The prevalence of neck-shoulder pain, back pain and psychological symptoms in association with daytime sleepiness – a prospective follow-up study of school children aged 10 to 15

2.1 Participants and study procedures

Participants were recruited from 31 eligible Finnish-speaking elementary schools in the city of Turku (total population: 1,90,000). There were n=1,097 children participating at the age of 10, 928 at the age of 12 and 568 at the age of 15. The rate of attrition (from baseline to 2nd follow-up) for girls was 36% and for boys 39%. In the drop-out analysis we used data from the age of 10 to compare those who did not complete the questionnaire at age 15 years with those who completed the questionnaire. For both those completing and drop out at age 15 years, there were no statistically significant differences in the amount of sleep, frequency of daytime sleepiness, back pain, and neck-shoulder pain when aged 10 years.

The contact teachers from each school informed the children and their parents about the study. Participation in the study was voluntary. When children and their parents had given a written informed consent to participate, the contact teacher distributed the questionnaires to the participating children during the school day. The questionnaires were filled in using either paper or electronic form (depending on the availability of computers at the schools). Both versions were piloted.

2.2 The questionnaire instruments

To assess sleep, pain and psychosocial symptoms the Health Behaviour in School-Aged Children (HBSC) study questionnaire by WHO (www.hbsc.org) was used. The questionnaire has been used in several previous studies [29].

The following questions were used to assess self-reported pain and psychological symptoms: “How often during the preceding 6 months have you had back pain/neck-shoulder pain/depression/irritability or bad temper/nervousness/anxiety/dejection?”. Symptoms were rated on a five-point frequency scale (seldom or never, once a month, once a week, more than once a week, almost daily). In the analysis, symptom frequencies of more than once a week and almost daily were combined and renamed as “frequent”. If at least one of the psychological symptoms (depression/irritability or bad temper/nervousness/anxiety/dejection) occurred at least weekly then we classified child to be symptomatic.

The question used to assess daytime sleepiness was: “How often have you felt tired in the daytime during the last week?” The response alternatives were: not at all, once, twice, several times or daily. For analysis the answers were categorized into two classes: rarely and frequently. Rarely included the responses not at all, once and twice, and frequently included the answers several times and daily.

2.3 Data analysis

Counts and percentages were calculated to describe the data in the tables. The frequency of neck-shoulder pain and back pain was analyzed with four ordinal categories (almost daily and at least once a week as combined and remained as frequent class).

The study goal was to examine whether associations between neck-shoulder pain, daytime sleepiness, gender and psychological symptoms occurred in children aged between 10 and 15 and whether any association changes over time for the same children answering the same questionnaire several times. The same modelling approach was used for back pain and neck-shoulder pain.

In statistical models neck-shoulder pain (assessed by asking the same children at 10, 12 and 15 years of age) was handled as response, and the association with explanatory variables (daytime sleepiness, psychological symptoms and gender) was examined. While the same children were examined at ages 10, 12 and 15, we used ordinal logistic regression for repeated measures analyses techniques to take into account the correlations between measurements (the SAS^® GLIMMIX procedure). We also included the interaction between daytime sleepiness and age, gender and age, and psychological symptoms and age, and we wanted to examine whether the associations of these factors changed over time. If a statistically significant effect was found, further contrasts were programmed to find out where the significant differences were. For example, if age had a significant effect, the effect on the measurement taken between the ages 10–12, 10–15 and/or 12–15 was examined.

Also, whether neck-shoulder pain occurring in 15-year-old children had associations with the symptoms or gender differences of 10-year-old children (i.e. whether symptoms can predict neck-shoulder pain for 15-year-old children) was studied. First, univariable analyses were performed with a chi-square test for all children and separately for boys and girls. Associations between neck-shoulder pain at the age of 15 and neck-shoulder pain, psychological symptoms and daytime sleepiness at the age of 10 were studied. After the univariable approach we modelled neck-shoulder pain at 15 years of age, including all these explanatory variables in the same ordinal logistic regression model (the SAS^® GLIMMIX procedure). This model was performed for boys and girls separately and also with a model where gender was one factor in the model. In the combined model we also included all interaction with gender.

p-Values less than 0.05 (two-tailed) were considered as statistically significant. The program used for statistical analysis was SAS^® System, version 9.3 and 9.4 for Windows (SAS Institute Inc., Cary, NC, USA).

2.4 Ethical issues

The study follows the national legislation (Medical Research Act 488/1999) and the general guidelines of research ethics (TENK 2004). The ethical commission of the Hospital District of Southwest Finland has approved the study (8/2004/232). The participants received both oral and written information about the study; they had time to consider their participation and knew about their right to withdraw from the study at any time. Both the guardians and children gave a written informed consent to participate.

3.1 Neck-shoulder pain, back pain and daytime sleepiness prevalence

The prevalence of weekly or more frequent neck-shoulder pain in children of the ages 10 and 12 was 20% (see Table 1). For 15-year-olds the prevalence had increased to 32% (F₂=11.04, p<0.0001). For girls the prevalence of weekly or more frequent neck-shoulder pain at the age of 15 was 41% and for boys it was 20% (F₂=5.6, p=0.0040). Overall, a greater proportion of girls compared to boys suffered from at least weekly neck-shoulder pain in all age groups (F₁=16.11, p<0.0001). A more detailed description can be found in Table 1.

Frequent back pain at the ages of 10 and 12 was 11%, and at the age of 15 it was 20% (F₂=17.7, p<0.0001) (see Table 1). There were no differences between girls and boys in back pain (F₁=0.04, p=0.84). Furthermore, changes in back pain over time did not differ significantly between boys and girls (gender×time interaction: F₂=1.37, p=0.25).

The daytime sleepiness of children occurring at least several times a week was on average 13% at the ages of 10 and 12, and 24% at the age of 15. There was a clear increase in daytime sleepiness at the age of 15 compared to 10 and 12 years of age (p<0.0001). The increase between age 12 and 15 was greater in girls compared to boys (p<0.0001). The results of the frequency of daytime sleepiness are seen in Table 2.

3.2 Explanatory variables which associated generally to neck-shoulder pain and back pain over the study follow-up

We also wanted to study the association between daytime sleepiness and neck-shoulder pain, as well as back pain. Overall, daytime sleepiness in all ages associated positively with the frequency of neck-shoulder pain (F₄=13.62, p<0.0001) and back pain (F₄=14.62, p<0.0001). The more that daytime sleepiness existed, the more neck-shoulder pain and back pain occurred (see Table 3).

In addition, positive associations between psychological symptoms and neck-shoulder pain (F₁=186.8, p<0.0001), as well as back pain (F₁=149.1, p<0.0001), were observed; subjects with psychological problems suffered from neck-shoulder pain and back pain more frequently. These associations did not change over time (neck-shoulder pain: F₂=0.27, p=0.77; back pain: F₂=0.28, p=0.76).

Univariable analysis showed significant association between daytime sleepiness, neck-shoulder pain and psychological symptoms at 10 years of age. The multivariable analysis between the ages of 10 and 15 revealed that only the daytime sleepiness of children at the age of 10 predicted neck-shoulder pain at the age of 15 (p=0.0029; Fig. 1A). At the age of 10 daytime sleepiness and psychological symptoms were strongly associated with neck-shoulder pain, but only daytime sleepiness (p=0.0029) at the age of 10 could predict neck-shoulder pain at the age of 15. Furthermore, this model was seen for boys (p=0.026) (Fig. 1C) but it was only indicative for girls (p=0.053) (Fig. 1B).

Figure 1:

The association between daytime sleepiness, a neck-shoulder pain and psychological symptoms at the ages of 10 and 15. (p-value with *, from univariable model and p-value with **, from multivariable model). 1A: All children. 1B: Girls. 1C: Boys.

3.3 Predictors for neck-shoulder pain and back pain occurred at 15-year old children

We also wanted to build a predictive model. We stated with univariable analysis between back pain, daytime sleepiness and psychological symptoms at 10 years of age, and back pain at 15 years of age (Fig. 2A). Univariable analysis also showed significant association between daytime sleepiness, back pain and psychological symptoms at 10 years of age (Fig. 2A). Multivariable ordinal logistic regression analysis showed that the predictive association of back pain suffered when 10 years old was statistically significantly different between girls and boys (p=0.0060). Girls had a significant association of back pain when 10 years old and when 15 years old (p=0.0004) (Fig. 2B), whereas the association for boys was not statistically significant (p=0.076) (Fig. 2C). In a multivariate model neither daytime sleepiness nor psychological symptoms significantly predicted back pain at 15 years old.

Figure 2:

The association between daytime sleepiness, back pain and psychological symptoms at the ages of 10 and 15. (p-value with *, from univariable model and p-value with **, from multivariable model). 2A: All children. 2B: Girls. 2C: Boys.

4.1 Strengths and limitations

The main strength of this study is the research design: a prospective follow-up of one age cohort consisting of a large sample size. Also, the follow-up time for the same children of 5 years – from pre-adolescence to adolescent years – is unique and additionally provides some predictive data to the description. As the data was collected during school hours and as schools have been reported to be good research settings (as basically all children attend school), the results present a very wide range of social backgrounds and a heterogeneous group of youth. Methodologically, the use of the HBSC study questions also supports the validity of the results of this study. The HBSC questionnaire has previously been widely used and reported to be acceptable for test-retest validity and stability [41].

Some limitations need to be considered when interpreting the findings. The assessment of the pain symptoms was only performed with adolescents’ self-reported data and did not include any diagnostic assessments. The drop out was fairly large during the follow-up time. The drop out was mainly caused by uncontrollable factors, such as children moving location or being absent from school at the study time. However, the drop-out analysis shows that there was no significant difference between the baseline results and the participants dropping out and those not dropping out.