Unresolved Conflicts

Cross-chapter contradictions found during assembly. Per the assembly brief, none of these have been resolved by the editor: both claims are preserved in place, and this section flags them for the owner to investigate against primary sources. (Contradictions within a single chapter — the CGRP direction dispute, the sex-ratio narrowing, weather triggers, the ONS placebo results, and so on — were already flagged in place by the chapters themselves and are collected in the consolidated Open Questions section below.)

1. Kudrow 1981 oxygen response rate: 75% vs 82%. Part I, §1.7 reports the landmark Kudrow oxygen trial (52 patients) as “75% got relief with a 7 L/min oxygen mask; 62% aborted their attack within 7 minutes.” Part V, §1.1.1, citing Guo et al. 2019, reports the same 1981 trial as “7 L/min for 15 minutes aborted more than 7 of 10 attacks in 82% of patients.” The two chapters relied on different secondary sources; the primary paper (Kudrow L, Headache 1981;21:1–4) should be checked directly.

2. Galcanezumab episodic trial (CGAL, NCT02397473) sample size: 106 vs 109. Part V, §3.6.1 gives n=106 randomised (49 galcanezumab / 57 placebo), sourced to the NEJM 2019 publication. Part VII, §5.1.1 gives n=109, sourced to the ClinicalTrials.gov registry record. Same trial, two authoritative sources, two numbers.

3. Diagnostic-delay figures attached to different countries. Part III, §4.7 lists “Japan 7.3 ± 6.9 years” among country-level mean delays. Part IV, §3 attributes a near-identical figure — “7.3 ± 8 years” — to an Italian clinic series (Cefalea a Grappolo, n=100), and gives Japan 8.1 years (Imai et al.) and 3.6–9 years (Japanese Headache Society data). Similarly, Part III lists “Serbia 7.8 ± 8 years” while Part IV lists “7.8 ± 8.3 years” for the Valladolid (Spain) registry. The overlapping numbers attached to different countries suggest at least one chapter’s attribution is off; both tables are preserved as found.

4. The Manzoni natural-history cohort: which study do the transition rates come from? Part IV, §1 describes a “ten-to-twenty-five-year natural-history follow-up (Manzoni et al.) of 123 episodic and 9 chronic patients” yielding 12.9% episodic→chronic and 32.6% chronic→episodic. Part VI, §4 describes “a 10-year follow-up of 189 patients” yielding “about 13%” and “about 30%” for the same two transitions. Part IV’s own reference list includes “Cluster headache — course over ten years in 189 patients” (PMID 1742772). Whether these are one study inconsistently described or two related cohorts is not resolved by either chapter.

5. What the Danish 48.3% smoking figure denotes. Part III, §4.4 reports the Danish Cluster Headache Survey (Lund et al. 2019) as finding 48.3% current smokers (with 74.5% ever-smokers) vs 9.0% of controls. Part IV, §5 describes the same survey’s figure as “current or past smoking prevalence of 48.3%.” These descriptions are incompatible; the primary paper defines which is right.

6. Lifetime single-bout course: 25% vs 17%. Part I, §2.1 states “about 25% of episodic-CH patients experience only one cluster period in their entire life.” Part IV, §1 cites a 60-patient follow-up in which remission after an assumed first cluster period occurred in “only 17%” after 3+ years. These may be measuring different constructs (lifetime course vs medium-term follow-up after a first bout), but the chapters do not reconcile them.

7. PEACE trial funding amount: disclosed or not? Part VI states the Australian PEACE psilocybin trial’s MRFF funding was “reported at AUD $800,000.” Part VII’s funding audit states the amount is “genuinely undisclosed in both public announcements found.” Directly contradictory sourcing about the same trial.

8. The chronic share of Western CH cohorts. Stated as “10–15%” (Part I, §2.1), “10–37% typically reported in Western cohorts” (Part I, §4.6), “10–20%” (Part IV, §3), with cohort-level scatter up to 27.1% (Part IV, §1). Each range traces to different underlying cohorts; no chapter reconciles them into one figure.

9. Evidence-tier label for the 2026 DMT survey abstract. Part VI tags the Schindler/Lenaburg/Wold interim abstract citizen sciencepeer-reviewed; Part VII tags the same abstract preprint / trial + citizen science. Labels preserved as found — the owner may want to settle one convention for conference abstracts.

10. The International Cluster Headache Questionnaire’s headline n. Part I cites the Burish 2021 pain-comparison analysis at n=1,604 respondents; Part VI states the questionnaire had “over 3,000 participants” when published in Headache in November 2021; Part V separately cites “Pearson et al. 2019, Cluster Headache Questionnaire, n≈3,251.” These are plausibly different analyses or waves of one instrument, but none of the chapters says so explicitly.