Skip to content

Curation log — major depressive disorder

2026-08-28 — Seed completed

  • Created: OPEN-QUESTIONS.md (8 seed questions) and this log, completing the existing INDEX.md and wiki/overview.md seed set; aligned the canonical table to the required Pages schema by adding a biomarkers-and-treatment-prediction page in place of the generic measurement/outcomes page.
  • Searches run (live PubMed E-utilities esummary): STAR*D original report and reanalysis; antidepressant network meta-analysis, severity analyses, publication bias, and relapse prevention; ketamine, esketamine, dextromethorphan-bupropion, psilocybin, ECT, and accelerated TMS trials; psychotherapy/combination and measurement-based care; CANMAT 2023; epidemiology, global burden, treatment gap, GWAS, serotonin-hypothesis umbrella review, and mortality.
  • Verification: all 24 PMIDs already used by the seed were retrieved from live PubMed in this session before these files were written.
  • Next: full build queued per CONDITIONS-ROADMAP.md — build the remaining 17 wiki pages and literature layer, and independently audit all claim–citation pairs.

2026-08-28 — Full build completed

  • Built: all 18 canonical wiki pages; every page has draft frontmatter, TL;DR, dense body, open questions, canonical related links, and references.
  • Literature layer: topic-grouped BIBLIOGRAPHY.md; five landmark paper notes (STAR*D, Cipriani 2018, Turner 2008, Berman 2000, SNT); guideline registry; statistics tables; and patient-voice method, verified organizations, themes, and annotated sources.
  • Evidence searches: live PubMed E-utilities searches for diagnosis/heterogeneity; epidemiology/burden; monoamines, HPA, BDNF, inflammation and circuits; genetics/GWAS/epigenetics; antidepressant efficacy, maintenance, adverse effects, withdrawal, publication bias and severity; psychotherapy/combination/exercise; TRD definitions and augmentation; ketamine/esketamine/dextromethorphan; ECT/TMS/tDCS/VNS; psychedelics; biomarkers/pharmacogenomics; suicide/mortality; guidelines; treatment preferences, functional recovery, and lived experience. All 110 unique wiki PMIDs were retrieved in this session, including every seed identifier retained in overview.md.
  • Trials: live ClinicalTrials.gov v2 query for MDD studies with recruiting, not-yet-recruiting, active-not-recruiting, enrolling-by-invitation status; selected NCT IDs and status were written only from that response.
  • Patient organizations: direct HTTP fetch verified DBSA, Mind, Beyond Blue, SADAG, Mental Health America, and Depression Alliance on 2026-08-28; ADAA (403) and SANE (no response) were explicitly excluded from the verified table.
  • Integration: replaced OPEN-QUESTIONS.md with 20 stable Tier 1/Tier 2 questions and 12 “Dots not yet connected” junctions; rebuilt INDEX.md; marked the condition built in CONDITIONS-ROADMAP.md.
  • Limitations at that build stage: APA guideline replacement status had not yet been resolved; patient-voice coverage was English-language and high-income weighted; all pages remained draft pending the required separate-session independent citation audit. The APA item and page status were resolved in the 2026-08-30 audit below.

2026-08-30 — Full-corpus deepening

  • Deepened all 18 canonical wiki pages in place: retained the verified seed/build content; added diagnostic-accuracy and conversion cohorts, stratified epidemiology and treatment coverage, immune/HPA/circuit and multi-ancestry mechanism evidence, comparative pharmacologic harms, efficacy estimands, psychotherapy delivery, TRD sequencing, rapid-acting treatments, neuromodulation, novel therapeutics, prediction standards, mortality, international guideline disagreements, trial-design failure modes, patient-defined outcomes, and quantified safety reassessment.
  • Resulting depth: 2,900+ wiki lines; every page is 151–183 lines and carries 25–33 unique live-retrieved PMIDs (383 unique wiki PMIDs total, up from 110). All pages remain draft; this was a build/deepening session, not the separate independent claim audit required for promotion.
  • Live PubMed work: 125 E-utilities searches across 31 query groups (diagnosis, screening, heterogeneity, epidemiology, course, burden, treatment gaps, mechanisms, multi-ancestry genetics, pharmacology/harms/withdrawal, efficacy/reporting bias, psychotherapy, TRD, ketamine/esketamine, neuromodulation, psychedelics/neurosteroids, biomarkers, suicide/mortality, guidelines, trials, experience, and safety). Final exact EFetch in four batches resolved all 386 PMIDs used anywhere in the depression condition; no identifier failed.
  • Live trial work: re-fetched all 27 NCT identifiers used in the condition through exact ClinicalTrials.gov v2 study endpoints and ran a live active-status MDD query returning 100 studies on its first page. Updated the relevant wiki snapshot language to 2026-08-30; registry status remains time-sensitive and does not establish local enrollment or efficacy.
  • Literature integration: expanded BIBLIOGRAPHY.md to cover all 386 PMIDs cited anywhere in the condition with cited-by fields; no condition PMID is missing from the bibliography. Expanded statistics/STATISTICS.md with diagnostic accuracy, longitudinal conversion, population heterogeneity, coverage, genetics, immune findings, treatment effects, adverse effects, and cause-specific mortality—keeping incompatible methods separate.
  • Guidelines: expanded the registry from 9 to 18 records with RANZCP, German NVL/S3, AFPBN/FondaMental, Korean Medication Algorithm, BAP, WFSBP, Japanese Society of Mood Disorders, USPSTF, and international tDCS guidance. Added an explicit international comparison to wiki/guidelines.md; re-fetched NICE NG222 successfully (HTTP 200) on 2026-08-30. Legacy documents are labeled rather than silently treated as current.
  • Research agenda: sharpened withdrawal, inflammatory-selection, and mortality questions with competing quantitative evidence; added OQ-21–OQ-25 on screening pathways, bipolar conversion, patient-weighted medication harms, task-sharing supervision, and accelerated-TMS dismantling; added four new “dots not yet connected.”
  • Mechanical verification: all page frontmatter remains status: draft; all 18 pages satisfy the 150–400-line and 25–60-citation targets; every body PMID has a page reference entry; every final PMID resolved live; all 27 NCT IDs resolved live; and every condition PMID appears in the bibliography.
  • Follow-up for the independent audit: verify every numerical claim against full text where abstracts do not report the estimand; independently reassess review overlap and sponsorship; inspect possible ePub/print year differences; rerun time-sensitive NCT statuses and the APA replacement watch item. The evidence ledgers deliberately preserve design limits but do not substitute for the separate claim-to-source audit.

2026-08-30 — Independent audit completed

  • Scope checked: every claim–citation pair across all 18 canonical wiki pages and the complete literature layer. Mechanical closure covered 2,964 wiki lines, 632 inline PMID occurrences, and 496 page-level unique claim–citation mappings; each page retained 151–183 lines and 25–33 unique cited PMIDs.
  • Live identifier verification: final exact PubMed E-utilities retrieval resolved 388/388 condition PMIDs with zero failures. Final exact ClinicalTrials.gov API retrieval resolved 27/27 NCT identifiers with a current overall status present for every study. No PMID or NCT identifier was supplied from memory.
  • Claims and numbers corrected: narrowed MDD heritability from 35–40% to approximately 35% to match the cited review; removed PMID 10770484, a duplicate letter/erratum record incorrectly presented as the underlying meta-analysis; and corrected six overstated study-design labels (two patient-experience studies, one psychedelic follow-up, and three safety reviews).
  • Stale absences re-searched: reran 16 junction searches plus targeted searches for diagnostic/predictive biomarkers, bipolar conversion thresholds, and TRD sequencing. Two former no-study claims were obsolete. Added the exploratory inflammation–iTBS study (PMID 42364721) and two randomized post-ketamine/esketamine CBT studies (PMID 34186531; PMID 42095692), while positively restating the remaining gaps as lack of externally validated clinical utility, powered interaction evidence, or definitive sequencing evidence as of 2026-08-30.
  • Literature artifacts repaired: the bibliography now covers exactly all 388 condition PMIDs, including 385 unique wiki PMIDs, with zero missing or extra entries; 85 stale cited-by reverse mappings were corrected. All five landmark-note DOI/PMID pairs and the statistics ledger's quantitative claims were rechecked against live PubMed records.
  • Guidelines and patient voice: resolved the prior APA ambiguity by distinguishing the American Psychiatric Association's 2010 legacy MDD guideline from the American Psychological Association's active 2019 guideline and 2025 update-panel notice. Re-fetched named patient organizations; removed Depression Alliance because its domain now redirects to an unrelated site, retained dated access results, and preserved the English-language/high-income sampling limitation.
  • Errors found and fixed: 1 wrong citation, 1 numeric overstatement, 6 study-design mislabels, 2 stale absence claims, 1 stale organization record, 1 unresolved guideline item, and 85 bibliography reverse-map defects. Every identified item was repaired in place.

2026-10-07 — Uncited dots grounded

Grounder: Grok. This entry records a grounding pass over the "Dots not yet connected" table. D-2 and D-3 already carried PMIDs and were left unchanged. Every PMID written in the rewritten rows was retrieved with PubMed E-utilities in this session, read, and checked for retraction. PubMed publication types contained no Retracted Publication and no Expression of Concern for any record cited below. Crossref updated-by was empty for those records except the two notices named here.

Counts. Started with 14 uncited rows (D-1 and D-4 through D-16). Rewrote 10 (D-1, D-4, D-5, D-6, D-8, D-9, D-11, D-12, D-13, D-16). Deleted 4 (D-7, D-10, D-14, D-15). IDs were not renumbered.

Deletions.

  • D-7 (post-ECT relapse | digital sleep/activity monitoring). The relapse side has a number: in modern continuation-pharmacotherapy studies, 51.1% (95% CI 44.7-57.4%) relapsed by 12 months and 37.7% (30.7-45.2%) by 6 months (Jelovac 2013, PMID 23774532). The monitoring side has no quantitative finding in this condition. The PubMed join below returned 0 records. The row was deleted because one side could not be grounded.
  • D-10 (suicide-attempt polygenic liability | dynamic intent and access to means). The incremental-utility question has been asked. In a prospective psychiatric emergency cohort of 333 European-ancestry adults, 28 (8.4%) attempted suicide within 6 months. Adding polygenic scores produced AUC 0.86 (95% CI 0.73-0.99) versus 0.84 (0.70-0.98) for the clinical model, and the improvement was not statistically significant (Lee 2024, PMID 39442166). The liability-scale SNP heritability of suicide attempt in the source GWAS is 5.7% (SE 0.003) across 43,871 cases (Docherty 2023, PMID 37777856). What remains, dynamic intent and access to means inside MDD, restates OQ-11. The row was deleted and the published test is reported here.
  • D-14 (register-based bipolar conversion | activation/withdrawal phenotypes). The activation/withdrawal side had no usable adult number as a predictor of conversion. A PubMed search returned 3 records. The quantitative one is a pediatric target-trial emulation, ages 4-17, n=43,677: 12-week mania risk difference 0.06% (95% CI -0.04 to 0.16) and 52-week risk difference 0.28% (0.12-0.44) for antidepressant initiation versus none (Virtanen 2024, PMID 37755835). It does not add a dynamic adult medication-response term to a register conversion model. The row was deleted.
  • D-15 (task-shared psychotherapy | patient-defined treatment burden). The junction restates OQ-24, which already proposes the supervision trial. Kerr 2023's quantitative result is the 19% EQ-5D-5L mapping used in D-8, which is not a treatment-burden score. The row was deleted.

Searches behind absence and narrowing claims. All were PubMed E-utilities esearch on 2026-10-07. Titles of every returned record were read. Abstracts named below were fetched and read.

  • D-1 individual-participant profiles: ("major depressive disorder"[tiab] OR depression[tiab]) AND ("symptom profile"[tiab] OR "symptom profiles"[tiab]) AND antidepressant[tiab] AND ("individual participant"[tiab] OR "individual patient"[tiab] OR "patient-level"[tiab]) — count 1, PMID 32931110, a review, not a symptom-profile-by-drug individual-participant analysis. STAR*D-specific: (STAR*D[tiab] OR "sequenced treatment alternatives"[tiab]) AND ("symptom profile"[tiab] OR "symptom profiles"[tiab]) AND (interaction[tiab] OR moderator[tiab] OR subgroup[tiab]) AND antidepressant[tiab] — count 0.
  • D-4 mediation: psilocybin[tiab] AND depression[tiab] AND (mediation[tiab] OR mediator[tiab]) AND (expectancy[tiab] OR unblinding[tiab] OR suggestibility[tiab] OR "psychological support"[tiab] OR psychotherapy[tiab]) — count 1, PMID 37652000, a perspective on headache and chronic pain.
  • D-5 taper symptoms: (CYP2C19[tiab] OR CYP2D6[tiab] OR pharmacogen*[tiab]) AND antidepressant*[tiab] AND (DESS[tiab] OR "discontinuation symptoms"[tiab] OR "withdrawal symptoms"[tiab] OR taper*[tiab]) AND depression[tiab] — count 0. Genotype has been tied to stopping the prescription (Baye 2026, PMID 42168153) and, in NESDA, not to switching or perceived side-effect count (Brouwer 2024, PMID 38797832). The zero count is specifically DESS, withdrawal symptoms, and taper.
  • D-6 conversion: "mixed features"[tiab] AND bipolar[tiab] AND (conversion[tiab] OR prospective[tiab]) AND depression[tiab] — count 7 (PMIDs 35985516, 32684621, 31691722, 31479581, 30597295, 29310066, 28612977). Titles were read. None is prospective bipolar conversion inside DSM mixed-features MDD. Mazzarini 2018 (PMID 29310066) says the cross-sectional design could not assess that conversion. Polygenic: "mixed features"[tiab] AND (polygenic[tiab] OR PRS[tiab] OR "polygenic risk"[tiab]) — count 1, PMID 31628696, bipolar disorder, n=402, no polygenic-score-by-maltreatment interaction for mixed features.
  • D-7 digital relapse monitoring: (electroconvulsive[tiab] OR ECT[tiab]) AND depression[tiab] AND relapse[tiab] AND (actigraph*[tiab] OR wearable[tiab] OR "digital phenotyp*"[tiab] OR "passive sensing"[tiab] OR smartphone[tiab] OR "sleep monitoring"[tiab]) — count 0.
  • D-8 co-primary: ("phase 3"[tiab] OR pivotal[tiab]) AND (depression[tiab] OR "major depressive"[tiab]) AND "co-primary"[tiab] AND (function*[tiab] OR "quality of life"[tiab] OR recovery[tiab] OR "patient-reported"[tiab]) AND (MADRS[tiab] OR antidepressant*[tiab] OR esketamine[tiab]) — count 0.
  • D-9 equity primary outcome: "measurement-based care"[tiab] AND depression[tiab] AND (equity[tiab] OR disparity[tiab] OR disparities[tiab] OR "low-income"[tiab] OR LMIC[tiab] OR "middle-income"[tiab]) — count 18, including Husain 2025 (PMID 40892412). Titles of all 18 were read. Husain was read in full: the trial is in Pakistan and the primary outcomes are time to response and time to remission. The row does not claim that no low-resource measurement-based-care trial exists.
  • D-11 exercise by insulin resistance: exercise[ti] AND depression[ti] AND (HOMA[tiab] OR "insulin resistance"[tiab]) AND (moderator[tiab] OR interaction[tiab] OR subgroup[tiab]) — count 0. Broader: exercise[tiab] AND depression[tiab] AND ("insulin resistance"[tiab] OR "HOMA-IR"[tiab] OR HOMA[tiab]) AND (response[tiab] OR moderator[tiab] OR subgroup[tiab] OR interaction[tiab]) — count 10, including Jeremiah 2020 (PMID 32995594), which pools adjunct lifestyle and dietary interventions.
  • D-12 sexual function and combination treatment: psychotherapy[tiab] AND (antidepressant*[tiab] OR pharmacotherapy[tiab]) AND combination[tiab] AND depression[tiab] AND ("sexual dysfunction"[tiab] OR "sexual function"[tiab] OR "sexual side"[tiab]) — count 4. The only trial in the set is Zajecka 2002 (PMID 12197452). The other three titles are a cirrhosis symptom review, a psychotropic dosing handbook, and a bupropion efficacy review. The naive claim that combination treatment and sexual function had never been joined was false. The row now asks for a preference-weighted net-benefit model.
  • D-13 screening trials: ("screening for depression"[ti] OR "depression screening"[ti]) AND (randomized[tiab] OR randomised[tiab]) AND (outcome*[tiab] OR remission[tiab] OR benefit[tiab]) — count 48. Titles of all 48 were read. Abstracts read in full for DISCOVER (PMID 38906611), GET.FEEDBACK.GP (PMID 38432236), CODIACS-QoL (PMID 31633746), SHARP (PMID 38408462), and the Spanish primary-care cluster trial (PMID 23497463). One further ACS paper in the set randomized escitalopram versus placebo among people already screen-positive, so it does not answer screen versus no screen. SHARP randomized supervision of follow-up inside a programme that already screened. The row does not claim that screening benefit is unstudied.
  • D-14 activation and conversion: ("treatment-emergent mania"[tiab] OR "antidepressant-induced mania"[tiab] OR "antidepressant-induced hypomania"[tiab]) AND (conversion[tiab] OR predict*[tiab]) AND ("major depressive"[tiab] OR unipolar[tiab]) AND (cohort[tiab] OR prospective[tiab]) — count 3 (PMIDs 37755835, 25882763, 15353888). Virtanen 2024 was read and is pediatric, as above. The other two titles are a ziprasidone crossover in bipolar spectrum disorder and a bipolar spectrum pilot study.
  • D-16 target-trial emulations: ("target trial"[tiab] OR "trial emulation"[tiab] OR emulated[tiab]) AND antidepressant*[tiab] AND (mortality[tiab] OR death[tiab]) AND depression[tiab] — count 6 (PMIDs 42235877, 40197402, 40062740, 38994727, 38032000, 36639294). Abstracts were read. Zhu 2024 (PMID 38032000) is initiation versus non-initiation and reports no association with all-cause mortality in CKD stages G3-G5; the mortality hazard ratio is not in the abstract. Tsai 2026 and Chen 2026 are the other survival contrasts cited in the row. PMID 40062740 compares SSRIs on osteoporosis. PMID 38994727 compares SSRIs on psychiatric admission and suicide. PMID 36639294 emulates sequencing strategies in STAR*D with remission as the primary outcome. The naive claim that no target-trial emulation in depression exists was false.

Retraction and correction checks.

  • Noetel 2024 (PMID 38355154, DOI 10.1136/bmj-2023-075847) has a Crossref correction, DOI 10.1136/bmj.q1024, PMID 38806193. The PMC full text (PMC11131084) was read. It removes a methods sentence about an internal reference standard and states that effects were standardized with the baseline standard deviation and that this did not materially change the findings. The walking-or-jogging estimate is cited with that note.
  • Goodwin 2025, "The role of the psychedelic experience in psilocybin treatment for treatment-resistant depression" (PMID 39706482, DOI 10.1016/j.jad.2024.12.061), has a Crossref erratum, DOI 10.1016/j.jad.2025.120352, PMID 41075579. PubMed has no abstract. Europe PMC has no full text. The publisher page returned HTTP 403, the Elsevier text-mining link returned an invalid-input error, and Unpaywall rejected the request. The corrigendum text was not read, so PMID 39706482 is not used for any number. The correlations in D-4 are taken from Goodwin 2026 (PMID 41881122), which reports them and whose Crossref updated-by field was empty.
  • Kalfas 2025 (PMID 40632531) has a PubMed erratum, PMID 40694362, titled "Omitted Disclosures." Crossref updated-by for the article DOI was empty. The erratum does not change the SMD. The two sources disagree on whether an update exists; the disclosure notice does not change the cited estimate.
  • Zhu 2024 (DOI 10.2215/CJN.0000000000000348) and Romera 2013 (DOI 10.1186/1471-244X-13-83) returned no Crossref updated-by field. The STAR*D emulation DOI 10.1016/j.biopsych.2022.09.028 likewise returned no updated-by field.

Unresolved. The text of the Goodwin 2025 corrigendum (PMID 41075579) could not be obtained from PubMed, Europe PMC, Crossref, Semantic Scholar, or the publisher. D-4 therefore does not cite that paper. No row was left in the uncited keyword form. Cochrane CENTRAL and SciELO were not searched; absence counts above are PubMed counts. - Final disposition: zero unresolved substantive verification flags; all 18 pages promoted to status: curated; the condition promoted to audited in CONDITIONS-ROADMAP.md. Registry status, guideline currency, and organization availability remain explicitly dated and require future refresh rather than constituting unresolved audit failures.

2026-10-07 — Independent audit of Grok grounding pass

Auditor: Codex. Author of audited changes: Grok. Scope: the 10 rows Grok rewrote from uncited rows (D-1, D-4, D-5, D-6, D-8, D-9, D-11, D-12, D-13, D-16) and its four deletions (D-7, D-10, D-14, D-15). D-2 and D-3 already had citations before Grok's pass and were outside scope. All 14 decisions were checked; all 39 unique PMIDs in the rewritten rows and the four additional deletion-evidence PMIDs were re-fetched through live PubMed E-utilities. After repair, all 50 unique PMIDs in the 13 in-scope active rows were fetched and their first author, population, result direction, and quoted numbers checked against the accessible PubMed abstracts. PubMed publication types showed no Retracted Publication or Expression of Concern; Crossref updated-by was checked for every final in-scope record and showed no retraction. The Noetel exercise review carries a correction, already named in D-11, rather than a retraction. No NCT identifier was written in the audited rows.

Verification failures and anchoring. Zero cited identifiers failed to resolve; zero were flagged as retracted or under an expression of concern. Two quantitative/source-use claims failed strict verification: D-11's exact HOMA-IR g, sample, and heterogeneity figures were not present in the accessible PubMed abstract, and full text could not be retrieved through Europe PMC, so those figures were removed; D-13 attached semistructured-interview sensitivity/specificity to the entire 100-study, 44,503-person IPD pool, although the source reports that estimate for a subset, now stated correctly. Two motivated anchoring instances were repaired: D-1 used an older personalization review to support a broad no-analysis claim despite later direct symptom-level drug comparisons; D-11 presented a pooled exercise-plus-supplement review as if it tested insulin-resistance-based selection, which it did not. D-4, D-6, and D-12 also had broader no-analysis/unpublished wording than their searches could establish; these were narrowed to the actual unanswered question. No retained row merely repeats the same claim in Dot A and Dot B, so no synonym-only row was deleted.

Stale absences: four. D-1's broad assertion missed the patient-level STAR*D/CO-MED differential symptom effects (Strobl 2025, PMID 40449742); the remaining question is prospective benefit from baseline-profile-guided antidepressant choice. D-5's depression-restricted zero-result query missed a CYP2C19 poor-metabolizer escitalopram withdrawal case report (Tran 2026, PMID 41646208); the remaining gap is comparative prediction of DESS/taper burden. D-7's deletion said the monitoring side could not be grounded, despite post-ECT polysomnographic REM latency associated with later symptoms (Grunhaus 1994, PMID 7986885), residual sleep predicting relapse (Lambrichts 2022, PMID 35933854), and a small post-ECT actigraphy study without relapse follow-up (Winkler 2014, PMID 24998016). D-14's deletion said there was no usable adult activation number, but a Taiwan register-biobank study reports 145 early antidepressant-induced mania events among 772 antidepressant-treated depression patients (Huang 2026, PMID 41274093). The source also prints 19.65%, which does not equal 145/772; the row uses the reported count without repeating that internally inconsistent percentage.

Deletion adjudication. D-7 restored as a passive sleep/activity prediction junction; the published sleep and actigraphy findings are named in the row, and the remaining external-validation question is explicit. D-14 restored as register conversion risk versus early treatment-emergent activation, with no claim that early activation itself establishes later bipolar conversion. D-15 restored: task-shared treatment completion and symptom change in rural India (Agrawal 2025, PMID 40866825) differ from patient-described responsibilities, health conditions, and stigma in Vietnam (Chau 2024, PMID 39695579); the patient-burden junction is distinct from OQ-24's supervision-intensity question. D-10 remains deleted. Its stated published test was verified: Lee 2024 (PMID 39442166) prospectively assessed added suicide-attempt polygenic scores in 333 psychiatric emergency patients, finding AUC 0.86 versus 0.84 without significant improvement; the source GWAS (Docherty 2023, PMID 37777856) reports 43,871 cases and liability-scale SNP heritability 5.7%. The remaining dynamic-intent question is covered by OQ-11. Thus three deletions reflected difficulty or an incorrect overlap judgment, and one was justified by a named published test.

Search reruns and limits. Every verbatim PubMed E-utilities query in Grok's entry was rerun on 2026-10-07; counts matched its entry (D-1: 1 and 0; D-4: 1; D-5: 0; D-6: 7 and 1; D-7: 0; D-8: 0; D-9: 18; D-11: 0 and 10; D-12: 4; D-13: 48; D-14: 3; D-16: 6). The misses arose from restrictive conjunctions, not from wrong counts. Further PubMed searches used these verbatim queries: D-1 prospective selection (depression[tiab] OR "major depressive disorder"[tiab]) AND ("symptom profile"[tiab] OR "symptom profiles"[tiab] OR "symptom-guided"[tiab]) AND (antidepressant*[tiab] OR pharmacotherapy[tiab]) AND (randomized[tiab] OR randomised[tiab] OR "treatment selection"[tiab]) (22 titles; none described a prospective among-antidepressant symptom-guided decision trial); D-5 comparative withdrawal (CYP2C19[tiab] OR CYP2D6[tiab] OR pharmacogen*[tiab]) AND ("withdrawal symptoms"[tiab] OR "discontinuation syndrome"[tiab] OR "discontinuation symptoms"[tiab] OR taper*[tiab]) AND antidepressant*[tiab] AND (cohort[tiab] OR trial[tiab] OR prospective[tiab]) (0); D-7 passive post-ECT relapse (electroconvulsive[tiab] OR ECT[tiab]) AND (actigraphy[tiab] OR wearable[tiab] OR "passive sensing"[tiab]) AND (relapse[tiab] OR recurrence[tiab]) (0); D-14 activation-to-conversion (depression[tiab] OR "major depressive"[tiab]) AND ("antidepressant-induced mania"[tiab] OR "antidepressant-induced hypomania"[tiab] OR "treatment-emergent mania"[tiab]) AND ("bipolar conversion"[tiab] OR "conversion to bipolar"[tiab] OR "later bipolar"[tiab]) (0). The D-13 low-resource screening search (depression[tiab] AND screen*[ti] AND (randomized[tiab] OR randomised[tiab]) AND ("low-income countries"[tiab] OR "low-resource"[tiab] OR Malawi[tiab] OR Pakistan[tiab])) returned one Malawi implementation trial, where both arms screened and follow-up supervision differed. These are PubMed-scoped findings, not claims that no study exists in any database; CENTRAL and regional databases were not searched.

Disposition: seven existing rows repaired (D-1, D-4, D-5, D-6, D-11, D-12, D-13); three deleted rows restored (D-7, D-14, D-15); zero rows newly deleted. The table now has 15 rows. It is fit for a novelty gate that treats these as candidate junctions and reruns the named, source-limited searches. It is not evidence of worldwide nonpublication, and the gate should not turn any zero-count PubMed query into an absolute absence claim.

2026-10-15 — Periodic literature sweep (Codex; independent audit pending)

  • Window and scope: publications dated 2026-08-28 through 2026-10-15, inclusive. Read the August build/audit and October grounding follow-ups, INDEX, OPEN-QUESTIONS, guideline/statistics/patient layers, CONVENTIONS, CLAUDE and SOURCES. Only depression files changed. No publishing or version-control commands were run.
  • PubMed: 15 major-topic searches and 15 matching CDSR searches covered diagnosis/heterogeneity, epidemiology/burden, mechanisms, genetics, antidepressants/withdrawal, efficacy/reporting bias, psychotherapy/exercise/task-sharing, TRD/augmentation, glutamatergic agents, neuromodulation, psychedelics/neurosteroids, biomarkers/prediction, mortality/safety, guidelines and patient experience. All used an explicit publication-date range. A broader depress* AND "Cochrane Database Syst Rev"[jour] query used the same range. These 31 searches returned 789 distinct IDs: EFetch yielded 784 journal-article records and five book records. Selected title/abstract triage prioritized syntheses, comparative trials, safety and earlier follow-ups; this was not a full-text review of every hit. Exact queries, translations, counts and IDs are in the published-path manifest.
  • Cochrane: no topic-specific in-window CDSR hits; the broader query found three reviews, concerning sexual-harassment education, lumbar supports and exercise after spinal cord injury, which did not justify MDD changes. The October-issue commentary on CANMAT exercise guidance led to supplementary, unrestricted searches and exact EFetch retrieval of the January 2026 Exercise for depression update (PMID 41500513). The initially quoted-title search returned zero; the broader exercise[ti] AND depression[ti] search retrieved it. This earlier review is retained as catch-up, not mislabeled an in-window publication. Europe PMC reports subscription-required full text; its XML endpoint did not deliver the review. No numeric review findings were imported or full-text appraisal claimed. The Cochrane Library and CENTRAL were not searched because of the known HTTP 403 barrier.
  • Beyond PubMed: Europe PMC REST searches were run for all 15 topics, with a native-syntax guideline fallback after the PubMed publication-type syntax returned zero. Every first page was complete (all counts below 1,000). The native guideline query returned 52 records. Supplementary exact PubMed EFetch calls covered Europe PMC PMIDs absent from the first journal-record set. Conference abstracts and preprints were encountered but none warranted addition. No DOI-only journal record was added. Full-text availability was checked through Europe PMC; Crossref was queried without personal contact information. Unpaywall was skipped because LIFEBANK_CONTACT_EMAIL was unset. No worldwide absence is inferred: SciELO and LILACS could not be searched (HTTP 403), limiting regional/transportability coverage.
  • Trial status/results: all 27 previously cited NCT IDs were retrieved from exact ClinicalTrials.gov v2 API endpoints. Celecoxib for immunometabolic depression (NCT05415397; updated 2026-09-21) and GM-2505 (NCT06236880; updated 2026-09-18) are now completed; the other tracked status labels remain consistent with the earlier snapshot. None of the 27 responses contains a results section. Earlier result snapshots were not stored, so no complete posting/withdrawal history is claimed. WHO ICTRP was attempted but returned 403; ISRCTN depression search and an exact ketamine record redirected to its holding page, so neither is reported as a completed registry search. The named EU register was successfully searched for the NHS psilocybin trial (EudraCT 2018-003573-97): protocol and results opened, with results first published 2026-09-26. Its additional ClinicalTrials.gov record (NCT04959253), queried live, has no results section. This is a real cross-registry difference, not evidence of no available results worldwide.
  • What survived triage: seven bibliography records. Full texts support focused wiki additions for negative satoprodil trials (Cohen 2026, PMID 42701110), NHS psilocybin feasibility and failed blinding (Rucker 2026, PMID 42562964), and a critique of CANMAT exercise prescriptions (Gierc 2026, PMID 42411994). Bibliography-only records are the earlier Cochrane update, the individualized-versus-conventional TMS-targeting trial (PMID 42581396), acute antidepressant QTc/IPD synthesis (PMID 42778211), and DESS construct-validity study (PMID 42814111). Their unavailable full texts prevent changes to quantitative estimates or claims that older gaps are resolved. PubMed publication-date retrieval includes September/October print issues with earlier online dates; Rucker first appeared online in August and Gierc in July. No landmark deep note was justified.
  • Retraction checks: all seven retained records were checked against PubMed publication types and Crossref updated-by; none showed a retraction or expression-of-concern flag. A September retraction notice for an aripiprazole/bupropion augmentation review was identified (PMID 42678910), and Crossref confirmed its original DOI is marked retracted. The original was not cited in depression; it was not added. A transient Crossref 429 on the QTc paper was retried successfully before retention.
  • Guidelines: checked official NICE, CANMAT, American Psychological Association, American Psychiatric Association and VA/DoD pages. NICE reports surveillance on 2026-01-30 with no recommendation update; CANMAT still lists its 2023 MDD update; the Psychological Association still describes the 2019 document and its 2025 update panel; the Psychiatric Association listing retains MDD as legacy. No replacement was identified in these checked listings. Narrative/expert-panel papers were not relabeled as society-issued guidance. The new exercise critique is distinguished from an endorsed guideline in REGISTRY and wiki/guidelines.
  • Statistics and patient material: added the WHO depression fact sheet dated 2026-09-11: modeled 2023 figures (approximately 322 million people; 5.2% of adults) remain separate from older country interviews and GBD series. Checked MHA dashboard/methodology, but imported no population prevalence from voluntary screens. Retrieved Mind's World Mental Health Day campaign on same-day access and MHA's October awareness/screening calendar; added annotated public sources without identifiable patient-story details. DBSA, Beyond Blue and SADAG homepages were also checked; no further condition-specific material survived those homepage checks. Several direct-page barriers were resolved through the browser retrieval tool, as recorded in the manifest.
  • Applied changes: four wiki pages (ketamine, psychedelics, trials, guidelines), BIBLIOGRAPHY, guideline REGISTRY, STATISTICS, patient-voice sources, OPEN-QUESTIONS and INDEX. OQ-9 now distinguishes public-healthcare feasibility from efficacy under credible masking; new stable OQ-26 asks which exercise prescription/supervision model is justified across severity. Existing October junction hypotheses were not reasserted as new absences. Edited pages carry 2026-10-15 frontmatter and draft status; the separate auditor may promote them after verification.
  • Mechanical validation: all newly written PMID literals matched primary identifiers in this session’s EFetch responses; every newly written NCT ID matched a fetched v2 record; all seven retained records had both retraction checks. New relative links resolve, wiki citations have reference entries, edited-page frontmatter is dated/draft, and the actual public-builder mapping/transform resolves the manifest link to sweeps/2026-10-15.json. No full-site generation/build was run because publishing is handled after this task and shared outputs are outside the condition-only edit scope. These checks are not the independent medical-content audit.
  • Next sweep: obtain the full Cochrane exercise update (current Europe PMC record says unavailable) and evaluate its certainty/sensitivity analyses before revising exercise effect claims; obtain full texts and tables for the targeting trial, QTc/IPD synthesis and DESS measurement study; then reassess OQ-7/OQ-25, medication safety and taper measurement. Follow the newly posted EU psilocybin results and its open-label extension; continue the two completed tracked studies' results watch and CYB003/HNK/accelerated-TMS registry checks. Revisit society update panels and seek locally led, non-English regional evidence without treating inaccessible registries or databases as negative searches.

2026-10-22 — Periodic literature sweep (Grok; independent audit pending)

  • Window and scope. Publication dates 2026-08-28 through 2026-10-22, inclusive, as requested, rather than only the week after the 2026-10-15 sweep. Read CONVENTIONS.md, CLAUDE.md, SOURCES.md, this log, INDEX, and OPEN-QUESTIONS before editing. Only depression files changed. No publishing or version-control commands were run. Assumption: records already in the 2026-10-15 manifest were not re-added; triage was limited to identifiers that manifest did not contain, plus the named follow-ups. Exact queries, translations, counts, and returned identifiers are in the search manifest.
  • PubMed. The same 15 topic searches as 2026-10-15, each with a matching "Cochrane Database Syst Rev"[jour] search, plus a broad depression CDSR search, a 2026 CDSR catch-up, an Entrez-date search for records entered 2026-10-16 to 2026-10-22, and a retraction-publication-type search. The first novel-topic call was discarded because an unquoted CYB003 token made E-utilities return phrasesnotfound and the client treated that as fatal; the quoted rerun returned 22 ids. The first retraction query was likewise rerun with quoted publication types and returned 0 in the window. The other publication-date topic searches returned 856 distinct PMIDs. The novel rerun's only id absent from the 2026-10-15 manifest, PMID 42896406, was already inside those 856. Of the 856, 67 were absent from the 2026-10-15 manifest. The Entrez-date search added 11 further PMIDs whose publication dates fell outside that window or outside the topic queries. ESummary was retrieved for the new and follow-up PMIDs. Abstracts were EFetched for the records considered for addition. This was not a full-text review of every new hit.
  • Cochrane. Every topic-specific CDSR search in the window returned 0. The broad in-window CDSR search returned the same three out-of-scope reviews as on 2026-10-15 (sexual-harassment education, lumbar supports, exercise after spinal cord injury). A 2026 CDSR catch-up found no new MDD treatment review beyond the already catalogued exercise update (PMID 41500513). Europe PMC still reports that review as subscription-required, not in PMC and not in Europe PMC full text. No exercise effect size was imported. cochranelibrary.com returned HTTP 403, so CENTRAL was not searched.
  • Beyond PubMed. Europe PMC REST searches were run for all 15 topics over the same publication window (resultType lite; every topic's hit count was fetched in full). A non-MEDLINE query returned 194 records (113 preprints and 81 PMC records). A preprint-only query returned 113. Conference abstracts in that PMC set, including lumateperone phase-3 subgroup posters, an LSD-microdosing RCT abstract, and a GLP-1 adjunct abstract, had no PMID and no DOI, so they are not citable here and were not used for efficacy claims. No preprint was added. Crossref was queried without a mailto. Unpaywall was skipped because LIFEBANK_CONTACT_EMAIL was unset. DOAJ returned HTTP 502 and was not a completed search. OpenAlex is_retracted:true for "major depressive disorder" in the window returned one record, the aripiprazole/bupropion retraction notice already logged on 2026-10-15 (doi:10.1371/journal.pone.0357331); it was not added. SciELO returned HTTP 403. LILACS (lilacs.bvsalud.org) did not return an HTTP status. No regional-literature absence claim is made.
  • Trials. All 28 NCT IDs already present in the condition markdown were re-fetched from exact ClinicalTrials.gov v2 study endpoints. None changed overall status. None of those 28 responses contained a results section. GM-2505 remains completed; its last update posted date moved from 2026-09-18 to 2026-10-19 (NCT06236880). CYB003 remains recruiting, last update posted 2026-10-16 (NCT06793397). Two additional records were fetched because papers named them: home tDCS NCT04953208 (overall status UNKNOWN, last update posted 2024-01-18, no results section) and CoMPACT NCT06895863 (RECRUITING, estimated enrollment 78, last update posted 2026-08-26, no results section). WHO ICTRP returned HTTP 200 and an HTML shell; no result count was retrieved. The ISRCTN query API returned HTTP 404. The ChiCTR search path returned HTTP 404, and the project page for the temporal-interference trial's named ChiCTR ID did not return readable text, so that ID is not treated as verified. CTRI returned a search form, not a result set. The legacy EU Clinical Trials Register depression search returned HTTP 200 and the page text "1,481 result(s)"; that is not a complete CTIS review. EudraCT 2018-003573-97 results were re-opened: version 1 remains current, publication date 26 September 2026. Its ClinicalTrials.gov record (NCT04959253) still has no results section.
  • Retraction checks. The five records added below were checked in PubMed publication types and in Crossref updated-by. None was a retracted publication or an expression of concern; every updated-by list was empty. A separate 2026 retracted-publication search returned PMID 42656927, a Cureus review of accelerated iTBS versus daily rTMS. PubMed types include Retracted Publication, and the PMC full text is titled as a retraction. Crossref updated-by for doi:10.7759/cureus.113431 was empty, so the two sources disagree. The review is not cited in this condition and was not added.
  • What survived triage. Four records changed pages. Dechantsreiter 2026 (PMID 42895746) is a multicenter double-masked home tDCS-plus-training trial: feasibility was met and the week-6 MADRS difference versus double sham was 0.89 (95% CI −1.37 to 3.15). The open PMC full text was read. Peng 2026 (PMID 42629468) is a Nature Medicine postmortem multiome study reporting a shift toward more neural stem cells and fewer neuroblasts in nonmedicated MDD (11 MDD and 19 control donors after quality control); the open full text was read and no cell-proportion odds ratio was copied because the numeric ratios sit in a figure. Faghani Jadidi 2026 (PMID 42890690) is the CoMPACT protocol, not a result; it sharpens OQ-7 and OQ-25 without answering them. Nabil 2026 (PMID 42880603) is a tACS GRADE review whose abstract concludes against routine use; pooled odds ratios were not copied because the full text is subscription-only. Wang 2026 (PMID 42880279), a sham-controlled temporal-interference trial, is bibliography-only for the same full-text reason. No landmark deep note was written. The GENE-YD youth pharmacogenetic feasibility trial (PMID 42897167) has a publication date of 1 August 2026, outside this window, and was not used for an efficacy claim.
  • Guidelines, statistics, patient material. NICE NG222 still shows a review on 30 January 2026 and no recommendation update. CANMAT still lists the 2023 MDD guidelines. VA/DoD still presents the 2022 MDD guideline. The American Psychiatric Association listing still labels the 2010 MDD guideline as legacy. The American Psychological Association guideline URL returned an Incapsula block and was not re-verified. The WHO depression fact sheet, dated 11 September 2026, still states approximately 322 million people, 4% of the population, and 5.2% of adults (4.1% men, 6.2% women; 5.4% at ages 70 and older). Beyond Blue's homepage now describes a sixth-year Connection Postcard mailing; Mind, MHA, DBSA, and SADAG homepages added no depression-outcome campaign. No identifiable patient story was retained.
  • Applied changes. Wiki: neuromodulation, neurobiology, trials, psychedelics, ketamine, guidelines. Also BIBLIOGRAPHY, guideline REGISTRY, STATISTICS, patient-voice sources, OPEN-QUESTIONS, and INDEX. Edited pages are status: draft with last-curated: 2026-10-22. OQ-7 and OQ-25 now name CoMPACT and still ask for a design that separates dose, spacing, and acceleration.
  • Next sweep. Read the Cochrane exercise update, the connectivity-guided iTBS trial (PMID 42581396), the QTc IPD synthesis (PMID 42778211), and the DESS paper (PMID 42814111) if full text appears, before changing those estimates. Watch results posting for NCT05415397, NCT06236880, and the EU psilocybin extension. Re-check the American Psychological Association page if it stops returning the Incapsula block. If a journal version of the lumateperone phase-3 pooled analysis or the LSD-microdosing trial acquires a PMID or a Crossref-resolved DOI, appraise it then. Do not treat CENTRAL, SciELO, or LILACS as searched.

2026-10-29 — Periodic literature sweep (Claude; independent audit pending)

  • Window and scope. Publication dates 2026-08-28 through 2026-10-29, as instructed. An Entrez-date search for records entered 2026-10-23 to 2026-10-29 caught records whose publication dates fall outside that window. Before editing, read CONVENTIONS.md, CLAUDE.md, SOURCES.md, this log, INDEX and OPEN-QUESTIONS. Only depression files changed. No publishing or version-control commands were run. Assumption: anything in the 2026-10-15 or 2026-10-22 manifest, or already cited by PMID in this condition, was not re-triaged. Queries, translations, counts, returned IDs and registry checks are in the search manifest.
  • PubMed. Ran the same 15 topic searches as 2026-10-22, each paired with a "Cochrane Database Syst Rev"[jour] search. Also ran the broad depression CDSR search, the 2026 CDSR catch-up, the Entrez-date search and two retraction searches. The topic searches returned 965 distinct PMIDs; 120 were new against both prior manifests and the condition's cited PMIDs. ESummary was read for all 120 and abstracts were EFetched for 10. E-utilities flagged the unquoted CYB003 token as phrase-not-found, but the novel-topic query still returned 26 IDs. One HTTP 429 was retried successfully.
  • Cochrane. Every topic-specific CDSR search returned 0. The broad depression CDSR search found one new record, a review of music-based interventions in haemodialysis (PMID 42929381). It is out of scope and was not added. No new MDD Cochrane review exists to supersede or contradict this condition. cochranelibrary.com returned HTTP 403, so CENTRAL was not searched.
  • Beyond PubMed. All 15 Europe PMC topic searches were re-run over the same window; hit counts are in the manifest. The non-MEDLINE query returned 213 records, 22 of them new since 2026-10-22: 4 PMC-only journal or preprint records and 18 preprint-server records (Research Square, PsyArXiv, Preprints.org and a new 10.64898 prefix). None was added; preprints were not used to support claims. Europe PMC fullTextXML returned HTTP 500 for both JAMA-group PMCIDs. NCBI PMC EFetch showed both are embargoed, with front matter only. OpenAlex is_retracted:true returned only the aripiprazole/bupropion retraction notice already logged on 2026-10-15. DOAJ returned HTTP 200, but only a year-level 2026 count (1,310), which was not triaged. Crossref was queried without a mailto. Unpaywall was skipped because LIFEBANK_CONTACT_EMAIL is unset. SciELO returned HTTP 403. LILACS served a homepage, but no search was run. No regional-literature absence claim is made.
  • Trials. All 30 NCT IDs already cited in this condition were re-fetched from the ClinicalTrials.gov v2 study endpoints. None changed overall status, and none had a results section. Three posted updates without a status change: NCT06136546 (last update posted 2025-05-07 → 2026-10-27), NCT06793397 (2026-10-16 → 2026-10-27) and NCT07226661 (2026-03-24 → 2026-10-28). NCT05415397 and NCT06236880 still have no posted results. New to the tracked set: NCT05870540 (BPL-003/mebufotenin), completed with 196 enrolled and results first posted 2026-09-10. Its posted primary-outcome LS-mean difference (−5.3; 95% CI −8.9 to −1.8; P = 0.0033) matches the JAMA paper. EudraCT 2018-003573-97 results are still version 1, published 26 September 2026. The legacy EU CTR depression search still reads "1,481 result(s)". WHO ICTRP served an HTML shell with no count. ISRCTN and ChiCTR returned HTTP 404 and CTRI refused the connection. The Korean CRIS ID KCT0010758, named in a bibliography-only paper, and the adolescent iTBS trial's ChiCTR ID were not opened.
  • Retraction checks. Every record added or newly cited (PMIDs 42931733, 42905697, 42924478, 42913960, 39964582, 42906705, 42925114) was checked against PubMed publication types and Crossref updated-by. None is retracted or carries an expression of concern; every updated-by list was empty. The in-year MDD retracted-publication search returned only PMID 42656927, which was already logged and is not cited.
  • What survived triage. Three wiki edits, all read from abstracts and key points only, never full text:
  • Feifel 2026 (PMID 42931733) is an industry-funded phase 2b trial of single-dose intranasal mebufotenin in TRD. Day-29 MADRS was −5.3 for 12 mg versus 0.3 mg, and most participants were discharge-ready within 2 hours. Added to the psychedelics and trials pages with explicit caveats: 8-week follow-up, a subperceptual comparator, and blinding data unread because the full text is embargoed. The registry results cross-check the primary endpoint.
  • Teutenberg 2026 (PMID 42905697) predicted 2-year relapse in 564 patients from 9 modalities. Balanced accuracy was at best 64.43%, and the best model was clinical-only. Added to the biomarkers page. That page was curated; the edit could not be checked against the embargoed full text, so it is now draft.
  • Lin 2026 (PMID 42924478) is a late-life NIBS network meta-analysis: bilateral ECT SMD 1.13 and 10-Hz left-DLPFC rTMS SMD 0.42, both moderate certainty. It fills the neuromodulation page's missing late-life section.
  • Ghimire 2026 (PMID 42913960) estimates $204.58 billion of 2023 US healthcare spending associated with MDD. It went to STATISTICS beside Greenberg 2021, with a caveat that the two differ in scope.
  • Ma 2026 (ketamine vs ECT meta-analysis of 6 RCTs; Epub 2025) and Jang 2026 (digital CBT RCT whose primary effect was analysis-sensitive) are bibliography-only.
  • Not added: a semaglutide explore/exploit secondary analysis (no efficacy outcome); an adolescent high-dose iTBS RCT (publication date 2026-07-29, outside the window; retrospectively registered; single-blind); a post hoc sex analysis of the psilocybin–escitalopram trial (n = 59); an open-label aripiprazole-versus-bupropion trial; and a 6-mA tDCS preprint.
  • No landmark note was written. Mebufotenin is a candidate once the full text and a phase 3 result exist.
  • Guidelines, statistics, patient material. No change on any page checked. NICE NG222 still shows "Last reviewed: 30 January 2026" with no update. CANMAT still lists 2023, VA/DoD 2022, and the American Psychiatric Association still marks its 2010 guideline as legacy. The American Psychological Association guideline URL again returned a 212-byte block page. The PubMed guideline query's three new records were two reference-book chapters and a five-person expert consensus panel (PMID 42925114). The panel was noted in REGISTRY but not catalogued. The WHO fact sheet (11 September 2026) still gives approximately 322 million people, 4%, and 5.2% of adults. Homepages for Mind (general "See Me Now" mental-health campaign, not depression-specific), MHA, DBSA, SADAG and Beyond Blue (the Connection Postcard drive, already recorded) offered no new depression-specific material. Patient-voice files were not changed.
  • Applied changes.
  • Wiki: psychedelics-and-novel-therapeutics, clinical-trials-landscape, biomarkers-and-treatment-prediction (curated → draft) and neuromodulation. All are status: draft with last-curated: 2026-10-29.
  • Also changed: BIBLIOGRAPHY (6 entries), guideline REGISTRY (2026-10-29 checkpoint), STATISTICS (2 rows plus a caveat), OPEN-QUESTIONS (OQ-9 and OQ-16 sharpened, none answered), INDEX and the new manifest.
  • Next sweep.
  • Re-check the PMC embargo for Feifel 2026 (PMC13711228) and Teutenberg 2026 (PMC13688201). Read the mebufotenin blinding assessment and secondary endpoints, and the relapse model's validation scheme and calibration, before adding detail.
  • Watch for a phase 3 mebufotenin registration and for results posting on NCT05415397 and NCT06236880.
  • Carry forward the older full-text follow-ups: PMIDs 41500513, 42581396, 42778211 and 42814111.
  • Re-try the American Psychological Association page.
  • Do not treat CENTRAL, SciELO or LILACS as searched.