It remains a difficult medical challenge to prevent the sudden cardiac death of athletes, typically defined as natural, unexpected death from cardiac arrest within one hour of the onset of collapse symptoms, excluding additional time on mechanical life support. (Wider definitions of sudden death are also in use, but not usually applied to the athletic situation.) Most causes relate to congenital or acquired cardiovascular disease with no symptoms noted before the fatal event. The prevalence of any single, associated condition is low, probably less than 0.3% of the population in the athletes’ age group, and the sensitivity and specificity of common screening tests leave much to be desired. The single most important predictor is fainting or near-fainting during exercise, which should require detailed explanation and investigation. The victims include many well-known names, especially in professional soccer, and close relatives are often at risk for similar cardiac problems.
Text under construction
The sudden cardiac deaths of 387 young American athletes (under age 35) were analyzed in a 2003 medical review:
|Hypertrophic cardiomyopathy||26%||Genetically determined|
|Commotio cordis||20%||Structurally normal heart, disrupted electrically by a blow to the chest|
|Coronary artery anomalies||14%||Exact mechanisms unknown; some association with other congenital CVS abnormalities|
|Left ventricular hypertrophy of undetermined origin||7%||Probable variant of hypertrophic cardiomyopathy|
|Ruptured aortic aneurysm (Marfan syndrome)||3%||Genetically determined; also associated with unusual height|
|Arrhythmogenic right ventricular cardiomyopathy||3%||Genetically determined|
|Tunneled coronary artery||3%||Congenital abnormality|
|Aortic valve stenosis||3%||Multiple causes|
|Atherosclerotic coronary artery disease||3%||Mainly acquired; dominant cause in older adults|
While most causes of sudden cardiac death relate to congenital or acquired cardiovascular disease, an exception is commotio cordis, in which the heart is structurally normal but a potentially fatal loss of rhythm occurs because of the accident of timing of a blow to the chest. Its fatality rate is about 65% even with prompt CPR and defibrillation, and more than 80% without.
Age 35 serves as an approximate borderline for the likely cause of sudden cardiac death. Before age 35, congenital abnormalities of the heart and blood vessels predominate. These are usually asymptomatic prior to the fatal event, although not invariably so. Congenital cardiovascular deaths are reported to occur disproportionately in African-American athletes.
After age 35, acquired coronary artery disease predominates (80%), and this is true regardless of the athlete’s former level of fitness.
Cardiomyopathies are generally inherited as autosomal dominants, although recessive forms have been described, and dilated cardiomyopathy can also be inherited in an X-linked pattern. Consequently, in addition to tragedy involving an athlete who succumbs, there are medical implications for close relatives. Among family members of index cases, more than 300 causative mutations have been identified. However, not all mutations have the same potential for severe outcomes, and there is not yet a clear understanding of how these mutations (which affect the same myosin protein molecule) can lead to the dramatically different clinical characteristics and outcomes associated with hypertrophic cardiomyopathy (HCM) and dilated cardiomyopathy (DCM).
Since HCM, as an example, is typically an autosomal dominant trait, each child of an HCM parent has a 50% chance of inheriting the mutation. In individuals without a family history, the most common cause of the disease is a “de novo” mutation of the gene that produces the β-myosin heavy chain.
Sudden cardiac death can usually be attributed to cardiovascular disease or commotio cordis, but about 20% of cases show no obvious cause and remain undiagnosed after autopsy. Interest in these “autopsy-negative” deaths has centered around the “ion channelopathies”. These electrolyte channels are pores regulating the movement of sodium, potassium and calcium ions into cardiac cells, collectively responsible for creating and controlling the electrical signals that govern the heart’s rhythm. Abnormalities in this system occur in relatively rare genetic diseases such as Long QT syndrome, Brugada syndrome, and Catecholaminergic polymorphic ventricular tachycardia, all associated with sudden death. Consequently, autopsy-negative sudden cardiac deaths (no physical abnormalities identified) may comprise a larger part of the channelopathies than previously anticipated.
Heritable connective tissue diseases are rare, each disorder estimated at one to ten per 100,000, of which Marfan syndrome is the most common. It is carried by the FBN1 gene on chromosome 15, which encodes the connective protein fibrillin-1, inherited as a dominant trait. This protein is essential for synthesis and maintenance of elastic fibers. Since these fibers are particularly abundant in the aorta, ligaments, and the ciliary zonules of the eye, these areas are among the worst affected. Everyone has a pair of FBN1 genes and, because transmission is dominant, those who have inherited one affected FBN1 gene from either parent will have Marfan syndrome. Although it is most frequently inherited as an autosomal dominant, there is no family history in 25% of cases.
Recruiting practices aimed at attracting athletes who are unusually tall or who have an unusually wide arm span (characteristics of Marfan syndrome) can increase the prevalence of the syndrome within sports such as basketball and volleyball.
After a disease-causing mutation has been identified in an index case (which is not always accomplished conclusively), the main task is genetic identification of carriers within a pedigree, a sequential process known as “cascade testing”. Family members with the same mutation may show different severities of disease, a phenomenon known as “variable penetrance”. As a result, some may remain asymptomatic, with little lifelong evidence of disease. Nevertheless, their children remain at risk of inheriting the disorder and potentially being more severely affected.
Screening athletes for cardiac disease can be problematic because of low prevalence and inaccurate performance of various tests that have been used. Nevertheless, sudden death among seemingly healthy individuals attracts much public and legislator attention because of its visible and tragic nature.
As an example, the Texas Legislature appropriated US$1 million for a pilot study of statewide athlete screening in 2007. The study employed a combination of questionnaire, examination and electrocardiography for 2,506 student athletes, followed by echocardiography for 2,051 of them, including any students with abnormal findings from the first three steps. The questionnaire alone flagged 35% of the students as potentially at risk, but there were many false positive results, with actual disease being confirmed in less than 2%. Further, a substantial number of screen-positive students declined repeated recommendations for follow-up evaluation. (Individuals who are conclusively diagnosed with cardiac disease are usually told to avoid competitive sports.) It should be stressed that this was a single pilot program, but it was indicative of the problems associated with large-scale screening, and consistent with experience in other locations with low prevalence of sudden death in athletes.
Sudden cardiac death occurs in approximately one per 200,000 young athletes per year, usually triggered during competition or practice. The victim is usually male and associated with soccer, basketball, ice hockey, or American football, reflecting the large number of athletes participating in these sustained and strenuous sports. For a normally healthy age group, the risk appears to be particularly magnified in competitive basketball, with sudden cardiac death rates as high as one per 3,000 annually for male basketball players in NCAA Division I. This is still far below the rate for the general population, estimated as one per 1,300–1,600 and dominated by the elderly. However, a population as large as the United States will experience the sudden cardiac death of a competitive athlete at the average rate of one every three days, often with significant local media coverage heightening public attention.
These athletes, in alphabetical order, experienced sudden cardiac death by age 40. Their notability is established by reliable sources in other Wikipedia articles.
- Mohamed Abdelwahab, 22 (2006), soccer
- Gaines Adams, 26 (2010), Amer. football
- Jaouad Akaddar, 28 (2012), soccer
- Davide Astori, 31 (2018), soccer
- Víctor Hugo Ávalos, 37 (2009), soccer
- Heath Benedict, 24 (2008), Amer. football
- Hédi Berkhissa, 24 (1997), soccer
- Viktor Blinov, 22 (1968), ice hockey
- Gilbert Bulawan, 29 (2016), basketball
- J. V. Cain, 28 (1979), Amer. football
- Sékou Camara, 27 (2013), soccer
- Alexei Cherepanov, 19 (2008), ice hockey
- Mitchell Cole, 27 (2012), soccer
- Jason Collier, 28 (2005), basketball
- Hugo Cunha, 28 (2005), soccer
- Renato Curi, 24 (1977), soccer
- Alexander Dale Oen, 26 (2012), swimming
- Shane del Rosario, 30 (2013), MMA
- Ben Idrissa Dermé, 34 (2016), soccer
- Lyle Downs, 24 (1921), Austral. football
- Patrick Ekeng, 26 (2016), soccer
- Bobsam Elejiko, 30 (2011), soccer
- Derrick Faison, 36 (2004), Amer. football
- Sebastian Faisst, 20 (2009), handball
- Miklós Fehér, 24 (2004), soccer
- Neil Fingleton, 36 (2017), basketball
- Marc-Vivien Foé, 28 (2003), soccer
- Matt Gadsby, 27 (2006), soccer
- Hank Gathers, 23 (1990), basketball
- Cristian Gómez, 27 (2015), soccer
- Michael Goolaerts, 23 (2018), cycling
- Larry Gordon, 28 (1983), Amer. football
- Herb Gorman, 28 (1953), baseball
- Rasmus Green, 26 (2006), soccer
- Sergei Grinkov, 28 (1995), figure skating
- Eddie Guerrero, 38 (2005), wrestling
- Frank Hayes, 35 (1923), horse racing
- Thomas Herrion, 23 (2005), Amer. football
- Cătălin Hîldan, 24 (2000), soccer
- Chuck Hughes, 28 (1971), Amer. football
- Flo Hyman, 31 (1986), volleyball
- Endurance Idahor, 25 (2010), soccer
- Robbie James, 40 (1998), soccer
- Daniel Jarque, 26 (2009), soccer
- Cristiano Júnior, 25 (2004), soccer
- Joe Kennedy, 28 (2007), baseball
- Darryl Kile, 33 (2002), baseball
- John Kirkby, 23 (1953), soccer
- Michael Klein, 33 (1993), soccer
- Wayne Larkin, 29 (1968), ice hockey
- Reggie Lewis, 27 (1993), basketball
- José Lima, 37 (2010), baseball
- David Longhurst, 25 (1990), soccer
- Nikola Mantov, 23 (1973), soccer
- Pete Maravich, 40 (1988), basketball
- Alex Marques, 20 (2013), soccer
- Jesse Marunde, 27 (2007), weightlifting
- Stan Mauldin, 27 (1948), Amer. football
- Conrad McRae, 29 (2000), basketball
- Fab Melo, 26 (2017), basketball
- Nilton Pereira Mendes, 30 (2006), soccer
- Igor Misko, 23 (2010), ice hockey
- Stéphane Morin, 29 (1998), ice hockey
- Piermario Morosini, 25 (2012), soccer
- Carl Morton, 39 (1983), baseball
- Damien Nash, 24 (2007), Amer. football
- Frederiek Nolf, 21 (2009), cycling
- Chaswe Nsofwa, 28 (2007), soccer
- Gábor Ocskay, 33 (2009), ice hockey
- Phil O’Donnell, 35 (2007), soccer
- Samuel Okwaraji, 25 (1989), soccer
- David Oniya, 30 (2015), soccer
- Alen Pamić, 23 (2013), soccer
- Pavão, 26 (1973), soccer
- Bruno Pezzey, 39 (1994), soccer
- Pheidippides, c. 40 (490 BC), marathon
- Petar Radaković, 29 (1966), soccer
- Mickey Renaud, 19 (2008), ice hockey
- Bernardo Ribeiro, 26 (2016), soccer
- Darcy Robinson, 26 (2007), ice hockey
- Brad Rone, 34 (2003), boxing
- Omar Sahnoun, 24 (1980), soccer
- Serginho, 30 (2004), soccer
- Ryan Shay, 28 (2007), marathon
- Dave Sparks, 26 (1954), Amer. football
- Cheick Tioté, 30 (2017), soccer
- Robert Traylor, 34 (2011), basketball
- Zeke Upshaw, 26 (2018), basketball
- Ginty Vrede, 22 (2008), kickboxing
- Frank Warfield, 35 (1932), baseball
- Chandler Williams, 27 (2013), Amer. football
- David “Soldier” Wilson, 23 (1906), soccer
- Sergejs Žoltoks, 31 (2004), ice hockey
- Cormac McAnallen, 24 (2004) GAA
- Cardiac Risk in the Young (UK charity)
- Lists of sportspeople who died during their careers
- van der Werf C, van Langen IM, Wilde AA (February 2010). “Sudden death in the young: what do we know about it and how to prevent?”. Circ Arrhythmia Electrophysiol. 3 (1): 96–104. doi:10.1161/CIRCEP.109.877142. PMID 20160177.
- Hastings JL, Levine BD (March 2012). “Syncope in the athletic patient”. Prog Cardiovasc Dis. 54 (5): 438–44. doi:10.1016/j.pcad.2012.02.003. PMID 22386295.
- Maron, Barry J. (September 11, 2003). “Sudden Death in Young Athletes”. New England Journal of Medicine. 349 (11): 1064–1075. doi:10.1056/NEJMra022783. PMID 12968091.
- Maron, BJ; Estes, NAM III (March 2010). “Commotio cordis”. New England Journal of Medicine. 362 (10): 917–927. doi:10.1056/NEJMra0910111. PMID 20220186.
- “Position Statement on Commotio Cordis”. US Lacrosse. January 2008. Retrieved 22 February 2017.
- Ferreira M, Santos-Silva PR, de Abreu LC, Valenti VE, Crispim V, Imaizumi C, Filho CF, Murad N, Meneghini A, Riera AR, de Carvalho TD, Vanderlei LC, Valenti EE, Cisternas JR, Moura Filho OF, Ferreira C (Aug 3, 2010). “Sudden cardiac death athletes: a systematic review”. Sports Med Arthrosc Rehabil Ther Technol. 2: 19. doi:10.1186/1758-2555-2-19. PMC 2923123. PMID 20682064.
- Maron BJ, Carney KP, Lever HM, Lewis JF, Barac I, Casey SA, Sherrid MV (March 2003). “Relationship of race to sudden cardiac death in competitive athletes with hypertrophic cardiomyopathy”. Journal of the American College of Cardiology. 41 (6): 974–980. doi:10.1016/S0735-1097(02)02976-5.
- Moore JR, Leinwand L, Warshaw DM (Jul 20, 2012). “Understanding cardiomyopathy phenotypes based on the functional impact of mutations in the myosin motor”. Circ Res. 111 (3): 375–85. doi:10.1161/CIRCRESAHA.110.223842. PMC 3947556. PMID 22821910.
- Westfal RE, Reissman S, Doering G (Jul 1996). “Out-of-hospital cardiac arrests: an 8-year New York City experience”. Am J Emerg Med. 14 (4): 364–8. doi:10.1016/S0735-6757(96)90050-9. PMID 8768156.
- de Vreede-Swagemakers JJ, Gorgels AP, Dubois-Arbouw WI, van Ree JW, Daemen MJ, Houben LG, Wellens HJ (Nov 1997). “Out-of-hospital cardiac arrest in the 1990s: a population-based study in the Maastricht area on incidence, characteristics and survival”. J Am Coll Cardiol. 30 (6): 1500–5. doi:10.1016/s0735-1097(97)00355-0. PMID 9362408.
- Kainulainen K, Karttunen L, Puhakka L, Sakai L, Peltonen L (January 1994). “Mutations in the fibrillin gene responsible for dominant ectopia lentis and neonatal Marfan syndrome”. Nat. Genet. 6 (1): 64–9. doi:10.1038/ng0194-64. PMID 8136837.
- Dietz HC, Loeys B, Carta L, Ramirez F (November 2005). “Recent progress towards a molecular understanding of Marfan syndrome”. Am J Med Genet C Semin Med Genet. 139C (1): 4–9. doi:10.1002/ajmg.c.30068. PMID 16273535.
- Armon K, Bale P (June 2012). “Identifying heritable connective tissue disorders in childhood”. Practitioner. 256 (1752): 19–23, 2–3. PMID 22916581.
- Raju H, Alberg C, Sagoo GS, Burton H, Behr ER (Nov 21, 2011). “Inherited cardiomyopathies”. BMJ. 343: d6966. doi:10.1136/bmj.d6966. PMID 22106372.
- Zeltser I, Cannon B, Silvana L, Fenrich A, George J, Schleifer J, Garcia M, Barnes A, Rivenes S, Patt H, Rodgers G, Scott W (Jun 15, 2012). “Lessons learned from preparticipation cardiovascular screening in a state funded program”. Am J Cardiol. 110 (6): 902–8. doi:10.1016/j.amjcard.2012.05.018. PMID 22704711.
- Harmon KG, Asif IM, Klossner D, Drezner JA (April 2011). “Incidence of sudden cardiac death in National Collegiate Athletic Association athletes”. Circulation. 123(15): 1594–1600. doi:10.1161/CIRCULATIONAHA.110.004622. PMID 21464047.
- Chugh SS, Reinier K, Teodorescu C, Evanado A, Kehr E, Al Samara M, Mariani R, Gunson K, Jui J (Nov–Dec 2008). “Epidemiology of sudden cardiac death: clinical and research implications”. Prog Cardiovasc Dis. 51 (3): 213–28. doi:10.1016/j.pcad.2008.06.003. PMC 2621010. PMID 19026856.
For the world (total population approx. 6,540,000,000), the estimated annual burden of sudden cardiac death would be in the range of 4–5 million cases per year.
- Link, MS; Estes, NAM III (May 2012). “Sudden Cardiac Death in the Athlete”. Circulation. 125 (20): 2511–2516. doi:10.1161/CIRCULATIONAHA.111.023861.
Sudden cardiac death (SCD) in athletes is always shocking, tragic, and stirs controversy regarding screening for pre-existing cardiovascular pathology. Could these deaths be prevented? What type of screening is required? To answer these questions, a better understanding of the causes of SCD in athletes is required. Traditionally, in the United States, hypertrophic cardiomyopathy (HCM) has been reported as the leading cardiovascular cause of death in athletes.1 However, more recent research and research from other countries has questioned whether sudden arrhythmic death syndrome (SADS) (death with a structurally and histologically normal heart, presumed to be electrical disease) may represent a larger proportion of deaths than previously thought. In addition, as autopsies are refined, there is an increasing proportion of hearts which show left ventricular hypertrophy (LVH) with or without myocardial fibrosis that do not meet the definition of HCM, prompting the question whether acquired LVH could be pathologic in some cases or whether this is an early expression of cardiomyopathy. Further, it seems death from cardiomyopathy may be more frequent in certain sports or races. A more nuanced look at the causes of SCD may offer new insights.
The cause of SCD is highly dependent on age, with electrical and structural disease more prevalent in populations <25 years of age and coronary artery disease (CAD) more common in athletes over the age of 35. In fact, CAD represents <10% of deaths in high school and college athletes and >85% of sport-related death in older athletes.1-4 Cardiovascular screening for congenital cardiac pathology versus acquired CAD should be examined separately; therefore, this discussion will focus on athletes aged 14-35.
While the cause of death may be related to age, the determination of cause is often dependent on the expertise and resources available at post-mortem examination. In the United States, a 2009 National Academy of Sciences Blue Ribbon Report concluded “coroners and medical examiner offices are struggling with inadequate resources, poor scientific training and substandard facilities and technology.”5 The charge of most coroners and medical examiners is to determine under what circumstances a death has occurred: “natural,” homicide or suicide. An exact medical cause of death, while important to the family for both peace of mind and for screening of surviving relatives, is not a priority and has possibly contributed to misperceptions.
It is against this backdrop that HCM was initially reported as the leading cause of SCD in US athletes based on a large registry. However, the determination of HCM was established primarily based on local autopsies, and the registry was based out of a HCM center utilizing passive surveillance for case identification. This report suggested up to 36% of SCD cases were due to HCM although the athletes with suspected cardiovascular events but no precise diagnosis of death (i.e., possible SADS) were dropped from the pool, thereby increasing the proportion of cases attributed to HCM. If the cardiac cases with unknown causes were included, HCM instead accounted for 24% of deaths in the cohort.1
In contrast, a study of unselected NCAA athletes reported a much larger proportion of SCD attributed to SADS with strictly defined HCM accounting for only 8% of deaths. Interestingly, if LVH with fibrosis or cardiomyopathies not meeting the definition of HCM were included, this number increased to 24% of deaths.2 Although this study had less risk of selection bias, it also relied on local post-mortem examinations of variable quality.
In the United Kingdom a referral center for cardiac pathology examines a large proportion of sudden deaths in athletes and reported 44% of the deaths in 18-35-year-old athletes were due to SADS.6 HCM accounted for only 8% of deaths with LVH with fibrosis accounting for another 14% of deaths. While this study provided a uniform and high quality post-mortem examination, only cases of sudden death referred from local coroners were reviewed at this center.6 This may have led to a skewed sample with more obvious causes of death such as HCM not being referred.
A study on the etiology of SCD with standardized, quality autopsies, an unselected population and active surveillance in active duty military personnel ages 18-35, often seen as comparable to athletes due to the regular, vigorous training required, provides additional insight. This study reported death due to SADS in 41% of cases, HCM in 13%, with another 8% of deaths due to left ventricular cardiomyopathies; a similar distribution to that seen in the United Kingdom.7
Indeed, a meta-analysis which included 34 studies on the cause of SCD in athletes showed HCM responsible for 10.3% of deaths and SADS represented 26.7%.8 This correlates to conditions which are found at screening. In a meta-analysis of over 47,000 athletes that reviewed conditions identified at screening, there were 160 conditions associated with SCD discovered: 42% Wolff-Parkinson-White (WPW); 11% Long QT Syndrome (LQTS); and 11% HCM, equating to a prevalence of WPW of 1 in 700 athletes and 1 in 4,200 for LQTS and HCM.9 While the prevalence of WPW and LQTS is consistent with published studies, the lower prevalence of HCM in the screened population may reflect pathology that has not yet manifested, as many of the studies included youth and high school athletes, or that HCM athletes were selected out because of the physical demands of the sport. Given the high prevalence of WPW reported compared to other conditions associated with SADS, and given molecular autopsy identifies channelopathies in SADS only 35% of the time,10 one could question if WPW is responsible for a larger proportion of SADS than previously appreciated. Because WPW is common and treatable, aggressive and earlier identification could be warranted if this were the case and this possibility should be explored with further research.
Finally, emerging research would suggest that the causes of SCD may be variable among different athlete groups. In studies of high risk groups such African Americans and male basketball players, it appears that left ventricular pathology is more frequent than in other athletes.2,3 This is important as HCM readily exhibits ECG abnormalities in 90% of cases and ECG changes may precede structural changes and further supports a targeted approach to screening.11
In the end, it seems there are several important takeaways. First, HCM may not cause SCD as commonly as previously thought; however, HCM and other left ventricular pathology remain an important cause of SCD, especially in high risk groups. Whether the left ventricular pathology is congenital or acquired and the role of exercise in both are intriguing questions. Second, WPW is prevalent in the athletic population, easily identified and often treatable, and questions arise if this is a more common cause of SCD than previously appreciated. Finally, it is indisputable that standardized high quality autopsies in combination with genetic testing and expert interpretation can further our understanding of the causes of SCD in young athletes and inform prevention strategies. Continued collaborative efforts to increase the understanding of both the incidence and etiology of SCD should be pursued.
- Maron BJ, Doerer JJ, Haas TS, Tierney DM, Mueller FO. Sudden deaths in young competitive athletes: analysis of 1866 deaths in the United States, 1980-2006. Circulation2009;119:1085-92.
- Harom K, Asif IM, Maleszewski JJ, et al. Incidence, cause, and comparative frequency of sudden cardiac death in national collegiate athletic association athletes: a decade in review. Circulation 2015;132:10-9.
- Harmon KG, Asif IM, Maleszewski JJ, et al. Incidence and etiology of sudden cardiac arrest and death in high school athletes in the United States. Mayo Clin Proc 2016;91:1493-1502.
- Marijon E, Tafflet M, Celermajer DS, et al. Sports-related sudden death in the general population. Circulation 2011;124:672-81.
- Committee on Identifying the Needs of the Forensic Science Community. Strengthening Forensic Science in the United States: A Path Forward. Washington, DC: The National Academies Press; 2009.
- Finocchiaro G, Papadakis M, Robertus JL, et al. Etiology of sudden death in sports: insights from a United Kingdom regional registry. J Am Coll Cardiol 2016;67:2108-15.
- Eckhart RE, Shry EA, Burke AP, et al. Sudden death in young adults: an autopsy-based series of a population undergoing active surveillance. J Am Coll Cardiol 2011;58:1254-61.
- Ullal AJ, Abdelfattah RS, Ashley EA, Froelicher VF. Hypertrophic cardiomyopathy as a cause of sudden cardiac death in the young: a meta-analysis. Am J Med 2016;129:486-96.
- Harmon KG, Zigman M, Drezner JA. The effectiveness of screening history, physical exam, and ECG to detect potentially lethal cardiac disorders in athletes: a systematic review/meta-analysis. J Electriocardiol 2015;48:329-38.
- Tester DJ, Medeiros-Domingo A, Will ML, Haglund CM, Ackerman MJ. Cardiac channel molecular autopsy: insights from 173 consecutive cases of autopsy-negative sudden unexplained death referred for postmortem genetic testing. Mayo Clin Proc 2012;87:524-39.
- Drezner JA, O’Connor FG, Harmon KG, et al. AMSSM position statement on cardiovascular preparticipation screening in athletes: current evidence, knowledge gaps, recommendations and future directions. Br J Sports Med 2017;51:153-67.