Methods
Sixty-two male professional soccer players, with ages between 18 and 36 years, participated in the present study. The athletes were recruited from the first and second division soccer teams of the largest Brazilian State Championship organised by the Brazilian Football Confederation (CBF), affiliated to FIFA.
Criteria for inclusion
All of the participants were registered at CBF; no musculoskeletal injury in the knee or thigh at the time of evaluation; and with scores ≥84 in functional performance questionnaires.
All subjects were informed about the objectives of the study and agreed to participate by providing their free and informed consent.
Evaluations
The functional performance evaluations were carried out as described below at the beginning of the preseason of the championships in the years 2013, 2014 and 2015. The training period before the start of the first official match of the competition was also included. The evaluations were conducted by trained physiotherapists.
History of injuries and previous professional activities
The data collected for the study referred only to the injuries in the knees and thighs that occurred in the last two years. Data on previous moderate or severe injuries in the knee–thigh complex that required the player to miss participation time of at least three matches, and 15–25 days of absence from practice, were recorded to provide current evidence on significant changes. The history of the athletes’ prior professional experience was also recorded for comparison with their current performance, with or without reports of musculoskeletal disorders and/or changes in the tests.
Lower extremity functional tests
The functional tasks used in this investigation were the single-hop test (SHT) and the cross-over hop for distance test (COHDT) with sensitivity (SHT: 53.0, COHDT: 88.0; 95% CI) and specificity (SHT: 72.0, COHDT: 47.0; 95% CI) already described by Logerstedt.34 These tests were applied to determine cut-off scores. Next, the athletes were placed on the kinetic dynamometer for the measurement of several variables related to muscular performance.
The SHT and COHDT, as described by Noyes et al, 35 were used to assess the symmetry of the lower limbs during horizontal hops with displacement postligament injuries. Maximum distance was measured from the posterior heel of the last hop.35
Three collections were made from each test, and the mean value was calculated for analysis. When the difference in symmetry between the limbs was <20%, the athlete was considered fit to perform the muscle performance tests. If the asymmetry between the limbs was >20%, the athlete was referred to CT exam and to a more detailed orthopaedic evaluation to investigate the possibility of a more acute lower limb musculoskeletal injury.
Isokinetic evaluation
Muscle activity of the players was evaluated using an isokinetic dynamometer Biodex System V.4 (Biodex Medical Systems, New York, USA) with sensitivity (45.5; 95% CI) and specificity (97.0; 95% CI) already described by Almosnino et al.36
Participants performed a warm-up exercise on an ergometric bicycle for 10 min. Each player was given a brief explanation to become familiar with the tasks. Pain intensity was assessed before and after the test using a visual analogue scale. This instrument is a semiobjective score system used to quantify the intensity of pain.37
A test was performed for each knee with the angle velocity of 60°/s (six repetitions of flexion and extension of the knee during the concentric phase with a rest interval of 90 s between sets). The evaluators encouraged the athlete during the entire test with verbal commands.
Isokinetic parameters with their respective values were obtained using a computer program to determine the peak torque (PT) and hamstrings/quadriceps ratio (H/Q ratio).
Characteristics of the athlete
Some characteristics that may be related to physical quality and effective performance of the athlete during matches were analysed to determine the relationship between the performance in clinical trials and the injury incidence rate throughout the season. These characteristics included age,38 body mass index (BMI),39 lower limb dominance, defined as the preferred leg used to kick the ball40 and position on the field: goalkeeper, defender, full-back, midfielder or centre-forward.41
Season injury follow-up
The thigh and knee complex injuries were diagnosed by the medical team from the soccer club and tabulated. Injuries to knee joint cartilage, meniscus and/or other ligaments that caused the player to miss training sessions or matches were computed together and separately. The injuries in other portions of the leg were not considered for this study.
A time-loss injury was defined by the FIFA Medical Assessment and Research Centre as ‘an injury that resulted in a player being unable to take a full part in future training or match play’.42 About 80% of injuries occur without physical contact or by the influence of external forces to the body. The high demand the musculoskeletal system of the lower limb is submitted to refers to the psychological factors.42
Statistical analysis
The qualitative variables were summarised as frequencies and expressed in percentages. The quantitative variables were described as measures of central tendency and dispersion, according to their distribution.
To define the exposure to injury, the athletes were classified as those with ‘no muscle imbalance’ or those ‘with muscle imbalance’, according to the relationship considered potentially predictive of injury: H/Q ratio >64% or <55% and differences >10% in peak of torque of the knee flexors and extensors between the limbs were also considered.
The incidence of injuries was defined as the number of new musculoskeletal injuries that occurred in each muscle groups or structures comprising the ligament complex and meniscus during the official championship matches subsequent to the evaluation (from February to May).
To estimate the raw OR adjusted for covariate of interest to detect risk factors for injury, a logistic regression model with random effects belonging to the class of generalised estimating equations models was proposed. This model represents the linear relationship between the response variable and the covariates.43
Scores for the risk of injury derived from the association between risk factors
Following the definition of the OR for each variable evaluated independently for the risk of injury during the preseason, a nomogram or score was constructed to calculate the injury predictions for each athlete and to give a total point score. For each of the variables, the corresponding number of points was read from the top scale. These were then summed to give a total point score, which was then readily translated into a probability of risk for the lower limb injury during the championship matches.
The nomogram outperforms the traditionally used staging systems because it considers multiple commonly available prognostic variables simultaneously, including the identification, calibration and stratification of risk factors according to a particular outcome. The nomogram provides a simple graphical representation of sophisticated statistical prediction models and has been accepted as a reliable tool for predicting clinical events and also for using in scientific studies.44 The information collected during the preseason was integrated in the nomogram. Consequently, the outcomes could be analysed more effectively to help guide prevention and intervention strategies.
Combining the risk factors for injuries in the proximal portion of the lower limbs
Blanch and Gabbet45 evaluated the relationship between the current weekly workloads the athletes were exposed to and the mean workload of the four previous weeks and established an equation to measure the probability of developing an injury at a later date. These findings have encouraged us to create a similar, consistent and easy-to-use instrument to gather a combination of risk factors, considering that a single risk factor would not be enough to predict injuries.
The present study was based on a previous study carried out by our research group. The risk of injury in Brazilian football players during the season was evaluated using preseason data collected from three constructs:
Professional profile, which included his professional history, current team and playing position;
Physical profile data, which consisted of BMI, age (the athletes were divided into three age groups: group A, ages between 18 and 24 years and 11 months; group B, ages between 25 and 29 years and 11 months; and group C, athletes over 30 years) and lower limb dominance.
Functional profile, which consisted of functional performance data from the lower limb. The isokinetic variables analysed were H/Q ratio and PT of the knee extensor and flexor muscles, SHT and COHDT.
According to our preliminary findings on risk of injury and the evaluation of each item referred to in the functional performance profile, we found a strong relationship between the incidence of injuries throughout the season in athletes who showed an H/Q ratio <55% or >64%. This was an interesting finding, considering that the use of H/Q ratio cut-off for lower limb muscle imbalance was set at <0.6 (60%),46 suggested by previous studies, still retained a large number of ‘false negatives’ for the athletes who have suffered injury, and with H/Q ratio of 60%.
The range between 55% and 64% in H/Q ratio proved to be more realistic in determining athletes at risk of injury that were out of this range. Therefore, the lower and higher values were used as references to obtain the equation score of the H/Q ratio in our equation. The PT values of flexor and extensor muscles were scored based on our previous findings of high risk of injury observed in the lower limb of athletes with PT values of flexor or extensor muscles of at least 10% less than those of the contralateral leg. In other words, when the difference of the PT values of flexor or extensor muscles was >10%, the score was different for the weakest lower limb compared with the limb with higher PT values.
The advantages of using a nomogram and its score system in football
Based on a parallel observational study to establish the risk factors, our hypothesis was that, with the use of a particular statistic methodology, a strong agreement could be established to score each variable, whether present or not, according to its relevance on the observation of individual risks. Thus, the combination of risk factors would produce a score for the level of exposure to predict future injuries. Even the variables with no significant relevance in the original study, although based on reports in the literature with correlation indexes for future injuries, obtained adequate scores for their findings. A score was obtained for each item of the profiles and for the strength ratio using a combined evaluation to obtain the score for probability of injury. The concordance index for the nomogram was 84%,44 which was enough to confirm the correlation between the score and the probability of injury. Figure 1 shows the nomogram and the variables, and table 1 shows the scores. Table 1 also shows the values for each score and the final score obtained for each lower limb and the probability of injury.
Nomogram, the variables and the scores. BMI, body mass index; H/Q, hamstrings/quadriceps; PT, peak torque.
The complexity of determining multiple risk factors and reflecting on their interactions in the athlete’s body to estimate which player deserves more attention and preventive care has hindered the introduction and choice of preventive protocols and athletes in need of extra support to meet the physical demands during the preseason.
The failure of the proposed preventive approach was the result of the need to optimise training performances in a short period of time. Therefore, those prevention protocols based on prior studies should be adapted to the original to reduce compliance and, consequently, influence the prevention effectiveness. Coaches, athletic trainers and physiotherapists with low compliance to carry out the previously proposed protocols that have produced positive outcomes47 showed the need for instruments to facilitate the tracking and quantification of athletes at risk of injury. The use of a nomogram that combines risk factor variables helps understand such risks and contributes to the work of sports medicine.
The use of a nomogram for football players
The nomogram proposed should estimate the score for each lower limb to establish the probability of injury based on the association of variables obtained from the athlete through the score defined for each answer and described in table 1.
H/Q ratio: If the H/Q ratio was <55% or >64%, the score was 7. If the H/Q ratio was between 55% and 64%, the score was 0.
Difference in PT of the knee extensor >10%: If the PT of the knee extensor was at least 10%, compared with the contralateral limb, the score was 100, or 0 when the difference was not >10%.
Difference in PT of the knee flexor >10%: If the PT of the knee flexor was at least 10%, compared with the contralateral limb, the score was 93, or 0 if the difference was not >10%.
BMI: Score was 10 if BMI was within normal range (18.50–24.99), 0 if it was considered overweight (25.00–29.99) or obese (≥30.00).
Age group: Score was 0 if the athlete was in group A (18–23 years and 11 months); 4 for group B (24–29 years and 11 months); and 15 for group C (≥30 years).
Dominance of the lower limb: Score was 5 for the non-dominant leg or 0 for the preferred leg used to kick the ball.
Professional history: Score was 29 if the athlete had played in the previous season and 0 if he was playing in a youth team.
Previous injuries: Score was 13 if the answer was yes (moderate/severe injuries) in the last two years or 0 if no injury.
Current championship level: Score was 23 if the elite team was playing in the championship or score 0 if the athlete was playing in a youth team.
Playing position: Score was 13 for goalkeeper, 43 for defender, 23 for full-back, 0 for midfielder and 11 points for centre-forward.
For the functional hop test (SHT and COHDT): The hop distances were set at 10 to 10 cm (SHT) and from 50 to 50 cm (COHDT) for the score. However, if the results achieved during the tests ranged between these values, the higher value was considered to establish the test score for these tasks (table 1).
Limitations of this study
Some of the main limitations of our study include the use/application of the nomogram only in players who participated of the largest Brazilian State Championship. According to the hierarchy established in previous studies on the use of clinical prediction models,48 this nomogram was evaluated and reported with up to level 3 evidence (on a scale of 1–4). It may be used clinically with criteria since the demographic data and professional profile of the study population were similar in relation to the nomograms validated internally and prospectively in a small sample. However, the level of evidence can reach up to 2 or 1 in further studies that will evaluate its validation using other samples or larger samples. Another limitation is that all the athletes were from Brazil. This aspect could have significant effects on the scores of athletes from other countries. However, several Brazilian football players are participating in several leagues around the world without significant differences in their injury rates compared with other athletes. Moreover, many athletes who participated in the original study have already played in European championships, which can avoid extrapolation bias of results.
