Introduction

Good balance is an essential quality in controlling body movement. However, as one gets older, the ability to maintain balance control deteriorates. This decline can be attributed to decline in the somatosensory, visual, and vestibular systems to varying degrees (Baloh et al. 1993; Enrietto et al. 1999; Skinner et al. 1984). The incidence of visual impairment has also been shown to increase rapidly with age (Congdon et al. 2004; Hong Kong Blind Union 2001) and it has been suggested that affected people may partially compensate by relying more on proprioceptive and/or vestibular cues for balance (Di Girolamo et al. 1999; Elliott et al. 1995). However, joint proprioception and vestibular acuity also decline with age (Skinner et al. 1984; Tsang and Hui-Chan 2004) as well as lower limb muscle strength (Sturnieks et al. 2008). As a result, balance control becomes a major problem in older individuals with visual impairments (Lee and Scudds 2003). While the previous studies investigated the balance control of sighted elderly persons, that of very old adults with varying degrees of visual impairment was scarcely studied. Whether or not balance control differs with varying degrees of visual impairment remains to be answered.

Methods

Subjects and study design

Seventeen low-vision subjects (1 male, 16 females; mean age 82.9 ± 5.4 years), and sixteen blind subjects (1 male, 15 females; mean age 82.2 ± 8.1 years) were recruited from residential care homes of the Hong Kong Society for the Blind. Fifteen sighted subjects (3 males, 12 females; mean age 79.7 ± 5.2 years) were recruited from the community. All the subjects were recruited by convenience sampling. All were independent walkers, independent in their activities of daily living, and able to communicate and follow the testing procedures. Candidates were excluded if they had any vestibular problems, symptomatic cardiovascular disease at a moderate exertion level, poorly controlled blood pressure, a history of neurologic disease (e.g., Parkinson’s disease), acute orthopedic problems that affected ambulation, or metastatic cancer.

All of the subjects underwent three screening tests. (1) functional ambulation classification (FAC) resulting in a classification of level 3 or above. (2) the visual acuity of the visually impaired subjects was extracted from their recent medical records, and the visual acuity of sighted subjects was checked using the Snellen eye chart. Low vision was defined as visual acuity of less than 6/18 but equal to or better than 3/60 in the better eye with the best possible correction. Blindness was defined as visual acuity of less than 3/60 in the better eye with the best possible correction. Normal vision was defined as visual acuity of better than 6/18 on the Snellen eye chart in the better eye with the best possible correction (World Health Organization 2007). (3) The Mini-Mental Status Examination (MMSE) was employed to screen subjects with cognitive dysfunction (Folstein et al. 1975). A modified physical activity questionnaire (Van Heuvelen et al. 1998), used to compare among three groups, allowed to categorize the subjects’ daily activities using metabolic index units (METs) as either light (<4 METs), moderate (4–5.5 METs), or heavy (>5.5 METs) (Tsang and Hui-Chan 2003).

The protocol was approved by the Ethics Committee of The Hong Kong Polytechnic University, and informed consent was obtained from all subjects. Those who could not sign used their fingerprint as a signature.

Test procedures

All the participants underwent the following assessments.

Muscle strength

The concentric isokinetic strength of the knee extensors and flexors of each subject’s dominant leg (the leg which they used for kicking a ball) were tested, at an angular velocity of 30°/s, using a Cybex Norm dynamometer. Before starting, the subjects performed a 5-min warm-up which included stretching the knee muscle groups. After three sub-maximal practice trials, five maximal concentric contractions of the knee extensors and flexors were recorded. The reliability of such testing has been established in a previous study with ICC ranging from 0.86 to 0.97 (Tsang and Hui-Chan 2005).

Sensory organization test

A sensory organization test (SOT) was used to measure the subjects’ ability to use somatosensory, visual, and vestibular information to control body sway when standing under reduced or conflicting sensory conditions. The subjects’ task was to stand on a force platform with as little sway as possible for 20 s under six sensory conditions. For each condition, three trials were performed. Body sway angle in the anteroposterior direction was estimated from the force data collected, and the three results were averaged for the statistical comparison. The reliability of this procedure has been established in a previous study with ICC ranging from 0.72 to 0.93 for the six conditions (Tsang et al. 2004).

Perturbed double-leg stance test

A perturbed double-leg stance test (PDLST) was used to assess the ability of the automatic motor system to quickly recover following an unexpected external disturbance. Subjects stood with the eyes opened and closed on a platform which suddenly and randomly applied an anteroposterior body sway angle of 3.2°, scaled to the subject’s height, in the forward or backward direction. Body sway was first recorded for 2 s before the platform translation, and the average sway observed served as a baseline. After each platform translation, the maximum body sway angle was estimated and the difference from the baseline value, termed the perturbed body sway angle was calculated. The perturbed body sway angles of three trials for each perturbation direction were averaged, and the averages were used to compare the three groups. A value of 12.5°, the theoretical anteroposterior sway stability limit, was assigned as the perturbed body sway angle if any subject fell during the platform translation (NeuroCom 2000). A “fall” was recorded if the subject touched the visual surround for support, or gained support by stepping forward, or required support by the investigators.

Five times sit-to-stand test

The five times sit-to-stand test (FTSTST) measures the time taken to complete five repetitions of standing from a chair. All sit-to-stand repetitions were performed from a hard chair 43 cm high and 47.5 cm deep without arm rests. Following the standardized instructions, each subject completed five trials with 1 min of rest between trials as needed to prevent fatigue. The first two trials were practice trials for familiarization purposes. The mean time of the last three trials was used for analysis. The time was recorded from when the subject’s back left the backrest on the first repetition and stopped when their back touched the backrest on the last repetition (Mong et al. 2010).

Statistical analysis

The groups’ average ages, heights, and weights, were compared using one-way analysis of variance (ANOVA). Post hoc analysis using Bonferroni’s adjustment was conducted if a significant difference was found. A Chi-square test was applied for among-group comparison of gender. A Kruskal–Wallis test was used to compare the FAC and physical activity level data. If a significant difference was found, a Mann–Whitney test was conducted for between-group comparison. For among-group comparison of the FTSTST times, one-way ANOVA was employed. Post hoc analysis using Bonferroni’s adjustment was again conducted if a significant difference was found. Multivariate ANOVA was used initially to explore how the performance on all the balance control and muscle strength tests differed among the three subject groups. If a statistically significant difference was found, univariate tests were used to analyze each component of these tests. Post hoc analysis using Bonferroni’s adjustment was conducted if a significant difference was found in the univariate tests. A significance level of 0.05 was chosen for the statistical comparisons.

Results

Subjects

Table 1 summarizes the demographic results and shows no significant difference in average age, gender, or weight among the three groups. One-way ANOVA showed an overall significant difference in height (p < 0.05). Subsequently, a post hoc test was conducted which showed that the sighted elderly were on average taller than those with low vision. The Kruskal–Wallis test showed an overall significant difference in average FAC and physical activity levels among the three groups (p < 0.05). The post-hoc test showed that the sighted elderly had significantly better average FAC scores than the blind subjects, and higher average physical activity levels than both of the other two groups with visual impairment.

Table 1 Demographic characteristics of the three groups
Full size table

Muscle strength

Multivariate ANOVA of the knee strength results indicated an overall statistically significant difference among the three groups (p = 0.025). The univariate tests showed statistically significant differences in the average concentric knee extensor peak torque-to-body-weight ratios (p = 0.015) but not for knee flexor. Compared to the low vision subjects (0.5 ± 0.2%), the sighted elderly achieved higher knee extensor ratios (0.8 ± 0.4%; p = 0.024). The torques of the blind subjects (0.7 ± 0.3%) were not, on average, significantly different from those of the low-vision subjects (p = 0.06) or the sighted subjects (p = 1.000).

Sensory organization test

The multivariate ANOVA of the SOT results showed an overall significant difference among the three groups (p < 0.001; Fig. 1). The univariate tests indicated differences among the groups in conditions one (p = 0.018) and four (p < 0.001). In condition one, sighted subjects had significantly higher equilibrium quotients (EQ) (mean = 93.9 ± 1.9%) than the blind subjects (mean = 91.5 ± 2.5%; p = 0.022). In condition four, the sighted subjects EQ (mean = 63.1 ± 24.5%) were significantly higher than those of the low vision (mean = 24.6 ± 25.7%) and blind subjects (17.1 ± 22.7%), both (p < 0.001).

Fig. 1
figure 1

Average equilibrium quotients on the SOT among sighted subjects, subjects with low vision and blind subjects. †Denotes difference significant at the p < 0.001 (*p < 0.05) level

Full size image

Perturbed double-leg stance test

The multivariate ANOVA of the PDLST results indicated an overall statistically significant difference among the three groups (p = 0.047; Fig. 2). The univariate tests showed that sighted subjects and the low-vision subjects achieved less body sway during the forward (mean values 4.0 ± 0.6° and 4.4 ± 1.7°, respectively) and backward (mean values 4.1 ± 1.0° and 4.6 ± 1.7°, respectively) platform translations with eyes open compared to the blind subjects (mean values 7.1 ± 3.5° and 7.1 ± 3.9°).

Fig. 2
figure 2

Results of the perturbed double-leg stance test among sighted subjects, subjects with low vision and blind subjects. EO, eyes open; EC, eyes closed. *Denotes a difference significant at the p < 0.05 level

Full size image

Five times sit-to-stand test

The three groups showed no significant differences in their average performance on the FTSTST (p = 0.185). The sighted subjects had an average time of 15.3 ± 3.3 s, the subjects with low vision had a time of 18.8 ± 8.4 s, while the blind subjects had a time of 15.1 ± 5.5 s.

Discussion

As might be expected, the sighted elderly attained better functional ambulation categories and scored higher in their physical activity levels than the visually impaired. A study by Crews and Campbell (2001) has previously reported that elderly people aged 70 and older with visual impairment often encounter activity limitations. This makes them less physically active and leads to reduced physical functioning. The findings of this study provide further evidence of greater decline in physical activity levels and functional ambulation in those with visual impairment. This amplifies the importance of finding a form of exercise that is suitable for this population to help improve physical functioning.

Also, these very old adults with visual impairment performed less well in certain balance control maneuvers and showed less knee extensor muscle strength than the sighted controls similar in age and gender distribution. The data confirm that vision plays an important role in controlling balance, especially in challenging circumstances. When standing on a firm surface during the SOT, all three groups achieved high EQ scores (>90%; Fig. 1) regardless of the visual conditions. This suggests that visual impairment may not influence significantly the postural sway of the elderly in unchallenging conditions. Indeed, a study by Fitzpatrick and McCloskey (1994) has shown that proprioception in the lower limbs is the most important sensory input for controlling postural sway in unchallenging conditions. However, when standing on a sway-referenced surface and allowed visual input, the sighted elderly attained significantly higher EQ scores (condition 4 of the SOT). In this condition, the subjects had to rely on visual inputs to maintain their balance, as the somatosensory inputs were misleading. This is in line with the results reported by Lord and Menz (2000), who concluded that visual impairment is strongly associated with increased sway when standing on a compliant surface. Poor performance in condition 4 of the SOT is an alarming sign and could account for the greater incidence of falls among those with visual impairment (Wallmann 2001). On the other hand, when visual input was distorted or taken away, the sighted elderly had no better balance performance than the visually impaired groups. In condition 5, subjects had to rely on vestibular inputs as there was no visual input allowed. The results confirm the importance of visual information from the environment as source of feedback for balance control (Stones and Kozma 1987). Outdoor surfaces are often compliant, which according to the results of this study, should increase the body sway of very old adults with visual impairment. Such unsteadiness in posture may explain the decreased physical activity of this group.

The SOT has been conducted with younger subjects (mean age 51.6 years) in a study by Chen et al. (2010). The average EQ score of our sighted subjects in condition 4 was 12.7% lower than that of younger group and that of our subjects with visual impairment was lower by 56%. This indicates that the balance abilities of visually impaired individuals decrease more than those of sighted individuals as they age. The dramatic difference in balance ability highlights the need for visually impaired persons to work at improving their balance control capability.

In the PDLST, the blind elderly swayed significantly more after being perturbed either forward or backward than the sighted and low-vision subjects when their eyes were open (Fig. 2). When there was no visual input allowed during the perturbation, no significant difference among the three groups was observed. These findings are consistent with the “two-mode” theory of vision (Held 1970). According to this theory, visual information reaches an individual in ambient and focal modes. Ambient vision is responsible for orientation and locomotion, while focal vision is responsible for object recognition and identification. This theory further asserts that controlling postural sway relies heavily on the ambient visual mode. Based on this theory, our sighted and low-vision elderly apparently made use of their ambient vision to react to an external perturbation, while the blind could not. In performing daily activities one has to react to unstable support surfaces, such as when stepping onto a moving escalator or walking in the aisle of a moving bus or train. Increased postural sway under perturbation may hinder subjects with visual impairment in such activities and discourage them from going outdoors.

A group led by Whitney has reported that the time on the FTSTST with optimal sensitivity and specificity for identifying a balance dysfunction in people older than 60 years is 14.2 s (Whitney et al. 2005). The mean times observed in this study were 15.3 s for the sighted subjects, 18.8 s for the low-vision subjects, and 15.1 s for the blind subjects, all of whom were around 80 years old. This would indicate that all our subjects probably had some degree of balance dysfunction.

Certain limitations of this study protocol restrict the generalisation of these results. Although each subject’s visual acuity was verified, other aspects of vision degraded with aging, such as contrast sensitivity, central processing of visual input and visual perception, were not tested and may make important contributions.

The study has shown, however, that elderly persons with low vision have poorer balance control than the sighted elderly. These results confirm that vision plays an important role for older adults in controlling balance in the more challenging circumstances encountered in daily living. Such subjects may benefit from adapted physical training.