The effects of an 8-week stabilization exercise program on lumbar movement sense in patients with low back pain

Participants

Sample size for the reliability study was estimated in order to obtain a target reliability coefficient (ICC) of 0.80 (95 % confidence interval: 0.6–1.0). Based on five experimental trials over two individual testing sessions, a minimum of 28 participants was needed considering the abovementioned requirements [13]. The study conducted by Lee et al. [14] is the only one reporting a statistically significant difference between both groups with regard to MPT in axial rotation, and using 24 patients with LBP and 24 control subjects. Based on the study conducted by Lee et al. [14] the present study have been conservative with a sample size of 29 patients with LBP and 30 control subjects. All participants were aged between 18 and 65 years old, and were recruited through newspaper advertisements and from physiotherapy clinics in Montreal, Quebec, Canada. All participants spoke French and English. Patients with LBP had lumbar or lumbosacral pain, with or without radicular pain, for at least four weeks (non-acute phase), and a score above 12 % on the Oswestry Disability Index (ODI) [15] to allow a minimal important change of 10 % [16] to occur. Patients with non-acute LBP were targeted because exercise is not a primary intervention strategy for acute LBP [1]. General exclusion criteria were: surgery of the pelvis or spinal column; a specific lumbar pathology (fracture, infection or tumor) or scoliosis; systemic or degenerative disease; body mass index over 30 kg/m²; having begun an exercise program in the last three months; pregnancy and claustrophobia. Additional exclusion criteria for patients with LBP were: having one positive neurological sign in two of three test categories: (1) Achilles and patellar tendon reflexes; (2) Reduced strength in myotomes; (3) Reduced sensation in dermatomes; and litigation relative to the back injury. Exclusion criteria for the control subjects were back pain in the preceding year or a history of back pain lasting more than 1 week.

Before testing, each participant was informed of all experimental procedures and provided their informed written consent. All procedures were approved by the ethics committees of the Centre for Interdisciplinary Research in Rehabilitation of Greater Montreal (CRIR) (ethical registration number: CRIR-738-0512).

Lumbar stabilization program

An 8-week lumbar stabilization exercise program (two 30-min sessions/week) was provided to the patients with LBP in local physiotherapy clinics, without any co-intervention allowed. The patients were encouraged to do the exercises at home. Briefly, the exercise program focused on motor control of the deep trunk muscles [17], followed by gradual inclusion of overloading exercises designed to improve endurance and strength of the paraspinal and abdominal muscles [18] (further details in the Additional file 1).

Questionnaires

The ODI was used to assess LBP-related disability [15] while the 11-point (0 to 10) numeric pain rating scale (NPRS) was used to assess the current, best and worst levels of pain intensity during the last week, so as to average the three ratings [19].

Lumbar proprioception assessment

Lumbar motion sense, or motion perception threshold (MPT), was evaluated using a custom built apparatus similar to ones used in previous studies [14, 20] (Fig. 1). The MPT test measured the smallest amount of axial trunk rotation a participant could perceive. This measure was preferred over the measure of lumbar joint position sense because previous findings suggest that this test better discriminates between patients and control subjects [14], and produces more reliable data [20]. During the test, the lumbar spine is passively rotated in the transverse plane (trunk rotation) by rotating the lower body (seat). The seat is positioned on a 16-inches diameter high quality ball bearing (Silverthin Bearing Group, Preston, WA, USA; model SG160CPO) designed to minimize vibration. Movement is driven by a stepper motor (Applied Motion Products, Watsonville, CA, USA; model HT34-504) at a slow, steady rate to minimize tactile cueing. The resolution of the angular measurement was 0.09°.

Participants were positioned in the apparatus so that the L5/S1 joint aligned with the stepper motor shaft. The participant’s upper body was secured to the backrest with a 4-point seatbelt to minimize vestibular feedback. Starting from the neutral (zero) position, the seat was rotated, either clockwise or counter-clockwise, at a constant rate of 0.2°/s. As soon as motion was perceived, participants stopped the rotation by pressing a switch, and immediately stated the direction of movement. To remove trials where the subject would have guessed the motion perceptions, the trial was rejected if the direction perceived did not correspond to the true direction [14]. This was done in an attempt to reduce potential noise in the data. Participants were returned to the neutral position following each trial. All trials were performed with eyes closed and noise-canceling headphones.

Prior to testing, participants were given five familiarization trials (or more; until the task was properly understood), randomized for direction. This was followed by ten experimental trials also randomized for direction (five trials per direction).

All testing was done at the start of the study (session 1) and after 8 weeks (session 2).

Home-exercise adherence

Home-exercise adherence was assessed with one question at the end of the 8-week clinical program: How many times have you done your exercises as prescribed in the last week? The frequency per week was divided by the recommendation of the physiotherapist to obtain a ratio, in accordance with the most common definition of adherence, which is defined as “the extent to which the patient follows medical instructions” [21]. The ratio may vary between 0 and 1, one being given when the frequency was equal or higher than prescription. The measurement of adherence was not carried out during the course of the 8-week clinical program to avoid influencing positively the adherence behavior (e.g. desirability bias).

Data analysis

MPT scores were expressed as absolute values, in degrees. No significant differences between left and right rotations were detected, so the data were combined. Outliers (outside the 2-interquartile range) were removed from the 10 MPT scores, to reduce variability. The mean of MPT value, for each participant, was used for analysis.

Reliability

For the control group, the 2-way ANOVA detected a DAY effect (P < 0.001) (Fig. 2), but no significant TRIAL effect (P = 0.833) or interaction (P = 0.663). The lack of a TRIAL effect indicates that no within-session learning was present, allowing us to average data across trials to increase reliability.

For the between-groups comparison (above), seven, eight, nine or ten trials were averaged per session (representing 2, 8, 18 and 72 % of the testing sessions, respectively), after having rejected the outliers. Consequently, the data used in the between-groups comparison are likely to be moderately reliable.

Effects of the stabilization exercise program and between-group comparisons

All patients with LBP have undertaken the 8-week exercise program and reached the third (final) phase of the exercise program. Home-exercise adherence measured at post-treatment was high, with a mean ratio of 0.86 (SD 0.21). For MPT, the only significant finding was a main effect of DAY (P < 0.001) (Table 3), showing significantly smaller MPT values at the second testing session (Fig. 2). The lack of a GROUP × DAY interaction reveals that the change in MPT measures over the 8-week period was not significantly different between the two groups. Figure 3 shows MPT scores of each participant for both testing sessions. The variability of the scores and variability of effects (time/treatment) were similar between groups.

Variable	ANOVA results (P values)
	GROUP	SEX	DAY	G × S	G × D	S × D	G × S × D
	(G)	(S)	(D)
NPRS	/	0.641	<0.001	/	/	0.513	/
ODI	/	0.429	<0.001	/	/	0.547	/
MPT	0.239	0.509	<0.001	0.303	0.271	0.253	0.088

NPRS numeric pain rating scale, ODI Oswestry Disability Index, MPT motion perception threshold; “/” = not applicable because the corresponding ANOVA model did not include a GROUP factor

A significant effect of DAY was also found for the clinical measures (NPRS, ODI) in patients with LBP, showing a decrease of pain and disability after the stabilization program (Table 3; Figs. 4 and 5). No significant SEX × DAY or SEX effects were found.

Partial correlation analyses revealed no significant correlations between changes in MPT scores and changes in NPRS (r = −0.06; P = 0.744) or ODI (r = −0.10; P = 0.602).

8-week test-retest reliability

The present reliability findings are the first attempt to assess the reliability of MPT data over the duration of a rehabilitation program (8 weeks). Two previous studies have reported reliability results for MPT measures taken with a device similar to the one used in the present study [8, 20]. The reliability of these previous data were high (ICCs over 0.77; SEM: 0.34°), but only permitted a within-day reliability analysis. The lower reliability of our data are at least partly explained by an apparent long-term learning effect (main effect of DAY), which suggest the presence of motor learning. The familiarization trials that were used at the beginning of each day were apparently good enough to eliminate possible within-session learning (no TRIAL main effect), but insufficient to eliminate between-day learning, the latter being likely explained by the effect of sleep on neural plasticity [26]. This is worth mentioning that the subject needs to be well concentrated because the chair movement is very difficult to detect. Introducing a familiarization day before the first testing session (double baseline measurement) might help to minimize the long-term learning effect.

Effect of 8-week stabilization exercise program on MPT

To the authors’ knowledge, this is the first study to assess the effects of an 8-week stabilization exercise program on MPT. While the patients in the study did show significant improvements in MPT following the exercise program, their improvement was comparable to that of a healthy control group who received no intervention. This suggests that patients’ improvements in MPT may be more attributable to having learned the testing procedure than to the exercise program. Consequently, it is not surprising that we observed no correlation between the change of MPT and the change in the clinical outcomes (NPRS, ODI).

One explanation for the apparent lack of a MPT training effect is that the stabilization program used in this study is not specifically designed to improve lumbar proprioception. These exercises do, however, include some required components (coordination, muscle performance and balance training) to improve proprioception [27]. These results may also be related to the fact that the patients with LBP in this study did not show MPT impairments at the beginning of the program, relative to the control group. It is also possible that different findings would have emerged if we had used a measure of active joint position sense, rather than passive MPT. Lumbar stabilization exercises may provide a better training stimulus for proprioception in situations in which the trunk muscles are involved in maintaining lumbar stability.

Between-group comparisons on MPT

There is still considerable discrepancy in the literature with regards to impairment of trunk proprioception in patients with LBP [20]. Considering that fundamental differences exist in the different available proprioception measures (MPT, active and passive joint position sense, force sense) [28], we will focus on MPT findings in the transverse plane (lumbar spine rotation) [8, 14, 20].

Previous studies have compared MPT between LBP patients and control subjects, using a device similar to the one used in the present study. Two studies showed significantly higher MPT in patients with LBP [8, 14], while one study showed no difference as in the present study [20]. However, the study conducted by Selfies et al. [20] involved young athletes with history of low back injury at the start of the study, which differ from patients with LBP in the present study. As patients with non-specific LBP are a very heterogeneous group, such conflicting results are not surprising. Unfortunately, the prevalence of patients with a clinically relevant proprioceptive deficit has never been substantiated with a representative sample of patients. Consequently, the number of patients required to study clinically-relevant subgroups of patients, based on their level of proprioceptive deficits, is unknown. Heterogeneity of MPT scores was not higher in the patient group, comparatively to the control group (Fig. 3), as the change of scores over time. Consequently, the presence of clinical subgroups is not obvious within this small sample of patients. The two studies that looked at sex comparisons showed no difference in position sense [8, 20], which concurs with the present findings.

Limitations

The test-retest reliability results are limited to healthy subjects. Patients with low back pain could have produced different findings. The healthy “control group” used in the present study does not correspond to a conventional control group composed of patients that have not received any treatment. Consequently, although unlikely, it would be possible that such a patient-based control group would have shown no learning of the motion perception measurement, which in turn would have enhanced the likelihood of detecting a GROUP × DAY interaction. As patients with non-specific LBP are a very heterogeneous group, a larger sample of patients is needed to explore possible clinical subgroups based on their level of proprioceptive deficits. This would have been valuable to study relationships with changes in clinical outcomes over treatment. There was no blinding in the assessment to which group the participant was in (LBP patient or control subject), although this type of objective assessment is not believed to be affected by the experimenter. Finally, a single baseline measurement was carried out. Future studies should considered multiple baseline measurements in lumbar MPT in order to wash-out learning effects to better isolate treatment effects.