SNOMED CT During the Admission: A Realistic Operating Model for Concurrent Care-Process Management, Quality Control, and Quality Assurance in a General Hospital

Abstract

The central implementation question is not whether SNOMED CT can improve coding at discharge, but whether it can become part of the care process itself. The answer is yes, but only under specific socio-technical conditions. SNOMED CT is most useful during admission when it is used as the semantic layer for working diagnoses, procedural events, pathway activation, clinical context, task orchestration, and real-time feedback, rather than as a retrospective abstraction exercise. Official SNOMED CT guidance explicitly frames the terminology as a means to “collect once and use many times,” and as an enabler of point-of-care analytics, decision support, and reporting that directly benefit clinicians and patients. In practice, this means that diagnoses and procedures must be recorded in small, workflow-specific SNOMED CT subsets; linked to timestamps, locations, and care states; and continuously reused for clinically relevant alerts, meaningful dashboards, and quality assurance (QA) during the admission itself.

The strongest use cases in a general hospital are time-critical emergency pathways and high-volume standardized elective pathways. The “First Hour Quintet” (FHQ) - cardiac arrest, severe trauma, stroke, acute respiratory failure, and cardiac chest pain - was defined in the European emergency medicine literature as a set of critical conditions in which rapid response and earliest treatment substantially modify outcome. Sepsis is not part of the historic FHQ, but it is operationally similar: it is a medical emergency in which early recognition and treatment improve outcomes. Elective hip replacement offers the opposite, but equally valuable, use case: a pathway with predictable milestones, PROMs, registry linkage, postoperative rehabilitation, and surveillance of complications and readmissions.

The main obstacles are well documented: (clinically irrelevant) documentation burden, poor user-interface design, overlong (clinically irrelevant) value lists, insufficient terminology governance, immaturity of postcoordination tooling, difficulty representing context, fragmented systems, weak maintenance processes, late or batch data entry, and the absence of strong feedback loops that show clinicians why clinically irrelevant structured capture matters. The implementation model proposed below therefore treats SNOMED CT not as a standalone coding project, but as a hospital operations capability and capacity built around interface terminology, curated clinically relevant reference sets, context handling, workflow integration, audit-feedback, and pathway-specific dashboards.

Introduction

Retrospective discharge coding has limited operational value during the admission because it is temporally late, often coarse-grained, and typically optimized for reimbursement, case-mix, registry submission, or statutory reporting. By the time a diagnosis is finalized after discharge, the opportunity to accelerate imaging, escalate a ward deterioration, trigger ICU review, or identify a pathway breach has already passed. A hospital that wants to manage care processes in real time therefore needs concurrent structured capture: the triage suspicion, the working diagnosis, the confirmed diagnosis, the performed intervention, and the relevant timestamps must all be recorded when care occurs, and must remain computable while the admission is still open. This is exactly the kind of “point of care reporting” and “point of care analytics” that SNOMED International describes as directly benefiting individual patients and clinicians.

This does not mean that every clinician should browse the full SNOMED CT hierarchy. SNOMED International’s own implementation guidance emphasizes constraining data entry with (relevant) reference sets, ordering search results, and using search/data-entry designs suited to the use case. The implementation literature reaches the same conclusion: clinicians accept SNOMED CT (codes) best when it is hidden behind familiar interface terms, auto-complete, templates, and focused subsets that do not interfere with workflow.

Accordingly, the practical aim for a general hospital is not “complete coding of everything in SNOMED CT.” The aim is narrower and more useful: capture the clinically relevant subset of diagnoses, procedures, findings, and states that effectively control operational flow, patient safety, and quality during the admission, then derive discharge abstractions, registry outputs, and analytics from that contemporaneous record. This is also aligned with the broader European and national framing of SNOMED CT as a shared semantic layer for meaningful EHRs, interoperability, effective search, and linkage to clinical knowledge and guidelines.

A Realistic General-Hospital Implementation Model

A realistic model has five layers.

First, a workflow-facing interface layer. Clinicians should see familiar clinical terms, not raw concept IDs. The hospital should build small clinically relevant SNOMED CT reference sets for Emergency Department (ED) triage, ward deterioration, sepsis, stroke, ACS/chest pain, trauma, respiratory failure, orthopaedic pre-op assessment, operative procedures, and postoperative complications. Search should be constrained to the relevant field and should prioritize common clinically relevant local terms and synonyms. 

Note: ward deterioration is acute clinical worsening in a patient who is already admitted to a regular hospital ward - that is, outside areas such as the ICU, operating theatre (OR), or ED resuscitation bay. It usually refers to a patient whose condition becomes worse after admission and therefore needs earlier recognition, escalation, and response.

Second, an information model that separates meaning from context. The same clinical meaning may be suspected, confirmed, ruled out, historical, post-procedural, or monitoring-related. SNOMED CT’s context-representation guidance exists precisely because clinical context must be represented in a disciplined way rather than improvised in free text. Operationally, the hospital should store: concept, assertion status, onset time, author, location, episode, and pathway state. That permits “suspected stroke” at triage to become “acute ischaemic stroke confirmed” after imaging without destroying the audit trail.

Third, an event-and-timestamp layer. Every pathway record should carry at least: time of presentation, time of pathway activation, time of specialist review, time of key diagnostic test, time of first definitive intervention, time of escalation/de-escalation, and disposition. SNOMED CT gives semantic coherence to the events, but process management requires timestamps and status transitions in addition to terminology. This is the step that converts structured documentation into pathway management rather than mere coding.

Fourth, a rules and feedback layer. Clinically relevant SNOMED CT-coded events should trigger pathway timers, task lists, alerts, and real-time dashboards. This is how the hospital moves from retrospective auditing to prospective (clinical process) control. Data-quality literature shows that timeliness, completeness, validity, consistency, and avoidance of duplicate capture are central to improvement; it also shows that feedback, interface redesign, workflow changes, and iterative Quality Improvement (QI) cycles are more effective than data cleaning alone.

Fifth, a competent maintenance and governance layer. The implementation literature is clear that SNOMED CT requires competent subset governance, mapping review, handling of inactive concepts, policies for uncoded terms, and competent local terminology expertise. Without this, data drift and concept misuse will quickly erode trust.

A practical rollout in a general hospital would be phased. Months 0–3: governance, pathway definitions, event dictionary, interface-term design, and three pilot dashboards. Months 4–9: production pilot in ED/ICU for FHQ and sepsis, and in orthopaedics for hip replacement. Months 10–18: expand to wards, theatre, radiology, and discharge derivation, with monthly Plan-Do-Study-Act (PDSA) review. That is a realistic scale because it limits early scope to high-value pathways while building the terminology and (competent) data-quality muscle the hospital will need later.

The First Hour Quintet as a Concurrent SNOMED CT Use Case

The "First Hour Quintet" (FHQ) refers to five critical, time-dependent emergency medical conditions - cardiac arrest, stroke, severe trauma, acute respiratory failure, and chest pain (including ACS) - defined by the European Resuscitation Council in 2002. These conditions require immediate, rapid prehospital intervention within the first hour to significantly improve patient survival and outcomes.  

The FHQ concept was developed for conditions in which Emergency Medical Services (EMS) and hospital systems can materially alter outcome through speed. For the hospital, that translates into a common operational model: the triage nurse or Emergency Department (ED) clinician records one SNOMED CT concept from an FHQ “suspicion” value set; this triggers the relevant pathway; subsequent SNOMED CT-coded assessments and procedures update the patient’s state; and the pathway remains visible until disposition or cancellation. The crucial design choice is to code the pathway state when it becomes operationally relevant, not only the final diagnosis at discharge.

For cardiac arrest, the minimum concurrent data set is arrest recognition, rhythm class, CPR started, first shock delivered, return of spontaneous circulation, airway status, and disposition. The FHQ literature explicitly identifies time to first shock as a key performance indicator, and contemporary Advanced Life Support (ALS) guidance continues to prioritize early interventions, minimal interruption of compressions, and early defibrillation. A realistic dashboard would therefore show active arrest cases, shockable-rhythm cases awaiting first shock, median time to first shock, Return of Spontaneous Circulation (ROSC) rate, survival to ICU transfer, and survival to discharge.

For severe trauma, the minimum dataset is trauma activation, haemodynamic status, major imaging requests, CT start, hot report, massive transfusion activation, theatre/interventional radiology activation, and disposition. In an  Major Trauma Centre (MTC)-style model, whole-body CT can be (reliably) targeted within 30 minutes of arrival; in a trauma-unit model, CT availability within 60 minutes of trauma-team activation is a realistic benchmark. Reporting timeliness is also operationally important: hot report within minutes and written report within one hour are service-management metrics, not merely radiology metrics.

Note: A Major Trauma Centre (MTC) is a (certified) specialized hospital unit providing 24/7 expert care for patients with severe, life-threatening injuries, such as major head, chest, or skeletal trauma.

Note: “Hot report within minutes” means an immediate provisional report, usually verbal or very brief, given by the radiologist right after the scan so the trauma team can act without waiting for the full formal report. In trauma guidance, this is often specified as about 5 minutes in major trauma centre workflows, or within 30 minutes in some trauma-unit guidance. Its purpose is rapid communication of the critical findings that change immediate management. “Written report within one hour” means a documented provisional or detailed radiology report entered into the reporting system within 60 minutes of the scan (or from the start of the scan, depending on the protocol). That written report supports formal clinical decision-making, handover, audit, and medico-legal documentation.

For stroke, the hospital should capture suspected stroke, National Institutes of Health Stroke Scale (NIHSS)-equivalent structured assessment if locally used, brain imaging ordered/performed/reported, stroke specialist review, thrombolysis decision, thrombectomy referral, and swallow screening. NHS England’s stroke service model requires brain imaging and interpretation within 60 minutes of arrival and stroke specialist review within 60 minutes; the AHA Target: Stroke programme uses door-to-needle within 60 minutes as a pragmatic benchmark. These are exactly the kinds of metrics that become visible only if the pathway is (in-line) coded at presentation rather than reconstructed later.

Note: "Time is brain" is a critical medical mantra emphasizing that in a stroke, every second counts. Each minute of untreated ischemic stroke causes about 1.9 million neurons to die, resulting in rapidly irreversible brain damage and loss of function. Immediate treatment is essential to save brain tissue and reduce disability or death.

For acute respiratory failure, the operational data set is respiratory failure suspected, blood gas taken, oxygen delivery mode, High-Flow Nasal Oxygen (HFNO)/Non-Invasive Ventilation (NIV)/intubation started, care location, escalation, and response assessment. British Thoracic Society (BTS) quality standards make two process expectations particularly useful for dashboarding: acute NIV should be delivered in appropriate designated areas, and eligible patients should start NIV within 60 minutes of the qualifying blood gas and within 120 minutes of arrival when presenting acutely. A respiratory-failure dashboard should therefore show cases eligible for NIV, location compliance, timing compliance, failure/escalation to intubation, and mortality.

For cardiac chest pain / acute coronary syndrome (ACS), the minimum data set is suspected ACS, first ECG, ECG interpretation state, serial troponin pathway status, cardiology review, reperfusion pathway activation where relevant, and disposition. Current ACS guidance remains explicit that ECG should be performed within 10 minutes of clinical contact and interpreted promptly by experienced clinicians. A general-hospital dashboard should therefore show median door-to-ECG time, percentage within 10 minutes, proportion escalated to invasive pathway, false-positive activations, and 30-day readmission or transfer outcomes.

Across all five pathways, three dashboards are needed. The operational dashboard is live and patient-level, showing active timers and breaches. The service dashboard is shift/day/week level, showing volume, compliance, and bottlenecks by location and team. The quality dashboard is monthly, showing adjusted trends, false-positive rates, completeness of structured capture, and outcome measures. This three-tier design mirrors what national SNOMED CT data-quality dashboards already (should) do: they do not merely count submissions, they assess whether the concepts are present, sufficiently specific, and located in the right parts of the data model.

Sepsis During Admission: A Model for the Ward as Well as the ED

Sepsis is the clearest example of why discharge-only coding is operationally insufficient. The Surviving Sepsis Campaign (SSC) states that sepsis and septic shock are medical emergencies and that treatment and resuscitation should begin immediately. The implementation problem in hospitals is often not knowledge of the bundle, but timely recognition on the floor, followed by rapid, coordinated action. The Society of Critical Care Medicine’s (SCCM) own implementation material explicitly addresses early identification of sepsis on hospital floors, not just in the ED.

A hospital should not use a single screening score as the sole trigger. The 2021 SSC guideline recommends against using qSOFA (Quick Sepsis-related Organ Failure Assessment) alone as the single screening tool. The safer design is a composite trigger: SNOMED-coded suspicion of infection or source, abnormal physiology, lactate result (LOINC), clinician concern, and organ dysfunction indicators. Once triggered, the pathway timer begins.

The core concurrent sepsis dataset should include: suspected sepsis or septic shock; suspected source; lactate ordered/resulted/repeated; blood cultures taken; antimicrobials started; fluids started and volume delivered; vasopressor started; ICU requested/admitted; source-control actions; and reassessment/de-escalation decisions. This aligns with the SSC hour-1 bundle structure and the 2021 guideline’s emphasis on early lactate measurement, immediate treatment, prompt antimicrobials - ideally within one hour for septic shock or high-likelihood sepsis - and 30 mL/kg crystalloid within the first three hours for sepsis-induced hypoperfusion or septic shock.

A realistic and relevant sepsis KPI set for a general hospital would include: time from recognition to lactate; time to first antibiotic; cultures before antibiotics where feasible; time to initial fluid resuscitation; vasopressor initiation for persistent hypotension; ICU admission within six hours when required; bundle completion; de-escalation when infection becomes unlikely; and mortality, ICU LOS, and 30-day readmission. The crucial distinction is that these KPIs should be computed from concurrent SNOMED CT events during the admission, not retrospectively inferred from discharge codes and timestamp archaeology.

The sepsis dashboard should exist in three forms. The ward dashboard shows patients with an active sepsis timer and overdue bundle elements. The sepsis command dashboard shows current load, active sources, antibiotic delays, and ICU demand. The monthly QA dashboard shows recognition-to-treatment distributions, pathway compliance by ward, false-positive triggers, and data completeness. This is also where SNOMED-based data-quality rules matter: overly generic concepts, wrong hierarchies, and non-SNOMED legacy values should be surfaced back to clinical teams rather than silently tolerated.

A Focus Clinic for Hip Replacements

Hip replacement is a different kind of admission pathway, but in many ways it is a stronger implementation starting point than emergency care because the workflow is predictable, multidisciplinary, and high-volume. National Institute for Health and Care Excellence (NICE) guidance covers care before, during, and after planned hip replacement; OECD and NHS PROMs programmes treat hip replacement as a canonical setting for patient-reported outcome measurement (PROM); and registry infrastructures such as the National Joint Registry (NJR) demonstrate how continuous structured capture supports safety, implant monitoring, and quality improvement.

In a general hospital “focus clinic,” the patient should enter a structured pathway before admission. The core preoperative dataset should include indication for surgery, comorbidity/risk factors, anaesthetic plan, baseline mobility, discharge plan, implant plan, and PROMs baseline. New South Wales’s (NSW) same-day hip and knee principles are especially useful because they make the pathway logic explicit: pre-operative education, multimodal analgesia, early mobilisation, follow-up after discharge, PROM collection, governance, and multidisciplinary coordination. Even if the hospital is not running same-day discharge for all patients, these are still the right operational components for a focused arthroplasty pathway.

The in-hospital SNOMED CT dataset should include the procedure actually performed, laterality, implant-related data where the local system supports it, antibiotic prophylaxis status, tranexamic acid and blood management steps if used, early mobilisation, physiotherapy milestones, pain status, complications, discharge readiness, and discharge destination. The key point is that these are not merely record items; they are the state machine of the pathway. If “mobilised within 24 hours” is not recorded, the dashboard should treat it as a pathway risk, not as a documentation nuisance.

A realistic and relevant KPI set for a hip-replacement focus clinic is the following. First, process KPIs: PROM completion before surgery and at follow-up, pre-op education completion, day-0 or day-1 mobilisation, Length of Stay (LOS), theatre cancellation due to bed flow, and registry submission completeness. Second, safety KPIs: Surgical Site Infection (SSI) surveillance, transfusion, VTE (venous thromboembolism), return to theatre (rework), delirium, and discharge with unresolved red flags. Third, outcome KPIs: PROM improvement, discharge home, 30-day readmission, 90-day complication/revision surveillance if the hospital tracks it, and patient experience. PROMs are central here because NHS and OECD sources both treat them as indicators of the quality and outcomes of hip replacement care. Registry completeness matters too: as a concrete exemplar, the NJR expects hospital compliance rates of at least 95%, and CMS (USA) uses unplanned readmission after elective primary Total Hip Arthroplasty (THA)/Total Knee Arthroplasty (TKA) as a hospital performance measure. SSI surveillance should be risk-adjusted and benchmarked, as recommended by ECDC surveillance protocol.

The hip clinic should have three dashboards. The pathway dashboard tracks each admitted patient against expected milestones. The service dashboard tracks throughput, LOS, early mobilisation, discharges, cancellations, and incomplete PROMs. The quality dashboard tracks PROM change, complications, SSI, registry completeness, readmissions, and surgeon/team variation. This is where SNOMED CT adds value during the admission: it turns perioperative care into a computable pathway rather than a pile of dictated notes.

Note: Unresolved red flags are signs of potential postoperative risk or clinical instability that are still present when discharge is being considered, but have not yet been adequately explained, treated, or stabilized. 

Obstacles to Using SNOMED CT During the Care Process

The first obstacle is documentation burden. Acute-care nurses spend a substantial proportion of their shift documenting, and poor EHR design/usability is associated with frustration, burnout, and reduced time for direct care. Structured entry can fail when it slows clinicians down or requires them to navigate long (irrelevant) lists with little immediate clinical benefit.

The second obstacle is terminology design failure. Broad domains such as “reason for admission” are difficult to subset cleanly; postcoordination remains challenging because tooling and practical maturity are limited; and concept selection becomes error-prone when the user is exposed to too much of the terminology. The implementation literature therefore repeatedly points toward interface terminology, clinically relevant curated subsets, hotlists, and auto-complete rather than raw browsing.

The third obstacle is poor data quality at the point of capture. Data-quality research shows that timeliness, completeness, accuracy, validity, and consistency are all vulnerable to workflow inefficiencies, interruptions, multiple systems, copy-paste, and batch processing after the event. NHS England’s SNOMED CT data-quality dashboard illustrates the same problem in practice: concepts may be present but still be too generic, in the wrong hierarchy, or not (validated) SNOMED CT concepts at all.

The fourth obstacle is weak governance and maintenance. SNOMED CT-enabled systems must routinely review inactive concepts, update mappings, decide what to do with uncoded terms, and maintain subsets as pathways evolve. Without a competent local terminology service, the hospital quickly accumulates drift, workarounds, and mistrust.

Note: Drift means the structured documentation gradually stops matching the intended standard over time. Workarounds means staff find unofficial ways to get their job done because the formal workflow is too slow, too rigid, or badly designed. Mistrust means clinicians, managers, or analysts stop believing the structured data are reliable enough to use for care management or quality assurance.

The fifth obstacle is misaligned incentives. Many hospitals still organize diagnosis capture around discharge abstraction, finance, or external reporting. When clinicians and nurses do not see a direct operational benefit - faster escalation, fewer missed bundle steps, better handoffs, cleaner dashboards - they rationally experience concurrent coding as “extra work.” That is why SNOMED CT projects that promise interoperability in the abstract but do not return value at the bedside often stall.

How to Facilitate SNOMED CT Use During the Healthcare Process

The enabling strategy is straightforward.

First, make documentation clinically useful immediately. A coded sepsis suspicion should start a timer; a coded stroke suspicion should place the patient on the stroke board; a coded non-invasive ventilation (NIV) start should satisfy a respiratory-pathway milestone. If structured capture does not change work in real time, adoption will remain weak.

Second, hide SNOMED CT (codes) behind good interface terminology. Use local synonyms, auto-complete, context-specific reference sets, and ordered clinically relevant pick lists. Avoid open-ended drop-downs. Keep required fields minimal and pathway-specific.

Third, separate concept from context. Record whether the diagnosis is suspected, confirmed, ruled out, historical, or complication-related. This is essential for admission-time care management because most early diagnoses are provisional.

Fourth, build a visible SNOMED CT data-quality programme. Hospitals should monitor completeness of structured capture, generic-code use, wrong-hierarchy use, mapping failures, and late entry. Feedback must go to the teams that generate the data. National examples show that SNOMED CT dashboards can meaningfully assess which organizations submit data, where the data land, how specific the concepts are, and where improvement is needed.

Fifth, treat terminology as an operational service, not a one-off IT build. The hospital needs clinical leadership, competent terminology management, pathway owners, dashboard ownership, and regular PDSA review. In the arthroplasty context, NSW’s (Australia) implementation guidance is explicit that governance, multidisciplinary leadership, and agreed pathways are essential. The same logic applies to FHQ and sepsis.

Conclusion

Structured recording of diagnoses and procedures in SNOMED CT can materially improve care-process management and quality assurance during a hospital admission - but only when it is used properly during care, with workflow-specific subsets, explicit context, timestamps, rules, and feedback. In that model, SNOMED CT stops being a discharge coding vocabulary and becomes the semantic control layer of the admission. It enables the hospital to know, in near real time, which patient is on which pathway, which critical step is overdue, which ward is missing bundle elements, which orthopaedic team has incomplete PROM capture, and where data quality is breaking down. That is the practical route by which concurrent SNOMED CT documentation can support not only secondary uses, but the healthcare process itself.

Bibliography (selection)

Evans, L., Rhodes, A., Alhazzani, W., Antonelli, M., Coopersmith, C. M., French, C., ... & Levy, M. (2021). Surviving sepsis campaign: international guidelines for management of sepsis and septic shock 2021. Critical care medicine, 49(11), e1063-e1143.

Levy, M. M., Evans, L. E., & Rhodes, A. (2018). The surviving sepsis campaign bundle: 2018 update. Intensive care medicine, 44(6), 925-928.

European Emergency Data Project. EMS Data-based Health Surveillance System: Project Report. European Commission, 2002.

SNOMED International. Data Analytics with SNOMED CT; Point of Care Analytics; Point of Care Reporting; Reference Set Guide; Implementation Guide for Context Representation. 2025–2026.

Lee, D., de Keizer, N., Lau, F., & Cornet, R. A Survey of SNOMED CT Implementations. Journal of Biomedical Informatics, 46(1), 87–96.

Cho, H., Nguyen, O. T., Weaver, M., Pruitt, J., Marcelle, C., Salloum, R. G., & Keenan, G. (2024). Electronic health record system use and documentation burden of acute and critical care nurse clinicians: a mixed-methods study. Journal of the American Medical Informatics Association, 31(11), 2540-2549.

Bush, R. A., Kuelbs, C., Ryu, J., Jiang, W., & Chiang, G. (2017). Structured data entry in the electronic medical record: perspectives of pediatric specialty physicians and surgeons. Journal of medical systems, 41(5), 75.

Lighterness, A., Adcock, M., Scanlon, L. A., & Price, G. (2024). Data quality–driven improvement in health care: systematic literature review. Journal of medical Internet research, 26, e57615.

 NHS England. National Stroke Service Model: Integrated Stroke Delivery Networks. 2021.

American Heart Association. Target: Stroke programme materials.

National Heart Foundation of Australia & Cardiac Society of Australia and New Zealand. ACS Guideline 2025.

British Thoracic Society. Quality Standards for Acute Non-Invasive Ventilation in Adults. 2018.

NICE. Joint Replacement (Primary): Hip, Knee and Shoulder (NG157). 2020, reviewed 2024.

OECD. International Assessment of the Use and Results of Patient-Reported Outcome Measures for Hip and Knee Replacement Surgery. OECD Health Working Paper No. 148, 2022.

NHS England / Digital. Patient Reported Outcome Measures (PROMs).

National Joint Registry. NJR: Hip and Knee Operations; How We Use Patient Data for Improvement.

Centers for Medicare & Medicaid Services. Hospital Readmissions Reduction Program.

European Centre for Disease Prevention and Control. HAI-Net Surgical Site Infection Protocol, Version 2.3. 2024.

NHS Scotland / regional trauma guidance. Imaging & CT Imaging; Sussex Trauma Network, Imaging for Trauma.

Sciensano / Belgian SNOMED CT National Release Centre materials. SNOMED CT. 2024.

Van Osta, Peter. An essay concerning a new healthcare.