Fitness, motor competence and body composition as correlates of adolescent neck/shoulder pain: an exploratory cross-sectional study

Data were collected from 1608 adolescents (783 females, 825 males) of mean (SD) age 14.06 (0.20) yrs, who were participating in the Western Australian Pregnancy Cohort "Raine" Study http://​www.​rainestudy.​org.​au. This project began with a cohort of women attending antenatal clinics at King Edward Memorial Hospital for Women, Perth, Australia between 1989 and 1991. The children have been followed at birth, 1, 2, 3, 5, 8, 10, and now 14 years of age. Inclusion criteria for the women were gestational age of 16–20 weeks, adequate English language to understand the implications of participation, and an intention to remain in Western Australia throughout follow-up. 2337 adolescents were eligible for the 14 year follow-up, and 1704 (72.9%) of these agreed to participate in some aspect of the follow-up. 1608 (68.8%) completed the data collection requirements for the analysis reported in this paper. There were no exclusion criteria for this part of the cohort. A comparison of the cohort with the Western Australian general population showed a higher proportion of high risk births, as would be expected for a major specialist hospital [19].

With the assistance of a research assistant, participants completed a laptop questionnaire at an assessment centre. The questionnaire contained 130 questions concerning a broad range of issues, many of which were not relevant to this study. Adolescents were asked about their experience of NSP, described as pain in the area of the posterior neck and upper trapezius, as diagrammatically defined by Kourinka et al. [20]. The relevant NSP questions were: Have you ever had neck/shoulder pain? ("yes" or "no"), Has your neck/shoulder been painful in the last month? ("yes" or "no"), and Did your neck/shoulder pain last for more than 3 months? ("yes" or "no"). The full questionnaire took about 1 hour to complete, and the NSP questions occurred in the first half. The life prevalence question is very similar to that used by Chiu and Leung [21], which was shown to be reliable and valid.

Information on diagnosed neck pain was obtained from a paper questionnaire given to the primary carer, which included the question, "Does your child have now, or has your child had in the past, any of the following health professional diagnosed medical conditions or health problems?". The primary carer had to indicate which medical diagnoses their child had experienced from a short list of general medical problems, which included "neck pain". This question was part of a questionnaire given to the primary carer, covering many other factors that are not relevant to this study.

A physical assessment of the child was carried out after the laptop questionnaire, and parts relevant to this study are described below. Height (m), body mass (kg), waist girth (cm) and arm girth (cm) were measured without shoes. Several physical performance tests were then carried out. Maximal aerobic capacity was estimated from heart rate recordings during sub-maximal cycle ergometry using the Physical Work Capacity 170 protocol [22]. The sustained back extension test [23] and the number of abdominal curls performed in 3 minutes [24, 25] were used to measure trunk muscle performance. Limb muscle performance was evaluated by standing long jump [24, 26], seated basketball throw [24, 27] and grip strength [26, 27]. Flexibility was tested using the shoulder stretch [28]. Motor competence was evaluated using the McCarron Assessment of Neuromuscular Development (MAND) [26]. The Neurodevelopmental Index (NDI) was derived from summing gender and age corrected scores from 10 tests of fine and gross motor skills, and then converting to a scale with 100 as the mean and a SD of 15. Four sub indices (muscle power, kinaesthetic integration, bimanual dexterity and persistent control) [26], each comprising two of the items, were calculated. Details of the 10 items used to generate the overall NDI and 4 sub indices are shown in table 1.

All of these physical performance tests have been previously validated in very similar forms [25‐27, 29‐31] except the shoulder stretch, which has acceptable face validity. Reliability of the same or similar versions of the tests is also good [26, 30‐32], although there are no reports on the shoulder stretch or the basketball throw.

Gender differences were analysed using independent t tests for each of the continuous variables, and Chi squared tests for the categorical variables. To facilitate the interpretation of non-linear relationships, continuous variables were banded into the bottom 25%, inter-quartile range and top 25%, and the proportion of subjects with neck pain in each segment were compared. Univariate logistic regression models predicting lifetime, last month, chronic (lasting more than 3 months) and diagnosed neck pain from each physical characteristic were calculated separately for males and females, with statistical significance set at p < 0.05, and the interquartile range of each continuous variable defined as the reference category. For the only binary variable (shoulder stretch), being able to perform the stretch was the reference category. Corrections for multiple univariate tests were not carried out as the multivariate results were the end point of the study.

Backwards stepwise likelihood ratio multivariate logistic regression models were used to evaluate the combined associations of performance factors for males and for females, with the probability for entry and removal of the likelihood ratio score statistic being p = 0.05 and 0.10 respectively. For each gender, 4 multivariate analyses for each of neck pain ever, last month, chronic and diagnosed were performed. Height and weight were included in an initial step, with abdominal curls, basketball throw, jump, back muscle endurance, PWC170, hand strength and shoulder stretch included in a second step, along with one of the body composition variables and either the NDI score or the 4 motor competence factor scores. The choice of body composition variables or motor competence variables was determined by the strength of univariate relationships with pain, and was determined for each of the eight multivariate analyses separately. The strength of the predictive ability of the model was estimated by Nagelkerke R². All statistical analysis was performed using SPSS version 13.

Adolescents provided written informed assent and their parent/guardian provided written informed consent prior to participation. The study was approved by the Human Research Ethics Committees of Curtin University of Technology and Princess Margaret Hospital.

NSP ever was experienced by 46.8% of the participants, NSP in the past month by 28.7%, 'chronic' (lasting more than 3 months) NSP by 8.2% and diagnosed NSP by 7.1%. Females had a significantly higher prevalence of NSP ever, month and chronic, but not diagnosed NSP (see table 2). 64.9% of those with chronic NSP also had experienced NSP within the last month and 20.9% of those with chronic NSP also had diagnosed NSP according to parental report.

Descriptive statistics for physical characteristics are given in table 2. Females obtained significantly higher mean scores for BMI, back endurance and the motor competence factors of Persistent Control and Bimanual Dexterity. A greater proportion of females could perform the shoulder stretch test. Males obtained significantly higher mean scores for waist girth, aerobic capacity, abdominal curl number, standing long jump, basketball throw, grip strength and the motor competence factor of muscle power. Males were also taller and heavier, with a lower BMI. There were no gender differences in arm circumference, NDI or Kinaesthetic integration.

This study analysed cross-sectional data only, so relationships identified could be the result of causality in both directions: NSP could be influenced by physical characteristics or vice versa. There is evidence that adults with back pain may experience a 'deconditioning' effect associated with pain inhibiting and restricting participation in work, leisure and household activities [33]. In contrast, there is evidence that poor back muscle endurance increases the risk of back pain episodes in manual workers [34].

Longitudinal data (currently being collected on this cohort) is required to elucidate the direction of any relationship. The remainder of this section discusses the cross-sectional results and suggests potential mechanisms for observed relationships.

Although a weak univariate relationship between low arm circumference and a lower risk of chronic NSP was observed in males, body composition was not associated with any form of NSP in either gender after multivariate analysis. This concurs with previous adolescent findings [35] and underlines the importance of multivariate analysis with a comprehensive range of covariates.

Higher aerobic capacity, after correction for other variables including body weight, was associated with a lower risk of NSP in the last month for females only, with a similar trend in NSP ever. The lower risk of NSP with improved aerobic capacity for NSP in females may be associated with increased levels of physical activity which is known to sometimes have a beneficial effect on spinal pain disorders [7]. This may also relate to a deconditioning mechanism, where females with NSP reduce their participation in physical activity and lose aerobic capacity. The lack of any relationships for males may indicate a differing mechanism or response to neck pain based on gender.

There were inconsistent associations between arm muscle performance and NSP after multivariate analysis. Greater upper body power, as measured by the basketball throw, was protective in females for both NSP ever and in the past month, but a risk factor for chronic NSP in males. The reason for this gender difference is unclear although other factors such as specific sport participation may influence these findings. Females (but not males) engaging in more dynamic arm activities have less pain [8, 36], and given that greater amounts of dynamic arm activities may increase upper body power, this may explain the pattern in females. The opposite pattern in males, with increased risk of chronic NSP in the most powerful quartile, may partly relate to greater arm activity not having a protective effect in males [8, 33], and also because their high arm power may be a proxy for greater overall physical activity levels (not just upper limb activity), which relates to greater NSP in males [37]. In contrast, Barnekow-Bergvist et al. [14] reported that greater arm endurance in adolescent males was related to a reduced risk of NSP in adulthood, which may relate to a deconditioning effect secondary to NSP.

Multivariate associations between NSP and leg power were very different to those with arm power. In females, a low jump performance decreased risk of NSP in the past month, effectively the opposite effect seen with upper body power. Aurvinen et al. [38] reported that higher overall activity levels may increase NSP risk in females. Since it is possible that higher overall activity may be associated with greater leg power, this may explain our finding of low leg power reducing risk. Although differing effects on NSP from arm activity levels and overall activity levels may initially appear paradoxical, it is possible that the relationship between overall activity levels and NSP is not direct but mediated by performance in sports that may increase risk of NSP. Similarly, diagnosed neck pain was associated with greater jump distance in males, although this was not observed for the other pain variables. This result may indicate a similar mechanism to that described in females.

A very similar pattern was observed between abdominal endurance and NSP ever after multivariate analysis, with better performance associated with greater risk of pain in females only. Mechanisms may be similar to those described for leg power. In contrast, Mikkelson et al. [16] reported that poorer female adolescent abdominal endurance was a risk factor. However, Mikkelson et al. [16] reported these outcomes in adulthood.

Less back muscle endurance was associated with a decreased risk of NSP ever in males after multivariate analysis, which was analogous to the findings for leg power and abdominal endurance in females, and may again relate to the males being involved in more vigorous physical activity [37]. Similarly, females with a diagnosis of NSP were more likely to have high back endurance, and this could relate to greater overall activity levels, as previously described. However, females with low back endurance also had a higher risk of diagnosed NSP. It is possible that these females were below a threshold of endurance at which any effects on spinal stability became important, or alternatively were experiencing a deconditioning effect as a result of the pain. However, this effect was not seen in males, who had lower back endurance overall.

One surprising multivariate finding was that poorer shoulder girdle flexibility, as measured by the shoulder stretch, was related to a significantly decreased risk of NSP in the past month in females. There was also a strong trend for the same effect on chronic NSP in males. Though counter-intuitive, there are reports of a relationship between lower shoulder rotational flexibility and greater upper limb activity levels in elite adult water polo [39] and volleyball players [40]. Greater amounts of dynamic upper limb activity have also been shown to reduce the risk of female adolescent NSP [8, 36] and so these separate findings may explain the overall association of reduced flexibility and lower risk of NSP observed in this study.

Males with higher levels of the motor competence factor of muscle power had a reduced risk of diagnosed NSP, and there was a trend for higher overall motor competence (NDI) to be associated with lower risk of NSP ever in males after multivariate analysis. This was expected, given that higher motor competence might have a protective effective on the musculoskeletal system [41]. However this relationship may be weakened by males with better motor competence being more involved in vigorous activities, as suggested by evidence that pre-pubescent children with higher motor competence engage in more vigorous play [42], and thus more likely to develop NSP [37]. This potential confounding may possibly explain the contradictory finding of lower persistent control being associated with a lower risk of diagnosed NSP. In contrast, poorer coordination may be a result of reduced motor practice as part of a reduction of activities associated with NSP.

In females, higher bimanual dexterity significantly increased risk of NSP in the past month. Bimanual dexterity relates to the co-ordination of fine motor skills across both arms, and might be developed by activities such as playing musical instruments, needlework, computer work or craftwork, which are known risk factors for female adolescent NSP [36].

Evidence from longitudinal studies [14, 16] demonstrates that physical performance in adolescence can influence the development of NSP in adulthood, although a deconditioning response to the presence of NSP is also possible. The predictive utility of the models in the current study was very low however, with Nagelkerke R² ranging from only 0.001 to 0.085. The lack of stronger relationships was not due to missed curvilinear relationships as these were accounted for in the analysis, and the study was not underpowered as weak relationships were detected.

This may indicate that either physical performance is not strongly related to NSP during adolescence or that the direction and mechanisms are more complex and other factors need to be considered. One of the strengths of our study was the broad range of physical variables adjusted for in the analyses, but certain possible confounders such as activity levels and sport participation were not included in the current analyses. Consideration of these may have either reduced or strengthened observed relationships, and should be attempted in further work. In addition, NSP was treated as a homogenous entity, but in reality it may have several sub-groups with different aetiologies. Real, but differing, associations between physical performance and each sub-group may thus have been lost in the current analysis. Further work towards subgroup identification is intended.

NSP is not a simple construct and thus four measures were used, including a parental report of health professional diagnosed NSP. Whilst parental report of diagnosed neck pain has limited detail and accuracy, it reinforced the self-report measure of NSP. Further, strong relationships could be expected to be more consistent across the different measures.

Associations were very different across genders, with no common effects seen. These differing gender effects may be the result of differences in the type and vigour of sporting activities [43], as well as anthropometric differences, and possible variation in underlying pain mechanisms and psychosocial effects. Whatever the cause, these differences emphasise the need to continue to consider gender in future work, as gender will be a possible confounder of many pain/physical characteristics relationships.