One hundred and twenty three clubs agreed to participate in the study, and we block randomised these, with four clubs in each block to an intervention or control group. To reduce potential confounding, we matched the clubs by region, playing level, and sex and number of players. The statistician (IH) who conducted the randomisation was not involved in the intervention. Box 1 gives details of the procedure used to recruit clubs.
We informed clubs allocated to the intervention group that they would receive a programme of warm-up exercises used to prevent injuries. We asked the clubs in the control group to do their training as usual during the season and informed them that they would receive the same programme as the intervention group at the start of the subsequent season.
Box 1: Recruiting clubs to the study
- All 145 clubs in the 16 year and 17 year divisions from central and eastern Norway, organised by the Norwegian Handball Federation, received an invitation to participate in the study during one eight month season (September 2002 to April 2003)
- The clubs practised one to five times per week and played between 20 and 50 matches during the season, depending on their ability and ambition
- The clubs were recruited from June to August 2002 through the website of the Norwegian Handball Federation, and a letter with information about the purpose and the design of the study went to the coaches, who also informed the players
The warm-up programme was developed by medical staff from the Oslo Sports Trauma Research Center and coaching staff from the Norwegian Handball Federation, and its feasibility had been tested in four clubs during the previous season. The programme included four different sets of exercises, each of increasing difficulty.
At the start of the league season (September), the clubs in the intervention group received one visit from an instructor from the handball federation. In addition, instructors followed up the clubs with a visit midway through the season (January). The instructors had been familiarised with the programme during a two hour seminar, in which they received theoretical and practical training on how to conduct the programme. The clubs received an exercise book, five wobble boards (disc diameter 38 cm; Norpro, Notodden, Norway, 2000) and five balance mats (40×50 cm2, 7 cm thick; Alusuisse Airex, Sins, Switzerland, 2000). The coaches were asked to use the programme at the beginning of every training session for 15 consecutive sessions and then once a week during the remainder of the season.
The main focus of the exercises was to improve awareness and control of knees and ankles during standing, running, cutting, jumping, and landing. The programme consisted of exercises with the ball, including the use of the wobble board and balance mat (box 2, , and ), for warm up, technique, balance, and strength.
Top: mat exercise. Middle: wobble board exercise. Bottom: mat pair exercise
Fig 2 Example of a strength exercise (“Nordic hamstring lowers”). Top: start position; a partner holds around the player's ankles. Bottom: The player falls slowly forwards, using hamstrings to resist the fall against the floor as long as possible (more ...)
The players were encouraged to be focused and conscious of the quality of their movements, with emphasis given to core stability and position of the hip and knee in relation to the foot (the “knee over toe” position). They were also asked to watch each other closely and give each other feedback during the training. They were instructed to spend 4-5 minutes on each exercise group for a total duration of 15-20 minutes.
Box 2: Programme of warm-up exercises used to prevent injuries
(30 seconds and one repetition each)
Jogging end to end
Backward running with sidesteps
Forward running with knee lifts and heel kicks
Sideways running with crossovers (“carioca”)
Sideways running with arms lifted (“parade”)
Forward running with trunk rotations
Forward running with intermittent stops
(One exercise during each training session; 4 minutes and 5×30 seconds each)
Planting and cutting movements
Jump shot landings
(On a balance mat or wobble board, one exercise during each training session; 4 minutes and 2×90 seconds each)
Passing the ball (two leg stance)
Squats (one or two leg stance)
Passing the ball (one leg stance)
Bouncing the ball with eyes closed
Pushing each other off balance
Strength and power
(2 minutes and 3×10 repetitions each)
One quadriceps exercise:
Squats to 80° of knee flexion
Bounding strides (Sprunglauf)
Jump shot—two legged landing
“Nordic hamstring lowers” (2 minutes and 3×10 repetitions each)
Data on injury and exposure were reported by the physiotherapists using a web based database in which all the data were coded anonymously. At the end of the season, the recorded data were confirmed, or if necessary corrected, by the coaches. Box 3 shows the definitions we used in registering injuries.
We defined the primary outcome as an acute injury to the knee or ankle. A secondary outcome was defined as any injury to the lower limbs. We also included secondary analyses of injuries overall (including all injuries) and injuries to the upper limb. We included all injuries reported after an intervention club had completed the first session of the training aiming to prevent injuries (and from the same date in the control clubs randomised in the same block), to compare the number of injured players and incidence of injury between the intervention group and the control group.
The number of injured players was based on data from individual players and the incidence of injuries on summary data of injuries and exposures for the whole group. Data on players who dropped out during the study period were included for the entire period of their participation.
Ten research physiotherapists who were blinded to group allocation recorded injuries in both groups, using definitions (box 3) and a standardised injury questionnaire described in our earlier study (Olsen OE, Myklebust G, Engebretsen L, Bahr R. Injury pattern in youth team handball: a comparison of two prospective registration methods. Submitted for publication to Scand J Med Sci Sports).
The physiotherapists were in contact with the coaches at least every month to record injured players and exposure data. They interviewed injured players, either in person or by telephone, and in most cases within four weeks (range one day to four months). They were responsible for roughly the same number of clubs from each of the groups (11 to 13 clubs each).
The coaches of the clubs receiving the intervention recorded compliance on a designated form as the number of injury prevention sessions, the duration of each session in minutes, and the average attendance of the players (in per cent). At the end of the season we also obtained information on prevention training conducted by the control clubs, including the types and volume of exercises used.
In youth team handball, the incidence of acute injuries to the knee and ankle is estimated to be 12 per 100 players per league season.11,15
From a pilot study conducted to determine the incidence of injury during the previous season (submitted for publication), we estimated that the cluster effects for club randomisation gave an inflation factor of 2.0 based on a cluster size of 15 and an intracluster correlation coefficient of 0.07. We then calculated that to achieve 90% power with α = 5% to detect a relative risk reduction of 50%, we would need 915 players in each group. Therefore, when we initiated the trial, we were hoping to include 60 clubs in each group (a total of 120 clubs; with an average of 15 players in each club).
We used Stata, version 8.0 (Stata Corporation, Lakeway Drive, Texas, 2003), for the statistical analysis. We undertook all statistical analyses according to a prespecified plan. We used the relative risk of the number of injured players according to the intention to treat principle to compare the risk of an injury in the intervention and control groups. Cox regression was our analysis tool for the primary outcome as well as the secondary outcomes, and we used the robust calculation method of the variance-covariance matrix,16
taking into account the cluster randomisation. We tested rate ratios with Wald's test. We used one way analysis of variance to estimate the intracluster correlation coefficient to obtain estimates of the inflation factor for comparison with planned sample size. We used the inverse of the difference between percentages of injured players in the two groups to calculate the number needed to treat to save one injury. We calculated exposures to training and matches and incidence of injury as described in our earlier study.
We used a z test based on the Poisson model to compare the rate ratio between the two groups (intervention v control), sex (female v male), severity of injury (slight, minor, moderate, major), and club activities (match, training).
Compliance and incidence of injury are presented as means with standard errors. Relative risk and rate ratio are presented with 95% confidence intervals. We regarded two tailed P values ≤ 0.05 as significant.