Response to Keele Team’s Response Letter

We would like to thank Hill et al (7) for taking the time to provide feedback on our recent publication in Pain (4). I would also like to thank the Editor for providing us the opportunity to respond to their comments.

Our clinical trial investigated the addition of behavioral treatments (graded exercise and graded exposure) to physical therapy determined by clinical prediction rules. Results from the trial suggest no additional benefit from the behavioral treatments for patients with acute/sub-acute low back pain at 4-week and 6-month outcomes (4). Hill et al (7) suggest that one conclusion drawn from our data is that psychosocial interventions should be put “out to pasture – page XXX.” We would like to clarify that these are their words, not ours. Such a strong recommendation is unwarranted based on data from a single clinical trial. Our own interpretation of these data included several other equally feasible reasons for our null findings and our suggestions for future study included methodology that incorporates psychosocial interventions (4).

Differences in interpretation aside, a primary issue that Hill et al (7) seem to have with our clinical trial is that the psychosocial sub-groups were determined in an “oversimplified” fashion. On this point we agree with the authors. In comparison to data available in 2008 for primary care (6) and physical therapy settings (3), the approach used to identify psychosocial sub-groups was simplified. In contrast, there were available data for physical therapy clinical prediction rules when this trial was being planned (1,5). We do, however, take umbrage with the description of this psychosocial sub-group identification as “arbitrary”. The same cut off score was used in a previous clinical trial reported by George et al (2). In this trial a differential treatment effect was observed such that those with elevated fear-avoidance beliefs had a greater benefit from graded exercise supplemented physical therapy. We believe the use of the same cut off score was an appropriate and necessary methodological choice for the follow up trial in which we attempted to replicate previously observed treatment effects.

A secondary issue that Hill et al bring up is that our clinical trial was potentially underpowered, an important concern when null findings are reported. We disagree with the authors on this point. The prospective power analyses were based on effect sizes observed in our trial reported in 2003 (2). The observed effect sizes and power for the current trial were calculated, and were much smaller than in the previous trial. These data were reported to demonstrate how quite small these effects were, not to justify the need for larger studies in this area. The current trial was adequately powered to detect clinically meaningful effects, but not adequately powered to detect statistical significance of miniscule effects.

Future study will determine if larger treatment effects are associated with more refined ways of determining psychosocial sub-groups, in comparison to readily available clinical prediction rules. This is a point we feel was made quite clearly in our manuscript; “Future study in patients with LBP should focus on improving the identification of patients at risk for developing chronic LBP and refining specific psychological targets that optimally reduce pain intensity in acute and sub-acute phase – page 157”(4).

Sincerely,