To the Editor:
I read with interest the article “Characterization of side effects sustained by chiropractic students during their undergraduate training in technique class at a chiropractic college: a preliminary retrospective study” by Macanuel et al. and I have a couple of comments:
It is a worthwhile cause to look into the safety of chiropractic students. This study is therefore certainly both timely and relevant. However, it is always important that the research we conduct and publish is carefully conducted and meticulously reported, in particular so in politically sensitive areas. Using a common Scandinavian phrase as an analogy, sometimes readers deal with written text “like the devil reads the bible”; in other words, deliberately taking something out of context and twisting the meaning. From this point of view, I have some problems with this report.
My main problem is the choice of vocabulary. The term “side-effect” is a general term that needs further definition, i.e. what type of side-effect was this and was it common or uncommon, mild or severe, acceptable or unacceptable. Thus side-effects can be transient and of no physical or pathological consequence, as is probably the case in most reactions following spinal manipulative treatment, i.e. the “normal” reactions. Side-effects can also be transient even if they are of a more sinister type, because Mother Nature allowed healing to occur. These would be “unsuitable” reactions but without further consequences. Unsuitable reactions, whether mild or severe, that persist are most likely signs of real injuries. The authors of the present article correctly use the term “side-effect” in the title but, unfortunately, also use the term “injury” in the text, as a synonym to “side-effect”. This, as explained, is not a good idea; it gives the reader (possibly the devil) the idea that students who treat each other may inflict irreversible injuries on each other – or perhaps on themselves.
My second problem has to do with the calculation of percentages, which requires a numerator (the number you find) and a denominator (out of so many), for example 12 students out of 120 (10%) who received a lumbar roll reported a side-effect. Unless you know the total number of treatments or persons submitted to treatments, you cannot determine whether the figure you report represents a large or a small proportion of the total number. For example, how would you know, whether it is less risky to treat the lumbar spine (35% of the reported “injuries”) than the cervical spine (44% of the reported “injuries”), unless you know the total number of cervical and lumbar spine adjustments? The same is true for procedures; perhaps the 1.9% of all “injuries” reported by students to occur after a drop-piece treatment was in fact very high, because almost no such procedures were performed, whereas the much more frequently reported “injuries” appearing after side-posture adjustments were relatively very rare, because they arose out of a very large number of such procedures.
My third problem is, related to the inequality of the recall periods between the participant groups from the different years of the study. According to Table 3, which I have recreated below, the 161 “injuries” occurred mainly in the second year of study.
According to the authors, technique classes were not common in the first year. Technique was probably an important aspect of the education from the second till the fourth year. It therefore appears, and the authors also point out, that year two is particularly dangerous.
However, participants in the survey were students from the second, third and fourth year of study. In other words, the fourth year students could report from the first to the fourth year, the third year students from the first to the third year, and the second year students from the first and second year. In other words, year two could be reported on by all but year three only by some and year four only by even fewer. Therefore, there is bound to be a higher number of events from the second year than from the third or fourth but this may have nothing to do with the proportion of treatment sessions that resulted in such reactions. Therefore, there is insufficient evidence to appoint year two as the most dangerous period.
| Year of study | Count | Percent |
|---|---|---|
| First year | 37 | 22.9 |
| Second year | 95 | 59.0 |
| Third year | 28 | 17.4 |
| Fourth year | 1 | 0.01 |
My fourth concern is that the authors do not seem to consider the potential consequences of the low response rate. Who was more likely to respond, the one with an unpleasant event or the one without? Could this have resulted in an over- or under-reporting of reactions?
And finally, I find the retrospective study design problematic. This may be acceptable if the recall period is not too long, but how can anybody be expected to remember details about these events up to three years ago? Even I, with the special interest I have in this research area, do not think that I would be able to recall when a reaction started, how long it lasted and how bad it was three years later.
My closing comment is that “preliminary” studies should keep a certain intellectual standard so as not to make it too easy for the devil when he sits down to read scientific articles.