Meta-analysis: Acupuncture for Osteoarthritis of the Knee

Context Previous studies have come to inconsistent conclusions about the effectiveness of acupuncture for treating knee osteoarthritis. Contribution This meta-analysis of 9 trials showed that sham-controlled trials identified no clinically meaningful short-term benefits in pain or function with acupuncture for knee osteo-arthritis, although trials that did not use a sham control identified some benefits. Implications The use of different types of comparisons (sham acupuncture vs. interventions in which the participant knew whether they were receiving acupuncture) explains the variability in the conclusions of published trials about the effectiveness of acupuncture for treating knee osteo-arthritis. Placebo or expectation effects probably account for the observed benefits. The Editors Osteoarthritis is the leading cause of disability among older adults (1, 2). The joint most commonly affected by osteoarthritis is the knee (3, 4). The prevalence, disability, and associated costs of knee osteoarthritis are expected to steadily increase over the next 25 years because of aging in the population (5). Nonsteroidal anti-inflammatory drugs (NSAIDs) and acetaminophen are the most commonly used pharmacologic agents for treating knee osteoarthritis (6, 7). However, according to a recent systematic review (8), NSAIDs are only slightly better than placebo in providing short-term pain relief and their effects are probably too small to be meaningful to patients (8). Furthermore, many NSAIDs are associated with considerable side effects (9). Gastrointestinal bleeding, the most clinically substantial effect (10), causes approximately 2200 deaths and 12000 emergency hospital admissions each year in the United Kingdom alone (11) and is of particular concern to older patients (10). Acetaminophen may have a better toxicity profile than that of NSAIDs (6). However, a recent systematic review (12) suggests that acetaminophen is modestly less effective than NSAIDs and that the clinical significance of acetaminophen is questionable because it results in only a 5% greater improvement from baseline in pain than does placebo in the short term. The evidence for nonpharmacologic treatments for knee osteoarthritis is generally sparse and inconclusive (13). However, 2 effective nonpharmacologic treatments are exercise (14) and weight loss (15). Some patients with osteoarthritis, however, may have difficulty exercising or losing weight. The need for additional safe and effective treatments for osteoarthritis is clear. Acupuncture is a safe treatment that has a low risk for serious side effects (1619). Given its safety, whether acupuncture is effective for treating osteoarthritis of the knee is a highly relevant question. Our objective was to conduct a systematic review and meta-analysis of the effects of acupuncture for treating knee osteoarthritis. Methods Data Sources and Study Selection We searched the MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials databases to January 2007 to identify randomized, controlled trials (RCTs). We combined acupuncture-related terms with osteoarthritis-related terms and limited the search to RCTs (20). We considered older RCTs that were included in previous reviews of acupuncture for osteoarthritis (2124) for inclusion. Two authors independently selected articles and resolved disagreements by discussion. Our selection criteria were published RCTs of acupuncture in patients who had received a diagnosis of knee osteoarthritis. We considered the 2 outcomes of pain and function. We applied no language restrictions. We included only RCTs in which the acupuncture treatment involved the insertion of needles into traditional meridian points. The needles could be inserted into tender points in addition to the traditional meridian points and could be electrically stimulated. We excluded RCTs of dry needling or trigger-point therapy. We also excluded RCTs that compared only 2 different forms of active acupuncture. We prespecified that trials have at least 6 weeks of observation. This criterion has not been validated as a threshold for study inclusion. However, we thought that RCTs with observation periods less than 6 weeks may also have methodological shortcomings (25, 26) that may exaggerate their results of benefits (27, 28). Data Extraction and Quality Assessment Two authors independently extracted data and resolved disagreements by discussion. They extracted information pertaining to the quality of the methods, participants, interventions, and outcomes (including adverse effects). When a study reported more than 1 pain or function outcome measure, we gave preference to the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain and function measures because the WOMAC is the most widely used and thoroughly validated instrument for assessing patients with knee osteoarthritis (2931). We contacted corresponding authors via e-mail and requested that they review the information we extracted from their studies, clarify any ambiguities, and supply missing information. We evaluated internal validity of the RCTs by using an 11-item scale developed by the Cochrane Back Review Group (32). We considered a score of 6 or more points to indicate high internal validity on the basis of data from ongoing research, in which 1 of the authors is involved. For the patient and outcomes assessor blinding items on the scale, we assigned sham-controlled trials 0.5 point rather than 1 point because we could not be certain that all shams were sufficiently credible in fully blinding patients to the treatment being evaluated. However, we assigned 1 point to sham-controlled trials that evaluated the credibility of the sham and found it to be indistinguishable from true acupuncture. Acupuncture Assessment No consensus exists on how best to assess treatment adequacy in acupuncture RCTs, and no methods have been validated (26). We used a method that involved assessing the adequacy of the following 4 aspects of the acupuncture treatment: choice of acupuncture points, number of sessions, needling technique, and experience of the acupuncturist. The adequacy of the sham intervention was also assessed by using an open-ended question. Two acupuncturists, who had previously used this adequacy assessment instrument for an earlier systematic review (33) on acupuncture, made these assessments. The acupuncturists assessed adequacy independently and achieved consensus by discussion. Assessments were based on only the description of the study population and the acupuncture procedure. The assessors were blinded to the results of the study and the publication. We asked the assessors to guess the identity of each study being assessed to test the success of the blinding. Data Synthesis and Analysis We placed RCTs into categories according to control groups, which were sham, usual care, and waiting list. We defined sham control as a sham intervention that was designed to be credible as the active treatment. We defined the usual care control as groups that received some additional standard care therapy that was not provided to the acupuncture group and waiting list control as groups that received no care while waiting for acupuncture. For our meta-analyses, we defined the short-term follow-up point as the measurement point closest to 8 weeks but no longer than 3 months after randomization. We defined the long-term outcome as the measurement point closest to 6 months but longer than 3 months after randomization. Because some RCTs used the visual analogue scale version of the WOMAC instrument and others used the Likert version, we used standardized mean differences as the principal measure of effect size so that the results of the RCTs could be combined. We calculated standardized mean differences (Hedge adjusted g) for all RCTs by using differences in improvements between groups divided by the SDs of improvements pooled from the 2 groups (34). For 3 RCTs (3537), we made some conservative assumptions to compute the standardized mean differences (Appendix Table 1). Appendix Table 1. Original Data from Included Studies and Assumptions Used to Derive the Meta-analysis Study Data* We used the DerSimonian and Laird (38) model, which is the random-effects model used in RevMan software, version 4.2 (Nordic Cochrane Centre, Copenhagen, Denmark) (39). This model estimates the average treatment effect by incorporating heterogeneity among clinically diverse trials with different, but related, treatment effects (40). When heterogeneity exists, the model (38) assigns smaller studies more weight than they would receive in a fixed-effects model (40). To evaluate statistical heterogeneity within our trial categories, we used I 2 tests on all outcomes included in our meta-analysis. We conducted sensitivity analyses for the short-term outcome of sham-controlled RCTs, restricting the analyses to RCTs assessed as adequate based on each item of the 11-item Cochrane scale (32) and the 4 aspects of the acupuncture treatment adequacy. We performed additional sensitivity analyses for funding source (industry vs. nonindustry) and follow-up length (3 months). We also evaluated whether the pooled effects of acupuncture met the threshold for minimal clinically important differences, defined as the smallest differences in scores that patients would perceive to be beneficial (29). The clinically relevant effects for knee osteoarthritis have been estimated to be standardized mean differences of 0.39 for WOMAC pain and 0.37 for WOMAC function (29). Role of the Funding Source The study was funded by the National Institutes of Health, National Center for Complementary and Alternative Medicine. The funding source had no role in the design, conduct, or reporting of the study or in the decision to submit the manuscript for publication. Results Study Characteristics We included 11 RCTs (3537, 4148) of 2821 patients with osteoarthritis (Figure 1). All studies but 1 (42) we