12-phase flow (12)

1. 1FRAME
   Apply Rome IV POSITIVE diagnosis of functional dyspepsia: ≥1 of bothersome postprandial fullness, early satiation, epigastric pain, epigastric burning, with NO structural disease (including upper endoscopy where indicated) explaining the symptoms; onset ≥6 months ago, criteria active ≥3 months. Assign subtype — PDS (postprandial fullness/early satiation ≥3 days/wk), EPS (epigastric pain/burning ≥1 day/wk), or PDS-EPS overlap. NOT a diagnosis of exclusion once the test-and-treat/EGD pathway is satisfied (Rome IV; Stanghellini 2016 PMID 27147122; ACG/CAG 2017 PMID 28631728)
   inputs: postprandial_fullness_early_satiety, epigastric_pain_burning, symptom_duration
   actions: workup.dyspepsia
   advance: Rome IV criteria met and PDS / EPS / overlap subtype assigned
2. 2ENTRY
   Recognise the presenting trigger — bothersome postprandial fullness/early satiation (PDS), epigastric pain/burning (EPS), known-FD therapy-review visit, positive H. pylori test with dyspepsia, or a normal EGD already performed (ACG/CAG 2017)
   inputs: postprandial_fullness_early_satiety
   advance: one entry trigger present
3. 3CONTEXT
   Capture age/sex (drives the test-and-treat vs early-EGD fork), family history of upper-GI cancer, NSAID/aspirin use, psychological comorbidity (PHQ-9/GAD-7 — bidirectional gut-brain axis), heartburn/IBS overlap, pregnancy status, penicillin allergy, prior dyspepsia therapy and response (ACG/CAG 2017; Maastricht VI 2022 PMID 35944925)
   inputs: age, sex, family_history_upper_gi_cancer, nsaid_use, psych_comorbidity, pregnancy_status
   actions: calc.phq9, calc.gad7
   advance: context captured
4. 4RED_FLAGS
   ALARM FEATURES override a functional diagnosis and mandate upper endoscopy ± targeted organic workup: age ≥60 with new-onset dyspepsia, unintentional weight loss, dysphagia, odynophagia, overt GI bleeding / iron-deficiency anaemia, persistent vomiting, palpable abdominal mass or lymphadenopathy, first-degree family history of upper-GI cancer. §5.5.2 BAYESIAN — PRE-TEST PRIOR: Rome IV FD population prevalence is ~10% with subtype split 61% PDS / 18% EPS / 21% overlap (Aziz/Sperber Lancet GH 2018 PMID 29396034); in a dyspeptic patient age <60 WITHOUT alarm features the prior probability of upper-GI malignancy is <1% and the EGD tumour-yield is only ~0.46% (vs 2.84% across an all-comers 2-week-wait cohort — PMID 31567636), so the negative organic LR is low and empiric non-invasive test-and-treat is justified WITHOUT upfront EGD. ORGANIC RULE-IN (LR+ ≥20, the load-bearing override): DYSPHAGIA is the single alarm feature whose presence overrides Rome IV positivity — in the Vakil systematic review & meta-analysis (Gastroenterology 2006 PMID 16890592, reference standard = EGD ± histology, pooled cross-sectional/cohort design) dysphagia in patients aged 36-74 is the ONLY alarm with a positive likelihood ratio >10 for upper-GI malignancy (alarm-feature specificity 93.8-99.8%, dysphagia PPV ~66.7%); against the <1% no-alarm FD prior an LR of this magnitude with high-prevalence cohorts (e.g. dysphagia OR ~3.1, age >55 OR ~9.5) drives a binned, age-conditional LR+ that exceeds 20 for new dysphagia in older patients — mandating EGD regardless of how cleanly Rome IV FD criteria are met. CONDITIONAL DEPENDENCY 1 (alarm-feature PPV ∝ age): the malignancy PPV of any alarm feature is strongly conditional on age — negligible in young patients, materially elevated ≥55-60 (age >55 OR ~9.5; alarm features and age are NOT independent, so do not multiply their LRs as if independent — use the age-binned alarm LR). ANY alarm feature, or age ≥60, raises the posterior probability of organic disease (peptic ulcer, gastric/oesophageal cancer) above the test-and-treat threshold and mandates EGD (ACG/CAG 2017 PMID 28631728; Vakil PMID 16890592)
   inputs: alarm_features, age, family_history_upper_gi_cancer
   actions: workup.ugib, workup.colorectal_screening, workup.achalasia, workup.eoe_workup
   advance: no alarm feature and age <60 (proceed test-and-treat), or alarm/age ≥60 rerouted to EGD/organic pathway
5. 5INITIAL_WORKUP
   Test-and-treat fork (the load-bearing decision; threshold = age cut-off 60 AND the alarm rule): age <60 AND no alarm features → non-invasive H. pylori testing (urea breath test or stool antigen), eradicate if positive — this strategy is at/above the cost-effectiveness threshold below the age cut-off (Ford test-and-treat-vs-endoscopy individual-patient-data MA PMID 15940619; ACG/CAG 2017 PMID 28631728); CBC for iron-deficiency anaemia (alarm); CMP/LFT and TSH if metabolic/hepatobiliary mimic or atypical emptying symptoms. Age ≥60 OR any alarm feature → upfront EGD with biopsy (rule out peptic ulcer, gastric/oesophageal malignancy, erosive disease; obtain gastric biopsy for H. pylori). §5.5.2 CONDITIONAL DEPENDENCY 2 (H. pylori test VALIDITY ∝ drug exposure & bleeding — the test result is NOT conditionally independent of recent therapy): UBT and stool-antigen sensitivity collapse to clinically unacceptable FALSE-NEGATIVE rates on PPIs (and on antibiotics/bismuth, and with active upper-GI bleeding) — the patient MUST be off PPI ≥2 weeks and off antibiotics/bismuth ≥4 weeks or a negative UBT/stool antigen has a near-1.0 false-negative LR and must not be trusted (Gatta PPI/antacid UBT-and-stool MA PMID 15128344; ACG H. pylori 2024 PMID 39626064); serology cannot distinguish active from past infection (low specificity) and is NOT preferred. Empiric PPI may begin pending H. pylori results in test-and-treat patients (timing the eradication test accordingly) (ACG/CAG 2017 PMID 28631728; Maastricht VI 2022 PMID 35944925)
   inputs: h_pylori_test, cbc, cmp_lft, tsh
   actions: panel.cbc, panel.cmp, panel.lft, cascade.labs_command
   advance: H. pylori status established (test-and-treat) or EGD completed (≥60 / alarm)
6. 6BRANCHING_WORKUP
   Second-tier directed testing when warranted: positive H. pylori → eradicate then confirm cure by urea breath test / stool antigen ≥4 weeks post-therapy (off PPI ≥2 wk) — confirmation of eradication is what predicts symptom benefit (eradicated NNT 4.5 vs ITT NNT 9-14; Ford 2022 PMID 35022266); predominant heartburn/regurgitation or PPI-refractory acid symptoms → GERD pathway (ambulatory pH/impedance off therapy); dysphagia → achalasia (HRM) / EoE (EGD biopsies ≥6 levels) pathway; refractory PDS with vomiting / suspected gastroparesis → gastric-emptying scintigraphy (4-h solid) to distinguish FD from gastroparesis; concurrent IBS bowel symptoms → IBS pathway. §5.5.2 CONDITIONAL DEPENDENCY 3 (organic EGD-yield ∝ age/alarm/region): even within Rome-IV-positive FD the diagnostic yield of EGD for clinically significant organic disease is conditional — very low in young no-alarm patients, rising with age, alarm features, and high background-prevalence regions (Rome IV criteria do NOT reliably exclude organic disease without the age/alarm gate; PMID 37409327, PMID 31567636) — so a Rome-IV-positive label does not lower the organic prior independently of the age/alarm stratum. CONDITIONAL DEPENDENCY 4 (symptom-overlap ∝ coexisting GERD/IBS — overlapping disorders inflate symptom counts and are NOT independent): FD-GERD overlap is frequent and the EGD yield in Rome IV FD is modified by coexisting reflux (PMID 35783630); ~30% of FD also meets Rome IV IBS — the gut-brain overlap means heartburn/bowel symptoms are correlated with FD, so a positive overlap screen does not independently raise the organic prior but does redirect therapy (shared neuromodulation) and routing. INDEPENDENCE NOTE: the only test treated as approximately conditionally independent of FD symptom status is the EGD/histology reference standard itself; alarm features, age, prior drug exposure, and overlap symptoms are explicitly modelled as dependent and are combined via age/exposure-binned LRs, never multiplied as independent. EGD is NOT repeated in the absence of new alarm features (ACG/CAG 2017; AGA white paper 2017 PMID 28529164)
   inputs: h_pylori_test, heartburn_regurgitation_overlap, ibs_overlap_bowel_symptoms
   actions: workup.ugib, workup.achalasia, workup.eoe_workup, workup.ibs_rome_iv
   advance: H. pylori eradication confirmed and structural / motility mimics excluded or rerouted
7. 7DIFFERENTIAL
   Distinguish functional dyspepsia (PDS / EPS / overlap) from: peptic ulcer disease (H. pylori / NSAID), GORD and FD-GERD overlap, gastric/oesophageal malignancy, gastroparesis (delayed 4-h scintigraphy — distinct entity though overlapping), chronic nausea-vomiting syndrome, biliary/pancreatic disease (chronic pancreatitis, sphincter-of-Oddi, gallstones), medication-induced dyspepsia (NSAID/aspirin/iron/bisphosphonate), coeliac disease, eosinophilic gastritis/EoE, IBS overlap, and centrally-mediated abdominal pain. Confirm Rome IV subtype and H. pylori-associated vs H. pylori-negative phenotype (Rome IV; ACG/CAG 2017)
   inputs: heartburn_regurgitation_overlap, ibs_overlap_bowel_symptoms
   advance: FD subtype confirmed and organic / overlap conditions excluded or co-managed
8. 8RISK_STRATIFICATION
   Stratify symptom burden (mild/moderate/severe by impact on quality of life, weight loss, healthcare utilisation), psychological comorbidity (PHQ-9 / GAD-7 — drives gut-brain neuromodulator and psychological-therapy selection; screen Q9 for suicidality), and refractoriness (failure of H. pylori eradication + PPI + ≥1 neuromodulator → GI/neurogastroenterology referral). Significant unintentional weight loss or dehydration re-enters the alarm/organic pathway (ACG/CAG 2017; Tack 2016 PMID 26538208)
   inputs: psych_comorbidity
   actions: calc.phq9, calc.gad7
   advance: symptom-burden tier + psychological burden + refractoriness documented
9. 9TREATMENT
   Stepwise, subtype-directed (ACG/CAG 2017; AGA white paper 2017): (1) H. pylori test-and-treat — eradicate if positive (potentially curative subset; bismuth quadruple preferred where clarithromycin resistance >15% or prior macrolide, clarithromycin triple only where resistance low/known-susceptible; confirm cure) — Ford 2022 PMID 35022266 NNT cure 14 / improvement 9 / eradicated 4.5. (2) PPI standard dose 4-8 weeks (especially EPS) — Pinto-Sanchez Cochrane PMID 29161458 NNTB 13; DEPRESCRIBE at 8 weeks if no response (protocol.deprescribing_ppi — step-down → on-demand → stop with rebound-acid counselling). (3) Prokinetic for PDS-predominant — metoclopramide SHORT-COURSE only (≤12 weeks; tardive-dyskinesia / EPS boxed warning), domperidone (QT — ECG/availability), acotiamide/itopride (regional availability — no US RxNorm). (3-prokinetic effect size: prokinetic vs placebo RR for global FD symptoms 0.81, 95% CI 0.74-0.89, NNT 7 — Pittayanon/Yuan Cochrane PMID 30335201). (4) Gut-brain neuromodulators — psychotropic-drug FD class effect: RR symptoms not improving 0.78, 95% CI 0.68-0.91, NNT 6, benefit limited to TCAs/antipsychotics (Ford psychotropic MA PMID 26567029); TCA amitriptyline/nortriptyline for refractory EPS (FDTT/Talley 2015 PMID 25921377 — adequate relief 53% amitriptyline vs 40% placebo vs 38% escitalopram, n=292; ulcer-like/EPS FD the responders; NO benefit if delayed gastric emptying — PMID 26116797), mirtazapine 15 mg for early satiety + weight loss (Tack 2016 PMID 26538208 — early satiation/QoL/nutrient-tolerance/weight all improved with large effect sizes; +3.9 kg / 6.4% body weight at 8 wk in FD with >10% weight loss), buspirone 10 mg TID for PDS / impaired accommodation (Tack 2012 PMID 22813445 — overall severity 7.5±1.3 vs 11.5±1.2 placebo P<.005, accommodation 229±28 vs 141±32 mL P<.05). (5) Psychological therapy (CBT, gut-directed hypnotherapy, psychotherapy — significant reduction in GI symptoms/anxiety across 16 RCTs/1550 pts, PMID 34105186). Adjuncts: H2RA, STW5/Iberogast. SPECIAL-POPULATION TREATMENT GATES: pregnancy/lactation — misoprostol CONTRAINDICATED (abortifacient), tetracycline & fluoroquinolone H. pylori components AVOIDED; first-trimester PPI exposure NOT associated with significant malformation excess (OR 1.10, 95% CI 0.95-1.26, PMID 37269915; Gill MA OR 1.12, 95% CI 0.86-1.45, PMID 19491869) so dietary measures ± lowest-effective PPI are the pregnancy-tier choice and H. pylori eradication is deferred until postpartum where possible; elderly — metoclopramide and TCAs are STOPP/START v3 potentially-inappropriate (extrapyramidal/anticholinergic/fall risk; PMID 37256475) → prefer secondary-amine nortriptyline at low dose, cap metoclopramide ≤12 wk or avoid; renal/hepatic impairment — reduce/avoid metoclopramide (renal accumulation→EPS) and adjust H. pylori-regimen components (eGFR via calc.ckd_epi_2021); penicillin allergy — amoxicillin-free eradication (bismuth quadruple first-line, levofloxacin-based alternative; ACG H. pylori 2024 PMID 39626064); paediatric pointer — paediatric FD is a separate route (not this adult engine). Stop offending NSAID/aspirin/iron/bisphosphonate; modern duodenal-eosinophilia mechanism supports PPI/dietary effects (Wauters 2021 PMID 33346007 — pantoprazole 40 mg ×4 wk reduces duodenal eosinophils/mast cells/permeability)
   inputs: epigastric_pain_burning, postprandial_fullness_early_satiety, penicillin_allergy, creatinine, pregnancy_status
   actions: protocol.deprescribing_ppi
   advance: subtype-directed stepwise regimen + gut-brain plan agreed with patient
10. 10DISPOSITION
    Outpatient primary-care management for the majority (test-and-treat + empiric stepwise therapy); refer to GI for age ≥60/alarm features (EGD), diagnostic uncertainty, refractory symptoms after H. pylori eradication + PPI + a neuromodulator, suspected gastroparesis, or restricted-prokinetic stewardship; psychiatry/psychology referral for severe affective comorbidity or positive suicidality screen (ACG/CAG 2017)
    inputs: alarm_features
    advance: care setting and referrals set
11. 11MONITORING
    Reassess symptom response at 4-8 weeks per agent; confirm H. pylori eradication by urea breath test / stool antigen ≥4 weeks post-therapy (off PPI ≥2 wk); review PPI at 8 weeks and DEPRESCRIBE if no benefit (protocol.deprescribing_ppi); metoclopramide — screen for extrapyramidal/tardive-dyskinesia signs and cap exposure ≤12 weeks; domperidone — baseline/interval QTc; TCA — anticholinergic burden, QTc at higher dose, fall/sedation risk (elderly STOPP); PHQ-9/GAD-7 trend on neuromodulator; renal function for dose adjustment (calc.ckd_epi_2021); re-screen for emergent alarm features (any new alarm → re-enter RED_FLAGS / EGD) (ACG/CAG 2017; ACG H. pylori 2024 PMID 39626064)
    inputs: creatinine, psych_comorbidity, h_pylori_test
    actions: calc.phq9, calc.gad7, calc.ckd_epi_2021, protocol.deprescribing_ppi
    advance: response assessed, eradication confirmed, and safety monitoring documented
12. 12FOLLOWUP
    Mild disease: review at 4-8 weeks then as needed with self-management education and return precautions (new alarm feature). Moderate-severe / refractory: structured follow-up q4-12 weeks during stepwise escalation; reinforce diet/lifestyle (small low-fat meals, reduce caffeine/alcohol, weight-bearing meals), the chronic relapsing-remitting natural history, and multidisciplinary coordination (dietitian, psychology, neurogastroenterology). Reassess the diagnosis if response is atypical or alarm features emerge (ACG/CAG 2017; AGA white paper 2017 PMID 28529164)
    inputs: psych_comorbidity
    advance: follow-up interval, self-management plan, and multidisciplinary referrals scheduled