Test-retest repeatability of questionnaire for pain symptoms for school children aged 10–15 years

1 Introduction

About 15%–25% of all children and adolescents report recurrent or chronic pain conditions [1]. The limbs (34%), head (26%), abdomen (16%) and back (15%) have reported to be the most common pain areas among 12–18-year old adolescents [2]. Already 6–8-year old children has been reported to have frequent pain most commonly in the lower limbs (40%) and head (31%) whereas back and neck-shoulder area pain prevalences were 7% [3]. Studies have shown that children with long-term pain report similar or more disabilities in daily life than children with chronic somatic disorders [2]. Disabilities manifest in social functioning, school participation, sports activities, and/or sleeping problems [4]. Pain in children and adolescents is also a burden on their families and to society [2]. Pain in adulthood is known to have severe economic consequences [5], but even in adolescence, long-term pain may have extensive financial consequences [6].

Reliable and valid pain assessment tools are needed for research and clinical purposes to accurately measure the frequency of pain in children. However, information about the repeatability of these questionnaires is limited. To study the prevalence and frequency of pain symptoms, self-report questionnaires are common instruments. Hakala et al. studied the changes in musculoskeletal pain in 189,894 school children in the Adolescent Health and Lifestyle survey in 1985–2001. They captured the prevalence of pain during the past half a year and the questions were categorized on four levels (seldom, once a month, one a week and daily). The test-retest reliability coefficient for back and neck pain were 0.48–0.67 and lower back pain ~0.60 [7], when the questionnaire was retaken after four to 6 weeks. Jannini et al. studied the frequency of musculoskeletal pain during the previous 3 months in the neck, upper limbs, chest, trapezoid muscles, lower back and lower limbs in children using a questionnaire earlier developed for adolescents [8]. The test-retest reliability of the questionnaire with a 10 day’s interval was substantial with a Kappa [κ] value of 0.72. The whole questionnaire contained 70 questions, and conclusions about the repeatability of pain questions only are not available.

El-Metwally et al. captured the information on musculoskeletal pain frequency (neck, upper limbs, chest, upper back, lower back, buttocks) in Finnish schoolchildren with a structured five-point Likert-scale questionnaire [9]. The test-retest reliability of the questionnaire in detecting children with pain at least once per week was good with a Kappa [κ] value of 0.90. Although this questionnaire has been widely used in research [9], [10], [11], its repeatability has not been studied in Finnish children. The purpose of this study was to assess the test-retest repeatability of the pain-related questions using the Finnish electronic version of this questionnaire, which categorizes pain frequency into five-levels among school-aged children.

2 Methods

3 Results

Of the 206 children from grades 4–9 who were invited into the study, 83% (n=181) filled in questionnaires twice and were included in the analysis. Seventy-seven children (41 boys and 36 girls) were from primary school (grades 4–6) and 104 children (40 boys and 64 girls) from lower secondary school (grades 7–9). The highest prevalence of pain symptoms observed were in the head (29%) and NS area (23%) (Table 1). In both the head and NS, the prevalence of pain was more than twice as high in lower secondary school students (37% and 30%) than in primary school students (16% and 13%).

The repeatability of the pain questionnaire is presented in Table 2. In the analysis with a five-point scale, questions about pain in NS, the head and lower extremities had good repeatability in all schoolchildren (ICC 0.67, 0.66 and 0.62, respectively). In primary school children, the ICC for the lower extremities (ICC 0.70) was good, and for NS (0.59), lower back (0.58) and the head (0.56) the ICC was fair. Other questions had poor repeatability (ICC<0.40). In the analysis of lower secondary school children, the ICC for pain in NS and the head were good (0.70, 0.69), and the stomach (0.59) and lower extremities (0.57) were fair, while the other areas remained poor.

In secondary analysis using a two-point scale and Kappa [κ] coefficient, the questions about pain in the head (0.70) and NS (0.68) had good repeatability in all schoolchildren. The agreement percentage varied between 88–94%. In primary school children, four pain area questions reached good repeatability with the following Kappa values: lower extremities (0.79), NS (0.67), lower back (0.67) and head (0.66). The other pain areas had moderate repeatability, except the upper extremities had fair repeatability (0.30). In lower secondary school children, questions about the head (0.69) and NS (0.67) were good, while the lower extremities (0.52), stomach (0.47) and upper extremities (0.45) were moderately repeatable. The agreement percentage in primary and secondary school varied between 92–98% and 83–92%, respectively.

4 Discussion

This study evaluated the test-retest repeatability of pain-related questions in a questionnaire for Finnish school children aged 10–15 years. Almost one third of the children reported head pain and a quarter reported pain in the NS area during the last 3 months. The most repeatable pain area questions were those related to the head, NS and lower extremities in all schoolchildren using an instrument with either a five-point or two-point scale.

In general, there was a tendency for higher frequencies of pain in lower secondary school children than in primary schoolchildren. Particularly, the frequencies of head and NS pain were almost three times higher in lower secondary school children than in primary schoolchildren. A previous cohort study in Finnish children showed similar results, where the prevalence of at least weekly experienced neck-shoulder pain and lower back pain was much more common in older age groups (14-, 16- and 18-year-old children) than in 12-year-old children [7]. Kappa is also dependent on the prevalence of the condition [15]. This can partly explain higher Kappa values in most frequent pain localizations. Thus, we should be careful when comparing Kappa values from different studies when the prevalence varies.

When studying frequencies in children and adolescents and the questionnaire as an instrument, the recall time can be an important factor. In a Finnish study, the test-retest reliability was measured with a four-to-6-week interval with a 6-month recall time over three different years [7]. The results for back and neck pain repeatability ([κ] 0.48–0.67) were similar compared to our study results. In a Brazilian study, to reduce the effects of memory bias, respondents, 10–19-year-old children were asked to report on musculoskeletal symptoms suffered during the previous 3 months [8]. The test-retest reliability of the questionnaire with a 10-day interval was substantial with a Kappa [κ] value of 0.72. In an American study of seventh- and eighth-graders, for evaluation of the internal consistency of a pilot survey with questions about back and neck pain, the test-retest with a 1-week interval, for evaluation of the internal consistency of a pilot survey questions for back and neck pain showed [κ] 0.54 as a mean [16]. In this study, the recall time was “present school year.” However, the age of the population and differences in the test-retest settings of these studies affect comparability to our study.

In addition to recall time, respondent characteristics and question characteristics may affect the reliability of responses in surveys [17]. When responding to the present pain questionnaire, children were asked to rate the frequency of their perceived pain on a five-point Likert-type scale. In different studies, the number of answer choices tends to vary, ranging from two response options up to seven. In this study, the main result remained similar on both ICC and Kappa analyses. In our secondary analysis, slightly better values for younger children in two-point scale Kappa analysis suggests that it would be easier for younger children to answer with fewer options. Reading comprehension becomes easier with age. Borgers et al. noticed that older children are better at reading and comprehension and thus produce better data quality [18]. Bae et al. reported, that the older children are, the more repetitive the answers are [19]. Our secondary analysis finding should be confirmed in a further study by collecting data with both a five-level and two-level classification and comparing these directly.

5 Limitations

The weakness of this study is its relatively small sample. The strength of the study was a high response rate in both age groups, partly because the data was collected during school days in classrooms. The electronic questionnaire ensured that there were no missing values. If the children failed to answer a question, the system did not allow them to proceed. Also, the time interval between two data collections was accurate. In school, children are also able to give more subjective answers, which are free from their parents’ opinions.

6 Conclusions

This electronic questionnaire, filled in during school hours with an interval of 2 weeks, had an acceptable repeatability to screen the frequency of pain both in primary and lower secondary school children in most prevalent pain areas in both age groups: pain in the head and NS. The five-point scale answering options were equally repeatable in younger and older children. These results are encouraging regarding the use of this measurement in further studies when measuring lower secondary school children.

7 Implications

It is important to be aware of the impact of health-related outcomes on children’s ability to be successful in their lives. With the help of a simple electronic questionnaire, it is possible to cost-effectively capture, for example, the prevalence and frequency of pain during the school hours. The identification of children’s pain symptoms accurately provides more possibilities to prevent and to minimize the chronic pain among schoolchildren.