Comparing Group and Individual Acupuncture Therapy for Treating Chronic Pain among Ethnically Diverse Patients with Low Incomes—AADDOPT-2

Comparing Group and Individual Acupuncture Therapy for Treating Chronic Pain among Ethnically Diverse Patients with Low Incomes—AADDOPT-2 - NCBI Bookshelf An official website of the United States government Here's how you know The .gov means it's official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site. The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely. Log in Show account info Close Account Logged in as: username Dashboard Publications Account settings Log out Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation Bookshelf Search database Books All Databases Assembly Biocollections BioProject BioSample Books ClinVar Conserved Domains dbVar Gene Genome GEO DataSets GEO Profiles GTR Identical Protein Groups MedGen MeSH NLM Catalog Nucleotide OMIM PMC Protein Protein Clusters Protein Family Models PubChem BioAssay PubChem Compound PubChem Substance PubMed SNP SRA Structure Taxonomy ToolKit ToolKitAll ToolKitBookgh Search term Search Browse Titles Advanced Help Disclaimer NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health. Comparing Group and Individual Acupuncture Therapy for Treating Chronic Pain among Ethnically Diverse Patients with Low Incomes—AADDOPT-2 M. Diane McKee , MD, MS, Benjamin Kligler , MD, MPH, Belinda Anderson , PhD, Elizabeth Chuang , MD, MPH, Mariel Connolly , BS, Qi Gao , PhD, Eric N. Gil , BA, Claudia Lechuga , MS, Mimi Kim , PhD, and Arya Nielsen , PhD. Author Information and Affiliations Authors M. Diane McKee , MD, MS, 1 Benjamin Kligler , MD, MPH, 1,2 Belinda Anderson , PhD, 1,3 Elizabeth Chuang , MD, MPH, 1 Mariel Connolly , BS, 1 Qi Gao , PhD, 1 Eric N. Gil , BA, 1 Claudia Lechuga , MS, 1 Mimi Kim , PhD, 1 and Arya Nielsen , PhD 4 . Affiliations 1 Department of Family and Social Medicine, Albert Einstein College of Medicine, New York, New York 2 Integrative Health Coordinating Center, US Veterans Health Administration, Washington, District of Columbia 3 Pacific College of Oriental Medicine, New York, New York 4 Department of Family Medicine and Community Health, Icahn School of Medicine at Mount Sinai, New York, New York Washington (DC): Patient-Centered Outcomes Research Institute (PCORI) ; 2020 Jun . Copyright and Permissions Copyright © 2020. Albert Einstein College of Medicine. All Rights Reserved. This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/ Structured Abstract Background: Chronic pain is common and often undertreated. Access to effective, nonpharmacologic therapies is limited, especially in low-income and ethnically diverse communities. Acupuncture therapy is an effective treatment option. Acupuncture therapy delivered in a group setting adapts individual session care to treat multiple individuals seated together in a room. Group delivery could reduce cost and expand access to low-income patients, but its effectiveness compared with individual sessions has not been established. Objectives: Using a noninferiority design, our specific aims were to (1) compare the effectiveness of group vs individual acupuncture to reduce pain and improve function among ethnically diverse, low-income primary care patients with chronic pain; and (2) use qualitative methods to understand and incorporate the patient experience of acupuncture to maximize the patient centeredness of the intervention and dissemination efforts. Methods: This was a randomized comparative effectiveness trial conducted in 6 primary care practices in the Bronx, New York, New York. Group acupuncture was delivered in multipurpose rooms, with patients seated in chairs. Patients with chronic (≥3 months) back, neck, or osteoarthritis pain were randomly assigned (N = 779) to 12 weekly sessions of acupuncture therapy; 706 patients initiated treatment via group (n = 346) or individual (n = 360) sessions. The primary outcome was pain interference with normal activities (“pain interference”), as measured by the Brief Pain Inventory (BPI) at 12 weeks. Secondary outcomes were the pain severity scale of the BPI; physical and mental well-being, as measured by the 10-item Patient-Reported Outcomes Measurement Information Systems (PROMIS-10) global health scale; and the Patient Global Impression of Change (PGIC). The intention-to-treat (ITT) population included all randomly assigned participants (N = 779); the per-protocol (PP) population included those who received ≥8 sessions of acupuncture during the 12-week intervention period (n = 450). The margin of noninferiority was defined as an absolute difference of δ = 10% (individual - group) in the proportion of patients demonstrating >30% improvement in pain interference between baseline and week 12. Results: Study participants were primarily Medicaid insured (75%) and reported poor/fair health (60%); over a third (37%) were disabled. The mean number of treatments was 8.0 for the group setting and 8.1 for the individual setting ( P = .56). Rates of loss to follow-up (and therefore, no outcome data collected) were similar in the 2 arms at 12 weeks (individual, 12.3%; group, 12.8%). For the primary outcome of pain interference, in analysis of the ITT population with available data, 37.5% of individual treatment arm (n = 297) and 30.3% in group treatment (n = 279) participants had ≥30% improvement (individual - group difference [ d ] = 7.2%; 95% CI, −0.6% to 15.1%). In the PP population with available data, ≥30% improvement was achieved for 39.7% of patients in the individual treatment arm (n = 209) vs 34.4% of patients in the group treatment arm (n = 194) ( d = 5.3%; 95% CI, −4.2% to 1.49%). For the secondary outcome of pain severity, in the ITT analysis, 34.8% of individual (n = 301) and 30.5% of group (n = 285) participants achieved ≥30% pain reduction ( d = 4.3%; 95% CI, −3.3% to 11.9%). In the PP analysis, >30% improvement was achieved in 39.2% of individual treatment participants (n = 213) compared with 36.3% of group treatment participants (n = 201; d = 2.8%; 95% CI, −6.5% to 12.2%). Similar results were observed when all ITT (N = 779) and PP (N = 450) participants were included in the analysis after applying multiple imputation to address missing data. ITT analysis of physical function scores showed mean increases from baseline of 3.6 in both arms ( d = 0.08; 95% CI, −0.97 to 1.13); mean changes in mental health scores were 1.5 in the individual therapy arm and 1.1 in the group therapy arm ( d = 0.48; 95% CI, −0.65 to 1.61). PGIC improvement rates were 37.5% in individual therapy and 35.0% in group therapy ( d = 2.5%; 95% CI, −5.2% to 10.2%). Noninferiority of group acupuncture was not demonstrated for the primary or secondary outcomes with a noninferiority margin of 10% because the 95% CI for the differences exceeded the noninferiority margin. Conclusions: Noninferiority of group acupuncture therapy compared with individual acupuncture therapy was not shown for pain interference, pain severity, physical or mental health function, or patient global impression of change. Limitations: The most significant limitations were (1) the lack of a usual care arm, and (2) the inability to blind outcome assessors to the study arm. Both were related to budget constraints. Additionally, the suboptimal physical setting, particularly for group acupuncture delivery, could have biased results away from noninferiority of group over individual treatment. Background Prevalence and Impact of Chronic Pain and Disparities in Health Outcomes The classification of pain continues to evolve based on expert opinion and the dimensional demands of ICD coding revisions. 1 A new and pragmatic classification of chronic pain developed for the upcoming 11th revision of the ICD defines “chronic primary pain” as pain in ≥1 anatomic regions that persists or recurs for longer than 3 months and is associated with significant emotional distress or significant functional disability (ie, interference with activities of daily life and participation in social roles) and that cannot be better explained by another chronic pain condition. 2 This new phenomenological definition, which was created because the etiology is unknown for many forms of chronic pain, recognizes aspects of chronic pain characterized by more than time frame alone. “Chronic musculoskeletal pain” is defined as persistent or recurrent pain that is nociceptive, that is, related to a disease process directly affecting bone(s), joint(s), muscle(s), or related soft tissue(s). 2 The prevalence of chronic pain conditions in the general adult US population is estimated to range from 11%-47% in large surveys. 3-10 Low back and neck pain, osteoarthritis (OA), and headache are the most common pain conditions in the United States and are leading global causes of disability in most countries. 11 Pain and pain care impact specific socioeconomic and demographic groups differently. 12 Risk factors for increased vulnerability to pain include having English as a second language; belonging to a racial or ethnic minority; having lower income, education, and health literacy; being female; being in an older age group; and geographic location. 7 Pain treatment for many of these groups is also limited in access and scope. 7 , 13 , 14 Although back pain is common across all primary care populations, low-income individuals and individuals belonging to racial and ethnic minority groups are impacted differently. For example, compared with white patients, African Americans have a lower likelihood of receiving comprehensive pain assessment and management. 13 Hispanics are also at higher risk for pain and pain undertreatment. These disadvantages are further compounded when limited English proficiency impacts communication with health care providers. 7 In every ethnic/racial category, women are more likely than men to report a wide range of chronic pain conditions. 7 , 9 , 15 , 16 Pain prevalence varies for women by age and race/ethnicity, 17 and women also experience disparities in pain care. 7 , 18 Living with chronic pain is associated with impairment of physical and psychological functioning 19-21 , lost productivity 22 , and lower socioeconomic status. 8 Chronic pain is often associated with multimorbidity or higher comorbidity load. 23 Individuals with pain have higher rates of functional limitations than do individuals without pain, and they develop functional limitations classically associated with aging at much earlier ages. 24 Functional limitations are directly associated with higher Medicare and Medicaid health care resource use and expenditures. 25 , 26 Effectiveness of Acupuncture Therapy in Treatment of Chronic Pain Acupuncture therapy has been shown to be effective in the treatment of chronic pain conditions, 27-30 including shoulder pain, 31 , 32 chronic low back pain, 33-36 neck pain, 36-38 knee pain from OA, 39-44 headache, 45 , 46 and temporomandibular disorders. 47 In an individual patient meta-analysis of 18 000 patients with chronic pain, including low back, neck, shoulder, OA of the knee, and headache/migraine, acupuncture was shown to be significantly better than sham treatment or usual care, 30 with persistence of 50% to 90% of benefit at 12 months after a course of treatment. 48 An update to that meta-analysis including 39 trials and 20 837 patients found that the effects of acupuncture persist over time, with only a 15% reduction in treatment effect at 1 year. 49 Acupuncture is recommended as a first-line option for the treatment of acute, subacute, and chronic low back pain by the American College of Physicians (ACP) in their 2017 Clinical Practice Guidelines 50 and by the National Institutes of Health (NIH) for low back pain and knee OA. 51 It is also among the primary treatment options recommended for patients with chronic low back and neck pain without serious pathology in the Global Spine Care Initiative Guidelines. 52 Preliminary Effectiveness of Acupuncture Delivered in Community Health Centers Our previous National Center for Complementary and Integrative Health-funded acupuncture trial, Acupuncture to Decrease Disparities in Outcomes of Pain Treatment (ADDOPT), demonstrated the feasibility and effectiveness of acupuncture treatment for chronic pain in an ethnically diverse and medically underserved population at high risk for health disparities. 53 , 54 This pragmatic quasi-experimental study was conducted in urban primary care health centers. Primary care providers (PCPs) referred patients with chronic pain due to OA, neck, or back pain to on-site acupuncture provided in weekly sessions. The feasibility of integrating acupuncture therapy in the urban primary care setting was supported by (1) success in referral (495 patients were referred over 2 years, with a large percentage of PCPs at each practice making referrals); (2) recruitment (47% initiated treatment; the most common reason for not initiating was schedule conflict/wait list); and (3) retention in care, as 71% participated in ≥5 treatments (mean [SD] overall, 8.0 [4.7] treatments). 53 In ADDOPT, using a pre-post design assessing within-group changes (without a control), we compared primary outcomes of pain, pain-free days, and quality of life (QOL) for patients during a pre-acupuncture phase (during which patients received usual care only) to the period after acupuncture was provided. Repeated-measures analyses of variance indicated that mean pain severity (as measured by the Brief Pain Inventory [BPI]) and physical health scores (12-Item Short Form Health Survey [SF-12]) changed significantly over time during the course of the study. Pain severity scores improved from baseline (6.8 at baseline vs 5.7 at 12 weeks and 5.8 at 24 weeks, respectively) as did physical well-being (31.8 at baseline vs 35.4 at 12 weeks and 35.2 at 24 weeks, respectively). Using hierarchical linear modeling methods, the reduction in pain severity between baseline and treatment phase was significant ( P < .001). Improvements in physical well-being were significant at 12 and 24 weeks after baseline ( P < .001). Almost one-third (32.4%) experienced a ≥30% improvement in pain. 54 ADDOPT demonstrated that acupuncture therapy is feasible and effective, specifically in an underserved and diverse population at risk for health outcome disparities. 53 , 54 However, high cost and lack of access to individual acupuncture treatment continue to pose barriers to widespread implementation in this patient population. 55 To address these issues, acupuncturists are offering group acupuncture in varied settings, including for patients with chronic pain. 56-59 Local group-based community acupuncture clinics reach individuals along a broad socioeconomic spectrum, facilitating access to treatment and reducing costs and disparities in care. 60-62 Preliminary Evidence of Effectiveness of Group Acupuncture Group acupuncture therapy is defined here as care given in settings where patients are treated simultaneously, in a staggered fashion, with participants situated near and in view of one another. The terms “community acupuncture” and “group acupuncture” are sometimes used interchangeably; however, a community acupuncture setting might also describe a series of treatment tables behind screens where patients are not situated in or as a group. In the group approach, 6-8 patients are treated simultaneously, in a staggered fashion, by a single acupuncturist over a 60- to 90-minute period. Patients are typically seated in a large room in comfortable chairs while receiving treatment; the acupuncturist conducts an interview with palpation, takes a history, and then treats each patient in turn, leaving the needles in place while moving on to treat the next person. Needles typically remain in place for 20 to 30 minutes, as they would in an individual session, and participants have the potential added benefit of social interaction during the treatment session. Controversy exists in the acupuncture and medical communities as to whether group acupuncture is as effective as individual treatment. 63 , 64 In group acupuncture, because patients are typically seated rather than lying down and they remain clothed, there is a greater emphasis on distal acupuncture points on the arms, legs, head, and neck; some acupuncturists feel this could decrease the effectiveness of the treatment. Because acupuncture therapy typically involves treatment of local and distal points together, the question of decreased effectiveness with reduced access to local points remains unresolved. Nonrandomized studies 57 , 59 , 65 , 66 as well as randomized studies 58 , 67 , 68 demonstrate that group acupuncture is accepted by patients. Our previous studies have shown the feasibility of providing acupuncture in a safety net care setting. 53 , 54 Our pilot nonrandomized trial of group acupuncture therapy for chronic pain in patients representing a cross-section of the New York City population, including but not limited to underserved patients, showed feasibility with clinically significant improvement in chronic pain and depression that persisted through the 24-week measure after the completion of an 8-week course of treatment. 67 Evidence Gap Acupuncture therapy has been predominantly studied as care delivered in an individual setting, where it is effective for chronic pain; 27 it has been shown to be effective and feasible for low-income, ethnically diverse patients with chronic pain when delivered in community health center (CHC) settings. 53 , 68 , 69 Cost (because most insurance plans do not cover acupuncture) and access to individual acupuncture treatment pose major barriers to widespread implementation in this patient population. To reduce cost, increase access, and meet patient demand, group acupuncture therapy is now being offered in settings across the United States, including for underserved and diverse populations at risk for health outcome disparities. However, studies have not been conducted to establish whether or not acupuncture delivered in the group setting is as effective as individual acupuncture for chronic pain, particularly for underserved chronic pain patients. Study Goal and Aims The Acupuncture Approaches to Decrease Disparities in Outcomes of Pain Treatment 2-Arm Comparative Effectiveness Trial (AADDOPT-2) sought to close this evidence gap, specifically to answer the important question of whether acupuncture for chronic pain delivered in a group setting is as effective as individual acupuncture in an underserved and ethnically diverse patient population. Our specific aims were to (1) compare the effectiveness of group vs individual acupuncture to reduce pain and improve function among ethnically diverse, low-income primary care patients with chronic pain; and (2) use qualitative methods to understand and incorporate the patient experience of acupuncture to maximize the patient centeredness of the intervention and dissemination efforts. Patient and Stakeholder Engagement Our stakeholder panel was composed of 3 patient partners from the CHCs in which the study was carried out (patients); a PCP from 1 of the CHCs (clinician); a pain psychologist (clinician); a national-level patient advocate for patients with pain (patients); the owner of a group/community acupuncture practice (clinician); a representative from the Montefiore Accountable Care Organization (ACO) (payer); and representatives from 2 national organizations representing pain practitioners from both conventional and complementary health professions (clinicians/policy makers). We felt that patients and patient advocates should represent the largest contingent and recruited these members through PCPs in the ambulatory care network. Local clinician and payer stakeholders were recruited through previous connection with our practice-based research network, and national-level participants were identified through our previous work in education and research on integrative approaches to pain management. We took a multipronged approach to stakeholder engagement, which also evolved during the course of the study. An initial stakeholder retreat was held before the study launch. In this session, our diverse group of stakeholders provided valuable input on issues such as the logistics of delivering acupuncture in group settings and anticipated challenges particularly related to the physical space we would be using. This face-to-face retreat created a sense of community between the investigators, acupuncturists, and stakeholders. Following this retreat, we invited all investigators, patient partners, and treating acupuncturists to attend our monthly team meetings, which provided an important opportunity to identify implementation issues and strategize solutions, while keeping the participant voice. Our patient partners were invited to receive acupuncture both in the group and individual settings early on after study launch as participant observers, and they provided invaluable feedback on the experience, particularly the group dimension. To demonstrate the value we placed on their participation and to promote ongoing engagement, we paid patient partners as consultants on a monthly basis during the study. The entire stakeholder group met quarterly by phone, with a rotating focus on implementation, dissemination, participant experience, and clinician perspective. We also found that these quarterly meetings of the entire national stakeholder group and the treating acupuncturists created an opportunity for the study acupuncturists to gain perspective about how the study fit into larger national efforts to improve pain outcomes and make nonpharmacologic options available. National stakeholders were paid on a quarterly basis. Finally, to be sure we were continuously including input from 1 of our most important stakeholder groups, our research acupuncturists, we implemented a weekly meeting of the treating acupuncturists that contributed to team building and added value in their ongoing development as clinicians working with a particularly challenging population with a high level of disability and comorbid illness. Discussion focused on issues identified by the treating acupuncturists and reported in weekly surveys, including challenges in the care setting, difficult patient presentations, and the use of the intervention manual. This group cohesion was very important for retention of treating acupuncturists and for the ongoing development of the manualized acupuncture procedure. In terms of the impact of stakeholder engagement on our study, the areas of most meaningful impact were in study design, processes, and outcomes; participant recruitment; and study rigor and quality. Many stakeholders have also specifically committed to helping with the dissemination of our final findings to promote the adoption of research evidence into practice. Some examples of stakeholder impact are provided below: Patient partners encouraged standardization of the patient experience in both arms by providing a consistent selection of music to establish ambience. They identified physical space issues that were uncomfortable for patients, as well as potential privacy issues, particularly in group settings. For example, they made specific suggestions regarding placement of chairs and furniture in the group treatment rooms and how to help participants be as comfortable as possible in mixed-gender groups. Patient partner contributions helped increase our sensitivity to the patient experience that allowed us to anticipate issues of social comfort and privacy in a group setting and likely contributed to participant retention in treatment. Patient partners provided feedback on all patient-facing materials, such as informational materials and consent documents. This improved patient understanding of the study and likely contributed to a sound, ethical informed consent process. In fact, the patient partners contributed early on to a name change for our trial in our promotional materials, as they felt our original name was too long and would be off-putting to participants; the name was changed from the unwieldy “Acupuncture Approaches to Decrease Disparities in Outcomes of Pain Treatment” to “Acupuncture for Chronic Pain.” Patient partners assisted as part of the research team in the development of the qualitative interview questions, ensuring that the questions were easily understood and tapped constructs that were salient for patients. A national clinician stakeholder who is a pain psychologist led a training session with the acupuncturists to build skills in interacting with primary care patients with challenging pain and mental health comorbidities in the group setting. Our primary care clinician stakeholder (a physician at 1 of the CHCs) helped ensure that our procedures were acceptable to the CHC staff and clinicians, minimizing burden while providing a highly valued new treatment option. At a critical juncture in the study, she helped us realize that more communication was needed between the study team and the referring physicians in order to streamline the process for study participants. This engagement helped us develop a workflow to communicate through the electronic medical record (EMR) in order for the physicians to be up to date on the status of their patients. This improved the patient experience and ensured that clinicians were informed of the status of their patients before and during treatment; thus, clinicians remained committed to referring patients. Regarding the impact of our payer representative from the Montefiore ACO, we have not yet been successful in developing a payment model that makes group acupuncture therapy sustainable for our health system. Although our CHCs were all very interested in continuing to offer acupuncture, our academic medical center is not prepared in the current fiscal environment to offer the treatment unless there is insurance reimbursement. Our payer stakeholder remains an influential advocate whom we hope in the future can help us incorporate acupuncture as a routine part of care for chronic pain in our system. Her involvement also helped support another important institutional development: Before the study, no licensed acupuncturist had been credentialed at our institution. We worked with senior leadership of the institution to develop a mechanism for credentialing acupuncturists. Based on this effort, a significant barrier to hiring licensed acupuncturists either for future studies or for routine clinical duties has now been removed. All stakeholders contributed to the creation of our dissemination plan, identifying relevant communities in pain management advocacy and acupuncture and a means to communicate study results to them outside of academic channels. For example, a stakeholder who is the executive director of a large national pain-oriented professional organization assisted in publishing an article on our study in progress in the organization's online newsletter. As another example, 1 of our patient partners attended a city-wide research symposium with the research team to participate in our poster presentation and share his perspective with the researchers in attendance. In summary, our stakeholders as a group had an enormous impact on this study. We believe that the patient partners' input as participant observers, early in the study, was especially important and resulted in a more patient-centered experience for participants. We did experience challenges with developing a way to pay patient partners. Our institution requires signed agreements for all individuals who serve as consultants. These agreements were designed primarily for academic consultants and included language that was difficult to understand and not entirely applicable to the role of a patient partner. We worked with the Albert Einstein College of Medicine to develop a more appropriate document that was still acceptable to the institution. We also realized that for some patient partners, the requirement to complete human participant training was too difficult. We learned that there were still many ways in which they could contribute and that such a requirement may not be worth the burden in future studies. These challenges have put us in a stronger position as we move forward to include patient stakeholders in future clinical research. Methods Study Overview AADDOPT-2 evaluated acupuncture for the treatment of chronic pain delivered either in individual or group visits in urban primary care health centers. Participants were eligible if they reported experiencing chronic pain (≥3 months) due to OA of any joint or to neck or back pain and were referred by their physician at a participating health center. Eligible patients were randomly assigned to individual or group treatment ( Figure 1 ). After randomization, participants were treated in weekly sessions for 12 weeks. The primary patient-centered outcome of pain interference and secondary outcomes of pain severity and health-related well-being were assessed at 12 weeks (primary end point) and 24 weeks (postintervention follow-up end point). Acupuncture therapy was given only in the 12-week treatment phase. Group and individual acupuncture sessions were of similar duration and were delivered by the same team of licensed acupuncturists. For aim 1, quantitative analyses were conducted for within-group and between-group comparisons of pain and function outcomes. In aim 2, to better understand participants' experience of the intervention, a subgroup of participants (n = 46) completed a semistructured interview after their 24-week assessment. We used qualitative analysis to identify themes related to the implementation of group and individual acupuncture in CHCs. The findings were used to improve ongoing intervention delivery in this trial and were disseminated to inform efforts to offer acupuncture therapy in similar settings. Figure 1 Study Overview. Study Design The AADDOPT-2 study was a randomized, nonblinded comparative effectiveness trial evaluating acupuncture for the treatment of chronic pain delivered in either individual or group visits. We chose to compare group and individual delivery methods of acupuncture therapy because no previous study had established whether group delivery is noninferior to individual delivery. We chose to conduct the study in primary care, and to continue usual care as coordinated by the PCPs, in order to reflect real-world patterns of care delivery, including heterogeneity of the use of medicines and approaches for pain. Acupuncture was provided on-site to minimize barriers to participation. We chose random assignment to the 2 arms to minimize bias related to self-selection. Participants could not be blinded due to the nature of the interventions. Study Setting Participating primary care practices are located in the Bronx, New York, a county composed of a large proportion (85.5%) of ethnic minority residents, of whom more than half (56.7%) are Hispanic. Nearly a third of the population lives below the poverty level. Table 1 describes the location, size, and demographics of the participating sites, all part of the ambulatory care network of Montefiore Medical Center, a large not-for-profit integrated delivery system. Montefiore provides comprehensive primary care to patients with a variety of insurance coverage; federally qualified health center (FQHC) sites see patients regardless of insurance status. We chose to work in practices that serve this diverse population because of the disproportionate burden of chronic pain and limited access to acupuncture therapy and other nonpharmacologic approaches for pain management. Table 1 Characteristics of Participating Practices. Participants PCPs at 6 participating health centers identified patients with eligible chronic pain conditions and, if they expressed interest, referred them to the study. Only patients who received primary care at 1 of the participating sites were eligible. Additional eligibility criteria included (1) a qualifying diagnosis of chronic pain (≥3 months) due to OA of any joint or to chronic neck or back pain related to noncancer diagnoses; (2) fluency in English or Spanish; (3) ability to provide a home or cell phone number; and (4) intention to be available for up to 24 weeks for follow-up. Eligibility was confirmed through a combination of physician referral (which indicated the source of pain) and eligibility screening by the research team. Clinicians indicated the qualifying diagnosis on the referral. During the screening process, we confirmed that patients had pain for 3 months or longer. We minimized the exclusion criteria so that participants would reflect typical primary care patients. Exclusions were current anticoagulant use and inability to provide informed consent due to mental illness or cognitive impairment. Interventions and Comparators or Controls Choice of Comparators: Group and Individual Treatment Arms Although group acupuncture and group medical visits are a developing model of care, 61 , 66 , 70 , 71 no previous study has directly compared individual acupuncture therapy sessions with care delivered in a group setting. Acupuncture therapy has been shown to be effective for chronic pain and is recommended for it as part of comprehensive pain care by the Centers for Disease Control and Prevention (CDC), 72 NIH, 51 US Food and Drug Administration (FDA), 73 ACP, 50 and The Joint Commission. 74 Our trial tested a hypothesis of noninferiority if delivered in a group setting compared with typical individual session care. Usual care Participants in both arms continued to receive clinical services for management of chronic pain, as coordinated by PCPs. Usual care includes a medical diagnostic evaluation, analgesic drug therapies, recommendations for physical activity, and sometimes, referral to specialist physicians or physical therapy. Usual care at these sites did not include acupuncture therapy (outside the study). Individual treatment arm Individual acupuncture sessions were scheduled on the half hour, with the acupuncturist simultaneously working 2 rooms during a shift. A patient would be set up in 1 room, have the initial part of their treatment, and rest alone while the practitioner attended to a patient in a proximal second room. The acupuncturist would then return to the first room to finish the treatment and/or discharge the patient. This process is common to outpatient acupuncture settings. Initial sessions lasted for up to an hour; follow-up sessions typically were 45 minutes. Primary care exam rooms were used for the individual sessions, with the addition of bodyCushions ( www.bodysupport.com ) so patients could be prone on an examination table if needed. Group treatment arm Participants randomly assigned to the group arm received treatment in a setting with up to 6 patients in the group at any 1 time. Initial treatments were scheduled every 20 minutes and follow-up treatments every 15 minutes. This allowed for participants receiving their first treatment to have approximately 20 minutes (at the start of their session) of direct contact time with the acupuncturist, and 15 minutes (at the start of the session) of direct contact time in follow-up treatments. Following the 20- or 15-minute contact time, some needles might have been retained for a period of time, which extended the intervention time. The acupuncturist would return to remove needles, apply ear seeds (nonpenetrating spheres that are natural, nontoxic botanical Vaccaria seeds for extended auricular acupressure), answer questions, and so on. The acupuncturist was present throughout the entire treatment period, allowing them to adjust or add treatment as needed (eg, if a participant reported discomfort or needed clarification about a pain-related symptom). In this way, we aimed for comparable direct access to the acupuncturist in the group and individual treatment arms. However, a participant might have been late or remained longer to rest or relax, and so participants were in the “group therapy” space for differing times. The group intervention was distinct in that even when the acupuncturist moved to the next patient, all patients remained proximal, within sight and hearing of the practitioner. Participants were seated in chairs in a large room (conference or multiuse rooms, depending on the site). Where appropriate, patients could lean and rest forward on a table to allow access to the dorsal body. Documentation and communication Acupuncturists recorded the specifics of each treatment in individual participant charts that were not part of the medical record. The acupuncturist created a simplified brief note for each visit that was filed in the patient's EMR as a scanned note. Acupuncture Therapy To systematically characterize our acupuncture interventions, we followed the Standards for Reporting Interventions in Clinical Trials of Acupuncture (STRICTA), 75 an extension of the CONSORT statement. 76 STRICTA criteria include details of acupuncture rationale, needling, treatment regimen, and practitioner background. The comparators were discussed previously, and the remaining components of STRICTA are discussed below. 75 See our paper describing the study protocol for information about study tools. 77 Rationale Research on acupuncture therapy presents a unique challenge in terms of standardization of intervention dosage, frequency, and delivery, while allowing for the clinical responsivity of a traditional East Asian medicine (TEAM). We intentionally incorporated aspects of acupuncture treatment associated with TEAM and classical Chinese medicine to better reflect usual care in real-world settings 78 , 79 that is consistent with typical practice in the United States. The treatment manual developed for the trial included acupuncture needling but also included procedures that typically accompany acupuncture treatment in clinical practice. Ours was a stepped process that took into account the challenge of introducing acupuncture to patients who were new to acupuncture and to research participation. The process allowed for typical palpatory preparation of an area for acupuncture needling but also accommodated patients who were needle phobic or needle shy. An early session could therefore include a rigorous intervention using manual acupuncture therapy techniques but no needles, with the introduction of limited needling at the next session. Decisions on the deployment of the manual steps were made by the research acupuncturists based on how a patient presented on each day. Every participant experienced acupuncture needling at some point during their course of care. Needles were used at more than a single session in nearly all participants. Responsive manualization We used a pragmatic, real-world approach that allows for individualizing treatment from a consensus-built array of options achieved through the creation of a treatment manual. 80 The manual was developed through a modified Delphi process 77 , 81 , 82 involving a broad range of experts from diverse backgrounds and practice styles with experience in individual and group settings, as well as our team of treating acupuncturists. Training in the use of the manual for the research acupuncturists began during the manual development process, in which they were active participants. The research acupuncturists also participated in clinical reviews of safe practice, needling depths, infection control, and acupuncture techniques, including tui na and gua sha . They were given a study orientation that included a review of the manual. The research acupuncturists were also active in practice, including in group settings. Because the manual was built to be responsive, it included a common set of acupuncture points and an array of optional points and techniques allowing treatments to be responsive to the heterogeneous and evolving nature of an individual's condition. Our process for ensuring fidelity involved following the manual steps or stages of intervention, wherein deviations from adherence to the manual steps or order of steps was permitted but had to be recorded with the rationale for doing so. During the weekly acupuncture team meetings, manual usage, ease of application, and need for deviation were discussed. This allowed for monitoring of manual deviation, which was always based on individual patient circumstances. All treatment followed guidelines for safety and correct methodology. 83 A paper describing the responsive manual has been published. 77 Details of preparation and needling Acupuncture needling in classical Chinese medicine is grounded in and informed by an interview and history taking of a presenting problem which, in classical acupuncture practice, also incorporates palpation. In our study, the interview included questions about the nature and history of the participant's presenting chronic pain problem as well as standard TEAM questions regarding temperature, sleep, urine, stool, and activity. Palpation is a manual “asking and listening” technique that informs, while also serving as an intervention. Palpation evaluates tissue tension, texture changes, temperature, surface humidity, tenderness, and responsivity; it identifies where there is pain, where it propagates, and if pressure exacerbates or relieves it. 84 , 85 Palpation was used to distinguish 2 types of pain from the perspective of TEAM: qi stasis pain (responsive to touch and resolved by needling, massage, movement, or hot shower) and “blood stasis” pain (unresponsive-to-touch pain that is fixed, persistent, or recurring and not resolved by needling, massage, movement, etc). The latter is an indication for gua sha . 86 Palpation was also used to identify the exact locations and sensitivities of acupuncture points. Needling included manual manipulation to de qi status, a sensory component felt by the patient as a twitch or sensation along a channel and perceived by the practitioner as “needle grasp” with or without fasciculation. 87 Needles would then be left in place for a period of 5 to 30 minutes or, depending on the needle site and treatment plan, removed. Because, as discussed previously, we were providing treatment typical of real-world settings where needles might not be retained at all or might be retained for periods ranging from as little as 5 to as much as 30 minutes, we allowed flexibility with the treating acupuncturists in determining the duration of needling. In our intervention, typical retention times were 10 to 20 minutes; variations in needle retention times were determined by patient presentation, response to needling time in a previous session, or by clinically unrelated reasons, such as patient lateness. The number of points treated varied from none to ≥20, at the discretion of the acupuncturist. Treatments involving no needles were rare and usually occurred if the participant was apprehensive about receiving acupuncture during the first visit. All participants received acupuncture in the majority of their treatment sessions. As explained previously, some type of manual acupuncture therapy technique was used in every session, even on the rare occasion when no needles were used. The manual also allowed a response to patients presenting with acute conditions not directly related to their main complaint, such as anxiety, emotional upset, and respiratory infections, per the TEAM approach of prioritizing acute conditions. Patients were also given general TEAM lifestyle recommendations in terms of hot and cold diet and the importance of movement. Referral back to the patient's PCP was made for new symptoms or conditions and for recommendations of counseling and physical therapy. All treatment details were recorded in the chart record. Other components of treatment The acupuncture therapy manual also provided for the incorporation of therapies often used with acupuncture, including palpation, tui na, gua sha , and auricular treatment. These allowed for intervention options for participants who were new to acupuncture, needle phobic, or needle sensitive or who experienced enough pain and anxiety that staging a slow introduction to needling was required. Palpation was used at initiation and throughout treatment if needed, where it became a technique of discovery, somatic rapport, and treatment. Tui na is a classical Chinese manual therapy often used in conjunction with acupuncture, particularly for muscle and joint pain, 88 , 89 to stimulate, relax, warm, and irrigate qi in painful areas. Tui na can be integrated into palpation or be used preferentially for patients who are sensitive to needles and should be included when the patient feels cold in an area of the body or cold overall. Gua sha is a TEAM technique, 90 defined as unidirectional press-stroking of a lubricated area of skin with a smooth, round-edged instrument to intentionally raise transitory therapeutic petechiae and ecchymosis 86 producing an anti-inflammatory and immune protective response. 91 Gua sha is used for any condition that involves persistent or recurring pain. Ear treatment with acupuncture needles or ear seeds was included to reduce anxiety and provide a systemic or targeted analgesic effect. To extend the effect of treatment between sessions, ear seeds could be affixed to ear points and retained for several days, a technique which has been shown to be effective for anxiety 92 , 93 and pain. 94 , 95 Dosage: session number, frequency, timing, and duration Participants were scheduled to attend acupuncture sessions weekly for 12 weeks. There were 18 weekly individual or group session time blocks across 5 primary care sites. We designed the trial to approach parity between arms in terms of the time present with the acupuncturist. Practitioner background We deployed a team of 6 licensed acupuncturists, all of whom had at least 1 year of experience and who were licensed in the State of New York and board certified by the National Certification Commission for Acupuncture and Oriental Medicine. The acupuncturists participated in the consensus process for constructing the manual and clinical review sessions in tui na and gua sha , as well as safe practice, needling depths, and infection control. Study Outcomes All instruments used are established measures with good reliability; all have been validated for use in English- and Spanish-speaking populations. The BPI and 10-item Patient-Reported Outcomes Measurement Information Systems (PROMIS-10) measures have also been validated specifically for telephone administration. 96 , 97 Primary Outcome Measure The BPI 98 , 99 has been used extensively in pain studies, facilitating comparisons with other populations and interventions. Its subscales assess pain severity and pain interference with function (including activity, mood, sleep, work, and life enjoyment). This trial is not adequately powered to report 2 primary end points, as this would have required specifying a smaller type I error rate. We selected pain interference as the primary outcome given that our focus was on chronic pain. 100-102 Specifically, the primary outcome was defined as ≥30% improvement in the BPI pain interference score between baseline and week 12 (pain severity is included as a secondary outcome). Improvement of ≥30% in the pain score is widely viewed as clinically significant. Secondary Outcome Measures Quality of life The PROMIS-10 global health measure was used as a secondary outcome measure. This scale is now widely used to evaluate QOL and functional status outcomes for chronic pain as well as other conditions. It includes global ratings of physical function and emotional distress, as well as perceptions of general health that cut across domains ( http://www.nihpromis.org/software/assessmentcenter ). 103 Patient Global Impression of Change Postintervention, participants completed the validated Patient Global Impression of Change (PGIC), a single-question 7-point categorical scale. 104 Medication use We tracked the use of opiate medications using 2 methods. Participants were asked at baseline, 12 weeks, and 24 weeks if they had a prescription for an opiate pain reliever from a physician. For those who indicated yes, they were asked how many days in the last week they used the opiate pain reliever. In addition, we extracted prescriptions for opiates written and refilled directly from the EMR (EPIC) using EMR extraction software (Clinical Looking Glass, Emerging Health Information Technology). Sample Size Calculation The power of this noninferiority trial was evaluated based on the primary outcome of response status, defined as a ≥30% improvement on the BPI pain measure between baseline and 12 weeks. A recent review confirms that 30% improvement in chronic pain represents a clinically important change. 105 In addition, we defined the margin of noninferiority to be a difference in response rates (individual therapy rate - group therapy rate) of δ = 10%. Noninferiority would be concluded if the upper bound of the 95% CI for the true difference in response rates (individual therapy rate - group therapy rate) is <10%. For example, if in the trial, the difference was estimated to be 2% with a corresponding 95% CI of −4% to 8%, then noninferiority would be concluded because the upper bound of the CI is <10%, indicating that a true difference is statistically unlikely. We estimated that with a sample size of 282 participants per group, the study would have 80% power with a type I error rate of 5% to conclude that group therapy is noninferior to individual therapy, under the alternative hypothesis that the true response rate in both groups is 35%. We anticipated a slightly higher response rate than the 30% rate observed in ADDOPT. We made this assumption based on planned changes to how the intervention would be delivered, specifically, the use of experienced acupuncturists rather than students to administer the therapy. Moreover, the typical response rate in the literature for acupuncture for chronic pain is in the 40% to 50% range. 49 Assuming that 20% of participants might be lost to follow-up, we planned to enroll approximately 350 participants per group (700 total). Time Frame for the Study The intervention lasted 12 weeks, a duration chosen based on experience of senior practitioners and the findings of large systematic reviews with meta-analyses showing that the dose of acupuncture, including the number of sessions, affects treatment outcome for chronic pain. 49 , 106 We chose weekly sessions, as this is common in practice and in the trials analyzed 49 , 106 ; weekly sessions also facilitate patient scheduling and minimize the burden of space requirements for the health centers. Data Collection Timing and Sources Table 2 provides an overview of the timing and approach of qualitative and quantitative data collection. Data were collected at screening (assessing eligibility within 2 weeks of referral whenever possible), baseline, 6 weeks (midtreatment), 12 weeks (end of treatment), and 24 weeks. The addition of the 24-week time point allowed an assessment of the sustainability of intervention effects. Table 2 Overview of Data Collection. For all data collection, questions were administered in English or Spanish, as preferred by the participant, and in as neutral a manner as possible to avoid bias. Quantitative Data Collection and Management (Aim 1) We employed the secure, HIPAA-compliant, web-based REDCap platform 105 as a study management database. We collected all data by phone due to limited space at the health centers. Evening and weekend hours facilitated reaching working adults for data collection. Participants did not receive incentives to attend acupuncture treatments but did receive modest incentives to complete the research interviews. Screening The study coordinator (SC) or research assistant (RA) completed a screening call with each patient, typically within 1 to 2 weeks of the receipt of referral. During the screening call, we confirmed eligibility and assessed the potential participant's interest in and availability for participating in the study. Baseline interview If patients were interested and eligible, we conducted a baseline research interview, also by phone, before randomization. The baseline interview included questions about demographics and a measure of depressive symptoms, the Patient Health Questionnaire-9 (PHQ-9), in addition to the outcome measures described below. Follow-up contacts and assessments Study staff conducted weekly calls to facilitate scheduling, decrease no-shows, and remind participants of upcoming data collection. Some participants chose to discontinue treatment; all were encouraged to continue providing outcome data even if they were no longer receiving acupuncture treatments. We mailed letters 1 week before the 12- and 24-week surveys were due to remind participants of the upcoming data collection. When participants could not be reached by phone for follow-up data collection or scheduling, we checked their EMR for new or additional phone numbers. Attempts to contact participants were made during daytime, evening, and weekend hours. When attempts to reach a participant by phone were unsuccessful, we mailed a letter to the participant requesting that he or she contact us. We recorded reasons for withdrawal from treatment and, when possible, reasons for withdrawal from the study. Qualitative Data Collection and Management (Aim 2) Our second aim related to understanding the patient experience of acupuncture in both individual and group settings. We conducted qualitative interviews with a subset of participants in both arms (n = 46). Working together with our 3 patient partners, we developed a guide for semistructured interviews to be used in this process. The questions aimed to elicit (1) the anticipation of acupuncture in group and individual sessions; (2) the experience of acupuncture during sessions, including treatment experiences, social interactions with acupuncturists and other patients, and the experience of the physical environment; and (3) perceptions of the outcomes of acupuncture. After pilot testing with the first 2 participants, the interview guide was modified. In 2 waves, we recruited consecutive participants who were willing to be interviewed. Interviews were conducted by phone by the SC or a research fellow in English or Spanish, as preferred by the participants. The 30- to 60-minute interviews occurred at least 24 weeks after the initiation of acupuncture or 12 weeks after completion, whichever was later. Purposeful sampling aimed to select patients from all sites and to include representation of both sexes and patients with both good and poor functional status. Interviews were audiotaped and transcribed verbatim by a professional transcription service and Spanish interviews translated into English before analysis. During the transcription, patient names were removed to protect privacy. Preliminary analysis of the first interviews suggested that the experience of acupuncture might be changing over time as the acupuncturists gained experience and logistical challenges were addressed. Due to this insight, we temporarily stopped further data collection. We then recruited a second wave so that the final sample would include many interviews of individuals whose treatment was initiated at least 6 months after launch. Preliminary data analysis showed that data saturation had been reached after 46 interviews (23 participants from each arm), with no additional themes emerging from the final set of interviews. Analytical and Statistical Approaches Randomization Approach Randomization was stratified by source of pain, (back pain vs other pain, such as neck or OA pain) to ensure that this prognostic factor would be equally distributed across treatment arms, given our expectation that patients with back pain might be more challenging to treat in a group. The randomization scheme was computer generated by the study statistician using a random number generator in the SAS software system. The patient's PCP provided the qualifying diagnosis on the referral form. The allocation sequence was provided to the network coordinator (who supervises the SC); this individual was not involved in recruitment or scheduling and did not have contact with participants. Once baseline data and consent were completed by the SC or RAs, each participant was randomly assigned and scheduled. Study allocation was not visible to the enrolling staff or provided to the patient until baseline data were collected. Methods to Minimize Bias in Data Collection We initially planned to keep all postrandomization data collectors blind to study assignment. However, we learned early on that it would not be feasible to keep the RA unaware of study assignment. One specific example was the need to conduct scheduling calls, which by definition revealed the study arm. We safeguarded against bias in other ways. The RA and SC were trained by the principal investigators (PIs) to conduct the interviews in as neutral a manner as possible. We observed at least 10 interviews by each to confirm that additional training was not required. Our RA and SC were both bilingual in Spanish and English; thus, there was little potential for bias in data collection based on language status. All data collection was conducted as interviewer-directed, rather than self-administered, surveys to facilitate the involvement of participants with lower literacy. Methods to Prevent and Monitor Missing Data We developed a robust study management database that included reminders and reporting to alert study staff that a participant was due for data collection. We employed numerous strategies to ensure that complete data were obtained. The PIs observed the data collection process in the early phase for both the SC and RA to ensure adherence to the data collection protocol. The study statistician systematically reviewed data to highlight inconsistent or missing data. We regularly reviewed the completeness of data at weekly team meetings. We obtained multiple contact phone numbers at recruitment to maximize accessibility throughout the follow-up period. When necessary, we reviewed the contact information available in the EMR. We provided higher incentives for the 12- and 24-week data points to encourage continued participation. Data collection was completed by phone to minimize travel burden, and the entire interview was kept as brief as possible (typically 10-15 minutes) for optimal acceptability to participants. All measures were collected as interviewer directed (rather than self-administered) to minimize incomplete survey data and facilitate the inclusion of individuals with lower literacy. Finally, we made a concerted effort to maintain participants in follow-up even when some chose to discontinue acupuncture treatment. Recording and Reporting Missing Data and Reasons for Dropout For all participants who opted to discontinue acupuncture, we recorded their reason for stopping treatment. For any participant who opted to withdraw from the study, we made every effort to document their reason for withdrawal. Assessing Data Source Adequacy Data sources for this project included (1) primary survey data collected at baseline and at 6-, 12-, and 24-week follow-up; (2) interview material from participants who completed qualitative interviews; and (3) prescribed medication use information obtained from the participants' EMR and self-report. We employed validated patient-centered measures to assess the primary outcome of pain interference and secondary outcomes of pain severity, global physical and mental health, and global impression of change. Quantitative Analysis Approach (Aim 1) Baseline analyses We anticipated that any baseline participant characteristics that might be potential confounders would balance across the treatment arms with randomization. Baseline characteristics that were evaluated included level of disability, age, sex, marital status, socioeconomic status, PHQ-9 score, and race/ethnicity. As expected, none of the key potential confounders were significantly different across treatment arms. Analysis of primary and secondary outcomes Analyses evaluated whether group acupuncture is noninferior to individual acupuncture in reducing pain and improving function. After extensive data checking and verification to identify and resolve the reasons for missing values, inconsistencies, and out-of-range values, we conducted analyses of outcomes according to the ITT approach. Because ITT analyses may be anticonservative in noninferiority trials (ie, noncompliance can make it easier to demonstrate noninferiority), per-protocol (PP) analyses were also conducted. We defined participation in the intervention PP as receiving ≥8 sessions of acupuncture during the 12-week intervention period. Handling of missing data Primary analysis was based on the available data at each time point. Data could be missing entirely (patient lost to follow-up) or partially (eg, skipped questions on the BPI, resulting in insufficient data to calculate that outcome). For the analysis of the primary outcome (pain interference) and secondary outcome (pain severity), multiple imputation using chained equations was applied to fill in the missing data. Forty complete data sets were generated; we analyzed these separately and combined the results using Rubin's rules. The variables included in the imputation model were baseline demographic and clinical characteristics, such as age, sex, place of birth, ethnicity, race, household income, level of education completed, health insurance, marital status, work status, and source of pain. Also included were auxiliary variables, such as baseline and 12-week scores for secondary outcomes, including PROMIS physical and mental health scores and PHQ-9 score. Additional analyses were conducted to assess the sensitivity of the results to the values of the covariates included in the imputation model. Primary outcome: change in pain interference The primary outcome was response to treatment, defined by ≥30% improvement on the BPI pain measure between baseline and 12 weeks. The difference in response rates between the arms was estimated, along with corresponding 2-sided 95% CIs. Noninferiority of the group approach relative to the individual approach was declared if the upper limit of the 95% ICI for the true difference in response rates (individual therapy rate - group therapy rate) was less than δ, the margin of noninferiority (defined as δ = 10%). The effects of the 2 treatment arms on pain were also compared by fitting analysis of covariance (ANCOVA) models with pain as the dependent continuous variable and with treatment arm and baseline value as predictor variables. Other outcomes Mean and median levels of other patient characteristics and outcomes measured at specific visits that are continuous variables, such as global health composite scores and pain scores, were estimated and compared between treatment arms using the 2-sample t test or Wilcoxon rank sum test, depending on the distribution of the data. The chi-square or Fisher exact test was used to evaluate bivariate associations between categorical variables and treatment arms. Subgroup analyses Additional analyses were conducted to evaluate whether the comparative effectiveness of the therapy varies by source of pain (back only, neck only, OA only, or multiple); baseline PROMIS physical health score (dichotomized at median); and treatment compliance (3 levels). Initially, subgroups were defined based on the aforementioned factors, and estimates of treatment effect were obtained separately in each subgroup using the methods described previously. These analyses were viewed as exploratory because sample sizes in some subgroups were limited. We also formally tested for the significance of any heterogeneous treatment effects across these factors by including main effects for the relevant factor and treatment arm, as well as corresponding interaction terms between the 2 variables in logistic regression models, with the dependent variable specified as response to treatment (≥30% improvement on the BPI pain interference and severity scores between baseline and 12 weeks). Secondary outcomes: pain severity, global health, PGIC, and opiate use Analysis of change in secondary outcomes of pain severity (as measured by the BPI) and global health (mental and physical health subscales of the PROMIS-10 global health measure) proceeded similarly to primary outcomes. The PGIC results were dichotomized into 2 categories, better/great deal better vs other responses. Opiate use analyses were conducted using both ITT and PP samples. Standard conversions published by the American Academy of Hospice and Palliative Medicine were used to calculate oral morphine equivalents from prescriptions extracted from the EMR. 107 , 108 Quantities were assumed to cover a 30-day supply, unless otherwise specified by the prescriber. For the purpose of analysis, participants were assumed to be taking all available doses as needed every day. The average daily doses of opioids were compared in the 3 months prerandomization and 3 months postrandomization using the Wilcoxon matched-pairs signed rank test, given that the data were not normally distributed. The proportions of participants with an opioid prescription during the 3 months prerandomization and 3 months postrandomization were compared using a McNemar test for proportions for paired data. Analytic approach for ad hoc analyses Additional exploratory analyses not anticipated initially are described below. Correlation of disability and opiate use Previous studies have demonstrated a relationship between opioid analgesic use and unemployment due to disability. These studies have been limited to mainly white European and North American populations. We performed a post hoc analysis to explore the relationship between self-reported opioid analgesic use and unemployment due to disability in our study population. Associations between sociodemographic variables, pain, QOL, and depression measures and unemployment due to disability were assessed using a chi-square or Fisher exact test for categorical variables and Student's t test for continuous variables. The α = .05 level of significance was used. To test the association between opioid pain medication use and unemployment due to disability, we built a multivariable logistic regression model with unemployment due to disability as the outcome. Variables associated with opioid pain medication use with a P value of α ≤.10 in the bivariate association analysis were also included in the model as covariates. Subgroup analysis for baseline depressive symptoms We also added 1 subgroup analysis that had not been prespecified, comparing individuals with baseline PHQ-9 scores suggesting clinically significant depressive symptoms (score ≥10) to those with scores <10. Qualitative Analysis Approach (Aim 2) Qualitative data were examined and categorized using an inductive thematic analysis strategy. 109 The first 14 transcripts were read in depth by 2 of the researchers, who independently identified coding categories. A consensus set of codes was derived by research team discussion. These revised codes were then applied independently by at least 2 researchers to the remaining transcripts, and the initial 14 transcripts were recoded. The data were coded in the Dedoose software program. Differences in coding were rare and were resolved with discussion. Once the coding was completed, the immersion/crystallization approach was used to look for patterns in the data. The process of immersion followed by crystallization was repeated, alternating with group reflective analysis with the research team, until all the data had been examined and the meaningful patterns and themes were extracted and described. The interviews were then reread to identify any disconfirming data. Finally, the findings were presented to our patient partners for member checking of the themes and conclusions. Preliminary data analysis showed that data saturation had been reached after 46 interviews, with no additional themes emerging from the final set of interviews. Changes to the Original Study Protocol Minimal changes were made to the original protocol. There were no changes to eligibility criteria or analytic approach. Sites One practice was unable to participate due to construction on-site and thus had no available space for group treatment. This site was replaced with another from the same network. In addition, at our busiest sites, we continued to add sessions to meet demand. One site could not offer treatment on-site due to space limitations; we still accepted referrals, but patients were treated at a nearby site. Data Collection With existing staffing, we were unable to keep data collectors blind to study arm. Based on input from our stakeholders, we added the Social Support and Pain Questionnaire (SPQ). 110 Results Participant Flow We followed the CONSORT approach for describing randomized clinical trials (see Figure 2 ). Of 1469 referrals received from primary care clinicians, we were able to contact and screen 1341 referrals (91.3%). Of the screened individuals, 41.9% either declined participation, were lost to follow-up before random assignment, or were found to be ineligible. The most common reason for ineligibility was “not interested” (n = 61). Of the 235 individuals who declined to participate, a few indicated discomfort with acupuncture or the idea of needles (n = 3, 1.3%), and a small proportion (n = 3, 1.3%) indicated discomfort with the possibility of being randomly assigned to group acupuncture. We randomly assigned 779 individuals to the group (n = 389) or individual (n = 390) arms. Of note, 73 (9.4%) of the individuals who were eligible, completed baseline data, and were randomly assigned never initiated acupuncture. In many of these cases, there was a wait list at the site to initiate acupuncture, and when a slot became available, the individual's interest or situation had changed. The wait time from randomization to initiating intervention varied from 0 to 311 days (mean, 26 days). Of the 73 individuals who did not initiate treatment, 12 were not available at the specific times during which acupuncture sessions were offered, and 9 could not be reached. Some told us they were no longer interested (n = 9), but most (43/73) agreed to start and then missed multiple appointments. We revised our screening procedures to be more confident that participants were truly interested in starting acupuncture and to conduct the baseline assessment and random assignment much closer to the date of initiation of treatment. Some participants completed <5 treatments. The reasons for this included that they were no longer interested in the treatment (n = 31); they were interested but unable to attend treatment sessions (n = 28) due to the offered times or their availability; we were unable to contact them to obtain a formal reason (n = 4); or no further appointments were scheduled after missing 3 consecutive appointments (n = 81). Very few withdrew consent for follow-up (n = 11), but a somewhat larger number who stopped treatment could not be reached for 12- or 24-week data collection, despite our efforts to continue data collection regardless of treatment status. The proportions of the total sample assessed at 6-, 12-, and 24-week follow-up are presented in Table 3 . The rates of loss to follow-up (no data collected) were similar in the 2 arms at 12 weeks (individual treatment, 12.3%; group treatment, 12.8%). Figure 2 Participant Flow in AADDOPT-2 (CONSORT Diagram). Table 3 Proportions of Sample Assessed at Each Time Point by Level of Participation in the Intervention. Missing primary outcome data The proportions of participants with either no survey or who skipped questions, resulting in insufficient data to calculate the pain outcomes (pain interference and pain severity) at each assessment point, are provided in Table 4 . A comparison of the demographics among those with and without missing primary outcome (pain interference) at 12 weeks is shown in Table 5 . Missing pain interference data were more common among those with a high school education or less, and the mean age of participants with missing data was 52.8 years compared with 55.6 years for those without missing data. See Quantitative Analysis Approach (Aim 1) for a description of the multiple-imputation approach for handling missing data for the pain outcomes. Table 4 Missing Pain Outcome Data at Each Assessment . Table 5 Comparison of Participants With and Without Missing Data for Pain Outcomes at Baseline and 12 Weeks. Participant Baseline Characteristics Table 6 presents the baseline characteristics for the sample of randomly assigned participants (N = 779) overall and by treatment arm. For the overall sample, the mean age was 54.8 years. Participants identified as black (35.3%), white (13.4%), and multiracial (12.3%). Over half identified as Latino (56.9%). Participants were primarily Medicaid insured (75%) and reported poor/fair health (60%). Half (50%) were born outside the United States, and 20.5% spoke Spanish as their primary language. Over a third (37%) were unable to work due to disability. Back pain was the most common reason for referral (68.5%), followed by OA (26.4%) and neck pain (17.7%). Participants reported a mean score of 14.8 (scale of 0-24, with higher scores reflecting higher perceived support) on the SPQ. This mean score is similar to that from a previous study of individuals with chronic orofacial pain, where the mean score was 15.4. 110 Of the referred participants, 21.6% had more than 1 qualifying diagnosis. At baseline, participants had a mean BPI pain severity score of 6.8 and mean BPI pain interference score of 6.1. A total of 26.0% reported having a prescription for an opiate pain reliever. For PROMIS-10, participants reported a mean score of 34.8 on the physical health scale and 42.6 on the mental health scale. Table 6 Participant Baseline Demographics. The group and individual treatment arm participants were similar with regard to demographics and baseline measures. Exceptions include that a lower proportion of the group treatment arm than the individual treatment arm reported disability or retirement benefit (45.5% vs 56.6%, respectively; P = .005). Of those who reported any opiate use, group treatment participants reported fewer days of opiate use in the past week than did individual treatment participants (4.3 vs 5.1 days, respectively; P = .02). Participation in the Intervention Among participants who initiated acupuncture (≥1 treatments), the mean number of treatments was 8.0 for group treatment and 8.1 for individual treatment arm participants ( P = .56). Of these, 80.7% attended ≥5 treatments. We defined a “full course” of treatment as ≥8 treatments, which was achieved by 63.7%. Individuals randomly assigned to the individual and group treatment arms were just as likely to be retained in treatment (79.5% of those in group treatment and 81.9% of those in individual treatment had ≥5 treatments). Study arm assignment was also not associated with ≥8 treatments (63% in both arms; see Table 7 ). A slightly higher proportion of individual treatment participants initiated treatment than did group treatment participants (ie, 7.7% of those in the individual arm and 11.1% of those in the group arm had no acupuncture treatment; P = .12; see Table 8 ). Table 7 Treatment Participation by Study Arm. Table 8 Treatment Initiation by Study Arm. Primary and Secondary Outcomes (Aim 1) Results from ITT and PP Analyses: Primary Outcome The ITT population included the entire sample of 779 individuals who were randomly assigned and completed baseline data, regardless of whether or not they initiated acupuncture. To address missing data, we performed analysis of the ITT population based on the available data as well as by using multiple imputation to include all 779 randomly assigned participants. We also report for each outcome the results for the subgroup of 450 participants (n = 220 in group treatment and n = 230 in individual treatment) who attended a “full course” of treatment, which we defined as ≥8 treatment sessions. Although the number of sessions required to constitute a “full dose” is still being debated, this prespecified threshold was based on expert opinion and a large systematic review. 106 This sample (n = 450, or 63.7% of the total sample who initiated treatment) is here referred to as the PP sample. Missing data for the PP analysis were handled similarly as in the ITT analysis. Our primary outcome was change in the pain interference subscale of the BPI at 12 weeks, specifically, the proportion with clinically significant improvement (≥30% change in mean score [responders]). Table 9 summarizes the results for the primary outcome for both the ITT sample and PP sample based on the available data. In the ITT sample, the difference in mean BPI pain interference scores among group treatment participants between baseline and 12 weeks was 0.8, compared with 1.2 for individual treatment participants. The between-group difference in pain interference at 12 weeks is −0.37 (95% CI, −0.77 to 0.3) after adjusting for the baseline value in the ANCOVA model (see Figure 3 ). A 2-point reduction in mean score on the BPI pain interference and pain severity scales is considered clinically significant (reflected in Tables 9 and 10 as the mean change from baseline to 12 weeks). 101 , 111 Table 9 Primary Outcome: BPI Pain Interference. Figure 3 Mean Pain Interference by Week (ITT and PP) . In the PP analysis, the mean change in BPI pain interference among group treatment participants was 1.1 between baseline and 12 weeks compared with 1.3 for participants in the individual treatment arm. The between-group difference in pain interference at 12 weeks was −0.16 (95% CI, −0.65 to 0.32) after adjusting for the baseline value ( Figure 3 ). In the ITT responder analysis for pain interference with available (nonmissing) data, 37.5% of participants in the individual treatment arm and 30.3% in the group treatment arm had ≥30% improvement (individual - group difference [ d ] = 7.2%; 95% CI, −0.6% to 15.1%). In the PP analysis, 39.7% of individual treatment and 34.4% of group treatment participants had ≥30% improvement ( d = 5.3%; 95% CI, −4.2% to 14.9%). Noninferiority of group acupuncture was not demonstrated for the primary outcome of pain interference because the upper bound of the 95% CI of the difference in response rates (individual acupuncture - group acupuncture) was 15.1% in the ITT analysis and 14.9% in the PP analysis, both of which exceeded the maximum acceptable difference of 10% (noninferiority margin). However, we cannot conclude that individual acupuncture is significantly better than group acupuncture either, because the 95% CI overlaps zero (no difference) in both the ITT and PP analyses. In fact, because the lower bound is −4.2% in the PP analysis, we cannot rule out that the response rate with individual acupuncture is worse by 4.2% compared with group acupuncture. We also measured pain interference at 24 weeks to assess maintenance of effect in the posttreatment period. In the ITT analysis at 24 weeks, 35.0% in the individual treatment arm, compared with 28.7% in the group treatment arm, had ≥30% improvement on pain interference ( d = 6.3%; 95% CI, −1.5% to 14.0%). In the PP analysis at 24 weeks, the proportion of responders for pain interference was 39.8% for individual treatment and 30.8% for group treatment. Noninferiority of group acupuncture was not demonstrated for main outcomes in the PP sample, with a noninferiority margin of 10% at 24 weeks ( Figure 4 ). Figure 4 Proportion of Participants Responding (Pain Interference) at 12 and 24 Weeks . Multiple imputation was applied to address missing data in the primary pain interference and secondary pain severity outcomes, defined as a ≥30% improvement on the BPI score between baseline and week 12. Forty complete ITT (N = 779) and PP (N = 450) data sets were generated using the method of chained equations, and the results were combined across data sets using Rubin's rules. The results are shown in Tables 9 and 10 . Noninferiority of group therapy could not be declared in either the ITT or PP analysis at 12 weeks because the upper limit of the 95% CI exceeded the margin of noninferiority in all cases. Results of ITT and PP Analyses: Secondary Outcomes Table 11 describes the results of secondary outcome measures by time point in both ITT and PP analyses of the available data. Table 11 Demographics of Self-reported Opiate Users. Pain severity In the ITT analysis, the mean change in BPI pain severity score among group treatment participants was 1.1 between baseline and 12 weeks, compared with 1.4 for individual treatment participants. The between-group difference in pain severity score at 12 weeks was −0.26 (95% CI, −0.61 to 0.09) after adjusting for the baseline value in the ANCOVA model. In the PP analysis, the mean change in BPI severity score among group treatment participants was 1.4 between baseline and 12 weeks compared with 1.7 for individual treatment participants. The between-group difference in pain severity at 12 weeks was −0.23 (95% CI, −0.65 to 0.18) after adjusting for the baseline value ( Figure 5 ). Figure 5 Mean Pain Severity by Week (ITT and PP) . Table 10a Secondary Outcome: BPI Pain Severity. Responder analysis of pain severity In the ITT sample, 34.8% of individual treatment and 30.5% of group treatment participants had ≥30% reduction in pain severity at 12 weeks ( d = 4.3%; 95% CI, −3.3% to 11.9%; see Table 11 and Figure 6 ). In the PP sample, the proportion with ≥30% reduction in pain severity at 12 weeks was 39.2% of individual treatment arm participants compared with 36.3% of group treatment participants ( d = 2.8%; 95% CI, −6.5% to 12.2%). Noninferiority of group acupuncture was not demonstrated for pain severity at 12 weeks with a noninferiority margin of 10%. We cannot conclude superiority of individual acupuncture either, because the 95% CIs (individual - group difference) for the response rates for this outcome overlap zero (−3.3). The results were similar after applying multiple imputation to fill in the missing data ( Table 10 ). Figure 6 Proportion of Participants Responding (Pain Severity) at 12 and 24 Weeks . In the ITT sample at 24 weeks, 25.4% of individual treatment and 22.8% of group treatment participants had a ≥30% reduction in pain severity ( d = 2.6; 95% CI, −4.4% to 9.6%). In the PP sample, the proportion was 27.7% of individual treatment arm participants compared with 24.5% of group treatment participants ( d = 3.2%; 95% CI, −5.4% to 11.9%). Noninferiority of group acupuncture was demonstrated for BPI pain severity at 24 weeks. PROMIS-10 global physical health In the ITT analysis of available data, the baseline physical health T-score was 34.8 in both arms. The unadjusted change in T-score between baseline and 12 weeks (controlling for baseline score) was 3.6 in both arms (estimate, 0.08; 95% CI, −0.97 to 1.13). Physical health scores at 24 weeks showed mean unadjusted changes from baseline of 2.7 in the individual treatment arm and 2.4 in the group treatment arm (estimate, 0.38; 95% CI, −0.69 to 1.45; Figure 7 ). Higher scores reflect better perceived physical health. A 2-point change is considered clinically significant. 112 , 113 Figure 7 Mean Physical Health Score by Week (ITT) . Table 10b Secondary Outcome: PROMIS-10 Physical Health. From baseline to 12 weeks, PP analysis of the physical health scale showed mean increases of 4.2 in the individual treatment arm and 3.8 in the group treatment arm (estimate, 0.36; 95% CI, −0.87 to 1.59); at 24 weeks, the mean changes in physical health scores were 3.1 in the individual treatment arm and 2.5 in the group treatment arm (estimate, 0.64; 95% CI, −0.61 to 1.9; Figure 8 ). Figure 8 Mean Physical Health Score by Week (PP) . PROMIS-10 global mental health In the ITT analysis, the mean unadjusted changes in mental health scores at 12 weeks were 1.5 (individual treatment) and 1.1 (group treatment) (estimate, 0.48; 95% CI, −0.65 to 1.61). The mean unadjusted changes in mental health scores at 24 weeks were 1.3 (individual treatment) and 0.8 (group treatment) (estimate, 0.52; 95% CI, −0.56 to 1.61; Figure 9 ). Higher scores reflect better perceived mental health. No minimal important difference for the mental health subscale has been established. Figure 9 Mean Mental Health Score by Week (ITT) . Table 10c Secondary Outcome: PROMIS-10 Mental Health. In the PP analysis, unadjusted mental health scores at 12 weeks showed a mean change from baseline of 1.4 in the individual treatment arm and 0.8 in the group treatment arm (estimate, 0.66; 95% CI, −0.61 to 1.93); mean changes in mental health scores at 24 weeks were 1.4 (individual treatment) and 0.4 (group treatment) (estimate, 0.94; 95% CI, −0.3 to 2.18; Figure 10 ). Figure 10 Mean Mental Health Score by Week (PP) . Table 10d Secondary Outcome: PGIC. Table 10f Secondary Outcome: Self-reported Opiate Use in the Last 7 Days . Responder analysis of PROMIS-10 A 2-point change in mean T-score for the physical health subscale is considered to represent a clinically important improvement. 112 , 113 As no minimal important difference for the mental health subscale has been established, we used a change of 1/2 the standard deviation (5 points) to define a “response.” PROMIS-10 global physical health In the ITT analysis, 59.5% of group treatment and 63.1% of individual treatment arm participants had clinically important improvement at 12 weeks for physical health ( d = 3.6%; 95% CI, −4.2% to 11.4%); at 24 weeks, 50.2% of group treatment and 55.4% of individual treatment participants still reported clinically important improvement ( d = 5.2%; 95% CI, −2.9% to 13.4%). In the PP analysis, 61.3% of group treatment and 67.7% of individual treatment participants had 2-point or greater improvement at 12 weeks ( d = 6.5%; 95% CI, −2.7% to 15.6%); at 24 weeks, 49.7% of group treatment vs 59% of individual treatment arm participants still reported a response ( d = 9.3%; 95% CI, −0.5% to 19%; Figure 11 ). Figure 11 Proportion of Participants With Clinically Important Improvement in Physical Health (≥2-pt Improvement) . PROMIS-10 global mental health In the ITT analysis, 26% of group treatment and 27.6% of individual treatment arm participants had clinically important improvement (defined as a 5-point change) at 12 weeks for mental health ( d = 1.6%; 95% CI, −5.6% to 8.7%); at 24 weeks, 25.3% of group treatment and 29.2% of individual treatment participants still reported clinically important improvement ( d = 3.9%; 95% CI, −3.4% to 11.1%). In the PP analysis, 23.6% of group treatment and 26.5% of individual treatment participants had a ≥5-point improvement at 12 weeks ( d = 2.9%; 95% CI, −5.4% to 11.2%); at 24 weeks, 22.8% of group treatment vs 28.9% of individual treatment arm participants still reported a response ( d = 6.1%; 95% CI, −2.5% to 14.6%; Figure 12 ). Figure 12 Proportion of Participants With Clinically Important Improvement in Mental Health (≥5-pt Improvement) . PGIC In the ITT analysis, the proportions who responded that they were better or a great deal better were 37.5% in individual treatment and 35.0% in group treatment ( d = 2.5%; 95% CI, −5.2% to 10.2%). This proportion dropped to 25.2% in individual treatment and 26.2% in group treatment participants at 24 weeks ( d = 1.0%; 95% CI, −8.1% to 6%). In the PP analysis, the proportions who responded that they were better or a great deal better were 43.4% in individual treatment and 43.2% in group treatment at 12 weeks ( d = 0.2%; 95% CI, −9.3% to 9.6%). This proportion dropped to 32.8% in individual treatment and 31.3% in group treatment participants at 24 weeks ( d = 1.5%; 95% CI, −7.5% to 10.6%). Noninferiority was demonstrated at 12 and 24 weeks ( Figure 13 ). Figure 13 PGIC (ITT and PP) . SPQ In the ITT analysis, the mean unadjusted changes in SPQ scores at 12 weeks were 1.0 (individual treatment) and 0.7 (group treatment) (estimate, 0.4; 95% CI, −0.28 to 1.07). The mean unadjusted changes in SPQ scores at 24 weeks were 0.5 (individual treatment) and 1.1 (group treatment) (estimate, −0.43; 95% CI, −1.08 to 0.27; Table 10e ). In the PP analysis, the mean unadjusted changes in SPQ scores at 12 weeks were 0.9 (individual treatment) and 1.1 (group treatment) (estimate, 0.05; 95% CI, −0.72 to 0.82). The mean unadjusted changes in SPQ scores at 24 weeks were 0.4 (individual treatment) and 1.0 (group treatment) (estimate, −0.25; 95% CI, −1.05 to 0.56). Higher scores reflect more perceived support. No minimum clinically important difference has been established for this measure. Table 10e Secondary Outcome: SPQ. Opiate use By self-report, the proportion of participants in the ITT sample who reported having a prescription for an opioid pain reliever at baseline was 21.3% overall (20.3% individual treatment, 22.3% group treatment). Participants taking opioid pain relievers had higher baseline pain scores, higher PHQ-9 scores, and lower physical health scores than the whole group of participants included in the intent-to-treat analyses. Those with an opioid prescription were more likely to be disabled and to have lower income ( Table 11 ). Self-reported opiate use over time by arm Table 12 and Figure 14 show the change over time in the proportion who reported using a prescribed opiate pain reliever in the past 7 days, by arm, for ITT and PP samples. There were no significant differences at each time point between individual and group arms in either the ITT or PP samples. In the ITT sample, there was no significant difference in the proportions of the total sample who reported using an opiate pain reliever at baseline vs 12 weeks (21.3% vs 18.1%, respectively; P = .17). In the PP sample, the results were similar, with no significant difference between baseline and 12 weeks for the total sample (18.4% vs 15.4%, respectively; P = .12). Table 12 Self-reported Opiate Use in the Past 7 Days (ITT and PP). Figure 14 Self-reported Opiate Use in Past 7 Days (ITT and PP) . Among all participants who reported using a prescribed opiate pain reliever, participants reported using it 4.7 days in the previous week at baseline. Table 13 shows the change over time in the mean days of use, for individual and group treatment arms, for the ITT and PP samples. For the ITT sample, there was no significant difference between the arms at 6, 12, or 24 weeks. There was no significant difference in mean days of use for the total sample at baseline and 12 weeks (4.7 vs 4.6, respectively; P = .17). For the PP sample, there was no significant difference between the arms at any time point. There was a modest decrease in the mean days of use for the total sample at baseline and 12 weeks (4.3 vs 3.9, respectively; P = .003; Figure 15 ). Table 13 Days of Opiate Use in the Past Week (ITT and PP). Figure 15 Mean Days of Opiate Use in Past Week (ITT and PP) . EMR-based analysis of opiate use Based on prescriptions extracted from the EMR, the proportion of participants with an opioid prescription in the ITT sample was significantly higher in the 3 months prerandomization than in the 3 months postrandomization for participants randomly assigned to individual acupuncture treatments (16.4% vs 11.0%, respectively; P = .003) but not for participants randomly assigned to group acupuncture (13.1% vs 14.3%, respectively; P = .39; see Table 14 ). Similarly, in the PP sample, the proportion of participants with an opioid prescription was significantly higher in the 3 months prerandomization than in the 3 months postrandomization for participants randomly assigned to individual acupuncture treatments (15.2% vs 9.6%, respectively; P = .012) but not for participants randomly assigned to group acupuncture (11.8% vs 12.7%, respectively; P = .68). Table 14 Proportion of Participants With Opioid Prescriptions (From EMR Data) for ITT and PP Samples. In the ITT sample, there was no significant difference in the average daily dose of opioid (in morphine milliequivalents) 3 months prerandomization and 1 to 3 months postrandomization for all participants who had any opioid prescription during the study period (24.4 mg vs 22.9 mg, respectively; P = .62). This finding did not significantly differ for participants randomly assigned to individual acupuncture treatment (32.4 mg vs 30.7 mg, respectively; P = .20) or group acupuncture treatment (16.5 mg vs 15.1 mg, respectively; P = .54). In the PP sample, there was no significant difference in the average daily dose of opioid 3 months prerandomization and 3 months postrandomization for all participants who had any opioid prescription during the study period (27.2 mg vs 23.4 mg, respectively; P = .43). Participants randomly assigned to individual acupuncture treatment had a decrease in dosage during the study period (39.1 mg vs 30.4 mg, respectively; P = .05), but those in group acupuncture treatment did not (13.5 mg vs 15.2 mg, respectively; P = .27; see Table 15 and Figure 16 ). Table 16 describes the mean change in morphine milliequivalents used 3 months prerandomization and 3 months postrandomization for each arm and the difference between treatment arms. There was no significant difference in the mean change of the doses for ITT or PP samples, although a trend was observed in the PP analysis ( P = .06). Table 15 MME From EMR Data Pre- and Postrandomization (ITT and PP). Figure 16 MME From EMR Data Pre- and Postrandomization by Arm (ITT and PP) . Table 16 Between-Group Mean Change in MME Used Pre- and Postrandomization. Subgroup Analyses Prespecified subgroup analyses explored whether the interventions were differentially effective for specific diagnoses (ie, neck, back, or OA pain) and participants with poor physical health (dichotomized at median score on the physical health scale of the PROMIS-10). We added another subgroup of interest, those with elevated depressive symptoms (PHQ-9 score <10 vs ≥10). The primary outcome of pain interference and secondary outcome of pain severity were dichotomized as respondent/nonrespondent. Both ITT and PP analyses were performed. Tables 17 to 19 present the proportion of each subgroup with ≥30% improvement in each arm. The subgroup analyses mostly mirror our overall finding that participants in the individual arm had slightly better rates of response than did group participants. Treatment effect differences across different sources of pain were not detected for pain severity or interference in the ITT and PP analyses (all P values for treatment × source of pain interaction terms exceeded .40). An interesting trend (although nonsignificant) is that group treatment participants had better outcomes than did individual treatment participants for pain interference in the neck pain subgroup (36.8% vs 30.8%, respectively [ITT] and 41.7% vs 33.3%, respectively [PP]). Among the subgroup with multiple diagnoses, group therapy participants in the PP sample had better response rates than individual therapy participants for both pain severity (37.1% vs 31.8%, respectively) and interference (38.2% vs 30.2%, respectively). Given the multiple tests conducted here, however, this could be a chance finding. Table 17 Responder Proportion by Pain Subgroup and Treatment Arms (ITT and PP). Low vs high depressive symptoms In the ITT sample, the proportion of all participants with clinically significant improvement was higher for individuals in the low-depressive-symptom subgroup (both arms) than in the high-depressive-symptom subgroup for both pain severity (37.0% vs 26.0%, respectively; P = .006) and pain interference (39.5% vs 25.7%, respectively; P = .0009). In the PP sample, this finding persisted for pain severity (42.7% vs 29.3%, respectively; P = .007) and interference (41.6% vs 30.3%, respectively; P = .026). In the ITT sample, among those with high depressive symptoms, a slightly higher proportion of group treatment participants responded for pain interference than did individual treatment participants ( d = 1.9%; 90% CI, −11.6% to 7.9%). Similarly, in the PP sample, a higher proportion of group treatment participants responded for pain interference than did individual treatment participants ( d = 9.4%; 90% CI, −22.2% to 3.4%; see Table 18 ). Table 18 Response Rates by PHQ-9 Subgroup. Low vs high physical health For all participants, the proportion with clinically significant improvement was higher for individuals with better physical health scores than for those with lower physical health scores for both pain severity (40.2% vs 22.4%, respectively; P ≤ .0001) and interference (44.4% vs 20.9%, respectively; P ≤ .0001) in the ITT sample. There was a similar finding in the PP sample for pain severity (45.3% vs 26.1%, respectively; P ≤ .0001) and interference (46.9% vs 23.2%, respectively; P ≤ .0001; see Table 19 ). Table 19 Response Rates by PROMIS-10 Physical Health Score Subgroup. Results of Ad Hoc Analyses Association of disability and opiate use Of the 765 participants who enrolled in the trial and completed baseline interviews, 150 patients were excluded because they reported being retired. An additional 11 participants were excluded due to missing data on opioid pain medication use. Participants who used opioid pain medication (n = 136) were more likely to be born in the United States (64.7% vs 46.8%, respectively; P = .002), to be male (25.0% vs 17.1%, respectively; P = .04), and to speak English as their primary language (84.3% vs 76.6%, respectively; P = .02). Opioid pain medication use was not associated with race or ethnicity. Participants who used opioid pain medication reported poorer function on the PROMIS-10 global mental health and global physical health measures, more symptoms of depression on the PHQ-9, and greater pain interference, but not greater baseline pain severity. Participants who used opioid pain medication were more likely to be unable to work due to disability (73.5% vs 39.3%, respectively; P < .0001) and to receive Supplemental Security Income (59.6% vs 37.2%, respectively; P < .001). When controlling for potential confounders in multivariable analysis, participants who used opioid pain medication had 3 times the odds of being unable to work due to disability compared with those who did not use opioids (odds ratio, 3.080; P < .001). Relationship of dose of acupuncture to pain severity For the total sample, a dose-response relationship was observed. Participants who received 1 to 4 treatments had a reduction in mean pain severity score of 0.63, compared with 0.78 for the group that received 5 to 8 treatments and 1.6 for the group that received 9 to 12 treatments. The mean changes from baseline to 12 weeks were significantly different across different dose groups ( P < .001; see Table 20a ). After adjusting for baseline score and randomization type, the mean changes in pain severity score remained significantly different between participants receiving 1 to 4 or 5 to 8 treatments and those receiving 9 to 12 treatments ( P < .001; see Table 20b ). Table 20a Relationship of Treatments Attended and Change in BPI Pain Severity Score for All Participants. Table 20b Linear Regression Model Coefficients for Change in BPI Pain Severity Score as the Outcome, Adjusting for Baseline Score, Randomization Type, and Number of TRTs Attended. When we evaluated the proportion of participants who had clinically significant improvement, a similar dose-response relationship was present. Participants with 9 to 12 treatments had the largest proportion with ≥30% improvement in pain severity (38.7%), compared with those with 5 to 8 treatments (25.3%) and 1 to 4 treatments (19.5%; see Table 21 ). Table 21 Relationship of Improvement of Pain Severity and Number of Treatments for All Participants. Qualitative Results—Patient Experience of Acupuncture (Aim 2) Patients in both study arms valued the pain relief, improved QOL, and relaxation experienced during acupuncture. Privacy and mixed-gender groups were cited as concerns by a minority of patients; however, most of those randomly assigned to the group setting noted that these concerns abated over time. Differences between arms included the depth of the relationship with the acupuncturist and misgivings related to the clinical treatment space. Patients who participated in qualitative interviews were generally similar to the study population. Responses are coded here by treatment type (G for group and I for individual), sex (F for female and M for male), language spoken (E for English and S for Spanish), and age in years (Y). Value of acupuncture Patients in both arms described valuing acupuncture as an additional option for pain relief. Even those who did not experience pain relief valued the opportunity to try a new modality. I'm very pleased with the results. Sincerely, they are very good. Good, good, good, like something that dropped from Heaven. (G/F/S/54Y) Well, just let me say that, for me, since I have had treatment with acupuncture, I've felt… I think the word is “useful.” Before this, I didn't feel useful for anything. I felt terrible. (G/F/S/54Y) Treatment experience The majority of patients in both arms experienced minimal discomfort, with only a few reporting significant pain. None stopped treatment early due to pain. Most of them I didn't feel at all and once, you know, if there was any, it was like a slight pinch like a mosquito bite and that's it. (I/F/E/74Y) Group acupuncture When asked to recall their initial enrollment in the study, most participants reported no worry about potentially being treated in a group setting. I don't care if the audience was there, you know? Their issue, I assume, is the pain, too. (G/M/E/68Y) A few patients in both arms recalled feeling uncomfortable before starting treatment with the idea of receiving acupuncture in a group setting. I'm very careful of anybody sharing my information out there, even with my body, you know, even though you're not completely naked but just anybody just looking at me in a weird kind of way, I'm very uncomfortable with. (I/F/E/32Y) For those who were actually assigned to group acupuncture, concerns about physical privacy and informational privacy were often relieved after experiencing the group setting. Patients assigned to group acupuncture were more likely to voice concerns about the physical environment of treatment sessions. Well, because [the acupuncturist] would come to each of us and she had a nice soft voice…Other people would hear but it was nothing, it was really nothing major that I couldn't say in front of other people. (G/F/E/53Y) Some patients in the group setting continued to experience mild discomfort in mixed-gender groups. Most patients, however, reported that mixed-gender groups did not affect their acupuncture experience. I felt a little more uncomfortable when there—when it was a man, you know, because…I had to…unbutton my shirt or whatever. I mean, they didn't see anything, but I just felt a little uncomfortable. (G/F/E/71Y) It wasn't difficult…sometimes you would have a woman sitting there and I thought, you know, it should be the women in a different room setting and men in a different room. (G/M/E/66Y) It was fine. There was 1 older man, that's it, and he would talk about his grandchildren. (G/F/E/59Y) Most patients assigned to group acupuncture described being able to achieve a state of deep relaxation during treatments. Very few patients assigned to group acupuncture, and none of those assigned to individual acupuncture, found it difficult to relax. Patients in both arms had generally positive things to say about their acupuncturists. Several patients assigned to individual acupuncture described meaningful bonds that developed with the acupuncturist. She didn't just talk about acupuncture, she made me feel important. (I/F/E/51Y) It was as if she were my psychologist, giving like emotional therapy, and I liked that. (I/F/S/63Y) It was hard to let her go, not to see her again. (I/F/E/58Y) In contrast, patients assigned to group acupuncture tended to describe the relationship with the acupuncturist in more general terms. I cannot remember her name at the moment, but she was very nice. (G/F/E/33Y) Sort of friendly. I didn't get too personal with her, but she was friendly. (G/M/E/66Y) Finally, groups varied in their social dynamics. Some groups interacted more and sometimes provided an extra layer of support for participants. Other groups remained more reserved, and interactions were described as inhibited. It's better in a group anyway, at least you got some company while you [are] sitting there and you can talk to somebody instead of being in a room by yourself with needles stuck in you staring at the walls, make you go crazy. (G/F/E/50Y) The group was good because I am with people who are worse than I am and who have more experience. That was like my source of support, and my encouragement to continue. (G/F/S/54Y) I was very mindful of that and didn't want to intrude on other people's experience. (G/F/E/33Y) One Spanish-speaking participant noted that language was a barrier to interactions with other group members. Discussion Context for Study Results Noninferiority of group to individual acupuncture was not demonstrated for the primary pain outcomes at 12 weeks in either the ITT or the PP analysis because the upper bound of the 95% CIs of the difference in response rates between individual and group treatment exceeded the noninferiority margin of 10% in both cases. Response rates for the primary outcome of pain interference and secondary outcome of pain severity were consistently slightly better in the individual treatment arm than in the group treatment arm. However, we cannot conclude that the individual acupuncture approach is superior to group acupuncture, because the 95% CIs also overlapped zero in both the ITT and PP analyses. The only analysis that demonstrated noninferiority was pain severity at 24 weeks in the ITT sample (a secondary outcome). Noninferiority was also not demonstrated for either the physical or mental health subscale of the PROMIS-10 on either ITT or PP analysis. The results on the PGIC measure mirrored the primary pain results, although on this measure, we did demonstrate noninferiority of group to individual treatment. We found clinically significant improvement (≥30% improvement in pain from baseline) in both the group and individual treatment arms for a substantial proportion of participants for pain severity and pain interference at 12 weeks in both our ITT and PP analyses. In the ITT sample, for pain severity, 34.8% in the individual treatment and 30.5% in the group treatment arm experienced a ≥30% improvement from baseline at 12 weeks; for pain interference, these numbers were 37.5% in individual treatment and 30.3% in group treatment. In the PP results, 39.2% of individual treatment and 36.3% of group treatment participants had a ≥30% improvement in pain severity at 12 weeks. For pain interference, 39.7% of individual treatment and 34.4% of group treatment participants had a ≥30% improvement. We found a large proportion with clinically meaningful improvement in the physical health subscale of the PROMIS-10 (59.5% for group and 63.1% for individual treatment in the ITT sample) but a much smaller proportion with clinically meaningful improvement in mental health (26% for group and 27.6% for individual treatment in the ITT sample). Overall, improvement in primary pain outcomes was greater for those with ≥8 treatments, for those with better functional status at baseline, and for those with the lowest level of baseline depressive symptoms. We also demonstrated for the total sample a significant dose-response relationship, in that participants having 1 to 4 treatments had a reduction in mean BPI pain severity score of 0.63, compared with 0.78 for 5 to 8 treatments and 1.6 for 9 to 12 treatments. Our ITT sample included 73 participants who were randomly assigned but never initiated acupuncture, as well as those who had very few treatments. As might be expected, a larger proportion of individuals in both arms had clinically significant improvement in pain outcomes in the PP analysis than in ITT analysis. Similarly, for secondary outcomes of physical health and PGIC score, overall improvement was larger in the PP sample. Finally, in our analysis of opiate pain medication use, by self-report, just over a quarter of the sample used opiates for their chronic pain. The proportion based on EMR data was slightly smaller, presumably due to some patients obtaining those prescriptions from physicians outside the integrated delivery system. Based on EMR data, opiate prescriptions declined in the individual treatment arm but not in the group treatment arm when comparing the 3 months preintervention to the 3 months postintervention. Overall, the proportion of participants experiencing clinically significant improvement in pain was smaller than that seen in the most definitive recent individual patient meta-analysis of acupuncture trials. Data analyzed from 39 trials (20 827 patients) showed that acupuncture was superior to sham and to a no-acupuncture control in the treatment of chronic musculoskeletal pain, including back and neck pain, OA, and chronic headache. 49 Using a pain reduction measure of ≥50%, response rates were 30% (no-acupuncture controls), 42.5% (sham acupuncture controls), and 50% (true acupuncture). Two major factors could have contributed to the overall lower response to our acupuncture intervention than that in the reported trials, which are detailed below. Treatment Setting For both individual and group arms, delivery in a busy CHC presented challenges. The physical plant was designed for needs of primary care, as opposed to the more optimal physical environment tested in most of the published clinical trials to date, and this might have affected the outcomes. Individual treatment occurred in medical examination rooms with examination tables that are not designed for a comfortable supine or prone position. We used body cushions to provide the option of a prone position, but this also made the table quite high, limiting access for differently abled participants. Group sessions were scheduled in multipurpose conference rooms with a table and plastic chairs. The limitations to physical access in group treatment were anticipated. Achieving the state of deep rest that some feel is an important aspect of the benefit of acupuncture is more difficult in this type of setting, particularly as it pertains to the group treatment arm. Study Population Our population was different in many ways from those in most clinical trials to date, in that our participants often had multiple significant comorbidities, including depression, higher levels of disability, lower functional status, and significant socioeconomic and biopsychosocial challenges. It is clearly reported in the literature that pain and pain care affect specific socioeconomic and demographic groups differently, 12 with increased vulnerability to pain and more limited access to pain treatment experienced by lower socioeconomic populations with a high comorbidity load. 7 , 13 , 14 Individuals with pain have higher rates of functional limitations than do individuals without pain; they develop functional limitations classically associated with aging at much earlier ages. 24 Functional limitations in multimorbid populations are directly associated with higher Medicare and Medicaid health care resource use and expenditures, 25 , 26 which is an indication of the challenges in treating pain populations with high comorbidity load and reduced functional status. Patients of lower socioeconomic status may also have lower perceived levels of control over pain, which is correlated with greater pain intensity. 114 Although we found a surprising gap in the comparative effectiveness literature regarding the question of whether underserved and highly comorbid populations generally show lower response rates to pain interventions than do other populations, it is reasonable to hypothesize that the functional limitations described could have contributed to the lower response rate. A recent noninferiority trial by Saper et al of yoga vs physical therapy (PT) for chronic low back pain (cLBP) in an underserved population found response rates very similar to those in our study, potentially supporting this hypothesis. 115 Pain reduction of ≥30% was gained by 35% of yoga participants at 12 weeks, noninferior to PT. These results were with a much more stringent set of exclusion criteria than we employed in our trial, excluding patients with comorbid rheumatoid arthritis or fibromyalgia, patients who had had a previous surgical procedure, patients with active worker's compensation or similar cases, and those who were not proficient in English. 116 The more flexible inclusion criteria in our study could have biased toward a lower response rate. Saper et al also noted that despite pain's disproportionate impact on minority and low socioeconomic status groups, few cLBP studies and even fewer nonpharmacologic intervention trials have targeted these populations. Furthermore, our previous NIH-funded study of individual acupuncture in this setting similarly found that roughly a third of participants had a ≥30% improvement in pain, 54 suggesting that this may be a more typical response in this population than the higher proportion seen in clinical trials in general that recruit participants with higher functional status and lower comorbidity load. Regarding the difference in outcomes between the group and individual treatment arms, there are very few published trials of group acupuncture to provide context for our results. A recent uncontrolled pilot trial by our group that provided group acupuncture in a more optimal setting to a somewhat less-challenging population did find a larger proportion of patients experiencing a clinically significant reduction in pain and depression than found in AADDOPT-2 in a more underserved population. 67 The issue of constraints on the physical environment might have contributed to the lower proportion benefiting in the group treatment arm. In particular, the challenge of treating patients seated in chairs with limited capacity for accessing acupuncture points on the trunk and upper legs might have affected our results. From a decision-making point of view, although we did find that group treatment was not as effective as individual treatment, it is clear based on our study that both individual and group acupuncture can be offered safely in the CHC setting and that a substantial proportion of patients with chronic pain will have clinically significant improvement. In light of the recent Joint Commission mandate that hospitals provide nonpharmacologic options for pain treatment, 74 and CDC, 72 FDA, 73 NIH, 51 and ACP 50 recommendations that acupuncture be included as part of a strategy to treat acute and chronic pain and to mitigate opioid risks, this is an important and impactful finding. From the point of view of both patients and clinicians, based on the rapid pace of recruiting and the high proportion of participants who followed through with at least 8 sessions in both the group and individual treatment arms, having acupuncture available at the primary care site is clearly a desirable option. Another interesting and important finding is that there was no difference in the average numbers of treatment sessions for participants in the 2 arms and no difference in the numbers of participants starting treatment postrandomization. In other words, even participants who might have had an initial preference for individual treatment reliably initiated and continued treatment in the group setting. This confirms the idea that group acupuncture is highly acceptable to patients and that differences in outcomes are not related to differences in participation. Because insurance coverage for acupuncture is still fairly limited, including for Medicaid and Medicare, the mechanism for financing acupuncture care in this particular population remains a challenge. Although participants requested treatment access beyond the trial and the leadership of the participating CHCs without exception expressed a desire to continue offering acupuncture—and although our stakeholder from the payer side (the Montefiore ACO) was also supportive—the lack of a business case to demonstrate a break-even outcome based on insurance reimbursement continues to be a major barrier. Generalizability Our study population was ethnically diverse and representative of many urban underserved populations nationally. Based on our success in integrating acupuncture services into the challenging setting at our CHCs, we feel that implementing the interventions in other primary care settings, such as large group practices, will be entirely feasible. The generalizability of our findings to other populations will need to be tested in further studies; in particular, the acceptability and effectiveness of the group treatment and the potentially differential effect of offering the group treatment in a more comfortable physical setting should be investigated. The effectiveness of individual acupuncture treatment has already been demonstrated in multiple clinical trials and large meta-analyses. Group acupuncture has reported positive results in the treatment of peripheral neuropathy, 59 cancer symptoms, 56 knee OA, 57 , 58 , 60 and chronic pain generally. 61 , 62 , 65-68 , 117 This being the first trial comparing individual with group treatment, it is somewhat difficult to draw definitive conclusions regarding the generalizability of the group approach used here to other populations; additional studies of group acupuncture in this and other populations are needed. Implementation of Study Results We demonstrated that both individual and group acupuncture can be successfully implemented in the primary care setting within a large urban ambulatory care network. The rapid pace of referral and recruiting for this study, even among a population previously unfamiliar with acupuncture, affirmed the strong demand for acupuncture as a nonpharmacologic option for chronic pain treatment and suggests that the service would be welcomed and used in other similar settings. One barrier we encountered was a lack of health system familiarity in credentialing and supervising licensed acupuncturists, despite the fact that this system had employed physicians certified in medical acupuncture. We were successful in overcoming this barrier and learned lessons that we feel would be helpful to both researchers and clinicians working to bring acupuncture to other large health system settings. A second similar barrier to CHC (FQHCs specifically) implementation was the fact that the scope of practice defined by the Health Resources & Services Administration (HRSA), which oversees FQHCs, did not initially include acupuncture. This required a somewhat involved process of working with the community board of the FQHCs to enable a revised scope of practice under the HRSA guidelines. Awareness of this requirement and the potential delays it could pose to implementation will be important for health centers looking to incorporate acupuncture treatment in the future. Another aspect of our study that has potential implications for implementation is our manualized protocol, which we developed through a consensus process at the start of the project. This manual, which we used to ensure both consistency and flexibility in the delivery of acupuncture, could potentially facilitate the implementation of acupuncture for pain in various other populations and clinical settings. Subpopulation Considerations The proportion of participants with clinically significant improvement in both pain severity and pain interference was higher, in both arms, for individuals in better physical health than in those with lower physical health scores. This held true in both the ITT and PP analyses. Similarly, the response rates on both subscales of the BPI in both analyses were higher for individuals in the low-depressive-symptom subgroup than in the high-depressive-symptom subgroup. Interestingly, group treatment was superior to individual treatment for those with elevated depressive symptoms. One might speculate that the social contact and support in the group setting contributed to this outcome, because it is known that social isolation can contribute to increased pain. Clearly, this would need to be tested in a future study. We also found a trend toward better outcomes in group vs individual treatment arms for participants with OA and neck pain; we did not find this for back pain. This preliminary finding raises the possibility that treating back pain in the group setting is more difficult, as access to points on the lower back is more challenging than in the individual setting. These findings are not definitive but may represent areas for future investigation or strategic implementation. For example, a health center could offer both group and individual options but help manage access to the individual slots by referring patients with neck or OA pain to the group sessions. Study Limitations From a design perspective, the most significant limitation relates to lack of a third arm representing usual care, as a 3-arm trial was not feasible within the constraints of our study budget. This limitation challenges the attribution of the participants' benefit to the acupuncture treatment. Although large individual data meta-analyses establish that acupuncture is effective for chronic pain compared with sham or usual care, our study design was not aimed at confirming these results but at comparing deployment in distinct settings. Additionally, as a pragmatic study, we included individuals with chronic pain who were referred by their physicians; we did not have a minimum BPI pain score for enrollment. This resulted in some individuals with less-severe pain being randomly assigned, though overall, the participants experienced high levels of pain. This might have diluted the true difference in efficacy between group and individual acupuncture, thereby making the treatment arms appear to be more similar and potentially increasing the probability of erroneously demonstrating noninferiority. However, this concern does not undermine our conclusion because we were still unable to demonstrate noninferiority of group acupuncture to individual acupuncture. Another limitation relates to the 73 individuals who were randomly assigned but never initiated treatment. For any participant who initiated treatment, we collected follow-up data regardless of the number of treatments attended. However, for those who never engaged in any treatment, we did not attempt to collect further data from them. Ideally, such data would have been collected; however, they were included with imputation in the ITT analysis. Our ability to ensure that data collectors were never aware of study arm assignment was compromised by limited staffing. Another limitation was that the sample size, though quite large, was not large enough for further subgroup analyses. We added the SPQ; however, the results should be interpreted with caution. We were unable to find evidence in the literature that the measure is sensitive to change; therefore, failure to demonstrate significant change over time may be due to limitations of the measure itself. Finally, although the proportions of individuals assessed at 12 and 24 weeks were relatively high given the population, there was still somewhat more loss to follow-up than would have been ideal. From an implementation perspective, another limitation of this study is the suboptimal physical setting of the group acupuncture delivery. This could have biased results away from noninferiority of group to individual treatment. In future implementation, this limitation could be mitigated while preserving the advantages of a group setting by the addition of more comfortable chairs and 1 or 2 mobile treatment tables to allow patients in the group setting to be prone or supine to facilitate access of mid-body points. We were unable to include patients who were not fluent in either English or Spanish. Thus, our participants are not fully representative of the patients seen in urban primary care (the proportion of patients in our health centers who are not fluent in either English or Spanish is between 5% and 10%, depending on the specific center). There were several limitations to use of EMR data to examine opioid analgesic use. First, participants likely obtained prescriptions outside our health system, which were unavailable for analysis. Second, the assumption was made that participants used all as-needed doses available to them by prescription, which is unlikely to be how participants actually use their medications. Future Research Several potential areas for further research emerge from our findings. First is the question of whether a more optimal environment for the group treatment might lead to greater benefit. It is possible that even in the CHC setting, one could make more comfortable chairs or mobile treatment tables accessible, which might affect outcomes. Second is the question of whether specific subgroups truly do benefit more from 1 of the 2 approaches, an example being the greater improvement we saw in more-depressed participants who were in the group treatment arm. Based on our qualitative data, another question that may require further research is the question of whether single-gender groups may have an advantage, particularly for women. Pain populations with lower functional status and high comorbidity load are characteristically sedentary. Pain and impaired function are often met with fear of activity and lack of familiarity with beneficial movement. An important question in the field that we hope to address in future research is whether combining an “active” intervention, such as yoga or tai chi, with the acupuncture treatment might lead to better and more sustained benefit. Conclusions AADDOPT-2 sought to answer the question of whether acupuncture for chronic pain delivered in a group setting is as effective as individual acupuncture in an underserved and ethnically diverse patient population at high risk for health disparities. Noninferiority of group to individual acupuncture was not demonstrated for the primary pain outcome (ie, pain interference) at 12 weeks in either the ITT or PP analysis. The ITT analysis did show clinically significant improvement (≥30% improvement from baseline) in both arms in pain severity at 12 weeks in 34.8% of individual and 30.5% of group treatment arm participants; for pain interference, these numbers were 37.5% for individual treatment and 30.3% for group treatment. The PP analysis showed higher proportions responding: For pain severity at 12 weeks, 39.2% of individual and 36.3% of group participants responded; for pain interference at 12 weeks, 39.7% of individual and 34.4% of group participants responded. Response rates for pain interference and pain severity were consistently slightly better in the individual treatment arm than in the group treatment arm; however, we cannot conclude statistically that individual acupuncture treatment is superior to group treatment. More than half of participants had clinically meaningful improvement in physical health on the ITT analysis (59.5% for group and 63.1% for individual treatment), while a much smaller proportion had clinically meaningful improvement in mental health (26% for group and 27.6% for individual treatment). Improvement in primary pain outcomes was greater for those with ≥8 treatments, for those with better functional status at baseline, and for those with the lowest level of baseline depressive symptoms. Qualitative findings further clarified the patient experience, as described previously. Our results should inform hospital and primary care leaders, as well as policy makers, in that they demonstrate that individual and group acupuncture can be offered safely in the CHC setting, that acceptability to patients and clinicians is very high, and that a substantial proportion of patients with chronic pain will have clinically significant improvement in both pain and well-being when treatment is delivered in individual and group settings. This finding is especially important In light of the recent mandates for nonpharmacologic options for pain treatment. 74 Although the results from group treatment are encouraging in that we did find nearly equivalent outcomes in the 2 arms, the fact that group delivery of acupuncture therapy was not noninferior to individual treatment suggests that further research is needed on the optimal strategy for delivering group acupuncture in this context before it is considered to be as effective as individual treatment. That said, the safety and acceptability of the group approach, the fact that a substantial proportion of the participants benefited, and the advantages group treatment offers in terms of access should all be of interest to decision-makers at the local, state, and national levels as they plan their strategy for widespread deployment of acupuncture therapy as a part of pain care in urban underserved patient populations. References 1. Fishman S, Ballantyne J, Rathmell JP, Bonica JJ. Bonica's Management of Pain. 4th ed. Lippincott, Williams & Wilkins; 2010. 2. Treede RD, Rief W, Barke A, et al. A classification of chronic pain for ICD-11. Pain. 2015;156(6):1003-1007. [ PMC free article : PMC4450869 ] [ PubMed : 25844555 ] 3. Blackwell DL, Lucas JW, Clarke TC. Summary health statistics for U.S. adults: National Health Interview Survey, 2012. National Center for Health Statistics. Updated January 14, 2017. Accessed January 14, 2017. https://www ​.cdc.gov/nchs ​/data/series/sr_10/sr10_260.pdf [ PubMed : 24819891 ] 4. Brown A. Chronic pain rates shoot up until Americans reach late 50s. Gallup Healthways. Published 2012. Accessed May 6, 2017. http://www ​.gallup.com ​/poll/154169/Chronic-Pain-Rates-Shoot-Until-Americans-Reach-Late-50s ​.aspx?ref=image 5. Gaskin DJ, Richard P. The economic costs of pain in the United States. J Pain. 2012;13(8):715-724. [ PubMed : 22607834 ] 6. Grol-Prokopczyk H. Sociodemographic disparities in chronic pain, based on 12-year longitudinal data. Pain. 2017;158(2):313-322. [ PMC free article : PMC5242384 ] [ PubMed : 28092650 ] 7. Institute of Medicine, Committee on Advancing Pain Research, Care and Education. Relieving Pain in America: A Blueprint for Transforming Prevention, Care, Education, and Research. National Academies Press; 2011. [ PubMed : 22553896 ] 8. Johannes CB, Le TK, Zhou X, Johnston JA, Dworkin RH. The prevalence of chronic pain in United States adults: results of an Internet-based survey. J Pain. 2010;11(11):1230-1239. [ PubMed : 20797916 ] 9. Kennedy J, Roll JM, Schraudner T, Murphy S, McPherson S. Prevalence of persistent pain in the U.S. adult population: new data from the 2010 National Health Interview Survey. J Pain. 2014;15(10):979-984. [ PubMed : 25267013 ] 10. Reyes-Gibby C, Aday L, Cleeland C. Impact of pain on self-rated health in the community-dwelling older adults. Pain. 2002;95(1-2):75-82. [ PubMed : 11790469 ] 11. GBD 2015 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and national incidence, prevalence, and years lived with disability for 310 diseases and injuries, 1990-2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet. 2016;388(10053):1545-1602. [ PMC free article : PMC5055577 ] [ PubMed : 27733282 ] 12. Tick H, Nielsen A, Pelletier KR, et al. Evidence-based nonpharmacologic strategies for comprehensive pain care: the Consortium Pain Task Force White Paper. Explore (NY). 2018;14(3):177-211. [ PubMed : 29735382 ] 13. Institute of Medicine. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care (with CD). National Academies Press; 2003. [ PubMed : 25032386 ] 14. National Academies of Sciences Engineering and Medicine. Pain Management and the Opioid Epidemic: Balancing Societal and Individual Benefits and Risks of Prescription Opioid Use. National Academies Press; 2017. [ PubMed : 29023083 ] 15. Hardt J, Jacobsen C, Goldberg J, Nickel R, Buchwald D. Prevalence of chronic pain in a representative sample in the United States. Pain Med. 2008;9(7):803-812. [ PubMed : 18346058 ] 16. Unruh AM. Gender variations in clinical pain experience. Pain. 1996;65(2-3):123-167. [ PubMed : 8826503 ] 17. Plesh O, Adams SH, Gansky SA. Racial/ethnic and gender prevalences in reported common pains in a national sample. J Orofac Pain. 2011;25(1):25-31. [ PMC free article : PMC3117989 ] [ PubMed : 21359234 ] 18. Campaign to End Chronic Pain in Women. Chronic pain in women: neglect, dismissal and discrimination. Published 2010. Accessed June 19, 2017. http://www ​.endwomenspain.org/ 19. Elliott A, Smith B, Hannaford P, Smith W, Chambers W. The course of chronic pain in the community: results of a 4-year follow-up study. Pain. 2002;99(1-2):299-307. [ PubMed : 12237208 ] 20. Gureje O, Simon G, Von Korff M. A cross-national study of the course of persistent pain in primary care. Pain. 2001;92(1-2):195-200. [ PubMed : 11323140 ] 21. Gureje O, Von Korff M, Simon G, Gater R. Persistent pain and well-being: a World Health Organization study in primary care. JAMA. 1998;280(2):147-151. [ PubMed : 9669787 ] 22. Goetzel R, Hawkins K, Ozminkowski R, Wang S. The health and productivity cost burden of the “top 10” physical and mental health conditions affecting six large U.S. employers in 1999. J Occup Environ Med. 2003;45(1):5-14. [ PubMed : 12553174 ] 23. Dominick CH, Blyth FM, Nicholas MK. Unpacking the burden: understanding the relationships between chronic pain and comorbidity in the general population. Pain. 2012;153(2):293-304. [ PubMed : 22071318 ] 24. Covinsky KE, Lindquist K, Dunlop DD, Yelin E. Pain, functional limitations, and aging. J Am Geriatr Soc. 2009;57(9):1556-1561. [ PMC free article : PMC2925684 ] [ PubMed : 19682122 ] 25. Koroukian SM, Schiltz NK, Warner DF, et al. Multimorbidity: constellations of conditions across subgroups of midlife and older individuals, and related Medicare expenditures. J Comorbid. 2017;7(1):33-43. [ PMC free article : PMC5556436 ] [ PubMed : 29090187 ] 26. Schiltz NK, Warner DF, Sun J, et al. Identifying specific combinations of multimorbidity that contribute to health care resource utilization: an analytic approach. Med Care. 2017;55(3):276-284. [ PMC free article : PMC5309172 ] [ PubMed : 27753745 ] 27. Vickers AJ, Cronin AM, Maschino AC, et al. Acupuncture for chronic pain: individual patient data meta-analysis. Arch Intern Med. 2012;172(19):1444-1453. [ PMC free article : PMC3658605 ] [ PubMed : 22965186 ] 28. Vickers AJ, Linde K. Acupuncture for chronic pain. JAMA. 2014;311(9):955-956. [ PMC free article : PMC4036643 ] [ PubMed : 24595780 ] 29. Yuan QL, Wang P, Liu L, et al. Acupuncture for musculoskeletal pain: a meta-analysis and meta-regression of sham-controlled randomized clinical trials. Sci Rep. 2016;6:30675. doi:10.1038/srep30675 [ PMC free article : PMC4965798 ] [ PubMed : 27471137 ] [ CrossRef ] 30. MacPherson H, Vickers A, Bland M, et al. Acupuncture for chronic pain and depression in primary care: a programme of research. Programme Grants for Appl Res. 2017;5(3):1-16. [ PubMed : 28121095 ] 31. Lathia AT, Jung SM, Chen LX. Efficacy of acupuncture as a treatment for chronic shoulder pain. J Altern Complement Med. 2009;15(6):613-618. [ PubMed : 19489707 ] 32. Lee SH, Lim SM. Acupuncture for poststroke shoulder pain: a systematic review and meta-analysis. Evid Based Complement Alternat Med. 2016;2016:3549878. doi:10.1155/2016/3549878 [ PMC free article : PMC4983325 ] [ PubMed : 27547224 ] [ CrossRef ] 33. Chou R, Deyo R, Friedly J, et al. Nonpharmacologic therapies for low back pain: a systematic review for an American College of Physicians clinical practice guideline. Ann Intern Med. 2017;166(7):493-505. [ PubMed : 28192793 ] 34. Liu L, Skinner M, McDonough S, Mabire L, Baxter GD. Acupuncture for low back pain: an overview of systematic reviews. Evid Based Complement Alternat Med. 2015;2015:328196. doi:10.1155/2015/328196 [ PMC free article : PMC4364128 ] [ PubMed : 25821485 ] [ CrossRef ] 35. Tang S, Mo Z, Zhang R. Acupuncture for lumbar disc herniation: a systematic review and meta-analysis. Acupunct Med. 2018;36(2):62-70. [ PubMed : 29496679 ] 36. Yuan QL, Guo TM, Liu L, Sun F, Zhang YG. Traditional Chinese medicine for neck pain and low back pain: a systematic review and meta-analysis. PLoS One. 2015;10(2):e0117146. doi:10.1371/journal.pone.0117146 [ PMC free article : PMC4339195 ] [ PubMed : 25710765 ] [ CrossRef ] 37. Willich SN, Reinhold T, Selim D, Jena S, Brinkhaus B, Witt CM. Cost-effectiveness of acupuncture treatment in patients with chronic neck pain. Pain. 2006;125(1-2):107-113. [ PubMed : 16842918 ] 38. Trinh KV, Graham N, Gross AR, et al. Acupuncture for neck disorders. Cochrane Database Syst Rev. 2006;(3):CD004870. doi:10.1002/14651858.CD004870 [ PubMed : 16856065 ] [ CrossRef ] 39. Berman B, Lao L, Langenberg P, Lee W, Gilpin A, Hochberg M. Effectiveness of acupuncture as adjunctive therapy in osteoarthritis of the knee: a randomized, controlled trial. Ann Intern Med. 2004;141(12):901-910. [ PubMed : 15611487 ] 40. Kwon YD, Pittler MH, Ernst E. Acupuncture for peripheral joint osteoarthritis: a systematic review and meta-analysis. Rheumatology. 2006;45(11):1331-1337. [ PubMed : 16936326 ] 41. Lin X, Huang K, Zhu G, Huang Z, Qin A, Fan S. The effects of acupuncture on chronic knee pain due to osteoarthritis: a meta-analysis. J Bone Joint Surg Am. 2016;98(18):1578-1585. [ PubMed : 27655986 ] 42. Zhang Y, Bao F, Wang Y, Wu Z. Influence of acupuncture in treatment of knee osteoarthritis and cartilage repairing. Am J Transl Res. 2016;8(9):3995-4002. [ PMC free article : PMC5040698 ] [ PubMed : 27725880 ] 43. Manyanga T, Froese M, Zarychanski R, et al. Pain management with acupuncture in osteoarthritis: a systematic review and meta-analysis. BMC Complement Altern Med. 2014;14:312. doi:10.1186/1472-6882-14-312 [ PMC free article : PMC4158087 ] [ PubMed : 25151529 ] [ CrossRef ] 44. Woods B, Manca A, Weatherly H, et al. Cost-effectiveness of adjunct non-pharmacological interventions for osteoarthritis of the knee. PLoS One. 2017;12(3):e0172749. doi:10.1371/journal.pone.0172749 [ PMC free article : PMC5340388 ] [ PubMed : 28267751 ] [ CrossRef ] 45. Linde K, Allais G, Brinkhaus B, et al. Acupuncture for the prevention of tension-type headache. Cochrane Database Syst Rev. 2016;(4):CD007587. doi:10.1002/14651858.CD007587.pub2 [ PMC free article : PMC4955729 ] [ PubMed : 27092807 ] [ CrossRef ] 46. Linde K, Allais G, Brinkhaus B, et al. Acupuncture for the prevention of episodic migraine. Cochrane Database Syst Rev. 2016;(6):CD001218. doi:10.1002/14651858.CD001218.pub3 [ PMC free article : PMC4977344 ] [ PubMed : 27351677 ] [ CrossRef ] 47. Wu JY, Zhang C, Xu YP, et al. Acupuncture therapy in the management of the clinical outcomes for temporomandibular disorders: a PRISMA-compliant meta-analysis. Medicine (Baltimore). 2017;96(9):e6064. doi:10.1097/MD.000000000000604 [ PMC free article : PMC5340435 ] [ PubMed : 28248862 ] [ CrossRef ] 48. MacPherson H, Vertosick EA, Foster NE, et al. The persistence of the effects of acupuncture after a course of treatment: a meta-analysis of patients with chronic pain. Pain. 2017;158(5):784-793. [ PMC free article : PMC5393924 ] [ PubMed : 27764035 ] 49. Vickers AJ, Vertosick EA, Lewith G, et al. Acupuncture for chronic pain: update of an individual patient data meta-analysis. J Pain. 2018;19(5):455-474. [ PMC free article : PMC5927830 ] [ PubMed : 29198932 ] 50. Qaseem A, Wilt TJ, McLean RM, Forciea M, Clinical Guidelines Committee of the American College of Physicians. Noninvasive treatments for acute, subacute, and chronic low back pain: a clinical practice guideline from the American College of Physicians. Ann Intern Med. 2017;166(7):514-530. [ PubMed : 28192789 ] 51. Nahin RL, Boineau R, Khalsa PS, Stussman BJ, Weber WJ. Evidence-based evaluation of complementary health approaches for pain management in the United States. Mayo Clin Proc. 2016;91(9):1292-1306. [ PMC free article : PMC5032142 ] [ PubMed : 27594189 ] 52. Chou R, Cote P, Randhawa K, et al. The Global Spine Care Initiative: applying evidence-based guidelines on the non-invasive management of back and neck pain to low- and middle-income communities. Eur Spine J. 2018;27(Suppl 6):851-860. [ PubMed : 29460009 ] 53. McKee MD, Kligler B, Blank AE, et al. The ADDOPT study (Acupuncture to Decrease Disparities in Outcomes of Pain Treatment): feasibility of offering acupuncture in the community health center setting. J Altern Complement Med. 2012;18(9):839-843. [ PMC free article : PMC3429271 ] [ PubMed : 22867026 ] 54. McKee MD, Kligler B, Fletcher J, et al. Outcomes of acupuncture for chronic pain in urban primary care. J Am Board Fam Med. 2013;26(6):692-700. [ PMC free article : PMC3979293 ] [ PubMed : 24204065 ] 55. Becker WC, Dorflinger L, Edmond SN, Islam L, Heapy AA, Fraenkel L. Barriers and facilitators to use of non-pharmacological treatments in chronic pain. BMC Fam Pract. 2017;18(1):41. doi:10.1186/s12587-017-0608-2 [ PMC free article : PMC5359906 ] [ PubMed : 28320337 ] [ CrossRef ] 56. Tofthagen C, Boses S, Healy G, Jooma N. Evaluation of group acupuncture for cancer-related symptoms: a retrospective analysis. J Palliat Med. 2015;18(10):878-880. [ PubMed : 26218367 ] 57. White A, Richardson M, Richmond P, Freedman J, Bevis M. Group acupuncture for knee pain: evaluation of a cost-saving initiative in the health service. Acupunct Med. 2012;30(3):170-175. [ PMC free article : PMC3470298 ] [ PubMed : 22914300 ] 58. White A, Tough L, Eyre V, et al. Western medical acupuncture in a group setting for knee osteoarthritis: results of a pilot randomised controlled trial. Pilot Feasibility Stud. 2016;2:10. doi:10.1186/s40814-016-0051-5 [ PMC free article : PMC5153913 ] [ PubMed : 27965830 ] [ CrossRef ] 59. Phillips KD, Skelton WD, Hand GA. Effect of acupuncture administered in a group setting on pain and subjective peripheral neuropathy in persons with human immunodeficiency virus disease. J Altern Complement Med. 2004;10(3):449-455. [ PubMed : 15253848 ] 60. Berkovitz S, Cummings M, Perrin C, Ito R. High volume acupuncture clinic (HVAC) for chronic knee pain--audit of a possible model for delivery of acupuncture in the National Health Service. Acupunct Med. 2008;26(1):46-50. [ PubMed : 18356798 ] 61. Chao MT, Tippens KM, Connelly E. Utilization of group-based, community acupuncture clinics: a comparative study with a nationally representative sample of acupuncture users. J Altern Complement Med. 2012;18(6):561-566. [ PMC free article : PMC3390970 ] [ PubMed : 22551075 ] 62. Chuang E, Hashai N, Buonora M, Gabison J, Kligler B, McKee MD. “It's better in a group anyway”: patient experiences of group and individual acupuncture. J Altern Complement Med. 2018;24(4):336-342. [ PMC free article : PMC6179144 ] [ PubMed : 29446994 ] 63. Rohleder L. Community acupuncture: making buckets from Ming vases. J Chinese Med. 2012;98:22-25. 64. Deadman P. Community acupuncture-making Ming vases from buckets: a reply to Lisa Rohleder. J Chinese Med. 2012;99:55-59. 65. Asprey A, Paterson C, White A. ‘All in the same boat': a qualitative study of patients' attitudes and experiences in group acupuncture clinics. Acupunct Med. 2012;30(3):163-169. [ PubMed : 22914301 ] 66. Tippens KM, Chao MT, Connelly E, Locke A. Patient perspectives on care received at community acupuncture clinics: a qualitative thematic analysis. BMC Complement Altern Med. 2013;13:293. doi:10.1186/1472-6882-13-293 [ PMC free article : PMC4231447 ] [ PubMed : 24168022 ] [ CrossRef ] 67. Kligler B, Nielsen A, Kohrrer C, et al. Acupuncture therapy in a group setting for chronic pain. Pain Med. 2017;19(2):393-403. [ PubMed : 28595273 ] 68. Kligler B, Buonora M, Gabison J, Jacobs E, Karasz A, McKee MD. “I Felt Like It Was God's Hands Putting the Needles In”: a qualitative analysis of the experience of acupuncture for chronic pain in a low-income, ethnically diverse, and medically underserved patient population. J Altern Complement Med. 2015;21(11):713-719. [ PMC free article : PMC4642820 ] [ PubMed : 26247238 ] 69. Davis RT, Badger G, Valentine K, Cavert A, Coeytaux RR. Acupuncture for chronic pain in the Vermont Medicaid population: a prospective, pragmatic intervention trial. Glob Adv Health Med. 2018;7:2164956118769557. [ PMC free article : PMC5896847 ] [ PubMed : 29662722 ] 70. Gardiner P, Dresner D, Barnett KG, Sadikova E, Saper R. Medical group visits: a feasibility study to manage patients with chronic pain in an underserved urban clinic. Glob Adv Health Med. 2014;3(4):20-26. [ PMC free article : PMC4104558 ] [ PubMed : 25105072 ] 71. Geller JS, Kulla J, Shoemaker A. Group medical visits using an empowerment-based model as treatment for women with chronic pain in an underserved community. Glob Adv Health Med. 2015;4(6):27-60. [ PMC free article : PMC4653596 ] [ PubMed : 26659335 ] 72. Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids for chronic pain--United States, 2016. JAMA. 2016;315(15):1624-1645. [ PMC free article : PMC6390846 ] [ PubMed : 26977696 ] 73. US Food and Drug Administration. FDA education blueprint for health care providers involved in the management or support of patients with pain (May 2017). Published 2017. Accessed May 22, 2017. https://www ​.fda.gov/drugs ​/news-events-human-drugs ​/fda-education-blueprint-health-care-providers-involved-management-or-support-patients-pain 74. Joint Commission enhances pain assessment and management requirements for accredited hospitals. Perspectives, the Official Newsletter of The Joint Commission. Published 2017. Accessed November 27, 2017. https://www ​.jointcommission ​.org/assets/1 ​/18/Joint_Commission ​_Enhances_Pain_Assessment ​_and_Management ​_Requirements_for_Accredited_Hospitals1 ​.PDF 75. MacPherson H, Altman DG, Hammerschlag R, et al. Revised STandards for Reporting Interventions in Clinical Trials of Acupuncture (STRICTA): extending the CONSORT statement. J Evid Based Med. 2010;3(3):140-155. [ PubMed : 21349059 ] 76. Schulz KF, Altman DG, Moher D. CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. PLoS Med. 2010;7(3):e1000251. doi:10.1371/journal.pmed.1000251 [ PMC free article : PMC2844794 ] [ PubMed : 20352064 ] [ CrossRef ] 77. Nielsen A, Anderson B, Citkovitz C, et al. Developing and employing a ‘responsive manualization' in the ‘Acupuncture Approaches to Decrease Disparities in Outcomes of Pain Treatment' comparative effectiveness study. Acupunct Med. 2019;37(3):184-191. [ PubMed : 30900480 ] 78. MacPherson H, Sherman K, Hammerschlag R, Birch S, Lao L, Zaslawski C. The clinical evaluation of traditional East Asian systems of medicine. Clin Acup Orient Med. 2002;3(1):16-19. 79. Witt CM, Huang WJ, Lao L, Berman B. Which research is needed to support clinical decision-making on integrative medicine?- Can comparative effectiveness research close the gap? Chin J Integr Med. 2012;18(10):723-729. [ PubMed : 22965697 ] 80. Schnyer RN, Allen JJ. Bridging the gap in complementary and alternative medicine research: manualization as a means of promoting standardization and flexibility of treatment in clinical trials of acupuncture. J Altern Complement Med. 2002;8(5):623-634. [ PubMed : 12470444 ] 81. Hsu CC, Sandford BA. The Delphi technique: making sense of consensus. Pract Assess Res Eval (PARE). 2007;12:10. https: ​//scholarworks ​.umass.edu/pare/vol12/iss1/10 82. Smith CA, Grant S, Lyttleton J, Cochrane S. Using a Delphi consensus process to develop an acupuncture treatment protocol by consensus for women undergoing Assisted Reproductive Technology (ART) treatment. BMC Complement Altern Med. 2012;12:88. [ PMC free article : PMC3416745 ] [ PubMed : 22769059 ] 83. Clean Needle Technique Manual: Best Practices for Acupuncture Needle Safety and Related Procedures. 7th ed. Council of Colleges of Acupuncture and Oriental Medicine; 2015. 84. Chen X, Chen B, Gu X, Shi Z, Chen Z. [Role of pressing hand in the clinical practice of acupuncture]. Zhongguo Zhen Jiu. 2017;37(11):1215-1217. [ PubMed : 29354960 ] 85. Ni Y, Rosenbaum RL. Navigating the Channels of Traditional Chinese Medicine. Oriental Medicine Center; 1996. 86. Nielsen A. Gua Sha: A Traditional Technique for Modern Practice. 2nd ed. Churchill Livingstone Elsevier; 2013. 87. Langevin HM, Churchill DL, Wu J, et al. Evidence of connective tissue involvement in acupuncture. FASEB J. 2002;16(8):872-874. [ PubMed : 11967233 ] 88. Bisio T, Butler F. Zheng Gu Tui Na: A Chinese Medical Massage Textbook. Zheng Gu Tui Na; 2007. 89. Pritchard S. Tui Na: The Chinese Massage Manual. Churchill Livingstone; 2009. 90. Nielsen A. Gua sha research and the language of integrative medicine. J Bodyw Mov Ther. 2009;13(1):63-72. [ PubMed : 19118794 ] 91. Kwong KK, Kloetzer L, Wong KK, et al. Bioluminescence imaging of heme oxygenase-1 upregulation in the Gua Sha procedure. J Vis Exp. 2009(30):1385. doi:10.3791/1385 [ PMC free article : PMC3149908 ] [ PubMed : 19718012 ] [ CrossRef ] 92. Wang SM, Peloquin C, Kain ZN. The use of auricular acupuncture to reduce preoperative anxiety. Anesth Analg. 2001;93(5):1178-1180. [ PubMed : 11682391 ] 93. de Lorent L, Agorastos A, Yassouridis A, Kellner M, Muhtz C. Auricular acupuncture versus progressive muscle relaxation in patients with anxiety disorders or major depressive disorder: a prospective parallel group clinical trial. J Acupunct Meridian Stud. 2016;9(4):191-199. [ PubMed : 27555224 ] 94. Asher GN, Jonas DE, Coeytaux RR, et al. Auriculotherapy for pain management: a systematic review and meta-analysis of randomized controlled trials. J Altern Complement Med. 2010;16(10):1097-1108. [ PMC free article : PMC3110838 ] [ PubMed : 20954963 ] 95. Yeh CH, Chiang YC, Hoffman SL, et al. Efficacy of auricular therapy for pain management: a systematic review and meta-analysis. Evid Based Complement Alternat Med. 2014;2014:934670. doi:10.1155/2014/934670 [ PMC free article : PMC4140110 ] [ PubMed : 25165482 ] [ CrossRef ] 96. Quach CW, Langer MM, Chen RC, et al. Reliability and validity of PROMIS measures administered by telephone interview in a longitudinal localized prostate cancer study. Qual Life Res. 2016;25(11):2811-2823. [ PMC free article : PMC6126915 ] [ PubMed : 27240448 ] 97. Zeng L, Sahgal A, Zhang L, et al. Patterns of pain and functional improvement in patients with bone metastases after conventional external beam radiotherapy and a telephone validation study. Pain Res Treat. 2011;2011:601720. doi:10.1155/2011/601720 [ PMC free article : PMC3197188 ] [ PubMed : 22110927 ] [ CrossRef ] 98. Tan G, Jensen MP, Thornby JI, Shanti BF. Validation of the Brief Pain Inventory for chronic nonmalignant pain. J Pain. 2004;5(2):133-137. [ PubMed : 15042521 ] 99. Badia X, Muriel C, Gracia A, et al. [Validation of the Spanish version of the Brief Pain Inventory in patients with oncological pain]. Med Clin (Barc). 2003;120(2):52-59. [ PubMed : 12570914 ] 100. Deyo RA, Dworkin SF, Amtmann D, et al. Report of the NIH Task Force on research standards for chronic low back pain. J Pain. 2014;15(6):569-585. [ PMC free article : PMC4128347 ] [ PubMed : 24787228 ] 101. Dworkin RH, Turk DC, Wyrwich KW, et al. Interpreting the clinical importance of treatment outcomes in chronic pain clinical trials: IMMPACT recommendations. J Pain. 2008;9(2):105-121. [ PubMed : 18055266 ] 102. Younger J, McCue R, Mackey S. Pain outcomes: a brief review of instruments and techniques. Curr Pain Headache Rep. 2009;13(1):39-43. [ PMC free article : PMC2891384 ] [ PubMed : 19126370 ] 103. Hays RD, Bjorner JB, Revicki DA, Spritzer KL, Cella D. Development of physical and mental health summary scores from the Patient-Reported Outcomes Measurement Information System (PROMIS) global items. Qual Life Res. 2009;18(7):873-880. [ PMC free article : PMC2724630 ] [ PubMed : 19543809 ] 104. Farrar J, Young JJ, LaMoreaux L, Werth J, Poole R. Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain. 2001;94(2):149-158. [ PubMed : 11690728 ] 105. Frahm Olsen M, Bjerre E, Hansen MD, Tendal B, Hilden J, Hrobjartsson A. Minimum clinically important differences in chronic pain vary considerable by baseline pain and methodological factors: systematic review of empirical studies. J Clin Epidemiol. 2018;101:87-106.e2. [ PubMed : 29793007 ] 106. MacPherson H, Maschino AC, Lewith G, et al. 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. doi:10.1371/journal.pone.0077438 [ PMC free article : PMC3795671 ] [ PubMed : 24146995 ] [ CrossRef ] 107. Fine PG, Portenoy RK, Ad Hoc Expert Panel on Evidence Review and Guidelines for Opioid Rotation. Establishing “best practices” for opioid rotation: conclusions of an expert panel. J Pain Symptom Manage. 2009;38(3):418-425. [ PMC free article : PMC4065110 ] [ PubMed : 19735902 ] 108. Knotkova H, Fine PG, Portenoy RK. Opioid rotation: the science and the limitations of the equianalgesic dose table. J Pain Symptom Manage. 2009;38(3):426-439. [ PubMed : 19735903 ] 109. Braun V, Clarke V. Using thematic analysis in psychology. Qual Res Psych. 2006;4(2):77-101. 110. Van Der Lugt CM, Rollman A, Naeije M, Lobbezoo F, Visscher CM. Social support in chronic pain: development and preliminary psychometric assessment of a new instrument. J Oral Rehabil. 2012;39(4):270-276. [ PubMed : 22115492 ] 111. Atkinson TM, Mendoza TR, Sit L, et al. The Brief Pain Inventory and its “pain at its worst in the last 24 hours” item: clinical trial endpoint considerations. Pain Med. 2010;11(3):337-346. [ PMC free article : PMC3806650 ] [ PubMed : 20030743 ] 112. Hays RD, Spritzer KL, Fries JF, Krishnan E. Responsiveness and minimally important difference for the patient-reported outcomes measurement information system (PROMIS) 20-item physical functioning short form in a prospective observational study of rheumatoid arthritis. Ann Rheum Dis. 2015;74(1):104-107. [ PMC free article : PMC3976454 ] [ PubMed : 24095937 ] 113. Lee AC, Driban JB, Price LL, Harvey WF, Rodday AM, Wang C. Responsiveness and minimally important differences for 4 patient-reported outcomes measurement information system short forms: physical function, pain interference, depression, and anxiety in knee osteoarthritis. J Pain. 2017;18(9):1096-1110. [ PMC free article : PMC5581239 ] [ PubMed : 28501708 ] 114. Vallerand AH, Crawley J, Pieper B, Templin TN. The perceived control over pain construct and functional status. Pain Med. 2016;17(4):692-703. [ PubMed : 26350223 ] 115. Saper RB, Lemaster C, Delitto A, et al. Yoga, physical therapy, or education for chronic low back pain: a randomized noninferiority trial. Ann Intern Med. 2017;167(2):85-94. https://www ​.ncbi.nlm ​.nih.gov/pmc/articles/PMC6392183/ [ PMC free article : PMC6392183 ] [ PubMed : 28631003 ] 116. Saper RB, Lemaster CM, Delitto A, et al. Yoga, physical therapy or education for chronic low back pain. A randomized noninferiority trial. Ann Intern Med. 2017;Supplement 1: Detailed Study Protocol. https://www ​.acpjournals ​.org/doi/suppl/10 ​.7326/M16-2579/suppl_file ​/M16-2579_Supplement_1.pdf [ PMC free article : PMC6392183 ] [ PubMed : 28631003 ] 117. Chuang E, Hashai N, Buonora M, Gabison J, Kligler B, McKee MD. “I'm trying to get rid of some pain. I didn't care about all of that”: patient experiences of group and individual acupuncture in a low-income, racially and ethnically diverse urban population. Paper presented at: 2017 North American Primary Care Research Group (NAPCRG) Annual Meeting; November 17-21, 2017; Montreal, Quebec, Canada. Related Publications Published Chuang E, Hashai N, Buonora M, Gabison J, Kligler B, McKee MD. “It's better in a group anyway”: patient experiences of group and individual acupuncture. J Altern Complement Med. 2018;24(4):336-342. [ PMC free article : PMC6179144 ] [ PubMed : 29446994 ] Chuang E, Gil EN, Gao Q, Kligler B, McKee MD. Relationship between opioid analgesic prescription and unemployment in patients seeking acupuncture for chronic pain in urban primary care. Pain Med. 2019;20(8):1528-1533. [ PMC free article : PMC7739953 ] [ PubMed : 30184213 ] McKee MD, Nielsen A, Anderson B, et al. Individual vs. group delivery of acupuncture therapy for chronic musculoskeletal pain in urban primary care—a randomized trial. J Gen Intern Med. 2020;35(4):1227-1237. [ PMC free article : PMC7174252 ] [ PubMed : 32076985 ] Nielsen A, Anderson B, Citkovitz C, et al. Developing and employing a ‘responsive manualization' in the ‘Acupuncture Approaches to Decrease Disparities in Outcomes of Pain Treatment' comparative effectiveness study. Acupunct Med. 2019;37(3):184-191. [ PubMed : 30900480 ] Acknowledgments We acknowledge our patient partners, who made invaluable contributions to the design and implementation of the study: John MacDonald, Linda Canales, and Adelaida Suarez. We also acknowledge the caring and expert treatment provided by our research acupuncturists: Donna Mah, Dana Moore, Patricia Botet, Valentina Duque, Amy Pagliarini, Selina Greene, and Susana Correia. Research reported in this report was funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award (#AD-1402-10857). Further information available at: https://www.pcori.org/research-results/2014/comparing-group-and-individual-acupuncture-therapy-treating-chronic-pain-among Original Project Title: Acupuncture Approaches to Decrease Disparities in Outcomes of Pain Treatment—A Two Arm Comparative Effectiveness Trial (AADDOPT-2) PCORI ID: AD-1402-10857 ClinicalTrials.gov ID: NCT02456727 Suggested citation: McKee MD, Kligler B, Anderson B, et al. (2020). Comparing Group and Individual Acupuncture Therapy for Treating Chronic Pain among Ethnically Diverse Patients with Low Incomes—AADDOPT-2) . Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/06.2020.AD.140210857 Disclaimer The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee. Copyright © 2020. Albert Einstein College of Medicine. All Rights Reserved. This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/ Bookshelf ID: NBK620248 PMID: 41529127 DOI: 10.25302/06.2020.AD.140210857 Share Views PubReader Print View Cite this Page McKee MD, Kligler B, Anderson B, et al. Comparing Group and Individual Acupuncture Therapy for Treating Chronic Pain among Ethnically Diverse Patients with Low Incomes—AADDOPT-2 [Internet]. Washington (DC): Patient-Centered Outcomes Research Institute (PCORI); 2020 Jun. doi: 10.25302/06.2020.AD.140210857 PDF version of this title (3.3M) In this Page Background Patient and Stakeholder Engagement Methods Results Discussion Conclusions References Related Publications Acknowledgments Other titles in this collection PCORI Final Research Reports Related information NLM Catalog Related NLM Catalog Entries PMC PubMed Central citations PubMed Links to PubMed Recent Activity Clear Turn Off Turn On Comparing Group and Individual Acupuncture Therapy for Treating Chronic Pain amo... Comparing Group and Individual Acupuncture Therapy for Treating Chronic Pain among Ethnically Diverse Patients with Low Incomes—AADDOPT-2 Your browsing activity is empty. Activity recording is turned off. Turn recording back on See more... Follow NCBI Twitter Facebook LinkedIn GitHub NCBI Insights Blog Connect with NLM Twitter Facebook Youtube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov