Participants with either correctable or noncorrectable visual impairment were more likely to be older and to receive a government social security pension, but were less likely to be married, to have higher qualifications, to own a home or be currently employed than participants without impairment
(Table 1) . They were also more likely to report a history of angina and stroke, but there were no significant differences in the prevalence of other medical conditions.
After adjustment for demographic and medical confounders, there was a trend toward lower SF-36 scores in participants with noncorrectable impairment than in those with correctable impairment
(Table 2) in five dimensions (
P trend ≤ 0.05) and in both physical and mental component scores (PCS
P trend = 0.01, MCS
P trend = 0.02).
After multivariate adjustment, participants with any noncorrectable visual impairment had significantly poorer SF-36 scores in five dimensions and in the mental component score
(Table 3) . Increasingly severe noncorrectable visual impairment was associated with deteriorating SF-36 profiles (the latter was only age–sex adjusted because of low sample numbers). Mental domains were more greatly affected than physical domains, as reflected by the trend in the summary component scores (PCS
P trend = 0.25, MCS
P trend = 0.001). When compared with participants without visual impairment (excluding participants with correctable visual impairment), participants with mild visual impairment had significantly lower scores in the physical-functioning, vitality, and mental health dimensions (
P = 0.02, 0.001, and 0.01 respectively), and participants with moderate to severe visual impairment had lower scores in role limitation due to physical problems, vitality, social functioning, and mental health and in the mental component score (
P = 0.02, 0.002, 0.001, 0.005, and 0.001, respectively). Similar results were also obtained when participants with correctable visual impairment were included in the reference group.