Title: An Adaptive Pragmatic Randomized Controlled Trial of Emergency Department Acupuncture for Acute Musculoskeletal Pain Management

Importance: Acute musculoskeletal pain is challenging to treat in the Emergency Department (ED), a novel setting to incorporate nonpharmacologic interventions, such as acupuncture, to reduce chronic pain. Objective: Determine effectiveness of ED acupuncture for treating acute musculoskeletal pain. Design: Pragmatic two-stage adaptive randomized clinical trial over one year. Stage 1 identified whether auricular acupuncture (AA) or peripheral acupuncture (PA) was more feasible and efficacious in the ED. Stage 2 assessed effectiveness of selected ED acupuncture intervention(s) on 1-hour pain reduction compared to no acupuncture (NA). Setting: Academic tertiary-care ED. Participants: Adults with acute ( ≤ 7 days) musculoskeletal pain in neck, back and/or extremities. Of 911 patients screened, 491 met inclusion criteria, 247 enrolled, 11 withdrew, and 236 were randomized (68 NA, 84 AA, 84 PA). Interventions: AA places ear needles based on battlefield acupuncture. PA places needles in head, neck, and extremity sites distal to pain. Acupuncture was delivered by licensed acupuncturists. All three arms received usual care from blinded ED providers. Main Outcomes: Primary outcome: 11-point Numeric pain rating scale at 1-hour. Secondary outcomes: Satisfaction with treatment; ED opioids given. Results: Interim analysis at end of Stage 1 found both acupuncture styles similarly feasible and efficacious, so Stage 2 continued all three arms with unequal allocation (from 1:1:1 to 1:2:2 NA:AA:PA). All three arms had similar demographics (mean age 46.1 years ±16.5 SD, 56.2% female, and 53.6% Black), ED baseline pain scores (AA 7.0 ± 2.3, PA 7.2 ± 2.3, NA 6.9 ± 2.2) and pre-baseline ED administration of opioid and non-opioid analgesics. Both AA (2.1 ± 2.9) and PA (1.6 ± 2.0) produced significantly greater 1-hour pain reductions than NA (


INTRODUCTION
Musculoskeletal disorders are the leading cause of pain and disability, affecting over 1.7 billion people worldwide. 1Chronic musculoskeletal pain impairs health, function and well-being, and is associated with chronic opioid use. 2,3As chronic pain begins as acute pain and is frequently marked by acute severe exacerbations, mitigating chronic pain requires effective acute pain management.Treatment is particularly challenging in emergency department (ED) settings, where acute pain is typically severe and associated with emotional and psychological stressors and diagnostic uncertainty. 4,57][8][9] Thus, more innovative and multi-modal approaches to ED musculoskeletal pain management are needed.
International organizations are increasingly calling for nonpharmacologic strategies, including acupuncture, to treat pain. 10Acupuncture is safe and effective for chronic pain conditions such as cancer pain, neck and low back pain, knee osteoarthritis, and headache. 11,12ts analgesic effects are postulated to be mediated through inflammatory and endogenous opioid pathways. 13Side effects are typically mild, transient needle site pain or bruising, and serious events are extremely rare. 14,15However, traditional private hour-long acupuncture sessions are less feasible in space-and time-limited ED environments.More efficient treatment styles have been developed for community clinic and military settings.These include auricular acupuncture (AA) with ear needles placed based on the battlefield acupuncture protocol, 16,17 and peripheral acupuncture (PA) where needles are limited to head, neck, and extremities distal to the painful site to treat pain. 18,19However, there is limited data on the efficacy of ED acupuncture for acute pain, and the acupuncture style best-suited to the ED environment remains unclear. 20erefore, the purpose of this study was to (1) identify which style of acupuncture is feasible, acceptable, and effective for treating acute musculoskeletal pain in the ED, and (2) determine effectiveness of that acupuncture style compared to control (no acupuncture) on pain reduction in the ED.

METHODS AND ANALYSIS: Study design and ethics
This pragmatic, two-stage adaptive randomized clinical trial was approved by the Institutional Review Board (Protocol # Pro00104140), registered on 2/7/2020 with clinicaltrials.gov(registration # NCT04290741) and released to the public on 2/28/2020.The detailed study methods were previously published. 21In brief, the first stage objective was to identify the more effective style of ED-based acupuncture, AA or PA, and the second stage was to assess effectiveness of the selected acupuncture style on pain reduction.A planned interim analysis at the end of Stage 1, based on the FDA-recommended criterion of probability of being the best treatment, 22,23 was reviewed by an independent data-safety monitoring committee (DSMC) to identify the acupuncture style and study adaptations for Stage 2. The only change from Stage 1 to Stage 2 was unequal allocation to treatment arms, described below.In addition, there were two phases to the study: Phase 1 focused on 1 hour ED acupuncture outcomes, followed by Phase 2 enrolling a larger number of patients and focused on the outpatient acupuncture clinic outcomes.This report is for Phase 1 ED outcomes only, as Phase 2 is ongoing.We used the SPIRIT and CONSORT checklists. 24,25udy setting and recruitment Trial participants were enrolled between February 10, 2020 and May 3, 2021 in a U.S. academic tertiary-care ED with 80,000 visits per year.Trained clinical research coordinators (CRCs) performed all patient screening, recruitment, informed e-consent, and enrollment procedures.Study acupuncturists explained and delivered acupuncture treatments 21 .

Inclusion and exclusion criteria
English-speaking adult ED patients with acute (≤7 days) new onset or exacerbations of chronic pain in the neck, back, arms and/or legs, deemed musculoskeletal by ED clinicians, were included. 23Exclusion criteria included: (1) no pain at triage; (2) contraindication to needles at acupuncture sites (e.g.skin infection); (3) unable to attend clinic; or (4) serious medical condition (e.g., active COVID-19 infection).

Randomization and Blinding
Subjects were randomized 1:1:1 in Stage 1 and 2:2:1 in Stage 2 to AA, PA or no acupuncture (NA) using a computer-generated unstratified block randomization sequence stored in a secure electronic file, with sequence electronically hidden and visible only at randomization by the acupuncturists.While participants and acupuncturists were not blinded, reasonable attempts were made to blind all other research and clinical personnel, described in detail elsewhere. 21

Interventions
Two different acupuncture interventions were performed by licensed acupuncturists to treat pain in the ED, described in detail elsewhere. 21(1) Auricular acupuncture (AA) places press needles in up to 5 bilateral ear sites corresponding to the battlefield acupuncture protocol. 16,17(2) Peripheral acupuncture (PA) places needles in select head, neck, and extremity sites distal to the pain based on acupuncturist clinical discretion. 26,27Acupuncture was delivered in the ED, then offered free in an outpatient clinic twice a week for one month after ED visit.During clinic, participants could receive PA, AA or both based on acupuncturist discretion and/or participant request.(3) The no acupuncture (NA) control group did not receive acupuncture in the ED and were asked to refrain from receiving acupuncture for one month after ED visit.As an attention control, all control group participants had the same initial evaluation and pre-treatment interaction with study acupuncturists prior to randomization as the two acupuncture groups.All study participants in all three groups also received usual care (e.g., pain medications) at the discretion of their blinded ED clinical team.

Data Collection and Management
Data was collected from participants in all arms at ED baseline, 1 hour post-treatment, and at 2 weeks and 1 month post-ED visit in a secure REDCap database, by iPad in ED and by email, text, or phone after the ED visit. 28Acupuncture treatment details based on STRICTA recommendations, 29 ED medications, and AEs were recorded by study personnel in REDCap.

Outcome Measures
The primary endpoints for Stage 1 were feasibility, acceptability, and safety.Feasibility was assessed by patient recruitment and retention rates.Acceptability was assessed by patientreported satisfaction.Safety was evaluated by adverse events (AEs), most commonly bleeding, bruising, or pain at needle sites. 26,30The primary endpoint for Stage 2 was change in current pain score (0-10 numeric rating scale, NRS) from ED baseline to 1-hour post-treatment.
Secondary measures included satisfaction with treatment (1-5 Likert scale) and medication use, including opioid and non-opioid medications, assessed by patient report and electronic medical record (EMR) data.Additional patient self-reported data included demographics, pain characteristics, non-medical opioid use using ASSIST, 31 the simplified graded chronic pain scale, 18,19 as well as pain interference, sleep disturbance, and physical function using the validated PROMIS-29 instruments. 14The question timeframe was modified from "over the past 7 days" to "over the past 24 hours (1 day)" for ED baseline assessments due to the acute duration (≤7 days) of pain.

Sample size
Sample size of 220 total subjects for Phase 1 was calculated using 90% power and α = 5%, based on a 2-stage adaptive design with 90 subjects allocated 1:1:1 to three arms (AA, PA, NA) in Stage 1, and the remaining 130 subjects to two arms assuming 1:2 control:treatment allocation for Stage 2, 23,32,33 adjusted for one planned interim analysis using O'Brien-Fleming type of alpha spending function 22 and a 10% drop rate, described in detail elsewhere. 21

Data Analysis and Statistical Methods
Primary analysis was performed based on intention-to-treat (ITT) including all randomized subjects combined from both stages with at least one follow-up evaluation.Descriptive statistics for continuous variables included number of subjects, means and standard deviations, medians and interquartile ranges, and minima and maxima.Descriptive statistics for discrete (categorical) variables included number and percentage of subjects in each category.
Because usual care for pain in this ED site starts immediately after triage, an exploratory analysis to assess within-group differences in response to usual care vs acupuncture examined differences in pain scores over time from triage to study baseline to 1 hour post-treatment across treatment groups using ANOVA, with the outcome being the difference-of-differences [(one-hour -baseline) -(baseline -triage)].All subjects with pain scores at one or more preand post-treatment timepoints were included; missing pain scores were not imputed.
All analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC).Unless otherwise noted, tests of hypotheses were two-sided with 5% level of significance.
Handling of Missing Data: For the primary analysis, 10 patients were missing change in 1-hour pain score, due to 1 missing baseline (Table 2) and 9 missing 1-hour pain scores (Table 3).A multiple linear regression model using baseline pain, sex, race, age, and duration of pain as predictor variables was fit to impute these missing change in pain scores.A quadratic linear regression model was fit using triage pain score to predict the single missing baseline pain score for use in the imputation model.Percentage change in pain at 1-hour also used these 10 imputed values.Additionally, two patients indicated a baseline 'current pain' score of 0 despite reporting pain at triage (Table 2), so a baseline pain score of 1 was used in the percentage change analysis.For secondary outcomes and repeated measures analysis, missing values were not imputed.
Analyses: As there were no significant differences in baseline characteristics between groups, the primary outcome was compared using unadjusted ANOVA.Secondary outcomes were compared using separate unadjusted ANOVAs for continuous variables, Kruskal-Wallis for ranked (Likert-scale) variables, and chi-squared for categorical variables.AEs were tabulated, ordered by frequency, and summarized by seriousness, severity, and possible association with acupuncture.Incidence rates of AEs were compared by Fisher's Exact Test.
Interim Analysis: One interim analysis was planned at the end of Stage 1 to assess feasibility based on patient recruitment and retention rates, safety based on AEs, and probability of being the more effective treatment arm analyzed from the 1-hour change in pain scores. 23An independent DSMC (biostatistician, emergency physician-researcher and medical acupuncturist) monitored trial safety and performance and recommended adaptations based on interim analysis.

RESULTS
At interim analysis, ED acupuncture was deemed feasible and safe by the DSMC.One acupuncture arm was not clearly superior, and both trended toward superior to control.Therefore, all three study arms were continued in Stage 2 with new allocation ratio 2:2:1 AA:PA:NA to determine effectiveness of AA and PA compared to NA.The study ended when the target sample size was achieved.
From February 10, 2020 to May 3, 2021, 911 patients were screened, 491 met inclusion criteria and were approached, 247 were enrolled, 11 withdrew prior to randomization, and 236 patients were randomized to one of three arms (68 NA, 84 AA, 84 PA, Figure 1).All subjects in the ITT population completed the study, and there were no protocol violations.The study population consisted of broad demographic characteristics representative of the typical ED population seeking care for musculoskeletal pain, 4 and were similar across the AA, PA and control arms (Table 1).Enrolled subjects had a mean age of 46.1 years (±16.5 SD, range: 19 -85).The most common self-identified race was Black (53.6%), and 7.2% identified as Hispanic or Latino.In addition, 56.6% of subjects were employed either full-time or part-time, whereas 21.7% reported being unemployed or unable to work.
Baseline pain and clinical characteristics were comparable across AA, PA, and control groups (Table 2).Overall, the most common primary pain locations were lower back (36.9%),legs (26.7%), and neck (14.4%).Most subjects (66.9%) reported having pain in more than one location, and 56.0% reported that their current painful condition was due to trauma or injury.The majority of subjects reported having at least some pain in the past 3 months (79.3%), and that pain had limited their life or work activities (66.5%).All three arms had similar ED baseline pain scores (AA 7.0 ± 2.3, PA 7.2 ± 2.3, NA 6.9 ± 2.2).Only 18% of study participants had ever tried acupuncture before.There were no differences in pre-baseline ED administration of opioid and non-opioid analgesics between the groups (Table 2).
As shown in Table 3, greater pain reductions at 1-hour post-treatment in the ED were observed with both AA (2.1 ± 2.9) and PA (1.6 ± 2.0) compared to NA (0.4 ± 1.9).Although not significant, there were fewer opioids ordered in the ED or prescribed on ED discharge among the acupuncture groups compared to control at ED discharge (Table 3).Participants in both the AA and PA groups reported comparably high satisfaction with their overall acupuncture experience at 1-hr with a mean score of 4.4 (±0.9) on a 1-5 Likert scale (Table 3).
Table 3 shows the mean pain score at triage, baseline and 1-hr post-treatment.Changes in pain score from triage to baseline were comparable across groups (-0.8 ± 1.9), indicating a similar effect of any usual care triage medications given.The AA and PA groups had significant decreases in pain at 1-hour post-treatment while the NA group did not, and the AA and PA pain scores were both significantly different from NA at 1 hour.For the exploratory analysis, the AA group showed significantly better within-group improvements after acupuncture than after usual care alone, while the PA group showed a similar trend that was not significant.Participants in both acupuncture groups also received fewer opioid and non-opioid analgesics post-baseline and at discharge than control, although these findings were not statistically significant.
Overall, there were few adverse events (AEs) and no serious adverse events (SAEs) reported (eTable).There was no significant difference in the number of subjects who reported AEs in the control (n=2, 2.9%), AA (n=3, 3.6%), and PA (n=0, 0.0%) groups in the ED.The most common AEs were transient pain, bleeding or bruising at needle sites, self-limited headache or brief episodes of anxiety.

Improved acute musculoskeletal pain management:
Acute musculoskeletal pain is a common complaint in the ED and challenging to treat, in part due to its severity. 34The high baseline pain scores in our study population support that observation.Effective management of acute pain is critically important to mitigate associated morbidity and disability; however, the current reliance on opioid medications presents substantial risks. 9,35Previous studies of acupuncture in the ED have shown greater improvements than sham acupuncture and similar benefits as medications for treating acute pain, 12,20,36 but have been limited by small sample sizes. 37Our study builds upon prior work with a uniquely large and diverse population of ED patients in a randomized controlled trial design.
Our study shows greater and clinically significant improvements in pain scores 1-hour after acupuncture compared to usual care alone.We found that two different acupuncture interventions were similarly effective, allowing for increased flexibility in terms of both patient preference and clinical implementation.
While not significant, there were fewer ED opioid administrations and discharge prescriptions after acupuncture.In addition, within-group changes in pain score were significantly better after auricular acupuncture + usual care than after usual care medications alone.9][40] One study, comparing acupuncture and intravenous morphine for acute pain in an ED setting, found that acupuncture was more likely to cause significant reduction in pain (≥50%), and faster, than morphine alone. 39nother study comparing acupuncture patients with those using non-steroidal anti-inflammatory drugs (NSAIDs) and physical therapy found that acupuncture patients were prescribed fewer opioids and had fewer ED visits. 40Thus, acupuncture may be an important treatment option for reducing opioid prescribing and subsequent use.

Delivery of acupuncture to a broad population in the ED environment:
Ours is one of the first randomized controlled trials of acupuncture to intentionally and successfully enroll a large number of underserved and minoritized groups in the U.S. [41][42][43] By minimizing the number of exclusion criteria that have historically excluded these populations, 44 and systematically approaching all potentially qualifying patients for the study, our study population is reflective of our general ED population. 45More than 50% of participants selfidentified as Black and 7% as Latino, more than half reported low income, and over half had public or no health insurance.These rates are higher than U.S. national averages reporting 36.1% with income under $50,000 46 and 8.3% with no health insurance. 47Furthermore, we successfully adapted acupuncture to a fast-paced, relatively chaotic ED environment by keeping treatments to 20 to 30 minutes and focusing on pain relief and needling of sites distal to the pain. 48,49Our ability to recruit and perform acupuncture on 236 ED patients over a one-year period demonstrates feasibility.Participants in both acupuncture intervention arms reported high patient satisfaction and minimal side effects, demonstrating broad acceptability across diverse populations.Our findings underscore those from recent work developing community acupuncture clinics for underserved populations reporting high participant interest in and satisfaction with acupuncture treatments, 17,42,50,51 supporting implementation of acupuncture more broadly to a wide range of patients.

Limitations
Limitations of this study include a single urban U.S. ED which may limit generalizability to other environments, such as rural ED's and other nations.However, studies from other countries have shown similar efficacy of ED acupuncture for treating pain, and ours adds to this body of literature as one of the largest U.S.-based studies.Second, treatment delivery by licensed acupuncturists may limit comparison to other ED-based trial protocols that trained other providers, like physicians, or environments with no available acupuncturists.Third, we did not include an active comparator or placebo group.However, all control group participants received the same initial contact and evaluation by study acupuncturists as the acupuncture groups to account for extra attention and service time effects.While every attempt was made to blind ED providers and outcomes assessors, patients could not be blinded to treatment assignment.Future research should include multi-site RCT's with varied ED settings to further evaluate acupuncture's efficacy across patient groups and practice environments.

Conclusions
These results indicate that both auricular and peripheral acupuncture are effective, feasible, and acceptable within the emergency department for musculoskeletal pain and should be further explored for more widespread implementation to improve ED pain management.

Table 1 : Patient Characteristics
1 Subjects were able to report more than one insurance type.

In the past 3 months, how often did pain limit your life or work activities?
1ASSIST scores are only reported among those answering Yes to non-medical opioid use in the ASSIST questionnaire.

Table 3 : Pain Outcomes at 1 Hour
1mputed for 1-hour change score calculations.1p-valuefrom overall ANOVA test.P-values for pairwise comparisons, adjusting for multiple comparisons using Patient-reported AEs in the ED included transient needle site pain or irritation (n=3), and non-acupuncture related episodes of fainting related to blood draw (n=1) and nausea related to opioid medication (n=1), respectively, as part of usual ED care.Some patients reported more than one adverse event.