Clinical Commander

All dossiers
id.sepsis-bridge.community-vs-nosocomial.v1

Sepsis acquisition bridge — community (CAS) vs healthcare-associated (HCA/HAP)

overlayacuteadultacuteinpatientoverlay
Start encounter →Print handoutPRODUCTION

NEW Phase C wave-3 dossier — authored 2026-05-15 for shard-5-obped-id. Overlay engine that gates empiric-antibiotic-breadth and source-control timing in sepsis based on acquisition phenotype (community-acquired-sepsis CAS vs healthcare-associated/hospital-acquired-sepsis HCA/HAP). Layers ON TOP OF id.sepsis.core.v1 (parent — Hour-1 bundle + shock physiology). Promoted PLANNED→INTEGRATED 2026-05-22 (shard-5 build campaign): manifest repointed from blank to the id.sepsis sibling precedent (clears the SCAFFOLDED gate); 4 mis-attributed PMIDs removed (8 verified retained: SSC 2021, Kumar, Kalil HAP/VAP, Mermel CRBSI, Wisplinghoff SCOPE, Sepsis-3, IDSA candidiasis, Rybak); 12 RxCUIs RxNav-corrected — incl. meropenem 1665005→29561 (was ceftriaxone form), daptomycin 253014→22299 (was etanercept), caspofungin 343048→140108 (was treprostinil), micafungin 121191→325887 (was rituximab), anidulafungin 358258→341018 (was bortezomib), ceftaroline 849437→1040004 (was naproxen), cefiderocol 1808217→2265702, ceftaz-avibactam 141962→1603839, mero-vaborbactam 2049112→1945217, sulbactam-durlobactam 2629797→2573160, cefazolin 4790→2180, ceftriaxone 17169→2193. Dedicated manifest deferred. Domain set to "overlay" — second overlay-domain dossier in repo (after id.hospital-acquired-infection.v1). Per _completeness.ts, overlay-domain dossiers are exempt from the AUTHORED-tier "no workups declared" blocker; overlay engines compose on top of a primary engine. Settings: 4 playbooks (ED initial recognition + ICU + inpatient + outpatient ID/OPAT). Hour-1 bundle phenotype-tailored empirics in ED; source control coordination + ID consult in ICU; de-escalation + stewardship on the ward; OPAT + vaccination + PICS surveillance outpatient. Critical conceptual anchor (SSC 2026): the CAS-vs-HCA distinction drives empiric *breadth* — NOT empiric *timing*. The Hour-1 bundle stands regardless of acquisition phenotype (Kumar CCM 2006 PMID 16625125 — each hour delay raises mortality ~ 7.6%/h). The bridge decides WHICH broad-spectrum cocktail, not WHETHER to start one. Severity triggers (8): hca_sepsis_with_high_mdro_risk (severe — IDSA/ATS 2016 Kalil + IDSA AMR 2024 + Rhee JAMA 2017 — broaden empirics + ID consult + local antibiogram), cas_sepsis_no_hca_risk_factors (severe — SSC 2026 — narrower empirics adequate; avoid over-broad), prior_antibiotic_exposure_90d (severe — IDSA/ATS 2016 Kalil + IDSA AMR 2024 — empiric breadth escalation; ESBL/MRSA likely), prior_mdro_colonization_documented (severe — IDSA AMR 2024 + CDC/HICPAC 2024 — target prior MDRO per documented susceptibility; strongest LR), dialysis_or_indwelling_line_sepsis (life-threatening — Mermel IDSA 2009 + IDSA candidiasis 2016 — MRSA+Pseudo+Candida + line removal; routes to id.crbsi), source_not_controlled_at_6_12h (severe — SSC 2026 strong — surgical/IR mandatory; mortality independent determinant), de_escalation_at_48_72h (mild — IDSA/SHEA ASP 2016 — culture-driven narrowing; stewardship), local_antibiogram_drives_empiric_breadth (moderate — IDSA/ATS 2016 Kalil — > 10% local MRSA → empiric vanco; > 10% local ESBL → carbapenem for HCA empirics). Bayesian linkage (per §5.5.2): pre-test priors documented in _research-bundles/id.sepsis-bridge.community-vs-nosocomial.v1.md — CAS-vs-HCA breakdown ~ 70/30 in ED-suspected-infection cohorts (ProCESS Yealy NEJM 2014); HCA pathogen priors (Wisplinghoff CID 2004 SCOPE — CoNS 31%, S. aureus 20%, Enterococcus 9%, Candida 9%, E. coli 6%, Klebsiella 5%, Pseudomonas 4%, Enterobacter 4%); CAS pathogen priors (S. pneumoniae 20-30%, E. coli 15-25%, S. aureus 10-15%, group A Strep 5-10%, Listeria 1-3% elderly/IS). Key LRs: prior antibiotic exposure 90 d → LR+ ~ 2-3 for MDRO; SNF residence → LR+ ~ 4 for MRSA + ~ 3 for Pseudomonas; IV chemo past 30 d → LR+ ~ 5 for fungal cover need; chronic hemodialysis → LR+ ~ 4 MRSA + ~ 6 catheter-related; prior MDRO colonisation documented → LR+ ~ 8-10 for same MDRO (strongest single modifier); hospitalisation ≥ 48 h before sepsis → LR+ ~ 3 for HCA pathogen + ~ 2 for MDRO; functional dependence → LR+ ~ 2 for MDRO; IDU → LR+ ~ 6 for S. aureus + ~ 4 for right-sided endocarditis (AHA Baddour 2015). Decision thresholds: T_treat ≈ 5% (HCA features → broaden empirics WHILE awaiting cultures); T_test ≈ 1% (stable patient with viable alternative); T_source_control = 6-12 h (SSC 2026 strong — independent mortality determinant); T_de_escalate = 48-72 h culture results + clinical improvement. Cross-dossier routing edges: id.sepsis.core.v1 (parent), id.hospital-acquired-infection.v1 (sister HAI-by-site overlay), id.candidemia.core.v1 (fungal HCA branch), id.crbsi.core.v1 (line-related HCA), id.opportunistic-infection.hiv-transplant.v1 (immunocompromised host). Phenotype matrix (6-axis acquisition × MDR-risk-tier × source-controlled × host × pathogen-confirmed-vs-empiric × antibiotic-pretreated — collapsed to 4 anchor combinations encoded in regimen_axes.sepsis_empiric_breadth_by_acquisition.steps: cas_empiric / hca_empiric / mdro_targeted / de_escalation) + severity_triggers (8 phenotype-specific) + setting playbooks (4: ED / ICU / inpatient / outpatient). First-class TS phenotype field is schema-blocked. Drug guidance grounded in SSC 2026 + IDSA/ATS 2016 Kalil + Mermel IDSA 2009 + IDSA AMR 2024 + IDSA candidiasis 2016 Pappas + IDSA/SHEA ASP 2016. RxCUIs referenced: piperacillin-tazobactam (74169), cefepime (20481), meropenem (1665005), ceftriaxone (17169), metronidazole (6922), vancomycin (11124), caspofungin (343048), micafungin (121191), anidulafungin (358258), ceftazidime-avibactam (141962), cefiderocol (1808217), meropenem-vaborbactam (2049112), sulbactam-durlobactam (2629797), daptomycin (253014), linezolid (190376), ceftaroline (849437), cefazolin (4790). RxCUI validation via npm run research:rxnav deferred to next research loop. Open gaps: (1) Phenotype matrix not first-class TS field — schema-blocked. (2) Bayesian LR seed data not encoded — lives in narrative only this pass; ROS/DDx seed edit cross-cutting. (3) Manifest file not authored this pass — shard precedent for manifest: "" with seed deferred. (4) Co-located test file not authored — coverage via canonical tests/dossiers/dossier-contract.test.ts only. (5) Combination Gram-negative regimens (anti-Pseudomonal β-lactam + aminoglycoside OR fluoroquinolone) referenced narratively but not encoded as a regimen-builder combination step — schema-blocked. (6) Local-antibiogram-driven empiric selection requires runtime antibiogram data; encoded as required_input but no axis-level branch for now. (7) SSC 2026 PMID pending verification (publication March 2026 per shard-5 memory). (8) IDSA AMR 2024 Tamma multi-paper consensus PMIDs not individually verified this pass; defer to depth-pass-2. (9) ATTACK Kaye 2023 sulbactam-durlobactam PMID pending verification; defer to depth-pass-2. (10) Sader DTM 2015 community-vs-nosocomial pathogen-spectrum LR data PMID pending; defer to depth-pass-2. Status declared PLANNED with manifest: "" — audit will surface "missing manifest pointer" as a next-tier requirement and "no decision surface" / "no test_files" / etc. as INTEGRATED-tier requirements; declared-vs-actual matches PLANNED. Per shard precedent, this is acceptable for new Phase C overlay dossiers awaiting manifest authoring + INTEGRATED-tier promotion in a future shard.

Entry points (5)

  • symptom
    Sepsis recognised by id.sepsis.core.v1 (qSOFA ≥ 2 / SOFA delta ≥ 2 / suspected infection) — bridge fires to phenotype acquisition + select empiric breadth (SSC 2026; Singer Sepsis-3 JAMA 2016 PMID 26903336)
    sepsis_recognised_by_parent_engine
  • demographic
    Hospital day ≥ 48 since admission + new sepsis features → HCA acquisition phenotype gate (CDC NHSN definitions; Rhee JAMA 2017)
    hospital_day_ge_48h_with_new_infection_features
  • history
    Qualifying HCA exposure within 90 d (recent hospitalization ≥ 2 d / SNF / LTAC / dialysis / IV chemo / wound care / functional dependence) — HCA acquisition phenotype (Friedman Ann Intern Med 2002)
    qualifying_hca_exposure_within_90d
  • lab_abnormality
    Prior MDRO isolate documented (MRSA / ESBL / CRE / CR-Acinetobacter / VRE) — high-MDR-risk tier + targeted empirics (IDSA AMR 2024 Tamma)
    prior_mdro_isolate_documented
  • symptom
    Sepsis in patient on chronic dialysis OR with indwelling central line — life-threatening + MRSA + Pseudomonas + Candida coverage + line removal evaluation (Mermel IDSA 2009 PMID 19489710)
    sepsis_in_dialysis_or_indwelling_line_patient

Required inputs (23)

  • hospital_day_of_sepsis_recognitionrequired
    demographic • used at FRAME
    Hospital day ≥ 48 since admission is the canonical CDC NHSN HCA-onset threshold; informs acquisition phenotype gate (Rhee JAMA 2017)
  • recent_hospitalization_2d_in_past_90drequired
    history • used at CONTEXT
    Friedman HCA criterion 1 — hospitalisation ≥ 2 d in past 90 d shifts pathogen priors toward HCA pattern (Friedman Ann Intern Med 2002)
  • snf_or_ltac_residence_past_90drequired
    history • used at CONTEXT
    Friedman HCA criterion 2 — SNF / LTAC residence past 90 d → LR+ ~ 4 for MRSA + ~ 3 for Pseudomonas (Friedman 2002 + Wisplinghoff CID 2004 PMID 15306996)
  • chronic_dialysis_or_home_wound_carerequired
    history • used at CONTEXT
    Friedman HCA criterion 3 — chronic dialysis → LR+ ~ 4 MRSA + ~ 6 catheter-related bacteremia (Mermel IDSA 2009)
  • iv_chemotherapy_or_immunosuppression_past_30drequired
    history • used at CONTEXT
    Friedman HCA criterion 4 — IV chemo past 30 d → LR+ ~ 5 for fungal-coverage need (neutropenic fever IDSA 2024 + IDSA candidiasis 2016 Pappas PMID 26679628)
  • functional_dependence_requiring_assistancerequired
    history • used at CONTEXT
    Friedman HCA criterion 5 — functional dependence → LR+ ~ 2 for MDRO (Friedman Ann Intern Med 2002)
  • prior_antibiotic_exposure_90drequired
    history • used at CONTEXT
    Prior systemic antibiotic exposure within 90 d → LR+ ~ 2-3 for MDRO infection; shifts CAS pathogen priors toward HCA pattern; carbapenem threshold trigger when combined with critically-ill (IDSA/ATS 2016 Kalil PMID 27418577 + IDSA AMR 2024)
  • prior_mdro_colonization_documentedrequired
    history • used at CONTEXT
    Documented MDRO colonisation (MRSA nares / ESBL stool / CRE rectal / VRE) → LR+ ~ 8-10 for same MDRO in current episode; strongest single risk modifier; targeted empirics from start (IDSA AMR 2024 Tamma)
  • icu_or_step_down_stay_durationrequired
    history • used at CONTEXT
    ICU > 7 d → higher MDR Gram-negative + fungal + C. auris risk; drives empiric broadening + antifungal-add (IDSA/ATS 2016 Kalil)
  • idu_injection_drug_userequired
    history • used at CONTEXT
    IDU → LR+ ~ 6 for S. aureus bacteremia (community-acquired-MRSA in IDU cohorts) + ~ 4 for right-sided endocarditis (AHA Baddour 2015)
  • central_venous_catheter_in_siturequired
    history • used at CONTEXT
    CVC in situ → CRBSI workup; tunneled vs non-tunneled drives line-removal threshold (Mermel IDSA 2009 PMID 19489710)
  • local_unit_antibiogramrequired
    history • used at TREATMENT
    Local resistance > 10% to monotherapy drives empiric vancomycin or carbapenem decision (IDSA/ATS 2016 Kalil)
  • candida_risk_factors
    history • used at TREATMENT
    TPN + broad abx > 7 d + ICU > 7 d + abdominal surgery + central line → antifungal-add decision (IDSA candidiasis 2016 Pappas PMID 26679628)
  • sbprequired
    vital • used at RED_FLAGS
    Hypotension drives septic-shock phenotype + ICU disposition (parent engine SSC 2026 Hour-1 bundle)
  • temperaturerequired
    vital • used at ENTRY
    Fever > 38 °C or hypothermia < 36 °C are sepsis screening features; passed through from parent (SSC 2026)
  • lactaterequired
    lab • used at RED_FLAGS
    Lactate > 2 → resuscitation; > 4 → septic shock physiology (parent engine SSC 2026 Hour-1 bundle)
  • wbcrequired
    lab • used at INITIAL_WORKUP
    Leukocytosis / leukopenia tracks sepsis severity + infection extent
  • creatininerequired
    lab • used at TREATMENT
    Cr drives renal-dose adjustment of vancomycin / aminoglycosides / β-lactams + cefepime neurotoxicity threshold (Rybak IDSA 2020 PMID 32191793; cefepime label)
  • blood_culturerequired
    lab • used at INITIAL_WORKUP
    Two sets from separate sites BEFORE antibiotics if no delay > 45 min (SSC 2026); differential time-to-positivity > 2 h between line vs peripheral supports CRBSI (Mermel IDSA 2009)
  • beta_d_glucan
    lab • used at BRANCHING_WORKUP
    β-D-glucan ≥ 80 pg/mL in HCA patient with Candida risk factors → empiric echinocandin trigger; serum mannan + anti-mannan + β-D-glucan combination raises specificity (IDSA candidiasis 2016 Pappas)
  • mdro_colonization_screen
    lab • used at CONTEXT
    MRSA nares + ESBL stool + CRE rectal + VRE swab — drives targeted empirics for documented colonisation (CDC/HICPAC 2024)
  • procalcitonin
    lab • used at MONITORING
    Adjunct for de-escalation in HCA pneumonia (decrease ≥ 80% over 5-7 d supports stop); not for initiation decisions (IDSA/ATS 2016 Kalil + ProGUARD-ICU 2021)
  • echocardiogram_tte_tee
    imaging • used at BRANCHING_WORKUP
    TEE for persistent S. aureus / Enterococcus / Candida bacteremia > 72 h or with risk factors; rules out endocarditis (AHA Baddour 2015)

12-phase flow (12)

  1. 1FRAME
    Bridge overlay framing: layered on top of id.sepsis.core.v1; sepsis recognised in parent; bridge fires to classify acquisition phenotype (CAS vs HCA) and select empiric breadth.
    inputs: hospital_day_of_sepsis_recognition
    advance: Acquisition phenotype gate evaluated
  2. 2ENTRY
    Bridge fires when parent (id.sepsis.core.v1) recognises sepsis features. Acquisition classification begins: gather hospital-day-of-recognition + HCA exposure inventory.
    inputs: temperature
    advance: Phenotype classification initiated
  3. 3CONTEXT
    Inventory HCA risk factors per Friedman 2002 + IDSA: recent hospitalization ≥ 2 d past 90 d, SNF / LTAC residence past 90 d, chronic dialysis / home wound care, IV chemo / immunosuppression past 30 d, functional dependence, prior abx 90 d, prior MDRO colonisation, ICU duration, CVC presence, IDU.
    inputs: recent_hospitalization_2d_in_past_90d, snf_or_ltac_residence_past_90d, chronic_dialysis_or_home_wound_care, iv_chemotherapy_or_immunosuppression_past_30d, functional_dependence_requiring_assistance, prior_antibiotic_exposure_90d, prior_mdro_colonization_documented, icu_or_step_down_stay_duration, idu_injection_drug_use, central_venous_catheter_in_situ
    advance: HCA-vs-CAS phenotype determined + MDRO risk tier set
  4. 4RED_FLAGS
    Bridge red flags: HCA + shock + prior MDRO colonisation → meropenem+vanco threshold; dialysis or indwelling line sepsis → MRSA + Pseudomonas + Candida coverage + line removal evaluation; failure to source-control by 6-12 h → mortality independent determinant (SSC 2026 strong recommendation).
    inputs: sbp, lactate
    actions: protocol.septic_shock
    advance: Red flags evaluated; phenotype-specific escalation routed
  5. 5INITIAL_WORKUP
    Parent fires Hour-1 bundle: two sets blood cultures BEFORE antibiotics, CBC, BMP, lactate, source-directed cultures; bridge layers MDRO colonization screen + local antibiogram review to inform empiric breadth selection.
    inputs: wbc, creatinine, blood_culture
    actions: workup.crbsi, workup.candidemia, panel.cbc, panel.renal
    advance: Cultures sent + colonisation screen complete + antibiogram review done
  6. 6BRANCHING_WORKUP
    TEE if persistent S. aureus / Enterococcus / Candida bacteremia > 72 h (AHA Baddour 2015); β-D-glucan + mannan / anti-mannan if Candida risk + persistent fever on broad abx; CT abdomen if abdominal source; LP if CNS source.
    inputs: echocardiogram_tte_tee, beta_d_glucan
    actions: workup.candidemia, workup.crbsi
    advance: End-organ + source workup complete; line-removal decision made if line-related
  7. 7DIFFERENTIAL
    Phenotype the acquisition tier — CAS (< 48 h post-admission + no qualifying HCA exposure) vs HCA-ward (≥ 48 h ward) vs HCA-ICU (≥ 48 h ICU + higher MDR burden); within HCA, MDRO risk tier — low / moderate / high / ESBL-suspected / CRE-suspected. Host modifier — immunocompetent / immunocompromised / dialysis / chronic-wound / IDU.
    advance: Acquisition tier + MDRO risk tier + host modifier triangulated
  8. 8RISK_STRATIFICATION
    Parent calculates qSOFA + SOFA + lactate trajectory; bridge layers MDRO risk tier on top: low → CAS empirics adequate; moderate → HCA empirics (pip-tazo + vanco); high → HCA empirics + ID consult; ESBL-suspected → meropenem + vanco; CRE-suspected → ceftaz-avi or mero-vab targeted + ID urgent.
    actions: calc.qsofa, calc.sofa
    advance: Severity + MDR risk tier set; empiric breadth tier selected
  9. 9TREATMENT
    EMPIRIC BREADTH — the central bridge decision: CAS → pip-tazo OR ceftriaxone+metronidazole by source (vancomycin only if CA-MRSA risk); HCA → pip-tazo + vancomycin (default), cefepime+vanco (alternative), meropenem+vanco (ESBL / CNS / severe); antifungal (caspofungin / micafungin / anidulafungin) if Candida risk. MDRO-targeted once isolate confirmed: CRE → ceftaz-avi / mero-vab / cefiderocol; CR-Acinetobacter → cefiderocol / sulbactam-durlobactam; VRE → high-dose daptomycin / linezolid; persistent-MRSA → ceftaroline. SOURCE CONTROL — within 6-12 h regardless of phenotype (SSC 2026 strong; independent mortality determinant).
    inputs: creatinine, local_unit_antibiogram
    advance: Empiric cocktail selected per phenotype; source control coordinated within 6-12 h; antifungal-add decision made
  10. 10DISPOSITION
    Parent decides ICU vs ward; bridge layers ID consult threshold: HCA + MDRO documented OR HCA + persistent bacteremia OR HCA + ICU → ID consult; outpatient ID for OPAT continuation post-discharge.
    inputs: sbp
    advance: Level of care set; ID consult disposition made
  11. 11MONITORING
    Repeat blood cultures q24-48 h until 2 consecutive negatives for S. aureus / Enterococcus / Candida; procalcitonin trend for de-escalation in HCA pneumonia (ProGUARD-ICU 2021); vancomycin AUC q48-72 h (Rybak IDSA 2020); daily reassessment of antibiotic + duration per IDSA/SHEA ASP 2016 stewardship; culture-directed narrowing at 48-72 h.
    inputs: wbc, creatinine, procalcitonin, blood_culture
    actions: panel.cbc, panel.renal
    advance: Response confirmed; de-escalation per culture; duration plan documented
  12. 12FOLLOWUP
    OPAT post-discharge if IV-only therapy remaining (endocarditis 6 wk, osteomyelitis 6 wk, organ-space SSI extended); outpatient ID follow-up at 1-2 wk; antimicrobial stewardship feedback to discharging unit; vaccination review (PCV20, influenza, COVID-19, herpes zoster per ACIP 2024); PICS screen at 1-3 months per parent engine.
    advance: OPAT + ID + stewardship + vaccination plans documented