Impact of Visual Impairment and Eye diseases on Mortality: the Singapore Malay Eye Study (SiMES)

We investigated the relationship of visual impairment (VI) and age-related eye diseases with mortality in a prospective, population-based cohort study of 3,280 Malay adults aged 40–80 years between 2004–2006. Participants underwent a full ophthalmic examination and standardized lens and fundus photographic grading. Visual acuity was measured using logMAR chart. VI was defined as presenting (PVA) and best-corrected (BCVA) visual acuity worse than 0.30 logMAR in the better-seeing eye. Participants were linked with mortality records until 2012. During follow-up (median 7.24 years), 398 (12.2%) persons died. In Cox proportional-hazards models adjusting for relevant factors, participants with VI (PVA) had higher all-cause mortality (hazard ratio[HR], 1.57; 95% confidence interval[CI], 1.25–1.96) and cardiovascular (CVD) mortality (HR 1.75; 95% CI, 1.24–2.49) than participants without. Diabetic retinopathy (DR) was associated with increased all-cause (HR 1.70; 95% CI, 1.25–2.36) and CVD mortality (HR 1.57; 95% CI, 1.05–2.43). Retinal vein occlusion (RVO) was associated with increased CVD mortality (HR 3.14; 95% CI, 1.26–7.73). No significant associations were observed between cataract, glaucoma and age-related macular degeneration with mortality. We conclude that persons with VI were more likely to die than persons without. DR and RVO are markers of CVD mortality.

Age-related eye diseases and VI is likely to increase substantially in Asia in the next few decades [41][42][43] . In this study, we describe the relationship of VI and major age-related eye diseases with all-cause and CVD mortality events in a population-based cohort of Asian Malay persons.

Methods
Study population. The Singapore Malay Eye Study is a population-based, cohort study, in which 3280 (response rate 78.7%) urban Malay adults aged 40 to 79 years living in Singapore were examined between August 2004 and June 2006. Study design and population details have been described elsewhere 44 . A total of 3273 participants were included in this analysis as visual acuity could not be successfully measured in 7 participants. All study procedures were performed in accordance with the tenets of the Declaration of Helsinki as revised in 1989. Written informed consent was obtained from the subjects and the study was approved by the Singhealth Centralised Institutional Review Board (CIRB approval number R1107/9/2014). Participants underwent a standardized interview, systemic and ocular examination, and laboratory investigations at baseline.

Assessment of visual impairment.
Presenting visual acuity (PVA) (with participants wearing their glasses or contact lenses, if any) and best-corrected visual acuity (BCVA) (with subjective refraction conducted by certified optometrists) were measured using the logMAR number chart (Lighthouse International) at a distance of 4 m. If no numbers were read at 4 m, the participants were moved to 3, 2, or 1 m, consecutively. If no numbers were identified, visual acuity was assessed as counting fingers, hand motions, light perception, or no light perception. BCVA and PVA were used in our analysis, although PVA is likely to be more relevant in the practical setting as this reflects the "true" visual acuity of the study population in their daily activities of living. VI was defined as greater than 0.30 logMAR in the better-seeing eye (US definition) 45 .

Assessment of age-related eye diseases.
Complete ophthalmic examinations of the anterior segment, fundus, and optic discs were conducted at the slit lamp using standardized protocols. Cataracts were assessed from lens photographs using the Wisconsin Cataract Grading System 46 and defined as nuclear cataract opacity 4 or greater, cortical cataract 25% or greater, posterior subcapsular cataract 5% or greater, or previous cataract surgery. Glaucoma was diagnosed and classified using the International Society Geographical and Epidemiological Ophthalmology scheme based on gonioscopy, optic disc characteristics, and/or visual fields results 47 . The presence of DR was graded from retinal photographs according to a modification of the Airlie House classification system as used in the Early Treatment Diabetic Retinopathy Study 48 . The presence of AMD was graded from retinal photographs according to the Wisconsin Age-Related Maculopathy Grading System 49 . Recent central retinal vein occlusion (CRVO) was characterized by retinal edema, optic disc hyperemia or edema, scattered superficial and deep retinal hemorrhages, and venous dilation. Old CRVO was characterized by occluded and sheathed retinal veins, or vascular anastomosis at the optic disc 50,51 . Branch retinal vein occlusion (BRVO) involved a more localized area of the retina in the sector of the obstructed venule and was characterized by scattered superficial and deep retinal hemorrhages, venous dilation, intraretinal microvascular abnormalities, and occluded and sheathed retinal venules 50,51 . The presence of any retinal vein occlusion (RVO) was defined as presence of CRVO or BRVO.

Assessment of other risk factors.
Participants underwent a standardized interview for socioeconomic measures (eg, income, education, type of housing), lifestyle risk factors (eg. smoking), medication use, and self-reported history of systemic diseases. Non-fasting venous blood samples were collected for analysis of cholesterol and glucose levels. Hypertension was defined as systolic blood pressure of 140 mm Hg or higher, diastolic blood pressure of 90 mm Hg or higher, or use of antihypertensive medication. Diabetes was defined as a random glucose level of 200 mg/dL (to convert to millimoles per liter, multiply by 0.0555), use of diabetic medication, or a physician's diagnosis of diabetes. Body Mass Index (BMI) was defined as body mass (kg) divided by the square of the height (m 2 ). History of CVD was ascertained through self-reporting (yes or no).

Assessment of mortality.
The unique national registration identity card numbers, together with date of birth and gender, of the 3273 members of the original cohort were matched with mortality records maintained by the National Registry of Diseases Office of Singapore (NRDO). Vital status as of 31 December 2012 was determined for all of the participants. Information on the date of death, all-cause deaths and CVD deaths for the participants were extracted. The underlying cause of death was reported using the International Classification of Diseases 9 codes. Statistical Analysis. Time-to-event was calculated for each participant from the date of examination where visual acuity was measured through December 31, 2012. Cox proportional hazard regression was used to investigate the associations of PVA, BCVA, age-related eye diseases and mortality. We constructed 2 multivariable models: 1) adjusted for age and gender; and 2) adjusted for age, gender, socio-economic status, hypertension, smoking, diabetes, BMI and history of CVD. This permits the evaluation of specific ocular diseases and VI on survival while controlling for other risk factors. Finally, each specific ocular condition was added to the model to determine its independent effect on mortality. Hazard ratios and 95% confidence intervals (CI) were presented for stratified PVA and BCVA groups. We regard P values of < 0.05 from 2-sided tests to indicate statistical significance. We further tested the interaction between DR and VI with mortality by including cross-product interaction term as independent variable (i.e. DR*VI) in the model. Statistical interactions were deemed significant if the P value for interaction was < 0.1, All statistical analyses were performed using STATA version 12.

Results
Baseline Characteristics. In total, 3273 participants were included. Of these, 992 (30.3%) met the definition of VI using PVA and 360 (10.9%) met the definition of VI using BCVA. Table 1 shows the baseline characteristics of participants according to their PVA in the better-seeing eye. Persons with VI tended to be females, have lower socioeconomic status, lower BMI and have a history of CVD, hypertension, diabetes, glaucoma, cataract, retinopathy, AMD, and RVO. Supplementary  Table 2). Persons with VI in terms of presenting VA (VA > 0.30) had poorer overall survival (Fig. 1) and higher CVD-cause mortality, compared with those without (Fig. 2). Similarly, persons with VI in terms of BCVA had higher all-cause and CVD-cause mortality. The results were similar when the WHO cut-off for visual impairment (VA < 6/18 in better-seeing eye) was used for presenting and best-corrected VA (Supplementary Table 2).
Age-related eye diseases and mortality. In terms of specific age-related eye disease (  (Fig. 3) and CVD-cause mortality (Fig. 4) respectively. In addition, RVO was only associated with increased  . None of the other major age-related eye diseases (i.e., cataract, glaucoma and AMD) was statistically significantly associated with mortality. In the stratified analysis (Supplementary Table 3), we found that the association between DR and all-cause mortality was stronger in subjects with VI (p-interaction = 0.035). However, no statistically significant interaction was found with CVD-cause mortality for DR and VI (p-interaction > 0.1).

Discussion
In this prospective population-based study, persons with VI were more likely to die than persons without VI, even while controlling for other clinical and socio-economical risk factors for mortality. In particular, DR and RVO were associated with increased risk of mortality, particularly CVD mortality. Our finding that VI is associated with higher risk of mortality is consistent with that found in several studies in the Western population (Table 4) 12,[15][16][17][18][19] and three in Asian studies [20][21][22] . As the above population-based studies span across different ethnicities, it is now highly convincing that VI has a close linked relationship with mortality. There are some possible explanations to justify this. First, visual acuity is a reflection of functional status and ocular health. Previous studies have clearly demonstrated that VI leads to functional problems and can contribute to loss of independent mobility 2-5 , falls 8 , accidents 7 and psychosocial implications such as depression 6 . These functional problems in turn may contribute to an increased risk of mortality. Second, VI can also be seen as being a marker of aging 52 and therefore it is not surprising that with aging comes higher incidence of mortality. Having said that, studies have not directly proven a causal relationship between VI and mortality, and thus the exact mechanisms for higher mortality rates associated with VI remain unclear. Our study adds to knowledge of the impact of VI not only on morbidity but also mortality. Reduction of VI such as cataract surgery may also be associated with a reduction in risk of mortality. For example, in the BMES, cataract surgery resulting in the correction of VI resulted in lower long-term mortality risk (HR, 0.55; 95% CI, 0.41-0.73) 53 . Further studies are essentially needed.
In our study, we also observed that major age-related retinal vascular conditions such as DR and RVO were associated with CVD mortality, even after adjusting for potential confounders. DR was also associated with higher risk of all-cause mortality. This positive association of DR with mortality is consistent with other cohort studies [54][55][56] . The ETDRS study also found that poor visual acuity was associated with mortality in type 1 diabetics, whereas worsening levels of DR was associated with increasing risk of mortality 57 . This is in line with the hypothesis that microvascular diseases such as retinopathy may reflect the macrovascular status in major organs such as the heart and kidney as they may share similar underlying pathophysiology 58 . The presence of DR has also been associated with other risk factors for CVD including hypertension and hyperlipidaemia 59 . When we performed interaction analysis, we found that the association between DR and all-cause mortality was stronger in subjects with VI, indicating that persons presenting with the coexistence of DR and VI were more likely to have increased risk of all-cause mortality. This further highlights the importance of prompt management of DR, which can be sight saving as well as life-saving. Vigilant screening for DR, which can be conveniently achieved in a non-invasive way by direct visualization of the retinal vessels of the eye, and early diagnosis for RVO may therefore play an important role in risk stratification of CVD. In the presence of retinopathy, prompt management of the ocular condition as well as optimization of associated vascular risk factors will be beneficial in disease control and preserving vision, hence reducing VI, which has also been shown to increase risk of mortality as above. We did not find an association between glaucoma, AMD, cataract and increased mortality risk, which was similar to the negative findings in many other studies.       The strengths of this study include its large sample size, long follow-up and standardized diagnostic protocols for cataract, AMD, DR, RVO and glaucoma. To our knowledge, this is also the first study evaluating the association of VI, age-related eye disease and mortality in an Asian Malay population. Limitations of our study include the possibility of some of the results being due to chance as there are a large number of confounding factors involved and not all may be accounted for. Further studies are also needed to investigate if VI has a direct or indirect causal effect on mortality. Longer duration of follow up may also be useful in confirming these associations. We also did not have data on the time of onset of cataract and time of surgery, which may confound the relationship between the presence of cataract and mortality, and whether cataract surgery had an effect on mortality risk.
In conclusion, persons with VI were more likely to die than persons without VI. The presence of DR and RVO were, in particular, markers of increased mortality and CVD risk.