Introduction The rapid evolution of minimally invasive surgery has had a positive impact on patient outcomes; however, it is reported to be associated with work-related musculoskeletal symptoms (WMS) in surgeons. Currently there is no objective measure to monitor the physical and psychological impact of performing a live surgical procedure on the surgeon.
Methods and analysis A single-arm observational study with the aim of developing a validated assessment tool to quantify the impact of surgery (open/laparoscopic/robotic-assisted) on the surgeon. Development and validation cohorts of major surgical cases of varying levels of complexity performed by consultant gynaecological and colorectal surgeons will be recruited. Recruited surgeons wear three Xsens DOT monitors (muscle activity) and an Actiheart monitor (heart rate). Salivary cortisol levels will be taken and questionnaires (WMS and State-Trait Anxiety Inventory) completed by the participants preoperatively and postoperatively. All the measures will be incorporated to produce a single score that will be called the 'S-IMPACT' score.
Ethics and dissemination Ethical approval for this study has been granted by the East Midlands Leicester Central Research Ethics Committee REC ref 21/EM/0174. Results will be disseminated to the academic community through conference presentations and peer-reviewed journal publications. The S-IMPACT score developed within this study will be taken forward for use in definitive multicentre prospective randomised control trials.
- OCCUPATIONAL & INDUSTRIAL MEDICINE
- Minimally invasive surgery
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Strengths and limitations of this study
This study measures aim to capture both the psychological and physical impact of performing surgery on the surgeon.
Data will be collected during live surgical cases.
Process evaluation will be used to enable development of the 'S-IMPACT' score which can be used in future surgical research studies.
Case recruitment and data collection may be impacted by theatre scheduling and disruptions in theatre workflow.
The work-related musculoskeletal symptom questionnaire is not able to quantify musculoskeletal symptoms.
Minimally invasive surgery (MIS), as compared with open surgery, for the management of many surgical conditions is associated with improved patient morbidity and mortality outcomes.1 2 However, it also appears that MIS is associated with significantly greater risk of work-related musculoskeletal symptoms (WMS) in surgeons,3 4 with 43–100% of surgeons from a wide range of specialties reporting symptoms after performing MIS when questioned.5
In particular, neck, back, arm/shoulder and leg pain5 are reported, as well as an increased incidence of degenerative cervical (17%) and lumbar (19%) spine conditions, damage to the rotator cuff (18%) and carpal tunnel syndrome (9%).6 The primary reasons for this association are reported to be prolonged static or uncomfortable postures while navigating anatomical areas that are not aligned with optimal ergonomic positions and the use of inflexible equipment that do not account for variable surgeon anthropometry (smaller glove size, shorter elbow height).7 Other factors reported to increase WMS risk, in particular with conventional laparoscopy (LAP), are high volume workload and high patient body mass index (BMI).8 A review of ergonomics in surgery identified that ‘table and monitor position, long-shafted instruments, and poor instrument handle design’ contribute extensively in the higher WMS reported in conventional LAP.9 Also, female surgeons with smaller glove sizes report significantly higher rates of shoulder/neck/upper back discomfort compared with male surgeons.10 Objective measures have shown that the range and magnitude of movements required to perform exercises laparoscopically increased with increasing BMI of the model2 and that these effects are greater in surgeons with less surgical experience.11 Using the LUBA ergonomic framework, it was shown that performing laparoscopic tasks on high BMI models increased the number and duration of non-neutral postures, which in turn could increase the risk of future WMS.12 The evolution and validation of miniaturised wearable motion sensor systems, such as Inertial Measurement Units (IMUs), have paved the way to measure surgeon ergonomics in a live setting.13 The portable nature, data storage capacity and available pre-existing validated software for data processing of these systems allow for objective analysis of surgeon kinematics without affecting theatre workflow and technical performance.13
The impact of surgery on the surgeon however, is not confined to musculoskeletal effects but also incorporates human factor issues, such as cognitive load and psychological stress.14 Surgeon stress can impair surgical performance15 and can impact on the theatre environment dynamics, increasing the risk of surgical complications or loss of attention. Laparoscopic procedures in particular, are reported to be stress-inducing.16 Measuring surgeon stress can be challenging since it requires incorporation of subjective and objective perception of a situation and including a self-reported aspect is important when assessing cognitive workload, as is the need for real-time assessment.17 The Imperial Stress Assessment Tool (ISAT) was developed to measure surgeon stress and combines continuous heart rate monitoring with salivary cortisol and a subjective questionnaire.18 It has shown that concurrent monitoring using a biomarker, as well as a subjective questionnaire, does not disrupt theatre workflow environment and is a feasible and valid option. The need for multiple indicators was highlighted in the study by Arora et al, who reported that subjective stress was identified by heart rate monitoring or salivary cortisol alone in 84% and 80% of procedures, respectively, but only in 70% by both.18
Since its introduction, robotic-assisted surgery (RA) has been reported to have greater benefits for surgeons’ musculoskeletal health as compared with conventional LAP,19–21 including decreased workload and upper body muscle motion.22–26 However, RA is not without ergonomics issues, typically from fixed position at the console,21 27 which instead is reported to have a greater impact on shoulder and arm movement.28 RA does appear to be associated with a lower rate of WMS compared with LAP and open surgery when operating on patients with obesity,8 29 and to be less stressful for the surgeon as compared to LAP.30 Although, the effect of increasing patient complexity and theatre organisational factors are not known.
Designing and conducting surgical trials can be very challenging due to the multiple sources of potential bias. Trial outcomes are understandably focused on patient outcomes and very few include an outcome measure assessing the impact of the intervention on the surgeon. A potential reason for this is the lack of a validated measurement of ‘impact’ that can objectively record the physical/psychological effect on the surgeon. ISSUE aims to develop the ‘S-IMPACT’ measure, which would enable surgeon health to be considered along with patient outcomes when evaluating interventions but could also be used to assess surgeon workload with a view to optimisation and reducing the risk of future injuries.
The overall aim of this study is to develop and validate a multifaceted assessment tool (called the S-IMPACT score) that objectively captures the real-time physical and psychological impact performing surgery has on the surgeon.
The primary objective is the proportion of surgical procedures (including open, robotic and laparoscopic) where all five measurement variables included in the proposed impact assessment tool are captured. Secondary objectives include answering feasibility questions of measuring individual aspects of the proposed tool, including data quality from heart rate and movement sensor monitors and questionnaire completeness (box 1).
Primary outcome measure
The proportion of surgical procedures where all five variables of the assessment tool are captured (continuous heart rate ((HR)) monitoring; salivary cortisol (taken preoperatively, intraoperatively and postoperatively); continuous movement sensor monitoring; State-Trait Anxiety Inventory (STAI) questionnaire and work-related musculoskeletal symptoms (WMS) questionnaire), taken preoperatively and postoperatively.
Secondary outcome measures
Development of a single value of ‘impact’ (high/intermediate/low) per procedure calculated by combining the five recorded variables.
Validation of the calculated ‘impact’ values by comparison of the surgeon’s subjective assessment of case complexity with objective measures.
Analysis of the impact on surgeons supervising surgical trainees, measured using HR monitoring and salivary cortisol levels intraoperatively.
Willingness of recruitment of patients and surgeons; sample size calculations for a future multicentre study comparing the impact of the route of surgery on the surgeon.
Primary/secondary outcomes for a future definitive multicentre study comparing the impact of the route of surgery on the surgeon.
Optimisation and validation of patient data recording measures to capture anaesthetic/operative/clinical events and outcomes.
Safety of measurement tools and recording devices in the theatre environment.
Optimisation of surgeon continuous HR monitoring during the procedure, quality and completeness of data collection and analysis.
Optimisation of positioning and recording of muscle movement/activity sensors and streamlining of data-analysis.
Optimisation and validation of WMS and STAI questionnaires and data analysis.
Optimisation of salivary cortisol collection preoperatively, intraoperatively and postoperatively and utility of results assessed.
Additional objectives include assessment of participant recruitment, both patients and surgeons. Potential primary outcome measures for a future randomised controlled trial comparing the impact of open/laparoscopic/RA surgery on the surgeon will also be explored. The results of the separate measures will be combined to give a single value of impact (high/intermediate/low) of a procedure on the surgeon, termed the S-IMPACT measure.
Methods and analysis
This is a single arm observational study designed to assess the feasibility of developing and using a multifaceted assessment tool for measuring the impact of surgery on the surgeon, termed S-IMPACT score. The study was due to open in November 2021 and be open for 2 years until November 2023. Recruitment period began in January 2022 and will end in February 2023.
The study will be conducted through the University Hospitals of Leicester (UHL) NHS Trust, in collaboration with University of Leicester and Loughborough University.
The S-IMPACT score
In order to capture both the physical and psychological impact of performing a major surgical procedure on the surgeon, five separate measures will be recorded: continuous heart rate monitoring; salivary cortisol; continuous motion tracking; State-Trait Anxiety Inventory (STAI) questionnaire; WMS questionnaire. The Actiheart 5 and Xsens DOT monitors have been chosen because of their size and weight, so as to be as unobtrusive as possible for the surgeon. Assessment of the potential impact of the monitoring equipment on the theatre environment has been conducted by the UHL Medical Physics department and the study procedures accommodate their safety recommendations. Simulation trials will be conducted at the time of baseline testing to ensure that they do not impact on the surgical ability of the surgeon.
Continuous heart rate monitoring
The Actiheart 5 monitor (CamNtech) measures 39.7×30.2×9.25 mm, weighs 10.5 g and has a long lasting battery life.31 It has been validated in several studies as an effective tool in measuring heart rate variability as a marker for stress and a psychosocial symptom marker in bio behavioural research.32 33 It will be placed at the centre of the chest, V1 ECG lead position (figure 1), enabling recording of heart rate variability. Heart rate and heart rate variability will be measured, in addition to minimum/maximum heart rate.
Salivary cortisol levels will be taken via a buccal swab immediately prior to starting the surgical procedure, after removal of the surgical specimen and after completion of the procedure. The swab testing and analysis will be performed by Medichecks (Nottingham) and company storage/transport procedures will be followed. The use of cortisol as a measure of stress has been validated to monitor stress in surgeons in previous studies.18 30
Continuous movement sensor monitoring
Three Xsens DOT sensors will be used to continuously monitor the kinematics of the surgeon during the surgical procedure. The sensors measure 36.3×30.35×10.8 mm, weigh 11.2 g and include a 3-axis accelerometer, 3-axis gyroscope and 3-axis magnetometer. The sensors have undergone validation, especially with upper body kinematics. Each monitor will be placed inside its own Faraday pouch (Todoxi) and attached to the surgeon using elasticated straps to the upper arms and upper back (figure 1). The muscle compartments to be monitored by the sensors have been selected based on the high rates of injuries noticed in these areas. The relative changes in the magnitude of movements during the procedure will be calculated.
The STAI questionnaire has been selected since this has been validated previously in the surgical setting.18 Literature reviews on available inventories have shown that the STAI questionnaire is widely used and is effective at the capture of the general state of anxiety as well as the ‘proneness to anxiety’.34 Subsequently a six-item version has also been developed and validated and this can be administered in more acute or surgical settings to measure relative changes in mental state.18 35 36 Furthermore, the STAI inventory has been used in conjunction with heart rate variability to prove the latter’s efficacy in demonstrating surgeon/operator stress as it showed a positive correlation between an increased sympathetic tone and surgeon stress levels.37
The Nordic WMS questionnaire38 will be used as a baseline and pre-questionnaires and post-questionnaires to survey the areas of concern and to capture changes in pain or discomfort post procedure. These aim to establish any baseline injuries as well as identify exacerbations in known areas of concern and development of new areas of discomfort.
The recorded data from the Actiheart and Xsens monitors will be downloaded using the manufacturers’ software to a tablet (Samsung Galaxy) before being transferred to a password protected University of Leicester laptop for analysis using the MATLAB software and Microsoft Excel. Analysis of the results of the five items will be performed individually, using previously validated scoring systems where possible, to give a score within the low, intermediate or high range. The five individual scores will then be combined to give an overall classification for the procedure, either low, intermediate or high impact.
Participants and recruitment
This study will recruit two cohorts of participants: surgeons and patients. Consultant colorectal surgeons and gynaecologists based at the UHL will be approached and those who register interest will be invited to participate in the study. The inclusion criteria for surgeons are: ability to consent, willingness to be monitored in cases where consent from the patient has also been obtained. Patients over 18 years old who are due to undergo a major intra-abdominal surgical procedure expected to last less than 2 hours duration under the care of a recruited surgeon will be invited to participate. The exclusion criteria include the inverse of the inclusion criteria and agreement to proceed by the supervising consultant on the day of the surgery. A minimum number of procedures per surgeon has not been set for the feasibility stage of the study as the primary focus is to identify and optimise the data collection in a live surgical setting with minimal disruption to theatre workflow.
Following recruitment, surgeons will attend for an initial session with the research team in order to perform baseline measures and to familiarise themselves with the monitoring equipment that will be worn during the planned surgery, to ensure no impact on their technical abilities. The session will involve: collection of their anthropometric data (including height, arm span, elbow height, age, sex) and years of surgical experience; baseline Nordic WMS and STAI questionnaires; completion of a series of standardised exercises on a laparoscopic simulator (LapAR, Innovus) and robotic simulator (Intuitive Surgical) with/without the Actiheart 5 monitor and the Xsens DOTs within Faraday pouches; and a semi-structured interview exploring their previous experience with MIS and WMS. The aim of the exercises on the laparoscopic/robotic simulators is to ensure that surgeons become familiar with wearing the monitoring equipment prior to undertaking a live surgical case and to confirm objectively that their technical surgical performance is not impaired by wearing the sensors. The LapAR has been chosen as the simulator on which to conduct this testing due to its performance tracking ability and haptics, giving high-fidelity simulation and performance metrics. The exercises will include key tasks from the LapPass (hoops/pegs) and specialty-specific procedures: either salpingectomy or appendicetomy (cutting skills) and vaginal vault closure (suturing). The ‘peg board’, ‘dots and needles’ and ‘threading the ring’ exercises were selected for the assessment on the DaVinci Si robotic simulator.
Patients will be identified by the researcher, confirmed as appropriate with the surgeon and approached by a member of the research team at their surgical pre-assessment appointment, typically 1–2 weeks prior to admission for surgery. This is to ensure that data collection does not affect clinical care provision or the theatre workflow or environment, led by the surgeon. The inclusion criteria for patient recruitment are those aged above 18 years, with capacity to consent and undergoing an intra-abdominal procedure that is expected to last less than 2 hours. There will be a minimum of 24 hours between approach and recruitment of patient participants. The researcher will consent the patient to study preoperatively and all due care will be taken to ensure that the research processes do not interrupt clinical care on the day of surgery.
Information on the patient undergoing surgery will be collected, including BMI, waist-to-hip ratio, medical comorbidities and previous surgery. The complexity of the surgical case (low/intermediate/high) will be categorised by the researcher taking into account factors including BMI, previous surgery and surgical procedure.
Preoperatively the surgeon will have their sensors placed and their preoperative data collected, including WMS/STAI questionnaires and salivary cortisol swab (figure 2). After the removal of the surgical specimen (bowel resection, uterus for hysterectomy), provided there are no urgent complications to attend to and patient safety is confirmed, the operating surgeon will have a salivary cortisol swab taken. Following completion of the procedure, the surgeon will repeat the WMS/STAI questionnaires and have another salivary cortisol swab taken. Preoperatively and postoperatively each surgeon will be asked to provide a subjective assessment of the overall complexity (low, intermediate or high) of the case. This is to ensure that the influence of other variables such as theatre team/organisation, assistant experience or equipment are accounted for within this response, and will be explored further in the qualitative aspect of this study. Surgeons will be recruited undertaking cases on their regular operating lists, ensuring familiarisation with the theatre set-up, equipment and surgical team. At the end of the case, the surgeon will participate in an interview focusing on exacerbation of any WMS symptoms and feasibility of data acquisition for this study on that particular operating list. The data collection will be led by a single researcher to avoid deviations from protocol.
The primary feasibility outcome is the proportion of surgical procedures with all five variables completely captured. An expected level of 90% was proposed and sample size of 65 surgical procedures produces a two-sided 95% CI with a width equal to 0.160 when the sample proportion is 0.900 (95% CI: 0.80 to 0.96).39–41 The data collected from these cases will be combined to determine the parameters for a tool that can be used to classify cases into ‘low’, ‘intermediate’ and ‘high’ impact (S-IMPACT) depending on the physical/psychological impact on the surgeon. A temporal validation to determine the ability of S-IMPACT to differentiate between low/intermediate/high complexity cases will be carried out with a sample size of 88 cases. This produces a two-sided 95% CI with a width equal to 0.20 for sensitivity (of identifying high impact cases), when the sample sensitivity is 0.90 and the prevalence (of high impact) is 0.50. The validation data will not be viewed until the risk score is finalised and a target recruitment for the validation cohort of 100 cases will be aimed for, to allow for case dropout.
Evaluation-related feasibility measures
Surgeon and patient recruitment will be monitored by comparing the number of potential and recruited participants. The number of cases recruited by individual surgeons and the completeness of the study measures per case will be recorded, including the proportion of surgical procedures with all five variables completely captured. Data from each of the measurements will be analysed to calculate the distribution of data and a threshold for ‘impact’.
Data analysis plan
The percentage of cases where all five variables are successfully collected will be calculated, along with recruitment rates and case mix data. Each case will be classified as low/intermediate/high in terms of complexity by the surgeon (subjective) and the researcher (objective). The Cohen’s weighted kappa test will be used to measure the level of agreement between the subjective and objective classification of case complexity. The sensitivity and specificity of each of the five variables will be analysed to identify their suitability as objective measures of surgical stress in a live surgical setting. Content validity will be established by analysing the trends shown by each variable in low/intermediate/high complexity cases. Construct validity will be determined by how well the S-IMPACT score can detect low/intermediate/high complex cases. Due to the intraexperience subsamples, normal distributions cannot be assumed therefore non-parametric analysis will be performed. When grouped by BMI level and surgical modality, a Friedman’s two-way analysis of variance will be applied. When grouped by independent variables (by experience level), a Kruskal-Wallis test will be applied.
Challenges experienced by surgeons participating in the study and undertaking the study measures will be explored in the semi-structured interviews. The interviews will collect data on surgeon’s views and situation factors that may have affected recruitment and the recording of data. There will be an initial interview at recruitment to the study to explore previous experience/views of WMS with different surgical routes. Interviews will be conducted at regular intervals during the recruitment period with each surgeon recruited to gain immediate feedback on symptoms, theatre set-up and the study measures. A further interview will be conducted with the surgeons towards the end of the recruitment period in order to determine the duration of any symptoms and further feedback on the study. Interview data will be analysed using Template Analysis42 and a deductive coding framework to inform refinement of the components of the S-IMPACT measure. A researcher who was not involved in the development and recruitment of ISSUE will code a number of transcripts in order to ensure independent assessment. The analysis will focus on three main areas: participant recruitment; measurement components; and adjustments needed for a future study using the S-IMPACT score.
Patient and public involvement
Patient and public involvement consultation was undertaken with consultant gynaecological surgeons and used to inform the study design.
This study aims to create and validate an impact score that can objectively measure the physical and psychological impact of surgery on the surgeon, without disrupting the theatre workflow environment. There is a career-long need for surgeons to improve their technical skills in order to improve patient outcomes; however, the impact of performing surgery on their long-term health is often not considered. The measures that we have included in ISSUE are aimed at measuring both magnitude and changes in physical and psychological workload in the live surgical setting, in order to produce an objective assessment.
Changes in heart rate are predominantly associated with physical workload,43 44 whereas psychological workload during surgery is better described by heart rate variability, as it is a reflection of the autonomic system adapting to situations.45 Heart rate variability has been applied as a measure of psychological workload in prospective studies exploring the impact of cognitive workload between open and laparoscopic surgery, where increased cognitive strain was associated with performing laparoscopic surgery.46 Similarly, when assessing the level of anxiety experienced by a cardiothoracic surgeon performing surgery or supervising a resident, the quantitative measure provided by the heart rate variability reflected the subjective assessment of the levels of anxiety reported by the surgeon at various points in the procedure.47 This suggests that heart rate variability is superior to heart rate when used as a marker for cognitive workload.
Cortisol has been identified as an accurate marker of acute stress and has been incorporated into the ISAT.18 Furthermore, salivary (tissue) cortisol has a longer half-life than plasma cortisol48 making it a viable marker in a live surgical setting, since it is not impacted by a delay in obtaining the sample at the time of removal of specimen, thereby allowing a greater window for sample collection.
Stress is not a linear concept but rather multifactorial because it includes, but is not limited to, operating time, skill level of assistant, complexity of case and disruptions to theatre workflow. Additionally, individual surgeons may be subject to additional stressors from their personal and work environments that are not related to the surgical procedure being performed. The Surgery Task Load Index (SURG-TLX) was developed with this in mind to measure cognitive workload and subjectively evaluate the ‘mental, physical and temporal demands’ as well as ‘task complexity, situational stress and distractions’.49 The S-IMPACT score will differ from the SURG-TLX in that it will include objective measures of physical impact as well as cognitive workload, with the rationale that these are not mutually exclusive factors and when combined may better predict future WMS risk. Prolonged static postures and heavy loading on the muscular compartments used during MIS may cause significant physiological stress.11 Patient factors, such as BMI, can have a profound impact on surgeon upper body kinematics and are important parameters to consider when monitoring physical stress from performing surgery.12 Studies objectively measuring upper body kinematics using IMUs (similar in function to the Xsens used in our study design) have shown that the mean jerk, angular speed and cumulative displacement magnitudes are positively correlated to an increase in the BMI of abdominal models and can impact the head, torso and both upper arms, regardless of the level of experience of the surgeon.11 An increase in the mean jerk suggests that the force applied and the frequency of movement are higher, suggesting that the upper body is working harder than when performing laparoscopic tasks on a model with a BMI within the normal range.11 Also, a higher angular speed reflects the need to constantly correct upper arm manipulation and a larger cumulative displacement in order to complete laparoscopic tasks on simulated high-BMI, which results in a greater distance being travelled by the equipment, suggesting a greater surgeon upper body workload.11 While electromyography (EMG) sensors have been established as effective monitors of muscle workload, they do not monitor movement, which is our focus. This is because muscular injuries are often sustained from holding postures aberrant to the neutral stance and IMUs are better placed to identify and correlate the physical impact of surgery on the surgeon.
Future applications of S-IMPACT could include assessment of surgeons and trainees in order to optimise their physical/psychological functioning while performing surgery, enabling additional support and ergonomic training to be targeted to areas of need. The identification of ‘harmful’ behaviours or postures could result in interventions and changes in working practices, with the aim of reducing future WMS. In this manner, an optimum ergonomic working environment can be developed to produce an enhanced/higher level performance by the operator as well as create a sustainable workforce within the healthcare sector.
Ethics and dissemination
Ethical approval has been granted by the HRA and Health and Care Research Wales, and the East Midlands Leicester Central Research Ethnics Committee for this study REC Ref 21/EM/0174. Publication and dissemination plan: Any data arising from this study will be published and presented in an open-access peer-review journal. The manuscript will be deposited with the University of Leicester, according to the University of Leicester’s policies (see http://www2.le.ac.uk/library/downloads/open-access/open-access-policy) and data sharing policies.
Individual participant data sharing statement: The data sets generated during and/or analysed during the current study are not expected to be made available as there is patient data, as well as personal information relevant to the participants, including their anthropometry.This data will be stored safely in an anonymised manner on the research drive of the University of Leicester. Direct access will be granted to authorised representatives from the Sponsor, host institution and the regulatory authorities to permit trial-related monitoring, audits and inspections.
Patient consent for publication
Contributors Study design: ELM, BS, MZ and AA. Research interviews: AA. Data analysis: ELM, AA, SB and MZ. Data interpretation: ELM, AA, MZ, SB and BS. ELM and AA wrote the first draft of the manuscript and all authors edited the final version. The funder does not have any roles or responsibilities in this study.
Funding This study is funded by an Intuitive Surgical Research grant RM60G0742.
Competing interests ELM has served on advisory boards for Inivata and GlaxoSmithKline; received speaker fees from GlaxoSmithKline; has received research grants from Hope Against Cancer, British Gynaecological Cancer Society and the East Midlands Clinical Research Network for unrelated work.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.