Higher age and present injury at the start of the season are risk factors for in-season injury in amateur male and female football players—a prospective cohort study

Inclusion criteria were teams (1) participating in the adolescent or adult 2020 series (male 5th–8th leagues (out of 8 leagues), female 3rd–5th leagues (out of 5 leagues), male and female 16–19 series) in one regional district (Östergötland) in Sweden), (2) with at least two scheduled training sessions per week. All players 14 years and older were eligible, including players who were injured at baseline.

Coaches for eligible teams (n = 251) were approached via e-mail and telephone and received oral and written information about the study before agreeing to participate. In teams where the coach had agreed to participate, all players (and guardians for players < 15 years) then received written information about the study and were inquired for participation in the study (Fig. 1). Participation was voluntary, and individual players could decline to partake in research even though their team was involved in the study. In order to initiate the online baseline form, the player had to verify that they had read and understood the information about study participation, and subsequent response to the questionnaires was taken as consent to participate.

Players responded to a baseline questionnaire (Supplementary Appendix 1) at the start of the football season including questions about previous (previous season or earlier) and/or present injury in any of the locations ankle, knee, hamstrings, or groin (primary injury outcomes of the randomised trial), if they participated in other sports during the football season, if they performed strength- and/or conditioning training during the football season and questions about injury beliefs, importance of success in sports and training load (response model 7-level Likert scale): “I expect I will sustain an injury sometime during this football season” (extremely likely–extremely unlikely); “What significance does sporting success have for your interest and motivation in sports?” (little importance–great importance); “How do you rate your current total training volume?” (extremely low–extremely high); How do you rate your current total training load (quantity and intensity of training? (extremely low–extremely high); How do you rate your current match load (frequency and time on the pitch)? (extremely low–extremely high); and I have a good balance between training/match load and recovery (strongly disagree–strongly agree).

Players responded to weekly questionnaires about occurrence of injury based on the Oslo Sports Trauma Research Center Overuse Injury Questionnaire (OSTRC-O2) [10], and exposure to training and matches. Conditional branching was used to allow for targeted questions, and the players answered a minimum of 1 (no football participation due to reasons other than injury or illness, e.g. holidays) and a maximum of 28 questions (for players who had participated in football activity and had sustained a new injury) each week. Multiple injuries could be reported. When a new injury was reported, additional questions were given regarding the nature of the injury (i.e. when the injury occurred (training/match/other activity), injury location, whether it was sudden or gradual onset). In case of sudden onset, they received an additional question about the type of injury. The questionnaire covered all physical complaint injuries. All sudden-onset injuries occurring during football training or matches and all gradual-onset injuries regardless of whether symptoms first appeared during football or other activities were included. Players who reported time-loss or substantial injury to the groin or hamstrings (sudden onset or gradual onset), knee or ankle (sudden onset) were contacted via telephone by a study physiotherapist who asked about the injury and filled in a standard injury report form. Injuries were defined in line with the International Olympic Committee consensus statement [5].

All questionnaires were distributed via online software (esMakerNX3 V 3.0). The baseline questionnaires, with two reminders, were distributed after each team’s acceptance to participate, and weekly questionnaires were distributed on each Sunday evening, with reminders sent to non-responders on Tuesday and Thursday the week after.

For calculation of the OSTRC severity score, the responses to each of the four questions of the OSTRC-O2 were transformed to a numerical value from 0 to 25, and these were summed to calculate a severity score from 0 to 100, where 0 represents no problems and 100 represents maximum problems [11]. Injuries were defined in line with the International Olympic Committee consensus statement [5] and covered any physical complaint injuries (irrespective of need of medical attention or time-loss), sudden- and gradual-onset injuries, time-loss injury (injury with reduced participation or absence from football training and/or matches as reported in question 1 of the OSTRC-O2), and medical attention injury (injury where the player sought medical advice or treatment). Substantial injury was defined as an injury with moderate or severe modifications in participation in football training and matches and/or moderate or severe effects on football performance, or inability to participate in football [10].

All analyses were conducted for male and female players separately. Descriptive statistics are presented with sum, mean ± standard deviation, and n (%). Weekly injury prevalence was calculated as the number of player reports where a player reported new or ongoing injury divided by the total number of eligible player reports that same week. Weekly injury prevalences are displayed as a moving average for 2 consecutive weeks, i.e. data for a certain week is the average of the current week and the preceding week (Fig. 2). Seasonal injury prevalence and substantial injury prevalence are presented with 95% confidence intervals (CI) (Table 1).

Injury incidence rates were calculated as total number of injuries divided by total exposure hours. Both new injuries and re-injuries were included, but injuries that were persistent from baseline were excluded. Seasonal injury incidence rates and time-loss incidence rates are presented as injuries per 1000 h exposure with 95% CI (Table 1).

Potential baseline risk factors (age, strength- and/or conditioning training beside regular football training, participation in other sports during football season, perceived importance of sporting success, self-rated training and match load, balance between training/match load and recovery, previous/present injury at start of the season, injury beliefs) for injury were analysed with Poisson regressions. The basic model included the number of new injuries during the season as the target variable, the natural logarithm of total exposure hours during the season as the offset variable, team as a random factor using variance component as covariance structure, and intervention group (extended Knee Control, Adductor group, Comparison group) as a fixed factor. Log link function was used in the models to estimate the risk of injury. The analyses were performed in two steps, based on the basic model. In step 1, each risk factor was entered as a fixed factor in the basic model. In step 2, significant risk factors from step 1 were entered in a final multivariable model (Supplementary Appendix 2a and 2b). Variance inflation factor (VIF) of each potential risk factor was calculated to check for multi-collinearity in the multivariable model (VIF values, range 1.1–2.1). As a sensitivity analysis, we analysed potential baseline risk factors for the total cohort and tested for two-way interactions of player sex and each risk factor. The level of significance was set at p < 0.05.

IBM SPSS Statistics for Windows (V28.0. Armont, New York) was used for all analyses. An experienced statistician was responsible for preparing the database and data analyses.

An a priori sample size calculation was made for the randomised trial [24], while no sample size calculation was made for the current analyses.

Previous or present injury at baseline was reported by 77.7% of male players and 79.2% of female players. Males reported a total of 95 injuries spread over 262 injury weeks, weekly prevalence 18.0% (95% CI 15.9–20.3) (Table 1). Based on a 20-week season and a squad of 15 players, there would be 20 injuries (12 time-loss) and 54 total injury weeks (33 substantial injury weeks) per team-season. Females reported a total of 350 injuries spread over 1206 injury weeks, weekly prevalence 23.9% (95% CI 22.6–25.3) (Table 1). Based on a 20-week season and a squad of 15 players, there would be 21 injuries (12 time-loss) and 72 total injury weeks (39 substantial injury weeks) per team-season. Weekly injury prevalence during the season with moving average for 2 consecutive weeks is displayed in Fig. 2.

Gradual-onset injuries accounted for 57.3% of the injuries in males and 65.9% in females. Among sudden-onset injuries, joint sprain/dislocation represented 17.6% of all injuries in males and 21.0% in females, while muscle injury accounted for 21.8% and 9.4%, respectively (Table 2).

The hip/groin was the most injured body location in male players, and the knee in females. Substantial injuries were most common in males in the hip/groin, ankle, posterior thigh, and knee, and in females in the knee, ankle, and lower leg/Achilles tendon (Fig. 3a). Time-loss injuries were most common in males in the knee, ankle, hip/groin, and anterior thigh, and in females in the knee, ankle, and lower leg/Achilles tendon (Fig. 3b). A matrix of the seasonal injury prevalence for each body location in relation to the OSTRC severity score is displayed in Fig. 4.

Step 1, basic model: For males, age was a significant risk factor for injury, with 7% increased incidence for each year of age increase. Present injury at the start of the season increased the risk of injury twofold. Estimation of present training volume was associated with injury incidence with 15% reduced injury rate for each unit of change towards high training volume (Table 3).

For females, age was a significant risk factor for injury, with 3% increased incidence for each year of age increase. Previous injury and present injury at start of the season increased the risk of injury 1.5 times and 1.8 times, respectively. Injury beliefs were associated with injury incidence with 7% reduced injury rate for each unit of change towards a belief that it is unlikely to sustain an injury (Table 3).

Step 2, multivariable model: For both males and females, only age and present injury were still significant in the multivariable model (Supplementary Appendix 2).

The sensitivity analysis of baseline risk factors for the total cohort showed that age, present injury at baseline, and estimation of present training volume were significant risk factors, with no evident difference between male and female players according to the non-significant two-way interactions of players’ sex and each risk factor (Supplementary Appendix 3).

A strength of the study is that a validated questionnaire previously used in sport surveillance was used to capture injuries [10, 11]. However, the baseline questionnaire containing questions about potential risk factors consisted of single Assessment Numeric Evaluation (SANE) questions that has not been validated and detailed information about the constructs cannot be determined. Though, in general, single-item scales have acceptable psychometric properties [43] and were appropriate in this prospective study with repeated surveys to limit response burden. Data are self-reported which entails uncertainty as to the validity of injury and exposure data. However, training and match exposure and injuries were reported weekly to limit recall bias. Moreover, players who reported time-loss or substantial injury to the groin or hamstrings (sudden onset or gradual onset), knee or ankle (sudden onset) were contacted via telephone by a study physiotherapist who asked about the injury and filled in a standard injury report form to validate the injury.

Other factors than those assessed in the current study may contribute to the injury risk. Risk factors for injury are complex and multifactorial, and a player’s injury profile depends on various internal risk factors and exposure to external risk factors [6, 7].

Clinical implications of the present study are that based on the considerable injury burden in amateur football, preventive measures should have high priority and injury-risk mitigation strategies may be differentiated in male and female teams, and also based on individual player characteristics. Due to differences in injury patterns between sexes, the primary target areas for IPEPs could be somewhat different between male and female amateur football players, i.e. more emphasis on the hip/groin in males, and emphasis on the knee in females. Moreover, since previous injury is a significant risk factor for injury during the season, a plausible interpretation is that rehabilitation after injuries might be insufficient, and players may start the season not fully physically and/or mentally prepared for play. Amateur football players with previous injury or present injury at the start of the season might need special attention from the coach and/or medical personnel to achieve individually adapted exercises and adjustment in training and/or workload. These adaptations may be of particular importance for older players.