Terminology for health problems

WHO defines health as ‘a state of complete physical, mental and social well-being’ and not merely the absence of disease or infirmity.12 Extending this definition, Clarsen et al 13 defined an athletic health problem as any condition that reduces an athlete’s normal state of full health, irrespective of its consequences on the athlete’s sports participation or performance or whether the athlete sought medical attention. This constitutes an umbrella term that includes, but is not limited to, injury and illness.

Health problems can have several consequences. A health problem that results in an athlete receiving medical attention is referred to as a ‘medical attention’ health problem, and a health problem that results in a player being unable to complete the current or future training session or competition as a ‘time-loss’ health problems.1 3–5 7 14 As not all health problems limit an athlete’s ability to participate nor require medical attention, broader definitions (self-reported, symptom-based or performance-based) will capture more health problems. Figure 1 illustrates these differences.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Distribution of health problems by consequences (not to scale). Adapted from Clarsen and Bahr.84

Defining injury and illness

Previous consensus statements on injury and illness in sport have proposed largely consistent definitions for injury and illness.1 3 5–10 14 15 Differences in definition stem from the specific sport or context for which statements were developed. For this consensus statement, we define injury and illness as follows:

Injury is tissue damage or other derangement of normal physical function due to participation in sports, resulting from rapid or repetitive transfer of kinetic energy.

Illness is a complaint or disorder experienced by an athlete, not related to injury. Illnesses include health-related problems in physical (eg, influenza), mental (eg, depression) or social well-being, or removal or loss of vital elements (air, water, warmth).

We acknowledge that there is not always a clear distinction between injury and illness. The consensus was that for injury, the primary mode involves transfer of kinetic energy, but other types of injury, such as sunburn or drowning, may have a different aetiology.

These definitions are meant to be inclusive; they embrace a broad array of injury-related and illness-related health problems that may affect an athlete. Depending on the goal of the monitoring activity, data recording may be limited to specific health problems that constitute a narrower subset of the above definitions (ie, via an operational definition). If the surveillance programme has a narrow scope (eg, to capture only concussions in school rugby), data recording can be limited to the specific injury type of interest.

Relationship to sports activity

Health problems may result:

1. directly from participation in competition or from training in the fundamental skills of a sport (eg, players colliding in a match, overuse from repetitive training or transmission of a skin infection from contact with another player);

2. indirectly from participation in activities that related to competition or training in a sport, but not during competition or a training session (eg, slipping, falling and sustaining an injury when in the Olympic village, developing an illness following international travel to a competition or an illness deemed to be related to an increased training load over a few weeks);

3. from activities that are not at all related to participation in sport, that is, would occur in the absence of participation during competition or training in the fundamental skills of a sport (eg, car crash, sudden cardiac arrest at home).

Depending on the purposes of the study, researchers may want to report health problems in these categories separately.

Mode of onset

Traditionally, health problems have been classified into those that have a sudden onset and those that have a gradual onset. Sudden-onset health problems were considered to be those that resulted from a specific identifiable event (eg, a collision between an athlete and an object causing a fracture). Gradual-onset problems, on the other hand, were considered to be those that lack a definable sudden, precipitating event as the onset (eg, a tendinopathy induced by repetitive movement).

The term ‘overuse injury’ is commonly applied to gradual-onset injuries. However, this term is used inconsistently in the literature16 17 and most injury surveillance systems do not define ‘overuse injury’.16

Health problems may have elements of both sudden onset and gradual onset. For example, a long-distance runner with an intensive training regime may have insufficient recovery, resulting in cumulative stress-related changes to the bone, but presenting as an acute tibia fracture without prior pain. The dichotomy between sudden onset and gradual onset, which most methods of data capture are based on, means such important nuances may be missed. One option to address this problem would be to classify health problems based on the underlying pathology, whether this indicates a single or repetitive pathogenic mechanism, based on imaging studies (eg, MRI, ultrasound) or tissue biopsies. However, routine capture of such detail in a reliable manner within a surveillance system is challenging.

Classifying the mechanism of injury

Mechanism of onset has typically been defined only in the context of sudden-onset injuries. Sudden-onset health problems can result from contact and non-contact mechanisms; this classification is discussed below and presented in table 2.

Direct contact mechanisms directly lead to the health problem in an immediate and proximal manner.

Indirect contact mechanisms also stem from contact with other athletes or an object. The force is not applied directly to the injured area, but contributes to the causal chain leading to the health problem.23–26

Non-contact mechanisms are those that lead to health problems without any direct or indirect contact from another external source. Gradual-onset injuries, by their nature, are non-contact.

We anticipate that subsequent sport-specific consensus statements will provide more detailed subclassifications to address specific features of contact mechanisms (eg, subclassification of contact with objects, such as ball, bat, net, gate). Future sport-specific statements may also give specific recommendations on other categories for classification related to injury causation (eg, rule infringements, particular movements or other sport-specific features). The International Classification of Disease (ICD) External Causes Chapter27 and the International Classification of External Causes of Injury28 provide specific codes that might be useful.

Multiple events and health problems

One of the particular features of sports epidemiology, compared with other settings, is the relatively high chance that an athlete will sustain more than one health problem over the follow-up period. This is illustrated in figure 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Examples of hypothetical prospectively collected injury/illness data (adapted from Finch and Marshall).114 The X indicates when a period of surveillance is ended because the athlete left, unrelated to health problems, before the end of the study period; this is called censoring.

The relatively common occurrence of multiple health problems in a single individual poses challenges for the reporting and analysis of sports injury and illness data.29 In particular, note that the number of athletes in a study is unlikely to be the same as the number of reported health conditions and both should be stated. When reporting the frequency (or proportion) of specific diagnoses or other characteristics, it is important to state clearly whether this is expressed as the proportion of all athletes followed up or the proportion of all injured athletes or the proportion of all reported injuries.

Subsequent, recurrent and/or exacerbation of health problems

Was a subsequent health problem related to previous health problems? This is an important question in the field. To know whether health problems follow previous health problems requires both sets of problems to be classified correctly using consistent terminology. This exercise can provide greater insight into the etiological factors that underpin subsequent health problems.30

Hamilton et al 31 provided a useful framework to categorise subsequent injuries/illnesses and exacerbations in sport (figure 3). More recent frameworks incorporate extensive criteria30 32 33 that require judgement by trained clinicians, which may be beyond the scope and capacity of many surveillance protocols. When reporting frameworks become more complex, there is a greater risk of data errors.34 In general, we do not recommend complex frameworks but they can be considered for sophisticated data collection and analysis where appropriate expertise and resources exist.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Classification tree for subsequent health problems (adapted from Hamilton et al 31). Definitions: (1) index injury (illness)=the first recorded injury (illness), (2) subsequent injury (illness)=any injury (illness) occurring after the index injury (illness) ((i) subsequent injury to a different location than the index injury (subsequent illness involving a different system than the index illness); (ii) subsequent injury to the same location but of a different tissue type than the index injury (subsequent illness of involving the same system but of a different type/other diagnosis) or (iii) subsequent recurrent injury (illness)=subsequent injury to the same site and of the same type as the index injury (subsequent illness involving the same system and type as the index illness). Third, fourth or more health problems should be assessed relative to the initial index health problem and all other previous ones (eg, second and third health problem).

The recommended subsequent injury terminology, adapted from Hamilton et al,31 includes noting whether subsequent injuries: (i) affect the same site but other tissues (eg, knee but meniscus instead of ACL alone) or (ii) affect other sites. Subsequent illness terminology31 notes whether the subsequent illnesses is the same system (eg, respiratory) but other diagnosis (eg, bronchospasm as distinct from a viral illness) or to other systems. The relevant definitions are shown in figure 3. Note that an injury may be subsequent to an illness and vice versa (eg, bones stress injury following diagnosis of an eating disorder, depression following a lengthy recovery after revision ACL reconstruction).

Subsequent injuries to the same location and tissue as the index injury are recurrences if the index injury was healed/fully recovered; they are exacerbations if the index injury was not yet healed/fully recovered. Subsequent illnesses to the same system and type as the index illness are recurrences if the individual has fully recovered from the index illness, and exacerbations if the individual has not yet recovered from the index illness. Healed/fully recovered from injury (or illness) is defined as when the athlete is fully available for training and competition (see ‘Severity of health problems’ section).

To illustrate how to classify subsequent injury, consider athlete ‘A’ who, following an ACL rupture and surgical reconstruction, presents late in the rehabilitation period, before return to play, with swelling and pain in the knee after a slip and fall injury resulting in a graft tear. This injury would be classified as an exacerbation of the index injury. In contrast, athlete ‘B’ rehabilitated successfully after an ACL reconstruction and returned to play; that player presents with pain and swelling in the same knee. If the diagnosis is a torn ACL graft, this would be classified as a recurrent injury. If the diagnosis is a meniscal tear (ACL graft intact), this is a local subsequent injury.

To illustrate how to classify subsequent illness, consider athlete ‘C’ who has withdrawn from sports participation due to an upper respiratory tract infection caused by influenza type A virus, which then progresses to a lower respiratory tract infection resulting in a diagnosis of viral pneumonia. As athlete ‘C’ is diagnosed with pneumonia before recovery and return to play, the diagnosis of pneumonia is an exacerbation of a recurrent illness. In contrast, athlete ‘D’, following full recovery from the upper respiratory tract infection and return to play, is diagnosed with pneumonia; this illness is a subsequent new illness.

Time to recurrence or exacerbation should be recorded in days (see ‘Severity of health problems’ section).

A minimum list of data items recommended when collecting information on subsequent injury or illness is shown in table 3.

Classifying sports injury and illness diagnoses

Injury and illness classification systems are used in sports medicine to:

1. Accurately classify and group diagnoses for research or reporting, allowing easy grouping into parent classifications for summary, so that injury and illness trends can be monitored over time or injury or illness incidence or prevalence can be compared between groups (eg, different teams, leagues, sports, sexes), potentially leading to risk factor and preventive studies.

2. Create databases from which cases can be extracted for research on particular or specific types of injuries and illnesses.

In the late 1980s, clinicians and researchers were using the ninth edition of the ICD.27 The ICD system is an important international standard, yet even the 11th edition, released in 2018, lacks some classifications important in sports injury and illness surveillance. Hamstring strain and exercise-associated postural hypotension are two examples.35–37 We encourage developers to include more sports medicine diagnoses in future revisions of the ICD.

In the early 1990s, in Canada and Australia, two alternate diagnostic coding systems were developed specifically for sports medicine and these have flourished into the most widely used systems in sports injury surveillance in the world today. Their ‘open access’ nature has allowed other researchers to use them free of charge (with acknowledgement). These diagnostic coding systems are the Sport Medicine Diagnostic Coding System (SMDCS) and the Orchard Sports Injury Classification System (OSICS). Both are based on initial codes to represent body area and further codes to represent injury type or pathology.

One advantage of these coding systems is that they are less cumbersome to apply than ICD codes, especially when built into electronic systems with drop-down menus taking advantage of the body area and tissue-type/pathology-type categories. The full ICD-11 coding system includes 55 000 codes, of which the majority are not relevant in sports medicine, compared with 750–1500 codes for versions of the SMDCS and OSICS.

When reporting aggregate injury data, we recommend using the categories for body areas (table 4) and tissue types and pathologies (table 5) outlined below. Table 6 illustrates how injury data can be reported using these categories. In addition, categories for organ system/region (table 7) and aetiology (table 8) are presented below for illnesses.

When recording injuries or illnesses, the diagnosis should be recorded in as much detail as possible given the information available and the expertise of the individual reporting. Acknowledging that some studies will rely on athlete self-report, or proxy report by parents, coaches or other non-medically trained staff, this consensus group also suggest categories to guide reporting of illnesses (table 9). When injury data are reported by athletes or non-medical staff, we recommend that reporting is limited to body area, as their reporting of tissue type and pathology is unreliable.38

To facilitate reporting based on diagnostic codes, a companion paper has been written with a supplemental Excel data file that provides a full list of revised SMDCS and OSIICS (Orchard Sports Injury and Illness Classification System) codes, along with a translation between both systems and the ICD system.39

Injuries—body area categories

Wherever possible, we tried to define body areas anatomically as either joints or segments. However, we made exceptions based on common clinical presentations in sport where needed. For example, hip/groin is an area we have defined, which is a combination of joint and part of a segment, and therefore not a singular anatomical region.

When one injury event results in more than one injury, the individual diagnoses should be recorded and classified separately. However, for injury incidence and prevalence reporting purposes this will be counted as one injury, and severity should be reported as the severity of the principal (most severe) injury (see below for further explanation).

Injuries—tissue-type and pathology-type categories

Using consensus methodology, we compared the ‘injury-type’ codes from the OSICS and SMDCS systems to arrive at definitions of injury types. We constructed this table to be a single table reflecting ‘injury types’ (as per OSICS) but split two columns into ‘tissue’ (as the broad area) and then ‘pathology’ type more specifically. This reflects the original approach taken in the SMDCS.

Recommendations: reporting injury characteristics

Injury characteristics are often reported in one table by region, one by injury type or both. Cross-tabulations depicting data by region and injury type, that is, combining the two into one table, often become large and unwieldy. It can leave many cells empty or with very few cases (which can then compromise confidentiality), unless the dataset is unusually large. Such tables often also provide insufficient information for research focused on specific areas or sports. For example, in a sport where knee sprains dominate, it may be desirable to report subgroups of these (eg, ACL, medial collateral ligament) at greater detail.

In many cases, a better reporting option is to combine region and type and diagnosis in one table such as in the example shown in table 6, where some categories have been collapsed at the level of body region (bold), some regions have been split further into injury types (subheader) and some even at the level of specific diagnosis (italics). It is expected that subsequent consensus statements on specific sports will provide recommendations on suitable, standard formats for each sport, to facilitate direct comparison of data on key injury types from studies on the same sport.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Risk matrix based on the duration of time loss illustrating the burden of match injuries among professional rugby teams in New Zealand between 2005 and 2018 (unpublished data). The darker the yellow, the greater the burden. The curved grey lines represent point with equal burden. The vertical and horizontal error bars represent 95% CIs. See also table 6, illustrating the same dataset in more detail.

Illness—categories for organ system and aetiology

Illness consensus categories are presented in tables 7 and 8. These are more detailed than the original versions of the SMDCS and OSICS. Our tables diverge from the ICD categorisation format, in which body systems and pathology types are grouped together. We believe that it is important to recognise that an illness, like an injury, affects both a body system and has a specific pathological type. A respiratory infection does not need to be considered either only as a respiratory condition or an infection, it is certainly both. Our recommended illness systems are similar to many of those in the ICD, but we have merged some systems, such as the upper respiratory system and nose/throat.

The professional background of those who report health data will influence the final data quality (see ‘Data collection methods’ section).40 When athletes themselves (or non-clinical recorders like coaching staff) are asked to capture illness data, they should be encouraged to record symptoms rather than attempt a diagnosis. Table 9 lists symptom clusters that are characteristic of various systems. We caution that this table requires additional validation and may be modified in future. Mapping symptoms to bodily systems sacrifices some accuracy; however, in circumstances where expert recorders are unavailable, it is better to have general systems diagnosis data than no data at all.

Recommendations: reporting illness characteristics

As was the case when we discussed reporting of injury data, we recommend against illness data being reported as cross-tabulations of organ system by aetiology type. A better option is to combine system/region and aetiology in one table, as in the example on injuries shown in table 6. Depending on the illness pattern of the sport/setting, some region categories may be collapsed, others split further into aetiology type and even to the level of specific diagnosis (where available), to highlight the most significant illnesses. We expect that subsequent sport-specific consensus statements will recommend useful standard formats for each sport.

Time loss from training and competition

The most widely used severity measure in sports medicine is the duration of time loss. It was recommended in previous consensus statements,3 5–8 44 and is relatively simple to capture, even when data collectors are non-expert—coaches, parents or athletes themselves.

When using this approach, we recommend that investigators record severity as the number of days that the athlete is unavailable for training and competition, from the date of onset until the athlete is fully available for training and competition.

The number of time-loss days should be counted from the day after the onset that the athlete is unable to participate (day 1), through the day before the athlete is fully available for training and competition. Therefore, cases where an athlete does not complete a particular competition or training session, but returns on the same or following day, should be recorded as 0 days of time loss (see table 10 for examples). We note that in some cases, time loss does not follow immediately after the health problem occurred and may be delayed and/or intermittent (table 10).

When athletes recover from health problems during periods with no planned training or competition (eg, during an end-of-season break), investigators should record the end date as when the athlete normally would have been ready for full training and competition participation.

When aggregating data across athletes, report severity as the total number of time-loss days, together with median and quartiles. Means and SD should be interpreted with care, given that the distribution of time loss days is likely to be right-skewed.

When reporting data separately in severity categories, we recommend using the following time bins: 0, 1–7 days, 8–28 days,>28 days.

If one injury event results in multiple injuries, injury severity should be based on the injury leading to the longest time loss (eg, if a downhill skier crashes and suffers two injuries, a concussion which takes 10 days to resolve and a tibia fracture which takes 120 days, time loss for the event is 120 days).

Athlete-reported symptoms and consequences

There are tools to measure injury and illness symptoms that directly address the second limitation of time loss discussed earlier, underestimating the effect of ongoing pain and symptoms that are below the time-loss threshold. A tool such as the Oslo Sports Trauma Research Center Questionnaire on Health Problems (OSTRC-H) complements time-loss measures of severity, as it also captures symptoms and functional consequences of injury and illness. This purpose-built instrument was devised in 201345 and updated in 2020,13 and has played an increasing role in sports injury and illness surveillance, especially in sports and settings where overuse injuries and illnesses represent a substantial burden on health and performance.48

The tool (which can be delivered via a mobile app) invites athletes to record reduced sports participation, training modifications, performance reductions and symptoms.45 Based on the response to these questions, researchers can calculate a severity score ranging from 0 to 100 at specific time points. These can be aggregated (summed as the area under the curve) to monitor injury and illness over time (figure 4). This is called the cumulative severity score. A limitation of this method is that severity score is an arbitrary number and it has not been thoroughly validated as a proxy for injury severity.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Example of severity scores being used to track the severity of three ‘typical’ health problems. Each black dot represents the weekly severity score. The area in orange represents a gradual-onset injury (cumulative severity score (the sum of weekly scores, the area under the curve): 1820), the black area represents a short-duration illness (100) and the dark red area represents an acute medial collateral ligament injury (362).45

Recording the severity of health problems based on clinical assessment

Investigators may also report the severity of health problems based on clinical outcomes such as the need for hospitalisation or surgery,10 42 retirement from sport, permanent disability, or death.10 49

Other severity measures

Depending on the sport setting and the purpose of data collection, investigators may also quantify severity in other ways.41 Function, performance and patient-reported outcome measures may be used to capture severity. Specific examples include:

• Functional measures, for example, International Classification of Functioning, Disability and Health.55

• Sports-related performance measures, for example, balance, strength and endurance. We include athletes reporting retirement from sport in this category.

• Patient-reported outcome measures, for example, ACL Quality of Life questionnaire,56 Knee injury and Osteoarthritis Outcome Score57 and Sport Concussion Assessment Tool.58

Tracking exposure for injury analyses

For injuries, exposure is generally quantified as the time during which athletes are at risk of injury (eg, minutes played), distance covered or a count of the number of specified events (eg, tackles, throws or jumps). In some sports, exposure is commonly expressed as the number of athletic participations (eg, games, races, training sessions), often referred to as ‘athletic exposures’. Table 2 in online supplementary appendix 1 provides a range of examples of exposure measures used.

In team sports, we recommend recording exposure for each individual within a team rather than merely estimating the number of matches the team plays and match duration (team exposure), because the former permits the researcher to examine individual risk factors. Results of all the individuals are then summed to provide exposure at the sport or team level.

As injury risk is often markedly different between training and competition, these exposures should be recorded and reported separately. To do this consistently, it is necessary to define competition and training, and to consider situations where applying the definition may be challenging.

We define competition as organised scheduled play between opposing athletes or teams of athletes, or athlete(s) competing (i) against time and/or (ii) to obtain a score (judged or measured).

We define training as physical activities performed by the athlete that are aimed at maintaining or improving their skills, physical condition and/or performance in their sport.

In many sports, it is common to simulate competition as a part of training. Examples include preseason ‘friendly scrimmages’ between two teams or dividing a single squad into teams that compete against each other. In general, this should be counted as training exposure. Additionally, activities such as warm up and cool down should be counted separately and reported as training injuries, even if occurring around competition.

It is likely that in some sports, these definitions will not be fully applicable. In such cases, we encourage sport-specific consensus groups to define what constitutes competition and training in that sport.

Tracking exposure for illness analyses

Because athletes remain at risk of developing illness even when they are not participating in sport, it is inappropriate to use exposure measures such as playing hours or movement counts to quantify illness risks (except for the rare cases of transmissible infections that are specific to participation in a sport, eg, scrum-pox). Instead, it is often most appropriate to use exposure measures based on the time athletes are under surveillance (eg, days or years), rather than time engaged in competition and training.

Recording exposure during multiday competitions

Multiday competitions, such as championships and tournaments, represent an exposure measurement challenge, particularly for injury analyses. Ideally, investigators should obtain accurate records of every athlete’s individual participation (eg, training and competition minutes) throughout the tournament. However, this is not always feasible. Acceptable exposure estimates can also be made by obtaining summary data from every team for each day of the tournament (eg, squad numbers). As a minimum standard, exposure can be estimated for each event by multiplying the number of registered athletes by the duration of the tournament (the number of days of competition). In multisport tournaments, this should be calculated for each sport. However, this approach assumes that all athletes have the same exposure and participate every day, which is rarely the case.

Training subcategories

Different types of training should, if possible, be recorded and reported separately. Training types can be generally categorised as follows:

1. Sport-specific training: sessions involving the techniques and/or tactics of the sport, usually supervised by of a coach.

2. Strength and conditioning: sessions solely composed of resistance training and/or conditioning training. In many cases, training sessions are mixed (sports-specific, but with addition of some strength and conditioning, eg, plyometrics, endurance). As a pragmatic consideration, any session containing sports-specific training should be categorised as such, even if the session includes some strength and conditioning, purely to streamline exposure tracking.

3. Other training sessions: sessions that include activities other than sport-specific training or strength and conditioning. These include recovery sessions (eg, low-intensity running and stretching), rehabilitation and postrehabilitation transition sessions (ie, postreturn to sport but prior to resuming normal training).

Sport-specific injury surveillance systems may need to depart from this guidance if there is a need to address a specific training concern, however, at a minimum, all training exposure that contain sports-specific training should be tracked.

Sport-specific injury surveillance systems are encouraged to develop specialised procedures for tracking the diversity of training exposures in their particular sport. Training programmes vary considerably among sports and many coaches intentionally design training programmes that integrate multidimensional training (eg, plyometric stretching, sports-specific training, light running) into a single session. In general, investigators should prioritise capturing specific data on the training activities considered to present the greatest health risk.

Wearable physical activity tracking devices enable investigators to capture large volumes of competition and training data at the elite level and from community sports participants across large sample groups. We encourage the use of these devices for tracking exposure. However, we caution that any device needs to be fit for purpose, and researchers should obtain evidence on their validity and reliability before data collected through these devices are used for injury surveillance.

Rates and proportions

Rates and proportions of injury and illness in studies of sports are usually reported as a count of ‘cases’ of the outcome of interest (the ‘numerator’), divided by a population at risk of developing the outcome (the ‘denominator’).60 Because research questions such as ‘How many players have suffered a knee injury?’, ‘What is the risk of getting injured in this sport?’ and ‘How does sport A compare with sport B for concussion risk?’ are very different, there are various ways of reporting risks relating to sports injury and illness. We explain some fundamental terms here.

Prevalence—How many? Prevalence is a proportion and refers to the number of existing cases divided by the total population at risk at a given point in time (point prevalence, eg, the proportion (percentage) of players in a volleyball team who—today—are suffering from patellar tendinopathy). It is a snapshot at one point in time, but can be repeated to determine changes in prevalence over time (eg, weekly). With serial measurements, it is possible to report, for example, the average prevalence over the course of the season and also to compare different stages of the season.

Period prevalence extends the concept of a single point in time to a window of time (eg, one season, a year). It refers to the proportion of athletes that has reported the condition of interest (eg, patellar tendinopathy) at any time during that given window. Notably, this includes people who already had the condition at the start of the study period as well as those who acquired it during that period.

Incidence—How often (do new cases occur)? Incidence is a rate and, as with any rate, time comes into play. Incidence refers to the number of new injuries/illnesses in the population that develop during a defined period of time. The term ‘incidence rate’ is synonymous, but we argue that it is a tautology—‘incidence’ is a rate.

Note that prevalence is calculated based on the number of athletes with a health problem, while incidence refers to the number of new health problems.

Recommendations: expressing risk in SIIS

Incidence-based measures usually represent more appropriate outcomes for sudden-onset conditions (eg, ankle sprains, ACL injuries) and prevalence-based measures for gradual-onset conditions (eg, asthma, patellar tendinopathy).46 Overuse injuries and pain problems such as low back pain and patellar tendinopathy are often chronic, with periods of remission and exacerbation. For example, in a professional volleyball team, there could be only one new case of patellar tendinopathy (so incidence will be low), yet 40% of the players (ie, nearly all pre-existing) could be affected by patellar tendinopathy during the season (period prevalence). Therefore, for such conditions, prevalence (the proportion of athletes affected) is a more appropriate measure than incidence (the number of new cases during the season).

Because sports and the activities that comprise them are so diverse, there is no single approach to expressing risk appropriately for all sports injury surveillance projects.61 In general, incidence-based measures that provide a standard time-window for the population at risk (injuries per hour) are preferable to measures for which the time at risk varies across individuals (injuries per athletic exposure, ie, per training session or match), because time-based measures better facilitate comparison across sports.

To provide numbers that are easy to interpret, avoiding small decimals, these data are typically reported as ‘per 1000 player hours’ (eg, concussion rate in men’s rugby study was reported as 4.7 per 1000 player hours, rather than 0.0047 per player hour).62 Such numbers allow risks to be compared (eg, How does concussion risk vary across contact sport codes?). We expect that subsequent sport-specific statements will recommend suitable, standard incidence-based measures for each sport. Table 2 in online supplementary appendix 1 provides a range of examples of risk measures.

If one injury event results in multiple injuries, these should only be counted as one when calculating overall injury incidence (eg, if a downhill skier crashes and suffers two injuries, a concussion and a tibia fracture, these are counted as one injury when calculating incidence).

Because of the difficulties in accurately measuring exposure to pathogens (which may be greater when not training or competing), illness risk should be estimated based on the entire period of exposure (eg, the duration of a competition, a ‘season of play’, a year), not athletic exposure only. We recommend reporting illness risk as either the incidence, the number of new illness cases divided by a period of time (eg, illnesses per 365 athlete-days),63 or as the period prevalence of illness, the proportion of athletes that were ill during a defined period.64 65

Where time-based measurements of exposure are unavailable, but participant numbers are available, crude rates of injury per number of participants per period can be derived. In such cases, we suggest that the incidence that may be most useful to permit population-level comparisons among sports or studies is ‘injuries per 365 athlete-days’.

Similarly, the proportion of participants with new or recurring injury or illness (ie, excluding pre-existing cases and exacerbations) during the event has been used to provide an impression of the risk associated with participation in each sport in both the summer and winter Olympic Games.64 65 However, this approach—period prevalence—can suggest widely different relative risks of activities that differ substantially in the amount of exposure participants experience.64 For example, exposure differs substantially between a football player and a sprinter. Period prevalence describes the absolute risk of participation in the Olympic Games, but not the relative risk (the risk of injury during 1 hour of football play vs 1 hour of marathon running).

Injury rates reported on a per-event (eg, per rugby tackle) basis provide information about how likely a particular aspect of play (‘event’) is to result in injury. Understanding events that both do66 and do not result in injury67 68 helps researchers identify injury prevention opportunities. In the absence of information about how frequently the event occurs within a sport and the average duration of the sport in which participants are exposed, rates per event also provide an incomplete view of the overall risks a sport poses. Using time-based and event-based denominators (eg, tackles in football codes) in parallel can help provide insights into both, which event (eg, tackle type) is most frequently associated with injuries and which event carries the highest risk when it occurs. To date, there have been relatively few injury surveillance studies in which such statistics have been provided together.29 For televised sports and those using new technology such as activity trackers, measurement of the duration of playing time and intensity for each individual is feasible, and coding of the number, characteristics and duration of activities each participant engages in (eg, tackles) is routine for some professional sports (eg, football). We include a real-life illustrative case of surveillance methods being used to investigate injury risk in rugby in online supplementary appendix 1.

Communicating risk to stakeholders

From clinical and practical perspectives, it is important that the end-users (the athletes, coaches and medical staff members) can make sense of the injury reports and increase the chances of having them participate in risk management plans. This can be done by expressing the injury incidence based on the concerned sport’s specifications. For instance, if an injury incidence for a specific muscle group (eg, hamstrings) is expressed as 0.9 injuries per 1000 hours of exposure, the incidence per player per season (0.28 injuries per player per season) could be multiplied by the average number of athletes per squad for the concerned sport (eg, 25 in football). This gives seven hamstring injuries per squad per season, a quantity which is more easily interpreted by end-users.

Another relevant measure, which is easy to communicate to managers, coaching staff and athletes, and which is associated with team performance in football,69 is player availability. Player match availability is calculated as the sum of player match opportunities (ie, the number of matches multiplied by the full size of the squad) minus the sum of player match absences due to injury or illness, and can be expressed as the average percentage over the period of interest (eg, one season). Training availability can be calculated in the same way.

We encourage sport-specific consensus statements to recommend relevant measures to communicate risk to relevant stakeholders.

Classification of sport categories

There are many ways of classifying and grouping sports. Any sports classification system used in surveillance should be clearly described in the methods section of reports. The description should permit other researchers to understand and replicate the process by which sports were grouped. The research problem being addressed should shape the classification system used, rather than vice versa.

From pen and paper to electronic solutions

Health problems and exposure can be captured using different methods ranging from paper copy data collection forms to a comprehensive web-based surveillance system, for example, internet platforms, apps or text messaging.51 90 91 96–100 The traditional pen and paper approach is often easy to implement,50 as it reduces the need for specific technical knowledge, equipment and related costs.51 101 Data can be verified and cleaned as they are manually entered.101

Electronic data capture reduces time for duplication of data entry,50 and associated entry errors.101 In terms of costs, there is potential long-term cost-effectiveness through the elimination of expenses linked to the printing, shipping, management and storage of physical documents.101

Web-based solutions allow instant and remote on-demand queries of real-time data (including end-users such as team medical staff), as well as integration with other data feeds (eg, performance, load, sleep). Web-based solutions should preferably be prototyped prior to being implemented in a larger injury surveillance setting. Full integration of surveillance reporting systems within clinical electronic medical record keeping systems has been used successfully in a number of professional elite leagues.102 While electronic solutions can lead to high response rates among athletes,90 91 98 there are also reports of poor athlete engagement,97 and thus demonstrates the importance of understanding uptake barriers. It is important to use surveillance tools that minimise intrusion into the daily activities of the data reporters (athletes, medical teams, coaches), for example, by limiting the number of questions to responders so that only essential data are captured. Another recommendation is to provide a clear incentive to athletes and teams to participate in injury surveillance, for example, by allowing continuous feedback within the data collection system (eg, performance data, load monitoring data) or sending regular reports back to the teams, athletes and other relevant stakeholders.103

Data collection methods must be adapted to the specific research question, the sport context and the skill set of the research team, and should follow strict quality standards. The quality of the surveillance system includes the quality of the forms (baseline, health problems and exposure) as well as the quality of the data collection procedure, implementation, data cleansing and analysis methods.51 The quality and usability of the forms and the data collection procedures should be examined before implementation. Reliability and validity should be analysed, and all translations should follow the standards of intercultural adaptation.104 105 The adherence to the data collection protocol as well as the completeness and consistency of responses should be monitored on a regular basis during implementation. Collaboration between research groups to share resources and joint data analytics can help advance the management of sport injuries/illnesses.106 Having data collection forms and related material available in free-to-access formats make it easier for sports bodies to participate in surveillance activities,51 and this consensus statement includes some sample forms as mentioned in online supplementary appendix 2.

Research ethics and data security

Research ethics govern the conduct of medical research and aim to protect the dignity, rights and welfare of human participants. They detail principles such as informed consent, data confidentiality, the use of research ethics committees and risks, burdens and benefits. Importantly, informed consent is the process in which permission is granted in full knowledge of the possible consequences (risks and benefits), for example, for their data to be used for research purposes. In some contexts, injury and illness surveillance may be regarded as an integral part of data audit and quality control processes, and—as long as individual patient data are fully deidentified—may not require informed consent. It is the duty of all researchers (and all other users of the data) to consider—and adhere to where appropriate—internationally recognised guidelines for research ethics (such as the Declaration of Helsinki107 and the Declaration of Taipei).108

Data protection governs how data are collected, shared, used and conserved, and aims to ensure that personal data are safe from unforeseen, unintended or malevolent use. Particular attention must be directed to the security of data stored on cloud-based systems and other electronic repositories. Researchers must adhere to the data protection regulations applicable to their context (such as General Data Protection Regulation in Europe).109