This is a one-season study in the 2025 football season, beginning in the pre-season (March) and ending after the competitive season (November). Football coaches and players are included in the study. A sub-sample of coaches will also be included in a qualitative sub-study after the 2025 season. As part of a process evaluation, club representatives, such as club chairmen, sports directors and youth managers, will also be targeted in the same district, as well as fitness coaches delivering the interventions to the coaches. Coaches will be targeted with most of the interventions, but players and club managers will also benefit from some of the support interventions.
Fitness coaches in a nation-wide organisation (Friskis\&Svettis) will be the primary intervention-deliverers and will be educated on the Knee Control+ programme exercises and support interventions to be able to hold workshops at their venues as well as to make site visits in the teams. The research group will be responsible for the education of fitness coaches and will deliver the digital workshops to the coaches and manage the social media account.
Co-design in development of the support material has been employed. During the 2024 season, a working group with representatives for coaches and players (for practical reasons age ≥ 18) was initiated. This working group (two coaches, one former player) provided feedback and new ideas during the development of the support material and the questionnaires for coaches and players.
The workshop concept was tested three times with fitness coaches leading workshops for football coaches during the autumn of 2024 in a district not included in the main study. One site visit was also accomplished and the fitness coaches gave feedback of their experiences to the research group after this.
In the planning and execution of the study, the research group will have regular contact with representatives for football clubs, the district football associations, the Swedish football association and the fitness coach association (Friskis\&Svettis) to get their strategic input on the work being done.
Both group constellations have mainly been active during the planning for the study, but some participants may also be invited to give their view on the results and the interpretation of results after the 2025 season. After the 2025 season and potentially before upscaling of the study, the workshop strategies will be followed-up with the fitness coach association and evaluate whether any changes are needed before opening up for national dissemination.
Data will be collected primarily through web-based questionnaires covering implementation outcomes, behavioural outcomes, injury outcomes, adverse events, as well as a process evaluation.
Sample size calculations for the primary outcome were completed using the Shiny CRT online tool for binary outcomes, based on a logistic regression model. Calculations assumed 3 teams per cluster (club); a coefficient of variation equal to 0.9 was applied to account for variation in cluster sizes. An exchangeable correlation structure was assumed, and base case calculations used an intracluster correlation (ICC) of 0.02. Sensitivity analyses considered ICC values between 0.01 and 0.05. The estimated proportion of weeks where teams use the Knee Control+ in the CG is 0.5 (50%), and in the IG 0.8 (80%) based on published data. Considering a design effect (DE) of 1.04, a cluster size of 3 teams will allow for detection of a between-group difference in the primary outcome of at least 0.3 (30%) with 80% power at the 5% level of statistical significance, with total 29 clusters (85 teams). For the same difference in outcome, varying the ICC results between 0.01 and 0.05 rendered sample sizes between 28-30 clusters (83-90 teams). Therefore, the aim is to include minimum 90 teams.
For the main injury outcome, ≥7 days time-loss injuries (secondary outcome), sample size calculations were completed based on a Poisson regression model using the Shiny CRT online tool. Calculations assumed 15 players per team (cluster in the secondary analysis), a coefficient of variation equal to 0.9 was applied to account for variation in cluster sizes. An exchangeable correlation structure was assumed, and base case calculations used an intracluster correlation (ICC) of 0.02. Sensitivity analyses considered ICC values between 0.01 and 0.05. The estimated injury incidence in the CG is 0.2 (20%), and in the IG 0.1 (10%) based on published data. Considering a design effect (DE) of 1.28, a cluster size of 15 players will allow detection of a between-group relative difference in the secondary outcome of at least 0.5 (50% injury reduction) with 80% power at the 5% level of statistical significance, with total 41 teams (607 players). For the same difference in outcome, varying the ICC results between 0.01 and 0.05 rendered sample sizes between 37-54 teams (541-806 players). This equals to total 14 clusters (range 12-18 clusters) at a club level.
Data on any football related injury will be collected regardless of need of care or absence from football training or matches ("all physical complaints" injury definition).
The primary outcome, use of Knee Control+, will be analysed at team-level as the proportion of weeks with sufficient use of the programme and compared between intervention group and control group using logistic regression. The limit for sufficient training frequency is set at Knee Control+ training at least 2 times per week. Since a gradual start of injury prevention training is preferred for beginners, for players 10-12 years use of ≥3 of 6 main exercises is deemed sufficient, whereas for players \>12 years, use of ≥4 of 6 main exercises is deemed sufficient to be defined as use of Knee Control+. Other measures of implementation will also be compared between groups and predominantly with parametric statistics.
Injury incidence rate (injuries per 1000 athlete exposures) and monthly prevalence rate (number of injured athletes/total number of athletes each month) will be presented with 95% confidence intervals, and incidence rate ratio and prevalence rate ratio will be calculated and compared between groups (according to intention-to-treat) using Generalized Linear Models with Poisson distribution, log link and the natural logarithm of total athlete exposures or total eligible weeks as offset denominator variables.
Qualitative interviews will be analysed using qualitative content analysis and an inductive approach.
