12-phase flow (12)

1. 1FRAME
   Hypothyroidism is a biochemical state requiring a mandatory phenotype split BEFORE levothyroxine: overt primary vs subclinical vs central vs non-thyroidal illness vs transient (thyroiditis/drug). Treatment target, treat-threshold, and safety steps differ by phenotype (ATA 2014 Jonklaas)
   inputs: serum_tsh, free_t4
   advance: Scope set to chronic ambulatory hypothyroidism with the 5-way phenotype frame explicit
2. 2ENTRY
   Elevated TSH on screen; low FT4 with low/normal TSH (central); low-specificity symptom cluster; established iatrogenic titration visit; pre-conception/pregnancy; thyroid-axis drug started (ATA 2014; ATA 2017; ETA 2013)
   inputs: serum_tsh, free_t4
   advance: Engine entered via a recognised trigger
3. 3CONTEXT
   Capture age (age-specific reference + elderly threshold), pregnancy/pre-conception status, weight (dosing), cardiac disease (start-low), known pituitary/hypothalamic disease (central pre-test), acute-illness status (NTIS), prior thyroid ablation/surgery (iatrogenic certainty), absorption-interacting and thyroid-toxic medications (ATA 2014; ATA 2017; ETA 2013)
   inputs: age, pregnancy_status, cardiac_disease, pituitary_or_hypothalamic_disease, acute_illness_status, concurrent_meds
   advance: Demographic, cardiac, pituitary, illness and medication context fully captured
4. 4RED_FLAGS
   Myxedema decompensation (hypothermia, depressed consciousness, hypoventilation, bradycardia, hyponatremia) → escalate OUT to endo.myxedema-coma.core.v1. Suspected central hypothyroidism with possible adrenal insufficiency → exclude/treat adrenal crisis and give glucocorticoid BEFORE levothyroxine; route to endo.hypopituitarism.core.v1 (ATA 2014; central CH not always mild — JCEM 2014)
   inputs: myxedema_decompensation_features, cortisol_axis
   actions: workup.hypothyroidism
   advance: Myxedema coma and adrenal-crisis-in-central screened and escalated/routed if present
5. 5INITIAL_WORKUP
   Serum TSH with reflex free T4 is the core test; the TSH/FT4 pattern is the decisive pivot: high TSH+low FT4 → overt primary; high TSH+normal FT4 → subclinical; low/normal TSH+low FT4 → central; low FT3/FT4 + low/normal TSH in acute illness → NTIS. Confirm a persistently abnormal TSH (repeat with TPO-Ab after 2–3 months for SCH) (ATA 2014 Jonklaas; ETA 2013 Pearce)
   inputs: serum_tsh, free_t4
   actions: panel.tsh, panel.thyroid, calc.tsh_reflex_interpretation, workup.hypothyroidism
   advance: TSH/FT4 pattern assigned to a phenotype and persistence confirmed
6. 6BRANCHING_WORKUP
   TPO-Ab for autoimmune aetiology + progression risk (~2× if positive); pregnancy → re-interpret against trimester-specific TSH reference; suspected central → full anterior pituitary axis + adrenal axis (route to hypopituitarism); transient suspicion (post-partum 2–12 mo, sub-acute thyroiditis, amiodarone/lithium/checkpoint-inhibitor/TKI/interferon) → characterise as transient vs permanent; associated lipid derangement screen (ATA 2014; ATA 2017; ETA 2013)
   inputs: tpo_antibody, pregnancy_status, pituitary_or_hypothalamic_disease, concurrent_meds
   actions: panel.thyroid, panel.lipid, workup.hypothyroidism
   advance: Aetiology, autoimmune status, central-axis and transient-vs-permanent characterised
7. 7DIFFERENTIAL
   MECE terminal split: overt primary hypothyroidism (Hashimoto / iatrogenic / iodine / congenital) vs subclinical primary vs central (secondary/tertiary) hypothyroidism vs non-thyroidal illness syndrome vs transient thyroiditis / drug-induced dysfunction. The TSH/FT4 pattern conditional on pregnancy / illness / pituitary status is decisive (ATA 2014 Jonklaas)
   inputs: serum_tsh, free_t4, tpo_antibody
   advance: Terminal phenotype + aetiology assigned
8. 8RISK_STRATIFICATION
   Subclinical treat-vs-observe engine: TSH >10 → treat (CHD mortality HR 1.58, 95% CI 1.10–2.27, Floriani/Rodondi); TSH 4.5–10 + symptoms → time-limited LT4 trial, stop if no benefit at 3–4 mo (Feller QoL SMD −0.11, 95% CI −0.25 to 0.03); age >65–70 → higher threshold/observe (TRUST null); oldest-old >80–85 → wait-and-see; pregnancy / TPO-Ab+ → lower threshold to treat (untreated pregnancy-loss RR 2.01 → treated OR 0.55) (ETA 2013 Pearce; TRUST Stott NEJM 2017; ATA 2017 Alexander)
   inputs: serum_tsh, free_t4, age, pregnancy_status, tpo_antibody
   actions: calc.tsh_reflex_interpretation
   advance: Treat-vs-observe decision made with the threshold rule that applies to this stratum
9. 9TREATMENT
   Overt primary: levothyroxine 1.6 µg/kg/day; elderly/CAD start 25–50 µg/day and up-titrate slowly; titrate to TSH at q6–8-week intervals; manage absorption interactions (separate from PPI/calcium/iron/fiber by 4 h, account for estrogen/malabsorption). Central: titrate to mid-normal FT4 NOT TSH, and only AFTER glucocorticoid if adrenal insufficiency possible. Subclinical: per the risk-stratification threshold. Pregnancy: increase dose ~20–30% as soon as confirmed, trimester-specific TSH targets. Persistent symptoms despite normal TSH: address chronic-disease support / associated autoimmune disease; LT4/LT3 combination is EXPERIMENTAL specialist-only (ratio 13:1–20:1, stop at 3 mo if no benefit). NTIS / transient: do NOT treat; retest after recovery / off insult. Myxedema coma → endo.myxedema-coma.core.v1 (ATA 2014 Jonklaas; ATA 2017 Alexander; 2012 ETA + 2021 ATA/BTA/ETA combination consensus)
   inputs: weight_kg, serum_tsh, free_t4, pregnancy_status, cardiac_disease
   actions: panel.tsh
   advance: Phenotype-appropriate levothyroxine plan (or deliberate no-treatment for NTIS/observed-SCH) documented
10. 10DISPOSITION
    Almost all managed outpatient. Route central hypothyroidism → endo.hypopituitarism.core.v1 (full axis + adrenal-first). Escalate myxedema decompensation → endo.myxedema-coma.core.v1 (ED/ICU). Pregnant patients co-managed with obstetrics (ATA 2014; ATA 2017)
    inputs: pituitary_or_hypothalamic_disease, myxedema_decompensation_features
    advance: Setting set and central/myxedema routing executed
11. 11MONITORING
    Primary: recheck TSH ~6–8 weeks after every dose change until stable, then every 6–12 months. Central: monitor FT4 (mid-normal target) — TSH is uninformative. Pregnancy: TSH approximately every 4 weeks in the first half then at least once mid-second/third trimester. Watch for over-replacement (suppressed TSH → AF risk HR 1.45 high-normal FT4; subclinical-hyperthyroid hip-fracture HR 1.36) (ATA 2014 Jonklaas; ATA 2017 Alexander; ETA 2013 Pearce)
    inputs: serum_tsh, free_t4, pregnancy_status
    actions: panel.tsh, panel.thyroid
    advance: Monitoring cadence individualised by phenotype and pregnancy status; over-replacement excluded
12. 12FOLLOWUP
    Lifelong replacement for permanent primary and central disease (transient may be weaned and retested). Reinforce adherence + consistent timing + absorption-interaction counselling; pre-conception planning for women of reproductive age (optimise pre-pregnancy, plan the ~20–30% bump); annual TSH once stable; reassess persistent-symptom patients (chronic-disease support, exclude associated autoimmune disease) before any combination trial (ATA 2014 Jonklaas; ATA 2017 Alexander; 2021 ATA/BTA/ETA combination consensus)
    inputs: pregnancy_status
    actions: panel.tsh
    advance: Lifelong / wean plan, adherence counselling and pre-conception plan booked