Article Text
PDF
Abstract
Aim To investigate the accuracy of the recently published international recommendations for ECG interpretation in young athletes in a large cohort of white and black adolescent soccer players.
Methods 11 168 soccer players (mean age 16.4±1.2 years) were evaluated with a health questionnaire, ECG and echocardiogram; 10 581 (95%) of the players were male and 10 163 (91%) were white. ECGs were retrospectively analysed according to (1) the 2010 European Society of Cardiology (ESC) recommendations, (2) Seattle criteria, (3) refined criteria and (4) the international recommendations for ECG interpretation in young athletes.
Results The ESC recommendations resulted in a higher number of abnormal ECGs compared with the Seattle, refined and international criteria (13.2%, 4.3%, 2.9% and 1.8%, respectively). All four criteria were associated with a higher prevalence of abnormal ECGs in black athletes compared with white athletes (ESC: 16.2% vs 12.9%; Seattle: 5.9% vs 4.2%; refined: 3.8% vs 2.8%; international 3.6% vs 1.6%; p<0.001 each). Compared with ESC recommendations, the Seattle, refined and international criteria identified a lower number of abnormal ECGs—by 67%, 78% and 86%, respectively. All four criteria identified 36 (86%) of 42 athletes with serious cardiac pathology. Compared with ESC recommendations, the Seattle criteria improved specificity from 87% to 96% in white athletes and 84% to 94% in black athletes. The international recommendations demonstrated the highest specificity for white (99%) and black (97%) athletes and a sensitivity of 86%.
Conclusions The 2017 international recommendations for ECG interpretation in young athletes can be applied to adolescent athletes to detect serious cardiac disease. These recommendations perform more effectively than previous ECG criteria in both white and black adolescent soccer players.
- adolescent athlete
- arrhythmogenic right ventricular cardiomyopathy
- ecg criteria
- hypertrophic cardiomyopathy
- screening
Statistics from Altmetric.com
- adolescent athlete
- arrhythmogenic right ventricular cardiomyopathy
- ecg criteria
- hypertrophic cardiomyopathy
- screening
Introduction
Competitive adolescent athletes comprise the largest cohort of exercising individuals in society and constitute an increasing number of athletes competing at national and international level. Although exercise-related sudden cardiac death (SCD) is uncommon, our recent study reported an incidence of 6.8/100 000 among UK-based adolescent soccer players1 which is considerably higher than previous estimates.2 Moreover, in an American series of 1866 SCDs in young athletes, over 70% of deaths occurred in those aged <17 years.3 The ECG is effective in detecting potentially serious cardiac disease in athletes,4–6 and numerous international sporting organisations undertake cardiac screening of adolescent athletes. Paradoxically, data relating to the adolescent athlete’s ECG are sparse and most ECG interpretation criteria are derived from and evaluated in adult athletes.7 8
The 2010 European Society of Cardiology (ESC) recommendations for athlete ECG interpretation9 are sensitive for detecting individuals harbouring serious cardiac diseases but have a high false positive rate, particularly in black athletes. Contemporary criteria account for ethnicity10 11 and other non-specific electrical patterns.6 12 Both the 2013 Seattle criteria and 2014 refined criteria show higher specificity than the ESC recommendations in adult athletes.6 13–16 International recommendations were devised by a group of international experts and published in 2017.17 Their overarching aim was to unify the recommendations for interpreting the athlete’s ECG. This study compared the effectiveness of those international recommendations (2017) with previous ECG criteria for young athletes in a large group of adolescent soccer players.
Methods
Setting
The English Football Association (FA) has conducted the largest cardiac screening programme for athletes in the UK since 1996 and encompasses 92 soccer clubs across all professional leagues in England. The entire cohort in this study comprised elite junior adolescent athletes (aged 15–17), who were talented soccer players excelling at youth level within the preceding 5 years. Athletes underwent mandatory cardiac evaluation including a health questionnaire, physical examination, ECG and echocardiogram. The process was conducted by cardiologists experienced in the cardiac assessments of athletes. Players requiring further investigations were investigated at specialist regional centres. Decisions relating to eligibility to play for individuals who harboured cardiac disease were discussed at an expert panel meeting, comprising 12 cardiologists and chaired by SS (principal investigator). This paper focuses on ECG interpretation in the same cohort of 11 168 athletes that we described in the New England Journal of Medicine in 2018.1
Participants
Between 1996 and 2016, 11 168 adolescent soccer players underwent pre-participation screening with a health questionnaire, physical examination, ECG and echocardiography. The health questionnaire inquired about cardiac symptoms or family history of premature (≤50 years) cardiovascular disease or SCD. Ethnicity was self-reported. Physical examination was performed by the club doctor.
Electrocardiogram
A standard 12-lead ECG was performed in the supine position during quiet respiration, using a GE Marquette Hellige recorder (Milwaukee, Wisconsin, USA) or Philips Pagewriter Trim III (Bothell, Washington, USA) at a paper speed of 25 mm/s and amplification of 0.1 mV/mm. Wave voltages were measured in each lead using callipers and a millimetre ruler as described elsewhere.18
Left ventricular (LV) and right ventricular (RV) hypertrophy was defined using the Sokolow-Lyon voltage criteria.12 19 T-wave inversion (TWI) was defined as a negative T-wave deflection ≥−0.1 mV. The QT interval was corrected for heart rate using Bazett’s formula.20
ECGs were initially analysed by regional cardiologists with expertise in sports cardiology based on a combination of personal experience prior to 2005, the original ESC recommendations from 2005 to 2010,21 and the 2010 ESC recommendations.9 All 11 168 ECGs were evaluated independently by the first author (AM) in 2016–2017 with retrospective application of 2010 ESC recommendations, Seattle criteria, refined criteria and international recommendations.6 9 13 The first author was initially blinded to any pathological conditions that were subsequently diagnosed with echocardiography. The ECGs of athletes with equivocal ECG findings were discussed with the principal investigator (SS) who made the final adjudication.
Echocardiography
Two-dimensional transthoracic echocardiography was performed by sonographers accredited by the British Society of Echocardiography in accordance with the standard American Heart Association and ESC protocols.22 23 Sonographers were blind to findings on physical examination and ECG. Standard views obtained with wall thickness, cavity dimension measurements and identification of the origins of the left and right coronary arteries were made in accordance with established guidelines.22 23 Each echocardiogram was reviewed by the respective sports cardiologist. Echocardiograms reported as abnormal or associated with an abnormal ECG were independently reviewed by the first (AM) and senior author (SS).
Further evaluation
Athletes with symptoms, suspicious family history, abnormal physical examination and abnormalities on the ECG or echocardiogram were investigated further as discussed previously.1 Hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), arrhythmogenic right ventricular cardiomyopathy (ARVC) and long QT syndrome were considered in accordance with predetermined criteria and established guidelines.24–30 The results of all investigations were sent to the FA headquarters for review.
Ethical approval
The protocol was reviewed against the requirements of the Research Governance Framework for Health and Social Care 2005, incorporating the Medicines for Human Use (Clinical Trials) Regulations 2004. The age of informed consent in the UK is 16 years, with parental or guardian consent sought for those under this age.
Diagnosis of disorders associated with sudden cardiac death
The diagnosis of hypertrophic cardiomyopathy was based on a left ventricular wall thickness ≥15 mm in any myocardial segment on echocardiography or cardiovascular MRI in the absence of another condition capable of producing left ventricular hypertrophy.24 25 31 The diagnosis of dilated cardiomyopathy was based on findings of a dilated left ventricle (left ventricular end-diastolic dimension for males >59 mm and for females >53 mm) and a reduced ejection fraction (<50%).26 27 The diagnosis of arrhythmogenic right ventricular cardiomyopathy was based on published criteria.28 32 The diagnosis of long-QT syndrome was based on a corrected QT interval (QTc) of ≥500 ms or a QTc 470–490 ms associated with abnormal T-wave morphology, paradoxical prolongation of the QT interval with exercise or a positive genetic test.29 30 The Wolff–Parkinson–White (WPW) ECG pattern was based on findings of a short PR interval and a slurred upstroke to the QRS complex.
Statistical analysis
Data are expressed as mean±SD or percentages as appropriate and analysed with SPSS software V.20. Continuous variables were tested for normality using the Kolmogorov-Smirnov test. Group differences were tested with the Student t-test or Mann-Whitney U test for normally and non-normally distributed variables, respectively. The χ2 test was used to compare the number of abnormal ECGs between the four ECG criteria, males and females and the two ethnic groups. Sensitivity, specificity, positive predictive values and negative predictive values were calculated for each of the four criteria according to the ethnic group. These calculations were performed using 2×2 contingency tables in GraphPad Prism software V.6.01 (La Jolla, California, USA).
Results
Demographics
In total, 10 163 (91%) athletes were white, 1005 (9%) were black (African/Caribbean) and 10 581 (95%) were male. White athletes were slightly younger than black athletes (16.4±1.2 vs 16.7±1.5 years; p<0.001) and had a smaller body surface area (1.89±0.47 vs 1.99±0.52 m2; p<0.001). All athletes performed 13±0.8 hours of exercise per week, including 6 hours of cardiovascular training, 2 hours of weight training and 5 hours of soccer in training and match-day games.
Prevalence of abnormal ECGs according to the four criteria
Based on the ESC recommendations, 1474 (13.2%) athletes had an abnormal ECG compared with 480 (4.3%) with the Seattle criteria, 294 (2.9%) with the refined criteria and 205 (1.8%) with the international recommendations (p<0.001). All four criteria were associated with a greater number of abnormal ECGs in black athletes compared with white athletes (figure 1).
Abnormal ECG rates in white and black adolescent athletes according to the four criteria. White athlete depicted by grey figure; black athlete depicted by black figure. ESC, European Society of Cardiology.
Specific ECG abnormalities in relation to the four criteria
An abnormal QTc interval accounted for 26% of abnormalities in accordance with the ESC criteria (figure 2). The less conservative QT limits in the Seattle, refined and international criteria reduced the prevalence of an abnormal QT interval by 89%. Consideration of left or right axis deviation as normal variants when present in isolation was responsible for the greatest reduction in abnormal ECGs between the Seattle criteria versus refined criteria and the Seattle criteria versus international recommendations (figure 2). Abnormal TWI was more prevalent with the ESC recommendations (3.0%) compared with the Seattle (1.3%) and refined criteria (2.1%). The international criteria revealed the lowest prevalence of abnormal TWI (1.1%) primarily because the juvenile pattern of TWI in V1–V3 in subjects aged <16 years is considered a normal variant in asymptomatic athletes without a family history of cardiomyopathy or premature SCD.17
Pie charts demonstrating the distribution of abnormalities among athletes with an abnormal ECG according to the four criteria. ESC, European Society of Cardiology; LBBB, left bundle branch block; RBBB, right bundle branch block; RVH, right ventricular hypertrophy; WPW, Wolff–Parkinson–White pattern. *Borderline ECG finding according to the specific guideline. **Right ventricular hypertrophy in the Seattle criteria included voltage criteria and right axis deviation.
Ethnic differences in ECG patterns
Black athletes revealed a higher prevalence of abnormal TWI compared with white athletes with the ESC recommendations (11.7% vs 2.2%, p<0.0001), the Seattle criteria (2.2% vs 1.2%, p=0.012) and the international recommendations (2.2% vs 0.9%, p=0.0005). The refined criteria regarded TWI beyond V1 as abnormal in white athletes, which led to a similar prevalence of abnormal TWI as black athletes (2.2% vs 2.1%, p=0.8029). Black athletes also demonstrated a higher prevalence of voltage criterion for left and right atrial enlargement and RV hypertrophy than white athletes. In contrast, voltage criterion LV hypertrophy was more common in white athletes (table 1).
Electrocardiographic features of adolescent athletes by ethnicity and sex
A juvenile ECG pattern (TWI V1–3) was present in 25 (1.8%) white athletes aged <16 years compared with 26 (0.3%) white athletes ≥16 years old (p<0.0001).
Sex differences in ECG patterns
Male athletes demonstrated a higher prevalence of sinus bradycardia, sinus arrhythmia and longer PR intervals than female athletes (table 1). Voltage criterion LV hypertrophy was four times more common in male athletes compared with female athletes (24.8% vs 6.3%; p<0.0001). There were no sex differences in the prevalence of atrial enlargement or axis deviation. Female athletes demonstrated a longer QTc interval than male athletes. Anterior TWI was also more common in females than males in leads V1–V3. Lateral TWI was rare in all white adolescent athletes, being present in just 0.3% males and none of the females.
Interobserver and intraobserver variability between ECG findings
A random selection of 1000 ECGs (every 10th ECG) were selected to assess intraobserver variation in ECG interpretation of the first author (AM). The same 1000 ECGs were read independently by the second author (HD) to assess interobserver variation. There was excellent agreement for intraobserver and interobserver variation for all four ECG criteria (ESC, Seattle, refined and international recommendations) with kappa values ranging from 0.93 to 0.97 for intraoberver and 0.79–0.85 for interobserver. The kappa values for interobserver and intraobserver variation for the international recommendations were 0.96 and 0.82, respectively.
Detecting cardiac disease with history, examination, ECG and echocardiography
Of the 11 168 athletes, 267 (2.39%) were diagnosed with cardiac pathology according to established criteria and pre-existing guidelines.24–30 Forty-two (0.37%) athletes demonstrated serious diseases associated with exercise-related SCD including HCM (n=5), ARVC (n=2), DCM (n=1), long QT syndrome (n=3), significant bicuspid aortic valve disease (n=3), anomalous coronary artery origins (n=2) and WPW ECG pattern (n=26)1 (table 2). There was a similar prevalence of serious disease among white and black athletes (0.36% vs 0.4%, p=0.9). All athletes with serious diseases were male except one female with a WPW ECG pattern. Of these, 36 (86%) demonstrated an abnormal ECG (table 3). Only three athletes (7%) reported symptoms. Thirteen (29%) athletes had an abnormal echocardiogram, comprising five with HCM, two with ARVC, one with DCM, three with significant bicuspid aortic valve disease and two with anomalous coronary artery origins. Six (14%) athletes with potentially serious disease demonstrated a normal ECG (one with ARVC, two with anomalous coronary artery origins and three with significant bicuspid aortic valve disease).
Clinical findings in athletes with cardiac conditions associated with sudden death
Athletes with cardiac pathology detected by history, physical examination, ECG and echocardiogram
A total of 225 athletes revealed congenital, septal or valvular abnormalities. Of these, three athletes (1%) reported symptoms, 74 (33%) had abnormal cardiovascular examination and 48 (21%) exhibited an abnormal ECG (table 3).
ECG abnormalities in athletes with overt cardiac disease
All five athletes with HCM and one athlete with ARVC showed abnormal TWI according to all four criteria (table 2). The athletes with long QT syndrome and WPW patterns were also detected by all four criteria.
Of the 68 athletes with bicuspid aortic valves, 17 (25%) had anomalies on the ECG, most of which would be considered abnormal according to the ESC 2010 recommendations including left axis deviation (n=8), left atrial enlargement (n=6), right axis deviation (n=2) and TWI in leads V1–V3 (n=1). Twenty-six (38%) of 62 athletes diagnosed with an atrial septal defect revealed an abnormal ECG according to the 2010 ESC recommendations including voltage criterion for RV hypertrophy (n=14), right axis deviation (n=9) and right atrial enlargement (n=3).
Left axis deviation was the only ECG abnormality in 2 of 29 athletes with aortic regurgitation; 1 of 13 with a ventricular septal defect (VSD) and 1 of 18 with a patent ductus arteriosus. Three of 24 athletes with mitral valve prolapse revealed voltage criterion for left atrial enlargement.
False-positive ECGs for the ESC recommendations, Seattle criteria, refined criteria and international recommendations
The false-positive rate for serious cardiac disease was 12.9% according to the ESC recommendations, 4% for the Seattle criteria, 2.6% for the refined criteria and 1.5% for the international recommendations (1.4% and 3.3% for white and black athletes, respectively).
Sensitivity and specificity of the four ECG criteria
For serious cardiac pathology, the ESC criteria demonstrated a specificity of 87% which improved to 96% with the Seattle criteria, 97% with the refined criteria and 98% with the international recommendations. There was a progressive increase in the positive predictive value from ESC recommendations, Seattle criteria, refined criteria and international recommendations in both white and black athletes (table 4).
Sensitivity, specificity and positive predictive value for serious cardiac pathology by each ECG criteria according to ethnicity, history, examination and echocardiogram
All four criteria revealed a relatively high sensitivity (86%) for detecting serious conditions associated with SCD. The ECG, however, revealed a low sensitivity for detecting all cardiac pathology primarily due to its inability to detect congenital structural abnormalities found mainly by echocardiography. The ESC recommendations and Seattle criteria were both associated with a sensitivity of 31% while the refined and international criteria were associated with a lower sensitivity of 23%. Exclusion of left axis deviation as a definitive abnormality largely accounted for the lower sensitivity of the refined and international criteria which was present in 12% of all 103 athletes with a congenital abnormality and abnormal ECG.
Further evaluation for cardiac disease
During a follow-up period of 10.6 years, there were eight deaths from cardiovascular disease, affecting males only.1 Two athletes were diagnosed with HCM during the screening process but continued to compete despite medical advice and both died during competition. The remaining six had a normal ECG and echocardiogram at 16 years of age and died 6.8 years (range, 0.1–13.2 years) after the initial screening process.1 Of these, five had postmortem evidence of a cardiomyopathy (HCM (n=3), idiopathic left ventricular hypertrophy (n=1), ARVC (n=1)) and one had a structurally normal heart.
Discussion
This study of over 11 000 adolescent athletes compared the performance of the ESC, Seattle, refined and international ECG criteria in adolescent athletes. Forty-two (0.37%) athletes were subsequently diagnosed with potentially serious disease associated with SCD, the prevalence of which is similar to that in adult athletes.5 All four criteria revealed a relatively high sensitivity (86%) for detecting serious disease. The number of abnormal ECGs dropped significantly from 13.2% using the ESC recommendations to 4.3% and 2.9% with the Seattle and refined criteria, respectively. The international recommendations reduced the number of abnormal ECGs to 1.9% and also revealed the greatest specificity for both white (99%) and black athletes (97%).
Association of ethnicity
The number of ECGs warranting further investigation in adolescent athletes was 40% lower in white athletes and over 60% lower in black athletes compared with those reported in a large cohort of British adult athletes6 when applying the refined criteria, and may reflect the lower lean body mass and inability to exercise to similar workloads as adults as well as the smaller number of cumulative years in training. Consistent with studies in adults, adolescent black athletes demonstrated a higher prevalence of abnormal ECGs compared with white counterparts irrespective of the ECG criterion employed.33 34 The prevalence of ECG abnormalities in black adolescent athletes was qualitatively similar to black adult athletes4 15 but to a lesser degree and was driven specifically by a higher prevalence of voltage criteria for left and right atrial enlargement and RV hypertrophy. The prevalence of these three anomalies was almost sixfold higher in black athletes compared with white counterparts. In this regard, the refined criteria and international recommendations had the greatest impact in reducing the number of false-positive ECGs in black athletes resulting in a fivefold reduction compared with the ESC recommendations. In contrast with the ESC and the Seattle criteria, there was no difference in the prevalence of abnormal TWI between black and white athletes after application of the refined criteria. This difference is explained by the fact that several white athletes showed anterior TWI exceeding beyond V1 which was considered abnormal with the refined criteria. As with the Seattle criteria, the international recommendations regard anterior TWI in V1–V2 as a normal finding35 which partly accounted for the reduction in abnormal ECGs in white athletes between the international recommendations and the refined criteria (0.9% vs 2.1%, p<0.0001).
Association of ECG abnormalities with age
T-wave inversion in leads V1–V3 is present in up to 10% of adolescents aged <14 years and is considered a normal variant. Previous studies in white adolescent athletes have shown that the prevalence of this juvenile pattern drops significantly after the age of 16 years.10 36 37 In this study, white athletes aged <16 years demonstrated a sixfold higher prevalence of TWI in V1–V3 compared with those aged ≥16 years. The international recommendations allow for the juvenile pattern and were associated with a lower number of abnormal ECGs compared with the Seattle or refined criteria. None of the white athletes aged <16 years with a juvenile ECG pattern revealed findings consistent with a cardiomyopathy at echocardiography. This observation reinforces the concept that the juvenile ECG pattern in white athletes <16 years is likely a marker of physical immaturity38 that warrants investigation only in symptomatic athletes and those with abnormal physical examination or family history of cardiomyopathy.
Association of ECG abnormalities with sex
Male athletes demonstrated a higher degree of training-related changes compared with female athletes with a greater prevalence of sinus bradycardia, sinus arrhythmia, LV hypertrophy and a longer mean PR interval. Female adolescent athletes revealed similar degrees of left and right atrial enlargement and axis deviation as male athletes. The greatest differences in females and males were due to TWI. Consistent with previous studies from our group, both white and black females revealed a higher prevalence of anterior TWI than male athletes.15 39 None of the female athletes with anterior TWI in V1–V2 (4.3%) or beyond V2 (2.2%) were diagnosed with a cardiomyopathy. Such TWI confined to the anterior leads V1–V3 in female athletes may be a normal variant, though this finding requires validation with longer-term follow-up studies.
Accuracy of the history, physical examination, ECG and echocardiography in detecting cardiac disease
The ECG identified 86% of all athletes with overt phenotypic features of disease associated with exercise-related SCD. Consistent with previous reports,40 history and physical examination were associated with a poor diagnostic yield (7% and 5%, respectively). This provides further evidence that screening that uses only history and physical examination will fail to identify most cardiovascular abnormalities associated with SCD. Echocardiography detected only 13 (31%) of all cardiac conditions that were present and can cause exercise-related SCD. The WPW ECG pattern accounted for 26 (62%) of those potentially lethal cases where the ECG is essential for diagnosis. This finding has important implications for screening young athletes for serious cardiac disease because several recent autopsies have shown that up to 40% of sudden cardiac deaths in athletes are associated with a structurally normal heart41–43 and considered to have resulted from an electrical abnormality.44
Echocardiography successfully identified all 225 athletes with structural abnormalities. Of these, only 3 (1.3%) were symptomatic, 74 (33%) had been identified by physical examination and 48 (21%) had an abnormal ECG with a false-positive rate of 2%. Based on these observations, the ECG could be considered as the most cost-effective method for detecting athletes with conditions associated with exercise-related SCD with the caveat that physicians interpreting these ECGs should be adequately experienced.
Five athletes with a normal cardiac screen died 6.8 years afterwards and revealed pathological findings consistent with cardiomyopathy. It is well established that some cardiomyopathies exhibit an age-related penetrance and predisposed individuals may not reveal electrical or structural features of the disease until early adulthood.
Limitations
Our cohort consisted entirely of soccer players; soccer is a dynamic, high-intensity sport associated with the largest number of SCDs in Europe. Furthermore, the ECG changes were not dissimilar from two previous studies of over 1000 adolescent athletes who had participated in a variety of sports.7 34
Adolescence is defined by the WHO as the transitional phase of growth and development between childhood and adulthood between the ages of 10 and 19 years. This study pertained to a narrower age range of 15 to 17 years. However, the performance of the international recommendations in this cohort are similar to those observed in almost 5000 predominantly white (85%) athletes aged 14–35 years.45
This study was cross-sectional in nature. Further longitudinal assessment would also be required in those individuals with abnormal patterns of TWI in the context of a structurally normal heart to assess for the development of a subsequent cardiomyopathy.
Conclusions
Understanding of the adolescent athlete’s ECG is essential, given that this age group is deemed at highest risk of exercise-related SCD. Based on this large study where all athletes were investigated with ECG and echocardiography, the international recommendations13 demonstrated the highest specificity (98%) and lowest false-positive rate (1.9%) compared with the other three sets of criteria to interpret ECGs in sportspeople.
What are the findings?
This study is the largest to validate the new international recommendations for ECG interpretation in over 11 000 adolescent athletes (aged 15–17) including 1005 black athletes and 587 female athletes, all of whom were investigated with ECG and echocardiogram.
The international recommendations demonstrated the highest specificity for white (99%) and black (97%) athletes with a sensitivity of 86%, compared with previous recommendations from the European Society of Cardiology, Seattle criteria and refined criteria.
This study adds to limited data on sex and ethnic differences in the ECG of adolescent athletes who form the largest proportion of exercising individuals in society.
The international recommendations can be applied to white and black adolescent athletes to detect serious cardiac disease. They are more accurate than previous recommendations.
Acknowledgments
The authors are grateful to The Football Association Cardiology Consensus panel for their support.
References
Footnotes
Contributors AM designed data collection tools, monitored data collection, cleaned and analysed the data, and drafted and revised the manuscript. HD, SG and GF designed data collection tools and drafted the manuscript. T-JY, CM, TK-A and JB monitored data collection and analysed the data. PB, ZF, EP, BE, BG, GP-W and AD contributed to data collection. MT, EB and MP revised the manuscript. SS initiated the project, monitored data collection for the whole study, analysed the data, and drafted and revised the paper. SS is the guarantor.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The project was reviewed and approved from an ethics perspective by the St. George’s Joint Research and Enterprise Office.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement No data are available.