ACT and ADJUST: Evaluating a Layered Emergency Triage Model Integrating Structured Scoring and Physician Judgment

ACT and ADJUST: Evaluating a Layered Emergency Triage Model Integrating Structured Scoring and Physician Judgment | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article ACT and ADJUST: Evaluating a Layered Emergency Triage Model Integrating Structured Scoring and Physician Judgment Gustav Siqueland, Bjørn-Jostein Singstad, Ingvild Billehaug Norum Viken, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6951401/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Triage in emergency departments (EDs) must extend beyond simply prioritizing patients for physician evaluation—it should also support the efficient allocation of limited clinical resources in fast-paced, high-pressure settings. This study evaluates a dual-layer triage model that integrates two complementary approaches: the structured, nurse-led Acuity Categorization Tool (ACT) and the physician-driven Adaptive Judgment of Urgency and Streaming (ADJUST). The model combines standardized initial assessment with dynamic, context-sensitive reassessment grounded in clinical insight and operational realities. Methods We retrospectively analyzed 53,651 consecutive ED encounters to assess the performance of ACT and ADJUST in identifying patients requiring intensive care unit (ICU) admission. Sensitivity, specificity, and reclassification rates were calculated using real-time triage documentation. To reflect the model’s distinct urgency thresholds, prognostic performance was evaluated exclusively for the highest acuity category (Red). Analyses included all encounters with complete data; documentation gaps in very high-acuity cases were acknowledged, as immediate clinical action often preempted formal triage. Results Among 409 ICU transfers, ACT demonstrated higher sensitivity (81.2%) than ADJUST (68.6%), aligning with its role in early detection. In contrast, ADJUST achieved greater specificity (96.0% vs. 85.9%), supporting more selective prioritization and improved resource alignment. Across the cohort, 45.3% of ACT classifications were downgraded and 3.6% upgraded by ADJUST. In a subset of 42,881 encounters with complete documentation, 20.9% of ICU patients initially flagged as Red by ACT were reclassified to lower acuity by ADJUST. Nearly half of all ADJUST evaluations occurred within 15 minutes of ACT. Conclusion This dual-layer triage model illustrates the complementary strengths of protocol-based scoring and physician judgment. ACT supports efficient and standardized initial triage, while ADJUST offers targeted refinement through clinical expertise and system-level awareness. The timing and reclassification patterns observed suggest that the two layers often functioned as an integrated, real-time decision-making process rather than discrete sequential steps. These findings align with performance benchmarks for established triage systems and suggest that integrated frameworks combining structured assessment with clinical discretion warrant further evaluation to improve acuity detection and care coordination in emergency care. Clinical trial number: not applicable. triage systems emergency department physician judgment acuity stratification layered triage model ICU prediction physician-assisted triage predictive validation Background Triage in emergency departments (EDs) serves a dual imperative: the timely identification of patients requiring urgent intervention and the efficient coordination of limited clinical resources. Widely used systems, such as the Manchester Triage System (MTS), provide a benchmark for sensitivity and specificity through structured, symptom-based algorithms. However, once initial categorization is complete, single-pass models may be less adaptable to evolving clinical presentations or shifting operational demands. To better support dynamic acuity detection and resource coordination, Vestfold Hospital Trust developed the ACT–ADJUST model in 2019—a lean, dual-layer triage framework that separates structured intake from clinician-led reassessment. The ACT tool was informed in part by the DAR triage system and further shaped through early collaboration with the Emergency Department at Slagelse Hospital in Denmark, whose experience with structured triage contributed to the model’s conceptual foundation. The system is grounded in the principle that front-line triage must be fast, structured, and easy to learn, yet flexible enough to support complex decision-making across the full continuum of emergency care. The system comprises two interdependent components: the Acuity Categorization Tool (ACT) and the Adaptive Judgment of Urgency and Streaming (ADJUST). ACT is a nurse-led instrument structured for consistency and operational simplicity. It integrates physiological scoring (NEWS2 for adults, PEWS for children), symptom-based triggers, and a discretionary escalation mechanism for professional concern. Its streamlined structure was intended to reduce cognitive load and subjective variation in early triage decisions, especially under conditions of high volume or staffing fluctuation. ADJUST introduces physician-led reassessment and allows for context-sensitive refinement of triage decisions. It accounts for evolving symptoms, treatment response, care goals, and department-level constraints such as crowding or staff capacity. In addition to modifying acuity classification, ADJUST also supports streaming—routing patients into appropriate care pathways with tailored resource allocation. This separation of roles enables more selective deployment of departmental resources and helps align care delivery with both clinical needs and systemic capacity. This layered strategy reflects a broader evolution in triage design. Elder et al. describe physician-assisted triage (PAT) as a means to improve throughput and reduce delays by incorporating early senior input [ 1 ]. Similarly, Christ et al. argue that modern triage must align urgency with system-level feasibility to ensure timely care delivery [ 2 ]. Real-world implementations such as TRIAD [ 3 ] and Swedish physician-led team triage model [ 4 ] demonstrate that structured clinician involvement can accelerate decision-making and improve flow. By decoupling standardized intake from discretionary reassessment, the ACT–ADJUST model offers a scalable approach that balances consistency in intake with flexibility in clinical reassessment. ACT serves as a consistent intake tool, while ADJUST introduces clinical reasoning, treatment trajectory, and operational context. This model is particularly suited for environments with high turnover or mixed clinical experience, where structured triage processes must coexist with real-time clinical insight. This study evaluates the prognostic performance of the ACT–ADJUST model in identifying patients requiring intensive care unit (ICU) admission—a pragmatic and validated proxy for high-acuity status. In addition to sensitivity and specificity, we examine reclassification patterns and timing dynamics to assess how the two triage layers function in practice. We also explore whether a layered model can reconcile the inherent tension between structured intake with context-driven judgment, ultimately improving acuity detection and resource coordination in complex emergency settings. Methods Study design and setting We conducted a retrospective cohort study of all emergency department (ED) encounters at Vestfold Hospital Trust, a regional hospital in Norway serving approximately 240,000 residents. The study period spanned from 1 September 2023 to 1 April 2025, beginning with the consistent implementation and electronic documentation of the dual-layer ACT–ADJUST triage model. Encounters were eligible if documentation from either triage layer was present. Paired-layer analyses were performed on the subset with both ACT and ADJUST recorded. Triage system The triage model comprises two sequential components: the Acuity Categorization Tool (ACT) and Adaptive Judgment of Urgency and Streaming (ADJUST). ACT is a four-level triage instrument applied by ED nurses during initial intake to rapidly prioritize patients based on structured criteria. It integrates predefined clinical triggers, early warning scores—NEWS2 for adults [ 5 ] and a Norwegian adaptation of PEWS for children—and a provision for professional concern, allowing discretionary escalation when clinical intuition suggests higher risk than structured inputs capture. ACT was designed to facilitate rapid and consistent prioritization across varying levels of clinical experience and patient volume. Full criteria are provided in Supplementary File 1. Following initial intake, ADJUST introduces physician-led reassessment. This layer incorporates evolving clinical information, such as symptom progression, treatment response, care goals, and department-level factors including capacity and staffing. Depending on clinical workflow, ADJUST was either applied simultaneously with ACT—particularly when physicians were present during intake—or documented later as a follow-up reassessment. Entries were made by attending physicians, whose experience ranged from junior residents to senior consultants. Outcome definition We defined the primary outcome as admission to the intensive care unit (ICU) within 24 hours of arrival to the emergency department—a pragmatic and validated proxy for high-acuity status in triage validation studies [ 6 ]. We excluded transfers to intermediate care and surgical units from the high-acuity category to avoid confounding from procedural workflows and bed allocation practices. This ensured consistency in evaluating both sensitivity and specificity against a clinically meaningful endpoint. Inclusion criteria and documentation All emergency department encounters with complete documentation from either ACT or ADJUST were eligible for inclusion. Analyses of sensitivity, specificity, and reclassification were restricted to encounters with paired entries from both triage layers. Encounters lacking triage documentation were excluded from performance analyses. These omissions primarily reflected very high-acuity scenarios—such as cardiac arrest or major trauma—where immediate clinical intervention superseded structured data entry. In such cases, retrospective documentation was often incomplete or absent, and these encounters were excluded from performance analysis due to the inability to validate real-time triage decisions. Triage metrics Sensitivity and specificity for ICU admission were calculated for both ACT and ADJUST based on real-time, uncorrected triage documentation. We assessed directional reclassification as changes in acuity level assigned by ADJUST compared to the initial ACT classification, either upward or downward. Ninety-five percent confidence intervals (CIs) were calculated using the Wald method. We interpreted prognostic accuracy using thresholds from Plante and Vance, where 90–100% is considered good to excellent and 80–89% acceptable for clinical decision-making [ 8 ]. As emphasized by Power et al., such interpretation should remain context-dependent, incorporating both statistical performance and clinical utility [ 9 ]. ROC analysis was not performed, as the ACT and ADJUST tools use predefined, ordinal triage categories rather than continuous scores. To reflect the model’s distinct acuity thresholds and operational mandates, sensitivity and specificity were calculated specifically for the highest acuity category (Red). Results Study cohort and documentation availability Between 1 September 2023 and 1 April 2025, a total of 53,651 emergency department (ED) encounters were recorded. Of these, 46,895 encounters (87.5%) had triage documentation from at least one layer and were included in the analysis. The remaining 6,756 encounters (12.6%) were excluded due to missing ACT or ADJUST data, including 111 of the 409 patients (27.1%) who were admitted to the intensive care unit (ICU). ADJUST documentation was specifically missing in 9,714 encounters (18.1%), and 5,701 encounters (10.6%) lacked triage entries from both layers. Among these, 62 ICU-admitted patients (15.2% of all ICU transfers) were excluded from sensitivity and specificity analyses due to incomplete documentation. Table 1 summarizes documentation availability and the distribution of ICU transfers across included and excluded groups. Table 1 Triage Documentation and ICU Transfers Group Total ICU Transfers % of ICU Transfers All cases 53,645 409 100% Included in ACT analysis 46,895 298 72.9% Excluded from ACT 6,751 111 27.1% Included in ADJUST analysis 43,932 280 68.5% Excluded from ADJUST 9,714 129 31.5% Excluded from both 5,701 62 15.2% The mean time from arrival to physician assessment was 78.0 ± 67.9 minutes, after excluding the top 5% of wait times. Several encounter types were excluded by design. Pediatric patients with non-urgent or non-surgical conditions were seen in a separate unit outside the ED workflow, and minor injuries were managed through a fast-track pathway that did not generate structured triage documentation. Triage documentation was frequently missing in high-acuity scenarios such as cardiac arrest or major trauma, where clinical intervention was initiated before structured triage entries were recorded. These encounters were excluded from prognostic performance analyses. Sensitivity for ICU transfers Sensitivity for identifying patients admitted to the intensive care unit (ICU) within 24 hours was calculated for both ACT and ADJUST using the subset of encounters with paired documentation. ACT demonstrated a sensitivity of 81.2% (± 4.4), while ADJUST had a sensitivity of 68.6% (± 5.4) (Table 2 ). These values were calculated for the Red triage category, which in the ACT–ADJUST model corresponds to immediate physician evaluation and continuous monitoring (see Section 3.6 for rationale). Table 2 Sensitivity of ACT and ADJUST for ICU Transfers System Sensitivity (%) 95% CI ACT 81.2 ± 4.4 ADJUST 68.6 ± 5.4 Legend: Sensitivity represents the proportion of ICU-transferred patients correctly identified by each triage method. Specificity for Non-Transferred Patients Specificity was calculated using ICU admission within 24 hours as the reference outcome. All other encounters—including those transferred to intermediate care units and surgical theatres—were classified as negative cases. ACT demonstrated a specificity of 85.9% (± 0.3%), based on 40,026 true negatives and 6,570 false positives. ADJUST demonstrated a specificity of 96.0% (± 0.2%), with 41,902 true negatives and 1,749 false positives (Table 3 ). Specificity was likewise calculated for the Red category, consistent with the model’s clinical thresholds. Table 3 Specificity for ICU Exclusion System False Positives (FP) True Negatives (TN) Specificity (%) 95% CI ACT 6,570 40,026 85.9% ± 0.3% ADJUST 1,749 41,902 96.0% ± 0.2% Legend: Specificity reflects the proportion of non-ICU patients correctly identified as not requiring high-acuity care. Triage classification and reclassification Among encounters with complete documentation from both ACT and ADJUST (n = 42,881), exact agreement in triage level occurred in 51.1% of cases (n = 21,903). ADJUST assigned a lower acuity level than ACT in 45.3% of cases (n = 19,440) and a higher acuity level in 3.6% of cases (n = 1,538) (Table 4 ). A full cross-tabulation of triage level assignments between ACT and ADJUST is provided in Supplementary File 2. Table 4 Reclassification Summary Between ACT and ADJUST Classification Outcome Count Percentage of Total Exact Match 21,903 51.1% Downgrade by ADJUST 19,440 45.3% Upgrade by ADJUST 1538 3.6% Legend: "Downgrade" refers to cases in which ADJUST assigned a lower acuity level than ACT. "Upgrade" refers to the opposite. Timing of physician reassessment Among the 42,881 encounters with complete documentation from both ACT and ADJUST, 20,856 (48.6%) had ADJUST entries recorded within 15 minutes of the ACT entry or earlier. This includes cases where ACT and ADJUST were completed jointly during intake, as well as instances in which ADJUST was timestamped before ACT. In such cases, assessments were typically performed simultaneously, with ACT documented later due to workflow or system entry sequence. In 22,025 encounters (51.4%) had ADJUST documented more than 15 minutes after ACT, reflecting a more sequential reassessment process. The mean time from ED arrival to physician assessment across all encounters was 78.0 ± 67.9 minutes, calculated after excluding the top 5% of wait times to reduce the influence of outliers. These findings suggest that in nearly half of encounters, ADJUST functioned as a co-primary triage input—either through structured collaboration or opportunistic early physician involvement. Performance within the paired documentation subset To assess the performance of ACT and ADJUST within their intended layered configuration, we conducted a secondary analysis limited to encounters with complete documentation from both triage layers (n = 42,881). A total of 231 patients in this subset were admitted to the intensive care unit (ICU) within 24 hours. Among these, ACT classified 182 patients as Red (sensitivity: 78.8%), while ADJUST classified 149 as Red (sensitivity: 64.5%). Of the 42,650 non-ICU encounters in this subset, ACT classified 37,017 as not Red (specificity: 86.8%), and ADJUST classified 41,031 as not Red (specificity: 96.2%). The core performance metrics are summarised in Table 5 . In this same subset, 38 ICU patients (20.9%) initially classified as Red by ACT were reassigned to a lower acuity category by ADJUST. A full cross-tabulation of acuity assignments within this subgroup is available in Supplementary File 2. Table 5 Prognostic performance within paired documentation subset Metric ACT ADJUST Sensitivity (ICU; n = 231) 78.8% 64.5% Specificity (non-ICU; n = 42,650) 86.8% 96.2% ICU patients downgraded from Resuscitation — 38 (20.9%) Legend: Sensitivity and specificity are calculated using ICU admission within 24 hours as the gold standard. “Red” denotes the highest acuity category requiring immediate physician contact and continuous monitoring. Downgrading refers to cases initially triaged as Red by ACT but reclassified as lower priority by ADJUST. These findings further support the importance of evaluating performance in the Red category separately from Orange. Unlike many triage systems where Red and Orange are treated as a unified "high urgency" group, the ACT–ADJUST model applies markedly different clinical expectations to these categories. Red triage mandates immediate physician evaluation and continuous monitoring, whereas Orange permits up to 30 minutes to physician review and lower-intensity surveillance. This separation is not merely conceptual but has direct implications for operational feasibility and patient safety. Grouping Red and Orange in performance metrics would obscure this clinically significant boundary and inflate sensitivity at the expense of interpretability. For this reason, we emphasize Red-only sensitivity and specificity to evaluate how well the model identifies patients requiring immediate escalation and critical care deployment. Discussion This study evaluated the performance of a layered triage model in identifying patients requiring intensive care. ACT supported early detection through structured intake, while ADJUST enabled refinement based on clinical judgment. When measured against ICU admission within 24 hours, ACT demonstrated higher sensitivity, consistent with its design as a front-line prioritization tool. ADJUST improved specificity, consistent with its role in streamlining resource use and reducing over-prioritization. In contrast to triage models that aggregate Red and Orange into a general “high urgency” category, the ACT–ADJUST framework applies distinct operational mandates to each level. The Red category requires immediate physician contact and continuous monitoring, whereas Orange allows up to 30 minutes for physician evaluation with intermittent observation. Reporting sensitivity and specificity for the Red category exclusively reflects these stricter clinical expectations and provides a more accurate view of the model’s ability to detect critical illness requiring immediate escalation. Moreover, the more flexible time window for Orange improves operational feasibility and supports realistic achievement of quality goals under routine emergency department conditions. Performance benchmarks for traditional systems such as the Manchester Triage System (MTS) typically show sensitivity between 80–86% and specificity from 84–91% in predicting urgent outcomes [ 10 ]. Within this study, ACT achieved comparable sensitivity and specificity, while the addition of ADJUST increased specificity to levels at or above the upper range reported for MTS. These results support the use of layered triage designs that combine standardized initial assessment with refined prioritization. The model’s separation of structured scoring and clinical reassessment introduces functional complementarity. Whereas systems such as the Manchester Triage System (MTS) rely on symptom categorization and predefined discriminator logic to guide nurse-led prioritization [ 10 ], ACT emphasizes operational clarity through physiological scoring and standardized acuity flags. ADJUST then reintroduces individualized appraisal, incorporating evolving symptoms, care goals, and system-level factors—an approach aligned with modern triage philosophies that emphasize both urgency and feasibility [ 2 ]. By decoupling structured intake from higher-order clinical judgment, the model reduces reliance on interpretive experience during triage while preserving flexibility for complex or ambiguous cases. This structure may be especially beneficial in settings with variable clinical experience or high turnover, where consistent early categorization can support safe and timely prioritization. This layered approach builds on earlier systems that integrate clinical oversight into triage workflows. Structured clinician involvement, as seen in physician-assisted triage (PAT), the TRIAD model, and physician-led team triage in Sweden, has been associated with improved flow and decision-making in emergency settings [ 1 – 4 ]. The ACT–ADJUST model formalizes this approach by clearly separating structured intake from discretionary reassessment. In the paired documentation subset, ADJUST reassigned nearly half of ACT’s classifications to a lower acuity level. While these downgrades may reflect stabilization, re-prioritization, or anticipated short symptom duration, they could also represent under-recognition of risk at the time of reassessment. The occurrence of 20.9% downgrades among ICU-admitted patients initially classified as Red by ACT highlights this ambiguity and illustrates the double-edged potential of clinical discretion. This dual possibility underscores both the value—and the risk—of discretionary reassessment in dynamic care environments. ADJUST entries occurred within 15 minutes of ACT in nearly half of cases, suggesting that nurse and physician assessments were completed together or in close succession. This timing pattern indicates that ACT and ADJUST frequently function as integrated inputs rather than strictly sequential steps. The use of ICU admission within 24 hours aligns with prior triage validation studies [ 6 ] and reflects the role of triage as a filter for immediate to near-term instability. This outcome is considered appropriate for model evaluation but excludes high-acuity care delivered outside the ICU and later deterioration events that may fall beyond the scope of triage to predict. Although various proxies have been used to assess triage validity, ICU admission remains a widely accepted outcome in this context despite known limitations [ 6 , 12 ]. The retrospective nature of the study also introduces variability in documentation, particularly in time-critical situations where structured triage was delayed or omitted. As a single-center analysis, generalizability may be limited, though reported metrics align with multicenter benchmarks for emergency triage performance [ 12 ]. Overall, the ACT–ADJUST model demonstrates how structured acuity detection and contextualized clinical judgment can be integrated to support timely, adaptive triage. This approach aligns with contemporary perspectives on emergency care as a dynamic system that requires both reproducibility and flexibility to meet changing clinical and operational demands [ 2 , 3 , 11 ]. Future studies should prospectively evaluate layered triage under varied operational conditions. Incorporating system-level indicators, reassessment timing, and patient-centered outcomes would help clarify the impact of dual-layer designs on safety, throughput, and efficiency. Comparative trials between layered and single-stage triage models may also clarify their respective contributions to decision-making in high-complexity emergency environments. Conclusions and recommendations This study demonstrates that a layered triage model integrating structured nurse-led assessment (ACT) with physician-led reassessment (ADJUST) offers complementary strengths in identifying patients requiring intensive care. ACT showed good sensitivity, supporting its utility for rapid, standardized prioritization, while ADJUST contributed high specificity by refining classifications based on evolving clinical context. By combining standardized intake with context-sensitive reassessment, the ACT–ADJUST model supports both consistency and adaptability in emergency care triage. These findings align with established benchmarks for triage performance and highlight the potential of layered approaches to improve acuity detection, reduce over-triage, and better coordinate limited clinical resources. Future research should prospectively evaluate layered triage models across diverse clinical settings and workload conditions. Comparative studies with single-stage systems and analyses incorporating reassessment timing, resource use, and patient-centered outcomes may further clarify the impact of dual-layer designs on safety, flow, and decision-making in complex emergency environments. Layered triage models such as ACT–ADJUST may offer a scalable strategy to bridge structured risk detection with individualized clinical judgment in modern emergency departments. Abbreviations ED Emergency Department TRIAD Triage rapid initial assessment by doctor ACT Acuity Categorization Tool ADJUST Adaptive Judgment of Urgency and Streaming ICU Intensive Care Unit NEWS National Early Warning Score CI Confidence Interval Declarations Ethics approval and consent to participate This study was approved by the internal ethics board at Vestfold Hospital Trust on 10 December 2020 (case number 20/03355). As a retrospective quality improvement study using de-identified registry data, informed consent was waived in accordance with national guidelines. Consent for publication Not applicable; no identifiable individual data are included. Availability of data and materials The data that support the findings of this study are available from the corresponding author upon reasonable request and with permission from Vestfold Hospital Trust. Competing interests The author declares that there are no competing interests. Funding This study received no external funding. It was supported internally by the Department of Emergency Medicine, Vestfold Hospital Trust. Authors’ contributions Gustav Siqueland led the study design, contributed to the development and implementation of the triage model, participated in ED trackboard integration, conducted the data analysis, and drafted the initial manuscript. Vidar Ruddox contributed to study design, interpretation of findings, and critical manuscript revisions. Bjørn Jostein Singstad supported data extraction, methodological review, and final revisions. Vetle Ellingsen Hauge and Rasmus Rimestad played key roles in developing and deploying the patient trackboard system that integrates the triage tool and enables structured data extraction. Ingvild Billehaug Norum and Therese Hamre Leet (emergency physicians), together with Eilin Solberg and Siren Jess (ED nurses), were instrumental in implementing and shaping the conceptual development of the ACT–ADJUST triage model. All authors approved the final manuscript and agreed to be accountable for all aspects of the work. Acknowledgements The authors thank the triage and Emergency Department staff at Vestfold Hospital Trust for their consistent contributions to real-time data entry and clinical documentation. We also acknowledge the clinical informatics and trackboard development teams for enabling system-integrated data capture—without whom this analysis would not have been possible. AI Disclosure Portions of this manuscript were edited using the AI language model ChatGPT (OpenAI) to improve clarity, grammar, and structure. No generative content, data interpretation, or scientific conclusions were produced by AI. All edits were reviewed and approved by the authors to ensure accuracy and integrity. Ethics Approval: Approved by the internal ethics board at Vestfold Hospital Trust on 10 December 2020 (case number 20/03355). References Elder E, Johnston ANB, Crilly J. Review article: Systematic review of three key strategies designed to improve patient flow through the emergency department. Emerg Med Australas. 2015;27(5):394–404. https://doi.org/10.1111/1742-6723.12450 . Christ M, Grossmann F, Winter D, Bingisser R, Platz E. Modern triage in the emergency department. Dtsch Arztebl Int. 2010;107(50):892–8. https://doi.org/10.3238/arztebl.2010.0892 . Choi YF, Wong TW, Lau CC. Triage rapid initial assessment by doctor (TRIAD) improves waiting time and processing time of the emergency department. Emerg Med J. 2006;23(4):262–5. https://doi.org/10.1136/emj.2005.025254 . Burström L, Nordberg M, Örnung G, Castrén M, Wiklund T, Engström M. Improved quality and efficiency after the introduction of physician-led team triage in an emergency department. Ups J Med Sci. 2016;121(1):38–43. https://doi.org/10.3109/03009734.2015.1100223 . Royal College of Physicians. National Early Warning Score (NEWS) 2: Standardising the assessment of acute-illness severity in the NHS. London: RCP; 2017. https://www.rcp.ac.uk/media/a4ibkkbf/news2-final-report_0_0.pdf . Covino M, Sandroni C, Della Polla D, De Matteis G, Piccioni A, De Vita A, et al. Predicting ICU admission and death in the Emergency Department: A comparison of six early warning scores. Resuscitation. 2023;190:109876. https://doi.org/10.1016/j.resuscitation.2023.109876 . Uriyama A, Urushidani S, Nakayama T. Five-level emergency triage systems: variation in assessment of validity. Emerg Med J. 2017;34(11):703–10. https://doi.org/10.1136/emermed-2016-206295 . Plante E, Vance R. Diagnostic accuracy of two tests of preschool language. Am J Speech Lang Pathol. 1994;3(1):57–64. https://doi.org/10.1044/1058-0360.0301.57 . Power M, Fell G, Wright M. Principles for high-quality, high-value testing. BMJ Evid Based Med. 2013;18(1):5–10. https://doi.org/10.1136/eb-2012-100645 . Zachariasse JM, Seiger N, Rood PPM, Alves CF, Freitas P, Smit FJ, et al. Validity of the Manchester Triage System in emergency care: A prospective observational study. PLoS ONE. 2017;12(2):e0170811. https://doi.org/10.1371/journal.pone.0170811 . Tsiftsis D, Tasioulis A, Bampalis D. Adult triage in the emergency department: Introducing a multi-layer triage system. Healthcare. 2025;13(9):1070. https://doi.org/10.3390/healthcare13091070 . Johansson A, Ekwall A, Forberg JL, Ekelund U. Development of outcomes for evaluating emergency care triage: a Delphi approach. Scand J Trauma Resusc Emerg Med. 2023;31(1):10. https://doi.org/10.1186/s13049-023-01073-1 . Additional Declarations No competing interests reported. Supplementary Files ACTADJUSTSupplementaryFile1.docx ACTADJUSTSupplementaryFile2.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6951401","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":476215186,"identity":"8eb568a3-faa1-413d-8838-f238b794777f","order_by":0,"name":"Gustav Siqueland","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBElEQVRIiWNgGAWjYDADNhDxgIGBsZ+BsYGQYogKsJYEIG9mA7FaGGBaNhwgoF4+Iv35g59tDPZ87DmGDxJq7shuPn+4TYJxhw1OLYY3cgwbe9sYEtt43hgbJBx7ZrztRiJQy5k03Fpm5DA28JxhSGCTyDGTSGA7nLjtBmOzAWPbYTxa0h82/jnDYA/UYv4j4d/hxM39B/FrkZdIMGzmqWBgbAPaAnTe4cQNDImND/BpMeB5YzhbpkIC6JdnxRKJfYeNZ9wAaklsw+0X+fb0Bx/fGNjYy7cnb/zw4dth2f7+4w8OfGzDHWIGB8CUBAM4UuAgAVMlwpYGopSNglEwCkbBiAYANblZvIVP88sAAAAASUVORK5CYII=","orcid":"","institution":"Vestfold Hospital Trust","correspondingAuthor":true,"prefix":"","firstName":"Gustav","middleName":"","lastName":"Siqueland","suffix":""},{"id":476215187,"identity":"58b05a22-27e1-4587-8ed3-eb1c0bbc8740","order_by":1,"name":"Bjørn-Jostein Singstad","email":"","orcid":"","institution":"Vestfold Hospital Trust","correspondingAuthor":false,"prefix":"","firstName":"Bjørn-Jostein","middleName":"","lastName":"Singstad","suffix":""},{"id":476215188,"identity":"422593a2-f903-4991-843f-57153d9aff08","order_by":2,"name":"Ingvild Billehaug Norum Viken","email":"","orcid":"","institution":"Vestfold Hospital Trust","correspondingAuthor":false,"prefix":"","firstName":"Ingvild","middleName":"Billehaug Norum","lastName":"Viken","suffix":""},{"id":476215189,"identity":"43f7bee7-f4b9-46d6-8e8f-89fc76bc6d84","order_by":3,"name":"Eilin Solberg","email":"","orcid":"","institution":"Vestfold Hospital Trust","correspondingAuthor":false,"prefix":"","firstName":"Eilin","middleName":"","lastName":"Solberg","suffix":""},{"id":476215190,"identity":"d14f993c-673b-430d-9704-e0d668f03e12","order_by":4,"name":"Siren Jess","email":"","orcid":"","institution":"Vestfold Hospital Trust","correspondingAuthor":false,"prefix":"","firstName":"Siren","middleName":"","lastName":"Jess","suffix":""},{"id":476215191,"identity":"4d83d463-ebe0-4119-a40a-e9d5f1ad9db7","order_by":5,"name":"Vetle Ellingsen Hauge","email":"","orcid":"","institution":"Vestfold Hospital Trust","correspondingAuthor":false,"prefix":"","firstName":"Vetle","middleName":"Ellingsen","lastName":"Hauge","suffix":""},{"id":476215192,"identity":"12f0b179-d703-4c1e-a255-0d70d22dd46f","order_by":6,"name":"Therese Hamre Leet","email":"","orcid":"","institution":"Vestfold Hospital Trust","correspondingAuthor":false,"prefix":"","firstName":"Therese","middleName":"Hamre","lastName":"Leet","suffix":""},{"id":476215193,"identity":"c82fc085-b768-40fe-a3f5-b84fe4a864b9","order_by":7,"name":"Vidar Ruddox","email":"","orcid":"","institution":"Vestfold Hospital Trust","correspondingAuthor":false,"prefix":"","firstName":"Vidar","middleName":"","lastName":"Ruddox","suffix":""}],"badges":[],"createdAt":"2025-06-22 21:23:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6951401/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6951401/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":86580436,"identity":"af4c62e7-c076-444b-b124-028be05cc2af","added_by":"auto","created_at":"2025-07-12 20:46:26","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":791421,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6951401/v1/2dd5eb81-600d-483d-95f2-f87bec701e63.pdf"},{"id":85575262,"identity":"e52dd190-ad0e-44b7-86cb-3e826c84e0b1","added_by":"auto","created_at":"2025-06-27 18:22:23","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":250092,"visible":true,"origin":"","legend":"","description":"","filename":"ACTADJUSTSupplementaryFile1.docx","url":"https://assets-eu.researchsquare.com/files/rs-6951401/v1/dad3938283316a1706e7c79f.docx"},{"id":85574892,"identity":"3980f99e-0edf-4246-92c0-30c10b5567fc","added_by":"auto","created_at":"2025-06-27 18:06:23","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":566058,"visible":true,"origin":"","legend":"","description":"","filename":"ACTADJUSTSupplementaryFile2.docx","url":"https://assets-eu.researchsquare.com/files/rs-6951401/v1/cae74b3b1b49df5b202c4739.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"ACT and ADJUST: Evaluating a Layered Emergency Triage Model Integrating Structured Scoring and Physician Judgment","fulltext":[{"header":"Background","content":"\u003cp\u003eTriage in emergency departments (EDs) serves a dual imperative: the timely identification of patients requiring urgent intervention and the efficient coordination of limited clinical resources. Widely used systems, such as the Manchester Triage System (MTS), provide a benchmark for sensitivity and specificity through structured, symptom-based algorithms. However, once initial categorization is complete, single-pass models may be less adaptable to evolving clinical presentations or shifting operational demands.\u003c/p\u003e \u003cp\u003eTo better support dynamic acuity detection and resource coordination, Vestfold Hospital Trust developed the ACT\u0026ndash;ADJUST model in 2019\u0026mdash;a lean, dual-layer triage framework that separates structured intake from clinician-led reassessment. The ACT tool was informed in part by the DAR triage system and further shaped through early collaboration with the Emergency Department at Slagelse Hospital in Denmark, whose experience with structured triage contributed to the model\u0026rsquo;s conceptual foundation. The system is grounded in the principle that front-line triage must be fast, structured, and easy to learn, yet flexible enough to support complex decision-making across the full continuum of emergency care.\u003c/p\u003e \u003cp\u003eThe system comprises two interdependent components: the Acuity Categorization Tool (ACT) and the Adaptive Judgment of Urgency and Streaming (ADJUST). ACT is a nurse-led instrument structured for consistency and operational simplicity. It integrates physiological scoring (NEWS2 for adults, PEWS for children), symptom-based triggers, and a discretionary escalation mechanism for professional concern. Its streamlined structure was intended to reduce cognitive load and subjective variation in early triage decisions, especially under conditions of high volume or staffing fluctuation.\u003c/p\u003e \u003cp\u003eADJUST introduces physician-led reassessment and allows for context-sensitive refinement of triage decisions. It accounts for evolving symptoms, treatment response, care goals, and department-level constraints such as crowding or staff capacity. In addition to modifying acuity classification, ADJUST also supports streaming\u0026mdash;routing patients into appropriate care pathways with tailored resource allocation. This separation of roles enables more selective deployment of departmental resources and helps align care delivery with both clinical needs and systemic capacity.\u003c/p\u003e \u003cp\u003eThis layered strategy reflects a broader evolution in triage design. Elder et al. describe physician-assisted triage (PAT) as a means to improve throughput and reduce delays by incorporating early senior input [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Similarly, Christ et al. argue that modern triage must align urgency with system-level feasibility to ensure timely care delivery [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Real-world implementations such as TRIAD [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] and Swedish physician-led team triage model [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] demonstrate that structured clinician involvement can accelerate decision-making and improve flow.\u003c/p\u003e \u003cp\u003eBy decoupling standardized intake from discretionary reassessment, the ACT\u0026ndash;ADJUST model offers a scalable approach that balances consistency in intake with flexibility in clinical reassessment. ACT serves as a consistent intake tool, while ADJUST introduces clinical reasoning, treatment trajectory, and operational context. This model is particularly suited for environments with high turnover or mixed clinical experience, where structured triage processes must coexist with real-time clinical insight.\u003c/p\u003e \u003cp\u003eThis study evaluates the prognostic performance of the ACT\u0026ndash;ADJUST model in identifying patients requiring intensive care unit (ICU) admission\u0026mdash;a pragmatic and validated proxy for high-acuity status. In addition to sensitivity and specificity, we examine reclassification patterns and timing dynamics to assess how the two triage layers function in practice. We also explore whether a layered model can reconcile the inherent tension between structured intake with context-driven judgment, ultimately improving acuity detection and resource coordination in complex emergency settings.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy design and setting\u003c/h2\u003e \u003cp\u003eWe conducted a retrospective cohort study of all emergency department (ED) encounters at Vestfold Hospital Trust, a regional hospital in Norway serving approximately 240,000 residents. The study period spanned from 1 September 2023 to 1 April 2025, beginning with the consistent implementation and electronic documentation of the dual-layer ACT\u0026ndash;ADJUST triage model.\u003c/p\u003e \u003cp\u003eEncounters were eligible if documentation from either triage layer was present. Paired-layer analyses were performed on the subset with both ACT and ADJUST recorded.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eTriage system\u003c/h3\u003e\n\u003cp\u003eThe triage model comprises two sequential components: the Acuity Categorization Tool (ACT) and Adaptive Judgment of Urgency and Streaming (ADJUST).\u003c/p\u003e \u003cp\u003eACT is a four-level triage instrument applied by ED nurses during initial intake to rapidly prioritize patients based on structured criteria. It integrates predefined clinical triggers, early warning scores\u0026mdash;NEWS2 for adults [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e] and a Norwegian adaptation of PEWS for children\u0026mdash;and a provision for professional concern, allowing discretionary escalation when clinical intuition suggests higher risk than structured inputs capture. ACT was designed to facilitate rapid and consistent prioritization across varying levels of clinical experience and patient volume. Full criteria are provided in Supplementary File 1.\u003c/p\u003e \u003cp\u003eFollowing initial intake, ADJUST introduces physician-led reassessment. This layer incorporates evolving clinical information, such as symptom progression, treatment response, care goals, and department-level factors including capacity and staffing. Depending on clinical workflow, ADJUST was either applied simultaneously with ACT\u0026mdash;particularly when physicians were present during intake\u0026mdash;or documented later as a follow-up reassessment. Entries were made by attending physicians, whose experience ranged from junior residents to senior consultants.\u003c/p\u003e\n\u003ch3\u003eOutcome definition\u003c/h3\u003e\n\u003cp\u003eWe defined the primary outcome as admission to the intensive care unit (ICU) within 24 hours of arrival to the emergency department\u0026mdash;a pragmatic and validated proxy for high-acuity status in triage validation studies [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. We excluded transfers to intermediate care and surgical units from the high-acuity category to avoid confounding from procedural workflows and bed allocation practices. This ensured consistency in evaluating both sensitivity and specificity against a clinically meaningful endpoint.\u003c/p\u003e\n\u003ch3\u003eInclusion criteria and documentation\u003c/h3\u003e\n\u003cp\u003eAll emergency department encounters with complete documentation from either ACT or ADJUST were eligible for inclusion. Analyses of sensitivity, specificity, and reclassification were restricted to encounters with paired entries from both triage layers.\u003c/p\u003e \u003cp\u003eEncounters lacking triage documentation were excluded from performance analyses. These omissions primarily reflected very high-acuity scenarios\u0026mdash;such as cardiac arrest or major trauma\u0026mdash;where immediate clinical intervention superseded structured data entry. In such cases, retrospective documentation was often incomplete or absent, and these encounters were excluded from performance analysis due to the inability to validate real-time triage decisions.\u003c/p\u003e\n\u003ch3\u003eTriage metrics\u003c/h3\u003e\n\u003cp\u003eSensitivity and specificity for ICU admission were calculated for both ACT and ADJUST based on real-time, uncorrected triage documentation. We assessed directional reclassification as changes in acuity level assigned by ADJUST compared to the initial ACT classification, either upward or downward.\u003c/p\u003e \u003cp\u003eNinety-five percent confidence intervals (CIs) were calculated using the Wald method. We interpreted prognostic accuracy using thresholds from Plante and Vance, where 90\u0026ndash;100% is considered good to excellent and 80\u0026ndash;89% acceptable for clinical decision-making [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. As emphasized by Power et al., such interpretation should remain context-dependent, incorporating both statistical performance and clinical utility [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eROC analysis was not performed, as the ACT and ADJUST tools use predefined, ordinal triage categories rather than continuous scores.\u003c/p\u003e \u003cp\u003eTo reflect the model\u0026rsquo;s distinct acuity thresholds and operational mandates, sensitivity and specificity were calculated specifically for the highest acuity category (Red).\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eStudy cohort and documentation availability\u003c/h2\u003e \u003cp\u003eBetween 1 September 2023 and 1 April 2025, a total of 53,651 emergency department (ED) encounters were recorded. Of these, 46,895 encounters (87.5%) had triage documentation from at least one layer and were included in the analysis. The remaining 6,756 encounters (12.6%) were excluded due to missing ACT or ADJUST data, including 111 of the 409 patients (27.1%) who were admitted to the intensive care unit (ICU).\u003c/p\u003e \u003cp\u003eADJUST documentation was specifically missing in 9,714 encounters (18.1%), and 5,701 encounters (10.6%) lacked triage entries from both layers. Among these, 62 ICU-admitted patients (15.2% of all ICU transfers) were excluded from sensitivity and specificity analyses due to incomplete documentation.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes documentation availability and the distribution of ICU transfers across included and excluded groups.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTriage Documentation and ICU Transfers\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGroup\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eICU Transfers\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e% of ICU Transfers\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll cases\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e53,645\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e409\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIncluded in ACT analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e46,895\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e298\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e72.9%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExcluded from ACT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6,751\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e111\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e27.1%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIncluded in ADJUST analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e43,932\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e280\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e68.5%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExcluded from ADJUST\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e9,714\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e129\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e31.5%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExcluded from both\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5,701\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e15.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe mean time from arrival to physician assessment was 78.0\u0026thinsp;\u0026plusmn;\u0026thinsp;67.9 minutes, after excluding the top 5% of wait times.\u003c/p\u003e \u003cp\u003eSeveral encounter types were excluded by design. Pediatric patients with non-urgent or non-surgical conditions were seen in a separate unit outside the ED workflow, and minor injuries were managed through a fast-track pathway that did not generate structured triage documentation.\u003c/p\u003e \u003cp\u003eTriage documentation was frequently missing in high-acuity scenarios such as cardiac arrest or major trauma, where clinical intervention was initiated before structured triage entries were recorded. These encounters were excluded from prognostic performance analyses.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eSensitivity for ICU transfers\u003c/h3\u003e\n\u003cp\u003eSensitivity for identifying patients admitted to the intensive care unit (ICU) within 24 hours was calculated for both ACT and ADJUST using the subset of encounters with paired documentation. ACT demonstrated a sensitivity of 81.2% (\u0026plusmn;\u0026thinsp;4.4), while ADJUST had a sensitivity of 68.6% (\u0026plusmn;\u0026thinsp;5.4) (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThese values were calculated for the Red triage category, which in the ACT\u0026ndash;ADJUST model corresponds to immediate physician evaluation and continuous monitoring (see Section 3.6 for rationale).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSensitivity of ACT and ADJUST for ICU Transfers\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSystem\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSensitivity (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eACT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e81.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026plusmn;\u0026thinsp;4.4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADJUST\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e68.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026plusmn;\u0026thinsp;5.4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eLegend: Sensitivity represents the proportion of ICU-transferred patients correctly identified by each triage method.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eSpecificity for Non-Transferred Patients\u003c/h2\u003e \u003cp\u003eSpecificity was calculated using ICU admission within 24 hours as the reference outcome. All other encounters\u0026mdash;including those transferred to intermediate care units and surgical theatres\u0026mdash;were classified as negative cases.\u003c/p\u003e \u003cp\u003eACT demonstrated a specificity of 85.9% (\u0026plusmn;\u0026thinsp;0.3%), based on 40,026 true negatives and 6,570 false positives. ADJUST demonstrated a specificity of 96.0% (\u0026plusmn;\u0026thinsp;0.2%), with 41,902 true negatives and 1,749 false positives (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eSpecificity was likewise calculated for the Red category, consistent with the model\u0026rsquo;s clinical thresholds.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSpecificity for ICU Exclusion\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSystem\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFalse Positives (FP)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTrue Negatives (TN)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSpecificity (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eACT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6,570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e40,026\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e85.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026plusmn;\u0026thinsp;0.3%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADJUST\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1,749\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e41,902\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e96.0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026plusmn;\u0026thinsp;0.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eLegend: Specificity reflects the proportion of non-ICU patients correctly identified as not requiring high-acuity care.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eTriage classification and reclassification\u003c/h2\u003e \u003cp\u003eAmong encounters with complete documentation from both ACT and ADJUST (n\u0026thinsp;=\u0026thinsp;42,881), exact agreement in triage level occurred in 51.1% of cases (n\u0026thinsp;=\u0026thinsp;21,903). ADJUST assigned a lower acuity level than ACT in 45.3% of cases (n\u0026thinsp;=\u0026thinsp;19,440) and a higher acuity level in 3.6% of cases (n\u0026thinsp;=\u0026thinsp;1,538) (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eA full cross-tabulation of triage level assignments between ACT and ADJUST is provided in Supplementary File 2.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eReclassification Summary Between ACT and ADJUST\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClassification Outcome\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCount\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePercentage of Total\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExact Match\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e21,903\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e51.1%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDowngrade by ADJUST\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e19,440\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e45.3%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUpgrade by ADJUST\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1538\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3.6%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eLegend: \"Downgrade\" refers to cases in which ADJUST assigned a lower acuity level than ACT. \"Upgrade\" refers to the opposite.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eTiming of physician reassessment\u003c/h2\u003e \u003cp\u003eAmong the 42,881 encounters with complete documentation from both ACT and ADJUST, 20,856 (48.6%) had ADJUST entries recorded within 15 minutes of the ACT entry or earlier. This includes cases where ACT and ADJUST were completed jointly during intake, as well as instances in which ADJUST was timestamped before ACT. In such cases, assessments were typically performed simultaneously, with ACT documented later due to workflow or system entry sequence.\u003c/p\u003e \u003cp\u003eIn 22,025 encounters (51.4%) had ADJUST documented more than 15 minutes after ACT, reflecting a more sequential reassessment process. The mean time from ED arrival to physician assessment across all encounters was 78.0\u0026thinsp;\u0026plusmn;\u0026thinsp;67.9 minutes, calculated after excluding the top 5% of wait times to reduce the influence of outliers.\u003c/p\u003e \u003cp\u003eThese findings suggest that in nearly half of encounters, ADJUST functioned as a co-primary triage input\u0026mdash;either through structured collaboration or opportunistic early physician involvement.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003ePerformance within the paired documentation subset\u003c/h2\u003e \u003cp\u003eTo assess the performance of ACT and ADJUST within their intended layered configuration, we conducted a secondary analysis limited to encounters with complete documentation from both triage layers (n\u0026thinsp;=\u0026thinsp;42,881). A total of 231 patients in this subset were admitted to the intensive care unit (ICU) within 24 hours.\u003c/p\u003e \u003cp\u003eAmong these, ACT classified 182 patients as Red (sensitivity: 78.8%), while ADJUST classified 149 as Red (sensitivity: 64.5%). Of the 42,650 non-ICU encounters in this subset, ACT classified 37,017 as not Red (specificity: 86.8%), and ADJUST classified 41,031 as not Red (specificity: 96.2%). The core performance metrics are summarised in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eIn this same subset, 38 ICU patients (20.9%) initially classified as Red by ACT were reassigned to a lower acuity category by ADJUST. A full cross-tabulation of acuity assignments within this subgroup is available in Supplementary File 2.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePrognostic performance within paired documentation subset\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eACT\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eADJUST\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSensitivity (ICU; n\u0026thinsp;=\u0026thinsp;231)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e78.8%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e64.5%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpecificity (non-ICU; n\u0026thinsp;=\u0026thinsp;42,650)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e86.8%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e96.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eICU patients downgraded from Resuscitation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e38 (20.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eLegend: Sensitivity and specificity are calculated using ICU admission within 24 hours as the gold standard. \u0026ldquo;Red\u0026rdquo; denotes the highest acuity category requiring immediate physician contact and continuous monitoring. Downgrading refers to cases initially triaged as Red by ACT but reclassified as lower priority by ADJUST.\u003c/p\u003e \u003cp\u003eThese findings further support the importance of evaluating performance in the Red category separately from Orange. Unlike many triage systems where Red and Orange are treated as a unified \"high urgency\" group, the ACT\u0026ndash;ADJUST model applies markedly different clinical expectations to these categories. Red triage mandates immediate physician evaluation and continuous monitoring, whereas Orange permits up to 30 minutes to physician review and lower-intensity surveillance. This separation is not merely conceptual but has direct implications for operational feasibility and patient safety. Grouping Red and Orange in performance metrics would obscure this clinically significant boundary and inflate sensitivity at the expense of interpretability. For this reason, we emphasize Red-only sensitivity and specificity to evaluate how well the model identifies patients requiring \u003cem\u003eimmediate\u003c/em\u003e escalation and critical care deployment.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study evaluated the performance of a layered triage model in identifying patients requiring intensive care. ACT supported early detection through structured intake, while ADJUST enabled refinement based on clinical judgment. When measured against ICU admission within 24 hours, ACT demonstrated higher sensitivity, consistent with its design as a front-line prioritization tool. ADJUST improved specificity, consistent with its role in streamlining resource use and reducing over-prioritization.\u003c/p\u003e \u003cp\u003eIn contrast to triage models that aggregate Red and Orange into a general \u0026ldquo;high urgency\u0026rdquo; category, the ACT\u0026ndash;ADJUST framework applies distinct operational mandates to each level. The Red category requires immediate physician contact and continuous monitoring, whereas Orange allows up to 30 minutes for physician evaluation with intermittent observation. Reporting sensitivity and specificity for the Red category exclusively reflects these stricter clinical expectations and provides a more accurate view of the model\u0026rsquo;s ability to detect critical illness requiring immediate escalation. Moreover, the more flexible time window for Orange improves operational feasibility and supports realistic achievement of quality goals under routine emergency department conditions.\u003c/p\u003e \u003cp\u003ePerformance benchmarks for traditional systems such as the Manchester Triage System (MTS) typically show sensitivity between 80\u0026ndash;86% and specificity from 84\u0026ndash;91% in predicting urgent outcomes [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Within this study, ACT achieved comparable sensitivity and specificity, while the addition of ADJUST increased specificity to levels at or above the upper range reported for MTS. These results support the use of layered triage designs that combine standardized initial assessment with refined prioritization.\u003c/p\u003e \u003cp\u003eThe model\u0026rsquo;s separation of structured scoring and clinical reassessment introduces functional complementarity. Whereas systems such as the Manchester Triage System (MTS) rely on symptom categorization and predefined discriminator logic to guide nurse-led prioritization [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], ACT emphasizes operational clarity through physiological scoring and standardized acuity flags. ADJUST then reintroduces individualized appraisal, incorporating evolving symptoms, care goals, and system-level factors\u0026mdash;an approach aligned with modern triage philosophies that emphasize both urgency and feasibility [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eBy decoupling structured intake from higher-order clinical judgment, the model reduces reliance on interpretive experience during triage while preserving flexibility for complex or ambiguous cases. This structure may be especially beneficial in settings with variable clinical experience or high turnover, where consistent early categorization can support safe and timely prioritization.\u003c/p\u003e \u003cp\u003eThis layered approach builds on earlier systems that integrate clinical oversight into triage workflows. Structured clinician involvement, as seen in physician-assisted triage (PAT), the TRIAD model, and physician-led team triage in Sweden, has been associated with improved flow and decision-making in emergency settings [\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. The ACT\u0026ndash;ADJUST model formalizes this approach by clearly separating structured intake from discretionary reassessment.\u003c/p\u003e \u003cp\u003eIn the paired documentation subset, ADJUST reassigned nearly half of ACT\u0026rsquo;s classifications to a lower acuity level. While these downgrades may reflect stabilization, re-prioritization, or anticipated short symptom duration, they could also represent under-recognition of risk at the time of reassessment. The occurrence of 20.9% downgrades among ICU-admitted patients initially classified as Red by ACT highlights this ambiguity and illustrates the double-edged potential of clinical discretion. This dual possibility underscores both the value\u0026mdash;and the risk\u0026mdash;of discretionary reassessment in dynamic care environments.\u003c/p\u003e \u003cp\u003eADJUST entries occurred within 15 minutes of ACT in nearly half of cases, suggesting that nurse and physician assessments were completed together or in close succession. This timing pattern indicates that ACT and ADJUST frequently function as integrated inputs rather than strictly sequential steps.\u003c/p\u003e \u003cp\u003eThe use of ICU admission within 24 hours aligns with prior triage validation studies [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] and reflects the role of triage as a filter for immediate to near-term instability. This outcome is considered appropriate for model evaluation but excludes high-acuity care delivered outside the ICU and later deterioration events that may fall beyond the scope of triage to predict. Although various proxies have been used to assess triage validity, ICU admission remains a widely accepted outcome in this context despite known limitations [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe retrospective nature of the study also introduces variability in documentation, particularly in time-critical situations where structured triage was delayed or omitted. As a single-center analysis, generalizability may be limited, though reported metrics align with multicenter benchmarks for emergency triage performance [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOverall, the ACT\u0026ndash;ADJUST model demonstrates how structured acuity detection and contextualized clinical judgment can be integrated to support timely, adaptive triage. This approach aligns with contemporary perspectives on emergency care as a dynamic system that requires both reproducibility and flexibility to meet changing clinical and operational demands [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFuture studies should prospectively evaluate layered triage under varied operational conditions. Incorporating system-level indicators, reassessment timing, and patient-centered outcomes would help clarify the impact of dual-layer designs on safety, throughput, and efficiency. Comparative trials between layered and single-stage triage models may also clarify their respective contributions to decision-making in high-complexity emergency environments.\u003c/p\u003e "},{"header":"Conclusions and recommendations","content":"\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003cp\u003eThis study demonstrates that a layered triage model integrating structured nurse-led assessment (ACT) with physician-led reassessment (ADJUST) offers complementary strengths in identifying patients requiring intensive care. ACT showed good sensitivity, supporting its utility for rapid, standardized prioritization, while ADJUST contributed high specificity by refining classifications based on evolving clinical context.\u003c/p\u003e \u003cp\u003eBy combining standardized intake with context-sensitive reassessment, the ACT\u0026ndash;ADJUST model supports both consistency and adaptability in emergency care triage. These findings align with established benchmarks for triage performance and highlight the potential of layered approaches to improve acuity detection, reduce over-triage, and better coordinate limited clinical resources.\u003c/p\u003e \u003cp\u003eFuture research should prospectively evaluate layered triage models across diverse clinical settings and workload conditions. Comparative studies with single-stage systems and analyses incorporating reassessment timing, resource use, and patient-centered outcomes may further clarify the impact of dual-layer designs on safety, flow, and decision-making in complex emergency environments.\u003c/p\u003e \u003cp\u003eLayered triage models such as ACT\u0026ndash;ADJUST may offer a scalable strategy to bridge structured risk detection with individualized clinical judgment in modern emergency departments.\u003c/p\u003e \u003c/div\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eED Emergency Department\u003c/p\u003e\n\u003cp\u003eTRIAD Triage rapid initial assessment by doctor\u003c/p\u003e\n\u003cp\u003eACT Acuity Categorization Tool\u003c/p\u003e\n\u003cp\u003eADJUST Adaptive Judgment of Urgency and Streaming\u003c/p\u003e\n\u003cp\u003eICU Intensive Care Unit\u003c/p\u003e\n\u003cp\u003eNEWS National Early Warning Score\u003c/p\u003e\n\u003cp\u003eCI Confidence Interval\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003cbr\u003e\u003c/strong\u003eThis study was approved by the internal ethics board at Vestfold Hospital Trust on 10 December 2020 (case number 20/03355). As a retrospective quality improvement study using de-identified registry data, informed consent was waived in accordance with national guidelines.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003cbr\u003e\u003c/strong\u003eNot applicable; no identifiable individual data are included.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003cbr\u003e\u003c/strong\u003eThe data that support the findings of this study are available from the corresponding author upon reasonable request and with permission from Vestfold Hospital Trust.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003cbr\u003e\u003c/strong\u003eThe author declares that there are no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003cbr\u003e\u003c/strong\u003eThis study received no external funding. It was supported internally by the Department of Emergency Medicine, Vestfold Hospital Trust.\u003c/p\u003e\n\u003ch4\u003eAuthors\u0026rsquo; contributions\u003c/h4\u003e\n\u003cp\u003eGustav Siqueland led the study design, contributed to the development and implementation of the triage model, participated in ED trackboard integration, conducted the data analysis, and drafted the initial manuscript.\u003c/p\u003e\n\u003cp\u003eVidar Ruddox contributed to study design, interpretation of findings, and critical manuscript revisions.\u003c/p\u003e\n\u003cp\u003eBj\u0026oslash;rn Jostein Singstad supported data extraction, methodological review, and final revisions.\u003c/p\u003e\n\u003cp\u003eVetle Ellingsen Hauge and Rasmus Rimestad played key roles in developing and deploying the patient trackboard system that integrates the triage tool and enables structured data extraction.\u003c/p\u003e\n\u003cp\u003eIngvild Billehaug Norum and Therese Hamre Leet (emergency physicians), together with Eilin Solberg and Siren Jess (ED nurses), were instrumental in implementing and shaping the conceptual development of the ACT\u0026ndash;ADJUST triage model.\u003c/p\u003e\n\u003cp\u003eAll authors approved the final manuscript and agreed to be accountable for all aspects of the work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003cbr\u003e\u003c/strong\u003eThe authors thank the triage and Emergency Department staff at Vestfold Hospital Trust for their consistent contributions to real-time data entry and clinical documentation. We also acknowledge the clinical informatics and trackboard development teams for enabling system-integrated data capture\u0026mdash;without whom this analysis would not have been possible.\u003c/p\u003e\n\u003ch4\u003eAI Disclosure\u003c/h4\u003e\n\u003cp\u003ePortions of this manuscript were edited using the AI language model ChatGPT (OpenAI) to improve clarity, grammar, and structure. No generative content, data interpretation, or scientific conclusions were produced by AI. All edits were reviewed and approved by the authors to ensure accuracy and integrity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics Approval:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eApproved by the internal ethics board at Vestfold Hospital Trust on 10 December 2020 (case number 20/03355).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eElder E, Johnston ANB, Crilly J. Review article: Systematic review of three key strategies designed to improve patient flow through the emergency department. Emerg Med Australas. 2015;27(5):394\u0026ndash;404. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/1742-6723.12450\u003c/span\u003e\u003cspan address=\"10.1111/1742-6723.12450\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChrist M, Grossmann F, Winter D, Bingisser R, Platz E. Modern triage in the emergency department. Dtsch Arztebl Int. 2010;107(50):892\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3238/arztebl.2010.0892\u003c/span\u003e\u003cspan address=\"10.3238/arztebl.2010.0892\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChoi YF, Wong TW, Lau CC. Triage rapid initial assessment by doctor (TRIAD) improves waiting time and processing time of the emergency department. Emerg Med J. 2006;23(4):262\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1136/emj.2005.025254\u003c/span\u003e\u003cspan address=\"10.1136/emj.2005.025254\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurstr\u0026ouml;m L, Nordberg M, \u0026Ouml;rnung G, Castr\u0026eacute;n M, Wiklund T, Engstr\u0026ouml;m M. Improved quality and efficiency after the introduction of physician-led team triage in an emergency department. Ups J Med Sci. 2016;121(1):38\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3109/03009734.2015.1100223\u003c/span\u003e\u003cspan address=\"10.3109/03009734.2015.1100223\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoyal College of Physicians. National Early Warning Score (NEWS) 2: Standardising the assessment of acute-illness severity in the NHS. London: RCP; 2017. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.rcp.ac.uk/media/a4ibkkbf/news2-final-report_0_0.pdf\u003c/span\u003e\u003cspan address=\"https://www.rcp.ac.uk/media/a4ibkkbf/news2-final-report_0_0.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCovino M, Sandroni C, Della Polla D, De Matteis G, Piccioni A, De Vita A, et al. Predicting ICU admission and death in the Emergency Department: A comparison of six early warning scores. Resuscitation. 2023;190:109876. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.resuscitation.2023.109876\u003c/span\u003e\u003cspan address=\"10.1016/j.resuscitation.2023.109876\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUriyama A, Urushidani S, Nakayama T. Five-level emergency triage systems: variation in assessment of validity. Emerg Med J. 2017;34(11):703\u0026ndash;10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1136/emermed-2016-206295\u003c/span\u003e\u003cspan address=\"10.1136/emermed-2016-206295\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePlante E, Vance R. Diagnostic accuracy of two tests of preschool language. Am J Speech Lang Pathol. 1994;3(1):57\u0026ndash;64. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1044/1058-0360.0301.57\u003c/span\u003e\u003cspan address=\"10.1044/1058-0360.0301.57\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePower M, Fell G, Wright M. Principles for high-quality, high-value testing. BMJ Evid Based Med. 2013;18(1):5\u0026ndash;10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1136/eb-2012-100645\u003c/span\u003e\u003cspan address=\"10.1136/eb-2012-100645\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZachariasse JM, Seiger N, Rood PPM, Alves CF, Freitas P, Smit FJ, et al. Validity of the Manchester Triage System in emergency care: A prospective observational study. PLoS ONE. 2017;12(2):e0170811. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0170811\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0170811\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTsiftsis D, Tasioulis A, Bampalis D. Adult triage in the emergency department: Introducing a multi-layer triage system. Healthcare. 2025;13(9):1070. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/healthcare13091070\u003c/span\u003e\u003cspan address=\"10.3390/healthcare13091070\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohansson A, Ekwall A, Forberg JL, Ekelund U. Development of outcomes for evaluating emergency care triage: a Delphi approach. Scand J Trauma Resusc Emerg Med. 2023;31(1):10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13049-023-01073-1\u003c/span\u003e\u003cspan address=\"10.1186/s13049-023-01073-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"triage systems, emergency department, physician judgment, acuity stratification, layered triage model, ICU prediction, physician-assisted triage, predictive validation","lastPublishedDoi":"10.21203/rs.3.rs-6951401/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6951401/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTriage in emergency departments (EDs) must extend beyond simply prioritizing patients for physician evaluation—it should also support the efficient allocation of limited clinical resources in fast-paced, high-pressure settings. This study evaluates a dual-layer triage model that integrates two complementary approaches: the structured, nurse-led \u003cem\u003eAcuity Categorization Tool\u003c/em\u003e (ACT) and the physician-driven \u003cem\u003eAdaptive Judgment of Urgency and Streaming\u003c/em\u003e (ADJUST). The model combines standardized initial assessment with dynamic, context-sensitive reassessment grounded in clinical insight and operational realities.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe retrospectively analyzed 53,651 consecutive ED encounters to assess the performance of ACT and ADJUST in identifying patients requiring intensive care unit (ICU) admission. Sensitivity, specificity, and reclassification rates were calculated using real-time triage documentation. To reflect the model’s distinct urgency thresholds, prognostic performance was evaluated exclusively for the highest acuity category (Red). Analyses included all encounters with complete data; documentation gaps in very high-acuity cases were acknowledged, as immediate clinical action often preempted formal triage.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAmong 409 ICU transfers, ACT demonstrated higher sensitivity (81.2%) than ADJUST (68.6%), aligning with its role in early detection. In contrast, ADJUST achieved greater specificity (96.0% vs. 85.9%), supporting more selective prioritization and improved resource alignment. Across the cohort, 45.3% of ACT classifications were downgraded and 3.6% upgraded by ADJUST. In a subset of 42,881 encounters with complete documentation, 20.9% of ICU patients initially flagged as Red by ACT were reclassified to lower acuity by ADJUST. Nearly half of all ADJUST evaluations occurred within 15 minutes of ACT.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis dual-layer triage model illustrates the complementary strengths of protocol-based scoring and physician judgment. ACT supports efficient and standardized initial triage, while ADJUST offers targeted refinement through clinical expertise and system-level awareness. The timing and reclassification patterns observed suggest that the two layers often functioned as an integrated, real-time decision-making process rather than discrete sequential steps. These findings align with performance benchmarks for established triage systems and suggest that integrated frameworks combining structured assessment with clinical discretion warrant further evaluation to improve acuity detection and care coordination in emergency care.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number: \u003c/strong\u003enot applicable.\u003c/p\u003e","manuscriptTitle":"ACT and ADJUST: Evaluating a Layered Emergency Triage Model Integrating Structured Scoring and Physician Judgment","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-27 18:06:18","doi":"10.21203/rs.3.rs-6951401/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"42f7d1dc-ed64-4f9e-ae58-9a5caee9bb12","owner":[],"postedDate":"June 27th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-07-12T20:38:18+00:00","versionOfRecord":[],"versionCreatedAt":"2025-06-27 18:06:18","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6951401","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6951401","identity":"rs-6951401","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}