Comparative effectiveness of Chuna manual therapy versus conventional usual care for non-acute low back pain: a pilot randomized controlled trial

This research adhered to Consolidated Standards of Reporting Trials (CONSORT) guidelines [16]. The Chuna Research Network (CRN) comprised four Korean medical institutions (two university-based Korean medicine hospitals and two spine specialty hospitals) and several expert discussions were conducted to devise a pilot protocol for conducting a trial (once a month). This multicenter, pragmatic, randomized controlled pilot trial, with three parallel arms, was designed to explore the feasibility of a trial to evaluate the clinical efficacy, safety, and cost-effectiveness of using CMT in patients with non-acute LBP. The study protocol was registered with the Clinical Research Information Service (CRIS identifier KCT0001850, 12 March 2016). Additionally, the protocol was approved by the institutional review board (IRB) of Pusan National University Korean Medicine Hospital on 11 March 2016 (IRB approval number 2016002) and had already been published [17]. After screening, participants were randomized into three groups (CMT, UC, and CMT + UC group) by central allocation and treated for 6 weeks consecutively. Other additional treatments (e.g., medications related to pain, acupuncture, procedures, or surgery) not specified in the protocol were not allowed during the 6-week intervention period. Study monitoring was carried out by the Contract Research Organization (CRO), which had no role in the research design and practice at each site.

The study was conducted in four major Korean medicine hospitals in Korea (Pusan National University Korean Medicine Hospital, Kyung Hee University Korean Medicine Hospital at Gangdong, Jaseng Hospital of Korean Medicine, and Mokhuri Neck and Back Hospital) from 28 March 2016, to 19 September 2016. Patients aged 19–70 years, with non-acute LBP, were considered on the basis of eligibility criteria. Patients were included in the study only when they met the following criteria: (1) non-acute LBP (with pain duration of 3 weeks or longer) requiring medical attention; (2) average numeric rating scale (NRS) score of more than 5 during the previous week; (3) aged from 19 to 70 years, inclusive; and (4) agreed to trial participation and provided written informed consent. Patients were excluded when they (1) were diagnosed with serious pathologic condition(s) that might cause LBP (e.g., spinal metastasis from tumor(s), acute fracture, spinal dislocation); (2) had undergone spinal surgery within the past 3 months; (3) were diagnosed with other chronic disease(s) that might interfere with the treatment effect or interpretation of the outcome (e.g., chronic renal failure); (4) were diagnosed with a progressive neurological deficit or had severe neurological symptoms; (5) had an inner fixation or stabilization device mounted through spinal surgery; (6) were currently taking steroids, immunosuppressants, medicine for psychological problems. or other medication(s) that might interfere with the study results; (7) had received CMT or medicine that may influence pain levels, such as nonsteroidal anti-inflammatory drugs (NSAIDs), within the past week; (8) were pregnant or were planning to become pregnant; or (9) were participating in other clinical studies or were otherwise deemed unsuitable by the researchers.

The exact procedures and details of this study have been published in a pilot protocol [17]. Briefly, the participants were recruited through advertisements, posters on hospital bulletin boards, and referrals from Korean Medicine doctors (KMDs) in hospitals. Potential participants were asked to answer questions and were evaluated by KMDs or by the clinical research coordinator to determine eligibility. If patients were eligible for trial participation in accordance with the inclusion and exclusion criteria, they were randomized per center and allocated to one of the three groups using block randomization (block size 3). A random sequence was generated by an independent statistician using SAS 9.3 (SAS Institute Inc., Cary, NC, USA). The participants enrolled at the four sites were randomly allocated to groups without stratification by site. Due to the dissimilarity of the interventions, blinding of physicians and participants to allocation of treatment groups was impossible, by nature of the interventions. Only outcome assessors, the statistician, and data analysts were blinded and conducted the outcome assessment in a separate room after treatments were performed by separate physicians. The electronic data that did not contain participants’ information or participants’ allocation were transferred to the statistician and data analysts. All allocations were concealed as far as possible.

Participants assigned to the concurrent CMT and UC group received UC treatment in addition to CMT treatment. Treatments involved the same method, frequency, session length, total duration, and number of sessions as in the individual treatment groups.

For the primary outcome, we measured NRS scores of LBP levels for the previous week. NRS scores ranged from 0 to 10, with the higher number indicating greater pain intensity. The secondary outcome included NRS scores for leg pain, evaluating functional status by using the Korean version of the Oswestry Disability Index (ODI) questionnaire [22]. The ODI questionnaire was used to measure LBP-related disability. It was composed of 10 questions, including questions on daily life, pain intensity, personal care, lifting, walking, sitting, standing, sleeping, social life, and travelling. The Patient Global Impression of Change (PGIC) was one of the secondary outcomes, which assessed comprehensive and global change in LBP and movement limitation due to pain [23]. The PGIC consisted of 7-level answers, where lower numbers indicated lower treatment satisfaction. The EuroQol-5 dimensions (EQ-5D) health survey was also used to assess secondary outcomes. The EQ-5D is composed of 5 dimensions assessing the current health state, consisting of mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each dimension was evaluated by 3-level answers, with the lower score indicating the patient has a better state of health. Patients’ quality of life was assessed using the validated Korean version of the EQ-5D [24]. The Health Utility Index III (HUI-III), including sight, hearing, speaking, walking, agility, emotion, cognition, pain, and quality of life values was used to calculate participants’ quality of life as with the EQ-5D [25]. Lumbar range of motion (ROM) was also measured for assessing improvements objectively [26]. The maximum lumbar spine angle between a perpendicular line was measured on flexion, extension, lateral bending, and lateral rotation, using a goniometer. The angle was recorded as 0° if a patient complained that lumbar movement was impossible due to pain. A 9-point Likert-scale credibility and expectancy questionnaire was used to assess treatment expectation at the first visit. Cost data were also investigated in this study, but the results of these investigations will be reported in a separate paper. All participants were followed up at 1, 3, and 4 weeks after the 6-week treatment periods. At each visit the participant was assessed before treatment, to record the outcomes of the previous treatment session.

Data were summarized using descriptive statistics: frequency was calculated as percentage for categorical variables and mean ± standard deviation (SD) was calculated for continuous variables. Differences in study participants’ characteristics were compared across subgroups using the chi-square test or Fisher’s exact test for categorical variables and analysis of variance or the Kruskal-Wallis test for continuous variables, as appropriate. The paired t test, independent t test, or Wilcoxon’s signed rank test were also employed to assess the differences between assessment points or between two groups. Analysis of covariance was employed to reduce error from inequality at baseline, using the baseline value as a covariate. We used the Shapiro-Wilk test to check whether the data distribution was normal. Intention-to-treat (ITT) and per-protocol (PP) analyses were performed and the last observed carried forward (LOCF) method was used to impute missing values. All statistical analyses were carried out using SPSS Statistics for Windows 22.0 (IBM Corp. Armonk, NY, USA) statistical software. All tests were two-tailed at the 5% significance level.

The sample size calculation method for this study was published in the protocol [17]. A significance level of 5% (α error), type 2 error of 20% (β error), power of 80%, and 25% drop-out rate were applied to the formula shown in the protocol; the number of participants required in total was 60 (20 per group). All statistical analyses were performed blinded and independently by a statistician.

To monitor the safety of CMT, UC, and CMT + UC treatment, participants were asked about adverse events (AEs) at each visit. If AEs occurred, physicians rated the relationship between each treatment and the outcome on a 6-point scale (1 = definitely related; 2 = probably not related; 3 = possibly related; 4 = probably not related; 5 = definitely not related; and 6 = unknown) and categorized into 3 levels using the Spilker classification (mild, moderate, severe) [27, 28]. If serious adverse events (SAEs) occurred during the study, unblinding was considered allowable and the physician would inform the relevant IRB and main study site (Pusan National University Korean Medicine Hospital) to decide whether the trial should be continued or terminated. Participants suffering AEs would receive appropriate medical attention and damage compensation.