Exertional SCD or SCD occurring at any time?

The primary objective of prevention programmes is early detection of disease, in order to prevent adverse health and catastrophic events from ever occurring. While cardiovascular screening in athletes is generally thought justified due to an increased risk of cardiovascular events during exertion, programmes designed to prevent SCD are not intended to prevent SCD only during sport or only during a school-sponsored activity, but rather to promote athlete health and prevent SCD at any time. Many of the pathologies associated with SCD in young athletes may present with lethal arrhythmias at rest or during sleep; therefore, it is the incidence of SCD inclusive of SCD occurring outside of sport, not just the incidence of SCD occurring during exertion, which is necessary to accurately identify athlete risk.

Case identification: mandatory reporting, media reports, registries and insurance claims

Mandatory reporting systems afford the most reliable results, although few of these exist. Registries depend on passive collection methods and are subject to ascertainment and selection bias. Many studies rely on the use of media reports to identify cases; however, this results in underestimating SCD incidence. In one retrospective cohort study from Denmark, only 20% of athlete deaths identified by death certificates were found through an extensive media search.6 In another retrospective cohort study of a high profile US college athlete population, media reports identified only 56% of total cases.3 Media reports are likely to be less reliable prior to widespread use of the Internet and in lower profile athletes (younger athletes, female athletes and non-revenue sports). Studies using catastrophic insurance claims as the only source of case identification also are likely to underestimate the incidence of SCD. Insurance claims provide information only on deaths which occur during school-sponsored sporting events, and do not include cases that occur during individual activity, unofficial practices or deaths which occur outside of sport. In fact, catastrophic insurance claims missed 80% of SCD cases in a study using an internal reporting structure for the NCAA. Consequently, incidence assessments based on media reports and catastrophic insurance claims should be considered a minimum estimate.

The importance of the denominator

Incidence calculations are based on a ratio, thus if the population being examined (the denominator) is not well defined the final estimate can be highly variable. The most accurate incidence estimates will be in a distinct group with well-defined demographics. Estimating the overall number of athletes based on population statistics, in large age ranges or multiple demographic areas without an overarching authority is fraught with potential error. Additionally, some studies may examine sudden deaths in a population as opposed to all deaths in a population.7 ,8 When comparing the relative risk of death, eliminating subtypes such as accidental deaths from the overall analysis significantly changes the ratios and can skew results making entities appear more or less common. Consequently, an accurate denominator is critical to precise incidence estimates.

Impact of age on SCD incidence

The rate of SCD varies considerably with age.10–13 In young athletes (<30 years old), the aetiology of SCD usually involves a genetic structural or primary electrical cardiac disorder. In older athletes, the rate of atherosclerotic heart disease increases rapidly, particularly after the age of 25–30. Thus, studies inclusive of large age ranges or a substantial proportion of older participants can vary significantly and may disproportionately affect incidence rates.14 ,15

Should SCA be included in incidence calculations?

Finally, some incidence calculations examine only SCD, while others also include sudden cardiac arrest (SCA) victims that survive. SCA is an important endpoint with potential morbidity, but it is more difficult to track using the most commonly employed methods. SCA is often not reported in the media and in most settings there is no requirement to report SCA. In early reports of SCD before public access defibrillation programmes, SCA events likely resulted in death, so the need to recognise SCA survivors and SCD victims in incidence calculations was less apparent. With the implementation of emergency action plans and the widespread use of automated external defibrillators (AEDs), survival rates from SCA in athletes now rival that of other public access defibrillation programmes, with one recent prospective observational study demonstrating an 89% survival rate.9 Survival from SCA may drastically decrease the overall incidence numbers of SCD. Primary prevention programmes are intended to identify athletes at risk and prevent SCA from ever occurring, so to properly evaluate screening programmes the inclusion of SCA and SCD in the assessment of incidence is warranted.

In fact, the identification of potentially lethal cardiac abnormalities during preparticipation screening appears to be universally supported. The American College of Cardiology (ACC), the American Heart Association (AHA), the European Society of Cardiology and the IOC all endorse that the identification of potentially serious medical conditions including cardiovascular disease that place athletes at risk for safe sports participation is the primary reason for screening.16–19 The latest edition of the Pre-Participation Physical Evaluation Monograph endorsed by six US medical societies and the AHA also states the primary objective of the preparticipation examination is to ‘screen for conditions that may be life threatening or disabling’.20 With such widespread agreement that the objective of screening is the identification of potentially life-threatening conditions, the effectiveness of preparticipation screening should be understood not just in terms of SCD but also in terms of SCA with survival.

Estimates of SCD in general populations of athletes

There are 13 studies reporting on SCD in athletes with 5 reporting on SCD incidence in general populations of athletes, most with broad age ranges from 8 to 406 ,8 ,21–23 (table 1). The first study to report on the rate of SCD in athletes was a 1995 study which employed a retrospective cohort design and utilised the National Center for Catastrophic Sports Injury Research database from 1983 to 1993 to evaluate sudden deaths in high school and college athletes. An overall SCD rate of 1:300 000 was estimated with a rate of 1:1 300 000 in women.21 While this effort highlighted an area without much previous research, this study relied on passive collection methods and media reports in the preinternet era for case identification, and undoubtedly cases were missed. In addition, the denominator was imprecise. However, 1:300 000 became the starting point for the incidence of SCD in athletes.

The largest report of SCD in a general population of athletes comes from the US Registry for Sudden Death in Athletes (USRSDA).8 ,24 The USRSDA has collected information on sudden death in athletes since 1980. The primary sources of information are media searches, electronic databases and self-reports, that is, next-of-kin. The number of cases identified has increased steadily by approximately 6% annually as media search strategies have improved.8 SCA and SCD were included which contrasted the previous study. The denominator of the athlete population was set at 10.7 million although there is no descriptor of how this was arrived at; in fact, in two previous reports using the same database incidence rates were not reported due to ‘unavoidable selection bias’ and the certainty of significantly underestimating the rate.25 ,26 In this study, the incidence of SCD for US athletes aged 8–39 from 2001 to 2006 was calculated at 1:164 000.8 The strengths of this retrospective cohort study are the long time frame of the registry and the number of cases identified. The limitations of the incidence calculation include a broad age range, underestimation of events and an approximated denominator.

Estimates of SCD in athletes in other parts of the world are impacted by the use of different methods for case ascertainment. In the Veneto region of Italy, there is a mandatory reporting system for juvenile sudden death with autopsies performed by specialist cardiovascular pathologists. In a prospective cohort study from 1979 to 1999, the reported rate of SCD was 1:47 000 in athletes aged 12–35.22 This study presented a more accurate case identification and a precise denominator and the rate was higher than in previous studies. In a subsequent report from the same Italian dataset, the initial rate of SCD from 1979 to 1981 was 1:28 000, and after implementation of a national screening programme, the rate of SCD decreased by 89% to 1:250 000 in 2006 suggesting benefit to a cardiac screening programme inclusive of ECG.27

The disparity in incidence numbers between Italy and the USA have been attributed to the differing ethnic populations with differing risks for SCD and has led some to suggest that the USA need not consider more intensive screening strategies.28 It is more likely that the differing SCD incidence is due to the disparate methods of measuring incidence: retrospective cohorts based on media reports, catastrophic insurance claims and registries subject to ascertainment bias with uncertain denominators, versus a prospective mandatory reporting system with a reliable denominator.

Other studies of SCD incidence using a broad age range of competitive athletes were performed in Israel and Denmark.6 ,23 The Israeli study was a retrospective cohort study which relied on the retrospective search of two major newspapers for case identification in athletes aged 12–44. The denominator was developed by ascertaining the number of registered athletes in 2009 and then extrapolating retrospectively, based on population statistics and the predicted growth in the percentage of the population engaged in athletic activity over time, the number of competitive athletes from 1985 onwards. The incidence of SCD was estimated to be 1:39 000 from 1985 to 1997 and 1:38 000 from 1998 to 2009.23 Critical weaknesses of this study include case identification using only a retrospective search of two newspapers, inclusion of a 24-year interval with variable media attention to athlete SCD, the broad age range with rates of SCD largely influenced by inclusion of older athletes and the imprecise denominator. The authors concluded that an ECG screening programme started in 1998 showed no benefit in the prevention of SCD in athletes, basing this claim on a flawed scientific foundation.4 The retrospective cohort study from Denmark reviewed death certificate and autopsy information for case identification and used population statistics as a denominator.6 The investigators reported a rate of SCD of 1:83 000 in athletes aged 12–35. The primary weakness in this study was the ability to accurately identify competitive athletes from the available information.

SCD in US college athletes

As the incidence of SCD varies considerably with age, the ability to examine populations with more narrowly defined age ranges is important. All studies on SCD incidence in competitive athletes outside the USA include broad age ranges. In the USA, three retrospective cohort studies specifically investigate the incidence of SCD in college athletes aged 17–24, with rates ranging from 1:43 000 to 1:67 0003 ,7 ,29 (table 2). In the first, a survey was sent to NCAA institutions regarding SCD and AED use over an average of 3.3 years and an SCD rate of 1:67 000 was calculated based on 244 responding institutions.29 In the second, a database of NCAA college athlete deaths was compiled from the Parent Heart Watch media database, the NCAA Resolutions database and NCAA insurance claims. This database looked retrospectively at 5 years of deaths from 2004 to 2008, identifying approximately nine cases of SCD per year,3 with the cardiovascular cause of these deaths confirmed by autopsy review in a subsequent study.30 NCAA participation statistics for race, sex and sport were used for the denominator. In this study an overall SCD rate of 1:43 000 was calculated; additionally, high-risk subgroups were identified such as men with a risk of 1:33 000, African-Americans with a risk of 1:18 000, basketball athletes with a risk of 1:11 000 and Division I male basketball players with an alarming risk of 1:3000.3 A similar study reviewed sudden deaths in NCAA athletes over a 10-year period from 2002 to 2011.7 In this study the USRSDA and NCAA Resolutions database was used for case identification, reporting approximately six cases of SCD per year with an incidence of 1:62 000.7 Thus, these three different studies using variable methodology all arrive at reasonably comparable figures for an incidence of SCD in college athletes of approximately 1:50 000, with some subgroups at higher risk.

SCD in US high school athletes

High school athletes aged 14–18 provide another well-defined age subgroup to determine incidence. Seven US studies examined high schools athletes with wide variation in the reported rate of SCD from 1:23 000 to 1:917 0001 ,2 (table 3). An initial incidence estimate of 1:217 000 in high school athletes came from Minnesota in a retrospective cohort study with case identification reliant on catastrophic insurance claim data.31 A recent follow-up study also limited to catastrophic insurance claims reported an incidence of 1:917 000 over the past decade.1 Catastrophic insurance claims only capture deaths which occur during school-sponsored events or practices and significantly underestimate the total incidence of SCD. Highlighting this point was a retrospective review of the very same population of Minnesota high school athletes but using media reports for case identification. Media reports identified 13 cases in Minnesota high school athletes (all in boys), including 6 cases of SCD and 7 additional cases of SCA in student athletes who survived. Catastrophic insurance claims detected only one of six deaths (17%) that occurred in athletes during the same study period and only one of four deaths which occurred while playing sports.5 The three exertional deaths missed by catastrophic insurance claims occurred in athletes when participating in club sports or unofficial practices. In a recent retrospective cohort study of Minnesota state high school athletes using the USRSDA, 13 cardiovascular related sudden deaths were identified over a 26-year period with an estimated incidence of 1:150 000.32 On closer examination and comparison of studies, catastrophic insurance claims identified only four deaths over 19 years and during an overlapping time period the USRSDA identified 13 deaths over 26 years, again emphasising the limitations of using catastrophic insurance claims as the sole means for case identification. Furthermore, insurance claims data limited only to deaths and not SCA will vastly underestimate the rate of major cardiovascular events in athletes.

Other studies have found higher rates of SCD in high school athletes. In a cross-sectional survey of 1710 high schools, an SCD rate of 1:46 000 was found, with a rate of 1:23 000 when SCA was included.2 While this study is limited by the potential for reporting bias, these numbers may underestimate risk as only events which occurred on high school campus were captured; events which occurred away from campus or while sleeping were not included. In two retrospective population-based cohort studies, up to one-third of SCD occurred while sleeping, illustrating the limits of only including school-based events.6 ,33 A recent 2-year prospective cross-sectional study of 2149 high schools demonstrated an overall SCA rate of 1:88 000.34 While the prospective monitoring provided more rigorous surveillance, this study also examined events only occurring on school campus. Lastly, a preliminary study examined SCA and SCD in high school athletes in six US states, representing 36% of the US high school population and almost 20 million athlete seasons over 5 years from 2007 to 2013.35 Media reports were used for case identification and the denominator was defined using data from the National Federation of State High School Associations (NFHS), with an SCA/SCD incidence rate of 1:63 000 calculated.35 The incidence in male high school athletes was 1:41 000 and in male basketball athletes was 1:33 000.35

In conclusion, studies of incidence in high school athletes are more difficult to interpret with a wide range of variables and methods accounting for different incidence rates. The methodological limitations of incidence studies in high school athletes must be recognised. Current assessments of incidence either focus on a fraction of the day (school hours),1 ,2 ,31 ,32 ,34 only account for SCD occurring with certain activities,1 ,31 ,32 include only SCD and not SCA with survival1 ,31 ,32 ,34 or rely on media reports for case identification.5 ,36 All these issues result in incomplete case identification and an underestimation of the incidence of SCD. Based on available studies, we believe that 1:50 000 to 1:80 000 should be considered a minimum estimate range for the incidence of SCD in high school athletes with male athletes at higher risk.

Comparisons to similar non-athlete populations

Fifteen studies report on the incidence of SCD in young, non-athlete populations, primarily generated by retrospective cohort studies using death certificates, coroner's databases and autopsies for case identification. SCD estimates range from 1:22 000 to 1:128 000, depending on the age of the population examined.6 ,10 ,14 ,33 ,36–40 There are several prospective population-based studies using emergency medical services databases which include SCA and SCD and provide more rigorous methodology to consider13 ,41–43 (table 4). The most comprehensive of these prospective studies spanned 30 years and reported an SCA/SCD rate of 1:69 000 in 14–24 year olds and 1:22 000 in 25–35-year olds in King County, Washington.43

Four studies have compared the relative risk of SCD or SCA in competitive athletes with the general population. It is generally accepted that intense physical exertion increases the likelihood of SCA in those with underlying cardiovascular disorders. An Italian prospective cohort study reported a 2.5 times relative risk for SCD in competitive athletes compared with age-matched non-athletes.22 In a prospective cohort study in France, a 4.5 times relative risk of sports-related SCD was reported in competitive athletes compared with recreational sports participants.32 Similarly, a prospective observational study in US high schools with over 4.1 million total student-years and over 1.5 million student athlete-years of surveillance found a 3.6 times relative risk for SCA on campus in high school student athletes versus student non-athletes.13 ,34 In contrast, in a retrospective review of death certificates and media reports in Denmark, investigators reported a rate of SCD in athletes aged 12–35 that was 3.3 times lower than the general population.6

The US military is perhaps the best source to calculate incidence rates of SCD in active individuals, due to the mandatory reporting of deaths with standard autopsy protocols and access to premortem medical records and clearly defined denominators. Although individuals in the military do not meet standard definitions of a competitive athlete, military personnel undergo regular, intense physical training, especially for recruits during basic training. Given the rich source of quality data and the similar median age to college athletes (19 years), these studies in military recruits are a good comparator. In a retrospective cohort study of 6.3 million recruit years over 25 years, the rate of SCD in military recruits was 1:10 000.44 In a follow-up study examining the rate of SCD in active duty military personnel who undergo regular physical activity but not necessarily the intense activity experienced during basic training by recruits, the rate of SCD was approximately 1:30 000 in persons under the age of 30, with the expected increase in overall incidence and death secondary to atherosclerotic heart disease in older military personnel.12 These military studies in exercising young adults with better methodology also support a higher SCD incidence in athletes than initial estimates.