Registry codes: SAL · NCK · AX-HS · converted from the registry dossier families/10-salivary-neck-axilla.md — the dossier remains the source of truth; this page is its reading surface.
30 cards · 🟢 11 verified · 🟡 12 qualified · 🔴 5 disputed · ⚪ 2 qualitative — global registry: 41 disputed cards carry their dispute on the card.
AX-HS-CTX-01 · Recurrence after incision & drainage ALONE in HS
No pooled measured rate exists.
Narrative upper-bound claim in peer-reviewed HS literature: recurrence "up to ~100%" — I&D relieves the acute abscess; it does not treat the disease. Handbook wording is qualitative only: *recurrence after drainage alone is the rule, not the exception.*
(Jemec 1996: 1.0%, 95% CI 0.4–2.2, 1-year, recall-based, مؤقت/abstract; Revuz 2008: 0.97%, 67/6,887, self-report, full text) · — 98/100,000 (95% CI 97–99), 47,690 cases in >48 M records (Garg 2017, full text) · pooled, bias-adjusted (95% CI 0.26–0.63), I²=99.97% (Jfri 2021, full text)
Prevalence under three different ascertainment methods — see Dispute D4
🔴Disputed. Sources disagree — the dispute table is on the card
Applies to — the population, exactly as printed
(a) European population symptom survey/examination; (b) US coded health records; (c) pooled meta-analysis
Basis — n / evidence base
See D4
Evidence tier (the registry's wording)
SR/MA + population studies
Dispute — the sources disagree
Jemec et al. · 1.0% (95% CI 0.4–2.2), 1-year · Danish general-population sample; history/recall + examination study design · 599 · provisional (abstract) · 1996
Revuz et al. · 0.97% ≈ 1% (67/6,887) · French representative sample ≥15 y; self-report questionnaire · 6,887 respondents · Full text · 2008
Garg et al. · 0.10% (98/100,000) · US EHR platform (Explorys), coded diagnoses only; single-code PPV 79.3% · 47,690 cases / >48 M patients · Full text · 2017
Jfri et al. SR/MA · 0.40% pooled, bias-adjusted (I²=99.97%); own split: population-based 0.3% vs clinical samples 1.7% · 16 studies, Western Europe/Scandinavia, USA, Australia · 38,082,054 · Full text · 2021
Why they differ
A complete difference in case ascertainment: the population survey captures undiagnosed patients; the coded records see only those who sought care, were diagnosed and coded — so most mild cases, and those misdiagnosed as 'boils', pass underneath them. And the famous 4.1% figure is not a third party: it is a point prevalence on examination in a young sample from a sexually transmitted diseases clinic (Jemec herself) — a selected group, not the general population.
Adopted
The dual-method wording: 'about 1% by the European population survey; ~0.1% by the American coded records — and the difference is a measurement method, not a disease geography'.
Justification
The pooled figure of 0.40%, with 99.97% heterogeneity, is an average across incompatible methods — publishing it alone hides the only certain truth here: that the figure is a child of its method.
Source · Wide excision · Limited / local excision · Deroofing · Counting unit & era · Studies · Year
Mehdizadeh et al. SR/MA (full text) · 13.0% (5–22) · 22.0% (10–37) ["local excision" — abstract wording "local incision" is an error in the abstract] · 27.0% (23–31) · Per study as defined; 1990–2015 literature; follow-up NOT pooled · 22 · 2015
Ovadja et al. SR/MA (full text) · 5.0% (3–9) · 26.0% (16–37) as "partial excision" · inside "partial" (lumped with I&D, laser, cryotherapy) · Per lesion/intervention; literature to 2019; median FU 20 mo (wide 23, partial 12) · 125 · 2020
Why they differ
Three documented reasons: (1) the unit of counting — Ovadja counts by lesion/intervention, not by patient; (2) the definition of 'partial' — in Ovadja a wide basket (including even abscess opening), while Mehdizadeh separates deroofing alone; (3) era and selection — newer studies and newer closure techniques in Ovadja. An honesty note: the crude proportions inside Ovadja itself (partial 34.7%, wide 16.2%) are higher than her modelled estimates (26%/5%) — the modelled ones are published as the authors published them, with this caveat.
Adopted
Both are published; and for the patient card, the inclusive range for wide excision (5–13 in every 100), with both parties named in the doctor card.
Justification
Two strong, equal sources with no decisive methodological tiebreaker — the case of rule 6 of the dispute protocol (the inclusive range, with both parties named). What they agree on (the ordering: wide < limited ≈ deroofing) is the clinical message, and it is stable across all opened analyses, including the unverified Riddle 2021 (secondary only: wide 8%, local 34% — not published).
NCK-LN-DIAG-01 · FNA performance — excellent for metastasis, insufficient alone for lymphoma
Metastatic lymphadenopathy diagnosed in 39/39 (100%) · correct lymphoma diagnosis in 73% · of 77 malignant aspirates, 35 (45%) were false-negative — 83% of those false negatives were lymphoid malignancies
by FNAC with immunocytochemistry; (94% if "suspicious of lymphoma" counts) — yet ALL lymphoma patients still required surgical (excisional) biopsy for full subtyping (Bosch QDU, expert cytopathology unit). Corroboration of the gap: in Chau's all-comer clinic, FNAC sensitivity for malignancy was only 49%, and
NCK-LN-DIAG-01 · FNA performance — excellent for metastasis, insufficient alone for lymphoma
🟢Verified. Confirmed in 2+ independent sources, full text opened
Applies to — the population, exactly as printed
Referred lymphadenopathy cohorts (Bosch n=372; Chau n=550, FNAC in 289)
Evidence tier (the registry's wording)
Two prospective cohorts (full texts opened)
Dispute — the sources disagree
Source · Sensitivity · Specificity · Definition of "positive" & case-mix · n · Status · Year
Tandon et al. SR/MA via DARE critical abstract · 89.6% (lymph-node subgroup 94.2%) · 96.5% · FNAC vs histology, head & neck masses, 38 studies; thresholds per primary study · 3,459 samples · provisional (critical abstract of the SR) · 2008
Chau et al. LNDC (full text) · 49% · — (accuracy framework differs) · C3–C5 counted positive, ALL-comers including lymphomas; 83% of false negatives were lymphoid malignancies · 289 FNAC of 550 · Full text (primary cohort) · 2003
Bosch et al. QDU (full text) · Metastatic disease 39/39 (100%) diagnosed · FN 1.3%, FP 0% among malignant cases · Expert unit: cytomorphology + immunocytochemistry/flow on every case · 372 · Full text (primary cohort) · 2014
Why they differ
Difference in the definition of positivity and the case mix: the Tandon pooling is dominated by metastatic tumours and salivary/thyroid glands with samples selected for histological comparison, while Chau's figure comes from an all-comers clinic in which lymphoma — the needle's structural weak point — is counted inside the denominator. They are not two contradictory measurements of the same performance; they are performances for different questions.
Adopted
No single sensitivity figure for the needle is published. The published rule (DIAG-01): the needle is excellent for metastatic spread, and insufficient on its own for lymphoma — supported by two full texts of two cohorts (Bosch + Chau), with the pooled Tandon figure in the side sources box tagged as provisional.
Justification
Settling with a single pooled figure would be read as a performance promise in both cases together — a promise the lymphoma data in every opened cohort refute.
Two prospective referral cohorts (full texts opened)
Dispute — the sources disagree
Older biopsy series, as summarised in AAFP reviews (Bazemore 2002: 54–85%; AAFP 2025 update: 54–84% — quote the version you cite) · 54–85% malignant · Patients whose supraclavicular nodes were selected for biopsy · Secondary reporting in opened full texts · 1980s series via 2002/2025 reviews
Chau et al. LNDC (full text) · 34.3% (12/35) · ALL GP-referred patients presenting with supraclavicular nodes to a rapid-access clinic · Full text (primary cohort) · 2003
AAFP reviews citing Fijten-lineage data · ~90% if >40 y vs ~25% if <40 y · Supraclavicular lymphadenopathy in biopsy-study framing · Secondary reporting (attribution chain tangled: Ferrer→Fijten; Richner→refs) · 1998/2010
Why they differ
Denominator difference, typically: biopsy series see only those already selected for biopsy (high prior suspicion), while the referral cohort sees everyone who attended. And the same lesson repeats in age: the '90% over forty' figures come from the framing of the biopsy studies.
Adopted
The doctor card displays both parties with their denominators explicitly; and the practical message stable across all sources: a supraclavicular node is the neck site of highest relative risk (RR ~3.7–4.7 in two referral cohorts, RISK-01) and is always biopsied early (THR-01 + the Richner rule: above/below the clavicle is not watched).
Justification
Relative risk is consistent across settings while absolute risk is a child of the denominator — so the relative is published as a figure, and the absolute conditional on its denominator.
NCK-LN-THR-01 · When a neck mass is at increased risk — act, don't watch (AAO-HNS 2017 CPG, adults ≥18 y)
≥2 weeks · fixation, firm consistency, size >1.5 cm, overlying ulceration · FNA instead of open biopsy
Increased malignancy risk if: mass present without significant fluctuation AND no infectious explanation, or duration uncertain (KAS 2a); or any of: (KAS 2b). At increased risk: contrast CT/MRI (strong recommendation); when diagnosis uncertain (KAS 7); cystic mass — keep evaluating, never assume benign (KAS 8); EUA of upper aerodigestive tract before open biopsy if still undiagnosed (KAS 10); do NOT routinely prescribe antibiotics without bacterial signs (KAS 1)
SAL-PAR-COMP-03 · Facial nerve palsy after total parotidectomy
transient 59.8%, permanent 4.0% · permanent 20.5%
Witt 2002 MA as quoted verbatim in an opened full text: — vs a large monitored single-centre series subgroup (abstract-only): transient 33.0%, (Jin 2019, n=794 all parotidectomies; malignancy and deep-lobe cases concentrated in the total-parotidectomy subgroup) — see Dispute D2
Transient / permanent, mixed benign+malignant case-mix in the underlying series
SAL-PAR-COMP-03 · Facial nerve palsy after total parotidectomy
🔴Disputed. Sources disagree — the dispute table is on the card
Applies to — the population, exactly as printed
Witt: PA parotid surgery by extent; Jin: benign+malignant, Samsung centre 2009–2016
Evidence tier (the registry's wording)
MA (secondary-quoted) + cohort (abstract-only)
Dispute — the sources disagree
Source · Transient · Permanent · Population / definition · n · Status · Year
Witt RL meta-analysis, as quoted in Salih 2022 (full text of quoting paper) · 59.8% · 4.0% · Parotid surgery for PA, by extent of resection · Not restated in quoting text · Secondary quotation · 2002
Jin et al., *BMC Surg* (abstract) · 33.0% · 20.5% · Total-parotidectomy SUBGROUP of an all-comers parotidectomy cohort (benign + malignant, monitored); permanent = persisting >6 months · 794 (whole cohort) · provisional (abstract-only) · 2019
Why they differ
Difference of stratum and confounding: Jin's whole cohort is a benign/malignant mix, and its 'total parotidectomy' group concentrates the malignant and deep tumours — which are themselves the cause of permanent weakness (sacrifice of the nerve or its adherence to the tumour), while Witt's tier is specific to benign adenoma surgery by extent of resection. They are not two measurements of the same risk.
Adopted
No single figure is published for the patient. The doctor card displays both parties with their denominators, and the written practical rule: the risk of total parotidectomy is quoted according to the reason for the resection (planned benign vs malignant/deep), not according to the operation's name alone.
Justification
Choosing either figure as a headline would be a transfer across different strata — the first prohibition of the golden rule. Jin's provisional figure also does not headline, per the 'provisional' rule.
SAL-PAR-COMP-04 · Gustatory sweating (Frey) after parotidectomy
(a) Pooled clinical: SP 13% (95% CI 7–25); ED 4% (3–6) · (b) Paired (Choi 2013, same 59 patients, same visit): subjective symptoms 33.9% (20/59) vs MSIT-positive 50.8% (30/59)
(Salzano 2025; ascertainment as reported per study, mostly clinical) · ; of MSIT-positive patients 63.3% reported symptoms
Two ascertainment layers, never blended: (a) pooled CLINICAL/symptomatic rates by technique; (b) paired same-patient comparison of symptoms vs Minor starch-iodine test (MSIT)
SAL-PAR-COMP-04 · Gustatory sweating (Frey) after parotidectomy
🔴Disputed. Sources disagree — the dispute table is on the card
Applies to — the population, exactly as printed
(a) PA meta-analysis population (COMP-01). (b) Unilateral parotidectomy (56 superficial, 3 total; PA 83%, Warthin 17%), assessed at 12 months
Evidence tier (the registry's wording)
SR/MA + paired cohort (both full texts opened)
Dispute — the sources disagree
Salzano 2025 SR/MA (full text) · SP 13% (7–25); ED 4% (3–6) · Clinical/symptomatic, as reported per primary study (unstandardised) · PA parotidectomy · SP 77/462; ED 25/941 · 2025
Choi 2013 — same patients, same visit · MSIT-positive 50.8% · Minor starch-iodine test (objective) · Same · 59 · 2013
Why they differ
A pure difference of measurement method: reported symptoms < a structured questionnaire result < a sensitive objective test. The test reveals subclinical reinnervation that most of its owners do not feel (only 63.3% of test-positive patients reported symptoms). The circulating '~80% on Minor's test' figure was not verified from an open text in this round, so it was removed from the body and recorded as provisional in the review log.
Adopted
For the patient card: the clinical tier (13% after superficial lobectomy), because it is what the patient actually feels; and for the doctor card: the three tiers with their measures, and Choi's dual figure (33.9%/50.8%) is the anchor of the explanation.
Justification
Mixing the test tier with the symptom tier in a single figure is exactly the definition error this dispute was built to expose.
SAL-PAR-RISK-01 · Malignancy share by gland (the classic gradient)
20–25% · 40–50% · 70–90% · >50% · sole adopted set
Parotid · Submandibular · Sublingual · Minor glands — (Thoeny gland-by-gland ranges, confirmed by the UpToDate check: ~25% · 40–45% · 70–90% · 50–75%); the Alsanie ranges and the merged sublingual+minor 50–81% are retained in D1 as dispute history only — adjudicated 2026-11-08, see D1
SAL-PAR-RISK-01 · Malignancy share by gland (the classic gradient)
🔴Disputed. Sources disagree — the dispute table is on the card
Applies to — the population, exactly as printed
Literature ranges as stated in two opened full texts + UpToDate topic check 2026-11-08 — see Dispute D1
Basis — n / evidence base
—
Evidence tier (the registry's wording)
Review-cited ranges, NOT one cohort's observed rates
Dispute — the sources disagree
Source · Parotid · Submandibular · Sublingual · Minor glands · Nature of the figures · Year
Thoeny, *Cancer Imaging* (full text) · 20–25% malignant ("most parotid tumours are benign (80%)") · 40–50% · combined with minor: 50–81% · combined (see left) · Literature ranges cited in a review · 2007
Alsanie et al., *Head Neck Pathol* — introduction (full text) · 15–35% · 41–45% · 70–90% · "more than half" malignant · Literature ranges cited in a multicentre paper's introduction — NOT that paper's own observed cohort rates · 2022
Why they differ
The two figures are not two competing measurements of the same stratum: both are *cited literature ranges* in reviews, gathered from series of different eras, countries and referral patterns; and one of them (Thoeny) merges the sublingual gland with the minor glands while the other separates them. An honesty note: Thoeny's arithmetic phrase ('80% benign' with a malignancy range of 20–25%) overlaps as published — it is carried verbatim and not corrected.
Adopted
Both ranges are published together in the doctor card, and in the patient card the older range split by gland (Parotid 20–25% · SMG 40–50%) with a note that 'the figures differ slightly between studies'.
Justification
There is no single opened primary source with a malignancy-by-gland table that outranks both parties (Alsanie's own cohort table was not opened in full in this round — recorded as a verification debt for the 2027-02-06 review); settling with a figure from either one claims a precision that does not exist. The direction (parotid least malignant ← sublingual highest) is agreed in both sources and is the carrying clinical message. [Note 2026-11-08: the display branch of this adoption was amended by the later adjudication below — its text and justifications remain here as they are, for documentation.]
Preuss 2007 (abstract-verified, مؤقت). Supporting pattern (abstract-verified): early paresis 18.7% falling to (Springborg 2013, n=139) — the same cohort measured at two time points. Review range for permanent paresis/paralysis: (Nahlieli 2017, full text — wording conflates injury incidence with permanence; use with the stated caution)
Transient vs permanent lower-lip depressor weakness
No defensible point estimate exists in the verified material
described as "possible but very rare" in the opened review; 0 events in the two series above; an abstract-only literature range of 0–1.4% exists (Carvalho 2015) but is not adopted
partial success (residual debris, symptom-free) 34.5%. Full-text corroboration of magnitude: complete elimination by ESWL alone in of cases (Nahlieli 2017 review, full text opened)