Pain, 34 (1988) 231-244 Elsevier 231 PA1 01284 Review A tiicle Meta-analysis of non-medical treatments for chronic pain Marguerite D. Malone and Michael J. Strube Crestview Treatment Center, Washington University, St. Louis, MO (U.S.A.), and University of Alabama, Birmingham, AL (V.S.A.) (Received 11 August 1987, revision received 25 April 1988, accepted 12 May 1988) Summary A meta-anaIysis was conducted on 109 published studies which assessed the outcome of various non-medical treatments for chronic pain. Of these studies, 48 provided sufficient information to calculate effect sizes. The remainder were examined according to proportion of patients rated as improved. Studies were compared as a function of type of treatment, type of pain, and type of outcome variable. In general, effect sizes were positive and of modest magnitude indicating the short-term efficacy of most treatments for most types of pain. This finding suggests that the effectiveness of treatments may be attributable not to the differences between treatments, but to the features they have in common. Mood and number of subjective symptoms consistently showed greater responses to treatment than did pain intensity, pain duration, or frequency of pain, indicating the importance of using a multidimensional framework for pain assessment. This finding also suggests that the benefit of psychological approaches to pain management may he in reducing the fear and depression associated with pain, rather than relieving the pain itself. The present study also highlights the advantages of meta-analytic reviews. Key words: Chronic pain; Meta-analysis~ Medical tr~tment; Non-medics treatment Introduction Growing concern over the side effects of pharmacological and surgical treatments for chronic pain has spurred an interest in non-medi- cal interventions for such conditions. These inter- ventions include physical therapy, transcutaneous electrical nerve stimulation, and a variety of psy- chological approaches to pain m~a~ement. Such a diverse array of treatments, presenting com- plaints and outcome measures, makes reviewing this area challenging. Some of these difficulties were described by Turner and Chapman [12,13] in their review. As a result of these difficulties, they were able to state only that biofeedback training was not appreciably better than relaxation train- ing for alleviating pain due to headache. They were not able to draw firm conclusions regarding other forms of treatment. A meta-analytic review of headache treatments by Blanchard et al. fl] reported similar results. In addition, they stressed the importance of baseline or control group data in pain studies, noting that the absence of these data made it difficult to evaluate the effectiveness of treatment. Trifiletti [ll] found no definitive evidence in the literature to support the consistent effectiveness of any treatment and appealed for use of more sensitive multi~mensional ap- proaches in future pain studies. This paper at- tempts to address the above appeals for multidi- mensional and control group data by using meta- analytic procedures to appraise the status of non- medical treatments for chronic pain. Meta-analysis refers to the statistical analysis of the summary findings of individual experiments [e.g., 2,3,6,10]. Thus, meta-analysis is a technique which organizes and extracts information differ- ently than the traditional narrative review, In fact, the surge of interest in meta-analysis grew out of dissatisfaction with the narrative review. Tradi- 03~-3959/88/$03.~0 0 1988 Elsevier Science Publishers B.V. (Biomedical Division)