Do antidepressant medications relieve chronic low back pain?

Journal of Family Practice, Dec, 1993 by Judith A. Turner, Mary C. Denny

Discussion

The literature does not demonstrate the superiority of antidepressant medication over placebo in decreasing low back pain or associated disability and depression. Of the three studies that examined the effects of pain, two[32,33] found no pain-relieving properties of antidepressant medication, and one study[31] found an effect for antidepressant medication. This effect was found only for patient self-ratings of frequency of certain daily activities; no effects were found on ratings made by clinical investigators. In one study[32] a placebo was found to be superior to trazodone in decreasing disability, and in the third study,[34] no placebo-antidepressant differences in effects on disability were found. In three studies[31-33] that included both depressed and nondepressed patients, antidepressant medication and placebo were found to equivalent in improving depression measure scores. One study[34] found antidepressant medication to be superior to placebo in decreasing analgesic medication use; two studies[31,32] found no such difference. No study reported consistent superiority of antidepressant medication over placebo across outcome measures. These published studies may be a biased sample of the studies conducted; there may well exist other studies that were never published because they failed to find a significant difference between the medication and the placebo. Such publication bias (studies that show a treatment effect are more likely to be published) has been demonstrated in other clinical research areas.[41-43]

Although we must conclude that there is at this date no convincing scientific evidence supporting the use of antidepressant medication for low back pain relief, we must also emphaize that the literature does not conclusively demonstrate that antidepressants have no specific effects on low back pain, disability, and associated problems. The studies generally had small samples sizes and may have lacked statistical power to detect possible differences between antidepressant and placebo groups. Pooling data across studies using meta-analytic techniques could potentially reveal differences that individual studies were not sufficiently powerful to detect. Thus, it is particularly unfortunate that differences among the studies (especially in terms of medications, doses, and outcome measures), insufficient reporting of data, and the small number (three) of randomized, placebo-controlled trials (the study by Pheasant et al[34] used a cross-over design) precluded the use of meta-analytic statistical techniques for this literature synthesis. These factors greatly limit the conclusions that can be drawn concerning the study questions. Variations in patient populations sampled and in whether patients were allowed to use other analgesics, as well as poor specification of study inclusion and exclusion criteria, further contributed to the inability to pool data across studies and draw generalize conclusions.

Subjects in the six studies were drawn from different populations. These included rehabilitation unit inpatients, family practice outpatients, outpatients in a low back pain clinic and in a pain clinic, Veterans Administration Medical Center outpatients, and community volunteers. It cannot be determined from the literature whether patients with low back problems seen in different settings (eg, primary practice vs pain clinic) may respond differently to antidepressants. Only one study enrolled subjects from family practice settings, although several studies recruited community volunteers, who might be similar to patients drawn from primary care populations.