You are here

Article Text

Article menu

Download PDFPDF

Original article
Are there risk factors for snowboard injuries? A case-control multicentre study of 559 snowboarders
  1. Rebecca M Hasler1,
  2. Simeon Berov2,
  3. Lorin Benneker3,
  4. Simon Dubler1,
  5. Jonathan Spycher3,
  6. Dominik Heim4,
  7. Heinz Zimmermann1,
  8. Aristomenis K Exadaktylos1
  1. 1Department of Emergency Medicine, Inselspital, University of Bern, Bern, Switzerland
  2. 2Department of Orthopaedic Surgery, Spital Interlaken, Interlaken, Switzerland
  3. 3Department of Orthopaedic Surgery, Inselspital, University of Bern, Bern, Switzerland
  4. 4Department of Surgery, Spital Frutigen, Frutigen, Switzerland
  1. Correspondence to Dr Lorin Benneker, Department of Orthopaedic Surgery, Inselspital Bern, University Hospital, CH-3010 Bern, Switzerland; lorin.benneker{at}insel.ch

Abstract

Objective To analyse risk factors leading to injuries during snowboarding.

Design A case–control multicentre survey of injured and non-injured snowboarders.

Setting One tertiary and two secondary trauma centres in Bern, Switzerland.

Methods All snowboard injuries admitted to our tertiary and two affiliated secondary trauma centres from 1 November 2007 to 15 April 2008 were analysed on the basis of a completed questionnaire incorporating 15 variables. The same questionnaire was applied in non-injured controls at valley stations after a snowboarding day during the same period. A multiple logistic regression was performed (dichotomous variables). Patterns of combined risk factors were calculated by inference trees.

Results 306 patients and 253 controls were interviewed. The following variables were statistically significant for the injured patients: low readiness for speed (OR 0.20, 95% CI 0.06 to 0.64, p=0.0037), bad weather/visibility (OR 19.06, 95% CI 2.70 to 134.73, p=0.0031) and old snow (OR 0.11, 95% CI 0.02 to 0.68, p=0.0323). Not wearing a helmet and riding on icy slopes emerged as a combination of risk factors associated with injury.

Conclusions Several risk factors and combinations exist, and different risk profiles were identified. Future research should be aimed at more precise identification of groups at risk and developing specific recommendations for each group—for example, a snow-weather conditions index at valley stations.

https://doi.org/10.1136/bjsm.2010.071357

Statistics from Altmetric.com

    Request Permissions

    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.

    Introduction

    Since the 1998 Olympic Winter Games in Nagano, snowboarding has established itself as a popular winter sport for young people and adults.1 In Switzerland, there are about 600 000 snowboarders who suffer an average of more than 300 injuries per day. Boarders now progress to more advanced skills and speeds faster than ever before, mainly due to better prepared slopes and high-tech equipment. This has allowed even moderately trained sportsmen and sportswomen to race down the slopes like professionals. High-energy falls or collisions may occur when snowboarders overestimate their abilities, take risks or underestimate slope conditions. These may result in severe injuries, such as head and spine trauma, and can be associated with global functional impairment giving rise to long-term disability.2

    Although there are many reports in the literature on the epidemiology, injury patterns and biomechanics of snowboarding injuries, data on risk factors are very limited.2,–,6 Previous studies have examined only a few individual risk factors or risk factors related to specific injuries.7,–,18 The large number of snow sports-related injuries and the rising costs of healthcare prompted our institution to explore ways of cooperating with other trauma centres and major insurance companies to increase the efficiency of injury prevention measures by identifying risk factors in snowboarding.

    Several studies have documented that the frequency and severity of snowboarding injuries are linked to the degree of experience and speed.1 2 10 15 However, our impression from past years and from the results of a pilot study is that many snowboarders suffer injuries at slow or moderate speeds, and that the probability of snowboard injuries may be linked to a concatenation of several risk factors (unpublished data from internal survey).

    To attempt to substantiate our impressions, we conducted a case–control multicentre survey of 559 injured and non-injured snowboarders during the 2007/2008 season. Our objective was to define statistically significant risk factors and risk factor combinations in snowboarding.

    Patients and methods

    Setting and interventions

    This was a multicentre case–control study between 1 November 2007 and 15 April 2008. Injured patients were interviewed at the emergency department (ED) of a level I trauma centre (the only tertiary trauma unit with a catchment area of 1.8 million people serving one alpine region) and at the EDs of two level II trauma centres. The control group consisted of subjects interviewed at valley stations at six different high-volume ski resorts in the catchment area of the three hospitals. The method of enrolment was to approach all snowboarders as they exited the valley station at the end of a snowboarding day (controls) as well as injured snowboarders in the ED (patients) and to invite them to respond to a 20 min questionnaire comprising questions on 15 potential risk factors.

    Patients and controls

    The patient group included 306 injured snowboarders of any age, independent of snowboarding level and snowboarding experience, who were admitted to the trauma centre of one of the three hospitals. The control group included 253 non-injured snowboarders of any age, independent of snowboarding level and snowboarding experience. Participants were interviewed using a standardised questionnaire in the EDs (patients) or when coming off the slopes (controls). Questionnaires were available in English, French and German. If young patients (<12 years) were not able to answer the questions, accompanying parents or relatives were interviewed on their behalf. Patients heavily intoxicated or unconscious were excluded. Participation in the study was voluntary and anonymous; confidentiality was guaranteed.

    Data were collected, stored, analysed and shared according to the ethics committee standards of the Inselspital Bern.

    Study measures

    We defined 15 primary outcome measures as possible risk factors: these included participant characteristics (age, gender, years of experience in snowboarding), behavioural aspects (readiness for risk and speed, abstinence from alcohol or drugs while snowboarding, duration of warming up), equipment (use of protectors, age of the snowboard, seasonal checking of the snowboard) and external conditions (slope and snow conditions, weather and visibility, experience of aggressive behaviour by other skiers or snowboarders).

    Readiness for risk and speed were measured on a self-reported visual analogue scale (VASrisk, VASspeed) from 1 to 10 (1 for minimum speed or risk and 10 for maximum speed or risk). We included three different types of protectors: (A) helmets, (B) back protectors and (C) wrist protectors.

    Seasonal checking of the snowboard was defined as a check by a sports store specialist in the current winter season.

    We included three types of slope conditions assessed by the individual snowboarder: (A) hard and icy, (B) soft and powdery and (C) slushy; and three types of snow conditions, also individually assessed: (A) fresh snow, (B) old snow and (C) artificial snow. Weather and visibility while snowboarding were classified as follows: (A) sunny weather and good visibility, or (B) cloudy weather and bad visibility. Aggressive behaviour on slopes included individually reported experiences of: (A) being deliberately kicked by another person, (B) being pressured by another person and (C) having witnessed an argument. See also appendix I.

    Statistical analysis

    Patients and controls are compared for the distribution of the covariates asked in the questionnaire. Crosstables are reported as counts and percentages within study groups. Ordinal variables are reported as median and IQR. Metric variables are reported as mean and SD or as median and IQR as appropriate.

    The influence of the risk factors was assessed using multiple logistic regression with OR and 95% CI. The level of significance was set to p<0.05. All calculations were made with R version 2.7.0 (a language and environment for statistical computing) (appendix II). For continuous or ordinal variables (scores) OR have to be based on a meaningful difference of the predicting variable. They were expressed as the ratio of the odds from the third to the first quartile based on cases and controls, representing a typical above average to a typical below average value.

    Conditional inference tree analysis was applied to elicit possible combinations of two or more risk factors. A tree-based model is a good exploratory tool to approximate a complex model. The advantages are clearness of interpretation and visualisation of complex interactions which are not covered by regression modelling. The disadvantage is the possible instability using strong correlated predictors and the splitting of continuous variables into classes. Details are described in appendices III and IV.

    Results

    Demographics

    Three hundred and six patients and 253 controls were included between 1 November 2007 and 15 April 2008. The mean ages were 20 years (range 9–68 years) in the patient group and 19 years (range 10–53 years) in the control group. Thirty-seven per cent (n=114) of the patients were female and 63% (n=192) male; 49% (n=124) of the controls were female and 51% (n=130) male. The mean snowboarding experience was 5 years (range 0–30 years) in the patient group and 7 years (range 0–30 years) in the control group. Nine patients and one control (<12 years) were interviewed via proxy respondents. Response rates were 90% for controls and 96% for patients.

    Types of injury

    Table 1 shows the body sites of injuries sustained by the patient group.

    View this table:
    Table 1

    Body sites of injuries

    Individual risk factors

    Table 2 shows the results for the individual categorical, ordinal and metric variables. More controls than patients wore protectors. More patients than controls had been boarding on icy snow and old snow, and more controls than patients on powder snow and fresh snow. More controls than patients had come across aggressive behaviour and offensive boarding styles on the slopes, and more patients had a defensive boarding style. The mean assessment of speed was higher for the controls than for the patients.

    View this table:
    Table 2

    Individual risk factors

    Multiple logistic regression analysis

    Significant relationships between injury and the following factors were found: low readiness for speed (OR 0.20, 95% CI 0.06 to 0.64; p=0.0037), bad weather/visibility (OR 19.06, 95% CI 2.70 to 134.73; p=0.0031) and old snow (OR 0.11, 95% CI 0.02 to 0.68; p=0.0323). A trend to an association with injury (within the 90% CI) was observed for the following factors: not experiencing aggressive behaviour on slopes, not wearing a helmet and consumption of drugs (table 3).

    View this table:
    Table 3

    Statistically significant and non-significant variables in the patient group

    Conditional inference trees

    Not wearing a helmet and riding on icy slopes emerged as a ‘typical’ combination leading to injury in snowboarders. Among the variables included, riding on icy slopes had the greatest influence on variability (figure 1).

    Figure 1
    Figure 1

    Example of a conditional inference tree used to postulate a specific group at risk by a combination of the most significant variables. Explanatory information for the interpretation of conditional inference trees is given in appendix IV.

    Discussion

    Many authors have speculated that the incidence of serious injuries in snow sports has changed over the past 10 years,1 13 but little research has been done into the risk factors that lead to such injuries. Our impression from our experience and an internal pilot study was that serious injuries also occur at slow or moderate speeds and that ski and snowboarding injuries are likely to be of multifactorial origin. We felt that the present case–control study in more than 500 injured and non-injured snowboarders would assist in clarifying this.19

    We identified several individual factors that showed an association with injury. However, significance emerged when the factors were analysed in isolation, disregarding the pattern observed for the other factors in both the patient and control group. A multiple regression analysis was therefore applied. It revealed the following potential risk factors: low readiness for speed, bad weather/visibility and boarding on old snow. This did not take into account, however, that a combination of factors might show an association with injury, and we therefore also applied conditional inference trees to our findings, which identified a ‘typical’ combination of factors in snowboarders that is more likely to be associated with injury: not wearing a helmet and riding on icy slopes.

    We found no strong association between injury and age, gender or years of experience with snowboarding. Generally, the probability of snow sports injury is associated with increased speeds.1,–,15 The distributions of ‘readiness for speed’ in our two groups overlapped substantially, in contrast with our finding of a highly statistically significant OR. Low readiness for speed was significantly more frequent in the patient group when a multiple logistic regression was applied. This complies with our impression from our ED that many patients suffer injuries at low or moderate speeds, and it also complies with our findings on risk factors in alpine skiing injuries, where lower readiness for speed was associated more often with injuries.19

    Many of the snowboarders in our sample felt intimidated by others on the slopes, but we did not detect any relationship between antisocial and aggressive behaviour of others and injury. As for road traffic accidents, where only a minority of injuries are caused by aggressive or antisocial drivers, or speeding, injuries while snowboarding seem to be attributable to other factors, including the ability to gauge slope conditions and adapt the speed of the descent accordingly. This should be one of the focus issues in snowboard training courses.

    As would be expected, snow and slope conditions have been reported to play a major role in the likelihood of injury.1 13 Although bad weather conditions are regarded as a potential risk factor, we were unable to find any publications showing a significant association between weather conditions and snowboard injuries.13 Our analysis in this study and in our previous study on risk factors in alpine skiing showed, however, that bad weather and visibility were reported more frequently in the patient group.19 Snowboarders should be made aware of boarding conditions. Snow, slope and weather conditions should be graded, and snowboarders should be informed at valley stations, and possibly even at selected points on the slopes, if conditions are changing. The development of a simple ‘snow quality-weather-injury risk’ grading would be highly desirable, as we also recommended for alpine skiers.19

    Alcohol consumption has become a part of the skiing and snowboarding culture. This and drug consumption have traditionally been regarded as risk factors for skiing and snowboarding injuries.16 17 This was, however, not consistent with the findings of our survey. Either alcohol consumption has become more moderate than previously assumed or our respondents did not answer this question honestly.13 An effect of alcohol consumption might have been more obvious if we had included heavily intoxicated patients. Another possible explanation is that our study was performed at the same time as new, more restrictive drinking-and-driving legislation was implemented, which proved to have positive effects on the drinking habits of a large proportion of the population.

    Although we saw no association between alcohol consumption and injury, we did observe a trend towards an association between drug consumption and injury (OR 17.54). In Switzerland, soft drug consumption (mainly cannabis) is not heavily sanctioned by law, is generally well tolerated by society and is very popular among young people. According to a survey of the Swiss Federal Health Department, Switzerland, the UK and USA have the highest cannabis consumption rates worldwide. About one-fifth of all 15- to 64-year-olds have used cannabis at least once in their life, and 25% profess to be regular consumers.20 Despite the results of our study and the tolerance towards soft drugs, it is likely that efforts to minimise alcohol and drug consumption when snowboarding might result in fewer injuries, as both lower the thresholds to risk and overestimation of one's personal abilities.

    Applying conditional inference trees to our findings showed that snowboarders without helmets on icy slopes might be regarded as a ‘typical’ group of snowboarders who are more likely to suffer injury. Compared with our findings in our previous study on alpine skiers, where we detected several different groups of typically injured skiers, snowboarders seem to behave more homogenously and we were only able to detect one group at risk. Not wearing a helmet is evidence of carelessness or bravado, so even this ‘typical’ group may consist of at least two different subtypes of snowboarders: careless, although perhaps good boarders who inadvertently find themselves on a particularly icy slope, or reckless boarders who deliberately seek out the icy slopes as a challenge. Whichever type, the members of this group are in particular need of information about safety aspects of snowboarding.

    Limitations

    Patients and controls were not interviewed consecutively, and participation was voluntary. Furthermore, heavily intoxicated or unconscious patients were excluded, which introduced some selection bias. This selection bias probably made alcohol and drug consumption appear more ‘protective’ against injury, because heavily intoxicated patients were not interviewed, while controls with only mild consumption of alcohol or drugs were able to participate in the interview after a snowboarding day.

    What is already known on this topic

    Although there are many reports in the literature describing the epidemiology and biomechanics of snowboard injuries, data on risk factors are very limited. Previous studies used the conventional approach of a matched patient-control design and looked at only a few single risk factors or risk factors related to specific injuries.

    What this study adds

    The conditional inference trees analysis has not been used before in such a study. Not only one but several risk factors and their combinations were identified: low speed, bad weather/visibility and boarding on old snow. A trend to an association with injury was observed for the variables not wearing a helmet and consumption of drugs.

    Patients were interviewed in one of the EDs after injury, and controls were approached at valley stations at the end of a snowboarding day. This method of enrolment was chosen for practical reasons, and the influence of the different sampling procedures in the study groups cannot be estimated. There is the possibility of greater variability in the patient group leading to wider CIs and more conservative conclusions. After an injury, patients may overestimate or underestimate their readiness for risk and speed leading to recall bias. The extent and effect of this kind of bias cannot be estimated. Some questions, like snow and slope conditions, were answered by self-estimation which can also cause information bias. Reporting bias can be expected in questions concerning alcohol and drug consumption. The VAS for the readiness for speed and risk has not been validated, and this may limit its value. Nevertheless, VAS is a widely used tool in clinical medicine and provided a good basis for assessments in this study. In general, VAS investigations have been validated for emotions and ‘feelings’ in the past, and the readiness for speed and to take risk is certainly influenced by feelings.21

    Conclusion

    Our study showed that low readiness for speed, bad weather/visibility and old snow might be risk factors for injury when snowboarding. Trends to an association with injury were observed for not wearing a helmet and consumption of drugs. It is likely that multiple factors are responsible for injuries, and we identified non-helmet wearers on icy slops as at risk. Future research should be aimed at identifying risk groups and developing recommendations or educational intervention programmes specific to these groups. In general, snowboarders should be made more aware of snow and weather conditions at valley stations. Training should include as a focus the ability to gauge slope conditions and the importance of protective equipment and the avoidance of drugs when snowboarding should also be stressed.

    Acknowledgments

    K Dopke, MPH, study coordinator, Department of Emergency Medicine, Inselspital, University Hospital Bern, A Schötzau, Dipl. Math, Schötzau und Simmen, Statistical Consulting, Basel.

    Appendix I. Fifteen primary outcome measure

    Patient and control characteristics

    1. (1). Age: years

    2. (2). Gender: F/M

    3. (3). Experience in snowboarding: years

    Behavioural aspects

    1. (4). Readiness for risk: VAS 1–10 (1 minimum risk; 10 maximum risk)

    2. (5). Readiness for speed: VAS 1–10 (1 minimum speed; 10 maximum speed)

    3. (6). Abstinence from alcohol while snowboarding: yes/no

    4. (7). Abstinence from drugs while snowboarding: yes/no

    5. (8). Duration of warm-up before the first ride: minutes

    Equipment

    1. Use of protectors:

      1. (a). Helmet

      2. (b). Spine protector

      3. (c). Wrist protector

    2. (10). Age of snowboard: Years

    3. (11). Seasonal checking of snowboard by a sports store specialist in the current winter season: yes/no

    External conditions

    1. Slope conditions:

      1. (a). Hard/icy

      2. (b). Soft/powdery

      3. (c). Slushy

    2. Snow conditions:

      1. (a). Fresh snow

      2. (b). Old snow

      3. (c). Artificial snow

    3. Weather and visibility:

      1. (a). Sunny weather and good visibility

      2. (b). Cloudy weather and bad visibility

    4. Experience of aggressive behaviour of other snowboarders:

      1. (a). Being voluntarily kicked by another person

      2. (b). Being pressured by another person

      3. (c). Having witnessed an argument

    Appendix II

    R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2008. ISBN 3-900051-07-0. http://www.R-project.org.

    Appendix III

    1. Hothorn T,
    2. Hornik K,
    3. Zeileis A
    . Unbiased recursive partitioning: a conditional inference framework. J Comput Graph Statist 2006;15:651–74.
    OpenUrlCrossRef

    Appendix IV

      1. Richard A
      . Chapter 3 Classification and Regression Trees (CART). In: Statistical Learning from a Regression Perspective. Springer Series in Statistics Berk, 2008:103, XVIII. Hardcover. ISBN: 978-0-387-77500-5. 360 p.
    3. Original research on inference trees. http://statmath.wu.ac.at/~zeileis/papers/Hothorn+Hornik+Zeileis-2006.pdf. Hothorn T, Hornik K, Zeileis A. Unbiased recursive partitioning: a conditional inference framework. J Comput Graph Statist 2006;15:651–74.

    References

    1. ↵
      1. Dann K,
      2. Kristen KH,
      3. Knoeringer M,
      4. et al
      . Snowboarding. History–injuries–risks–new materials–tournament on-site services–prevention. Orthopade 2005;34:433–40.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Franz T,
      2. Hasler RM,
      3. Benneker L,
      4. et al
      . Severe spinal injuries in alpine skiing and snowboarding: a 6-year review of a tertiary trauma centre for the Bernese Alps ski resorts, Switzerland. Br J Sports Med 2008;42:55–8.
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Prall JA,
      2. Winston KR,
      3. Brennan R
      . Severe snowboarding injuries. Injury 1995;26:539–42.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Chissell HR,
      2. Feagin JA Jr.,
      3. Warme WJ,
      4. et al
      . Trends in ski and snowboard injuries. Sports Med 1996;22:141–5.
      OpenUrlPubMedWeb of Science
    5. ↵
      1. Pigozzi F,
      2. Santori N,
      3. Di Salvo V,
      4. et al
      . Snowboard traumatology: an epidemiological study. Orthopedics 1997;20:505–9.
      OpenUrlPubMedWeb of Science
    6. ↵
      1. Hagel B
      . Skiing and snowboarding injuries. Med Sport Sci 2005;48:74–119.
      OpenUrlPubMed
    7. ↵
      1. Mueller BA,
      2. Cummings P,
      3. Rivara FP,
      4. et al
      . Injuries of the head, face, and neck in relation to ski helmet use. Epidemiology 2008;19:270–6.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Scott MD,
      2. Buller DB,
      3. Andersen PA,
      4. et al
      . Testing the risk compensation hypothesis for safety helmets in alpine skiing and snowboarding. Inj Prev 2007;13:173–7.
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Hagel B,
      2. Pless IB,
      3. Goulet C,
      4. et al
      . The effect of helmet use on injury severity and crash circumstances in skiers and snowboarders. Accid Anal Prev 2005;37:103–8.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Langran M,
      2. Selvaraj S
      . Increased injury risk among first-day skiers, snowboarders, and skiboarders. Am J Sports Med 2004;32:96–103.
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Williams R,
      2. Delaney T,
      3. Nelson E,
      4. et al
      . Speeds associated with skiing and snowboarding. Wilderness Environ Med 2007;18:102–5.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Chamarro A,
      2. Fernández-Castro J
      . The perception of causes of accidents in mountain sports: a study based on the experiences of victims. Accid Anal Prev 2009;41:197–201.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. McBeth PB,
      2. Ball CG,
      3. Mulloy RH,
      4. et al
      . Alpine ski and snowboarding traumatic injuries: incidence, injury patterns, and risk factors for 10 years. Am J Surg 2009;197:560–3; discussion 563–4.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Genelin A,
      2. Kathrein A,
      3. Daniaux A,
      4. et al
      . Current status of spinal injuries in winter sports. Schweiz Z Med Traumatol 1994:17–20.
    15. ↵
      1. Westphal K
      . [Carving ski and snowboard: trauma trend in a trendy sport. Fast and with sharp edge into the clinic]. MMW Fortschr Med 2007;149:10–12.
      OpenUrlPubMed
    16. ↵
      1. Barnas C,
      2. Miller CH,
      3. Sperner G,
      4. et al
      . The effects of alcohol and benzodiazepines on the severity of ski accidents. Acta Psychiatr Scand 1992;86:296–300.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Sherker S,
      2. Finch C,
      3. Kehoe EJ,
      4. et al
      . Drunk, drowsy, doped: skiers' and snowboarders' injury risk perceptions regarding alcohol, fatigue and recreational drug use. Int J Inj Contr Saf Promot 2006;13:151–7.
      OpenUrlCrossRefPubMed
    18. ↵
      1. Sulheim S,
      2. Holme I,
      3. Ekeland A,
      4. et al
      . Helmet use and risk of head injuries in alpine skiers and snowboarders. JAMA 2006;295:919–24.
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Hasler RM,
      2. Dubler S,
      3. Benneker LM,
      4. et al
      . Are there risk factors in alpine skiing? A controlled multicentre survey of 1278 skiers. Br J Sports Med 2009;43:1020–5.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      Federal Department of Foreign Affairs, General Secretariat, Presence Switzerland. Soft drugs, 2010. http://www.swissworld.org/de/bevoelkerung/alkohol_und_drogen/weiche_drogen/ (accessed 26/7/2010).
    21. ↵
      1. Aitken RC
      . Measurement of feelings using visual analogue scales. Proc R Soc Med 1969;62:989–93.
      OpenUrlPubMedWeb of Science
    View Abstract

    Footnotes

    • Competing interests None.

    • Provenance and peer review Not commissioned; externally peer reviewed.