Cost-Effectiveness of a Host-Response Point-of-Care Assay for Early Sepsis Diagnosis in UK Emergency Departments

Abstract

Background

Early recognition of sepsis in emergency departments (EDs) improves outcomes and reduces healthcare resource use. However, current clinical practice (CCP), largely based on the National Early Warning Score 2 (NEWS2) and clinical judgement, has limited diagnostic sensitivity at presentation. Consequently, some patients with early sepsis are not promptly identified, resulting in delayed treatment, disease progression, and increased use of high-cost care. Rapid point-of-care host-response diagnostics may complement clinical assessment by identifying patients at risk who do not initially meet conventional clinical thresholds.

Objective

To estimate the cost-effectiveness of adding a rapid LPS-induced TNF-α release point-of-care assay (TARA) to current clinical practice, compared with procalcitonin (PCT) plus CCP and CCP alone, from the UK National Health Service (NHS) and Personal Social Services perspective, in line with NICE reference case methods.

Methods

A decision-analytic model simulated diagnostic classification, timing of sepsis recognition (early, delayed, late), initiation of antibiotic treatment, hospital admission (general ward or intensive care unit [ICU]), discharge, and readmission among adults presenting to UK EDs with suspected sepsis without shock. Model inputs included diagnostic accuracy, clinical outcomes, health-related quality of life, and costs derived from published literature and UK-relevant sources. The base-case time horizon was lifetime, to fully capture the QALY consequences of prevented deaths and long-term sequelae (ESRD, amputation); a 6-month scenario analysis was also conducted. Deterministic and probabilistic sensitivity analyses assessed parameter uncertainty.

Results

In the deterministic base-case analysis, TARA plus CCP was associated with lower total costs and higher quality-adjusted life-years (QALYs) compared with both comparator strategies. Cost savings were approximately £370 per patient versus PCT plus CCP and £1,822 per patient versus CCP alone, with incremental QALY gains of 0.0834 and 0.0836, respectively. In the probabilistic analysis, mean cost savings were £391 versus PCT plus CCP and £1,835 versus CCP alone. Benefits were driven primarily by reductions in false-negative diagnoses, leading to earlier treatment, lower ICU use, and reduced downstream hospital costs.

Conclusions

Integrating a rapid host-response point-of-care assay into ED sepsis pathways in the UK was associated with lower total costs (savings of approximately £370 per patient versus PCT plus CCP and £1,822 per patient versus CCP alone) and marginally higher quality-adjusted life-years (QALY gains of 0.0834 and 0.0836, respectively), driven by reductions in missed diagnoses and downstream use of high-cost care.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s40291-026-00860-7.

Key points for decision makers

Early sepsis diagnosis in emergency departments remains challenging, and missed cases are associated with delayed treatment, increased intensive care utilization, and higher healthcare costs for the NHS.

Adding a rapid host-response point-of-care assay to current clinical practice was modelled to improve early identification of sepsis while reducing downstream use of high-cost resources, particularly ICU admissions and length of stay.

From an NHS and Personal Social Services perspective, the strategy was cost saving and associated with modest health gains, with a high probability of cost-effectiveness at conventional UK willingness-to-pay thresholds.

These findings support further UK-based validation and consideration of host-response point-of-care diagnostics as a system efficiency intervention within emergency sepsis pathways.

Introduction

Sepsis is a leading cause of preventable hospital mortality in the UK, with an estimated 245,000 cases annually and over 48,000 deaths [1]. A comprehensive study analyzing over 426,000 intensive care unit (ICU) sepsis admissions from 261 ICUs between 1988 and 2019 reported a 30-year overview of sepsis epidemiology and outcomes using data from the UK APACHE II and Intensive Care National Audit and Research Centre databases [2]. The study included standardized information on demographics, comorbidities, diagnoses, APACHE II scores, and mortality. Key findings revealed a 2.6-fold increase in sepsis as a proportion of ICU cases (rising from 7.4% to 19.6%), a higher number of admissions originating from emergency departments (ED), shifts in sources of infection, and a trend toward older but less acutely ill patients. Initiatives such as the Surviving Sepsis Campaign, an international quality improvement program launched in 2002 (and last updated in 2026) that develops evidence-based management guidelines and promotes their implementation across healthcare systems worldwide, have supported earlier diagnosis of less severe cases, earlier interventions, improved ICU care, and better patient outcomes [3].

Despite advances in sepsis management in recent years, early sepsis diagnosis or triage still depends on nonspecific clinical variables included in the National Early Warning Score 2 (NEWS2). NEWS2 is widely used and can predict acute deterioration, but its accuracy for detecting early sepsis at presentation is suboptimal: for instance, a NEWS2 threshold (≥5 points) yields an estimated sensitivity of only ~65% for sepsis in ED patients [4]. This approach misses a significant proportion of early cases, patients who may appear less ill but are at high risk for disease progression, organ dysfunction, and death. These individuals continue to face an unacceptably high risk of developing multiple organ dysfunction and mortality [5]. Timely diagnosis in sepsis is critical, as delays in antibiotic administration correlate strongly with increased mortality [6, 7]. Despite established screening tools such as the quick Sequential Organ Failure Assessment (qSOFA) and the NEWS2, diagnostic performance remains suboptimal, particularly for early-stage sepsis without overt or clinically evident organ dysfunction [8]. Emerging point-of care biomarker diagnostics offer a promising approach to this clinical challenge, as recent studies have shown biomarker-enhanced clinical scores improve the diagnostic accuracy of clinical scores alone.[9–11]

The rapid lipopolysaccharide (LPS)-induced TNF-release assay (TARA) is a blood-based assay, which identifies sepsis within a few hours of ED presentation. TARA is designed to detect endotoxin tolerance: in patients with nascent sepsis due to bacterial infection, immune cells fail to release TNF-α upon ex vivo LPS stimulation [12]. By integrating this tool into the initial work-up, clinicians could better differentiate between bacterial sepsis and non-infectious causes of illness, allowing more precise and timely interventions. This indicates TARA may help catch cases that clinical judgement or NEWS2 would otherwise miss, potentially initiating antibiotics and supportive care sooner in those patients. This analysis focuses specifically on bacterial sepsis; the LPS-based mechanism of TARA detects the host response to bacterial endotoxin and is not designed to detect fungal infections, which are therefore outside the scope of this model.

Given the high burden of sepsis and the critical importance of timely treatment, any new diagnostic needs to be evaluated not only for clinical efficacy but also for its cost-effectiveness in the health system. Sepsis care can be resource-intensive, especially if diagnosis is delayed and patients progress to septic shock or organ failure requiring ICU admission. Prior research has shown that delayed sepsis recognition sharply increases treatment costs and worsens outcomes. Health systems like the NHS must therefore balance the benefits of improved sepsis detection with the costs of implementing new testing strategies. This study presents a comprehensive cost-effectiveness analysis of integrating the point-of-care TARA assay into usual care for suspected sepsis in the UK ED, compared with the relevant alternatives (procalcitonin (PCT)-based screening and standard care without biomarkers). The model adheres to UK National Institute for Health and Care Excellence (NICE) guidelines for economic evaluations [13–16] and considers both clinical and financial implications from the National Health Service (NHS) and Personal Social Services (PSS) perspective. In addition to improving survival, earlier and more accurate diagnosis has implications for antimicrobial stewardship: while timely antibiotic administration is essential to reduce mortality in sepsis, improved diagnostic specificity may reduce unnecessary or inappropriate broad-spectrum antibiotic use in patients who do not have bacterial infection, thereby contributing to efforts to limit the emergence of antimicrobial resistance, a priority aligned with the UK's national action plan on antimicrobial resistance.

Methods

Model Structure

A decision-tree model was constructed in Microsoft Excel to simulate the clinical trajectories of adults presenting to EDs with suspected sepsis but without septic shock. The model included three diagnostic strategies which were defined as: 1) Current Clinical Practice (CCP): Standard care using NEWS2 or qSOFA, clinical judgement, and microbiological cultures; 2) PCT + CCP: Adds PCT biomarker testing to CCP; and 3) Point-of-care TARA + CCP: Integrates point-of-care TARA as an adjunct diagnostic test alongside CCP. Performance of the diagnostic strategies was measured based on sensitivity and specificity. For combined diagnostic strategies, the model assumes that a patient is identified as positive if either the clinical assessment component or the adjunct biomarker component is positive. Therefore, the diagnostic performance of a combined strategy does not directly correspond to the standalone sensitivity and specificity of the biomarker. In the case of PCT + CCP, the sensitivity of the combined pathway is driven by the higher sensitivity component, CCP in this case.

The model considered the point of ED presentation and followed patients through diagnosis, initial treatment, hospitalization (general ward or ICU), discharge, and potential readmission. The decision-tree model simulates the typical diagnosis and treatment pathway for adult patients presenting with possible sepsis without shock at the ED in the UK (Fig. 1). Patients were categorized based on diagnostic test results and stratified into three groups: (1) early sepsis identification; (2) delayed sepsis identification and (3) late sepsis identification, as described in Table 1. Base-case patient distributions are presented in Table 1. Deterministic and probabilistic sensitivity analyses were performed following CHEERS 2022 reporting guidelines [17].

Fig. 1.

Model schematic. Decision tree schematic of the diagnostic and treatment pathway for adults presenting to the emergency department with suspected sepsis without septic shock. All patients undergo current clinical practice assessment, with results assumed to be available within 1 hour. In the add-on testing strategies, procalcitonin or TARA results are assumed to be available within 3 hours. Based on diagnostic results and timing of recognition, patients with sepsis are classified as early identification, delayed identification, or late identification. Early identification corresponds to patients correctly identified by current clinical practice at presentation. Delayed identification corresponds to patients initially missed by current clinical practice but subsequently identified by the add-on test within 3 hours. Late identification corresponds to patients missed by both current clinical practice and the add-on test, with antibiotic treatment initiated after 3 hours following clinical deterioration or later recognition. Downstream outcomes include admission to the general ward or ICU, discharge, readmission, and long-term complications among late-identified sepsis survivors, including end-stage renal disease and amputation. CCP current clinical practice; ED emergency department; ICU intensive care unit; PCT procalcitonin; TARA LPS-induced TNF-α release assay.

Table 1.

Four Subgroups of Patients in the Model

Subgroup	Description of patients	Expected treatment pathway and clinical outcomes
Early sepsis identification	In all arms: CCP true positive patients	Patients receive antibiotics treatment early (assumed within 1 hour) Patients are admitted to general ward or ICU Patients have an increased mortality due to sepsis Patients are at risk of being readmitted following discharge
Delayed sepsis identification	Only in the TARA/PCT + CCP arms: CCP false negative patients within 1 hour, followed by TARA/PCT true positive results within 3 hours	Patients receive delayed antibiotics treatment (assumed within 3 hours) Patients may be observed in the ED or sent home after the CCP false negative result Once the patient’s symptoms deteriorate, they return to the ED and are admitted to the general ward or ICU Patients have an increased mortality due to delayed sepsis diagnosis Patients are at higher risk of being readmitted following discharge, compared with the early sepsis identification group
Late sepsis identification	CCP alone arm: CCP false negative patients within 1 hour TARA/PCT + CCP arms: CCP false negative patients within 1 hour, followed by TARA/PCT false negative results within 3 hours	Patients receive late antibiotics treatment (beyond 3 hours) and may be observed in the ED or sent home Once the patient’s symptoms deteriorate, they return to the ED and are admitted to the general ward or ICU Patients have at the highest increased mortality due to late sepsis diagnosis Patients are at highest risk of being readmitted following discharge, compared with the early sepsis identification and delayed sepsis identification subgroups Patients are at risk of having sepsis-induced ESRD and amputations
No sepsis	In all arms: true negative patients In all arms: false positive patients	Patients with a true negative test result will be sent home and assumed to require no further treatment Patients with a false positive test result will receive antibiotics treatment and are admitted to the general ward or ICU. In the base case, 100% are assumed to be admitted to the general ward Patients have an increased risk of mortality while in the general ward Upon discharge they receive the same mortality rate as the general population These patients do not have a risk of being readmitted for sepsis following discharge and are not at risk of sepsis-induced ESRD and amputation

CCP current clinical practice; ED emergency department; ESRD end-stage renal disease; ICU intensive care unit; TARA LPS-induced TNF-α release assay; PCT procalcitonin.

Time Horizon and Perspective

The base-case time horizon was lifetime, to fully capture the mortality differential between diagnostic strategies (in-hospital mortality: 4.5% early, 13.8% delayed, 30.7% late identification) and long-term sequelae including end-stage renal disease (ESRD) and amputations [18, 19]. A 3.5% annual discount rate was applied to costs and QALYs in line with NICE reference case guidance. A 6-month scenario analysis was also conducted for comparison with prior estimates. The analysis adopted the perspective of the UK healthcare system, incorporating NHS and Personal Social Services costs. This perspective was chosen because it represents the NICE reference case, ensuring comparability with other technology appraisals conducted for decision-making in the UK. Wider societal costs (e.g., productivity losses) were not included, consistent with NICE guidelines. PSS costs were included in line with the NICE reference case. These encompass community care needs (e.g., domiciliary support following hospital discharge), rehabilitation services (e.g., physiotherapy and occupational therapy following ICU admission), and long-term care costs associated with chronic sequelae of sepsis, particularly end-stage renal disease and post-amputation care. PSS unit costs were derived from the Unit Costs of Health and Social Care (PSSRU 2023). Given the long-term morbidity frequently observed in sepsis survivors, inclusion of these components was considered essential to capture the full economic burden of the condition beyond the acute episode.

LPS-induced TNF-α Release Assay description

The rapid LPS-induced TNF-alpha release point-of-care assay (SeptiLoop®, Loop Diagnostics S.L., Barcelona, Spain) is CE-marked under the IVDR (Class C) and holds UKCA certification for near patient use. It is intended as an indirect qualitative test to aid in identifying patients at risk of LPS-induced immunosuppression during suspected bacterial bloodstream infection, including sepsis. The assay stimulates whole blood ex vivo with lipopolysaccharide (LPS) and, after a 2.5-hour incubation, detects tumor necrosis factor-alpha (TNF-α) on a lateral flow strip. In individuals with preserved innate immune function, immune cells respond to LPS by releasing TNF-α, yielding a visible test line. In patients with sepsis who exhibit endotoxin tolerance, immune cells fail to release TNF-α under the same stimulus, resulting in a negative test line that is interpreted as immunosuppression. The model assumes the TARA test is processed on an individual patient basis, with a time-to-result of 2.5 hours from sample receipt; this is the intended single-sample workflow for which the device is CE/UKCA certified. In practice, EDs may batch-process samples, which could extend the effective time-to-result beyond 2.5 hours for individual patients. This assumption is addressed in the sensitivity analysis.

Input Parameters

Diagnostic accuracy parameters for current clinical practice (CCP), PCT + CCP and point-of-care TARA + CCP were obtained from published indexed and internal data file studies, and UK literature, as detailed in Table 2 and Supplementary Table 1 [12, 20–22]. Diagnostic accuracy inputs were selected based on their relevance to the modelled population, clinical setting and diagnostic pathway. Preference was given to studies conducted in adult ED patients with suspected sepsis, reporting both sensitivity and specificity, and using diagnostic criteria aligned with contemporary sepsis definitions or routine clinical practice. Where more than one plausible estimate was available, alternative values were explored in scenario analyses.

Table 2.

Diagnostic performance of the different methods for early sepsis diagnosis

Diagnostic strategy	Sensitivity	Specificity	Source
CCP	0.71	0.60	Zhang, et al. (2021)[36]; NEWS2 ≥ 5 as sepsis detection tool
TARA + CCP	0.91	0.80	TARA IFU & SSP data on file (2025); trial was conducted in patients with suspected sepsis in the ED
PCT + CCP	0.71*	0.87	Tsalik et al. (2012)[22], Trial of sensitivity in ED patients with early sepsis

CCP current clinical practice; ED emergency department; NEWS National Early Warning Score; PCT procalcitonin.* The model uses the best sensitivity output of the two inputs.

The prevalence of sepsis at presentation, the probability of ICU admission, lengths of stay for general ward and ICU, readmission risk, and in-hospital and post-discharge mortality rates were sourced from contemporary observational studies and UK-relevant evidence [23–26]. Health-related quality of life was represented using utility weights for the acute ward and ICU states and post-discharge survivorship, with decrements for readmission, ESRD and amputation applied where appropriate. Unit costs, sources, and PSA distributional assumptions for all cost parameters are detailed in Supplementary Tables 1 and 2. The low aggregate cost contribution of amputation reflects a probability-weighted expected value: the absolute per-patient probability is approximately 0.22%, such that the population-level cost impact is minimal despite a unit procedure cost of £11,007. All data sources and assumptions are documented in the manuscript tables and references. This structure and sourcing mirror the reporting standards used in the reference model. Patients were split between those with and without sepsis (30% vs. 70%). Within the sepsis cohort, diagnostic timing impacted ICU admission risk, mortality, length of stay, and probabilities of readmission, ESRD, or amputation. All costs in the analysis were updated to 2023 values and encompassed a comprehensive range of healthcare expenditures.

Analytical Approach

Incremental cost-effectiveness ratios (ICERs) were calculated as the incremental cost per QALY gained. One-way sensitivity analysis (OWSA) identified influential parameters. Probabilistic sensitivity analysis (PSA) using Monte Carlo simulations provided estimates of uncertainty and cost-effectiveness acceptability curves (CEACs). Probabilistic sensitivity analysis was conducted using separate pairwise simulations comparing TARA + CCP with CCP alone and TARA + CCP with PCT + CCP. Each pairwise PSA used 10,000 iterations. Because the current model structure was implemented as a pairwise decision model, the probabilistic results and CEACs are interpreted within each pairwise comparison rather than as a single fully incremental three-strategy probabilistic analysis.

Ethics approval and consent to participate

Ethical approval and informed consent were not required for this study, as it is a model-based economic evaluation using aggregated data from previously published studies and anonymized secondary sources, without involvement of individual patient-level data.

Results

Point-of-care TARA combined with CCP demonstrated both clinical benefits and substantial system-level savings. In the base-case analysis results suggested that TARA + CCP compared with PCT + CCP and CCP alone increase total QALYs and decrease total direct costs in patients presenting to the ED with possible sepsis without shock in the UK. The base-case results and incremental results are presented in Table 3.

Table 3.

Deterministic model outputs for the three diagnostic strategies

	TARA + CCP	PCT + CCP	CCP alone
Costs (£)
Diagnostic tests	86.49	27.03	0.00
Antibiotic	21.11	26.75	26.75
Resource use (general ward, ICU)	3684.11	4090.35	5569.26
Readmission	103.25	96.78	96.78
End-stage renal disease	10.61	34.17	34.17
Amputation	0.16	0.53	0.53
Total costs	3905.72	4275.62	5727.50
Outcomes
Life-years	22.5443	22.3615	22.3615
QALYs	10.1360	10.0526	10.0524
End-stage renal disease per 100,000 patients	1.48	4.75	4.75
Amputation per 100,000 patients	1.52	4.89	4.89
Diagnostic quadrant distribution (% of total cohort)
True positive	27.3%	21.3%	21.3%
False negative	2.7%	8.7%	8.7%
True negative	56.0%	60.9%	42.0%
False positive	14.0%	9.1%	28.0%
Hospital resource use per patient
Mean days in general ward	4.039	2.642	5.099
Mean days in ICU	0.396	0.964	0.964

CCP current clinical practice; ESRD end-stage renal disease; ICU intensive care unit; PCT procalcitonin; QALY quality-adjusted life-year; TARA LPS-induced TNF-α release assay.

Compared with PCT + CCP, it yielded an incremental gain of 0.0834 QALYs and reduced mean costs by £369.90. Compared with CCP alone, point-of-care TARA achieved a gain of 0.0836 QALYs and a reduction of £1,821.78 per patient (Table 3). These per-patient differences are consistent in direction across all sensitivity analyses; however, a formal population-level budget impact analysis was beyond the scope of this study and would be required to quantify system-level implications. Both comparator strategies, CCP alone and PCT + CCP, had a false-negative rate of 8.7%. This reflects a conservative modelling assumption whereby PCT + CCP was assumed not to improve sensitivity beyond CCP alone, because the published PCT sensitivity estimate used in the model was lower than that of CCP alone. The use of TARA + CCP reduced the percentage of false negatives to 2.7%. Patients receiving TARA+ CCP had higher LYs and a reduced number of sepsis-induced ESRDs and amputation.

Point-of-care TARA + CCP is highly robust across multiple scenarios. Its dominance holds even when input values are pushed to upper and lower plausible bounds. The OWSA confirms that ICU-related costs and sepsis-related mortality are key cost-effectiveness drivers, see tornado plot in Figs. 2 and 3. The mean total costs, mean QALYs, incremental costs, incremental QALYs, net monetary benefit and ICERs from 10,000 PSA iterations are summarized in Table 4. In the pairwise probabilistic analyses, TARA + CCP remained dominant versus both CCP alone and PCT + CCP, with lower costs and higher QALYs in each comparison. The Supplementary Figure 1 presents the cost-effectiveness acceptability curves. The probabilistic ICERs were very similar to the deterministic ICERs. In OWSAs, both ICERs remained in the Southeast quadrant (being more effective and less costly than its comparators). Both ICERs were most sensitive to ICU unit cost, in-hospital mortality for patients with late sepsis identification, post-discharge mortality for patients with delayed and late sepsis identification, and utility weight during the general ward stay. The sensitivity of the cost-effectiveness results to small incremental QALYs was attributed to the formula used to calculate the ICER, which is the division of incremental costs by incremental QALYs (Figs. 4 & 5). Given this formula, when the incremental QALYs are small, even slight changes can lead to substantial variation in the ICER. The main drivers of cost-effectiveness were reduced ICU use and lower overall hospital resource costs, alongside a lower incidence of sepsis-related complications such as end-stage renal disease and amputations. For the NHS, these outcomes imply direct cost avoidance in high-cost critical care settings. These findings highlight not only strong statistical support for adoption but also a consistent direction of benefit under uncertainty.

Fig. 2.

Tornado diagram: Point-of-care TARA + CCP vs. CCP Alone. Deterministic sensitivity analysis comparing TARA + CCP with CCP alone. The x-axis displays the incremental cost-effectiveness ratio (ICER, £ per QALY gained). Each bar shows the range of the ICER when the corresponding parameter is varied between its lower and upper values. The three values shown in parentheses beside each parameter label represent the lower bound, base-case (midpoint) value, and upper bound, respectively. The base-case ICER value is the same for all parameters within this figure and is indicated by the vertical reference line. Negative ICER values reflect scenarios in which TARA + CCP is associated with lower costs and higher QALYs compared with CCP alone, and should therefore be interpreted as dominance rather than as a conventional positive cost-effectiveness ratio. The parameter labelled “Sensitivity of CCP alone” refers to variation in the sensitivity of the clinical assessment component of the comparator pathway; when CCP sensitivity is lower, more patients with sepsis remain initially unidentified and may benefit from TARA. CCP current clinical practice; ICER incremental cost-effectiveness ratio; QALY quality-adjusted life-year; TARA LPS-induced TNF-α release assay.

Fig. 3.

Tornado diagram Point-of-care TARA + CCP vs. PCT+ CCP. Deterministic sensitivity analysis comparing TARA + CCP with PCT + CCP. The x-axis displays the incremental cost-effectiveness ratio (ICER, £ per QALY gained). Each bar shows the range of the ICER when the corresponding parameter is varied between its lower and upper values. The three values shown in parentheses beside each parameter label represent the lower bound, base-case (midpoint) value, and upper bound, respectively. The base-case ICER value is the same for all parameters within this figure and is indicated by the vertical reference line. Negative ICER values reflect scenarios in which TARA + CCP is associated with lower costs and higher QALYs compared with PCT + CCP, and should therefore be interpreted as dominance rather than as a conventional positive cost-effectiveness ratio. The parameter labelled “Sensitivity of PCT + CCP” refers to variation in the sensitivity of the combined comparator pathway; when PCT + CCP sensitivity is lower, the incremental clinical and economic value of TARA increases. CCP current clinical practice; ICER incremental cost-effectiveness ratio; PCT procalcitonin; QALY quality-adjusted life-year; TARA LPS-induced TNF-α release assay.

Table 4.

Probabilistic sensitivity analysis results for pairwise diagnostic strategy comparisons

Comparison	Incremental cost (£)	Incremental QALY	Mean NMB (£)	ICER (£/QALY)
TARA + CCP vs PCT + CCP	−391	0.0857	2533	− 4563	Dominant
TARA + CCP vs CCP alone	−1,835	0.0872	4015	− 21037	Dominant

Values are based on 10,000 probabilistic simulations. Incremental costs, incremental QALYs, net monetary benefit, and ICERs are reported at a willingness-to-pay threshold of £25,000 per QALY gained. CCP current clinical practice; ICER incremental cost-effectiveness ratio; NMB net monetary benefit; PCT procalcitonin; QALY quality-adjusted life-year; TARA LPS-induced TNF-α release assay.

Fig. 4.

Cost-effectiveness plane: Point-of-care TARA + CCP vs. CCP Alone. Cost-effectiveness plane from the probabilistic sensitivity analysis comparing TARA + CCP with CCP alone. Each point represents one probabilistic simulation (n = 10,000). The x-axis shows incremental QALYs and the y-axis shows incremental costs in pounds sterling (£). The dashed line represents the willingness-to-pay threshold of £25,000 per QALY gained. The red point indicates the deterministic base-case estimate, and the yellow point indicates the probabilistic mean estimate. Most simulations fall in the southeast quadrant, indicating that TARA + CCP is generally associated with lower costs and higher QALYs compared with CCP alone. CCP current clinical practice; PSA probabilistic sensitivity analysis; QALY quality-adjusted life-year; TARA LPS-induced TNF-α release assay; WTP willingness to pay.

Fig. 5.

Cost-effectiveness plane: Point-of-care TARA + CCP vs. PCT + CCP. Cost-effectiveness plane from the probabilistic sensitivity analysis comparing TARA + CCP with PCT + CCP. Each point represents one probabilistic simulation (n = 10,000). The x-axis shows incremental QALYs and the y-axis shows incremental costs in pounds sterling (£). The dashed line represents the willingness-to-pay threshold of £25,000 per QALY gained. The red point indicates the deterministic base-case estimate, and the yellow point indicates the probabilistic mean estimate. Simulations falling below the willingness-to-pay threshold indicate that TARA + CCP is cost-effective compared with PCT + CCP; simulations in the southeast quadrant indicate dominance, with lower costs and higher QALYs. CCP current clinical practice; PCT procalcitonin; PSA probabilistic sensitivity analysis; QALY quality-adjusted life-year; TARA, LPS-induced TNF-α release assay; WTP willingness to pay.

Discussion

This model-based economic evaluation shows that integrating point-of-care TARA into UK ED diagnostic protocols has implications beyond incremental cost-effectiveness. Its higher diagnostic sensitivity enables earlier treatment, reducing downstream ICU admissions and length of stay. When scaled to the annual number of sepsis presentations in UK EDs, these modelled reductions translate into a substantial release of critical care bed-days, with potential relevance for managing seasonal capacity pressures, including winter demand. These effects could alleviate NHS bed pressures, ease ED overcrowding, and release scarce intensive care capacity for other critically ill patients.

From an economic perspective, point-of-care TARA was both less costly and more effective than current alternatives, confirming that improved diagnostic accuracy can be achieved without additional financial burden. Although the term “dominance” is standard in health economic evaluation, the practical clinical implication is clear: better outcomes at lower cost. Point-of-care TARA reduced average costs by ~32% compared with CCP alone, largely due to shorter ICU stays and fewer complications such as ESRD and amputations. These findings are consistent with evidence showing that delayed sepsis diagnosis sharply increases costs. Although incremental QALY gains were modest, reflecting the diagnostic nature of the intervention, the cost-effectiveness conclusions were driven primarily by avoided high-cost care, particularly ICU admission and prolonged hospitalization, rather than by health utility gains, and should be interpreted accordingly.

NEWS2 remains the UK standard for sepsis triage, serving as a valuable non-specific tool in prehospital and ED settings, particularly when presentations are unclear [27]. NEWS2 demonstrates high sensitivity and specificity for predicting early mortality and organ dysfunction but is less reliable for in-hospital and 30-day mortality [4]. Its accuracy in ED populations is inconsistent, especially in patients without overt organ dysfunction. Although international guidelines endorse NEWS2, evidence indicates that biomarkers outperform it in identifying sepsis [5, 9, 11].

Point-of-care TARA complements NEWS2 by detecting cases missed by scoring alone. In doing so, it reduces diagnostic errors of omission,[28] which are associated with delayed treatment, prolonged hospital stays, and increased costs [29, 30]. Rather than replacing NEWS2, point-of-care TARA enhances it by providing an objective host-response signal at the point of care. Pragmatically, a NEWS2 score ≥5 identifies severely ill patients who require urgent senior review to determine whether sepsis is present and initiate treatment [27]. For optimal performance, NEWS2 should be used for continuous monitoring rather than single-point assessment [31].

From a policy perspective, these findings highlight the potential value of improved early diagnostic sensitivity in sepsis pathways and may inform future NHS decision-making, alongside prospective UK validation and implementation studies. Missed or delayed sepsis diagnosis is a major source of avoidable harm and litigation. By reducing false negatives from 8.7 to 2.7%, point-of-care TARA may lower medico-legal risk, improve compliance with national safety standards, and enable earlier treatment. Its higher sensitivity translates into fewer cases progressing to multi-organ failure and improved outcomes. Although generating more false positives than PCT, the reduction in missed cases supports earlier and more accurate triage. This facilitates targeted therapy, prevents both undertreatment and unnecessary prolonged antibiotic use, and directly advances the UK antimicrobial resistance strategy. While probabilistic sensitivity analyses were conducted and demonstrated a consistent direction of effect, further work should explicitly report cost-effectiveness acceptability across a range of willingness-to-pay thresholds under alternative implementation scenarios to support policy decision-making.

Sepsis outcomes are strongly linked to timely diagnosis. Mortality is higher when sepsis is not recognized at admission compared with cases present on admission (25.6% vs. 11.4%) [32]. Delayed diagnosis is associated with worse outcomes and markedly higher costs. Hospital costs rise with severity, and patients diagnosed after admission incur the greatest expense. While the largest number of cases and total costs occur in the lowest severity group (sepsis without organ dysfunction), both mortality and per-patient costs increase substantially with delayed recognition.

This study also extends the economic evidence base beyond PCT. Previous UK-focused evaluations of PCT primarily demonstrated value in guiding antibiotic stewardship and reducing unnecessary antimicrobial use, but not in improving early diagnostic accurac [33–35]. Adding procalcitonin (PCT) to current clinical practice did not reduce false-negative classifications compared with CCP alone because PCT has limited sensitivity at ED presentation, particularly in early or non-shock sepsis, where delayed biomarker kinetics result in many true cases remaining below diagnostic thresholds. In contrast, point-of-care TARA increases sensitivity at first presentation, reducing underdiagnosis without markedly increasing false positives. This represents a novel contribution that complements, rather than duplicates, prior biomarker assessments.

Finally, in the context of rising NHS emergency demand and the imperative of antimicrobial stewardship, point-of-care TARA provides a practical means to improve efficiency [27]. Earlier and more accurate triage enables targeted therapy, reducing both undertreatment and inappropriate prolonged broad-spectrum antibiotic use. By decreasing unnecessary ICU admissions and supporting focused treatment, point-of-care TARA may alleviate NHS capacity pressures while advancing antimicrobial resistance goals.

Implementation of a point of care immune response assay in NHS EDs is likely to affect workflow primarily through test ordering at arrival and clinical decision making once results are available. In the model, current clinical practice results are assumed available within 1 hour and the add on test results within 3 hours, enabling a delayed identification pathway for patients initially missed by current clinical practice. Differences in ED staffing, laboratory connectivity, and protocol adherence across NHS Trusts may influence the real-world timeliness of actions triggered by results, which could lead to heterogeneity in effectiveness and cost savings. Equity considerations are therefore relevant: Trusts with constrained ED capacity or delayed escalation pathways may gain less benefit unless implementation is accompanied by standardized training, clear decision protocols, and monitoring of time to antibiotics and admission decisions.

Limitations and future directions

A key limitation is that diagnostic accuracy inputs were obtained from different sources. The TARA + CCP data were derived from a prospective diagnostic accuracy study rather than an RCT, and the comparisons with PCT + CCP and CCP alone are therefore indirect. Differences in study populations, sepsis prevalence, clinical workflows and reference standards may affect the comparability of diagnostic performance estimates. This uncertainty was addressed through scenario, probabilistic and threshold analyses.

Despite its strengths, this study has several limitations inherent to model-based economic evaluations of diagnostic technologies. First, diagnostic performance inputs for point-of-care TARA were derived from clinical studies conducted outside the UK and from data on file, which may limit their generalizability to UK ED populations. Although parameter uncertainty was explored through extensive sensitivity analyses, the results remain dependent on assumptions regarding comparative diagnostic accuracy and the relationship between earlier sepsis recognition and downstream clinical outcomes. Second, the model simplifies real-world ED pathways and cannot fully capture operational factors such as clinical inertia, variability in escalation practices, staffing constraints, or delays between test results and clinical action. The model assumes that in patients with a false-negative result on both clinical assessment and TARA, deterioration sufficient to trigger re-recognition of sepsis occurs beyond the 2.5-hour test window, consistent with evidence that hemodynamic progression in non-shock sepsis typically evolves over 4–6 hours. Importantly, the false-negative pathway applies to patients in whom both clinical assessment and the adjunct test are reassuring; patients with high clinical suspicion would be retained for observation regardless of the biomarker result, in line with NHS sepsis pathway guidance (NICE NG51). Differences in ED organization and resource availability across NHS Trusts may therefore influence the magnitude of any real-world effectiveness and cost savings. Third, long-term outcomes such as utility decrements associated with end-stage renal disease and amputations were informed by published international literature and may not fully reflect UK-specific survivorship trajectories. In addition, while the analysis focused on direct healthcare costs in line with NICE reference case requirements, broader system-level impacts such as antimicrobial resistance and productivity losses were not explicitly modelled.

Future work should prioritize prospective UK-based validation of diagnostic performance and workflow impact in the intended ED population. Specifically, multicenter studies comparing point-of-care TARA, procalcitonin, and current clinical practice within the same patient cohort are needed to minimize cross-study heterogeneity in diagnostic accuracy estimates. Such evaluations should prospectively capture test turnaround times, time to antibiotic initiation, admission and escalation decisions (ward versus ICU), and downstream resource use. Parallel collection of economic endpoints would enable refinement of model assumptions and assessment of real-world cost-effectiveness under routine NHS implementation conditions.

Conclusions

This model-based economic evaluation suggests that adding point-of-care TARA to current clinical practice in UK EDs could be associated with lower healthcare costs and marginal health gains by reducing missed sepsis cases and subsequent escalation to high-cost care. The direction and magnitude of these effects are driven by assumptions regarding diagnostic accuracy, timing of recognition, and the modeled relationship between earlier treatment and downstream outcomes. Given that incremental QALY gains were modest relative to the magnitude of cost savings, ICER estimates are numerically negative and should be interpreted as indicating dominance rather than as conventional positive cost-effectiveness ratios.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature.

Declarations

Conflict of interest

Some authors are employees of, or have received research support from, Loop Diagnostics S.L., StartUB! – University of Barcelona, Spain, the developer of the SeptiLoop® assay evaluated in this study. Loop Diagnostics S.L. is an early-stage academic spin-off originating from the University of Barcelona (StartUB!), with limited commercial activity at the time of the study. The economic model was developed by RTI Health Solutions using prespecified methods and assumptions, funded by the academic grant previously declared. The sponsor provided funding and access to data but had no role in the selection of comparators, model structure, analysis, interpretation of results, or the decision to submit the manuscript for publication. All authors had full access to the model outputs and take responsibility for the integrity and accuracy of the analysis.

Author Contributions

EHJ, QCC, EGJJ, and EPM conceived the study. JL, KZ, CMD, CK contributed to model development and data analysis. All authors contributed to interpretation of the results, drafting and critical revision of the manuscript, and approved the final version.

Acknowledgements

The authors thank RTI Health Solutions for their contribution to the development of the economic model.

Funding

This research was funded by the academic grant Ayudas a Doctorados industriales (DI-2024-00112 & DI-2024-00051), with the support of the Pla de doctorats industrials of the Department of Research and Universities of the Generalitat de Catalunya. The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and in the decision to submit the manuscript for publication.

Ethics approval

Not applicable. This study is a modelling analysis based exclusively on published literature and anonymized aggregate data; no primary data collection involving human participants was conducted.

Consent to participate and Consent for publication

Not applicable. No individual patient data, interviews, or identifiable information were used in this analysis.

Code Availability (Availability of Materials and Data)

The decision-analytic model was developed in Microsoft Excel. The model structure and key assumptions are described in the manuscript. The full model file is not publicly available but may be shared with academic researchers upon reasonable request, subject to confidentiality agreements. The data supporting the findings of this study are derived from a combination of published sources and aggregated clinical data used as model inputs. All parameter values, assumptions, and data sources are reported in the manuscript tables and references, and supplementary material. Clinical data on file used to inform diagnostic performance parameters are not publicly available due to commercial and confidentiality restrictions but are available from the corresponding author upon reasonable request and subject to data sharing agreements.