12-phase flow (12)

1. 1FRAME
   Thyrotoxicosis is an AETIOLOGY question to be answered before definitive therapy: Graves vs toxic multinodular vs toxic adenoma vs thyroiditis (subacute/painless/post-partum/amiodarone-type-2/checkpoint) vs amiodarone-type-1 vs factitious vs iodine-induced vs TSH-oma. RAIU + TRAb + thyroglobulin + Doppler is the Bayesian discriminator set. Thyroid storm is screened and ROUTED OUT to endo.thyroid-storm.core.v1 (ATA 2016 Ross)
   inputs: serum_tsh, free_t4_free_t3
   advance: Scope set to ambulatory thyrotoxicosis with the aetiology-discrimination frame explicit and storm excluded as the engine target
2. 2ENTRY
   Suppressed TSH on screen; thyrotoxic symptom cluster; new AF / unexplained tachycardia; orbitopathy eye signs; painful goitre; pregnant/pre-conception thyrotoxic; amiodarone/checkpoint surveillance; subclinical low-TSH-normal-FT4 (ATA 2016 Ross; ATA 2017 Alexander; EUGOGO 2021 Bartalena)
   inputs: serum_tsh, free_t4_free_t3
   advance: Engine entered via a recognised trigger
3. 3CONTEXT
   Capture age (apathetic geriatric, age≥65 lowers subclinical threshold), pregnancy/lactation (RAI contraindicated, PTU 1st-tri), iodine exposure (RAIU conditional dependence + Jod-Basedow), amiodarone/ICI, cardiac disease (AF/IHD), orbitopathy severity + smoking (EUGOGO disease-modifier), prior thionamide adverse reaction (ATA 2016 Ross; ATA 2017 Alexander; EUGOGO 2021 Bartalena)
   inputs: age, pregnancy_status, iodine_exposure, amiodarone_or_ici, cardiac_disease, orbitopathy_severity, smoking_status
   advance: Demographic, pregnancy, iodine, drug, cardiac, orbitopathy and smoking context fully captured
4. 4RED_FLAGS
   Thyroid storm (hyperthermia + altered mental status + CHF/GI dysfunction, Burch-Wartofsky high) → escalate OUT to endo.thyroid-storm.core.v1. Febrile sore throat on a thionamide → STOP drug + urgent CBC (agranulocytosis baseline ~0.2–0.5%, ATA 2016 27521067). The agranulocytosis prior is itself CONDITIONAL — not flat: (a) TIME — risk concentrates in the first ~90 days of therapy (most events 20–41 days from start) and recurs on rechallenge/dose-escalation, so a febrile patient in month 1–3 has a far higher pre-test probability than one stable at 2 years; (b) DOSE — higher methimazole dose raises risk (dose-dependent for MMI); (c) DRUG/AETIOLOGY — amiodarone-induced thyrotoxicosis multiplies ATD-agranulocytosis risk: incidence 22 vs 1.79/1000 person-years, adjusted HR 5.70 (95% CI 2.14–15.21) vs non-AIT thyrotoxicosis (Gershinsky Thyroid 2019 30648930) — treat the febrile AIT patient on a thionamide as agranulocytosis until the CBC excludes it. PTU fulminant hepatotoxicity / MMI cholestasis → stop + LFTs. Sight-threatening orbitopathy (dysthyroid optic neuropathy / corneal breakdown) → urgent IV glucocorticoid + ophthalmology (EUGOGO 2021) (ATA 2016 Ross; EUGOGO 2021 Bartalena; Gershinsky 30648930)
   inputs: storm_decompensation_features, cbc_for_agranulocytosis
   actions: workup.hyperthyroidism_deep, panel.cbc
   advance: Storm, agranulocytosis, hepatotoxicity and sight-threatening orbitopathy screened and escalated/routed if present
5. 5INITIAL_WORKUP
   Confirm with TSH + free T4 and free T3; the TSH/FT4/FT3 pattern is the first pivot (suppressed TSH + high FT4/FT3 = overt; suppressed TSH + normal FT4/FT3 = subclinical; non-suppressed TSH + high FT4 = TSH-oma/interference). TRAb at first visit is the Graves rule-in pivot with a CONDITIONAL-ON-ASSAY-GENERATION likelihood ratio (Tozzoli meta-analysis 22776786): 2nd-gen TRAb sens 97.1% / spec 97.4% → LR+ ≈ 37, LR− ≈ 0.030; 3rd-gen sens 98.3% / spec 99.2% → LR+ ≈ 123, LR− ≈ 0.017 (Tozzoli reports a TRAb-positive subject 1367–3420× more likely to have Graves at study prevalence). Report which assay generation the lab uses before applying the LR — a borderline 2nd-gen result is less decisive than the same 3rd-gen result. calc.tsh_reflex_interpretation structures the TSH/FT4/FT3 pattern (ATA 2016 Ross; ETA 2018 Kahaly; Tozzoli 22776786)
   inputs: serum_tsh, free_t4_free_t3, trab_tsi
   actions: panel.tsh, panel.thyroid, panel.hormone, calc.tsh_reflex_interpretation, workup.hyperthyroidism_deep
   advance: Overt-vs-subclinical pattern assigned and TRAb resulted
6. 6BRANCHING_WORKUP
   TRAb-positive → Graves (proceed to treatment, no scan mandatory). TRAb-negative or equivocal → radioiodine uptake & scan: diffuse↑ Graves / focal-patchy↑ toxic adenoma or TMNG (→ workup.thyroid_nodule) / LOW uptake → thyroiditis vs factitious vs iodine-induced. RAIU LR set (ATA 2016 Ross 27521067): elevated diffuse/focal uptake LR+ ≈ 20–25 for autonomous overproduction (Graves/toxic nodular) vs near-zero (<3%) suppressed uptake LR+ ≈ 18–24 for destructive/exogenous/iodine-induced thyrotoxicosis — RAIU is the strongest single wired discriminator in the dossier (the master MECE split). CONDITIONAL DEPENDENCY (critical): RAIU is UNINTERPRETABLE when the iodine pool is expanded — recent iodinated CT contrast, amiodarone, or povidone-iodine drives uptake artefactually LOW regardless of true aetiology, collapsing its LR to ~1; in that state and in pregnancy (RAIU contraindicated) the discriminator stack falls back to TRAb (assay-generation LR above), thyroglobulin, and color-flow Doppler. Low-RAIU branch: thyroglobulin (LOW = factitious — high LR+ for exogenous-hormone vs all endogenous causes), ESR/CRP + neck tenderness (subacute de Quervain LR+ high), iodine history (Jod-Basedow), post-partum timing. Thyroid ultrasound + Doppler separates amiodarone type 1 (hypervascular → thionamide) from type 2 (absent flow → glucocorticoid-responsive) — Doppler is the SUBSTITUTE master pivot exactly when iodine load has voided RAIU (Tanda 2008 18410546) (ATA 2016 Ross; Tanda 2008)
   inputs: radioiodine_uptake_scan, thyroglobulin, esr_crp, thyroid_ultrasound_doppler
   actions: workup.thyroid_nodule, panel.thyroid, panel.inflammation, workup.hyperthyroidism_deep
   advance: RAIU pattern + TRAb + thyroglobulin + Doppler resolve a terminal aetiology
7. 7DIFFERENTIAL
   MECE terminal aetiology: Graves (TRAb+, diffuse↑RAIU, orbitopathy) vs toxic multinodular goitre (patchy↑RAIU) vs toxic adenoma (focal↑RAIU) vs thyroiditis — subacute de Quervain (painful, ↑ESR, low RAIU) / painless / post-partum / amiodarone type 2 / checkpoint-inhibitor — vs amiodarone type 1 (iodine-induced) vs factitious (LOW thyroglobulin) vs iodine-induced Jod-Basedow vs TSH-secreting adenoma (non-suppressed TSH). RAIU is the master split; TRAb / tenderness+ESR / thyroglobulin / Doppler are the named pivots (ATA 2016 Ross)
   inputs: radioiodine_uptake_scan, trab_tsi, thyroglobulin
   advance: Terminal aetiology assigned (RAIU + TRAb + Tg + Doppler decisive)
8. 8RISK_STRATIFICATION
   Overt vs subclinical. SUBCLINICAL TREAT-DECISION is an explicitly CONDITIONAL threshold — the same TSH value carries a different action depending on age + AF/cardiac + bone status (the threshold LOWERS as risk modifiers accumulate): treat if TSH <0.1 mIU/L AND (age ≥65, OR AF/cardiac disease, OR osteoporosis, OR symptomatic postmenopausal); TSH 0.1–0.4 mIU/L → treat only if a high-risk modifier is present, else observe + repeat in 3–6 mo; TSH 0.1–0.4 with NO modifier → observe. Quantitative modifier weights wired in: endogenous subclinical hyperthyroidism (TSH <0.10) hip-fracture HR 1.36 (95% CI 1.13–1.64) and any-fracture HR ~1.28 (Blum JAMA 2015 26010634); higher fT4 (toward/above range) incident-AF HR 1.45 (95% CI 1.26–1.66) highest vs lowest quartile in the Thyroid Studies Collaboration IPD (Baumgartner Circulation 2017 29061566) — so an osteoporotic or AF-prone patient is treated at a TSH where a low-risk 40-year-old is merely observed. Graves orbitopathy EUGOGO activity/severity tier (mild vs moderate-severe active vs sight-threatening) selects the orbitopathy ladder and contraindicates RAI in active/severe GO. Remission likelihood by TRAb / goitre / age / T3 (Azizi risk-score, c-index 0.78, 38165576) (ATA 2016 Ross; EUGOGO 2021 Bartalena; Blum 26010634; Baumgartner 29061566)
   inputs: serum_tsh, free_t4_free_t3, age, cardiac_disease, orbitopathy_severity
   actions: calc.tsh_reflex_interpretation
   advance: Overt/subclinical-treat decision made and orbitopathy severity tier set
9. 9TREATMENT
   Aetiology gate first: destructive thyroiditis (subacute / painless / post-partum / amiodarone-type-2 / checkpoint) and factitious get NO thionamide — beta-blocker ± NSAID/steroid (subacute) and supportive only. Graves / toxic nodular: beta-blocker symptom control (propranolol also blunts T4→T3) + methimazole FIRST-LINE (titration or block-replace), titrate to euthyroid; definitive therapy = RAI (contraindicated in pregnancy / severe active orbitopathy / lactation) or thyroidectomy (large goitre / coexisting suspicious nodule / severe active orbitopathy / patient choice / high-volume surgeon). PTU only 1st-trimester pregnancy & storm (agranulocytosis ~0.2–0.5% both drugs; PTU hepatotoxicity > MMI). Graves orbitopathy ladder: smoking cessation + selenium (mild) → IV methylprednisolone ± mycophenolate (moderate-severe active, EUGOGO 4.5 g/12 wk) → teprotumumab (OPTIC proptosis response 83% vs 10%, NNT 1.36; first-line for proptosis-predominant per ATA/ETA 2022). Pregnancy ATD-SWITCH branch (quantified): PTU in the 1st trimester then switch back to MMI from the 2nd trimester, lowest dose to upper-normal maternal FT4, TRAb for fetal/neonatal risk — driven by the Danish nationwide cohort (n=817,093): MMI/CMZ-exposed combined adjusted OR 21.8 (95% CI 13.4–35.4) for the characteristic embryopathy (choanal/oesophageal atresia, omphalocele, omphalomesenteric duct anomaly, aplasia cutis), overall MMI/CMZ birth-defect OR 1.66 (1.35–2.04), PTU OR 1.41 (1.03–1.92), and shifting MMI↔PTU within early pregnancy OR 1.82 (1.08–3.07) — hence switch BEFORE conception or as early as possible, not mid-1st-trimester (Andersen JCEM 2013 24151287). Amiodarone: type 1 thionamide ± perchlorate, type 2 glucocorticoid — and counsel that ATD-agranulocytosis risk is ~6× higher in amiodarone-induced thyrotoxicosis (Gershinsky 30648930) so CBC vigilance is intensified (ATA 2016 Ross; ETA 2018 Kahaly; ATA 2017 Alexander; EUGOGO 2021 Bartalena; OPTIC 2020 Douglas; Andersen 24151287; Gershinsky 30648930)
   inputs: trab_tsi, pregnancy_status, orbitopathy_severity, lft_baseline
   actions: workup.hyperthyroidism_deep
   advance: Aetiology-appropriate plan (thionamide+BB+definitive for Graves/nodular, OR supportive-only for destructive/factitious, OR orbitopathy ladder) documented
10. 10DISPOSITION
    Almost all managed outpatient endocrinology. Escalate thyroid storm / impending decompensation → endo.thyroid-storm.core.v1 (ED/ICU). Surgical candidates → high-volume thyroid surgeon; sight-threatening orbitopathy → urgent oculoplastics + endocrinology. Pregnant patients co-managed with obstetrics (ATA 2016 Ross; ATA 2017 Alexander)
    inputs: storm_decompensation_features, orbitopathy_severity
    advance: Setting set and storm / surgical / orbitopathy / obstetric routing executed
11. 11MONITORING
    On thionamide: TFTs (FT4/FT3 — TSH lags) every 4–6 weeks until euthyroid then every 2–3 months; baseline + symptom-driven CBC (agranulocytosis) and LFTs (hepatotoxicity); counsel febrile-sore-throat STOP rule. TRAb at 12–18 months predicts Graves remission vs relapse (Azizi). Post-RAI / post-thyroidectomy: surveil for the near-universal hypothyroid endpoint → route to endo.hypothyroidism.core.v1. Subacute thyroiditis: expect hypothyroid phase then recovery — serial TFTs. Watch over-treatment (iatrogenic hypothyroidism) (ATA 2016 Ross; ETA 2018 Kahaly)
    inputs: free_t4_free_t3, serum_tsh, trab_tsi
    actions: panel.tsh, panel.thyroid, panel.cbc, panel.lft
    advance: Monitoring cadence individualised by aetiology and therapy; agranulocytosis/hepatotoxicity counselling given; post-ablation hypothyroid surveillance booked
12. 12FOLLOWUP
    Graves: assess remission after 12–18 mo MMI (TRAb-guided) — relapse → definitive therapy or long-term low-dose MMI (Azizi: long-term recurrence 17% vs conventional 56%; juvenile 4-yr cure 88% vs 33%). Lifelong levothyroxine after RAI/thyroidectomy (route to endo.hypothyroidism.core.v1). Pregnancy: TRAb in pregnancy → fetal/neonatal thyrotoxicosis surveillance; pre-conception counselling (switch to PTU plan / consider definitive before pregnancy). RAI long-term modest solid-cancer-mortality counselling (Kitahara). Return precautions: fever/sore throat, jaundice, eye pain/vision change, palpitations, pregnancy (ATA 2016 Ross; ETA 2018 Kahaly; ATA 2017 Alexander)
    inputs: trab_tsi, pregnancy_status
    actions: panel.tsh
    advance: Remission/definitive-therapy decision, lifelong-LT4 plan, pregnancy-TRAb surveillance and return precautions booked