Determination of pain intensity risk factors among school children with nonspecific low back pain

Background

LBP affects up to 80% of the population at some time during their lives [1]. Although LBP has generally been believed to be uncommon before the age of 20, the prevalence of LBP among school children and adolescents has been reported to be high in different parts of the world, mostly in Western countries, where it varies from 10–40%. The NHANES II (National Health and Nutrition Examination Survey Series) reported the onset of LBP before the age of 20 in 11% of the general population [2]. LBP prevalence rates were found to be 28–31% in 2 studies carried out on school children in Kuwait and Tunisia. These studies show that the prevalence of LBP in children is high, equaling that of adults by the end of the growth period [3,4].

LBP has a significant economic impact on individuals, society, and quality of life. Many studies have analyzed the risk factors associated with LBP in children and adolescents to describe risk factors profiles [3,5–8]. To our knowledge there have been no reports on NLBP and related factors among school children in Turkey. The aim of this study was to investigate possible factors associated with pain intensity among Turkish school children with NLBP, aged 10–18 years, and to determine the relationship between related factors and pain intensity using the regression tree method.

Material and Methods

THE QUESTIONNAIRE:

All school children completed the questionnaire during school time under the supervision of interviewers. Two methods of enquiry about NLBP were used, namely “a direct question” and “a pre-shaded manikin question”:

Students who answered both of the questions as “yes” were classified as having LBP, and pain intensity was recorded. The questionnaire contained 22 questions.

STATISTICAL ANALYSIS:

The regression tree method (RTM) was used to determine risk factors that may affect pain intensity. RTM is a tree-based model, and is more useful than traditional statistical methods when a data set is large and when there are many variables. Moreover, the RTM takes into consideration interactions among variables, and is not affected by high correlations between risk factors. There is no assumption about distribution shapes of risk factors, but outcome variable should be numeric [10,11].

In RTM the association between risk factors (x) and outcome variable (y) (pain intensity, in this study) are examined by a schematic representation. The main idea of RTM is to obtain homogeneous subgroups. Homogeneous groups are constituted according to the adequate cut-off values of the risk factors. At the beginning, all individuals are collected in 1 group called a “root node”. Homogeneous groups that come into being based on recursive binary splitting are termed “terminal node” [12]. The homogeneous group means that this group is sufficiently homogeneous and cannot be split any more.

Splitting continues until the tree reaches maximum size, and then passing the selection of adequate tree structure stage, called “pruning”. The maximum tree is not used for every data set because of its overfit structure. After pruning, the tree is termed an optimal tree. In the optimal tree, the values that take place in the terminal nodes give the mean and variance of that group [13].

Results

The descriptive statistics are shown (Table 1) as mean ±SD and as frequencies and percentages. The overall mean and standard deviation of pain intensity was 2.58±0.86 (minimum=1, maximum=5).

Among the risk factors used in this study, duration of studying, type of bed, transportation to/from school, and BMI score were found to have a significant effect on pain intensity, while sex, studying posture, regular exercise habit, and bag handling were not significant (Figure 1).

As will be seen from Figure 1, 6 homogeneous groups are defined by the RTM according to pain intensity, with an increasing order. These are as follows:

Discussion

It has become clear that a high prevalence of LBP occurs not only in adults, but also in children/adolescents [3]. More recently, cross-sectional and longitudinal studies have focused on NLBP in children [2,3].

The prevalence has been reported to vary from 10% to 40% in the literature, but the authors found it to be 46.7% in their previous cross-sectional study of 624 school children/adolescents 10–18 years old [14].

Among the 8 risk factors included in the study, 4 were found to be important regarding pain intensity in children with NLBP. These factors are as follows:

Korovessis et al. in 2004 [17] reported that dorsal pain increased with increasing backpack weight among children. We found no significant relation between bag handling and NLBP in our study.

Lee and Chiou found that “poor sitting habit” were statistically associated with LBP [18]. In our study, studying posture was not found to be an important factor.

We also found that the sex of the children was not an important factor in NLBP intensity.

Conclusions

Results from the literature, as well as our study, show that taking parents’ and teachers’ concerns seriously is of vital importance. Therefore, health care providers should evaluate school children carefully and make accurate observations in terms of risk factors, including duration of studying, type of bed, transportation to/from school, and obesity, to predict any severe musculoskeletal problems, especially NLBP. Finally, physical factors and musculoskeletal risk factors are especially important in terms of NLBP in school children. Further studies are needed to investigate psychosocial risk factors and their relationships with NLBP in school children.