Study design and setting

This study was a 22-week, outcome assessor-blind, randomized, controlled, parallel trial from April 2020 to February 2022. The study protocol has been published previously [39] (Additional file 1). The ethical committee of the Iran University of Medical Sciences (IUMS) approved this study (IR.IUMS.REC.1398.1041). Participants signed a reviewed and approved informed consent document. We followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.

Participants

We sought to recruit participants from hospitals, advertisements, and social media in Tehran, Iran. All participants were examined by a specialist for eligibility. Inclusion criteria were as follows: 18–50 years old; having permanent or intermittent local pain between L1 and gluteal fold for three months or more and a Numeric Rating Scale (NRS) score of 3–7/10 in the last week; the Oswestry Disability Index (ODI) score of 20–60; the Tampa Scale for Kinesiophobia (TSK) score > 37; elementary level of education and native Persian speaking. The exclusion criteria were having a specific medical diagnosis (such as fracture, canal stenosis), rheumatoid disease, fibromyalgia, neuropathy, progressive neurological disease, headache, dizziness, nausea, epilepsy, migraines, and mental disorders; having a history of lumbar surgery in the past three years; Beck’s Anxiety Inventory score > 26; Beck’s Depression Inventory-II score > 29; participating in other therapies during the present research and pregnancy. Then they participated in an initial screening interview to provide an overview of the study. Informed written consent was obtained from all participants before the baseline examination. After completing baseline examinations, individuals who agreed to participate in the study were assigned to either the experimental or active control group. A total of 246 patients with NSCLBP presented to participate in the present study, and researchers excluded 176 participants based on exclusion criteria.

Randomization and blinding

After performing a baseline assessment, participants were randomly assigned (1:1) to an experimental or active control group. A computer-generated randomization sequence was performed using a stratified permuted block allocation (block size of 4). The stratification factor was gender. The outcome assessor who evaluated the outcome of the study was blind to the allocation of the groups. An analyzer independent of the research team and blinded to participant assignment monitored and analyzed the study data. In addition, the person who assigned participants was instructed not to disclose the status of the assignment to patients at any time. Details of the study protocols have been previously published.

Outcome measures

The primary outcome was pain, assessed by NRS at baseline, after six weeks of treatment, and after 10 and 22 weeks of follow-up. The NRS uses an 11-point scale to measure pain intensity [40].

The secondary outcomes included pain cognitions, disability, quality of life, active lumbar forward flexion, and brain function. Brain function was measured using the EEG power spectra analysis. EEG was recorded (bandwidth is 0-0.70 Hz, the pass filter is 0.05–60 Hz, and the sampling rate is 512 Hz) with a 64-channel amplifier (EB Neuro, Italy). The cap with 19 Ag–AgCl electrodes (impedance below 20 KOhm) was positioned according to the international 10–20 system at Fp1, Fp2, F7, F3, Fz, F4, F8, T3, C3, Cz, C4, T4, T5, P3, Pz, P4, T6, O1, O2. We placed references and ground electrodes on the mastoid processes (A1 and A2) and the Fpz region. EEG signals were recorded under two conditions: three minutes of an eyes-open resting state and 20 s of active lumbar forward flexion. The absolute and relative (percentage of total EEG power) EEG power spectra analyses of the following frequency domains were calculated: delta (1–3.8 Hz), theta (4.0–7.8 Hz), alpha (8.0–13.8 Hz), beta (14.0–34.8 Hz), and gamma (35–50 Hz).

The following questionnaires were used to assess pain cognitions: The Pain Catastrophizing Scale [41]; the Tampa Scale for Kinesiophobia [42]; and the Fear Avoidance Beliefs Questionnaire [43]; Quality of life and disability were measured using the 36 Health Status Survey (SF-36) [44] and the Oswestry Disability Index Questionnaire [45].

Study group interventions

Participants in each group received 12 treatment sessions delivered twice weekly for 6 weeks from their trained independent physiotherapists. The treatment in each session lasted equally long in the two groups.

Experimental group intervention

Participants in the experimental group received multidimensional physiotherapy. This treatment included psychoeducation, graded exposure, postural correction and electrotherapy as a passive warm-up. In this study, we performed twelve 30-minute psychoeducation sessions based on cognitive behavioral therapy (CBT). Each session was conducted using the question-and-answer method. The physiotherapist first provided the desired content, then posed a question regarding the covered topic to ensure the patients’ attendance and active involvement in each session. At the end of each session, the patient was given a task to do at home and present in the next session. Out of the 12 sessions, three were dedicated to educating the patient on pain neurophysiology and achieving therapeutic goals. The contents of these sections were designed based on a summary of the book Explain Pain [46]. The other sessions consisted of anxiety management, interpersonal conflict management, problem-solving training, coping strategy training, pain flare-ups management, medication abuse management, enhancing the ability to cope with labeling and stigma, empowering one to create a daily sleep routine, and training relaxation techniques [47]. Each topic is discussed in one session. Table 1 shows the details of psychoeducation treatment in the experimental group. Participants also received recommendations for postural correction, and gradual initiation of movements and activities that are feared. We asked participants to repeat the practice of each session at home once a day. Electrotherapy, including 20 min of transcutaneous electrical nerve stimulation (TENS) at 100 Hz frequency in the lumbar region with a hot pack and ultrasound with a frequency of 1 MHz and intensity of 1.2 W/cm2, was performed on the paravertebral muscle for 5 min.

Table 1 Description of psychoeducation sessionsActive control group intervention

Participants in the active control group received usual physiotherapy including traditional back school education (three sessions), general trunk exercises, and electrotherapy as a passive warm-up. Back school education covered basic anatomy and biomechanics of the spine, common causes of spinal pain, nociceptive pain processing, and ergonomic counseling based on the inherent postural strain associated with various postures and daily activities (including standing, sitting, and lifting). As such, the education sessions prepared the patients for a symptom-contingent, biomedical approach to daily activity and exercise therapy. In session four, general exercise therapy was started with a specific emphasis on treating dysfunctional muscles and joints. Different therapeutic goals were pursued (microcirculation, mobility, endurance, strength of the abdominal and paraspinal muscles) depending on what emerges from the clinical reasoning as the most dominant peripheral dysfunction. The progressive exercise program mainly entailed an increase in exercise intensity, and an evolution towards functional activities and more physically demanding tasks while keeping the spine in physiologically neutral positions to minimize strain imposed upon the spinal structures. All exercises were performed in a symptom-contingent way (Additional file 3). They also received electrotherapy, the same as the experimental group.

Training of the physiotherapists and treatment fidelity

The two independent physiotherapists provided interventions in each group. The physiotherapist who conducted treatment in the active control group had ten-year experience in musculoskeletal treatment. The specific training for CBT-based psychoeducation was developed to enhance the standardization. The training involved (1) participating in learning sessions and interactive classes with an expert psychologic in the field of psychoeducation on pain, as well as clinical workshops where the management of NSCLBP as a multidimensional disorder was discussed (2) assessing and treating actual patients with NSCLBP (3) assessing and treating patients with NSCLBP in front of an expert.

A fidelity evaluation was conducted in which the physiotherapists were observed while assessing and treating actual patients from the study. For each participant, session-by-session documentation of treatment content was recorded by the physiotherapists. Regular training, supervision, and feedback were provided to the physiotherapists throughout the study to facilitate the successful delivery of both treatments. Physiotherapists provided demographic details and information about their training at each session. For qualitative evaluation, some sessions of both intervention groups were observed and audio recorded.

Sample size calculation

The sample size was calculated using G*Power software 3.1.9.4 (Düsseldorf, Germany) based on the effects on pain in the pilot study (partial η2 = 0.04, α = 0.025, power = 0.95) and accounted for F tests and 10% loss to follow-up after 22 weeks, resulting in a total sample size of 70 individuals [48].

Statistical analysis

Stata 14.2 (StataCorp LLC; College Station, TX, USA) and SPSS (version 22) were used to analyze all data. We performed an intention-to-treat analysis for participants who dropped out of the study at follow-up. Continuous and categorical baseline variables were summarized using mean (standard deviation), median (interquartile range), and frequency (percentages) to determine descriptive statistics. A mixed model ANOVA/ANCOVA was used to determine the time effect, group effect, and time × group interaction effect between groups, using the Bonferroni post hoc method and adjusting for differences in patients’ characteristics at baseline where appropriate were done. The assumption of homogeneity and sphericity were checked by Levene’s and Mauchly’s tests respectively. When the assumption of sphericity is violated, Greenhouse-Geisser corrections are used. Similar analyses examined the treatment effect for all secondary outcomes. Subgroup analyses were planned for the pain catastrophizing variable (mild, moderate, and severe pain catastrophizing). We also calculated the effect size (Cohen’s d) for each pairwise comparison, using the pooled SD of baseline scores, where 0.2 was considered a small effect, 0.5 a moderate effect, and 0.8 a large effect [49]. Statistical significance was indicated at p ≤ 0:05 (2-sided), and the confidence interval was set at 95%. The statistician was blinded to group allocation.

Deviations from the registered trial protocol

We made some deviations from our protocol. In calculating the follow-up period of one and four months, we considered the end of the treatment period, which was finally corrected 10- and 22 weeks after randomization. In the design phase of the trial, we designed a three-blind study, but in the pilot study, it was not possible to implement the first phase of blinding, so we conducted the study as an outcome assessor blind study.