12-phase flow (12)

1. 1FRAME
   Classify: HAP = pneumonia ≥48 h after admission AND not incubating at admission; VAP = pneumonia ≥48 h after endotracheal intubation. Distinguish from ventilator-associated tracheobronchitis (VAT — purulent secretions + clinical signs but NO new infiltrate). Exclude COVID and other viral etiologies (out of scope — non-COVID engine). HCAP is a retired category (Kalil PMID 27418577)
   advance: HAP vs VAP vs VAT classification made + non-COVID confirmed
2. 2ENTRY
   Triggered by NEW or progressive radiographic infiltrate ≥48 h PLUS clinical deterioration PLUS ≥1 of: fever/hypothermia, leukocytosis/leukopenia, purulent secretions, rising FiO2/PEEP. The new infiltrate is necessary — its absence redirects to VAT or non-infectious mimic
   inputs: age
   advance: New-infiltrate + clinical + (pending) micro triad initiated
3. 3CONTEXT
   Capture severity vitals + the MDR-risk stratification that decides empiric breadth: prior IV abx ≤90 d (strongest), septic shock at VAP onset, ARDS preceding VAP, ≥5 d hospitalisation before VAP, acute RRT before VAP, prior MRSA/Pseudomonas, and — primary — the LOCAL UNIT ANTIBIOGRAM (MRSA ≥10-20% or GNR monotherapy resistance >10%). Capture immunocompromise, aspiration risk, post-surgical status, hepatic/renal function, allergy/interactions (Kalil PMID 27418577)
   inputs: sbp, rr, spo2, temp, fio2, prior_iv_abx_90d, mrsa_colonisation_or_prior, pseudomonas_colonisation_or_prior, septic_shock_or_high_mortality_risk, hospital_unit_resistance_pattern, immunocompromised, aspiration_risk, post_surgical_status, liver_disease, current_meds
   advance: MDR-risk tier + severity + comorbidity profile captured
4. 4RED_FLAGS
   Septic shock, ARDS, hemodynamic instability, multi-organ failure → immediate SSC 2021 sepsis bundle within 1 h (cultures before abx, lactate, 30 mL/kg crystalloid, broad empirics). Shock or lactate ≥4 mandates DUAL anti-pseudomonal coverage (Kalil PMID 27418577; SSC 2021 Evans PMID 34599691)
   inputs: sbp, spo2
   actions: protocol.septic_shock
   advance: Sepsis bundle initiated or no shock physiology
5. 5INITIAL_WORKUP
   CXR (new/progressive infiltrate); CBC + diff; BMP + Cr; lactate; RESPIRATORY culture BEFORE antibiotics (BAL quantitative if intubated OR endotracheal aspirate semi-quantitative — NO mortality difference per Canadian Critical Care 2006; both acceptable per Kalil); blood cultures × 2; baseline procalcitonin; Legionella/Strep urinary antigen if community-onset overlap. Do NOT delay antibiotics for sampling if septic
   inputs: cxr, wbc, creatinine, lactate, sputum_culture_or_BAL, blood_cultures
   actions: workup.hap_vap, panel.cbc, panel.renal
   advance: Cultures sent + empirics started within 1 h if septic
6. 6BRANCHING_WORKUP
   If non-resolving at 72 h: CT chest (abscess, empyema, non-resolving consolidation); bronchoscopy/BAL with quantitative culture; thoracentesis if effusion (rule out empyema → source control). Actively exclude mimics that look like non-resolving pneumonia: PE (disproportionate hypoxia, route pulm context), ARDS (bilateral, P/F-defined), atelectasis (post-op, rapid change with recruitment), cardiogenic edema (BNP/echo/diuretic response). If immunocompromised: galactomannan/β-D-glucan, PJP PCR, CMV, respiratory viral panel
   inputs: immunocompromised
   actions: panel.lft, panel.inflammation
   advance: Branch tests obtained when triggered OR alternative dx routed
7. 7DIFFERENTIAL
   Pathogen axis: MDR Gram-negative (Pseudomonas, Acinetobacter, ESBL Klebsiella/E. coli, CRE) vs MRSA vs MSSA vs Enterobacterales vs anaerobes (aspiration overlap) vs Legionella vs respiratory virus — per Kalil Table 3. Disease axis: HAP/VAP vs VAT (no infiltrate) vs ARDS vs aspiration pneumonia vs atelectasis vs cardiogenic pulmonary edema vs PE — pivot is new infiltrate + clinical + micro positivity
   advance: Working pathogen category + disease entity assigned
8. 8RISK_STRATIFICATION
   qSOFA / SOFA (Sepsis-3 Singer JAMA 2016) for sepsis severity; CPIS noted but limited (poor specificity, inter-observer variability — NOT used as a stand-alone gate, Kalil); identify septic shock / high mortality risk for DUAL anti-pseudomonal coverage. Melsen IPD meta-analysis: VAP attributable mortality ~13% overall, higher in surgical + mid-severity patients (PMID 23622939)
   inputs: sbp, rr
   actions: calc.qsofa, calc.sofa
   advance: Severity tier + MDR-risk tier classified
9. 9TREATMENT
   Empiric anti-pseudomonal β-lactam (pip-tazo 4.5 g IV q6h ext-infusion OR cefepime 2 g IV q8h OR meropenem 1 g IV q8h) keyed to local antibiogram; ADD MRSA cover (vancomycin AUC24/MIC 400-600 — Rybak 2020 PMID 32191793 — OR linezolid 600 mg q12h) if MRSA risk/unit prevalence; ADD a second anti-pseudomonal class (aminoglycoside or anti-pseudomonal FQ) only if septic shock OR ≥1 MDR-GNR risk factor OR unit GNR monotherapy resistance >10%. Renal/hepatic dose-adjust. De-escalate at 48-72 h on culture/susceptibility; 7-day course unless complicated (Chastre PMID 14625336; Pugh PMID 26301604; Kalil PMID 27418577)
   inputs: creatinine, liver_disease
   actions: calc.ckd_epi_2021, calc.cockcroft_gault
   advance: Empirics + renal/hepatic adjust + de-escalation plan documented
10. 10DISPOSITION
    ICU for VAP, septic shock, ARDS, P/F ≤200, vasopressor; ward for HAP if hemodynamically stable, no shock, FiO2 ≤0.4; step-down when off pressors, FiO2 ≤0.4, stable, tolerating PO/IV completion
    advance: Level of care set
11. 11MONITORING
    Clinical response reassessed at 48-72 h (defervescence, oxygenation, secretion volume/purulence, leukocyte trend); culture-driven de-escalation; PCT trend assists short-course stop (de Jong SAPS PMID 26947523); daily IV→PO suitability; vancomycin AUC24 (creatinine q48h); AKI + linezolid-thrombocytopenia surveillance
    inputs: procalcitonin
    advance: Improving at 48-72 h OR non-response workup triggered
12. 12FOLLOWUP
    Total 7-day duration if responding (Chastre non-inferior mortality 18.8% vs 17.2%, more antibiotic-free days, less MDR among recurrences — PMID 14625336; Pugh +4.02 antibiotic-free days, MDR recurrence OR 0.44 — PMID 26301604). Extend to 10-14 d ONLY for non-fermenting GNR (Pseudomonas/Acinetobacter — Chastre NF-GNB recurrence 40.6% vs 25.4%; Pugh OR 2.18), structural lung disease, immunocompromise, empyema/abscess, or slow response. Reinforce VAP prevention bundle; post-ICU follow-up
    advance: Duration set + prevention plan documented