Healing as a therapy for human disease: a systematic review.

Abstract

To assess, from published clinical trials, the evidence for the use of healing as a complementary medical intervention in human disease. Limited to studies involving random assignment to a treatment group consisting of "healing," broadly defined, or to a concurrent control group. All randomized trials published up to the year 2000, were identified from MEDLINE, CINAHL, BIDS-EMBASE, the CISCOM complementary medicine databases and from bibliographic references of published articles. Copies of all published studies were obtained, data were extracted, and methodological quality (Jadad) scores were derived where possible. Fifty-nine randomized clinical trials (RCTs) were found comparing healing with a control intervention on human participants. In 37 of these, healing was used for existing diseases or symptoms (22 existed as fully accessible published reports, 10 as dissertation abstracts only, and 5 as "preliminary" investigations with limited evidential value). The 22 full trials (10 reporting a "significant" effect of healing compared with control) constitute an extremely heterogeneous group, varying greatly in the method and duration of healing; the medical condition treated; the outcome measure employed; and the control intervention used. Many trials had a number of methodological shortcomings, including small sample sizes, and were inadequately reported. Only 8 studies (5 with a significant outcome for healing) had a maximum methodological quality score of 5, and in 10 studies this score was 3 or less. Two trials-both large scale and methodologically sound-were replicates, and each found a significant beneficial effect of intercessory prayer on the clinical progress of cardiac patients. Eleven of the 15 dissertation abstracts and pilot studies reported nonsignificant results for healing compared with control, a finding that probably reflects the relatively small sample sizes and the likelihood of type II errors. The significant heterogeneity found in this group of trials makes categorization problematic and inhibits the pooling of results by meta-analysis or similar techniques to obtain a global estimate of the "treatment effect" of healing. No firm conclusions about the efficacy or inefficacy of healing can be drawn from this diverse group of RCTs. Given the current emphasis on evidence-based medicine, future investigations should be adequately powered, appropriately controlled, and properly described. These future investigations would most usefully consist of: (1) pragmatic trials of healing for undifferentiated conditions on patients based in general practice and (2) larger RCTs of distant healing on large numbers of patients with well-defined measurable illness.