Hypothermia risk factors in patients with burns during emergency presentations: protocol for a retrospective cohort study Authors: Dhairya Deven Vayada, Maleea D Holbert, Brandon Meikle, Brett Paul Dyer, Carl Lisec, Marc Schnekenburger, Paul Baker, Monique Bertinetti, Andrew J A Holland, Roy Kimble, Anne Darton, Daniel Isacson, Varun Harish, Natalia Adanichkin, Rebecca Schrale, Linda Quinn, Bernard Carney, Bronwyn Griffin Published: 2026 ABSTRACT AbstractIntroductionBurn injuries constitute a significant health concern, requiring immediate first aid to mitigate further tissue damage and complications. Most countries worldwide recommend application of 20 min of cool running water (20CRW) within 3 hours of the burn as the cornerstone in burn first aid management. Despite its widespread acceptance and proven benefits in reducing the severity of burns and subsequent interventions, concerns regarding the risk of hypothermia following this intervention persist, representing at least a perceived barrier to the delivery of 20CRW. When it does occur, hypothermia in patients with burns has been associated with higher mortality rates, even after controlling for burn injury severity. Developing an understanding of the incidence of post-burn hypothermia following 20CRW, with a specific focus on potential predictive and/or causative factors, is quintessential.Methods and analysisA retrospective cohort study of all adult and paediatric patients with thermal burn injuries presenting to one of 11 participating Australian or New Zealand hospitals between 1 January 2024 and 31 December 2024 will be conducted. The primary outcome is the incidence of hypothermia in patients with burns following their arrival at the emergency department (ED). Secondary outcomes include influence of burn first aid cooling, risk factors influencing hypothermia, impact of hypothermia on clinical patient outcomes and incidence of hypothermia in patients with burns in non-ED settings.Ethics and disseminationEthical approval was granted by the Children’s Health Queensland Human Research Ethics Committee (CHQHREC; HREC Ref No: HREC/25/QCHQ/114285) as well as Health and Disability Ethics Committees, New Zealand (HDEC; Ref No: 2026 EXP 23892). The study findings will be formally disseminated through peer-reviewed journals and conference presentations. STRENGTHS AND LIMITATIONS OF THIS STUDY This retrospective cohort study investigates hypothermia in patients with burns in Australia and New Zealand, providing broad geographic and climate representation across the region. This retrospective cohort study investigates hypothermia in patients with burns in Australia and New Zealand, providing broad geographic and climate representation across the region. Potential confounders will be systematically identified using directed acyclic graphs constructed prior to data analysis, minimising the risk of ad hoc confounder selection and supporting a rigorous causal inference framework. Potential confounders will be systematically identified using directed acyclic graphs constructed prior to data analysis, minimising the risk of ad hoc confounder selection and supporting a rigorous causal inference framework. The estimated sample of 5000 burn presentations captured across a defined 12-month period enhances the study’s capacity to conduct robust subgroup analyses, partially offsetting the variability inherent in a multisite retrospective design. The estimated sample of 5000 burn presentations captured across a defined 12-month period enhances the study’s capacity to conduct robust subgroup analyses, partially offsetting the variability inherent in a multisite retrospective design. There will be variability in how and when body temperature is measured across sites, due to local policies and practice, which may impact the estimation of the incidence of hypothermia. There will be variability in how and when body temperature is measured across sites, due to local policies and practice, which may impact the estimation of the incidence of hypothermia. The lack of standardisation in first-aid delivery, including factors such as water temperature, duration and the location where first aid is provided, may limit the ability to draw definitive inferences between burns first-aid and hypothermia. The lack of standardisation in first-aid delivery, including factors such as water temperature, duration and the location where first aid is provided, may limit the ability to draw definitive inferences between burns first-aid and hypothermia. Background and rationale In addition to protecting the body from outside insult and pathogens, the skin plays a crucial role in regulating body temperature, termed thermoregulation. Where skin coverage is disrupted, such as following a burn injury, thermoregulation may be affected, potentially resulting in the development of hypothermia.1 The current recommendation following a burn injury is the application of 20 min of cool running water (20CRW) in most countries.110 Concerns about the risk of hypothermia have been identified as a barrier to the delivery of this recommendation.11 There is an association between hypothermia in patients with burns and poor patient outcomes,1214 including higher rates of mortality.13 At present, hypothermia is poorly defined in the literature, which prevents accurate calculation of the incidence of post-burn hypothermia. For example, when defined as a body temperature below 36°C, the reported incidence is as high as 62.9%,15 however, at a body temperature below 35°C, hypothermia is calculated to occur in only 1.6% of cases.12 Regardless of its definition, there is evidence to suggest that post-burn hypothermia may result in poorer patient outcomes, including increased length of hospital stay and greater mortality,13 although these outcomes are varied. In paediatric patients with burns, age and burn total body surface area (TBSA) percentages have been identified as significant predictors of hypothermia.13 In addition to developing hypothermia as a direct result of skin loss following a burn injury, the application of CRW as burn first-aid has also been questioned as a potential causative factor. Most burn care and treatment bodies around the world, including the Australian and New Zealand Burn Association, European Burn Association and British Burn Association, recommend applying 20CRW to all thermal burns within 3 hours of the injury.8 9 Despite the widespread adoption and advocacy of this practice, and its proven benefit in lessening burn progression,1 it has been met with caution in some areas due to the concern that it may result in an increased risk of developing hypothermia. Few studies have been conducted investigating the association between CRW application and hypothermia, and these have produced conflicting results. One study found an association between hypothermia and length of cooling with water in adult patients with burns >20% TBSA.15 An earlier study, however, did not find an association between prehospital first aid cooling with water and development of hypothermia in a cohort of adult and paediatric patients with burns.12 This highlights the need for an in-depth review of the incidence of post-burn hypothermia, with a specific focus on potential predictive and/or causative factors, such as treatment with 20CRW. Furthermore, as little is known about the effects that hypothermia may have on patients with burns, understanding the clinical outcomes of patients with burns with hypothermia is important. This study aims to fill this critical gap in the literature. It is anticipated that the results of this investigation can be used to further support the use of 20CRW following burn injuries or suggest caution in its application in certain circumstances and/or climates. Methods Aims The primary aim of this study is to determine the incidence of hypothermia, defined as a recorded body temperature below 36°C, in adult and paediatric patients with thermal burns presenting to emergency departments (EDs) across Australia and New Zealand. The secondary aims of this investigation include: (1) to examine the influence of burn first aid cooling on the development of hypothermia in patients with burns; (2) to assess the influence of hypothermia on outcomes of the patient with burns (such as length of hospital stay, intensive care unit (ICU) admission and mortality) and (3) to determine the potential causal risk factors for development of post-burn hypothermia and develop a prognostic model for these factors in Australian and New Zealand patients. Objectives The following seven objectives will be pursued in this investigation: Conduct a retrospective cohort review of patient electronic medical records (EMRs) from n=11 tertiary EDs (n=10 Australian, n=1 New Zealand) over a 12-month period to quantify the incidence of hypothermia among patients with acute burns. Conduct a retrospective cohort review of patient electronic medical records (EMRs) from n=11 tertiary EDs (n=10 Australian, n=1 New Zealand) over a 12-month period to quantify the incidence of hypothermia among patients with acute burns. Determine the association between application of CRW of any duration and the incidence of hypothermia in patients with acute burns presenting to the ED. Data will be stratified for running water duration, including less than 20 min, 20 min and more than 20 min. Determine the association between application of CRW of any duration and the incidence of hypothermia in patients with acute burns presenting to the ED. Data will be stratified for running water duration, including less than 20 min, 20 min and more than 20 min. Evaluate clinical and demographic characteristics (eg, burn size, depth, location, patient age) associated with hypothermia in patients with acute burns. Evaluate clinical and demographic characteristics (eg, burn size, depth, location, patient age) associated with hypothermia in patients with acute burns. Assess environmental and situational factors (eg, time of year, incident location, geography, remoteness) that may contribute to the development of hypothermia in this patient population. Assess environmental and situational factors (eg, time of year, incident location, geography, remoteness) that may contribute to the development of hypothermia in this patient population. Determine the impact of the mode of transport to the ED (eg, fixed wing, helicopter, road ambulance, boat) on the incidence and severity of hypothermia in patients with acute burns. Determine the impact of the mode of transport to the ED (eg, fixed wing, helicopter, road ambulance, boat) on the incidence and severity of hypothermia in patients with acute burns. Develop a predictive model/prognostic model for hypothermia risk in patients with acute burns based on identified risk factors. Develop a predictive model/prognostic model for hypothermia risk in patients with acute burns based on identified risk factors. Examine the impact of hypothermia on outcomes of the patient with burns—specifically focusing on the length of hospital stay, rate of ICU admissions and mortality. Examine the impact of hypothermia on outcomes of the patient with burns—specifically focusing on the length of hospital stay, rate of ICU admissions and mortality. Project design A retrospective cohort study will be conducted for adult and paediatric patients with burns who presented to one of 11 participating Australian or New Zealand tertiary hospitals between 1 January 2024 and 31 December 2024 with thermal burn injuries. The planned start date of the study is 15 April 2026, and the planned end date is 31 December 2026. Setting Data will be collected from 11 sites, including 10 in Australia and 1 in New Zealand. Efforts have been made to include a wide range of geographical locations with varying climates, temperatures and referral pools. These sites are: Concord Hospital: Concord Repatriation General Hospital is a major teaching hospital in Sydney, New South Wales (NSW). It offers a wide range of services including the NSW Statewide Severe Burn Injury Service. Concord Hospital: Concord Repatriation General Hospital is a major teaching hospital in Sydney, New South Wales (NSW). It offers a wide range of services including the NSW Statewide Severe Burn Injury Service. The Children’s Hospital at Westmead: This hospital is the largest paediatric centre in NSW, providing specialist care to children from across the state and beyond. It offers a comprehensive range of services including the New South Wales Paediatric Burns Unit. This site will collect paediatric burns data. The Children’s Hospital at Westmead: This hospital is the largest paediatric centre in NSW, providing specialist care to children from across the state and beyond. It offers a comprehensive range of services including the New South Wales Paediatric Burns Unit. This site will collect paediatric burns data. Royal North Shore Hospital: Located in St Leonards, NSW, this major public teaching hospital provides a wide range of services including major trauma, severe burns, spinal cord injuries and neonatal intensive care. Royal North Shore Hospital: Located in St Leonards, NSW, this major public teaching hospital provides a wide range of services including major trauma, severe burns, spinal cord injuries and neonatal intensive care. Queensland Children’s Hospital: This hospital in South Brisbane, Queensland, is the only specialist children’s hospital in Queensland. It offers a full range of paediatric services including emergency care, intensive care and specialist medical and surgical services including the Pegg Leditschke Children’s Burns Centre. This site will collect paediatric burns data. Queensland Children’s Hospital: This hospital in South Brisbane, Queensland, is the only specialist children’s hospital in Queensland. It offers a full range of paediatric services including emergency care, intensive care and specialist medical and surgical services including the Pegg Leditschke Children’s Burns Centre. This site will collect paediatric burns data. Royal Brisbane and Women’s Hospital: Located in Herston, this is Queensland’s largest teaching and research hospital. This hospital contains the Professor Stuart Pegg Adult Burns Centre, the major adult burn referral centre for Queensland, Northern New South Wales, Northern Territory and the Pacific Islands. Royal Brisbane and Women’s Hospital: Located in Herston, this is Queensland’s largest teaching and research hospital. This hospital contains the Professor Stuart Pegg Adult Burns Centre, the major adult burn referral centre for Queensland, Northern New South Wales, Northern Territory and the Pacific Islands. Royal Adelaide Hospital: The largest hospital in South Australia, the Royal Adelaide Hospital provides tertiary and secondary healthcare services. The Adult Burns Service at the Royal Adelaide Hospital provides specialised acute, surgical and reconstructive care to patients with burns. Royal Adelaide Hospital: The largest hospital in South Australia, the Royal Adelaide Hospital provides tertiary and secondary healthcare services. The Adult Burns Service at the Royal Adelaide Hospital provides specialised acute, surgical and reconstructive care to patients with burns. Women’s and Children’s Hospital: This hospital specialises in women’s and children’s health, offering emergency, inpatient and outpatient care in South Australia. The Burns Service at the Women’s and Children’s Hospital provides inpatient and outpatient burn treatment for children from birth to 16 years. This site will collect paediatric burns data. Women’s and Children’s Hospital: This hospital specialises in women’s and children’s health, offering emergency, inpatient and outpatient care in South Australia. The Burns Service at the Women’s and Children’s Hospital provides inpatient and outpatient burn treatment for children from birth to 16 years. This site will collect paediatric burns data. Royal Hobart Hospital: The Royal Hobart Hospital is Tasmania’s largest hospital and a major teaching facility. The Tasmanian Burns Service is coordinated through the Tasmanian Burns Unit based at the Royal Hobart Hospital. The Tasmanian Burns Unit provides acute, surgical, rehabilitation and reconstructive burn treatment for children and adults as well as statewide burn services. Royal Hobart Hospital: The Royal Hobart Hospital is Tasmania’s largest hospital and a major teaching facility. The Tasmanian Burns Service is coordinated through the Tasmanian Burns Unit based at the Royal Hobart Hospital. The Tasmanian Burns Unit provides acute, surgical, rehabilitation and reconstructive burn treatment for children and adults as well as statewide burn services. Alfred Hospital: The Alfred in Melbourne, Victoria is a major tertiary referral hospital known for its emergency and trauma services, ICU and statewide services including the Victorian Adult Burns Service. Alfred Hospital: The Alfred in Melbourne, Victoria is a major tertiary referral hospital known for its emergency and trauma services, ICU and statewide services including the Victorian Adult Burns Service. Royal Children’s Hospital Melbourne: This hospital is a leading paediatric hospital in Melbourne, Victoria, providing comprehensive health services for children and adolescents including the Royal Children’s Hospital Burns Service. This site will collect paediatric burns data. Royal Children’s Hospital Melbourne: This hospital is a leading paediatric hospital in Melbourne, Victoria, providing comprehensive health services for children and adolescents including the Royal Children’s Hospital Burns Service. This site will collect paediatric burns data. Middlemore Hospital: Located in Ōtāhuhu, Auckland, New Zealand, Middlemore Hospital is a leading healthcare facility in South Auckland. The hospital is home to the National Burn Centre, which provides specialised care for patients with burns including acute treatment, surgical interventions and comprehensive rehabilitation. This site will collect both adult and paediatric burns data. Middlemore Hospital: Located in Ōtāhuhu, Auckland, New Zealand, Middlemore Hospital is a leading healthcare facility in South Auckland. The hospital is home to the National Burn Centre, which provides specialised care for patients with burns including acute treatment, surgical interventions and comprehensive rehabilitation. This site will collect both adult and paediatric burns data. Eligibility criteria The following criteria will be used to determine whether cases are eligible for inclusion in this study. As a retrospective cohort study, participants will not be formally recruited. Inclusion criteria All adult (patients >16 years) and paediatric patients with burns presenting to one of the participating hospitals (see above) with at least superficial dermal thermal burn injuries (eg, scald, contact, flame, flash). All adult (patients >16 years) and paediatric patients with burns presenting to one of the participating hospitals (see above) with at least superficial dermal thermal burn injuries (eg, scald, contact, flame, flash). Presentation to a participating ED within 12 hours of sustaining their burn. Presentation to a participating ED within 12 hours of sustaining their burn. Burns of all sizes and of all anatomical regions are eligible. Burns of all sizes and of all anatomical regions are eligible. Patients who present to a participating ED between 1 January 2024 and 31 December 2024. Patients who present to a participating ED between 1 January 2024 and 31 December 2024. Exclusion criteria Burn injuries classified as superficial epidermal or epidermal in depth (ie, erythema only). Burn injuries classified as superficial epidermal or epidermal in depth (ie, erythema only). Patients with non-thermal burn injuries (eg, electrical, chemical and friction injuries) and those presenting to hospital greater than 12 hours post-burn. Patients with non-thermal burn injuries (eg, electrical, chemical and friction injuries) and those presenting to hospital greater than 12 hours post-burn. Patient presentations outside of the specified 12-month study period. Patient presentations outside of the specified 12-month study period. While there are no exclusion criteria for time to initial temperature measurement in the ED, patients with acute burn without temperature measures taken in the ED will be excluded from the investigation. While there are no exclusion criteria for time to initial temperature measurement in the ED, patients with acute burn without temperature measures taken in the ED will be excluded from the investigation. Outcome measures Primary outcome The primary outcome measure for this retrospective cohort review is the incidence of hypothermia in patients with burns following their arrival at the ED. For the purposes of this investigation, hypothermia will be defined as a recorded temperature of less than 36°C. This temperature threshold has been used in previous studies assessing hypothermia in patients with burns.13 16 Secondary outcomes In addition to the incidence of hypothermia, the following secondary outcomes will be investigated: Influence of burn first aid cooling: comparison of hypothermia incidence between patients with burns who received running water first aid and those who did not—application of CRW, regardless of duration, will be examined. Influence of burn first aid cooling: comparison of hypothermia incidence between patients with burns who received running water first aid and those who did not—application of CRW, regardless of duration, will be examined. Risk factors influencing the development of hypothermia: identification and analysis of demographic, clinical, environmental and situational factors associated with the development of hypothermia in patients with acute burn. Risk factors influencing the development of hypothermia: identification and analysis of demographic, clinical, environmental and situational factors associated with the development of hypothermia in patients with acute burn. Impact of hypothermia on clinical patient outcomes, including:Average duration of hospital stay (days).Rate of ICU admissions and length of ICU stay (days).Mortality rate. Impact of hypothermia on clinical patient outcomes, including: Average duration of hospital stay (days). Average duration of hospital stay (days). Rate of ICU admissions and length of ICU stay (days). Rate of ICU admissions and length of ICU stay (days). Mortality rate. Mortality rate. Incidence of hypothermia in patients with burns in non-ED settings: comparison of hypothermia incidences in prehospital settings, retrieval settings, ICUs (for admitted patients) and operating theatres (for those requiring immediate surgical interventions). Incidence of hypothermia in patients with burns in non-ED settings: comparison of hypothermia incidences in prehospital settings, retrieval settings, ICUs (for admitted patients) and operating theatres (for those requiring immediate surgical interventions). Data collection Data sources Data sources will include ED databases such as the Emergency Department Information System (EDIS), Cerner FirstNet, TrackED and Epic, as well as local EMRs. Additional potential data sources will include locally stored burns databases. Variables and data sources are presented in online supplemental file 1. Data collection procedures Access to ED data will be obtained from all participating sites following required ethics approvals. Patients presenting with complaints of burns or burn-based International Classification of Diseases version 10, Australian Modification (ICD-10-AM) codes will be identified from site ED databases (eg, EDIS, Cerner FirstNet, Epic). Health service data (ie, hospital admission, patient management and discharge disposition) will be collected from either EMR or local burn databases to obtain missing values for variables. Variables ED data will include time from initial injury to arrival in the ED; admission to hospital; length of stay in the ED; temperature on arrival to the ED; hourly temperature measures; mode of arrival to the ED; first aid received; location of first aid provision (eg, at home with ambulance service, following arrival to the ED). As burn first aid cooling can sometimes be given in stages (eg, 10 min at home, 10 min in the ED), prehospital first aid durations, referral centre/other first aid cooling and first aid cooling durations given in the ED will also be captured. Health services data (inpatient clinical outcome data) will include inpatient length of stay; ICU admission (paediatric and adult); complications; operating theatre requirements and ICD procedure codes. Data sources include EMR, supplemented with hospital-based local burns registry data. Please see online supplemental file 1 for the complete list of variables required for this research. For the purposes of this investigation, hypothermia will be consistently defined across all participating sites as a recorded body temperature below 36°C, in accordance with the threshold specified in the primary outcome measure. The mode of temperature assessment (eg, oral, tympanic, rectal, axillary, temporal artery) will be extracted and reported where this information is available within the EMR. Data analysis Sample size As a retrospective chart review, a direct sample size calculation will not be completed. This study aims to capture all burn patient presentations to participating EDs. Based on hospital reports of annual acute burn presentations and referrals, this study is likely to include an estimated 5000 patients. It is anticipated that this sample will allow the primary outcome to be explored. Statistical analysis Descriptive statistics will be calculated for each of the identified variables. Data will be described as frequencies, percentages and medians (quartiles) for non-parametric data. Crude incidence rates (IRs) and 95% CIs will be calculated to determine rates of hypothermia in the sample. Multivariable regression models will be used to assess the association between potential causal risk factors (eg, CRW application, TBSA%, depth, age, rurality, mode of presentation to ED and other comorbidities) and hypothermia and will also be used in the development of the prognostic model. Model-building strategies will be matched to whether the study question relates to causal inference or prognostic model development.17 For objectives 2, 3, 4, 5 and 7, potential confounders relating to each causal effect estimate of interest will be identified using separate directed acyclic graphs (DAGs), which will be constructed prior to data analysis following a thorough literature review and after consulting subject-matter experts.18 Sensitivity analyses with alternative adjustment sets will be used to test the robustness of results to misspecification of the DAG-implied adjustment sets.19 E values will be used to assess sensitivity to unmeasured confounders.20 Sensitivity analyses will also be conducted to assess the impact of alternative temperature thresholds for hypothermia (eg, 35°C). For objective 6, existing predictors of hypothermia will be identified through a literature review and nominated for inclusion in the model. Following this, LASSO will be used for variable selection and shrinkage.21 Internal-external cross-validation will be performed to assess the site-specific performance of the model, and then random-effects meta-analysis will be used to assess pooled performance.22 Performance of the prognostic model will be assessed using calibration-in-the-large, calibration slope and C statistic.21 Data will be analysed using R.23 Safety considerations As this research is a retrospective review of patient EMRs, with no active participant recruitment or intervention allocation, there are few safety considerations. The only risk to participants is a potential confidentiality and privacy breach. Governance and other relevant approvals (ie, Public Health Act approval) will be obtained from hospital and health service data custodians to access patient medical records. Patient data will be deidentified and confidentiality of patient information will be preserved at all times. Access to patient data will only be available to necessary members of the research team and stored in accordance with data storage best practices. No identifiable or potentially re-identifiable patient information will be stored or published in manuscripts arising from this research. Data management and storage Patient data (including data from the ED, hospital EMR or local burn registry) will be collected from multiple sites and stored securely in a password-protected Microsoft Excel file and/or REDCap database. While the data collected will be from multiple sites across Australia, the database will be located on a Queensland Health server. Permission to extract identifiable data will be obtained from the relevant local ethics committees and/or statewide health services. Data access will be limited to those researchers directly involved in the study. Data analysis will be performed by the researchers with consultation from a statistician. Patient and public involvement There is no direct patient and public involvement in the design, conduct, reporting or dissemination of this research. Patient EMRs will be accessed to collect relevant information; however no contact will be made between researchers and patients. Ethics and dissemination National mutually agreeable human research ethics approval has been granted by the Children’s Health Queensland Hospital and Health Service Human Research Ethics Committee (Australia) (HREC Reference number: HREC/25/QCHQ/114285) as well as through the Health and Disability Ethics Committees (New Zealand), (HDEC; Ref No: 2026 EXP 23892). This study will be performed in accordance with the ethical principles of the Declaration of Helsinki, ICH GCP for Guidance on Good Clinical Practice and National Health and Medical Research Council National Statement on Ethical Conduct in Research Involving Humans.2426 No part of the study will commence prior to all HREC and governance approvals. Any changes to this protocol will be submitted to the relevant ethics committees as an amendment for approval before being actioned. Discussion This study aims to address a critical evidence gap in the delivery of first aid following burn injury, namely, the relationship between the application of 20CRW and the development of hypothermia. Caution surrounding hypothermia development and causing hypothermia is a known barrier adherence to best practice 20CRW delivery.11 This study is the first multicentre retrospective cohort study to examine hypothermia in patients with burns presenting to EDs across Australia and New Zealand. This provides a unique opportunity to encompass various geographic and climate factors, which may assist in enhancing the generalisability of 20CRW first aid guidelines. With an estimated sample size of 5000 burn presentations, the study allows for robust exploratory subgroup analyses and detailed statistical analysis. As a retrospective cohort review, this study is dependent on the quality of data collected—missing and incomplete data may impact the quality of the study. Furthermore, variability across sites based on local policies and practices may impact how and when the body temperature is measured, which may introduce potential measurement bias.12 As temperature measurement method is not consistently recorded within EMR systems, the authors are unable to fully standardise this variable across sites. The mode of assessment will be reported where available, and this constraint will be explicitly acknowledged as a limitation in the resulting publication. In addition, real-world provision of 20CRW has variations—such as duration, method and water temperature—and depends on resource availability, awareness and the ability to provide first aid, which may introduce potential confounders that impact the analysis.14 However, the sample size of the study enhances the study’s capacity to address these variations and allow for stable estimates across subgroups. Previous large-scale observational studies have demonstrated that large datasets can reduce the influence of site-level and procedural variability.27 Supplementary material