There were 3040 (93%) participants who were still alive out of 3271 participants in the baseline survey. Of 3040 remaining participants, 2594 (85%) attended the follow up survey. Participation was not significantly related to visual outcomes such as decreased visual acuity, cataract, or glaucoma. In addition, it was not significantly related to sex even at the univariate level, and the p value was only 0.04 for age in the multivariate level.10
The only factors other than age related to participation was country of birth and language spoken at home, where non-English speakers and people born in Greece, Malta, or Cyprus were significantly less likely to participate.10
However, the rates of participation based on those who were still alive at the follow up survey were 73% for non-English speakers, 84% for English speakers, 72% for people born in Greece, Malta, or Cyprus, and 79% for people born in other places. The effects of reduced best corrected visual acuity in the better eye on deaths and falls in the Melbourne VIP project have been previously examined,10,16
where the predictors were from the baseline survey.
Of 2594 participants who attended the follow up survey, 2530 (98%) had complete data records for presenting visual acuity. Of 159 participants with bilateral vision loss, 84 (53%) had correctable vision loss and 75 (47%) had non-correctable vision loss. Of 302 participants with unilateral vision loss, 165 (55%) had correctable vision loss and 137 (45%) had non-correctable vision loss. Also 64 (47%) of 137 with non-correctable unilateral vision loss had best corrected visual acuity of less than 6/7.5 in the better eye. It should be noted that the frequency of an outcome was reported only for those who had a complete record for that outcome (tables 1 and 2). For example, only 137 (78%) of 159 participants with bilateral vision loss had a complete record for the outcome of nursing home placement. Similarly, 281 (93%) of 302 participants with bilateral vision loss had a complete record for the outcome of getting help with chores.
For all outcomes we included only age, sex, and vision loss in the univariate analyses, and we further selected other significant predictors by the backward selection method in addition to these predictors. From the univariate analyses, both correctable bilateral or unilateral vision loss were not significantly associated with any outcomes except those of visual functions (table 3). Moderate to severe non-correctable unilateral vision loss did not produce more significant results than non-correctable unilateral vision loss. Thus we carried out the multivariate analyses for only non-correctable unilateral and bilateral vision loss. Non-correctable vision loss remained significant or insignificant in the best selected multivariate models as they were in the corresponding univariate models for all outcomes, and the magnitudes of the odds ratios from the best selected multivariate models were essentially the same as those from the univariate models.
Table 3 Univariate odds ratios (95% CI) adjusted by age and sex of correctable, non-correctable, or moderate to severe non-correctable unilateral or bilateral vision loss versus normal vision
We presented the univariate and multivariate odds of vision loss verus normal vision (tables 3 and 4). Non-correctable unilateral vision loss was associated with vision related activities, falling, and dependency issues. Non-correctable unilateral vision loss with best corrected visual acuity of less than 6/7.5 in the better eye increased the odds of having health/emotional problems in comparison with normal vision (univariate OR
4.94, 95% CI 1.13 to 21.6). On the other hand, non-correctable bilateral visual loss was not associated with falling, but it was associated with dependency, nursing home placement, emotional wellbeing, and visual tasks. Non-correctable unilateral vision loss gave a twofold to fivefold increase in the odds of having problems in reading the telephone book, reading the newspaper, watching television, and seeing faces. Non-correctable bilateral vision loss gave a sixfold to 41-fold increase in the odds of having these problems.
Table 4 Odds ratios (95% CI) of non-correctable unilateral or bilateral vision loss versus normal vision from the best selected multivariate models (all analyses include age and sex)