Is being a regular player with fewer teammates associated with musculoskeletal pain in youth team sports? A cross-sectional study

The present cross-sectional study was carried out concomitantly with a previous study [14]. We hypothesized that regular players with fewer teammates relative to the required number of players in team sports have a higher probability of having pain compared with non-regular players with more teammates. This study protocol was previously approved by the research ethics committee of the Physical Education and Medicine Research Center UNNAN (H19-7-23-5). The study procedure, analysis, and description were reported according to the Strengthening the Reporting of Observational Studies in Epidemiology statement [17].

All junior high schools (seven schools; seventh to ninth grades; students’ age: 12–15 years) and high schools (three schools; 10th–12th grades; students’ age: 15–18 years) in Unnan City (population: 43,520; area: 553.4 km²), Shimane, Japan, participated in the previous study [14]. In October 2008 and 2009, self-administered questionnaires were distributed to all students (2,271 students in 2008 and 2,211 students in 2009) and returned through schools. Students who received special needs education and those with invalid (blank) responses were excluded. The total response rate was 82.1% (3,680 responses from two surveys in 2008 and 2009). Third (final) grade students in junior high schools and high schools in 2008 were excluded from the study. These students were excluded because they had stopped their organized sports activity (if they participated) since July to the end of the academic year (March) to focus on academic work, in accordance with Japanese school custom. Additionally, students pursuing individual sports (n = 331), those pursuing no sports (n = 630), and those with invalid responses (missing data) on the type of sport or playing status (n = 223) were excluded. We used only the first-year data of each individual (i.e., first and second grades in 2008 and first grades in 2009) to avoid overlap. As a result, 632 eligible responses were analyzed for the present study (Fig. 1). An explanation of the students’ right to decline to answer any questions and general information about the study were provided in a letter to parents or guardians of the students who answered the questionnaire. Students signed their name on the questionnaire to give their assent to participate in the study and returned the completed questionnaire in an anonymous envelope. Individual answers were not shared with teachers or sports coaches.

Self-administered questionnaires included questions on sex, age, weight and height (used for calculating body mass index [BMI] in kg/m²), organized sports activity (h/wk from hours on weekdays and the weekend), sleep time (h/d from time in and out of bed), and screen time (h/wk of TV viewing and game playing time on weekdays and the weekend). For students who participated in organized sports activity, questions regarding the nature of the sport in which they were registered (e.g., baseball, basketball) and whether they were regular players (Do you currently participate in competitions as a regular player or represent your team? [yes/no]) were also asked. In this study, organized sports activity was defined as sports activities that occurred after school programs on weekdays and/or weekends or activities that were organized by sports clubs.

To estimate the opportunity to actively participate in organized sports competitions and games and the possible physical burden among regular players with fewer teammates, a teammate quantity index (TQI) was calculated as follows: TQI = [number of teammates in their grade]/[required number of players for the sport]. The registration information regarding the number of teammates in sports clubs was reported by the schools that managed the clubs. The required numbers of players for each sport were assigned as follows: baseball and softball = 9; basketball = 5; soccer = 11; and volleyball = 6. For example, in the situation of a baseball team with 14 students in the second grade, the TQI for these second grade students is 14 divided by 9 (i.e., TQI = 1.6). A higher TQI indicates more teammates in the same grade in their team with consideration for each sport. TQI was dichotomized as high or low, using the median value (1.3) as the cutoff.

MSP was assessed using a questionnaire [see Additional file 1]. Details of the questionnaire are described elsewhere [14]. Briefly, pain was defined by the frequency and the part of the body area that was painful. Pain should be present at least once a week in at least one part of the body. The test-retest reliability over 7 days of the pain questionnaire was tested and found to be acceptable (Cohen’s kappa for pain in any location = 0.67). The agreement test (criterion validity test) with a face-to-face interview by health professionals (e.g., public health nurse, health fitness programmer) who were trained by an orthopedist also showed moderate agreement (Cohen’s kappa = 0.52).

We compared the prevalence of pain in students with different playing status (regular or non-regular player) and team status (lower TQI or higher TQI), using a chi-squared test.

Our hypothesis was that regular players with fewer teammates relative to the required number of players in the relevant sport have a higher probability of having pain, compared with non-regular players with more teammates. First, we evaluated whether playing status (regular or non-regular player) or team status (number of teammates) was associated with the prevalence of pain separately, using multivariable-adjusted Poisson regression [18]. Second, we tested the interaction between playing status (regular or non-regular players) and the TQI for prevalence of pain. Additionally, we defined the joint categories of playing status and team status as 1) non-regular players with a high TQI, 2) non-regular players with a low TQI, 3) regular players with a high TQI, and 4) regular players with a low TQI. We examined the association of these categories with the prevalence of pain, using Poisson regression. Location-specific (upper limb, lower back, and lower limb) pain analyses were also performed. Poisson regression was used to calculate prevalence ratios (PRs), which are more easily interpreted than odds ratios [19]. Adjustments were made by sex, age, BMI, sleep length, screen time, and school. BMI, sleep length, and screen time were included in the model as categorical variables divided by tertiles within each grade with the lowest groups as references. School was included in the model as a fixed effect rather than a random effect. The fixed effect was used because all school (cluster) level confounders were able to be adjusted, regardless of whether they were measured [20]. School was considered to be more appropriate than school class considering the scarce distribution of outcomes in each class. Additionally, students within the same schools shared some potential confounders of pain beyond classes and grades (e.g., sports clubs, physical education teachers, sports facilities, and environment).

Missing information was processed using multiple imputation (n datasets = 10) under the missing at random assumption [21, 22]. Each imputation was based on regression models including variables used in the main regression analyses. The 10 imputed datasets were analyzed independently and combined for inference, accounting for variability of imputation [21]. For sensitivity analyses, the analyses were repeated using only cases with no missing values (n = 541). Analyses (two-sided α < 0.05) were carried out using SAS version 9.3 (SAS Institute Inc., Cary, NC).