Reliability & Validity

Reliability

In order to calculate reliability of the resulting scales, internal consistency (Cronbach’s α) was used. Table 6 shows internal consistency for all scales of the AEQ in comparison to the scales of the KPI. For better comparison of the behavioral endurance scales of both instruments, we decided also to report the reliability score of the combined behavioral endurance scale of the AEQ in addition to the sub-scales humor/distraction and pain persistence. Internal consistency was above r = 0.80 for all scales except catastrophizing (r = 0.78), which consists only of three items and for the behavioral endurance sub-scales humor/distraction (r = 0.78) and pain persistence (r = 0.76). In sum, first data showed good to excellent internal consistency for all sub-scales of the AEQ.

Criterion-related validity

In order to compare the findings of the AEQ with the results of a reference criterion, self-reported pain intensity and disability were used as criterion-related variables, as they are most used as outcomes in low back pain studies (Bombardier, 2000; Ostelo et al., 2008). As we expected that FAR as well as ER will represent maladaptive responses to pain that lead to maintenance of pain and promote the development of chronic pain problems, both groups of variables should show positive correlations with pain intensity. As suggested by the AEM outlined above, FAR variables should also be positively related to disability whereas ER should be negatively associated with self-reported disability. Results of bivariate correlations revealed that in accordance to these expectations most FAR and ER variables were positively related to pain intensity, although the correlations were small (between r = 0.16 and 0.28). The highest correlations were observed for both, FAR and ER variables and self-reported disability, as expected, in an opposite direction (see Table 7). FARs corresponded positively with disability, while affective and behavioral ERs were negatively related (between r = .24 and .45). Further, the FAR variables ADS, ASAS and APAS were positively related to the number days with sick listing whereas the ER variables PMS and PPS were negatively related to sick listing behavior. The four CPG grades differ with respect to pain intensity and self-reported disability. CPG 1 represents low pain intensity and low disability, CPG 2 high pain intensity but low disability, CPG 3 represents moderate degrees of disability and CPG 4 high disability, both irrespective of pain intensity. The CPG grades are therefore suitable for testing suggestions of the AE model. It was hypothesized, that endurance-responses will be highest in CPG 2 as patients with high scores on endurance scales will withstand their daily activities despite severe pain or at the costs of an increase of pain. In contrast, FAR responses are supposed to be highest in CPG 4 which was represented by high disability. Separate analyses of variance (ANOVAs) and ANCOVAs revealed significant group differences as expected for the FAR variables ADS, HHS and both avoidance sub-scales, but not for CTS (see Table 8). The ANOVAs remained significant after Bonferroni adjustment for multiple comparisons. Further, the behavioral ER scales, especially the BES total score and the HDS sub-scale showed higher scores in CPG 2 compared to CPG 4 and 3. Positive mood was lowest in CPG 4 and significantly different to CPG 1. TTS was highest in CPG 2, as expected, but this difference was only significant for the comparison with CPG 1.

Content validity

Most AEQ scales revealed associations in the expected direction. The affective responses ADS and PMS showed the highest correlations with BDI-Depression and PASS-anxiety with positive relationships for ADS (r = .50 and .54) and negative for PMS (r = -.49 and -.45). Furthermore, ADS was positively correlated with work-related fear-avoidance beliefs (r = .41) and, to a lesser degree to beliefs of kinesophobia (r = .19), measured with the TSK. As expected, there were also negative medium correlations between the PMS and fear-avoidance beliefs (r = -.29 and -.31). In addition, there were medium to high correlations between most of the cognitive and behavioral sub-scales of the AEQ and BDI, PASS, FABQwork and TSK, with positive correlations for FAR and negative for ER. The APAS, measuring avoidance of physical activities, was the only scale that revealed a significant correlation with FABQ-activity (r = .23). The TSS was only related to depression (r = .26), but not to other fear-avoidance variables.