Oppel's comments regarding alternative control groups are good suggestions for future research but would have required more time and funds than were available (e.g., recruitment of naive subjects, provision of sham massage). An attempt was made to dilute the subjects' pre-existing expectations by indicating in the advertisements that subjects might receive one or more treatment modalities. Dropout rates also partially reflect pre-existing expectations of treatment.² Each group experienced a similar number of dropouts (l or 2 subjects per group).

Oppel's concerns about the accuracy of reporting the self-rated measures and the possible provider influence on subjects' perceptions are valid, and both were addressed in the article. Measures were clearly stated as self reported or observer recorded, and unknown provider effects were stated as a limitation of the study. In my review of the literature I found no study that employed a truly objective measure of subacute back pain (e.g., laboratory investigations).

Sedergreen's first 3 comments relate to subject inclusion and characteristics. An attempt was made to produce a sample representative of the typical patient load of massage therapists. The screening protocol was reviewed and approved by several staff physicians, and history-taking and physical examination also helped to rule out contraindications to massage therapy as well as the presence of exclusion criteria. Ancillary tests are appropriate when indicated and should not be routine.³

Sedergreen was also concerned about the potential influence of the nonblinded providers of sham laser treatment. This was not reported as a double-blinded study, nor was double blinding feasible. One finding not in the published report was that at post-test, 8% of the subjects in the sham laser group indicated that they had no pain as compared with 5% in the exercise and education group. Both providers of the exercise and education believed exercise to be an effective remedy for subacute low back pain. In this study there is no clear link between the nonblinded treatment provider and subjects' self-reported outcomes.

It is true that medication use was not considered during randomization; however, only 6 subjects indicated analgesic use and they were fairly evenly dispersed among the 4 groups. Each of these 6 subjects scored within the 95% confidence interval of their group mean at each time.

In terms of secondary gain, the case histories revealed that no patients were receiving disability payments or compensation for their low-back pain, and this issue was thus not mentioned.

Regarding interaction, this study revealed that some part of the interaction between massage therapist and patient is beneficial within a specified treatment protocol. It was not within the scope of this study to determine the mechanism of remediation.

This study provided some evidence of the effectiveness of massage therapy for some patients with subacute low-back pain. One randomized controlled trial cannot provide conclusive evidence for treatment effectiveness; more research is clearly needed.