The Efficiency of Multiple Impulse Therapy for Musculoskeletal Complaints

Survival analysis is an important tool for understanding the treatment course of a condition. Its use by Collins et al 1 exemplifies misunderstanding of the rules that enhance external validity and credibility of clinical designs.

This is an uncontrolled retrospective case series on a heterogeneous population, without blinding, with rate of recovery selectively contrasted against the literature. Literature control and statistical analysis cannot alter that fact. Such methods are unintended for clinicians, rather than are a basis for hypothesis development. It provides no real evidence of treatment effectiveness.

The researchers discuss chronicity but ignore severity. Evidence exists that subgroups with differing severity/chronicity combinations respond differently. With severity being omitted, understanding of the dominant subgroup is lost. Instead, the authors cite disagreement over interpretation of the visual analog scale in the study of pain itself. The authors admit obtaining severity data that methodologically should define the population. At the same time, they used literature that categorized severity to contrast perceived differences in outcome. The preponderance of evidence supports pain scale use. If the authors did not like the visual analog scale, other ways to report severity exist. Besides making the designing of the next study more difficult, these omissions give little confidence in the interpretation of results.

From the authors' references, different chronicity/severity patient mix skewed the result to lengthier duration of treatment. Quoting from the original work, 2

In general, patients with less severe low back pain participated less often or did not complete the follow-up study. … Approximately three of every four patients whose index episode ended before the end of the follow-up period had one or more relapses within a year. The analysis resulted in a model with four variables predicting the duration of the low back pain, including ‘the duration of the low back pain preceding the initial visit,’ ‘receiving physical therapy,’ ‘pain intensity’, and ‘history of back surgery.’

By failing to report severity, an alternate explanation of results is that lesser severity patients were studied. The Kaplan-Meier analysis accurately reflects the recovery but will not show effectiveness.

Further contrasts made are misleading. The comparison study designs were different and unrelated. They had a predetermined follow-up interval and examined pain results. Collins et al report regular questioning on “relief,” a significantly different question from quantitative self-reporting pain, to which you expect a different answer. When at “total relief,” the study participation ended. This ignores the “recurrent” pain issues as is reflected by van den Hoogen et al2 and many others. Patients in practice and in clinical trials report occasional days of complete relief as a part of the episode course. Mixing basis of comparisons, as appears to have occurred, is a form of reporting bias that may well exaggerate the apparent outcome.

In all, this is an observational retrospective case series with some interesting results on an incompletely specified sample. Patient responses may be confounded with efforts to please their clinician (eg, exaggerating relief to their clinician—a known source of bias), and severity issues were not reported. The logic in comparing results to others is faulty. These limitations and their effects are largely ignored in the enthusiasm of the report.