12-phase flow (12)

1. 1FRAME
   PCOS is a DIAGNOSIS OF EXCLUSION on the Rotterdam 2/3 rule: exclude thyroid / prolactin / non-classic CAH (17-OHP) / Cushing / acromegaly / androgen-secreting tumour / hypothalamic amenorrhoea / pregnancy BEFORE scoring Rotterdam (H/O/P, 2 of 3). Adolescent criteria are STRICTER (H+O, no morphology) (2023 Int’l PCOS Guideline, Teede; 2025 adolescent recommendations)
   inputs: age, reproductive_intent
   advance: Two-gate frame explicit: exclusion ladder named before Rotterdam scoring
2. 2ENTRY
   Oligo/amenorrhoea ± infertility; hirsutism/acne/alopecia; elevated androgens on testing; incidental polycystic ovaries on US; anovulatory subfertility; adolescent with hyperandrogenism + irregular cycles (2023 Int’l PCOS Guideline; Legro PPCOS-II NEJM 2014)
   inputs: age, reproductive_intent
   advance: Engine entered via a recognised trigger
3. 3CONTEXT
   Capture age (adolescent stricter-criteria flag), reproductive intent (fertility-seeking vs not — the master regimen fork), pregnancy status, BMI/weight, US-MEC inputs (VTE / migraine-with-aura / uncontrolled HTN / smoking ≥35 y), mood/eating-disorder screen, OSA/MASLD/CV risk (2023 Int’l PCOS Guideline)
   inputs: age, reproductive_intent, pregnancy_status, bmi_weight, vte_migraine_htn_smoking, mood_eating_disorder
   advance: Demographic, reproductive, contraceptive-eligibility and comorbidity context fully captured
4. 4RED_FLAGS
   RAPID virilisation (deepening voice, clitoromegaly, frontal balding, short symptom duration) + VERY high total testosterone (commonly cited >150–200 ng/dL / >5.2–7 nmol/L) = androgen-secreting ovarian/adrenal tumour — NOT PCOS, urgent imaging. Also: severe untreated metabolic decompensation, active eating disorder, or suicidality on mood screen (2023 Int’l PCOS Guideline; GWAS depression causal link 30566500)
   inputs: virilisation_tempo, total_and_free_testosterone
   actions: panel.androgen, workup.pcos
   advance: Tumour-virilisation pivot and psychiatric/metabolic emergencies screened and escalated if present
5. 5INITIAL_WORKUP
   EXCLUSION-FIRST labs: calculated free testosterone / free androgen index (biochemical H), TSH (thyroid mimic — high → endo.hypothyroidism.core.v1, suppressed → endo.hyperthyroidism.core.v1), prolactin (→ endo.hyperprolactinemia.core.v1), EARLY-FOLLICULAR 17-OHP (NCAH pivot), 75 g OGTT (preferred metabolic screen), ± overnight dexamethasone / IGF-1 only if Cushing/acromegaly features. CONDITIONAL DEPENDENCY 4 — the 17-OHP likelihood is jointly conditioned on (a) cycle timing (must be EARLY-FOLLICULAR — luteal-phase values are uninterpretable), (b) assay (LC-MS/MS yields lower values than immunoassay; an immunoassay-derived ≈6 nmol/L ≈ LC-MS/MS 3.3 nmol/L) and (c) basal-vs-ACTH-stimulated: a correctly-timed basal 17-OHP ≥6 nmol/L (≈2 ng/mL) screens NCAH at sens 96% / spec 96% → LR+ ≈24, LR− ≈0.04 (Chesover PMID 32110745); cosyntropin-STIMULATED 17-OHP AUC 0.95, sens 86% / spec 88% → LR+ ≈7.2 confirms, and a stimulated peak ≥30 nmol/L near-perfectly identifies genetically-confirmed NCAH (Ueland PMID 35725670: 12.5% of PCOS-like patients had NCAH). Without correct timing the LR is undefined. (2023 Int’l PCOS Guideline, Teede; PMIDs 32110745, 35725670)
   inputs: total_and_free_testosterone, serum_tsh, serum_prolactin, early_follicular_17ohp, ogtt_75g
   actions: panel.androgen, panel.tsh, panel.hormone, panel.glucose_a1c, calc.ferriman_gallwey, workup.pcos, workup.hypothyroidism, workup.hyperprolactinemia
   advance: All mimics excluded (thyroid/prolactin/17-OHP normal; no Cushing/acromegaly/tumour) — true PCOS candidate confirmed
6. 6BRANCHING_WORKUP
   Score Rotterdam ONLY after exclusion: H (modified Ferriman–Gallwey ≥ population threshold OR calculated free T/FAI), O (cycle <21 or >35 d, <8/year, or anovulation despite regular cycles), P (transvaginal ≥20 follicles/ovary with ≥8 MHz transducer OR volume ≥10 mL, OR AMH substitution IN ADULTS only). CONDITIONAL DEPENDENCY 1 — the AMH likelihood is assay- and age-conditioned: in ADULTS pooled sens 0.79 / spec 0.87 → LR+ ≈6.1, LR− ≈0.24 (van der Ham 2023-guideline MA PMID 38944177; Anand 13,509-subject MA PMID 35394100 DOR 24); when AMH SUBSTITUTES for the PCOM criterion within an otherwise-met Rotterdam set spec rises to 0.99 → LR+ ≈93 (Zhao MA PMID 31414908, DOR 1634) — the strongest wired discriminator, valid ONLY in adults with a standardised assay. In ADOLESCENTS the same test collapses (sens 0.66 / spec 0.78–0.81 → LR+ ≈3.0; Tsukui MA PMID 35675999) so AMH/US morphology is NOT applied. CONDITIONAL DEPENDENCY 2 — Rotterdam itself is conditionally NULL until GATE-1 exclusions (TSH both directions, prolactin, early-follicular 17-OHP) are normal; scoring before exclusion has no valid post-test meaning. CONDITIONAL DEPENDENCY 3 — if bleeding is heavy/prolonged the PALM-COEIN category gates routing: AUB-O with hyperandrogenism stays here, a structural (PALM) cause routes OUT to gyn.abnormal-uterine-bleeding.core.v1. Assign phenotype A–D. PCOM NOT applied within 8 y of menarche / in adolescents (2023 Int’l PCOS Guideline; 2025 adolescent recommendations; PMIDs 38944177, 31414908, 35394100, 35675999)
   inputs: amh_level, transvaginal_ultrasound
   actions: calc.rotterdam_pcos, calc.ferriman_gallwey, calc.palm_coein, panel.hormone, workup.pcos, workup.abnormal_uterine_bleeding
   advance: Rotterdam 2/3 met after exclusions AND phenotype A–D assigned (or PCOS excluded / adolescent at-risk flagged / structural AUB routed out)
7. 7DIFFERENTIAL
   MECE terminal split: PCOS (Rotterdam 2/3 after exclusion) vs non-classic CAH (17-OHP pivot) vs androgen-secreting ovarian/adrenal tumour (rapid virilisation + very high T pivot) vs hypothalamic amenorrhoea (low LH/FSH, low BMI, energy deficit, NO hyperandrogenism) vs thyroid dysfunction vs hyperprolactinaemia vs Cushing/acromegaly vs primary ovarian insufficiency. Phenotype A–D sub-classification within PCOS (2023 Int’l PCOS Guideline, Teede)
   inputs: total_and_free_testosterone, early_follicular_17ohp, serum_tsh, serum_prolactin
   advance: Terminal diagnosis + (if PCOS) phenotype A–D assigned
8. 8RISK_STRATIFICATION
   Phenotype metabolic tiers: A (H+O+P) and B (H+O) highest insulin-resistance/metabolic burden; C (H+P, ovulatory) intermediate; D (O+P, non-hyperandrogenic) lowest. CONDITIONAL DEPENDENCY 5 — metabolic-risk magnitude is CONDITIONED on phenotype: the GDM (OR 2.02), IGT/T2DM, dyslipidaemia and endometrial priors are not phenotype-flat — they concentrate in the hyperandrogenic A/B phenotypes and are attenuated in D, so the OGTT/lipid pre-test prior and the metformin/GLP-1 treatment threshold are set per phenotype, not uniformly. CONDITIONAL DEPENDENCY 6 — endometrial-cancer risk (OR 2.79; <54 y OR 4.05, Barry 2014) is conditioned on the bleeding state: it applies only under CHRONIC UNOPPOSED estrogen (oligo/amenorrhoea NOT on COC/progestin), and is removed once ≥4 withdrawal bleeds/year or continuous progestin is established. Overlay strata: obesity, OGTT result, dyslipidaemia, GDM risk, pregnancy, adolescent (2023 Int’l PCOS Guideline; PMIDs 35172306, 24688118)
   inputs: bmi_weight, ogtt_75g, reproductive_intent, age
   actions: calc.rotterdam_pcos, panel.lipid, panel.glucose_a1c
   advance: Phenotype + metabolic + reproductive + endometrial-risk tier documented and treatment goal set
9. 9TREATMENT
   GOAL-STRATIFIED (not a single ladder): (1) lifestyle first-line for ALL (≥5–10% weight loss improves ovulation/androgens/metabolic); (2) menstrual/hyperandrogenism → combined OC first-line behind a US-MEC gate; (3) hirsutism → spironolactone after ≥6 mo inadequate COC or COC contraindicated, ALWAYS with contraception; (4) metabolic/anovulation adjunct → metformin (LB OR 1.59); (5) weight → GLP-1 RA if obesity-dominant (BMI −2.42), STOP pre-conception; (6) FERTILITY → LETROZOLE first-line (LB 27.5% vs clomiphene 19.1%, RR 1.44, Legro PPCOS-II) → clomiphene ± metformin → gonadotropins/laparoscopic ovarian surgery → IVF (metformin reduces OHSS, RR 0.46); (7) endometrial protection → cyclic progestin/COC if oligo/amenorrhoea (2023 Int’l PCOS Guideline; PMIDs 25006718, 29183107, 33347618, 37583655, 39178623, 24688118)
   inputs: reproductive_intent, pregnancy_status, bmi_weight, vte_migraine_htn_smoking
   actions: calc.us_mec, panel.androgen
   advance: Goal-appropriate regimen documented (lifestyle ± COC/anti-androgen/metformin/GLP-1 RA, or letrozole-led fertility plan, or deliberate observation)
10. 10DISPOSITION
    Almost all managed outpatient (primary care / endocrinology / gynaecology / reproductive endocrinology by goal). Route exclusion-positive patients OUT: thyroid → endo.hypothyroidism.core.v1; prolactin → endo.hyperprolactinemia.core.v1; pregnant/pre-pregnancy → ob.gdm.core.v1 (GDM screening). Urgent referral for suspected androgen-secreting tumour; psychiatry for active eating disorder/suicidality (2023 Int’l PCOS Guideline)
    inputs: reproductive_intent, pregnancy_status
    advance: Setting set and exclusion/pregnancy/tumour/psychiatric routing executed
11. 11MONITORING
    OGTT every 1–3 years (annually if high-risk / prior GDM / pre-pregnancy); fasting lipids periodically; weight/BMI each visit; mood/eating-disorder reassessment; endometrial surveillance (ensure ≥4 withdrawal bleeds/year or progestin protection in oligo/amenorrhoea — endometrial-Ca OR 2.79); on COC re-check BP and re-screen US-MEC; on spironolactone monitor K and confirm ongoing contraception; on metformin watch GI tolerance + B12; on GLP-1 RA confirm discontinuation before any conception attempt (2023 Int’l PCOS Guideline; PMIDs 24688118, 36720508)
    inputs: ogtt_75g, bmi_weight, pregnancy_status
    actions: panel.glucose_a1c, panel.lipid, panel.androgen
    advance: Monitoring cadence individualised by phenotype, treatment goal and pregnancy status
12. 12FOLLOWUP
    Lifelong metabolic + reproductive surveillance. Pre-conception optimisation: weight, glycaemia, stop GLP-1 RA / teratogenic anti-androgen, plan letrozole-led ovulation induction, and route to ob.gdm.core.v1 for early + 24–28-week OGTT (GDM OR 2.02). Reinforce lifestyle, endometrial protection in oligo/amenorrhoea, mood support, and CV-risk follow-up; re-evaluate adolescent "at-risk" patients at maturity (2023 Int’l PCOS Guideline; 2025 adolescent recommendations; PMID 35172306)
    inputs: reproductive_intent, pregnancy_status
    actions: panel.glucose_a1c
    advance: Lifelong surveillance, pre-conception plan and adolescent re-evaluation booked