8

coping. These results were stable over 12 months and clinically significant. Despite the fact that the CBT treatment did not directly focus on pain behaviours, the present results support a moderate benefit of the treatment on behaviours (ES = 0.57, 0.49 adjusted). Apparently, changing patients' beliefs is a critical aspect of treatment, regardless of whether they are directly targeted or derived from the observation of the patients' own behaviour [38]. Significant changes for OBT were found with respect to pain and physical and behavioural variables. In accordance with previous reports [14, 21], healthy behaviours were successively increased and pain behaviours were decreased. The OBT, notably, achieved statistically significant reductions in physician visits (50%) in direct contrast to the AP group, which almost doubled the number of visits. CBT, however, produced only a modest and not statistically significant reduction of physician consultations. These results suggest that the OBT treatment may not only provide clinical benefits but also produce significant reductions in health care utilisation. As hypothesised, the analysis of clinical significance demonstrated that CBT had a relatively greater effect in the reduction of affective distress and catastrophising, whereas OBT had a relatively greater effect in the reduction in functional limitations, pain behaviours, and solicitous spouse behaviour. These data support the validity of the treatments. Regarding pain intensity and coping, CBT and OBT showed similar effects. CBT focused on changes of cognitions with the effect of reduced cognitive factors of pain, whereas OBT focused on behavioural changes and reached reductions of physical and operant components of pain. Although it is not surprising that CBT and OBT reached comparable effects in pain reduction, the focus of treatment used to accomplish these outcomes varied substantially. Taken together, these data suggest that CBT might be especially beneficial in patients with high levels of cognitive maladaptation to the pain and high affective distress, whereas OBT might be especially efficacious in patients with high levels of pain behaviours and low physical functioning. Future research will have to determine to what extent this is true or whether it might be beneficial to combine elements of both treatments. Unexpectedly, the AP group significantly deteriorated during the study, with a worsening of symptoms on all outcome measures, which may explain the large proportion (50%) of AP patients' terminating treatment prematurely. Unstructured discussion about problems aligned with coping with chronic pain appeared to lead to increased pain, functional limitations, emotional distress, and pain behaviours. This might be due to the disease-oriented, solicitous behaviour of group members which may have reinforced pain and pain behaviour. The deterioration of the AP was not limited to the pre-post comparison but persisted up to 12 months after treatment termination. One possible explanation for the large dropout in the AP group might be that the treatment was simply not credible or was viewed as unsatisfactory. This explanation was not supported by the results on patient satisfaction. There were no significant differences in patient satisfaction among the three groups. The long-lasting deterioration of the AP group was unexpected and needs to be explored. If confirmed, these results suggest that, at least for a subset of patients, discussions about pain and possibly informal support groups may be detrimental. There are several limitations in this study, and the large number of dropouts in the AP group is disconcerting indeed. The unexpected detrimental effects of the AP group are a significant concern and may have contributed to the large ESs reported. Because this result may have overestimated the true effects of the behavioural treatments, we also computed ESs for the carried-forward baseline data of the dropouts. The ESs that were obtained are still moderate to large, suggesting a substantial effect of the behavioural treatments. The interpretation of the results of the ANOVAs on the subgroups must be considered with caution due to the small sample sizes. Larger studies

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20