KEY WORDS: Aetiology, ankle sprain, injury cause, injury mechanism, overuse knee injury.

Basketball continues to increase in popularity at all levels of play, from recreational to professional (Zvijac and Thompson, 1996) and remains immensely popular, not just in the United States, but throughout the world (Cantwell, 2004). This tendency is also present in Belgium, with a 173% increase in the number of participants over the last 30 years (Van der Aerschot et al., 2004). This growing participation level in basketball is mainly the result of the attention paid by the media to the NBA competition in the US (Zvijac and Thompson, 1996) and of the inception of the WNBA in 1997 (Deitch et al., 2006). A sport such as basketball, which is traditionally perceived to be safer and is considered to be a non contact sport, (Gomez et al., 1996) has so far not been the focus of much research. More attention has gone to more popular sports like football and to sports which are physically more demanding and carry a higher risk of very serious injuries at all levels of participation (Gomez et al., 1996). Nevertheless, basketball has gained more attention these last 10 years in the scientific literature of injury surveillance (Deitch et al., 2006; Hickey et al., 1997; McKay et al., 2001a; b; Meeuwisse et al., 2003; Messina et al., 1999; Starkey, 2000). Basketball appears to have the highest frequency of injuries among non contact sports; it is even referred to as being more dangerous, with a higher injury risk, than contact sports (Conn et al., 2003; Finch et al., 1998; Finch and Mitchell, 2002; Luidinga and Rogmans, 1985; Rogmans and van Weperen, 1986; Ytterstad, 1996). Consequently, the intensity and aggressiveness of the game should not be underestimated (Meeuwisse et al., 2003), because the contemporary game of basketball puts full emphasises on the speed and power of competitors (Starkey, 2000). Strength and quickness are necessary to control an opponent's position, "muscle" a rebound, or "power" a shot, all of these are prerequisites for a successful basketball career (Starkey, 2000). As the sport grows, in terms of numbers of participants and intensity, so does the number of injuries (van Mechelen et al., 1992). So far, there is not a great deal of data to be found on the injury susceptibility of basketball players in European countries (Benazzo et al., 2001; Colliander et al., 1986; Huguet and Begué, 1998). Furthermore, the data are collected through retrospective investigations and concern either male professional players in France (Huguet and Begué, 1998), young players in Italy (Benazzo et al., 2001) or Swedish elite players (Colliander et al., 1986). Studies concerning the epidemiology of basketball injuries have been very popular in the US (Anderson et al., 2003; Arendt and Dick, 1995; Deitch et al., 2006; Gomez et al., 1996; Henry et al., 1982; Messina et al., 1999; Prebble et al., 1999; Starkey, 2000; Zelisko et al., 1982) and research has focused mainly on professional levels (Deitch et al., 2006; Henry et al., 1982; Hickey et al., 1997; Starkey, 2000; Zelisko et al., 1982) or high school players (Gomez et al., 1996; Messina et al., 1999). We can conclude that the level of evidence in Europe concerning the epidemiology of basketball is very low. Before translating injury data resulting from epidemiological studies performed in the US to European players, we have to bear in mind that the game of basketball as played in the US is different from its European counterpart, which is partly caused by the different rules maintained by the NBA and FIBA (Fédération Internationale du Basketball Association), although we have yet to prove that the difference in rules between both leagues does indeed lead to changes in the injury risk or pattern.

An important purpose of sports injury epidemiology however, is to supply knowledge on injuries that occur frequently and which have serious consequences, and to describe their aetiology in order to provide a basis for preventive measures (van Mechelen et al., 1992). Studies on basketball injury incidence in Europe have been retrospective and reliable information from season-long studies is scarce (Colliander et al., 1986; Benazzo et al., 2001; Huguet and Begué, 1998).

The purpose of this study was to examine the overall incidence of basketball injuries, both acute and overuse, at all levels of play, and to describe the factors associated with ankle sprains and overuse knee injuries. For this purpose a season-long prospective cohort study in a population of competitive basketball players was conducted.

Team recruitment
The goal was to perform research in Flanders on both men and women in senior competitive teams on all levels of play. A total of 12 teams, 2 on each level, were randomly contacted by phone. The teams were invited to participate voluntarily. Two additional teams (regional level men) were informed about the project by the already participating teams, and volunteered to take part.

Subjects
In total, 14 teams of different senior competitive levels professional (only 2 men's teams), national (8) and regional (4) agreed to prospectively register information about both acute and overuse injuries during one basketball season. The total sample consisted of 164 players, 81 men and 83 women, and represented 2.5% of the 'total basketball population in Flanders'. In order for subjects to be included in the study the following criteria applied: the subjects had to play for the senior team and this team had to be the player's original team. Any youth and other players which only joined in occasionally, were excluded.

Organizational aspects
The study was conducted in accordance with the institutional rules for human research and the Declaration of Helsinki for Medical Research involving human subjects. Written informed consent was obtained from each player. During the previous season, a supervisor randomly contacted 14 teams by phone to introduce the project to them. If the team agreed to participate, one team designate (TD), who was either a physician, a physical therapist or a trainer/coach, was appointed within the team. Meetings with these TDs were established in the pre- season to provide the necessary information. During the season preparation period, the TDs organized meetings with all the players to inform them about the practical issues. The TDs were constantly observed and motivated through monthly visits by the supervisor. The TDs completed the injury forms in accordance with the applied injury definitions every time a basketball player sustained an injury.

Injury definitions
An acute injury was defined as being a basketball accident with a sudden, direct cause/onset, which required at least minimum (medical) care including, e.g. ice, tape, etc. and which caused the injured player to miss out on at least 1 training or game session (Verhagen et al., 2004b). Muscle cramps and mild bruises were excluded from the definition. An athlete sustained an overuse injury when he/she suffered a physical discomfort with an insidious onset, which caused pain and/or stiffness of the musculoskeletal system, and which was present during and/or after the basketball activity. Such an injury was only reported if it persisted for at least 3 basketball active days. Malaise and illness were excluded.

Data collection
All information was collected through questionnaires. A standard questionnaire, used to collect the demographic information at baseline (e.g. playing position, years of basketball experience, etc.), was completed by each player. The information about acute and overuse injuries was gathered through an injury registration form, with different questions for acute and overuse injuries because of their intrinsic differences. Each form consisted of 3 parts: part 1 concerned the circumstances involved and was filled out by the injured players themselves, part 2 collected the medical diagnosis in cases where a doctor was consulted and part 3 contained questions on time loss and was completed by the injured players. At the end of the season, a checklist was handed out to check whether all injury forms were accounted for and to determine whether follow-up was necessary when the athlete had not yet gained full recovery at the end of the season.

Exposure time
Exposure time was determined for each team separately, not for each player. To measure the exposure hours, attendance lists for games and training sessions were developed. The attendance lists for training sessions included the dates on which they took place, their duration as well as the number of participating players. Trainers were asked to report whether or not the player was present, or if he did not play for the whole time of the training session. The total number of games played by every team was collected. Exposure time was calculated as follows: each game lasts 40 minutes of actual play and there are always 5 players on the court. Also, the duration of the warming-up for each separate game played was reported.

Injury incidence
The incidence of injury was expressed per 1,000 basketball exposure hours. An injured athlete was reintroduced into the study after recovery from an injury. Consequently, a player could sustain one or more different injuries.

Injury severity
To express the severity of acute injuries, the days of basketball inactivity were calculated and presented in weeks (w) and days (d). The days of basketball inactivity were defined as the number of days for which the player was not able to play or train because of the injury sustained.

The NAIRS (van Mechelen et al., 1992) was used to define minor, moderate and severe injuries. To determine the severity of overuse injuries, the 4-point scale of Puffer and Zachazewski, 1988 was applied. A fifth point was added to the scale for overuse injuries not corresponding to one of the other 4 categories.

Statistical analysis
Relative Risks (RR) [95% CI] were calculated to analyse differences between acute and overuse injuries, men and women, game versus training, offense versus defense and risky movements. To compare the different player positions Odds Ratio (OR) [95% CI] were achieved. RR were determined where exposure time was available and OR were calculated in cases where exposure time could not be estimated.

Subjects
All participating players of the 14 teams completed the study. The sample characteristics are shown in Table 1.

Given the study design, only 32.3% of the players remained injury-free throughout the season, and 37.2% sustained more than 1 injury.

Injury incidences
A total exposure of 16,002h for men and 7,034h for women was reported throughout the 32 week season, during which a total of 226 injuries occurred (Table 2). The players sustained significantly more acute than overuse injuries (RR=3.8 [95% CI: 1.2-2.1]). Relative risks show a significant higher risk for acute than overuse injuries in men (RR=2.9 [95% CI: 1.2-2.5]), but not in women (RR=5.7 [95% CI: 1.0-2.2]). A significant difference is found for the injury incidence between men and women for the total number of injuries (RR=8.0 [95% CI: 1.3-2.3]), acute injuries (RR=5.1 [95% CI: 1. 2-2.3]) and overuse injuries (RR=2.9 [95% CI: 1.3-2.9] with women at higher risk.

Comparing the competitive levels, the relative risks reveal that the national level shows significantly lower risks than the regional level, for the total number of injuries (RR=0.7 [95% CI: 0.5-0.9]) as well as for acute injuries (RR=0.6 [95% CI: 0.4-0.8]). In men, players of the professional level sustained fewest injuries (0.6/1,000 hours) and the national level showed significantly less injuries compared to the regional level (RR=0.5 [95% CI: 0.4-0.8]). In women, a significantly lower risk for acute injuries was found in the national level compared to the regional level (RR=0.5 [95% CI: 0.3-0.8]).

Acute injuries
With a total of 139 acute injuries, the acute injury incidence was 6.0/1,000 hours (95% CI: 5.0 - 7.0) (Table 3). The mean absence from basketball activity after an acute injury was 2w 5d (± 5w 1d), with the longest inactivity period for acute knee injuries (7w 2d ± 9w 1d). With a relative risk of 23.7 (95% CI: 18.6 - 30.1) the risk of acute injuries was higher for game play than for training (Table 2).

In terms of acute injuries, the lower extremity was the most commonly injured body region, with 71 injuries (i.e. 51. 1% of all acute injuries) (Table 3). Ankle injuries (n = 34), all sprains, accounted for most of these. The risk of ankle sprains differed significantly between game and training (RR=27.2 [95% CI: 16.7-44.2]) and between women and men (RR=3.7 [95% CI: 1.9-7. 0]), the risk being higher during game and in women.

Overuse injuries
During the season, 87 overuse injuries were reported (Table 3). The overall incidence of overuse injuries was 3.8/1,000 hours (95% CI: 3.0 - 4.6). Knee (39.1%) and back (16.1%) were the main overuse injury localisations. Analysis reveals that 88.2% of the knee overuse injuries can be defined as 'anterior knee pain' (AKP), according to the definition of Thomee et al., 1999. With 17% of the players seeking medical care, 80% were diagnosed as suffering from the so called 'Jumper's knee', and 20% as having a cartilage lesion. The risk of AKP was significantly higher in female than male players (RR=2.3 [95% CI: 1.1-4.7]).

Ankle sprains and overuse knee injuries
With an equal absolute number of 34 injuries and an incidence of 1.5/1,000 hours (95% CI: 1.0-2.0), ankle sprains and overuse knee injuries were the most common injury types. The incidence differed significantly when compared to almost all other injuries, except for back injuries (acute and overuse) (RR=0.8 [95% CI: 0.5-1.4]) and finger sprains (RR=0.6 [95% CI: 0.4-1.1]).

Ankle sprains
For ankle sprains, re-injuries accounted for 52.9% and new injuries for 47.1%. Most of the re-injuries occurred after 1 year or more (50%). The categories 'more than 1w' (RR=0.1 [95% CI: 0.02-0.6]) and '6 months or more' (RR=0.2 [95% CI: 0.05-0.9]) were significantly less frequently seen. During games, the relative risk for ankle sprains was significantly higher in offense than in defense and was 2.1 (95% CI: 1.0-4.2) (Figure 1). There are no significant differences measured between the risk of ankle sprains and the playing position in offense. Landing on an opponent's foot was responsible for the occurrence of most of the ankle sprains, and significantly more so than non contact mechanisms (RR=2.1 [95% CI: 1.0-4.2]) (Table 4). Jumping tasks carry a significantly higher risk for sustaining ankle sprains compared to sudden changes of direction, running to score (both RR=4.5 [95% CI: 1.7-12.1]), and passing & receiving (RR=9.0 [95% CI: 2.7-30.0]). Not only do contact mechanisms account for the highest injury incidence of ankle sprains, but the cause of all other acute injuries is also significantly higher for contact mechanisms than for non contact mechanisms (RR=2.3 [95% CI: 1.6-3.4]). Relative risks showed that contact with an opponent was significantly more common than all other injury causes considered separately (e.g. contact ball, non contact, etc.) (data not shown).

Overuse knee injuries
The self-reported causes for AKP were: exercise loads being too high (56.7%), monotony of exercise (10.0%) and previous trauma (3.3%). AKP was seen in players of all playing positions (Figure 2). The prevalence of AKP divided by player position showed that forward players had the lowest prevalence (12%), followed by guard players (20%), and center players (26%). A significant difference between the prevalence of AKP was only found between forward and center players (OR=0.5 [95% CI: 0.2-0.9]), with forward players at lower risk. The severity of AKP showed that almost 40% of the players with AKP complained of symptoms that met stage 3 and 4, 20% had symptoms according to stage 2, and only 6.7% showed signs corresponding to stage 1 at the moment of injury reporting (Puffer and Zachazewski, 1988).

This one season prospective cohort study showed an overall basketball injury incidence of 9.8/1,000 hours. A review of the literature on sports epidemiology reveals inconsistencies in data collection methods between different studies and few prospective epidemiological studies in basketball have been performed so far (Gomez et al., 1996; McKay et al., 2001b; Meeuwisse et al., 2003; Messina et al., 1999) but none of these within European countries. Also, there is significant variability in the definitions used (Messina et al., 1999). Because of the strict definitions applied here, we found a higher injury rate than Yde and Nielsen, 1990 at club level; however, McKay et al., 2001b showed a higher injury incidence of 24.7/1,000 hours at national/elite level compared to our results. Incidence observed in our study was also considerably higher compared to other prospective investigations (Meeuwisse et al. 2003, Messina et al., 1999). The injury incidence found at the professional level in our study, is more than twice as high as the injury incidence found by Deitch et al., 2006, Huguet and Begué, 1998 and Starkey, 2000, but is more or less in line with the results of Henry et al., 1982, although our outcome is slightly higher.

Not all participating players were injured, but as seen in our study (67.7%), 44.7% (Meeuwisse et al., 2003) to 69% (Henry et al., 1982) of all players are affected by injuries.

The lower the level of play, the higher the risk of sustaining injuries, as is shown by the relative risks. This difference can be caused by many things, such as e.g. skill level and age, to name a few. The true reason for this difference however, cannot be deduced from this study.

With an incidence of 1.5/1,000 hours in the present study, ankle sprains, along with overuse knee injuries, were by far the most common type of injuries. The high prevalence of ankle injuries is supported by findings from previous epidemiological studies, which have also shown that these injuries are common (Colliander et al., 1986; Deitch et al., 2006; Gomez et al., 1996; Henry et al., 1982; Hickey et al., 1997; Huguet and Begué, 1998) and that sprains of the lateral ligaments make up the majority of these injuries (Apple et al., 1982; Deitch et al., 2006; Henry et al., 1982; Hickey et al., 1997). Knee injuries have also been recognized as being common in basketball (Colliander et al., 1986; Deitch et al., 2006; Henry et al., 1982; Hickey et al., 1997; Huguet and Begué, 1998) and patellar tendonitis or Jumper's knee has long been known to be particularly prevalent in high level basketball players (Henry et al., 1982; Hickey et al., 1997; Lian et al., 2005; Starkey, 2000; Zelisko et al., 1982). In our study, the injury incidence for ankle sprains was similar compared to overuse knee injuries. This is why further analysis in this paper has focused on the general and sports specific circumstances involved in these two kinds of injuries.

In our study we were not able to compare the risk for new or recurrent ankle sprains, since we had no information on previous ankle sprains of the uninjured sample. The injury rate of re- (47.1%) and new injuries (52.9%) was comparable. McKay et al., 2001a found that a history of ankle injuries was the best predictor for the occurrence of ankle injuries. Meeuwisse et al. (2003), however, could not confirm a significant difference between the risk for re- or new injury.

Significantly more ankle sprains were sustained during games. Meeuwisse et al. (2003) also found games to be more dangerous for the occurrence of ankle sprains, but only for ankle sprains resulting in 7 or more session losses. Ferretti et al., 1992 suggested that the increased frequency of injuries in games is caused by of the high-intensity level of competition and because of the maximum effort that is expended during games. The athlete is at maximum risk, which might make athletes more vulnerable to injury.

Our results showed that ankle sprains were particularly seen in offense, but no significant difference could be found between the different playing positions. The study by Meeuwisse et al. (2003) revealed lowest risk for forwards and highest for center players, but could not prove a statistically significant difference.

Landing on an opponent's foot in this study has been identified as the first major inciting event causing ankle sprains, which is mainly the result of jumping tasks. In second place, ankle sprains were brought on by sudden changes of direction. An earlier study performed by McKay et al., 2001a, also showed that the two main risk factors for ankle sprains in basketball are landing or jumping on someone else's foot or making a sharp cutting manoeuvre.

Not only in ankle sprains, but also in all other acute injuries, contact mechanisms were reported significantly more as a cause of injury, and contact with an opponent was seen significantly more than non contact mechanisms. Also Meeuwisse et al. (2003) registered more injuries resulting from contact than from non contact, and Zelisko et al., 1982 found a high prevalence of injuries caused by contact with another player.

Our study shows a high incidence of AKP, and Jumper's knee was the most common diagnosis. The main reported causes for AKP were high training loads or monotony of exercise. Anderson et al., 2003 demonstrated the existence of a temporal relationship between training load and injury, which suggested a causative link. An increase in injuries occurred during times of increased training loads. Center players seem to have a significantly higher prevalence of AKP than forward players.

This was the first prospective cohort study on the epidemiology of basketball injuries performed in a European country. As in all epidemiological studies, there were pitfalls we tried to avoid. We had to rely on the motivation and honesty of the players and TDs for filling out every injury sustained and answering the injury definitions. It appears that our injury incidences were quite high compared to those found in the existing literature; this, we think, sheds a positive light on our research. The high injury rate found is also a result of the injury definitions used in this study. The importance of carefully defining injury, meticulously collecting data, characterizing exposure and calculating risks and rates has previously been well described (Gomez et al., 1996). Since we consider injury definitions as the gateway for injury reporting, injury definitions used here were very strict and included minor, moderate as well as serious injuries. This could also explain the high injury incidence found in the present study. The investigated sample can be considered rather small. In research however, one has to decide which methodological issues will be applied, and whether they appropriately answer the questions and purposes put forward at the beginning of the study. The small sample size allows for a close follow-up, resulting in a zero drop-out rate and an injury reporting which covers minor and severe injuries as well as acute and overuse injuries. Although a longer data collection time span would have provided more injury data, the period of investigation as determined at the start of the study was set up to ensure close follow-up of this kind of population. Exposure measurement for each individual should be taken into account in the future to calculate the differences between the playing positions using RR instead of OR. Also the reporting of previous ankle sprains is a valuable tool which should be considered in the future.

• Ankle sprains are the most common acute injuries in basketball with the inciting event being landing on an opponent's foot or changing direction.
• Anterior knee pain is the most common overuse injury. Etiologic factors are well described in literature, but prevention strategies are lacking.
• Acute knee injuries account for the highest inactivity and should therefore also be prevented.
• Most of the injuries are due to contact mechanisms and therefore the definition of basketball as a non contact sport is questionable.
• Highest injury risks are found in women and in the lower levels.