An assessor-blinded randomized controlled trial was approved by the Ethics Committee of Khon Kaen University, Thailand (HE 612259) and the University of Public Health, Myanmar. The study proposal followed the CONSORT checklist and was registered at clinicaltrials.in.th (registration number: TCTR20180822001; August 21, 2018). It was conducted at the Physiotherapy Department, Yangon Orthopedic Hospital, Myanmar between December 2018 and April 2019. The intentions and processes of the study were explained to the eligible participants, and they were asked to sign an informed consent form before their participation.
Participants
The eligibility criteria of participants for the current study included: age 20–50 years old, subacute NSLBP (6–12 weeks), no radiating leg pain, with moderate pain (VAS: 3–7) and a disability score of 19% or greater as evident from the modified Oswestry Disability Questionnaire (MODQ).
The postural control system can also be affected by aging, (decline in muscle strength, sensory functioning, or in speed of sensorimotor responses) and deteriorates from the age of 50 onwards [42, 43]. Balance may decline beyond 50 years of age due to the aging process, so, the upper age of participants in this study was limited to 50 years old. In the present study, patients with a pain score over 3/10 and with no more than 7/10 pain were selected because the outcome measures used in our study may not be suitable for patients with severe pain, especially the task of single leg standing [44].
Participants were excluded if they had: neuromuscular disorders, spine or other joint deformity, recent lower extremity injuries (within 6 months), brain injury, pregnancy, body mass index > 30, conditions that affect balance (drugs, alcohol consumption, visual and vestibular disorders), or compromised exercise performance (hypertension, ischemic heart disease, diabetes mellitus).
Sample size was calculated in relation to the primary outcome of the study: “joint repositioning error”, with the mean difference of (μ1 – μ2 = 0.35 points) and a pooled variance estimation (σ2 = 0.15) between the CSE and STE groups from the pilot study with 80% power and significance level set at α = 0.05. An initial sample size of 13 in each group was required by calculation using the formula: n = (zα/2 + zβ)2 × σ2/(μ1 – μ2)2, where, we set zα/2 = 1.96 and zβ = 1.28. Allowing for a dropout rate of 15%, at least 36 participants (18 in each group) were recruited in this study. All eligible participants were randomly assigned to either the CSE or STE group (allocation ratio 1:1).
Baseline assessment of the outcome variables was performed by one physical therapist with 20 years of work experience and one Senior Consultant Radiologist after gaining patient consent. Another physical therapist with 7 years of working experience assigned the participants to either the CSE group or the STE group using a random allocation with a block size of six. Pre-generated random assignment schemes were enclosed in sealed envelopes and each participant chose their preferred envelope. Participants were asked not to participate in other treatment during the study period.
Intervention procedure
Participants completed the exercise program for 30-min sessions, with three sessions per week for up to 4 weeks. In this study, 8–10-week CSE or STE programs were shortened to 4 weeks as our pilot study demonstrated that proprioception improvement occurs after 4 weeks of the exercise program. Each exercise was repeated 10 times, with 10 s holds, followed by a five-minute rest interval. The researcher (SSH) trained participants in the CSE group with the CSE program and the other experienced physical therapist trained participants in the STE group with the STE program. Participants in the CSE and STE groups completed either the CSE or STE training program. Participants were asked to perform their exercise routine as daily home exercises for 15 min. Other exercises were not permitted. Participants were asked to record details of their exercise practice on a log sheet, over the study period, to monitor their compliance. Additionally, the researcher made a phone call to all participants every week to encourage them to continue their home exercises during the study period.
Core stabilization exercise
Participants in the CSE group received CSE training using the treatment approach described by Puntumetakul et al. (2013) with the exercise period shortened from10 weeks to 4 weeks [45]. Details of the CSE program are described in Additional file 1: Appendix 1.
There were two stages of CSE. In the first stage (weeks 1–2), the treatment emphasis was on isolating low-load activation of the TrA and LM muscles with an abdominal drawing-in maneuver (ADIM) technique. Subsequent training involved co-contraction of these muscles in low load conditions. To activate TrA in the first week, the participants were positioned in prone lying on a bed with a small pillow placed under their ankles. A pressure biofeedback device (Chattanooga Australia Pty Ltd., Brisbane, QLD, Australia) was used to provide visual feedback and was set at 70 mmHg, it was placed under the participants’ lower abdomen. The participants were asked to draw their lower anterior abdominal wall “up-and-in” towards the spine. If they performed the exercise successfully, pressure was lowered 6 to 10 mmHg. Isolated activation of LM was stimulated by raising the contralateral arm when performing ADIM seated on a chair. Manual contact provided feedback for LM activation. In the second week, co-contraction of the TrA and LM muscles with controlled movements of lower extremities was performed in supine lying and sitting positions.
The second stage (weeks 3–4), emphasized increasing the accuracy and duration of exercise. Co-contraction of the TrA and LM muscles with controlled movements of upper and lower extremities was performed and progressed to high load positions. Co-contraction of the TrA and LM muscles was performed while sitting on a balance board and while lying supine in the third week. In the fourth week of the exercise program, co-contraction of the TrA and LM muscles was performed in the quadruped position and standing.
During exercise performance, participants were trained to self-monitor by palpating for the contraction of TrA and LM. For contraction of TrA, their index and middle fingers palpated the area 2 cm medial to their anterior superior iliac spines, and for LM, participants placed their index and middle fingers near their L5 spinous process. The accuracy of contractions of the TrA and LM muscles was re-evaluated at every exercise session by the researcher physiotherapist.
Strengthening exercise
Participants in the STE group underwent the STE program, which activated the back and abdominal muscles for extension and flexion, respectively. This program was adapted from that described by Koumantakis et al. (2005), with the exercise duration shortened from 8 to 4 weeks [32]. The program progressed based on individual performance, from lying to the quadruped position. The exercise program for the strengthening exercises is described in Additional file 1: Appendix 2.
In the first week, the participants were asked to train their upper abdominal muscles by performing partial sit-ups from the supine position with knees bent (crook lying) and to train their back extensors in the prone lying position by raising their trunk. In the second week, the participants were asked to undertake lower limb movement while training their upper abdominals and performing bridging from the supine lying position to train their back extensors.
The participants were asked to perform hip lifts in the side-lying position to train their oblique abdominal muscles and perform extension of one leg in the quadruped position to train back extensor muscles during the third week of the exercise program. In the 4 week, participants were asked to perform full abdominal crunches from the supine position and alternate arm and leg lifting from the quadruped position to train the back extensor muscles.
Outcome measures
Proprioception was the primary measure in this study, and balance, muscle thickness of TrA and LM, and pain-related outcomes were used as secondary measures. All outcome measures were evaluated at baseline and 4 weeks after the intervention by two assessors who did not know the participant group’s assignment and intervention. An experienced physical therapist assessed proprioception, balance, and pain-related outcomes. Researcher (EEK) performed Rehabilitative Ultrasound Imaging (RUSI) using standardized tools to assess TrA and LM muscle thickness.
Primary outcome
Proprioception
Proprioception was assessed with joint repositioning error as a primary outcome (Fig. 1). The method used followed the process described by Puntumetakul et al., 2018 [36]. “The participant was in the sitting position (90° hips and knees) with feet on the ground, hands on thighs and the examiner guided the participant into the neutral lumbar spine position. As the start point of measurement, the center of the 10-centimeter tape-measure was positioned on sacral segment 1 (S1). A laser pointer with a stable base was sited directly on the start point. The examiner instructed the participant to remember the target position, and then to perform the maximum anterior and posterior pelvic tilt twice, maintaining five seconds in each position,” and then returning to the neutral target position [36]. Deviance from the start point was measured in centimeters as a joint repositioning error. Feedback about any error was not given to the participant. Prior to the assessment, the participants practiced the repositioning test twice. This examination procedure was executed three times with 1 minute rest intervals. The mean values were used for analysis.
Repositioning error can be described by the absolute error, the constant error, and the variable error. We chose to calculate absolute error as it reflects accuracy [46], represents error magnitude [47], and is the most commonly used measure [10]. Inter-rater reliability for the proprioception test was excellent [ICC 3, 1: (0.92), (p < 0.001)] [48].
Secondary outcomes
Balance
The Romberg test was used to assess balance. Participants were asked to stand erect and were observed by the examiner for 1 minute to reduce ceiling effects. The participants were tested in four conditions: two each on a stable and an unstable surface. The conditions were single-leg standing with eyes open (SEOS, SEOUS) and with eyes closed (SECS, SECUS) [49]. The starting position for participants was standing on the dominant leg, as determined by the football kicking test [50], while the other leg was positioned with the hip in neutral, 90° knee flexion, and arms crossed at chest level. For each condition, the individual was asked to remain in this position and the standing time was recorded. This was repeated three times per condition, with a 30-s rest interval between each trial and a one-minute resting interval between either condition (eyes open or eyes closed).
The assessor monitored the participants carefully during the assessment. Postural retention was considered impossible when the arms moved, the stance foot moved, the lifted foot touched the floor, there was large body sway, eyes were open during eyes-closed trials, or there was a loss of balance that required the assessor to physically prevent a fall [51, 52]. Inter-rater reliability for the balance tests was excellent [ICC 3, 1: (0.92–1.00), (p < 0.001)] [48].
Rehabilitative ultrasound imaging
RUSI was used to evaluate the muscle thickness of the TrA and LM muscles. Images of the TrA and LM muscles were acquired with a Color Doppler Ultrasound machine (Ultrasonic/Canada50nl*OP) and a 5-MHz curvilinear/convex array transducer. Measurement was performed at the thickest part of the muscles and measured in the sagittal plane at the same location for each measurement time point. Image acquisition was performed three times in each measurement condition and the mean value was used for analysis.
The muscle thickness of TrA was measured in two conditions: at rest and during the ADIM (Fig. 2). The participants remained supine in a crook lying position. The transducer was placed along the midaxillary line just superior to the iliac crest, using the measurement protocol described by Teyhen et al. (2005) [53]. TrA thickness was measured between the superficial and deep borders of the muscle, made visible by the hyperechoic fascial lines. To control for the effect of respiration, all images were taken at the end of normal exhalation.
The muscle thickness of the LM muscle was measured in two conditions: at rest and during a submaximal contraction, according to the technique described by Kiesel et al. (2007) [54]. The participants remained in a prone lying position and pillows were placed under their hips to reduce lumbar lordosis. The spinous processes of L4/5 were palpated and marked with a pen prior to imaging. The subjects were instructed to relax the paraspinal musculature, electro-conductive gel was applied, and the transducer was placed transversely over the spinous process of L4/5. Measurement was taken between the most posterior portion of the L4/5 zygapophyseal joint and the inner border of the LM (Fig. 3).
The test-retest reliability of the RUSI measurement for muscle thickness of both sides of the TrA and LM muscles was tested by a senior consultant radiologist with 20 years of work experience (EEK) before the study in patients with subacute NSLBP. The results showed excellent reliability of muscle thickness measurements of TrA [ICC 3, 1: (0.96–0.99), (p < 0.001)], and LM [ICC 3, 1: (0.95–0.99), (p < 0.001)], respectively. Muscle thickness at rest, and the percentage of change [(contraction – rest)/ rest ˟ 100] of TrA and LM muscles were determined and analyzed.
Pain-related outcomes
The visual analogue scale (VAS) was used to assess pain [55], the MODQ was used to measure functional disability [56], and the Tampa Scale for Kinesiophobia (TSK) was used to evaluate fear of movement [57].
Statistical analysis of the study
For analyzing all the data in this study, STATA version 10.1 (Stata Corp, 4905 Lakeway Drive College Station, Texas 77,845, USA) was used. Demographic data are presented as means (standard deviation) and numbers (percentage). Descriptive statistics, independent sample t-test, and Chi-square tests were used to analyze participant characteristics. Normality of distribution for all data was tested by the Shapiro-Wilk test. The results of this study are presented as mean, standard deviation (SD), and 95% confidence interval (CI). Within-group comparisons on all data were analyzed using the paired t-test. Between-group comparisons were calculated using analysis of covariance (ANCOVA) for adjusting the baseline data. The effect size of the magnitude of standardized difference between groups was calculated using Cohen’s “d” = M1-M2/ SDpooled. p < 0.05 was considered a statistically significant level.