Controlled trials, systematic reviews and meta-analyses: acupuncture effective for patients with knee osteoarthritis

ABSTRACT
There are multiple randomized controlled trials that have investigated the use of acupuncture in patients with knee osteoarthritis, and all but one of them came to a very different conclusion than Hinman et al. Recent systematic reviews and meta-analyses provide strong and overwhelming evidence for the effectiveness of acupuncture in the treatment of knee osteoarthritis. Additionally, unlike 11 of the 13 randomized controlled trials in the Cochrane Database systematic review that found acupuncture effective in patients with knee osteoarthritis, Hinman et al did not radiographically confirm the clinical diagnosis of knee osteoarthritis. Therefore, it was with interest and some concern that we read the study by Hinman et al published in the October 2014 issue of JAMA titled “Acupuncture for Chronic Knee Pain: A Randomized Clinical Trial.” We credit the authors for using a Zelen-design that likely reduced the risk of selection bias during recruitment. However, there are a number of methodologic issues that appear to impact both the internal and external validity of this study. Thus, for the primary outcomes of pain and physical function, the conclusion made by Hinman et al that “needle and laser acupuncture were no more efficacious than sham laser acupuncture” should be viewed cautiously.

MANUSCRIPT
According to the Cochrane Database systematic review on acupuncture for peripheral joint osteoarthritis, Manheimer et al1 found acupuncture to be associated with a statistically significant and clinically meaningful short term improvement in OA pain when compared to a wait list control. Additionally, this systematic review,1 which included 12 RCTs of patients with knee OA and 4 trials of patients with either knee or hip OA, reported statistically significant reductions in pain following acupuncture in patients with knee OA at 6 months when compared with sham acupuncture. Moreover, in an individual patient data meta-analysis of 9 RCTs comparing real, sham and no acupuncture for chronic pain conditions, Vickers et al2 reported superior outcomes using real acupuncture in the treatment of knee OA. Additionally, a very recent meta-analysis3 of 11 high-quality RCTs concluded that real acupuncture provides a significant reduction in pain immediately following treatment compared to other physical treatment methods, including sham acupuncture.

Notably, one earlier trial4 found the addition of real acupuncture to a course of advice and exercise for the treatment of knee OA provided no additional improvement in the WOMAC pain subscale at 6 months when compared to sham acupuncture. The data from Foster et al4 also failed to demonstrate a significant relationship between patient treatment preferences and clinical outcomes or patient expectations and pain at 6 and 12 months. However, the results of Foster et al4 should be viewed cautiously due to the limited number of treatment sessions in the acupuncture protocol compared to other studies,5-8 which may have rendered the true acupuncture intervention suboptimal, a concession that the authors independently made. More importantly, unlike other trials,5-7,9-11 the subjects in the Foster et al4 trial (and now the Hinman et al12 trial) did not have radiographically confirmed knee osteoarthritis. Nevertheless, despite the methodological differences of the Foster et al4 trial, a cost-utility analysis of the Foster et al4 data by Whitehurst et al13 concluded that advice and exercise plus real acupuncture delivered by physical therapists still provided a cost effective use of health care resources.

It is of major concern that the description of the acupuncture treatment provided by Hinman et al,12 in both the methods section and eTable 2 of the supplement, is incomplete at best, and more importantly, it clearly does not meet the internationally established Standards for Reporting Interventions in Controlled Trials of Acupuncture (STRICTA14) or the Consolidated Standards for Reporting Trials (CONSORT15,16). More specifically, the depth of needle insertion was not disclosed, the amount and kind of manual needle manipulation was not described, the elicitation of a deqi response was apparently not attempted during needle insertion, and unlike most of the recent RCTs showing positive outcomes in patients with knee OA, the use of electroacupuncture was not employed. Additionally, the number and placement of needles were not standardized between practitioners or patients, ranging between just 4 local points at the knee to 31 possible acupoints including the ear, vertex of the head, foot, ankle, elbow, cervicothoracic junction, and thoracolumbar junction (for patients with knee osteoarthritis!); thus, the internal validity of this study appears jeopardized and more importantly, given the inadequate description of the acupuncture intervention, replication of this study by another research team in a similar or different patient population would seem impossible.16,17 Moreover, the 20 minute treatment sessions employed by Hinman et al are shorter in duration than what has previously been used in successful acupuncture trials in patients with knee osteoarthritis; hence, this could additionally render the protocol suboptimal and perhaps explain why Hinman et al found no significant between-group difference. There are several original trials, literature reviews and meta-analyses that support our position on each of the above items.

Four years ago, in the British Medical Journal, Glasziou et al17 reported, “The results of thousands of trials are never acted on because their published reports do not describe the interventions in enough detail.” Furthermore, a review of 80 studies selected for the journal Evidence Based Medicine, as both valid and important for clinical practice, found that clinicians could replicate the intervention in only half of the studies.16 More specifically, in his paper titled “Taking healthcare interventions from trial to practice”, Glasziou et al17 stated “the methods section in a protocol should provide a description of the intervention(s) that is sufficiently detailed to enable people with appropriate expertise to reproduce them…including the dose….” The 2010 update of the CONSORT statement15 requires authors to describe “the interventions for each group with sufficient details to allow replication, including how and when they were actually administrated.”

Contrary to the CONSORT and STRICTA guidelines,15,18 Hinman et al12 did not describe the real acupuncture intervention with sufficient detail such that it could be critically appraised, appropriately analyzed, or even replicated by practicing clinicians or by researchers in future trials. The STRICTA guidelines18 state, “The depth of insertion, whether standardized or individualized, should be expressed…in terms of anatomic depth, for example subcutaneous tissue, fascia, muscle, or periosteum; or in millimeters;” however, we still do not know whether Hinman et al12 employed superficial (3-5 mm depth), deep (15-20+ mm depth), subcutaneous, intramuscular, or periosteal acupuncture treatments. The STRICTA guidelines18 state, “Where a trial protocol includes individualization of treatment, the rationale for the treatments should be documented. Whatever the active intervention, the sources that justify the underlying rationale must be explicit, whether these are from the literature, expert clinical and research panels, practitioner surveys, or any combination of sources;” however, Hinman et al12 did not provide adequate justification of the underlying rationale used to support the individualized point selection from the 31+ acupoints in the foot, ankle, vertex of the head, elbow, cervicothoracic junction or thoracolumbar junction for patients with a primary complaint of knee pain. The STRICTA guidelines18 state, “Needle stimulation techniques, when used, should be clearly described. For manual stimulation such techniques include lifting, thrusting or rotating the needle to manipulate the de qi sensation” should be recorded; however, we don’t know if and how the needles were manually manipulated during or even after insertion (at 5 or 10 minute intervals as has been traditionally used) in the Hinman et al12 trial. The STRICTA guidelines18 recommend, “the de qi sensation in traditional Chinese acupuncture…should be documented;” yet, we don’t know if the elicitation of a de qi response was even attempted during needle insertion in the Hinman et al trial that reportedly “used a combined Western and traditional Chinese medicine style of acupuncture.”12 Lastly, although Hinman et al12 reported the mean number of years of general acupuncture experience by the eight family practice medical physicians who delivered the interventions, contrary to the STRICTA standards (“details of [practitioner] expertise in treating the specific condition being evaluated”18 should be disclosed), no evidence of prior experience of these medical physicians using acupuncture for specifically knee osteoarthritis was provided. Certainly it has been well established that the background of the acupuncture practitioner will influence the nature of the acupuncture treatment and likely the outcome also.14

In a systematic review of acupuncture trials for knee OA, White et al19 defines adequate dosage as when “at least 6 treatments were given, at least one per week, with at least four points needled for each painful knee for at least 20 minutes, and either needle sensation (de qi) achieved in manual acupuncture, or electrical stimulation of sufficient intensity to produce more than minimal sensation”. According to this criteria,19 the acupuncture protocol used by Hinman et al12 seems inadequate because no manual manipulation of the needles or electric needle stimulation was provided. Needle stimulation seems to play a significant role in the outcomes of acupuncture treatments. Although the concept of de qi, defined as a dull ache, heaviness, distension, numbness, tingling, cramping, pressure, fullness, spreading, warmth, or cooling, is still a topic of controversy in the literature, the vast majority of acupuncture trials have attempted to elicit a de qi response.20,21 While some authors contest that needle sensation has no affect on pain relief, other authors consider it a gross marker of acupuncture effectiveness.22 Notably, a recent randomized control trial by Choi et al23 found de qi secondary to needle rotation led to a significantly greater analgesic effect compared with de qi secondary to increased needle depth. This finding seems to coincide well with Langevin’s work that demonstrated the production of needle grasp or de qi with unidirectional winding is associated with collagen fiber mediated mechano-transduction of fibroblasts and subsequently up regulation of enzymes and proteins required for repair and healing.23-27 Therefore, it is not surprising that a recent Cochrane Review by Manheimer et al1 found only 2 of 13 trials for peripheral joint osteoarthritis did not attempt to elicit a de qi response during or following needle insertion. Considering real electroacupuncture (compared with sham electroacupuncture) targeting mainly local points at the knee has been found effective in patients with knee osteoarthritis in multiple controlled trials and systematic reviews,1,3,5-7,9-11 has been shown to modulate blood flow (26% increase in arteriolar diameter and a 17% increase in mean arterial pressure) to the knee joint through autonomic reflexes and nitric oxide,28 and has been found with fMRI to activate the hypothalamus and deactivate the limbic system (locally over the anterolateral aspect of the knee just below the joint line and with elicitation of a strong deqi response),29we find it rather odd that Hinman et al12 made no mention of rotating, winding, pistoning or manipulating needles in any way manually in an attempt to elicit a de qi response as part of the treatment protocol. Thus, a high risk of under dosage seems to be apparent yet again in the Hinman et al12 study.

We further question the decision of Hinman et al12 to not incorporate electric stimulation into their acupuncture treatments. Of all of the acupuncture characteristics investigated by MacPherson et al30 in a meta-analysis of 17,922 patients with chronic pain in randomized controlled trials, a sensitivity analysis revealed that “trials allowing for electric stimulation had a significantly stronger effect of acupuncture compared to sham controls”. While the authors did not find significantly better outcomes for electric acupuncture versus sham or non-acupuncture treatments, the univariate meta-regression analysis demonstrated a strong trend for better outcomes with electroacupuncture.30 Electroacupuncture has been associated with the added release of endogenous opioids such as enkephalin and dynorphin31,32 and stress hormones such as cortisol.33 In addition, electroacupuncture has been found to result in more widespread activation of the pain centers of the brain including the somatosensory cortex, the anterior middle cingulate, and the nucleus raphe pontis not seen in manual acupuncture.34 Additionally, a recent randomized controlled trial of patients with knee osteoarthritis found electroacupuncture provided superior outcomes when compared to diclofenac.35 Therefore, it is not surprising that after examining 13 RCTs consisting of 2428 patients with knee OA that Manheimer et al1 concluded, “our sensitivity analysis suggests that electric stimulation may be associated with better outcomes, and the two sham-controlled trials11,35 in this review that used electric stimulation of all local knee points showed the greatest benefits.”1 Therefore, given the existing literature,1,30,32-35 it appears that the manual acupuncture employed by Hinman et al12 was less than optimal for patients with knee osteoarthritis.

Additionally, Hinman et al12 did not standardize the number and placement of needles between practitioners, between treatment sessions, or between patients. Leaving selection of 31 possible acupoints (see eTable 2 of the supplement) to the discretion of each practitioner, at each site, on each patient, on each session, for patients that apparently had the same or similar condition (i.e. knee osteoarthritis) with acupoint locations dramatically varying from the ear, vertex of the head, foot, ankle, elbow, cervicothoracic junction, and thoracolumbar junction appears to have severely jeopardized the internal validity of the Hinman et al12 study. The real acupuncture intervention clearly was not standardized and this assuredly makes replication of the Hinman et al trial almost impossible.

A recent exploratory review attempted to delineate the most important aspects of acupuncture for the treatment of chronic pain, including osteoarthritis.30 In addition to patient expectation, MacPherson et al30 found the vast majority of acupoints that have been utilized in previous trials for the treatment of knee OA are in the vicinity of the knee on the affected lower extremity; moreover, “the data do not support the idea that needling distant points, either in the feet or the hands, is likely to improve the patient’s response.”30 However, the family practice physicians delivering the acupuncture intervention in the Hinman et al12 trial had the freedom to insert needles in up to 16 non-local or distal acupoints (see eTable 2 in the supplement), including the vertex of the head, ear, foot, ankle, elbow, cervicothoracic junction and/or thoracolumbar junction.1 Given that 59% of the clinicians in Hinman et al12 trial practiced a purely traditional Chinese style of acupuncture and 28% utilized a combination approach of Western and traditional Chinese medicine, we would be interested to know what proportion of the acupoints needled were not within the vicinity of the knee over the 8-12 week intervention period—this alone could possibly explain the lack of a between-group effect.

While some have suggested an effect size ceiling of 4 needles,4 a recent meta-analysis of 17,922 patients found that an increased number of needles was associated with better outcomes in patients suffering from osteoarthritis.30 In fact, of the 29 acupuncture trials used to treat chronic pain, only 4% and 25% used 1-4 and 5-9 needles, respectively; whereas, 38% and 33% used 10-14 and 15-20 needles, respectively.30 Furthermore, this same meta-analysis found the average number of needles utilized across 8 acupuncture RCTs involving patients with knee OA was 11.30 Interestingly, the knee OA trial with the lowest effect size by Foster et al4 used the least number of needles and did not require needling of any points local to the knee itself; whereas, the trial reporting the largest between-group effect size by Witt et al7 averaged the most needles per treatment session and required a minimum of 6 needles placed at local knee acupoints. In contrast, Hinman et al12 used just 4 local acupoints on the first session and it remains unknown how many local acupoints in the vicinity of the knee were utilized for the remaining 7-11 sessions of the intervention period.

Longer treatment durations (time in which the needles are left in situ) seem to be associated with superior outcomes in acupuncture trials for patients with knee osteoarthritis.5,7,11 Hinman et al12 appears to avoid scrutiny by relying on comments made in a recent meta-analysis by MacPherson et al,30 in which the authors suggested that longer treatment durations have a smaller effect than sham controls. Nevertheless, in this same meta-analysis,30 56% of the trials evaluated used acupuncture treatment durations of at least 30 minutes and only 4% of the studies reported treatment durations of 15-19 minutes. More specifically, the average treatment duration used by acupuncture trials for the treatment knee OA in the MacPherson et al30 meta-analysis was 26.5 minutes. It is remains unclear as to whether the duration or “average length of a session” as described by Hinman et al12 represents the total treatment session (including positioning, set up etc.) or the duration the needles were actually in situ. Interestingly, a 2014 literature review found that needles inserted to treat a variety of neuromuscular conditions including low back pain, neck pain, carpal tunnel, plantar fasciitis, knee OA, shoulder pain and headaches were left in situ between 5 and 40 minutes.36 Given that the literature for knee OA overwhelming advocates leaving needles in situ for 205,6,9,10,19 to 30 minutes,7,11 it seems quite likely that the acupuncture dosage employed by Hinman et al12 was suboptimal.

Lastly and of major importance, unlike 11 of the 13 randomized controlled trials in the Cochrane Database systematic review1 that found acupuncture effective in patients with knee osteoarthritis, Hinman et al12 did not radiographically confirm the clinical diagnosis of knee osteoarthritis.37 Notably, Foster et al4 was one of the two studies that did not require radiologic verification of knee OA, and like Hinman et al,12 this study reported no between-group difference. Rather than a homogeneous group of patients with knee osteoarthritis, it seems possible that a heterogeneous group of patients with knee pain of mixed etiologies could have made up the samples in both the Foster et al4 and Hinman et al12 trials.

CO-AUTHORS:

James Dunning, DPT, MSc, MMACP (UK), MAACP (UK), FAAOMPT
Director, AAMT Fellowship in Orthopaedic Manual Physical Therapy
President, Alabama Physical Therapy & Acupuncture, Montgomery, AL

Raymond Butts, PhD, DPT, MSc (NeuroSci), Cert. SMT, Cert. DN
Senior Faculty, AAMT Fellowship in Orthopaedic Manual Physical Therapy
Senior Physical Therapist, Research Physical Therapy Specialists, Columbia, SC

Thomas Perreault, DPT, OCS, Cert. SMT, Cert. DN
Senior Faculty, AAMT Fellowship in Orthopaedic Manual Physical Therapy
Senior Physical Therapist, Portsmouth Physical Therapy, Portsmouth, NH

Matthew Grubb, DPT, MS, Cert. SMT, Cert. DN, ATC, CSCS
Director of Physical Therapy, Putnam Physical Therapy, Cookeville, TN
Fellow-in-Training, AAMT Fellowship in Orthopaedic Manual Physical Therapy

Rachel Grubb, DPT, Cert. SMT, Cert. DN
Senior Physical Therapist, Putnam Physical Therapy, Cookeville, TN
Fellow-in-Training, AAMT Fellowship in Orthopaedic Manual Physical Therapy

REFERENCES

  1. Manheimer E, Cheng K, Linde K, et al. Acupuncture for peripheral joint osteoarthritis. The Cochrane database of systematic reviews. 2010(1):CD001977.
  2. Vickers AJ, Cronin AM, Maschino AC, et al. Acupuncture for chronic pain: individual patient data meta-analysis. Arch Intern Med. Oct 22 2012;172(19):1444-1453.
  3. Corbett MS, Rice SJ, Madurasinghe V, et al. Acupuncture and other physical treatments for the relief of pain due to osteoarthritis of the knee: network meta-analysis. Osteoarthritis Cartilage. Sep 2013;21(9):1290-1298.
  4. Foster NE, Thomas E, Barlas P, et al. Acupuncture as an adjunct to exercise based physiotherapy for osteoarthritis of the knee: randomised controlled trial. Bmj. Sep 1 2007;335(7617):436.
  5. Berman BM, Lao L, Langenberg P, Lee WL, Gilpin AM, Hochberg MC. Effectiveness of acupuncture as adjunctive therapy in osteoarthritis of the knee: a randomized, controlled trial. Ann Intern Med. Dec 21 2004;141(12):901-910.
  6. Scharf HP, Mansmann U, Streitberger K, et al. Acupuncture and knee osteoarthritis: a three-armed randomized trial. Ann Intern Med. Jul 4 2006;145(1):12-20.
  7. Witt C, Brinkhaus B, Jena S, et al. Acupuncture in patients with osteoarthritis of the knee: a randomised trial. Lancet. Jul 9-15 2005;366(9480):136-143.
  8. Witt CM, Jena S, Brinkhaus B, Liecker B, Wegscheider K, Willich SN. Acupuncture in patients with osteoarthritis of the knee or hip: a randomized, controlled trial with an additional nonrandomized arm. Arthritis Rheum. Nov 2006;54(11):3485-3493.
  9. Berman BM, Singh BB, Lao L, et al. A randomized trial of acupuncture as an adjunctive therapy in osteoarthritis of the knee. Rheumatology (Oxford). Apr 1999;38(4):346-354.
  10. Mavrommatis CI, Argyra E, Vadalouka A, Vasilakos DG. Acupuncture as an adjunctive therapy to pharmacological treatment in patients with chronic pain due to osteoarthritis of the knee: a 3-armed, randomized, placebo-controlled trial. Pain. Aug 2012;153(8):1720-1726.
  11. Vas J, Mendez C, Perea-Milla E, et al. Acupuncture as a complementary therapy to the pharmacological treatment of osteoarthritis of the knee: randomised controlled trial. Bmj. Nov 20 2004;329(7476):1216.
  12. Hinman RS, McCrory P, Pirotta M, et al. Acupuncture for chronic knee pain: a randomized clinical trial. Jama. Oct 1 2014;312(13):1313-1322.
  13. Whitehurst DG, Bryan S, Hay EM, Thomas E, Young J, Foster NE. Cost-effectiveness of acupuncture care as an adjunct to exercise-based physical therapy for osteoarthritis of the knee. Phys Ther. May 2011;91(5):630-641.
  14. MacPherson H, White A, Cummings M, Jobst K, Rose K, Niemtzow R. Standards for reporting interventions in controlled trials of acupuncture: The STRICTA recommendations. STandards for Reporting Interventions in Controlled Trails of Acupuncture. Acupunct Med. Mar 2002;20(1):22-25.
  15. Chen YL, Yang KH. Consort 2010. Lancet. Jul 24 2010;376(9737):230.
  16. Glasziou P, Meats E, Heneghan C, Shepperd S. What is missing from descriptions of treatment in trials and reviews? Bmj. 2008;336(7659):1472-1474.
  17. Glasziou P, Chalmers I, Altman DG, et al. Taking healthcare interventions from trial to practice. Bmj. 2010;341:c3852.
  18. MacPherson H, White A, Cummings M, Jobst KA, Rose K, Niemtzow RC. Standards for Reporting Interventions in Controlled Trials of Acupuncture: the STRICTA recommendations. J Altern Complement Med. Feb 2002;8(1):85-89.
  19. White A, Foster NE, Cummings M, Barlas P. Acupuncture treatment for chronic knee pain: a systematic review. Rheumatology (Oxford). Mar 2007;46(3):384-390.
  20. Bovey M. Deqi. Journal of chinese medicine. 2006;81:18-29.
  21. Hui KK, Nixon EE, Vangel MG, et al. Characterization of the “deqi” response in acupuncture. BMC Complement Altern Med. 2007;7:33.
  22. Benham A, Johnson MI. Could acupuncture needle sensation be a predictor of analgesic response? Acupunct Med. Jun 2009;27(2):65-67.
  23. Choi YJ, Lee JE, Moon WK, Cho SH. Does the effect of acupuncture depend on needling sensation and manipulation? Complement Ther Med. Jun 2013;21(3):207-214.
  24. Langevin HM, Bouffard NA, Badger GJ, Churchill DL, Howe AK. Subcutaneous tissue fibroblast cytoskeletal remodeling induced by acupuncture: evidence for a mechanotransduction-based mechanism. J Cell Physiol. Jun 2006;207(3):767-774.
  25. Langevin HM, Bouffard NA, Badger GJ, Iatridis JC, Howe AK. Dynamic fibroblast cytoskeletal response to subcutaneous tissue stretch ex vivo and in vivo. Am J Physiol Cell Physiol. Mar 2005;288(3):C747-756.
  26. Langevin HM, Bouffard NA, Churchill DL, Badger GJ. Connective tissue fibroblast response to acupuncture: dose-dependent effect of bidirectional needle rotation. J Altern Complement Med. Apr 2007;13(3):355-360.
  27. Langevin HM, Churchill DL, Cipolla MJ. Mechanical signaling through connective tissue: a mechanism for the therapeutic effect of acupuncture. Faseb J. Oct 2001;15(12):2275-2282.
  28. Loaiza LA, Yamaguchi S, Ito M, Ohshima N. Electro-acupuncture stimulation to muscle afferents in anesthetized rats modulates the blood flow to the knee joint through autonomic reflexes and nitric oxide. Auton Neurosci. May 31 2002;97(2):103-109.
  29. Wu MT, Sheen JM, Chuang KH, et al. Neuronal specificity of acupuncture response: a fMRI study with electroacupuncture. Neuroimage. Aug 2002;16(4):1028-1037.
  30. MacPherson H, Maschino AC, Lewith G, Foster NE, Witt CM, Vickers AJ. Characteristics of acupuncture treatment associated with outcome: an individual patient meta-analysis of 17,922 patients with chronic pain in randomised controlled trials. PLoS One. 2013;8(10):e77438.
  31. Ulett GA, Han S, Han JS. Electroacupuncture: mechanisms and clinical application. Biol Psychiatry. Jul 15 1998;44(2):129-138.
  32. Mayor D. An exploratory review of the electroacupuncture literature: clinical applications and endorphin mechanisms. Acupunct Med. Dec 2013;31(4):409-415.
  33. Ahsin S, Saleem S, Bhatti AM, Iles RK, Aslam M. Clinical and endocrinological changes after electro-acupuncture treatment in patients with osteoarthritis of the knee. Pain. Dec 15 2009;147(1-3):60-66.
  34. Napadow V, Makris N, Liu J, Kettner NW, Kwong KK, Hui KK. Effects of electroacupuncture versus manual acupuncture on the human brain as measured by fMRI. Hum Brain Mapp. Mar 2005;24(3):193-205.
  35. Sangdee C, Teekachunhatean S, Sananpanich K, et al. Electroacupuncture versus diclofenac in symptomatic treatment of osteoarthritis of the knee: a randomized controlled trial. BMC Complement Altern Med. Mar 21 2002;2:3.
  36. Dunning J, Butts R, Mourad F, Young I, Flannagan S, Perreault T. Dry needling: a literature review with implications for clinical practice guidelines. Phys Ther Rev. Aug 2014;19(4):252-265.
  37. Altman R, Asch E, Bloch D, et al. Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association. Arthritis Rheum. Aug 1986;29(8):1039-1049.