Introduction
Back pain is among the most common musculoskeletal complaints seen in primary care [
1]. The prevalence of back pain has been rising continuously for many years and the financial burden on society is increasing [
1‐
3]. Although our population is aging globally, the elderly are often excluded from research on back pain and the influence of psychological factors [
4]. Psychological factors, in particular beliefs about back pain, have been shown to play a major role for the course of back pain [
5‐
8]. Negative and irrational beliefs are associated with persistent back pain [
5].
Beliefs and attitudes towards back pain can be measured with questionnaires. The Back Beliefs Questionnaire (BBQ) was developed by Symonds et al., with the aim to make a new instrument to measure beliefs and attitudes related to back pain [
5]. The authors developed a 14-item self-report questionnaire to investigate beliefs about the inevitable consequences of back pain [
5]. BBQ has been used to predict recovery rate from back pain [
9,
10], in population studies assessing public attitudes and as an outcome to assess effectiveness of educational campaigns [
11‐
13]. To our knowledge, the BBQ has been translated into Arabic [
14,
15], German [
10], Chinese [
16,
17] and French [
18]. Most of these methodological studies have shown good test-retest reliability and validity. However, if BBQ is to be used as a measurement outcome in addition to a predictor, test-retest reliability in terms of minimal detectable change (MDC), standard error of measurement (SEM) and limits of agreement (LOA) needs to be established. These estimates are useful as they provide an interpretation of measurement error according to the absolute score of an instrument [
19], however, few studies have investigated these properties [
14,
15].
The BBQ has recently been translated into Norwegian, but assessment of test-retest reliability and validity has not yet been performed. The psychometric properties of any scale may be affected by translation into another language, hence, it is important for the scale to be evaluated psychometrically. Therefore, the aim of this study was to evaluate the psychometric properties of the Norwegian version of the BBQ in terms of test-retest reliability, construct validity and internal consistency when used on elderly patients with a new episode of back pain.
Discussion
The Norwegian version of the BBQ shows acceptable psychometric properties in elderly patients with a new episode of back pain. Our results indicate that BBQ can be used in both clinical settings and research with the purpose of assessing beliefs about back pain. This is in line with former assessments of BBQ in other languages [
14,
15,
31]. Our study is the first report to evaluate psychometric properties in the Norwegian version of the BBQ. Additionally, the fact that our research was conducted in elderly patients with back pain contributes important knowledge to a field in which most research has been conducted in younger populations [
4,
32].
Fears and beliefs leading to avoidance have been shown to negatively influence the prognosis of back pain and increases the risk of developing chronic disability [
9,
33]. With BBQ as an examination measurement, it is possible to detect negative beliefs in patients with back pain. Early detection will allow primary healthcare workers to provide back pain patients with clarifying information pertaining to their irrational beliefs. This positive influence may have an important socioeconomic impact worldwide.
Our study sample completed the BBQ with a mean score of approximately 30 on a scale ranging from 9 to 45; this is a relatively high score, reflecting optimistic beliefs. This score is higher than some other studies which report a mean score ranging from 21 to 26 [
14,
15,
18,
34], but similar to a study from Australia which reports a mean BBQ score of 30.7 [
33]. The low level of negative beliefs in our sample of elderly people might have been influenced by different coping strategies, reduced pain perception and it might be argued that some elderly patients believe pain to be a normal part of the ageing process and have more realistic beliefs [
35,
36]. Only 35% of our participants were working, and one could speculate that retired participants may experience fewer consequences due to an episode of back pain as they are unconcerned by the responsibilities of employment and with taking sick leave. Since previous studies have been conducted on different populations, such as healthcare workers [
17], healthcare students [
16], healthy individuals [
11], and younger patients [
18] as well as in different cultures [
14,
15,
34], it is difficult to make any direct comparisons with this study. Furthermore, back beliefs can change rapidly, which can influence evaluations using the BBQ. While we found a low correlation between the BBQ and pain, in their research, Bostick and associates found that BBQ participants achieved lower scores when experiencing acute and severe pain, while a 1-week history of mild back pain resulted in higher scores [
31]. The ICC was considered to be acceptable (0.71) according to our chosen classification, which suggests that the BBQ is a reliable outcome measure in our population [
19]. In earlier studies, the BBQ has demonstrated ICC with results ranging from 0.80–0.89 [
10,
14‐
17]. Few authors specify the chosen effects model and measure regarding the ICC, which can influence the outcome results. Measuring back beliefs with retest might be challenging. There is a potential risk of recall bias if the participant has a short interval between the test and the retest, and a risk of possible change in back pain status when the time between tests is long. A short interval was chosen in this study, since the high number of questionnaires completed at T0 would most likely reduce recall bias of BBQ at T1 2 days later
. The ICC is influenced by the variation between the patients – heterogeneity resulting in a high ICC value – and substantiates the importance of assessing measurement errors [
37]. MDC, which expresses an error estimate given in the scale’s unit, resulted in 10.5 of a possible 45 points. MDC determines the smallest within-person change to ensure that the change is larger than the measurement error, and 10.5 points will provide an estimate of this limit when using the BBQ as an outcome measure. The results (Fig.
1) from LOA show that the large measurement error was equally spread across the whole scale range. This implies that when using the BBQ as an outcome to evaluate change during a treatment or clinical course, an observed change below 10.5 points can not be distinguished from measurement error, regardless of baseline value. This estimate of measurement error is large taking into consideration the scale from 0 to 45. To the authors’ knowledge, few studies have investigated the measurement error of the BBQ. Alamrani and coworkers obtained a somewhat lower MDC (5.9) and SEM (2.1) values influenced by their high ICC value (0.88) [
15]. Due to the high measurement error in this study, more reports should investigate of measurement errors for the BBQ, which may increase our confidence in utilizing the questionnaire as an outcome measure.
A high internal consistency was found for BBQ, even though the Cronbach’s alpha coefficient is sensitive to the number of items in the scale, and questionnaires with fewer than 10 items can result in a value that is too low [
38,
39]. The results are consistent with other studies conducted on BBQ, although our values of 0.82 (test) and 0.80 (retest) are slightly higher than in most previous studies. Previous publications have demonstrated Cronbach’s alphas ranging from 0.70 to 0.80 [
5,
10,
14‐
17,
31] and reflects the homogeneity of each statement.
Construct validity is an important element of the validity of a questionnaire. As there were no comparable questionnaires for evaluating back beliefs in our prospective cohort, construct validity was assessed by testing predefined hypotheses about expected correlations to other measurements in our study. The hypotheses were based on existing literature on the BBQ and its assumed relation to similar or non-similar constructs. As expected, a good correlation between the BBQ and the FABQ-PA was found, similar to the original study by Symonds et al., in Britain [
5]. Other studies have been conducted on the BBQ and FABQ-PA with populations from different cultures and backgrounds, including Arabic and Chinese low back pain patients [
14,
34] and Chinese healthcare students and workers [
16,
17]. Their correlation analyses differ from ours and demonstrates low values ranging from − 0.02 [
16] to − 0.35 [
34]. These results show that healthcare professions and cultural background and origin are important aspects to consider when evaluating back beliefs and fear avoidance behavior due to physical activity. The moderate correlation between the BBQ and RMDQ was also as hypothesized, while the moderate correlation between the BBQ and PCS was slightly lower than expected. Two other studies have investigated the relationship between the Oswestry Disability Index and back beliefs and have found correlation between high disability status and negative back beliefs [
33,
40]. To the authors’ knowledge, no previous studies have assessed the relationship between pain catastrophizing (PCS) and back beliefs (BBQ), making this study the first to investigate the correlation between these two scales. The divergent validity was shown by the low correlation found between the BBQ and NRS, and our results are consistent with previous research. Other studies investigating the relationship between the BBQ and pain are finding a low correlation, demonstrating that the degree of pain experienced is not related to pessimistic back beliefs [
15,
17,
18,
33]. The exception is that patients with a 1-week history of severe back pain might have more pessimistic back beliefs [
31]. The correlation analysis confirmed 75% of the predefined hypotheses, indicating a good construct validity [
19].
One limitation of this study is that we could not prohibit participants from seeking medical advice or treatment between recruitment, baseline testing and retesting, and their back beliefs may therefore have been influenced by healthcare or alternative care practitioners. Furthermore, participants were recruited in primary care, and due to practical and economic considerations, there are no data on potential study participants that declined to participate.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.