To the Editor:
In metaphorical terms, a problem arises when good apples are pooled with bad apples; they all get tarred with the same brush. This principle applies when all studies pertaining to spinal injections of steroids are pooled, as if they are all equal. Subsequently, the lay press publicizes sweeping conclusions such as “injecting any liquid, even plain saline solution, works just as well.”1 Such statements bring all injections into disrepute.
Admirably, Bicket et al.2 used an ingenious statistical exercise to explore the conjecture that epidural injections of other agents are not fair controls as epidural injections of steroids. However, in their exploration, they pooled data on cervical and lumbar injections, on image-guided injections and blind injections, and on interlaminar, caudal, and transforaminal injections; they even included studies that did not involve steroids. Given that these various targets and techniques differ with respect to pathology, anatomy, technical accuracy, and evidence base, such pooling might not be legitimate, and at least clouds the true picture.
Prominent among the studies analyzed is that of Ghahreman et al.3 which, indeed, the authors rank as rigorous. In the statistical analysis, this study stands out as an outlier; but it is also different in other respects. It is one of the few studies included in the review that used transforaminal injections, and it is the only study that actually addressed prospectively the very question being explored by the meta-analysis. In that regard, its results happen to contradict the conclusions of the review. It showed that the efficacy of transforaminal injection of steroid is significantly greater than that of transforaminal injection of nonsteroid. The authors of the review have referred to a conclusion that detecting a difference between treatment and control groups would not be practical but have not stated that this conclusion related specifically to long-term (12 months) outcomes.
It would have been more courteous, and more informative, had the authors stratified their analysis by region and by technique. Their conclusions might still apply to classical, blind epidural injections, but they would not apply to lumbar transforaminal injections. Lumbar transforaminal injection of steroids is significantly more often effective than transforaminal injection of either local anesthetic or saline, and intramuscular injection of either steroids or saline, and by the same magnitude in all cases.
It may be that these data could be overturned by future studies, but at present, they are the only direct data on this procedure. Those data defy the sweeping generalizations of the review, which are sensationalized by the lay press, and which serve the purpose of those who wish to deny reimbursement for epidural injections.
Meta-analysis of circumstantial evidence does not constitute proof; is not a substitute for well-designed controlled trials that address the issue. It serves only to raise an intriguing proposition worthy of studies that prospectively test it. In this regard, doubts may be raised about the propriety of interlaminar injections. However, lumbar transforaminal injections do not deserve to be tarred with the same brush, for the current data show that they are different. Let us not allow insurers to deny a treatment for which there is decent evidence of efficacy.
The review included another study of lumbar transforaminal injections, which was used to show favorable effect of epidural steroid over nonepidural steroid4 ; yet in this study, no steroid was used. The study used etanercept, which is neither a steroid nor an analgesic, and which has been shown to be no more effective than saline.5 It is curious why the review was contaminated by inclusion of this study.
None of the authors has a direct conflict of interest. Drs. Engel, Kennedy, and Bogduk serve as volunteer members of the International Spine Intervention Society (ISIS) Standards Division, and Drs. Kennedy, MacVicar, and Bogduk are volunteer members of the ISIS Board of Directors. Their views do not necessarily represent the views of ISIS.