Article Text
Abstract
Introduction In children, respiratory distress due to upper airway obstruction (UAO) is a common complication of extubation. The quantitative cuff-leak test (qtCLT) is a simple, rapid and non-invasive test that has not been extensively studied in children. The objective of the ongoing study whose protocol is reported here is to investigate how well the qtCLT predicts UAO-related postextubation respiratory distress in paediatric intensive care unit (PICU) patients.
Methods and analysis Air Leak Test in the Paediatric Intensive Care Unit is a multicentre, prospective, observational study that will recruit 900 patients who are aged 2 days post-term to 17 years and ventilated through a cuffed endotracheal tube for at least 24 hours in any of 19 French PICUs. Within an hour of planned extubation, the qtCLT will be performed as a sequence of six measurements of the tidal volume with the cuff inflated then deflated. The primary outcome is the occurrence within 48 hours after extubation of severe UAO defined as combining a requirement for intravenous corticosteroid therapy and/or ventilator support by high-flow nasal cannula and/or by non-invasive ventilation or repeat invasive mechanical ventilation with a Westley score ≥4 with at least one point for stridor at each initiation. The results of the study are expected to identify risk factors for UAO-related postextubation respiratory distress and extubation failure, thereby identifying patient subgroups most likely to require preventive interventions. It will also determine whether qtCLT appears to be a reliable method to predict an increased risk for postextubation adverse events as severe UAO.
Ethics and dissemination The study was approved by the Robert Debré University Hospital institutional review board (IRB) on September 2021 (approval #2021578). The report of Robert Debré University Hospital IRB is valid for all sites, given the nature of the study with respect to the French law. The results will be submitted for publication in a peer-reviewed journal.
Trial registration number NCT05328206.
- paediatric intensive & critical care
- respiratory therapy
- hospitals, public
- paediatric intensive & critical care
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
- paediatric intensive & critical care
- respiratory therapy
- hospitals, public
- paediatric intensive & critical care
STRENGTHS AND LIMITATIONS OF THIS STUDY
This is the largest multicentre prospective observational study to date investigating the ability of the quantitative cuff-leak test to predict the risk of postextubation upper-airway obstruction responsible for respiratory distress in critically ill children extubated after cuffed ventilation.
The trial is being conducted in 19 paediatric intensive care units throughout France and is therefore expected to produce highly generalisable results.
Using a primary outcome that combines a therapeutic measure with a clinical score reduces the risk of bias caused by interobserver variability.
The management of postextubation respiratory distress is not standardised and follows usual protocols in each participating centre.
The use of pre-extubation corticosteroid therapy, which may affect cuff-leak test results and the risk of postextubation upper airway obstruction, is not standardised and follows usual protocols in each participating centre.
Introduction
Invasive mechanical ventilation (IMV) is very often required in patients admitted to the paediatric intensive care unit (PICU). Respiratory distress, the main reason for PICU admission, requires IMV in 30%–40% of cases.1 2 Moreover, IMV may be needed after surgery or for non-respiratory life-threatening conditions such as neurological, haemodynamic or metabolic failure.1–3 In a marked departure from historical practice, cuffed endotracheal tubes are now recommended in children.4 5 Complications that may develop during IMV or after extubation include ventilator-associated pneumonia, airway injury, laryngeal ischaemia and postextubation respiratory distress (PERD).6–8
The main cause of PERD in paediatric patients is upper airway obstruction (UAO) due to laryngeal oedema, which results in a characteristic high-pitched sound known as stridor. UAO-related PERD (UAO-PERD) occurs after 5%–30% of extubations in children. The severe form may require reintubation in 2%–5% of cases, and is also associated with the need for non-invasive ventilation (NIV), prolonged PICU stay and tracheal stenosis.9–12 Moreover, before extubation, each patient should be assessed for the risk of UAO-PERD. One risk marker is the result of the cuff-leak test (CLT). This simple, rapid and non-invasive test consists in measuring the air leak after deflation of the endotracheal tube cuff. Leakage is expected to occur through the space left free around the tube. In the event of oedema, however, the tube is in contact with the airway after cuff deflation and leakage does not occur. Thus, the greater the leak, the lower the risk of UAO-PERD. The latest PALISI network guidelines recommend performing the CLT before extubation in children, despite underlying evidence of only very low certainty.13
The qualitative variants of the CLT consists in listening for expired air around the endotracheal tube when the cuff is deflated. Sensitivity for predicting PERD is low, probably due to the subjective nature of the assessment.14–16 For the quantitative CLT (qtCLT), the expiratory tidal volume (exVT) is measured with the cuff inflated then deflated. The difference between the two values reflects the size of the leak.17 In a meta-analysis of studies in adults, the qtCLT had 87% specificity (95% CI: 0.82% to 0.90%) and 62% sensitivity (95% CI: 0.49% to 0.73%) for UAO.16 However, of the 28 included studies, 5 used only the qualitative CLT and 1 either the qualitative CLT or the qtCLT, possibly decreasing the estimated sensitivity. Moreover, no standardised definition of UAO was used. We are aware of a single paediatric study providing information on the performance of the qtCLT in predicting postextubation stridor.17 A leak of less than 11% was 61% sensitive, 53% specific and 59% accurate. The corresponding values for an ultrasound-measured air column width between the vocal cords of less than 0.8 mm were 93%, 86% and 91%. In this study, postextubation stridor was defined as ‘a high-pitched inspiratory wheeze requiring medical intervention (corticosteroid therapy or reintubation) within 24 hours of extubation and associated with respiratory distress’. Although clear and simple, this outcome is not fully satisfactory: respiratory distress is not defined objectively, the use of NIV after extubation is not considered among the medical interventions for respiratory distress, and the time interval for stridor to develop is only 24 hours. Moreover, the study included 400 patients, all of whom were recruited at a single centre with more than half admitted for surgical causes. Consequently, there is a need for a large study capable of providing definitive data on the performance of the qtCLT in predicting UAO-PERD after extubation of PICU patients.
We designed a multicentre observational study with a large sample size, obtained through the participation of 19 PICUs, to evaluate the performance of the qtCLT for predicting UAO-PERD. We used a strong definition of UAO-PERD combining a requirement for intravenous corticosteroid (IV-CS) therapy to decrease the laryngeal oedema and/or ventilator support by high-flow nasal cannula (HFNC) and/or by NIV or repeat IMV with a Westley score ≥4 with at least one point for stridor at each initiation.
Methods
Study design
This is an observational, non-randomised, prospective, cohort study that is being conducted in 19 French PICUs (final version n°1). Enrolment began on 5 October 2022 and is expected to last 18 months, the planned sample size being 900. The occurrence of UAO-PERD is recorded during the first 48 hours after extubation, and additional outcomes are recorded until day 28.
The study was approved by the appropriate ethics committee on September 2021 (approval #CEER-RD 2021-578) and was registered prior to initiation on ClinicalTrials.gov (NCT05328206).
A monthly newsletter about the progress of the study is sent to all investigators.
Participants
The inclusion criteria are full-term birth, age at least 2 days up to including 17 years, intubation with a cuffed endotracheal tube, IMV for at least 24 hours and oral informed consent to study participation. The exclusion criteria are tracheostomy before extubation, indication for long-term NIV, history of upper airway abnormalities, upper airway surgery within 1 month before the assessment of eligibility, decision to withdraw or withhold life-sustaining treatment and previous inclusion in the study (figure 1). We also, recommended that investigator systematically check for the absence of supraglottic obstruction.
Patient flow chart. HFNC, high-flow nasal cannula; IMV, invasive mechanical ventilation; IV-CS, intravenous corticosteroid; NIV, non-invasive ventilation; PERD, postextubation respiratory distress; PICU, paediatric intensive care unit; UAO-PERD, PERD related to upper airway obstruction.
Patients and public involvement
The patients, their parents and the public had no role in designing this protocol or in writing the present manuscript. They have no role in conducting the study and will not be involved in analysing, interpreting or disseminating its results. The study findings will be published in a peer-reviewed journal and reported at one or more scientific meetings.
Recruitment
Eligible children will be identified by the study investigator at each participating PICU (box 1), who explains the study and requests oral informed consent by the parents or legal guardian and, when able to understand, the patients (figure 1).
List of the 19 study sites
Réanimation pédiatrique - CHU Robert Debré, Paris
Réanimation et surveillance continue médicochirurgicale - CHU Necker-Enfant Malades, Paris
Anesthésie et Réanimation pédiatrique - CHU Necker-Enfant Malades, Paris
Réanimation néonatale et pédiatrique - CHU Armand Trousseau, Paris
Réanimation pédiatrique et Néonatale - CHU Kremlin Bicêtre
Réanimation pédiatrique - CHU Raymond Poincaré, Garches
Réanimation pédiatrique - CH Marie Lannelongue, Le Plessis-Robinson
Réanimation pédiatrique - CHRU Jeanne de Flandre, Lille
Réanimation médicale pédiatrique - CHRU Nancy Brabois
Réanimation pédiatrique - CHU Hospices Civils de Lyon
Réanimation et soins intensifs pédiatrique - CHU Grenoble Alpes
Anesthésie et Réanimation pédiatrique - CHU de la Timone, Marseille
Médecine néonatale et Réanimation pédiatrique - CHU Clermont-Ferrand
Réanimation néonatale et pédiatrique - CHU Toulouse
Réanimation pédiatrique – CHU Pellegrin Bordeaux
Maladie cardio-vasculaires congénitales – CHU Haut-Lévêque Bordeaux
Réanimation pédiatrique et unité de surveillance continue - CHRU de Tours
Réanimation et soins intensifs pédiatriques - CHU de Caen
Réanimation pédiatrique, surveillance continue pédiatriques et réanimation des cardiopathies congénitales - CHU de la Réunion Pôle Nord
Study intervention
If extubation is planned, qtCLT will be performed according to the protocol developed by Miller and Cole and used by El Amrousy et al, within 1 hour before extubation (table 1).17 18 The ventilator will be set to assist-volume control mode with a tidal volume (VT) of 8–10 mL/kg of predicted body weight, up to 450 mL. Six consecutive exVT measured by the ventilator will be recorded with the cuff inflated then deflated. The mean exVT values (in mL) are computed for the values obtained with the cuff inflated and deflated. Leak volume is defined as the difference between exVT inflated and exVT deflated. Leak percentage (%) is calculated as follows: 100 × (exVT inflated − exVT deflated)/ exVT inflated.17 19
Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) checklist: enrolment, interventions and assessments
Extubation will be performed according to each centre’s procedure. No specific instructions were given for the study. The intensivist will record the Westley score immediately after extubation and then over the next 48 hours before the initiation of IV-CS and/or ventilatory support (HFNC, NIV or reintubation). The value serves to define UAO-PERD, as described below in the section on the primary outcome. Thus, if none of these treatments is given, the Westley score will be recorded only once, immediately after extubation (figure 1).
Surveys completed by the participating PICUs indicated that 80% used automatic leak compensation, regardless of the ventilator model. Automatic leak compensation relies chiefly on the minute volume and therefore starts only after more than 1 min. In our study, the qtCLT will be performed within less than 1 min. Consequently, automatic leak compensation cannot affect the measured values. We therefore require no changes to the ventilator setups that are part of standard practice in each participating PICU.
Study objectives
The primary study objective is to assess the performance of qtCLT in predicting severe UAO-PERD with onset within 48 hours after extubation.
The secondary objectives are to determine the frequency of extubation failure; identify risk factors for severe UAO-PERD within 48 hours after extubation; describe the frequency and effects of pre-extubation IV-CS within 12 hours; and develop a score predicting severe UAO-PERD within 48 hours after extubation, which would subsequently undergo an external validation study.
Primary Endpoint
The primary endpoint is the cumulative incidence of UAO-PERD within 48 hours after planned extubation. UAO-PERD is defined as the initiation of IV-CS and/or ventilatory assistance (HFNC, NIV or reintubation) with a Westley score ≥4, including at least 1 point for stridor (indicating laryngeal obstruction).
Each participating PICU is asked to ensure that, to the extent possible, the primary endpoint is assessed by a person different from the person who performs the qtCLT.
Secondary endpoints
The secondary endpoints are the cumulative incidence of extubation failure, defined as reintubation for respiratory distress within 48 hours after extubation combined with a pre-reintubation Westley score ≥4 with at least 1 point for the stridor item; the frequency of IV-CS therapy given within 12 hours before extubation; the frequency of UAO-PERD in patients with versus without this treatment; the number of PICU-free days by day 28 after enrolment; in-PICU mortality; and, in patients who died in the PICU, the time to death. The predictive score for UAO-PERD will be assessed by computing the area under the receiver operating characteristic curve (AUROC).
Data collection
The study data will be collected prospectively by the investigator at each participating PICU, using a secure online database (CleanWEB 2022 Telemedicine Technologies, Boulogne-Billancourt, France). For each patient we will record the age (months); weight (kg), sex; reason for PICU admission; Paediatric Index of Mortality 3 (PIM-3; %) and Paediatric Logistic Organ Dysfunction 2 (PELOD-2; %) scores; internal diameter of the cuffed endotracheal tube (mm); history of failed extubation during the same PICU stay prior to enrolment; if planned intubation (in the operating room); emergency intubation; out-of-hospital intubation; number of intubations during the index PICU stay; whether IV-CS therapy was given within 12 hours before extubation; history of failed extubation during the same PICU stay before enrolment; and IMV duration (hours). The qtCLT results (mean of the six exVT values with the cuff inflated and mean of the six exVT values with the cuff deflated); date and time of extubation and Westley score immediately after extubation will be collected; as well as the use within 48 hours after extubation of IV-CS and/or ventilatory assistance (HFNC, NIV or reintubation), with the time to initiation. All the Westley score items will be precisely reported for each evaluation. The PICU stay length, number of PICU-free days by day 28 and in-PICU mortality with the time to death are recorded.
Statistical analysis
Sample size estimation
Based on the literature, the expected proportion of patients with severe UAO-PERD is 15%.14 17 20 21 Therefore, 900 patients (135 with vs 765 without UAO-PERD) are needed to estimate the qtCLT AUROC with a two-sided 95.0% CI of <0.10 for AUROC values greater than 0.75 (eg, 95% CI of 0.071 if the AUROC equals 0.90). Recruiting 765 patients without PERD will allow the identification of a cut-off having 90% specificity, with an exact two-sided 95% CI of 0.044.
The analyses will be repeated in the following age subgroups: (2 days–2 years), (2 years–8 years) and (8 years–17 years). The smallest expected ROC curve width for AUROC values greater than 0.75 is 0.199 (0.141 for an AUROC of 0.90) within each age subgroup. The cut-off having 90% specificity will have an exact two-sided 95% CI of 0.09 within the same age subgroup.
Based on data recorded in each of the 19 participating PICUs during the year preceding study initiation, 18 months will be required to recruit 900 patients meeting all the study inclusion criteria and none of the study exclusion criteria.
Statistical analysis principles
All enrolled patients will be included in the analysis. Descriptive statistics will be computed for the included patients, as median (IQR) for quantitative variables and as number (percentage) for qualitative data.
Primary endpoint
The AUROCs for the leak as a percentage and as a volume associated with the occurrence of UAO-PERD will be estimated with their 95% CIs as described by DeLong and DeLong, in the overall population and in each of the above-defined age subgroups.22 Bootstrapping will be used to check 95% CI boundaries. Age-adjusted AUROC values will be estimated according to the age subgroups defined above.
The analyses will consider death within 48 hours after extubation and reintubation due to non UAO-PERD, as defined in the protocol, as competing risks. Missing data will be handled by multiple imputation, and an analysis of cases with no missing data will also be performed.
Secondary endpoints
The cumulative incidence of extubation failure, defined as reintubation performed within 48 hours after extubation after documenting a Westley score ≥4, including at least 1 point for stridor, will be estimated with its 95% CI, using Gray’s estimator and considering two competing risks, namely, death and reintubation due to causes other than UAO-PERD. This cumulative incidence will be estimated in the overall population and in each age sub-group.
The risk factors for UAO-PERD will be investigated using a logistic model predicting the presence or absence of respiratory distress. Only inclusion variables will be considered. Continuous variables may be transformed to ensure that the effect of the variable is log-linear. In the event of a non-linear effect, spline modelling will be considered. We will analyse the impact of inadapted internal diameter of the cuffed endotracheal using the Khine formula.
The frequency of IV-CS therapy within 12 hours before extubation will be determined. Potential associations between this treatment and the occurrence of UAO-PERD will be assessed using a weighted propensity score approach. Risk factors for UAO-PERD within 48 hours after extubation will be sought by building a logistic model for predicting UAO-PERD. Only baseline variables, that is, variables recorded at the admission and the extubation, will be considered. Continuous variables may be transformed to ensure log-linearity of their effect. If the effect is non-linear, spline modelling will be considered. The characteristics and possible over-optimism of the predictive score thus developed will be evaluated using bootstrapping and cross-validation.
The PICU stay length, number of PICU-free days by day 28 and in-PICU mortality with the time to death will be reported as descriptive statistics.
Ethics and dissemination
The study was approved by the institutional review board of the Robert Debré University Hospital in Paris, France, on September 2021 (approval #2021-578). Specific agreements have been made with all participating centres and the report of Robert Debre Committee is valid for all sites, given the nature of the study with respect to the French law.
As soon as a new centre is added, a new approval will be submitted to the Robert Debre Ethics Commitee. Only after approval can the centre be added. Thus, the 19 participating centres had ethical approval to conduct the study. This study requires a signed ‘no objection’ agreement to the use of the data, as it is considered by the ethics committee, in accordance with French law, to be a non-interventional study, as it is observational, devoid of risk, does not modify patient care and the procedures performed are the usual ones. Finally, prior enrolment, information and verbal consent will be obtained from the parents and from the minor if his/her age and condition allow him/her to understand. We have specific patient consent form for minor (one from 7-year-old to 11-year-old and another 12-year-old to 17-year-old) and parents. The results will be submitted for publication in a peer-reviewed journal and reported at one or more scientific meetings.
Discussion
The paucity of available evidence has not allowed the development of strong recommendations about performing a qtCLT before the extubation of paediatric patients.13 To our knowledge, this is the first multicentre study evaluating the performance of the qtCLT in predicting the risk of UAO-PERD after extubation in the PICU. The age range is broad, from 2 days up to including 17 years. As there does not seem to be a clear high-risk age group in the literature (varying between studies, either under 2 years or under 5 years), we decide to include patients up to and including 17 years and perform subgroup analyses by age. The main research question is whether the qtCLT reliably predicts the risk of UAO-PERD, in children without upper airways abnormalities, thereby providing information likely to improve patient management. We excluded children with upper airway abnormalities or surgery, first because we focused on the effect of an endotracheal tube on healthy airway. Second, the Westley score is not adapted for children with such abnormalities. Finally, these cases are more likely to receive known preventive measures (corticosteroid), which introduce bias into analysis. In addition, the study will identify risk factors for UAO-PERD and prospectively collect information on current ventilator weaning and postextubation practices in French PICUs. Finally, we will develop a score for predicting UAO-PERD that should be suitable for use in daily practice. This score will require external validation in a separate study.
No validated method for assessing UAO-PERD in paediatric patients has been reported to date. UAO-PERD is often defined as the onset of stridor or the need for reintubation within 48 hours after planned extubation.10 23 Although reintubation is a strong objective marker, it is required in less than 10% of paediatric patients with UAO-PERD, and this proportion is declining, notably due to the increasing use of NIV for managing ventilator weaning.10 11 24 Consequently, using only reintubation to define UAO-PERD would require a very large sample size. Moreover, reintubation may be needed for reasons other than UAO. On the other hand, isolated stridor is not associated with worse patient outcomes. Only stridor combined with evidence of respiratory distress, with or without a need for intervention (IV-CS and/or HFNC, NIV or reintubation), is relevant. The Westley score, first described by Westley et al in 1978 assesses the severity of UAO in acute conditions such as croup, has since been validated on numerous occasions as a reliable tool for clinicians at the bedside.25–28 It consists of 5 items, scored from 0 (absent) to 5 (maximum), and it is now the most widely used score in the clinical and therapeutic evaluation of respiratory distress induced by UAO.20 26 Because of the intrinsic qualities of the Westley test and the pathophysiological and semiological similarity between acute laryngitis and postextubation UAO, this score has been used in a paediatric study of UAO.29
Our strong definition includes respiratory distress defined as requiring intervention (IV-CS and/or ventilatory assistance) with a Westley score ≥4, with at least 1 point for stridor. The use of the Westley score at each initiation makes it possible to distinguish between the use of support for inspiratory and non-inspiratory reasons and to neglect the planned or systemic use of postextubation support. However, to differentiate PERD due to UAO versus other causes, the Westley score has to include at least 1 point for stridor. Thus, use of the interventions is linked to objective evidence of respiratory distress.29–31 UAO-PERD shares pathophysiological similarities with croup. Based on the study of Yang et al, we chose 4 as the cut-off for defining severe UAO-PERD.26 The combined use of treatments in the definition decreases the risk of bias due to inter-observer variability in Westley score determination.
Several factors support the feasibility of conducting and completing our study. First, the project was presented several times at meetings of the French PICU research network on respiratory conditions (Groupe francophone de reanimation et urgences pédiatriques, GFRUP). This dissemination effort allowed us to obtain the participation of 19 PICUs, accounting for over 85% of all PICUs in France. Second, the study carries no risk to the participants. The qtCLT is a safe procedure that is part of standard care. All other investigations and treatments used in the study patients are also part of standard care. Third, all local investigators followed a training session on protocol procedures and Westley score determination. Finally, given that the primary outcome is assessed after only 2 days, the number of missing data for the primary outcome is expected to be very small.
Our study design has limitations. First the qtCLT is associated with incredible amount of variation, depending on the respiratory effort or diaphragm weakness or the status of the patient during cuff deflation, or the ventilators used and accuracy of VT measurements. These limitations are a strong issue and explain why the American and European Society guidelines for weaning adult and peadiatric patients off mechanical ventilation are based on very low-quality evidence for performing the qtCLT.13 32 In fact, we chose to give priority to carrying out a test that could be used on a daily basis, at the patient’s bedside, without the need for additional equipment. This choice therefore does not allow us to study the validity of the qtCLT intrinsically. This is a conscious decision, and although it is a limitation of the study, it is also a strength, as it allows us to include many patients in many centres. Depending on results, a question to be discussed will concern the technical conditions for performing the leak test and the relevance of repeating the study with more precise and specifically dedicated equipment.
Second, extubation management was left to the discretion of each centre. In order to limit the impact, we recorded pre-extubation measures such as use of intravenous corticosteroids, and intubation procedure. Third, the management of UAO-PERD will probably vary across participating PICUs, given the absence of guidelines for initiating IV-CS, HFNC, NIV or repeat IMV. However, our use of a definition that comprises not only these interventions but also the physical findings, as reflected by the Westley score and stridor subscore, will limit this potential source of bias. To limit interventions to those that increase patient morbidity, we decided not to include inhaled racemic epinephrine in the study protocol, which is almost systematically used by several PICU teams whatever the patients’ respiratory status after extubation. Finally, we decided to focus only on the quantitative test to avoid repeating the many studies that have already been done in this area, with conflicting results, probably due to the many limitations associated with this test. If we carried out the two tests in parallel, there would be a high risk of obtaining different results between the two, without knowing what to make of them, given that the limitations of each test are very different.
Conclusion
This protocol for a prospective multicentre observational study of qtCLT performance in predicting UAO-PERD has important strengths and is therefore expected to determine whether the result of qtCLT is a risk factor for the development of PERD due to UAO. If this is the case, the qtCLT could be used in an RCT to guide the management of extubation in paediatric patients. Among these strengths is the composite definition of UAO-PERD combining clinical features, reflected by the Westley score and presence of stridor, with a range of treatment requirements that includes postextubation NIV. External validation of the predictive score designed based on the study results will be performed subsequently.
Ethics statements
Patient consent for publication
References
Footnotes
Contributors BL, AH, JN, AM, GG, GP, ML and SD contributed to conceive and prepare the study. BL, MR-R and SD contributed to design the study. BL and AH wrote the first draft of the manuscript. JN, AM, GG, GP, ML, MR-R and SD contributed to draft the manuscript then to review it for important intellectual content. BL, AH, MR-R and SD are contributing to manage the study and to collect, analyse and interpret the data. All authors have read the final version of this manuscript and approved its submission to BMJ Open.
Funding This work was supported by the non-profit international organisation Réseau Mère Enfant de la Francophonie (RMEF – grant number “Not Applicable”).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.