First, the screening process relied upon self-reported criteria. Such reporting is unreliable. Even when supplemented by information from physician files (which only occurred in selected cases in this study) it may be incomplete.

Second, significant pathology was not reliably excluded. The management of mechanical back pain is not necessarily the same as that of back pain from metastatic or metabolic disease, for example. The patient may be unaware of either of these circumstances, which moreover may not be apparent from a plain radiograph.

Third, the ages of the subjects were not defined; only the mean was reported. Approaches to management of back pain may vary considerably between patients who are 35 years old and 70 years old.

Fourth, although the patient was supposedly blinded to the sham nature of the laser therapy, it is not reported that the operator of the equipment was similarly blinded. The potential for unconscious communication of the ineffectiveness of this treatment is substantial.

Fifth, patients were asked to refrain from analgesic use only on the days that they were being evaluated. Since some took medication and others did not, 2 subsets of patients existed, the distribution of which wasn't necessarily randomized.

Finally, there was no screening to determine the presence or absence of secondary gain issues such as compensation or avoidance behaviours.

It may be argued that the interaction between a massage therapist and a patient is particularly vulnerable to producing a placebo response, in which case the obligation of researchers in this field to disprove such bias is substantially increased. Massage may well feel nice but there is scant evidence that it should be considered therapy.