The Attitudes and Beliefs of Physiotherapists Treating Back Pain : Development and validation of the Norwegian version of the Pain Attitudes and Beliefs Scale (PABS)

BACKGROUND: Low back pain (LBP) is a common musculoskeletal condition. constituting a significant health care problem in developed societies. The available literature suggests that negative attitudes and maladaptive beliefs of clinicians can serve as obstacles for the delivery of optimal care to patients with LBP. Furthermore, the back-pain beliefs of physiotherapists are found to influence the back-pain beliefs and the illness perceptions of their patients, with a profound effect on patient outcome. Evaluation of these attitudes and beliefs is necessary for the implementation of a comprehensive and more systematic patient management, in line with evidence-based clinical guidelines. The Pain Attitudes and Beliefs Scale for Physiotherapists (PABS-PT) is a questionnaire, originally developed in Dutch, that aims to measure physiotherapists’ attitudes and beliefs by differentiating between two dominant treatment orientations in musculoskeletal care on two subscales, one representing a biomedical, the other representing a biopsychosocial treatment orientation. OBJECTIVE: The purpose of the PhD thesis was to develop a Norwegian version of the PABS-PT, comprising a stepwise validation process in four parts: 1. Translation from the Dutch version into Norwegian and examination of structural validity and internal consistency. 2. Examination of internal construct validity based on modern test theory with Rasch modelling. 3. Examination of discriminative validity of the scale. 4. Examination of content validity. The four parts of this validation effort are covered in four papers. In PAPER I, the PABS was translated and cross-culturally adapted, followed by an examination of the underlying dimensionality and internal consistency. Data for exploratory factor analysis were collected from 647 physiotherapists responding to a cross-sectional web-based survey. Analysis revealed a two-factor structure and 36 items were reduced to 19 items, 13 items loading on the biomedical factor and 6 items loading on the biopsychosocial factor. Internal consistency was found to be sufficient for the biomedical but too low for the biopsychosocial subscale. The two factors accounted for low explained variance, which may be indicative for problematic construct validity. Therefore, in PAPER II, construct validity of the PABS-PT was examined more closely by applying Rasch modelling to the data collected in a sample of 667 physiotherapists. Rasch model analysis resulted in an improved scale with two strictly unidimensional subscales, each holding seven items, and with invariant item ordering and free from any form of misfit. The improved PABS allows for ordinal raw scores to be transformed into interval-level scores, rendering a greater accuracy to compare scores between groups of persons and justifying the use of parametric calculations, like means and differences. However, low separation indexes indicated limitations regarding the PABS’ ability to differentiate between clinicians with a traditional biomedical treatment orientation and clinicians with a biopsychosocial orientation. Therefore, in PAPER III we examined the discriminative validity of the PABS in more detail. A construct validation by hypothesis testing was performed using a sample of 662 physiotherapists. Twenty-four a priori hypotheses were formulated about expected differences between known subgroups of physiotherapists, based on evidence from research. Discriminative validity was considered adequate when at least 75% of the hypotheses were confirmed. Analysis showed that discriminative ability of each separate subscale was insufficient. Furthermore, subgroup analysis of respondents with high biomedical and low biopsychosocial scores (and vice versa) identified the presence of extreme treatment orientations. Differences in treatment orientation among physiotherapists were very small, indicating that Norwegian physiotherapists may basically be similar in their treatment orientation. Alternatively, they gave socially desirable responses, or the PABS is just not able to detect any differences between them. To explore the possible presence of social desirability or homogeneity in responses, the next step was to qualitatively examine the content validity of the PABS by assessing the relevance, comprehensibility and especially the comprehensiveness of the items. Therefore, in PAPER IV, we performed cognitive interviews using the Three-Step Test-Interview (TSTI) method and asked eleven physiotherapists how they understood and interpreted the items of the PABS. Our results indicated that all items were relevant and important for physiotherapists. Five of fourteen items had ambiguous formulations, but these can be handled with some slight modifications. The biomedical subscale appeared to be a comprehensive representation of biomedical treatment orientation. The biopsychosocial subscale, however, lacks comprehensiveness, as it is not able to capture important aspects of contemporary biopsychosocial best practice care. Measurement of biopsychosocial treatment orientation may therefore be incomplete. CONCLUSION: We developed a 14-items Norwegian version of the PABS and subjected it to a scrutinous validation process. We improved the scale performance by rendering the subscales strictly unidimensional, free from misfit and with an invariant, hierarchical item ordering. We were able to improve the item performance by suggesting some slight alterations for better comprehensibility. We identified two major shortcomings: the scale in its original form has poor discriminative ability and the biopsychosocial subscale has limited comprehensiveness, as it does not capture important aspect of contemporary biopsychosocial best practice care. RELEVANCE AND RECOMMANDATIONS: To provide broader insights into clinicians’ attitudes and beliefs, we recommend complementing the PABS with other questionnaires measuring related biopsychosocial constructs, such as practitioners’ confidence, patient-centeredness and knowledge of modern pain neurophysiology. Item performance may be improved by including the suggested item modifications. In addition, the applicability of the scale may be extended by combining high scores on one subscale and low score on the other, allowing the identification of physiotherapists with extreme (biomedical or biopsychosocial) treatment orientations