Agreement between arterial and end-tidal carbon dioxide in adult patients admitted with serious traumatic brain injury.

BACKGROUND: Low-normal levels of arterial carbon dioxide (PaCO2) are recommended in the acute phase of traumatic brain injury (TBI) to optimize oxygen and CO2 tension, and to maintain cerebral perfusion. End-tidal CO2 (ETCO2) may be used as a surrogate for PaCO2 when arterial sampling is less readily available. ETCO2 may not be an adequate proxy to guide ventilation and the effects on concomitant injury, time, and the impact of ventilatory strategies on the PaCO2-ETCO2 gradient are not well understood. The primary objective of this study was to describe the correlation and agreement between PaCO2 and ETCO2 in intubated adult trauma patients with TBI. METHODS: This study was a retrospective analysis of prospectively-collected data of intubated adult major trauma patients with serious TBI, admitted to the East of England regional major trauma centre; 2015-2019. Linear regression and Welch's test were performed on each cohort to assess correlation between paired PaCO2 and ETCO2 at 24-hour epochs for 120 hours after admission. Bland-Altman plots were constructed at 24-hour epochs to assess the PaCO2-ETCO2 agreement. RESULTS: 695 patients were included, with 3812 paired PaCO2 and ETCO2 data points. The median PaCO2-ETCO2 gradient on admission was 0.8 [0.4-1.4] kPa, Bland Altman Bias of 0.96, upper (+2.93) and lower (-1.00), and correlation R2 0.149. The gradient was significantly greater in patients with TBI plus concomitant injury, compared to those with isolated TBI (0.9 [0.4-1.5] kPa vs. 0.7 [0.3-1.1] kPa, p<0.05). Across all groups the gradient reduced over time. Patients who died within 30 days had a larger gradient on admission compared to those who survived; 1.2 [0.7-1.9] kPa and 0.7 [0.3-1.2] kPa, p<0.005. CONCLUSIONS: Amongst adult patients with TBI, the PaCO2-ETCO2 gradient was greater than previously reported values, particularly early in the patient journey, and when associated with concomitant chest injury. An increased PaCO2-ETCO2 gradient on admission was associated with increased mortality.

Clinical descriptors of disease trajectories in patients with traumatic brain injury in the intensive care unit (CENTER-TBI): a multicentre observational cohort study.

BACKGROUND: Patients with traumatic brain injury are a heterogeneous population, and the most severely injured individuals are often treated in an intensive care unit (ICU). The primary injury at impact, and the harmful secondary events that can occur during the first week of the ICU stay, will affect outcome in this vulnerable group of patients. We aimed to identify clinical variables that might distinguish disease trajectories among patients with traumatic brain injury admitted to the ICU. METHODS: We used data from the Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI) prospective observational cohort study. We included patients aged 18 years or older with traumatic brain injury who were admitted to the ICU at one of the 65 CENTER-TBI participating centres, which range from large academic hospitals to small rural hospitals. For every patient, we obtained pre-injury data and injury features, clinical characteristics on admission, demographics, physiological parameters, laboratory features, brain biomarkers (ubiquitin carboxy-terminal hydrolase L1 [UCH-L1], S100 calcium-binding protein B [S100B], tau, neurofilament light [NFL], glial fibrillary acidic protein [GFAP], and neuron-specific enolase [NSE]), and information about intracranial pressure lowering treatments during the first 7 days of ICU stay. To identify clinical variables that might distinguish disease trajectories, we applied a novel clustering method to these data, which was based on a mixture of probabilistic graph models with a Markov chain extension. The relation of clusters to the extended Glasgow Outcome Scale (GOS-E) was investigated. FINDINGS: Between Dec 19, 2014, and Dec 17, 2017, 4509 patients with traumatic brain injury were recruited into the CENTER-TBI core dataset, of whom 1728 were eligible for this analysis. Glucose variation (defined as the difference between daily maximum and minimum glucose concentrations) and brain biomarkers (S100B, NSE, NFL, tau, UCH-L1, and GFAP) were consistently found to be the main clinical descriptors of disease trajectories (ie, the leading variables contributing to the distinguishing clusters) in patients with traumatic brain injury in the ICU. The disease trajectory cluster to which a patient was assigned in a model was analysed as a predictor together with variables from the IMPACT model, and prediction of both mortality and unfavourable outcome (dichotomised GOS-E ≤4) was improved. INTERPRETATION: First-day ICU admission data are not the only clinical descriptors of disease trajectories in patients with traumatic brain injury. By analysing temporal variables in our study, variation of glucose was identified as the most important clinical descriptor that might distinguish disease trajectories in the ICU, which should direct further research. Biomarkers of brain injury (S100B, NSE, NFL, tau, UCH-L1, and GFAP) were also top clinical descriptors over time, suggesting they might be important in future clinical practice. FUNDING: European Union 7th Framework program, Hannelore Kohl Stiftung, OneMind, Integra LifeSciences Corporation, and NeuroTrauma Sciences.

Low Tidal Volume Ventilation Is Poorly Implemented for Patients in North American and United Kingdom ICUs Using Electronic Health Records.

BACKGROUND: Low tidal volume ventilation (LTVV; < 8 mL/kg predicted body weight [PBW]) is a well-established standard of care associated with improved outcomes. This study used data collated in multicenter electronic health record ICU databases from the United Kingdom and the United States to analyze the use of LTVV in routine clinical practice. RESEARCH QUESTION: What factors are associated with adherence to LTVV in the United Kingdom and North America? STUDY DESIGN: This was a retrospective, multicenter study across the United Kingdom and United States of patients who were mechanically ventilated. METHODS: Factors associated with adherence to LTVV were assessed in all patients in both databases who were mechanically ventilated for > 48 h. We observed trends over time and investigated whether LTVV was associated with patient outcomes (30-day mortality and duration of ventilation) and identified strategies to improve adherence to LTVV. RESULTS: A total of 5,466 (Critical Care Health Informatics Collaborative [CCHIC]) and 7,384 electronic ICU collaborative research database [eICU-CRD] patients were ventilated for > 48 h and had data of suitable quality for analysis. The median tidal volume (VT) values were 7.48 mL/kg PBW (CCHIC) and 7.91 mL/kg PBW (eICU-CRD). The patients at highest risk of not receiving LTVV were shorter than 160 cm (CCHIC) and 165 cm (eICU-CRD). Those with BMI > 30 kg/m2 (CCHIC OR, 1.9 [95% CI, 1.7-2.13]; eICU-CRD OR, 1.61 [95% CI, 1.49-1.75]) and female patients (CCHIC OR, 2.39 [95% CI, 2.16-2.65]; eICU-CRD OR, 2.29 [95% CI, 2.26-2.31]) were at increased risk of having median VT > 8 mL/kg PBW. Patients with median VT < 8 mL/kg PBW had decreased 30-day mortality in the CCHIC database (CCHIC cause-specific hazard ratio, 0.86 [95% CI, 0.76-0.97]; eICU-CRD cause-specific hazard ratio, 0.9 [95% CI, 0.86-1.00]). There was a significant reduction in VT over time in the CCHIC database. INTERPRETATION: There has been limited implementation of LTVV in routine clinical practice in the United Kingdom and the United States. VT > 8 mL/kg PBW was associated with worse patient outcomes.

A Solution to the Cerebral Perfusion Pressure Transducer Placement Conundrum in Neurointensive Care? The Dual Transducer.

Intracranial pressure is routinely monitored in most intensive care units caring for patients with severe neurological insults and, together with continuous arterial blood pressure measurement, allows for monitoring of cerebral perfusion pressure (CPP). CPP is the driving pressure of blood flow to the brain and is used to guide therapy. However, there is considerable inconsistency in the literature regarding how CPP is technically measured and, more specifically, the appropriate placement of the arterial pressure transducer. Depending on patient positioning and where the arterial pressure transducer is placed, the mean arterial pressure used for CPP calculation can vary widely by up to 15 mm Hg, which is greater than the acceptable variation in target ranges used clinically. Physiologically, the arterial pressure transducer should be placed at the level of the foramen of Monro for CPP measurement, but it is commonly set at the level of the right atrium for systematic measurement. Mean arterial pressure measurement at the level of the right atrium can lead to overestimation and potentially critically low actual CPP levels when the head is elevated, and measurement at the level of the foramen of Monro will underestimate systemic pressures, increasing the risk of excessive and unnecessary use of vasopressors and fluid. At the Karolinska University Hospital neurointensive care unit, we have used a split dual-transducer system, measuring arterial pressure both at the level of the foramen of Monro and at the level of the right atrium from a single arterial source. In doing so, we work with constants and can monitor and target optimum arterial pressures to better secure perfusion to all organs, with potentially less risk of cerebral ischemia or overuse of vasopressors and fluids, which may affect outcome.

Therapy Intensity Level Scale for Traumatic Brain Injury: Clinimetric Assessment on Neuro-Monitored Patients Across 52 European Intensive Care Units.

Intracranial pressure (ICP) data from traumatic brain injury (TBI) patients in the intensive care unit (ICU) cannot be interpreted appropriately without accounting for the effect of administered therapy intensity level (TIL) on ICP. A 15-point scale was originally proposed in 1987 to quantify the hourly intensity of ICP-targeted treatment. This scale was subsequently modified-through expert consensus-during the development of TBI Common Data Elements to address statistical limitations and improve usability. The latest 38-point scale (hereafter referred to as TIL) permits integrated scoring for a 24-h period and has a five-category, condensed version (TIL(Basic)) based on qualitative assessment. Here, we perform a total- and component-score analysis of TIL and TIL(Basic) to: 1) validate the scales across the wide variation in contemporary ICP management; 2) compare their performance against that of predecessors; and 3) derive guidelines for proper scale use. From the observational Collaborative European NeuroTrauma Effectiveness Research in TBI (CENTER-TBI) study, we extract clinical data from a prospective cohort of ICP-monitored TBI patients (n = 873) from 52 ICUs across 19 countries. We calculate daily TIL and TIL(Basic) scores (TIL24 and TIL(Basic)24, respectively) from each patient's first week of ICU stay. We also calculate summary TIL and TIL(Basic) scores by taking the first-week maximum (TILmax and TIL(Basic)max) and first-week median (TILmedian and TIL(Basic)median) of TIL24 and TIL(Basic)24 scores for each patient. We find that, across all measures of construct and criterion validity, the latest TIL scale performs significantly greater than or similarly to all alternative scales (including TIL(Basic)) and integrates the widest range of modern ICP treatments. TILmedian outperforms both TILmax and summarized ICP values in detecting refractory intracranial hypertension (RICH) during ICU stay. The RICH detection thresholds which maximize the sum of sensitivity and specificity are TILmedian ≥ 7.5 and TILmax ≥ 14. The TIL24 threshold which maximizes the sum of sensitivity and specificity in the detection of surgical ICP control is TIL24 ≥ 9. The median scores of each TIL component therapy over increasing TIL24 reflect a credible staircase approach to treatment intensity escalation, from head positioning to surgical ICP control, as well as considerable variability in the use of cerebrospinal fluid drainage and decompressive craniectomy. Since TIL(Basic)max suffers from a strong statistical ceiling effect and only covers 17% (95% confidence interval [CI]: 16-18%) of the information in TILmax, TIL(Basic) should not be used instead of TIL for rating maximum treatment intensity. TIL(Basic)24 and TIL(Basic)median can be suitable replacements for TIL24 and TILmedian, respectively (with up to 33% [95% CI: 31-35%] information coverage) when full TIL assessment is infeasible. Accordingly, we derive numerical ranges for categorising TIL24 scores into TIL(Basic)24 scores. In conclusion, our results validate TIL across a spectrum of ICP management and monitoring approaches. TIL is a more sensitive surrogate for pathophysiology than ICP and thus can be considered an intermediate outcome after TBI.

Common Data Elements for Disorders of Consciousness: Recommendations from the Working Group on Physiology and Big Data.

BACKGROUND: The implementation of multimodality monitoring in the clinical management of patients with disorders of consciousness (DoC) results in physiological measurements that can be collected in a continuous and regular fashion or even at waveform resolution. Such data are considered part of the "Big Data" available in intensive care units and are potentially suitable for health care-focused artificial intelligence research. Despite the richness in content of the physiological measurements, and the clinical implications shown by derived metrics based on those measurements, they have been largely neglected from previous attempts in harmonizing data collection and standardizing reporting of results as part of common data elements (CDEs) efforts. CDEs aim to provide a framework for unifying data in clinical research and help in implementing a systematic approach that can facilitate reliable comparison of results from clinical studies in DoC as well in international research collaborations. METHODS: To address this need, the Neurocritical Care Society's Curing Coma Campaign convened a multidisciplinary panel of DoC "Physiology and Big Data" experts to propose CDEs for data collection and reporting in this field. RESULTS: We report the recommendations of this CDE development panel and disseminate CDEs to be used in physiologic and big data studies of patients with DoC. CONCLUSIONS: These CDEs will support progress in the field of DoC physiologic and big data and facilitate international collaboration.

Mining the contribution of intensive care clinical course to outcome after traumatic brain injury.

Existing methods to characterise the evolving condition of traumatic brain injury (TBI) patients in the intensive care unit (ICU) do not capture the context necessary for individualising treatment. Here, we integrate all heterogenous data stored in medical records (1166 pre-ICU and ICU variables) to model the individualised contribution of clinical course to 6-month functional outcome on the Glasgow Outcome Scale -Extended (GOSE). On a prospective cohort (n = 1550, 65 centres) of TBI patients, we train recurrent neural network models to map a token-embedded time series representation of all variables (including missing values) to an ordinal GOSE prognosis every 2 h. The full range of variables explains up to 52% (95% CI: 50-54%) of the ordinal variance in functional outcome. Up to 91% (95% CI: 90-91%) of this explanation is derived from pre-ICU and admission information (i.e., static variables). Information collected in the ICU (i.e., dynamic variables) increases explanation (by up to 5% [95% CI: 4-6%]), though not enough to counter poorer overall performance in longer-stay (>5.75 days) patients. Highest-contributing variables include physician-based prognoses, CT features, and markers of neurological function. Whilst static information currently accounts for the majority of functional outcome explanation after TBI, data-driven analysis highlights investigative avenues to improve the dynamic characterisation of longer-stay patients. Moreover, our modelling strategy proves useful for converting large patient records into interpretable time series with missing data integration and minimal processing.

The lower limit of reactivity as a potential individualised cerebral perfusion pressure target in traumatic brain injury: a CENTER-TBI high-resolution sub-study analysis.

BACKGROUND: A previous retrospective single-centre study suggested that the percentage of time spent with cerebral perfusion pressure (CPP) below the individual lower limit of reactivity (LLR) is associated with mortality in traumatic brain injury (TBI) patients. We aim to validate this in a large multicentre cohort. METHODS: Recordings from 171 TBI patients from the high-resolution cohort of the CENTER-TBI study were processed with ICM+ software. We derived LLR as a time trend of CPP at a level for which the pressure reactivity index (PRx) indicates impaired cerebrovascular reactivity with low CPP. The relationship with mortality was assessed with Mann-U test (first 7-day period), Kruskal-Wallis (daily analysis for 7 days), univariate and multivariate logistic regression models. AUCs (CI 95%) were calculated and compared using DeLong's test. RESULTS: Average LLR over the first 7 days was above 60 mmHg in 48% of patients. %time with CPP < LLR could predict mortality (AUC 0.73, p = < 0.001). This association becomes significant starting from the third day post injury. The relationship was maintained when correcting for IMPACT covariates or for high ICP. CONCLUSIONS: Using a multicentre cohort, we confirmed that CPP below LLR was associated with mortality during the first seven days post injury.

Moving from non-emergency bleeps and long-range pagers to a hospital-wide, EHR-integrated secure messaging system: an implementer report.

INTRODUCTION: Obsolete bleep/long-range pager equipment remains firmly embedded in the National Health Service (NHS). OBJECTIVE: To introduce a secure, chart-integrated messaging system (Epic Secure Chat) in a large NHS tertiary referral centre to replace non-emergency bleeps/long-range pagers. METHODS: The system was socialised in the months before go-live. Operational readiness was overseen by an implementation group with stakeholder engagement. Cutover was accompanied by a week of Secure Chat and bleeps running in parallel. RESULTS: Engagement due to socialisation was high with usage stabilising approximately 3 months after go-live. Contact centre internal call activity fell significantly after go-live. No significant patient safety concerns were reported. DISCUSSION: Uptake was excellent with substantial utilisation well before cutover indirectly supporting high levels of engagement. The majority of those who previously carried bleeps were content to use personal devices for messaging because of user convenience after reassurance about privacy. CONCLUSION: An integrated secure messaging system can replace non-emergency bleeps with beneficial impact on service.

Comparative Effectiveness of Mannitol Versus Hypertonic Saline in Patients With Traumatic Brain Injury: A CENTER-TBI Study.

Increased intracranial pressure (ICP) is one of the most important modifiable and immediate threats to critically ill patients suffering from traumatic brain injury (TBI). Two hyperosmolar agents (HOAs), mannitol and hypertonic saline (HTS), are routinely used in clinical practice to treat increased ICP. We aimed to assess whether a preference for mannitol, HTS, or their combined use translated into differences in outcome. The Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI) Study is a prospective multi-center cohort study. For this study, patients with TBI, admitted to the intensive care unit (ICU), treated with mannitol and/or HTS, and aged ≥16 years were included. Patients and centers were differentiated based on treatment preference with mannitol and/or HTS based on structured, data-driven criteria such as first administered HOA in the ICU. We assessed influence of center and patient characteristics in the choice of agent using adjusted multi-variate models. Further, we assessed the influence of HOA preference on outcome using adjusted ordinal and logistic regression models, and instrumental variable analyses. In total, 2056 patients were assessed. Of these, 502 (24%) patients received mannitol and/or HTS in the ICU. The first received HOA was HTS for 287 (57%) patients, mannitol for 149 (30%) patients, or both mannitol and HTS on the same day for 66 (13%) patients. Two unreactive pupils were more common in patients receiving both (13, 21%), compared with patients receiving HTS (40, 14%) or mannitol (22, 16%). Center, rather than patient characteristics, was independently associated with the preferred choice of HOA (p-value <0.05). ICU mortality and 6-month outcome were similar between patients preferably treated with mannitol compared with HTS (odds ratio [OR] = 1.0, confidence interval [CI] = 0.4-2.2; OR = 0.9, CI = 0.5-1.6, respectively). Patients who received both also had a similar ICU mortality and 6-month outcome compared with patients receiving HTS (OR = 1.8, CI = 0.7-5.0; OR = 0.6, CI = 0.3-1.7, respectively). We found between-center variability regarding HOA preference. Moreover, we found that center is a more important driver of the choice of HOA than patient characteristics. However, our study indicates that this variability is an acceptable practice given absence of differences in outcomes associated with a specific HOA.

Towards autoregulation-oriented management after traumatic brain injury: increasing the reliability and stability of the CPPopt algorithm.

PURPOSE: CPPopt denotes a Cerebral Perfusion Pressure (CPP) value at which the Pressure-Reactivity index, reflecting the global state of Cerebral Autoregulation, is best preserved. CPPopt has been investigated as a potential dynamically individualised CPP target in traumatic brain injury patients admitted in intensive care unit. The prospective bedside use of the concept requires ensured safety and reliability of the CPP recommended targets based on the automatically-generated CPPopt. We aimed to: Increase stability and reliability of the CPPopt automated algorithm by fine-tuning; perform outcome validation of the adjusted algorithm in a multi-centre TBI cohort. METHODS: ICM + software was used to derive CPPopt and fine-tune the algorithm. Parameters for improvement of the algorithm were selected based on qualitative and quantitative assessment of stability and reliability metrics. Patients enrolled in the Collaborative European Neuro Trauma Effectiveness Research in TBI (CENTER-TBI) high-resolution cohort were included for retrospective validation. Yield and stability of the new algorithm were compared to the previous algorithm using Mann-U test. Area under the curves for mortality prediction at 6 months were compared with the DeLong Test. RESULTS: CPPopt showed higher stability (p < 0.0001), but lower yield compared to the previous algorithm [80.5% (70-87.5) vs 85% (75.7-91.2), p < 0.001]. Deviation of CPPopt could predict mortality with an AUC of [AUC = 0.69 (95% CI 0.59-0.78), p < 0.001] and was comparable with the previous algorithm. CONCLUSION: The CPPopt calculation algorithm was fine-tuned and adapted for prospective use with acceptable lower yield, improved stability and maintained prognostic power.

Diagnosing acute kidney injury ahead of time in critically ill septic patients using kinetic estimated glomerular filtration rate.

INTRODUCTION: Accurate and actionable diagnosis of Acute Kidney Injury (AKI) ahead of time is important to prevent or mitigate renal insufficiency. The purpose of this study was to evaluate the performance of Kinetic estimated Glomerular Filtration Rate (KeGFR) in timely predicting AKI in critically ill septic patients. METHODS: We conducted a retrospective analysis on septic ICU patients who developed AKI in AmsterdamUMCdb, the first freely available European ICU database. The reference standard for AKI was the Kidney Disease: Improving Global Outcomes (KDIGO) classification based on serum creatinine and urine output (UO). Prediction of AKI was based on stages defined by KeGFR and UO. Classifications were compared by length of ICU stay (LOS), need for renal replacement therapy and 28-day mortality. Predictive performance and time between prediction and diagnosis were calculated. RESULTS: Of 2492 patients in the cohort, 1560 (62.0%) were diagnosed with AKI by KDIGO and 1706 (68.5%) by KeGFR criteria. Disease stages had agreement of kappa = 0.77, with KeGFR sensitivity 93.2%, specificity 73.0% and accuracy 85.7%. Median time to recognition of AKI Stage 1 was 13.2 h faster for KeGFR, and 7.5 h and 5.0 h for Stages 2 and 3. Outcomes revealed a slight difference in LOS and 28-day mortality for Stage 1. CONCLUSIONS: Predictive performance of KeGFR combined with UO criteria for diagnosing AKI is excellent. Compared to KDIGO, deterioration of renal function was identified earlier, most prominently for lower stages of AKI. This may shift the actionable window for preventing and mitigating renal insufficiency.

Bayesian model selection for multilevel models using integrated likelihoods.

Multilevel linear models allow flexible statistical modelling of complex data with different levels of stratification. Identifying the most appropriate model from the large set of possible candidates is a challenging problem. In the Bayesian setting, the standard approach is a comparison of models using the model evidence or the Bayes factor. Explicit expressions for these quantities are available for the simplest linear models with unrealistic priors, but in most cases, direct computation is impossible. In practice, Markov Chain Monte Carlo approaches are widely used, such as sequential Monte Carlo, but it is not always clear how well such techniques perform. We present a method for estimation of the log model evidence, by an intermediate marginalisation over non-variance parameters. This reduces the dimensionality of any Monte Carlo sampling algorithm, which in turn yields more consistent estimates. The aim of this paper is to show how this framework fits together and works in practice, particularly on data with hierarchical structure. We illustrate this method on simulated multilevel data and on a popular dataset containing levels of radon in homes in the US state of Minnesota.

Development of a Reinforcement Learning Algorithm to Optimize Corticosteroid Therapy in Critically Ill Patients with Sepsis.

BACKGROUND: The optimal indication, dose, and timing of corticosteroids in sepsis is controversial. Here, we used reinforcement learning to derive the optimal steroid policy in septic patients based on data on 3051 ICU admissions from the AmsterdamUMCdb intensive care database. METHODS: We identified septic patients according to the 2016 consensus definition. An actor-critic RL algorithm using ICU mortality as a reward signal was developed to determine the optimal treatment policy from time-series data on 277 clinical parameters. We performed off-policy evaluation and testing in independent subsets to assess the algorithm's performance. RESULTS: Agreement between the RL agent's policy and the actual documented treatment reached 59%. Our RL agent's treatment policy was more restrictive compared to the actual clinician behavior: our algorithm suggested withholding corticosteroids in 62% of the patient states, versus 52% according to the physicians' policy. The 95% lower bound of the expected reward was higher for the RL agent than clinicians' historical decisions. ICU mortality after concordant action in the testing dataset was lower both when corticosteroids had been withheld and when corticosteroids had been prescribed by the virtual agent. The most relevant variables were vital parameters and laboratory values, such as blood pressure, heart rate, leucocyte count, and glycemia. CONCLUSIONS: Individualized use of corticosteroids in sepsis may result in a mortality benefit, but optimal treatment policy may be more restrictive than the routine clinical practice. Whilst external validation is needed, our study motivates a 'precision-medicine' approach to future prospective controlled trials and practice.

Can process mapping and a multisite Delphi of perioperative professionals inform our understanding of system-wide factors that may impact operative risk?

OBJECTIVES: To examine whether the use of process mapping and a multidisciplinary Delphi can identify potential contributors to perioperative risk. We hypothesised that this approach may identify factors not represented in common perioperative risk tools and give insights of use to future research in this area. DESIGN: Multidisciplinary, modified Delphi study. SETTING: Two centres (one tertiary, one secondary) in the UK during 2020 amidst coronavirus pressures. PARTICIPANTS: 91 stakeholders from 23 professional groups involved in the perioperative care of older patients. Key stakeholder groups were identified via process mapping of local perioperative care pathways. RESULTS: Response rate ranged from 51% in round 1 to 19% in round 3. After round 1, free text suggestions from the panel were combined with variables identified from perioperative risk scores. This yielded a total of 410 variables that were voted on in subsequent rounds. Including new suggestions from round two, 468/519 (90%) of the statements presented to the panel reached a consensus decision by the end of round 3. Identified risk factors included patient-level factors (such as ethnicity and socioeconomic status), and organisational or process factors related to the individual hospital (such as policies, staffing and organisational culture). 66/160 (41%) of the new suggestions did not feature in systematic reviews of perioperative risk scores or key process indicators. No factor categorised as 'organisational' is currently present in any perioperative risk score. CONCLUSIONS: Through process mapping and a modified Delphi we gained insights into additional factors that may contribute to perioperative risk. Many were absent from currently used risk stratification scores. These results enable an appreciation of the contextual limitations of currently used risk tools and could support future research into the generation of more holistic data sets for the development of perioperative risk assessment tools.