Do antidepressant medications relieve chronic low back pain?

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

[TABULAR DATA OMITTED]

Is there a significant difference between antidepresent medication and a placebo in decreasing analgesic medication use? Three studies examined this question. Pheasant et al[34] found amitriptyline to be superior to atropine placebo; Alcoff et al[31] and Goodkin et al[32] found no difference between antidepressant medication and inert placebo. In the Pheasant et al study,[34] analgesic use was assessed by having a pharmacist count the unused prescribed analgesic pills every 2 weeks for 6 weeks. Prescribed analgesic pills every 2 weeks for 6 weeks. Prescribed analgesics were acetaminophen or aspirin, with or without codeine, according to patients' preferences. In the study by Alcoff et al,[31] analgesic use was measured by patient self-report of the average number of pain pills or tablets consumed per day in the past week. Goodkin et al[32] reported only on reductions in narcotics intake and not on other analgesic medication use. Their article did not state what statistical analyses were performed to compare reductions in narcotics intake or how narcotics use was operationally defined. None of these studies reported any subject exclusion criteria for the presence of substance abuse or dependence at the start of the treatment trials.

Are analgesic effects of antidepressants independent of antidepressant effects? Only the two studies by Ward and colleagues[35,36] in which all subjects were clinically depressed, examined whether changes in depression correlated significantly with changes in pain. The first study,[36] involving 26 subjects who completed a trial of doxepin or desipramine, reported a trend toward a significant relation between changes in pain and depression (r = .26, P = .099). However, the paper did not state which pain and depression measures were used in this analysis. The 1986 article by Ward,[35] which included these subjects plus nine additional subjects, did report a significant association. Decreases on the Hamilton Depression Rating Scale[39] (from a mean of 26.2 to 9.1) correlated .61 (P [less than] .001) with decreases in pain severitty (from 5.7 to 3.1 on a scale of 0 to 10) and .47 (P [less than] .003) with decreases in pain frequency (from 76.5% to 54.6% of the time). No study examined whether antidepressants were differently effective in relieving pain in depressed as compared with nondepressed patients.

Quality of Studies

Details of how patients were randomized to treatment groups were generally not reported. Although all studies were "double-blind," no study presented data on the number of patients who were able to correctly guess whether they were prescribed active medication or placebo. Only Goodkin et al[32] reported information concerning whether clinical evaluators were able to correctly guess medications. In no study was outcome clearly assessed independently of the treating clinician. This is a major problem because clinician ratings may be biased by guesses as to what treatment the patient is receiving based on knowledge of the side effects reported. Even when outcome is assessed by patient questionnaires, if these are completed in the presence of the treating clinician, subject responses may be biased by demand characteristics of that situation. No antidepressant-placebo comparison study presented information concerning the number of subjects diagnosed as clinically depressed by standardized criteria such as the DSM-III-R,[40] although earlier studies, of course, were performed before these were available. No study reported the results of power calculations, and all studies had fairly small samples; thus, the studies may have lacked sufficient power to detect true differences between antidepressant and placebo treatments. Attrition rates were high in all studies except the one by Goodkin et al.[32] Furthermore, only Goodkin et al[32] attempted to obtain and analyze outcome data for all patients who began the study regardless of compliance.