Participants
Twenty eight male participants (14 healthy individuals and 14 patients with LBP) between the ages of 21–50 were asked to participate in the study. Healthy individuals consisted of those who were free of LBP for the six months previous to the study, whereas patients with LBP had a history of LBP for at least three consecutive days over the last three consecutive weeks prior to testing. Individuals with a known neurological disorder, scoliosis or other deformity, inflammatory or degenerative arthropathy, connective tissue disease, or a history of spinal surgery were excluded from the study. Individuals with current or previous neck pain in the past three weeks were also excluded. Participants were asked to avoid engaging in any type of resistive exercise for the 48 hours prior to testing. All participants signed the informed consent form. The procedures used were in accordance with the institutional research ethics board. The clinical trial was registered at ClinicalTrials.gov (NCT00754585). Data were collected in the Biomechanics and Elastography laboratory at the Canadian Memorial Chiropractic College (CMCC).
Protocol and instrumentation
A Grahl Duo Back™ office chair (Rohde & Grahl, Steyerberg/Voigtei, Germany), fixed in position to prevent it from swivelling or rolling, was used for the study. The arm rests were lowered so that they were not used and to ensure the maximum amount of loading was transferred to the seat pan of the chair. The chair had all the features of a typical ergonomic office chair but it was unique in that the back rest did not provide any specific lordotic support and was split vertically, providing access to the midline for sensor attachment.
At the beginning of the data collection, participants were asked to stand in a neutral position “with their arms by their side, weight evenly balanced, and looking straight ahead”. Kinematic data were collected over 30 seconds of neutral static stance for comparison against the seated conditions. Participants sat in the office chair and in the same seat pan but with a lumbar support pillow (“Logic Back™”, Mediflow Inc., Toronto) to test the effect of the back rest profile on comfort and lumbar and thoracolumbar postures. The lumbar support is a portable device, convex in the anterior direction and contoured with an arched opening above the seat pan, that provides space for the bulk of the posterior pelvic tissues (Figure 1). The back frame of the device is constructed of a solid plastic, curved side-to-side and is relatively rigid. The frame acts as a bow which is strung anteriorly by four adjustable straps. These straps provide an elastic, anterior projection above the buttock tissues. A band affixes the device to the chair back. The lumbar support was “fitted” to each participant prior to testing by having the individual sit in the chair in a relaxed fashion, with the hips and knees flexed to 90°, feet flat and looking straight ahead. The pelvis was pushed all the way back into the aperture of the pillow, and the individual’s lumbar spine rested against the back rest. The straps were tensioned to participant preference.
Study participants sat for 30 minutes in the regular chair with vertical (90°) back support and for 30 minutes in the chair with additional lumbar support while watching a video on a computer screen placed directly in front of them in the mid-sagittal plane. The angle of visual gaze was controlled by the height of the computer monitor which was placed 15 cm above waist height for each participant. The participants’ feet rested on an adjustable foot rest such that their hips and knees were flexed to 90°. There were seven minutes of rest between conditions, whereby participants were asked to stand and move freely about. The order of conditions was randomized. All sources of metal (e.g., belts, keys in pockets, etc.) were removed prior to testing to minimize any interference with the electromagnetic equipment.
Postural measure
Electromagnetic sensors (Polhemus Liberty® system, Colchester, Vermont) were placed in the midline over the spinous processes at the junctions of the neck and upper back (T1), mid and lower back (T12), centre of the lower back (L3), and over the spinal base at the sacrum (S2). The spinal configuration was represented as a series of linkages connected by nodes that allowed for bending in the sagittal plane at the landmark pivots (Figures 2 and 3). The thoracic spine was treated as a single segment, rigid body, while the lumbar spine was modelled as a two segment linkage. The sensors were sampled at 240 Hz which allowed for the continuous and automatic monitoring of landmark positions and orientations in space. This system has an accuracy of 0.15° RMS. Difference in orientation between the thoracic link and the upper lumbar link were used as a surrogate for the thoracolumbar angle at T12 while the upper lumbar and lower lumbar links at L3 served to estimate the lumbar lordosis angle.
Comfort
Using methods inspired by Fenety et al. [22], the seat–user interface pressure distribution in the current study was measured using a pressure mapping system (CONFORMat®, Tekscan Incorporated, Boston). The sensor mat is an ultra-thin (0.00400, 0.10 mm) flexible printed circuit with 1024 individual sensing elements or cells organized in a 32 × 32 array with a density of 0.5 sensels/cm2. Before the study, the pressure mat sensels were preconditioned, equilibrated, and calibrated using the Tekscan Inc. uniform pressure vacuum pump and Tekscan Inc. user guide. During the collection, the pressure mat was placed only on the specific seat surface to measure the CoP at the buttock-chair interface. It was covered with a sheet that was fixed at the ends to prevent slipping of the mat and participant bias by observing the mat. Data were recorded at 60 Hz and fed into a PC computer. The first two minutes of data were removed from analysis to ensure that the individual was “settled” prior to calculation of the CoP.
Each trial was divided into three epochs of equal duration. Using MatLab 2007b (version 7.5.0.342, Mathworks Inc., Natick, MA), a circle of best fit was calculated for each epoch using a least squares method. The radius was of particular interest as it gave a measure of the overall CoP shifting, such that the larger the radius, the more the shift, and the greater the objective measure of discomfort. MatLab code inspired by Gander et al. [23] was used to calculate a best-fit circle that minimized algebraic error (Figure 4).
To obtain a subjective measure of comfort and to get an idea of how the pillow would affect comfort in other areas of the body, individuals were asked to complete a body map questionnaire [24]. They reported their level of discomfort in various areas of the body according to a visual analog scale (VAS) at baseline and after sitting in the chair with the lumbar support and without. Participants were asked to mark along a 100 mm line where their level of discomfort is “right now”, 0 mm being “none” and 100 mm being the “worst possible”. VAS measures were collected for the neck, upper back/back of shoulders, mid back, low back, buttocks, thighs and lower legs.
Data analyses
Data were selected for three distinct epochs (minutes 2–4; minutes 15–17; and minutes 27–29) for analysis to represent behaviour over the full 30-minute test interval. Significance in statistical comparisons was set at p < 0.05. To calculate the sample size for a three-factor repeated measures design, we used an approximate approach based on a paired t-test for the comparison between chair support. We set out to detect a large effect size ([25]; d = 0.8) with power of 0.8 and a significance level set at 0.05 within each group (healthy participants versus those with LBP). A sample size of 14 was required for each group.
Posture and objective measure of comfort
Three-factor repeated measures ANOVAs (with group, condition and epoch) were used to identify any significant main effects or their interactions on the lumbar and thoracolumbar angles. Similarly, a three-factor repeated measures ANOVA was used to identify any significant main effects or their interactions on the Least Squares Radius (LSR). There were two levels of group (healthy and LBP), three levels of condition (standing, lumbar support and standard chair), and three time intervals (epochs 1, 2 and 3).
Subjective measure of comfort: VAS
One-factor repeated measures ANCOVAs (baseline VAS measure as covariate) were used to identify any effect of condition (standing, lumbar support and standard chair) on VAS scores for each group separately (healthy individuals and patients with LBP).
T-tests (paired and unpaired where appropriate) employing Holm’s method of p-value adjustment were used for all post-hoc pair-wise comparisons following significant ANOVA/ANCOVA results. The R-Project statistical software version 2.12.1 was used for all data analyses (The R Foundation for Statistical Computing, Institut für Statistik und Wahrscheinlichkeitstheorie, Vienna, Austria).