Effectiveness of workplace active rest programme on low back pain in office workers: a stepped-wedge cluster randomised controlled trial ======================================================================================================================================== * Yamato Tsuboi * Tomohiro Oka * Kiyomasa Nakatsuka * Tsunenori Isa * Rei Ono ## Abstract **Objective** This study aimed to investigate the effectiveness of workplace active rest programme (WARP) on chronic low back pain (LBP) among office workers. **Design** A closed cohort, stepped-wedge cluster randomised trial was conducted. The total duration of the study was 16 weeks (4 weeks for each step). Sequence allocation was randomised, but no one was blinded. **Setting** This study was conducted in three offices in a Japanese electronics company. One office was for the administrative department, the others are for the engineering department. **Participants** We recruited 29 office workers with LBP greater than 3 months. LBP due to specific injury or disease was excluded. The median age was 38 years, and 26 (90%) were male. All participants completed the study. **Interventions** In the intervention phase, participants performed WARP comprising frequent stand-up and individualised brief exercise/physical activity during work. Physical therapists held an LBP workshop and developed tailor-made programmes before introducing WARP. We instructed participants to perform WARP at five timings during work. Control phase was set before the intervention and participants stayed as usual. **Primary and secondary outcome measures** The primary outcome was pain intensity of LBP assessed using the Brief Pain Inventory. The secondary outcomes were work productivity loss measured using the Work Limitations Questionnaire, LBP disability assessed using the Roland-Morris Disability Questionnaire, psychosocial subscale assessed using the STarT Back Screening Tool and physical activity measured using triaxial accelerometers. These outcomes were collected at baseline and at 4-month follow-up evaluation. **Results** In the intention-to-treat analysis, WARP did not show any significant effects on pain intensity (*β*, 0.01; 95% CI −0.50 to 0.52) and on the secondary outcomes. The median adherence to WARP was 28.6% (IQR, 16.8–41.1), which was equal to 1.43 times per day. No adverse effect was observed. **Conclusions** The present study was unable to confirm the effectiveness of active rest in improving LBP. Hence, further study needs to investigate its effectiveness. **Trial registration number** UMIN000033210. * pain management * occupational & industrial medicine * back pain ### Strengths and limitations of this study * This study is the first pragmatic trial conducted in a real-world setting that investigates the feasibility and effectiveness of active rest. * All participants completed the workplace active rest programme (WARP). * Adherence to WARP was lower than expected. * Because recruited office workers had relatively mild low back pain, we were unable to confirm whether WARP is effective among office workers with severe low back pain. ## Introduction Low back pain (LBP) is a prevalent health problem among office workers1 2 and is the leading cause of decreasing healthy life expectancy worldwide.3 Moreover, LBP results in a large socioeconomic burden due to work productivity loss and medical expenses.4 5 In terms of both individual and social impact, LBP among office workers is a crucial problem which should be tackled. Office workers are workers who stay in prolonged sitting position during most of their working time.6 7 Prolonged sitting is one of the causes of LBP, which is also due to several factors such as increased disc pressure,8 decreased trunk mobility9 and less posture variation.10 Although previous studies have investigated the effectiveness of ergonomic intervention and back support, these are considered ineffective in improving LBP.11 12 Recently, the use of standing desk has been shown to be effective in improving LBP,13 but it has the following limitations: it requires a lot of space and is costly. Therefore, easy-to-use solutions are required in the workplace. Active rest (taking a break with exercise/physical activity in the workplace) could possibly improve LBP because it has the following characteristics: (1) sedentary break by standing up, which can prevent prolonged sitting; and (2) exercise/physical activity, which is recommended in the LBP guidelines.14 15 A previous study showed that office-based stretching (10–15 min/session, 3 times/week) was effective in reducing the occurrence of musculoskeletal discomfort when compared with no intervention.16 However, in our study, we developed a shorter exercise programme involving frequent sessions (a few minutes per session, five times per day, except on weekends) because we aimed to promote frequent standing to break the habit of prolonged sitting. Although a positive effect of active rest on LBP has been shown in a laboratory study,17 its effectiveness in the real-world setting is still unknown. We hypothesised that there is a difference in effectiveness between laboratory and real-world setting. Thus, the present study aimed to investigate the effectiveness of the workplace active rest programme (WARP) on chronic LBP and work productivity loss among office workers in a real-world setting. ## Methods ### Study design The present study was conducted according to the extension of the Consolidated Standards of Reporting Trials 2010 statement for stepped-wedge cluster randomised controlled trial (SW-CRT).18 We used a closed cohort SW-CRT involving randomisation of clusters to different sequences. SW-CRT is a crossover design with repeated measurement, in which clusters switch from control to intervention condition. SW-CRT is a suitable study design if we assume that the intervention will do more good than harm, hence making it unethical to withhold the intervention from a control group. Thus, because it is morally acceptable and beneficial for participant recruitment, we introduced the SW-CRT design.19 Moreover, this is the pragmatic design, which increases statistical power and decreases needed clusters compared with those in parallel CRT.20 The present clinical trial was registered with University Hospital Medical Information Network (UMIN) Clinical Trials Registry. As figure 1 shows, we conducted the present study in three offices (clusters) in a Japanese electronics company. We set three sequences, where an office switched from the control condition to the intervention condition one by one. The total duration of the study was 16 weeks (4 weeks for each step). Evaluation was conducted at baseline and at four points during the last week of each step. Due to the study’s closed cohort design, participants assessed in different periods were the same participants. ![Figure 1](http://bmjopen.bmj.com/https://bmjopen.bmj.com/content/bmjopen/11/6/e040101/F1.medium.gif) [Figure 1](http://bmjopen.bmj.com/content/11/6/e040101/F1) Figure 1 Diagram of stepped-wedge cluster randomised controlled trial design. ### Patient and public involvement Office workers with LBP were not involved in developing the research question, but we consulted them about the design of the study (especially the intervention programme) in terms of feasibility and applicability by joining the employees’ health committee. During the trial, they helped us to hold an LBP workshop by arranging a room and equipment. We asked them to assess the burden of the intervention before they joined the study. We already disseminated the results of our study to the participants and reported them at the employees’ health committee. ### Participants’ recruitment We recruited 29 participants from three offices of a Japanese electronics company in July 2018. Three offices were separated from one another. First, participants were approached by the public health nurse working in this company. When they expressed interest in the study, the public health nurse introduced them to us. Subsequently, the researchers explained the study to the participants, and the participants provided informed consent for inclusion in the study. Office workers were eligible for the present study if they have the following characteristics: (1) full-time workers (all workers worked in the same day shifts), (2) engaged in desk work greater than 4 hours/daily working time (self-reported)21 and (3) had LBP for greater than 3 months. The location of LBP was defined as pain between the 12th rib and the inferior gluteal folds.22 Exclusion criteria were as follows: (1) LBP caused by fracture and trauma injuries, infectious diseases and internal organ disorders and (2) difficulty participating in the study due to medical or surgical disease. Cluster-level eligibility criteria were as follows: (1) an office where most workers were engaged in desk work and (2) supervisors granting permission in the performance of the study. Whereas office A was for an administrative department, office B and office C were for engineering department. All participants provided written informed consent for inclusion in the study. ### Randomisation and blinding Offices were randomly assigned to one of the three successive sequences (one office per sequence) after all clusters and participants were recruited. A researcher who was not involved in the recruitment performed random allocation using computer-generated random numbers and coded information about offices. To prevent contamination, both clusters and participants were not informed of the time the intervention started and the detailed programme of the intervention until 2 weeks before the intervention started. We also asked the participants exposed to the intervention not to disclose the programme content to other workers. Due to the nature of the present study, participants, intervenient and outcome assessors (self-reported) could not be blinded. The data analyst was also not blinded to group allocation. ### Intervention In the intervention phase, we offered WARP in two parts. First, we held the LBP workshop (group), followed by the introduction of active rest in the workplace. The LBP workshop was held when the group moved from the control phase to the intervention phase. The purposes of the LBP workshop were as follows: to allow the participants to understand LBP and sedentary behaviour, develop customised exercise programme and explain how WARP is performed after the workshop. The LBP workshop was held at the company’s gymnastics room after work for 90 min by two or three physical therapists (PTs) (PTs with expertise in LBP, at least 3 or more experience years) including the primary researcher (YT). To avoid inconsistency on workshop contents in PTs, we discussed and agreed with its contents before workshop. We disseminated leaflets about the contents of the LBP workshop to the participants. First, we gave lecture on the following: (1) LBP causes and interventions using a biopsychosocial model and (2) the impact of sedentary behaviour on health (death, non-communicable diseases and LBP). Second, evaluation was performed using a physical examination and an interview sheet (a brief file is described in online supplemental figure 1). We evaluated trunk flexion and extension (comfortable direction), static trunk posture (sagittal plane, lordosis/kyphosis), Thomas test (flexibility of the iliopsoas muscle),23 finger-floor distance test (spine and hip joint movement) and one-finger test (positive result indicates sacroiliac joint pain)24 and asked if the participants felt painful sensations when sitting or standing. Third, individualised exercise programmes were developed based on the results of the evaluation. Some exercises were recommended based on the results of the physical examination and interview sheets (online supplemental figures 1 and 2). We prepared six types of exercise focusing on spine and hip stretch and training, which can improve spine and hip joint mobility and decrease lumbar disc pressure (trunk extension exercise, stretching of the iliopsoas and hamstrings, abdominal oblique, erector spinae muscles, thoracolumbar fascia). We selected these exercises because these can be briefly performed by the participants when they stand up. We let them perform the recommended exercises during workshop after they had seen the demonstration. If participants had difficulty in performing the exercise, we individually helped them. ### Supplementary data [[bmjopen-2020-040101supp001.pdf]](pending:yes) At the end of the workshop, we explained to the participants how and when WARP is performed. Participants were instructed to perform WARP at five timings (just before the work starts, morning break, lunch break, afternoon break, after the work is finished). A chime ringed at these five timings, and we asked them to stand up and perform their exercises for a few minutes after the chime ringed. We also recommended performing WARP other than the five fixed timings. However, the participants were not required to perform the programme. We explained the content of WARP and introduced some brief exercises to other workers in the same office. It enables participants to easily perform exercise at the workplace because they understand what they do. Additionally, to determine if problems occurred after performing WARP, researchers visited each office once a month. ### Control When the participants were in the control phase, we did not perform any intervention to the participants (usual work). ### Primary outcome The primary outcome was LBP intensity. We used the pain intensity subscale of the Brief Pain Inventory (BPI), which is well validated and reliable among patients with non-cancer pain including LBP.25 26 BPI consists of four questions rating pain intensity separately at ‘worst’, ‘least’, ‘average’ and ‘now’ during the last 24 hours using 11-point Numerical Rating Scale (NRS), ranging from 0 (no pain) to 10 (worst imaginable pain). Finally, the mean of these four items was used as the BPI score: BPI score=(worst+least+average+now)/4. A Japanese version of BPI has good validity and reliability.27 At the time of trial registration, although we had planned to evaluate weekly LBP intensity, we changed to once in 4 weeks evaluation. This is because weekly evaluation was not feasible at this company in terms of responders’ burden in answering questionnaires. ### Secondary outcome The Roland-Morris Disability Questionnaire (RDQ) is a validated 24-item questionnaire that assesses disability due to LBP, such as ‘I change position frequently to try and get my back comfortable’.28 29 Each item is scored either 0 or 1, with all scores summed to a total between 0 and 24 (a higher score indicates greater disability level). The STarT Back Screening Tool is a validated screening tool that predicts future disability level.30 31 We used the five-item psychosocial subscale of the STarT Back Screening Tool, including fear of movement, depressive symptom, catastrophic attitude, anxiety and pain distress. Scores ranged from 0 to 5 (a higher score indicates higher possibility of future disability level). The Work Limitations Questionnaire (WLQ) is a validated 25-item questionnaire that evaluates work productivity loss due to physical/psychological issues.32 33 The WLQ is composed of the following four subscales: (1) time management (difficulty in performing job tasks in a timely manner and in scheduling tasks); (2) mental-interpersonal demands (difficulty in performing cognitive job tasks and in interacting with colleagues); (3) physical demands (ability to perform job tasks involving body strength, movement, endurance, coordination and flexibility); and (4) output demands (work quantity and quality reduction and timeliness of completed work). Additionally, ‘not applicable’ was also provided as a response option and treated as a missing value. All subscales scores were converted to percentage, from 0% (least limited) to 100% (most limited). Work productivity loss (%) was calculated from the weighed sum of the four subscale scores using a validated algorithm ranging from 0% to 24.9%. A higher score indicates higher level of work productivity loss. To measure physical activity and sedentary behaviour, we distributed triaxial accelerometers (Active style Pro HJA-750C, Omron Healthcare) to the participants during each step. Details of the accelerometer measurement procedure were described elsewhere.34 35 Participants were instructed to wear triaxial accelerometers on their waist during only working time for 5 days. Data were recorded in 60 s epoch. In addition to the number of steps, time spent in moderate-to-vigorous physical activity (MVPA, ≤3.0 metabolic equivalent; METs), light physical activity (1.5 < MET <3.0) and sedentary behaviour (MET ≤1.5) was calculated using R V.3.5.2. Days with at least 4 hours of wearing time or 75% of working hours were considered a valid day,36 and we included the data with at least one valid day in the analysis. Non-wear time was defined as a period with continuous zero count lasting over 60 min. ### Other measurements We collected demographic data such as age, sex, height, weight and body mass index. Participants were asked whether they were ever diagnosed with the following conditions: lumbar disc herniation, lumbar canal stenosis, lumbar compression fracture, trauma, spinal metastasis, fibromyalgia, rheumatoid arthritis and infectious spondylitis. Participants also reported the status of their analgesic administration (none, rarely, sometimes, often and always), consultation in orthopaedic clinics or complementary medicine for LBP (none, once, twice, three times, four times and greater than five), sleep quality (very good, fairly good, fairly bad and very bad), and other musculoskeletal pain including neck, shoulder, elbow, wrist, hip, knee and foot (NRS). At the final follow-up evaluation (T4 evaluation in figure 1), participants answered about their satisfaction (satisfied very much, satisfied, normal, dissatisfied, dissatisfied very much) and free opinion about WARP. ### Adherence To evaluate adherence to WARP, we asked participants to keep diaries on whether they performed WARP or not in each five timing. Adherence is calculated 100% if they performed WARP at all five timings during the whole intervention phase. Because WARP is a programme at the workplace, we did not include holidays when assessing adherence. ### Sample size We calculated the sample size using formula specific for stepped-wedge design.20 The primary outcome difference and SD were set as 2.0 and 2.5, respectively.37 The following assumed parameters were used: cluster size=10, intracluster correlation coefficient=0.05, number of step=3, number of baseline measurement=1, measurement after each step=1, two-sided α level=0.05 and *β*=80%. To detect a 2-point difference in primary outcome, a total of 22 participants were needed. Considering dropout, we estimated 30 participants as the required sample size, and 29 participants actually joined the present study. Although we set cluster size as 10 before recruitment, the actual size of the two clusters was 8. We conservatively performed sample size calculation by changing some parameters. However, the required sample size was not changed (22 participants) even if it is 8 participants. Therefore, this difference would not affect the results of our study. ### Statistical analysis For the characteristics of participants, categorical variables were presented as frequency and percentage and continuous variables as mean±SD. If the distributions of the continuous variables were skewed, data were presented as median (range or IQR). We performed both intention-to-treat (ITT) and per-protocol analyses to investigate both the effectiveness and the efficacy of WARP. The primary analysis was ITT analysis because this study aimed to investigate the pragmatic effectiveness of WARP in a real-world setting. Regarding ITT analysis, we performed linear mixed effect model for all outcomes, setting the intervention as the fixed effect, individual and office as the random effect, and calendar time as the confounding factor. Regarding per-protocol analysis, we also performed the linear mixed effect model for all outcomes after excluding participants whose adherence to WARP was 28.6% (median) or less. Unstandardised coefficients and 95% CIs were calculated. All statistical analyses were performed using Stata/IC V.15.1 software. P<0.05 was considered statistically significant. ## Results We recruited 29 office workers from three offices in July (figure 2). As planned, office A performed the intervention in the first period (August), office B in the second period (September) and office C in the third period (October). All participants continued WARP until the end (no dropout) of the study. Twenty-eight participants completed the baseline and each follow-up evaluation (T1–T4). Only one participant did not answer the T3 evaluation, but answered other evaluations. ![Figure 2](http://bmjopen.bmj.com/https://bmjopen.bmj.com/content/bmjopen/11/6/e040101/F2.medium.gif) [Figure 2](http://bmjopen.bmj.com/content/11/6/e040101/F2) Figure 2 Flow chart for stepped-wedge cluster randomised trial. The median age was 38 years, and 26 (90%) were male (table 1). The median pain intensity assessed using BPI was 2.0 (IQR, 0.8–2.2), and the median score on RDQ was 1.0 (0.0–2.0). Only two participants had clinic or alternative care, and only one participant often received analgesic medication. The median proportion of sedentary time was 79.6% (68.1–84.1). The median productivity loss estimated by WLQ was 2.2% (0.8–5.9). Regarding the difference in characteristics of the three offices, participants were younger in office C than in other offices. Pain intensity was lighter in office B than in other offices. View this table: [Table 1](http://bmjopen.bmj.com/content/11/6/e040101/T1) Table 1 Characteristics of the participants The median adherence to WARP was 28.6% (16.8–41.1), which is equal to 1.43 times per day (figure 3). Participants with higher adherence had relatively higher pain intensity, disability due to LBP and higher work productivity loss (online supplemental table 1) compared with those with lower adherence. Furthermore, low adherence was related to longer duration of WARP (adherence, office A