focused absence/current-guideline searches and the resulting prevention, RA-ILD and guideline records; directly fetched all 29 ClinicalTrials.gov v2 records; and re-fetched all patient-organization URLs. Final unresolved identifier counts: PMID 0; NCT 0; DOI mismatch across six landmark notes 0.
Citation/reference repairs: rebuilt all 18 wiki reference sections and the 257-entry bibliography from live PubMed metadata; normalized incorrect secondary-author labels; removed nine unused wiki references; corrected the overview tapering question to cite RETRO/TARA rather than remission criteria. Final reference-title/author mismatches: 0; wiki body/reference-set mismatches: 0; duplicate bibliography PMIDs: 0; cited PMIDs missing from bibliography: 0.
Structural repairs: moved 11 evidence-map blocks out of unrelated clinical tables and into their intended sections, then validated every Markdown table and canonical wiki link. Final malformed-table count: 0; misplaced evidence-map rows: 0; noncanonical wiki links: 0.
Stale evidence corrected: incorporated live 2026 StopRA (PMID 40884017), APIPPRA/ALTO (PMID 41576971) and five-year TREAT EARLIER (PMID 42392130) results; replaced vague prevention claims with overall and subgroup effect estimates; dated the lifestyle-prevention evidence gap after fresh searches; and added current APLAR, ALAT, BSR and South African guidance found in the refreshed guideline search.
Stale absence corrected: the prior claim that no accessible South African national patient organization was found was false. The Arthritis Foundation of South Africa site resolved live and was added with its national helpline, education, support and advocacy services.
Final state: all 18 pages promoted to status: curated; [unverified] flags: 0; unresolved substantive issues: 0. ClinicalTrials.gov UNKNOWN labels remain where the live registry itself returns UNKNOWN; they are documented registry state, not unresolved evidence flags.
Post-promotion integrity pass: found and fixed one reference-order mismatch on biomarkers-and-tissue-precision.md after the STRAP update; rechecked all 18 body/reference sequences, reference numbering, Markdown tables, internal links, page density, bibliography coverage, six note DOIs and roadmap/index status. Final errors: 0; pages span 169–187 lines with 35–42 verified references each.
Searches run: 17 topic-specific PubMed E-utilities searches covering classification/diagnosis, epidemiology/burden, preclinical prevention, genetics/environment/mucosa, synovium, treat-to-target/remission, conventional DMARDs, biologics, JAK inhibitors, difficult-to-treat disease, comorbidity, RA-ILD, guidelines, biomarkers, trials, patient-reported outcomes and safety, all restricted to publication dates 2026-08-28 through 2026-08-31; one broad RA title/abstract query returned 16 records. A separate live title query recovered a relevant GBD 2023 LMIC analysis published 2026-08-21 but missed by the previous build. Live abstracts and metadata were fetched for every retained PMID.
PubMed triage: five records survived. Wang et al. quantified GBD 2023 RA burden across 129 LMICs (PMID 42495527); Fernández-Codina et al. compared oral and parenteral early glucocorticoid exposure in 2,222 CATCH participants (PMID 42665531); Dalix et al. reported a TNFA-methylation methotrexate-response model requiring external validation (PMID 42664067); Ikeda et al. reported two-year observational baricitinib outcomes including 88 D2T patients (PMID 42667647); and Sugiyama et al. described five-year CKD progression in 601 patients (PMID 42667508). Case reports, incidental RA mentions, an animal celastrol study, a cross-sectional miRNA diagnostic candidate, a small immune-checkpoint-inhibitor cohort and elbow arthroplasty outcomes did not survive triage. No paper was landmark-level; no deep note was created.
ClinicalTrials.gov: individually re-fetched all 29 cited NCT records through API v2 and compared overall status, enrollment, last update and presence of posted results with the audited 2026-08-30 state. No tracked record changed status or gained/lost posted results. A broader RA update query identified two RA-inclusive early-phase records updated on 2026-08-28: PF-08154225 across four autoimmune diseases (NCT07782450) and MK-1045 across SLE/RA (NCT07363590); both were individually fetched and added with explicit multi-disease/phase-1 caveats.
Guidelines, statistics and patient organizations: official EULAR, ACR, BSR, DGRh/AWMF and targeted web searches found the new German DGRh 2026 S3 DMARD guideline, which replaces the 2018 S2e document and adds specific de-escalation and RA-ILD positions. It was added to literature/guidelines/REGISTRY.md and synthesized in wiki/guidelines.md. The LMIC GBD analysis and four observational studies were added to literature/statistics/STATISTICS.md. Searches of NRAS, Arthritis Foundation, Arthritis Society Canada and German patient-organization material found no post-cutoff campaign or patient-experience source substantial enough to change the patient-voice layer.
Files changed: updated wiki/clinical-trials-landscape.md, wiki/guidelines.md, literature/BIBLIOGRAPHY.md, literature/guidelines/REGISTRY.md, literature/statistics/STATISTICS.md, OPEN-QUESTIONS.md (OQ-12, OQ-21 and OQ-29), and INDEX.md. The two edited wiki pages retain status: curated because every new claim was verified directly against the live registry or official guideline source; both carry last-curated: 2026-08-31.
Next sweep: check whether NCT07782450 or NCT07363590 opens a separately reported RA cohort; monitor the DGRh S3 journal publication and independent appraisal, especially its early antifibrotic/RA-ILD and DMARD de-escalation positions; look for external validation of the TNFA methylation response model; and continue the prior RESET-RA and APIPPRA/ARIAA durability watch.
2026-08-31 — Independent audit of the 2026-08-31 sweep (auditor: Grok; author audited: Codex)¶
Scope: the two wiki pages the sweep edited (clinical-trials-landscape.md, guidelines.md) and every sweep-touched line in BIBLIOGRAPHY.md, STATISTICS.md, guidelines/REGISTRY.md, OPEN-QUESTIONS.md (OQ-12, OQ-21, OQ-29) and INDEX.md. Pages the sweep did not edit were not re-audited; they retain the 2026-08-30 audit.
Live identifier verification (this session): PubMed E-utilities esearch plus efetch for PMIDs 42495527, 42665531, 42664067, 42667647 and 42667508 — 5/5 resolved with matching titles, first authors and abstracts. ClinicalTrials.gov API v2 batch plus individual protocol fetches for all 31 unique NCT IDs on the trials page — 31/31 resolved. Direct retrieval of the DGRh press release and AWMF register 060-004. A targeted PubMed search for a DGRh 2026 S3 journal article returned 0 records, matching the sweep's watch-list stance.
Claim–citation pairs checked: 42. Twenty numeric/qualitative claims from the five new papers against live abstracts; bibliography metadata for those five records; protocol-level claims for NCT07782450 and NCT07363590 (status, phase, enrollment, conditions, RA eligibility); overall-status match for the 29 previously tracked NCT IDs; six DGRh recommendation claims against the official 2026-08-26 press release and AWMF 060-004; INDEX counts (262 unique bibliography PMIDs, 31 NCT IDs, 27 registry documents, 94 statistics rows).
Verified without change: Wang 2026 (PMID 42495527) — 12.6 million prevalent cases in 129 LMICs in 2023 and a 4.1-fold upper-middle- versus low-income prevalence ratio, correctly scoped to LMICs rather than global GBD. Fernández-Codina 2026 (PMID 42665531) — 2,222 CATCH participants; 12-month glucocorticoid use 47% after early oral, 26% after parenteral and 8% after no early glucocorticoid; adjusted OR 9.8 and 4.1 versus no early glucocorticoid; OQ-12 correctly flags confounding by indication. Dalix 2026 (PMID 42664067) — 201 early untreated RA; combined AUC 0.760 versus clinical-only 0.699; correctly not treated as an externally validated selector. Ikeda 2026 (PMID 42667647) — 353 overall / 88 D2T; 24-month retention 58.2% / 42.0%; CDAI remission 36.4% / 22.9%. Sugiyama 2026 (PMID 42667508) — CKD-stage progression in 106/601 over five years. NCT07782450 is NOT_YET_RECRUITING phase 1/2, estimated n=54, RA eligible in parts 1a/1b. NCT07363590 is RECRUITING phase 1, estimated n=21, SLE and RA. All 29 previously listed NCT statuses match the live registry. DGRh S3 replaces the 2018 S2e document (AWMF 060-004); 1–3-month assessment, 3-/6-month adjustment, glucocorticoid stop generally within three months, gradual DMARD reduction without complete withdrawal, and a first dedicated RA-ILD recommendation all match the official release.
Absence re-checks: no PubMed-indexed DGRh 2026 S3 journal publication yet. NRAS/Arthritis Foundation/Arthritis Society Canada material retrieved for 2026 includes pre-cutoff items (SMILE-RA modules February, Healthcare Champion Awards April, Work Without Barriers July) and the annual September RA Awareness Week; nothing post-2026-08-28 required a patient-voice change.
Precision fix (1):guidelines/REGISTRY.md labelled PMID 39389412 as “Daien et al.” Live ESummary returns Fautrel B as first author (Daien C last author), matching the wiki page and bibliography. Corrected the registry row. Pre-existing secondary-author label on a file the sweep edited; not introduced by the five new papers.
Verdict: the sweep was clean. No fabricated identifier, no numeric or directional mismatch, no stale absence, and the two new trial rows are correctly caveated as multi-disease early-phase programs. Promotion: both edited wiki pages remain status: curated. Unresolved substantive issues: 0. [unverified] flags: 0.
Next sweep should also chase: whether NCT07782450 or NCT07363590 opens a separately reported RA cohort; the DGRh S3 journal publication and independent appraisal; external validation of the TNFA methylation model; RESET-RA and APIPPRA/ARIAA durability.