Visual Impairment and Mobility Performance in Older Adults
Visual Impairment and Mobility Performance in Older Adults
At baseline, 169 (7%) of SEE Study participants were categorized as VI, and 2,351 (93%) were categorized as NVI. The VI participants were significantly older and, after adjustment for age, were more likely to be black, to have a lower Mini–Mental State Examination score, to have diabetes, and to report depressive symptoms compared with NVI participants (Table 1).
For all outcomes, performance speeds declined at each study visit, and the VI had slower speeds at each time point than did the NVI (Table 2, models 1a, 2a, and 3a). We extended our models to include an interaction between years since baseline and visual impairment status to assess whether speeds declined at different rates in the VI and the NVI (Table 2, models 1b, 2b, and 3b). However, the interaction terms were not significant for any of the 3 speed outcomes, indicating that the change in speed over time was similar for the VI and the NVI over the 8-year period. For example, the interaction term between visual impairment status and speed walking up steps was 0.00 steps/second (95% confidence interval (CI): −0.01, 0.01) (Table 2, model 1b), indicating that the change in speed over time for this task was the same for the VI and the NVI. However, after including this interaction term in the model, the VI still had slower speeds at each time point. For walking down stairs, the VI remained 0.08 steps/second slower than the NVI (95% CI: −0.10, −0.05) (Table 2, model 1b).
Other factors were significantly related to slower speeds, including age, sex, race, the number of other comorbid conditions, and the presence of depressive symptoms. These results were largely unchanged in the models that included an interaction between visual impairment status and years since baseline.
For each year of observation, the odds of being classified as disabled for all 3 tasks increased by approximately 20% (Table 3, models 4a, 5a, and 6a). Additionally, the VI were about twice as likely to be classified as disabled than the NVI over the study period after adjustment for all other covariates. We included an interaction between years since baseline and visual impairment status to determine whether the odds of having mobility disability increased at different rates in the VI and the NVI (Table 3, models 4b, 5b, and 6b). However, this interaction was not significant for any of the mobility disability outcomes, indicating that the difference in the odds of having mobility disability between the VI and the NVI remained the same over the study period.
To determine the potential affect of losses to follow-up on our results, we modeled the cross-sectional odds of being lost to follow-up compared with the odds of not being lost to follow-up at each study visit after baseline. From baseline to the 2-year visit, the VI were not more likely than the NVI to be lost to follow-up, although those classified as having disability walking up stairs (ORlost at 2-year visit = 1.9, 95% CI: 1.3, 2.7), walking down stairs (ORlost at 2-year visit = 1.7, 95% CI: 1.2, 2.4), and walking 4 m (ORlost at 2-year visit = 1.7, 95% CI: 1.1, 2.4) were more likely to be lost than those not reporting these difficulties. The interaction terms assessing differential loss to follow-up of VI participants with slowest speeds were not significant for any of the performance-based measurements (data not shown). Models predicting losses to follow-up at the 6-year and 8-year visits had the same inference; the odds of being lost were not significantly different by visual impairment status, and the interaction terms assessing differential loss to follow-up of VI participants with slowest speeds were not significant for any of the performance-based measurements.
To examine how changing the criteria used to determine mobility disability would affect the results, we shifted the cutpoint for defining mobility disability from 1 standard deviation to speeds below 0.5 standard deviations of the population baseline mean. This meant that disability was redefined as speeds slower than 0.8 steps/second walking up stairs, 0.7 steps/second walking down stairs, and 0.7 m/second walking 4 m. After we shifted the disability cutpoint, the VI were more likely than the NVI to be classified as having mobility disability for all 3 outcomes (ORdisabled walking up steps = 1.7, 95% CI: 1.2, 2.3; ORdisabled walking down steps = 1.7, 95% CI: 1.2, 2.3; and ORdisabled walking 4 m = 1.8, 95% CI: 1.3, 2.4). However, the interactions between visual impairment status and years since baseline were not significant for any of the outcomes (data not shown).
We explored how changing our definition of visual impairment affected our results and shifted this definition to best-corrected distance visual acuity worse than 20/60 in the better-seeing eye. This alternate cutpoint was chosen because it is the visual acuity criterion for visual impairment used by the World Health Organization. The inference and resulting speed estimates were largely the same as in our primary analyses and indicated that the VI had significantly slower performance speeds than the NVI (βwalking up steps = −0.10 steps/second, 95% CI: −0.14, −0.06; βwalking down steps = −0.14 steps/second, 95% CI: −0.20, −0.09; and βwalking 4 m = −0.09 m/second, 95% CI: −0.13, −0.05). Additionally, the interactions between the new category of visual impairment status and years since baseline were not statistically significant (data not shown).
Results
Participant Characteristics
At baseline, 169 (7%) of SEE Study participants were categorized as VI, and 2,351 (93%) were categorized as NVI. The VI participants were significantly older and, after adjustment for age, were more likely to be black, to have a lower Mini–Mental State Examination score, to have diabetes, and to report depressive symptoms compared with NVI participants (Table 1).
Visual Impairment Status and Performance Speeds
For all outcomes, performance speeds declined at each study visit, and the VI had slower speeds at each time point than did the NVI (Table 2, models 1a, 2a, and 3a). We extended our models to include an interaction between years since baseline and visual impairment status to assess whether speeds declined at different rates in the VI and the NVI (Table 2, models 1b, 2b, and 3b). However, the interaction terms were not significant for any of the 3 speed outcomes, indicating that the change in speed over time was similar for the VI and the NVI over the 8-year period. For example, the interaction term between visual impairment status and speed walking up steps was 0.00 steps/second (95% confidence interval (CI): −0.01, 0.01) (Table 2, model 1b), indicating that the change in speed over time for this task was the same for the VI and the NVI. However, after including this interaction term in the model, the VI still had slower speeds at each time point. For walking down stairs, the VI remained 0.08 steps/second slower than the NVI (95% CI: −0.10, −0.05) (Table 2, model 1b).
Other factors were significantly related to slower speeds, including age, sex, race, the number of other comorbid conditions, and the presence of depressive symptoms. These results were largely unchanged in the models that included an interaction between visual impairment status and years since baseline.
Visual Impairment Status and Mobility Disability
For each year of observation, the odds of being classified as disabled for all 3 tasks increased by approximately 20% (Table 3, models 4a, 5a, and 6a). Additionally, the VI were about twice as likely to be classified as disabled than the NVI over the study period after adjustment for all other covariates. We included an interaction between years since baseline and visual impairment status to determine whether the odds of having mobility disability increased at different rates in the VI and the NVI (Table 3, models 4b, 5b, and 6b). However, this interaction was not significant for any of the mobility disability outcomes, indicating that the difference in the odds of having mobility disability between the VI and the NVI remained the same over the study period.
Losses to Follow-up
To determine the potential affect of losses to follow-up on our results, we modeled the cross-sectional odds of being lost to follow-up compared with the odds of not being lost to follow-up at each study visit after baseline. From baseline to the 2-year visit, the VI were not more likely than the NVI to be lost to follow-up, although those classified as having disability walking up stairs (ORlost at 2-year visit = 1.9, 95% CI: 1.3, 2.7), walking down stairs (ORlost at 2-year visit = 1.7, 95% CI: 1.2, 2.4), and walking 4 m (ORlost at 2-year visit = 1.7, 95% CI: 1.1, 2.4) were more likely to be lost than those not reporting these difficulties. The interaction terms assessing differential loss to follow-up of VI participants with slowest speeds were not significant for any of the performance-based measurements (data not shown). Models predicting losses to follow-up at the 6-year and 8-year visits had the same inference; the odds of being lost were not significantly different by visual impairment status, and the interaction terms assessing differential loss to follow-up of VI participants with slowest speeds were not significant for any of the performance-based measurements.
Sensitivity Analyses
To examine how changing the criteria used to determine mobility disability would affect the results, we shifted the cutpoint for defining mobility disability from 1 standard deviation to speeds below 0.5 standard deviations of the population baseline mean. This meant that disability was redefined as speeds slower than 0.8 steps/second walking up stairs, 0.7 steps/second walking down stairs, and 0.7 m/second walking 4 m. After we shifted the disability cutpoint, the VI were more likely than the NVI to be classified as having mobility disability for all 3 outcomes (ORdisabled walking up steps = 1.7, 95% CI: 1.2, 2.3; ORdisabled walking down steps = 1.7, 95% CI: 1.2, 2.3; and ORdisabled walking 4 m = 1.8, 95% CI: 1.3, 2.4). However, the interactions between visual impairment status and years since baseline were not significant for any of the outcomes (data not shown).
We explored how changing our definition of visual impairment affected our results and shifted this definition to best-corrected distance visual acuity worse than 20/60 in the better-seeing eye. This alternate cutpoint was chosen because it is the visual acuity criterion for visual impairment used by the World Health Organization. The inference and resulting speed estimates were largely the same as in our primary analyses and indicated that the VI had significantly slower performance speeds than the NVI (βwalking up steps = −0.10 steps/second, 95% CI: −0.14, −0.06; βwalking down steps = −0.14 steps/second, 95% CI: −0.20, −0.09; and βwalking 4 m = −0.09 m/second, 95% CI: −0.13, −0.05). Additionally, the interactions between the new category of visual impairment status and years since baseline were not statistically significant (data not shown).
Source...