3.1. Participants
Twenty patients with LBP (LBP group) and twenty healthy individuals(control group) were recruited in the present study. All study participants were female and right-handed. There were no significant differences in age, weight, or height between two groups (P≥0.05 for all) (Table 1). However, the LBP group had significantly higher BMI and abdominal circumference, and more falls in the previous 12 months (Table 1).
3.2 Postural performance in the single or dual tasks
Five participants in the LBP group could not complete the one-leg stance in a single task or dual task, so the data of 15 participants with LBP were used in the mixed model repeated-measure ANCOVA. Table 2 shows the mean (SD) of COP parameters in different combinations of postural difficulty and cognitive difficulty. Table 3 presents a summary of ANCOVA results for all data of postural performance in the single or dual tasks.
With regard to balance performance, the covariate of BMI AP:[F(1,31)=1.122, P=0.298,ƞ2p=0.035], ML:[F(1,31)=0.076,P=0.784,ƞ2p=0.002], sway area:[F(1,31)=0.115, P=0.737,ƞ2p=0.004], sway length:[F(1,31)=0.454, P=0.505,ƞ2p=0.014] and abdominal circumference AP: [F(1,31)=1.039, P=0.316,ƞ2p=0.032], ML: [F(1,31)=0.516 P=0.478,ƞ2p=0.015], sway area: [F(1,31)=0.003, P=0.956,ƞ2p=0.000], sway length:[F(1,31)=0.010,P=0.920,ƞ2p=0.000] were not significant in all the COP parameters. The COP parameters were significant between two groups AP:[ F(1,31)=49.260, P<0.001,ƞ2p=0.614], ML:[F(1,31)=115.997, P<0.001,ƞ2p=0.789], sway area: [F(1,31)=82.414, P<0.001,ƞ2p=0.727], sway length:[F(1,31)=48.600, P<0.001,ƞ2p=0.611]. The main effects of postural difficulty were significant in AP:[F(1.297,40.197)=8.254, P<0.005,ƞ2p=0.210],ML:[ F(1.202,37.250)=3.920, P<0.001,ƞ2p=0.112], and sway length:[F(1.425,44.175)=3.128, P=0.069, ƞ2p=0.092], but not significant in sway area:[F(1.160,35.972)=1.662, P=0.207, ƞ2p=0.051]. The main effects of cognitive difficulty were not significant in all COP parameters AP: [F(1.607,49.823)=0.192, P=0.778, ƞ2p=0.006], ML:[F(2,62)=2.554, P=0.086,ƞ2p=0.076],sway area:[F(2,62)=0.518, P=0.582,ƞ2p=0.016 and sway length:[ F(2,62)=0.415, P=0.653,ƞ2p=0.013].
The group × postural difficulty × cognitive difficulty effect was only significant in sway length:[F(2.615,81.067)=3.044, P=0.040,ƞ2p=0.089]. Post hoc analysis showed that the LBP group had larger sway length than the control group in the dual task (P<0.05) but not in the single task, when standing on a force platform with eyes open or closed. But the LBP group showed larger sway length than the control group in both the single and dual tasks (P<0.05) in one-leg stance. The group × postural difficulty effects were significant in all COP parameters AP: [F(1.297,40.197)=6.862, P=0.008,ƞ2p=0.181],ML:[F(1.202,37.250)=17.359,P<0.001,ƞ2p=0.435], sway area:[F(1.160,35.972)=9.142, P=0.003,ƞ2p=0.228] and sway length: [F(1.425,44.175)=7.198, P=0.005, ƞ2p=0.188]. Post hoc analysis showed that the LBP group had larger COP parameters than the control group in three tasks with different postural difficulties (P<0.05). The cognitive difficulty × group effects were significant in ML:[F(2,62)=3.448, P=0.038, ƞ2p=0.100], sway area: [F(2,62)=15.065, P<0.001,ƞ2p=0.327], sway length:[F(2,62)=5.346, P=0.007, ƞ2p=0.147], and marginally significant in AP: [F(1.607,49.823)=2.994, P=0.070, ƞ2p=0.089]. Post hoc analysis for the cognitive difficulty × group effects showed that the LBP group had larger COP parameters than the control group in three tasks with different cognition difficulties (P<0.05). LBP participants showed larger COP parameters in the dual tasks with high and low cognitive difficulties than those in single task (P<0.05), whereas the control participants only displayed larger COP parameters in the dual task with higher cognitive difficulty than those in single task (P<0.05)(Fig. 4). Both LBP and control participants showed larger COP parameters in the dual task with higher cognitive difficulty that the dual task with lower cognitive difficulty (P<0.001). No significant postural difficulty × cognitive difficulty effects were found in COP parameters. These results suggested that compare to the healthy older people, the older people with LBP had poor postural performance reflected by larger COP parameters regardless of any postural or cognitive difficulties. Compared with the single task, the LBP participants’ postural control were decreased in the dual task, even though the difficulty level of the cognitive task was low. The control participants’ postural balance , however, were poor when the difficulty level of the cognitive task was high. The postural balance of control groups, however, was decreased when the difficulty level of the cognitive task was high.
3.3 Cognitive performance in the postural tasks
The cognitive performance results in the postural tasks are shown in Table 4.The covariates of BMI (F = 7.258, P = 0.011), and abdominal circumference (F=11.123, P=0.002) were significant in the condition of auditory arithmetic task with a one-leg stance. BMI and abdominal circumference, however, were not significant in other conditions (P≥0.05). After adjustment of the covariates of BMI and abdominal circumference, the group main effect (F = 6.011, P = 0.019) and the interaction effect of group × postural difficulty (F = 6.859, P = 0.003) were significant, whereas other main effects and interaction effects were not significant (Table 4). Post hoc analyses for the interaction effect of group × postural difficulty showed that the LBP group had lower percentage accuracy than that of the control group only in the one-leg stance condition (P < 0.05)(Table 5). However, there was no significant between-group difference in two-leg stance with eyes-open or eyes-closed (P≥0.05).
3.4 Associations between COP parameters and the number of falls
The associations between the COP parameters in all conditions and the number of falls are shown in Table 6. Significant associations between sway area (R=0.386)/sway length (R=0.482) and the number of falls in a single task were shown only in the eye-closed condition (P < 0.05). The correlations between sway area and the number of falls in dual task 1 (postural task and auditory arithmetic task) were significant in eyes-open (R=0.314) and eyes-closed (R=0.323) conditions (P < 0.05). The correlations between sway length (R=0.445,R=0.331,R=0.347) and the number of falls in dual task 2 (postural task and serial-7s arithmetic task) were significant in the eyes-open, eyes-closed, and one-leg stance conditions(P < 0.05). The other associations between the COP parameters and the number of falls were not significant(P≥0.05).