Acta Ophthalmologica 2014
Cardiovascular risk factors associated with agerelated macular degeneration: the Tromsø Study Maja G Erke,1,2 Geir Bertelsen,1,2 Tunde Peto,3 Anne K Sjølie,4,5 Haakon Lindekleiv2 and Inger Njølstad2 1
Department of Ophthalmology and Neurosurgery, University Hospital of North Norway, Tromsø, Norway Research Group of Epidemiology of Chronic Diseases, Department of Community Medicine, Faculty of Health Sciences, University of Tromsø, Tromsø, Norway 3 NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, Head of Reading Centre, London, UK 4 Department of Ophthalmology, Odense University Hospital, Odense, Denmark 5 Brain and Circulation Research Group, Department of Clinical Medicine, Faculty of Health Sciences, University of Tromsø, Tromsø, Norway 2
ABSTRACT. Purpose: To examine associations between cardiovascular risk factors and agerelated macular degeneration (AMD). Methods: A population-based, cross-sectional study of Caucasians aged 65–87 years was conducted in Norway in 2007/2008. Retinal photographs were graded for AMD. Multivariable logistic regression analyses were performed based on questionnaires addressing habits of smoking, alcohol consumption, physical activity, health and medication; and physical examination comprising anthropometric measurements, blood pressure and blood sampling. Cardiovascular disease status was obtained from a validated end-point registry. Results: Gradable photographs were available for 2631 participants, of whom 92 (3.5%) subjects had late AMD. In the multivariable analysis of late AMD, significant interactions were found between sex and the variables age, triglyceride level, use of lipid-lowering drugs and physical exercise. Current daily smoking was significantly related to late AMD in both sexes (odds ratio (OR) 4.06, 95% confidence interval (CI) 1.69–9.76 and OR 3.59, 95% CI 1.17–11.04, women and men, respectively) compared with never smokers. Higher number of pack years was associated with the presence of large drusen (>125 lm) (OR 1.04, 95% CI 1.01– 1.09 per 5 years). Higher systolic blood pressure (OR 1.06, 95% CI 1.01–1.12 per 5 mmHg), overweight (OR 2.87, 95% CI 1.13–7.29) and obesity (OR 2.92, 95% CI 1.06–8.03), physical exercise duration (OR 0.41, 95% 0.18–0.96 for 30 min or more compared with less) and frequency (OR 0.46, 95% CI 0.23–0.92 for weekly or more often compared to less) were associated with late AMD in women only. Conclusions: Smoking was strongly associated with AMD, in line with results from other populations. Also, late AMD was related to higher systolic blood pressure, physical inactivity, overweight and obesity in women. Key words: epidemiology – age-related macular degeneration – smoking – risk factors – physical activity – hypertension – body mass index – sex difference
Acta Ophthalmol. 2014: 92: 662–669 ª 2014 Acta Ophthalmologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
doi: 10.1111/aos.12346
662
Introduction Among elderly people, age-related macular degeneration (AMD) is a major threat to vision worldwide (Resnikoff et al. 2004; Rudnicka et al. 2012). In Denmark, blindness due to AMD has declined in the years 2000 through 2010 (Bloch et al. 2012). The Danes share genetic origin with and are comparable to Norwegians socioculturally. The authors suggested that the new anti-angiogenic treatments for neovascular AMD could be partly responsible for the decline. These treatments have, however, imposed a considerable economic and logistic burden on the healthcare system due to high numbers of patients needing frequent intravitreal injections (Day et al. 2011). Currently, no available treatment exists for the dry end stage of AMD, geographical atrophy. These aspects combined with increasing longevity and ageing populations are urgent incentives for detecting potential modifiable risk factors to reduce the future pressure of this blinding disease (Lindekleiv & Erke 2012). Due to the similarities between drusen deposition and the development of atherosclerosis (Curcio et al. 2011), traditional cardiovascular risk factors have been suggested as risk factors for AMD. Persons with AMD are reported to have increased risk of stroke and
Acta Ophthalmologica 2014
coronary heart disease in some studies (Tan et al. 2008; Hu et al. 2010; Wieberdink et al. 2011), but previous epidemiologic studies evaluating risk factors for cardiovascular disease and AMD have not yielded consistent relationships or strong associations (Connell et al. 2009). The exception is the strong evidence for increasing age and smoking as factors common for both diseases (Connell et al. 2009; Chakravarthy et al. 2010; Rudnicka et al. 2012). There are some evidences for high blood pressure (Age-related Eye Disease Study Group 2000; Klein et al. 2003; van Leeuwen et al. 2003) and obesity (Delcourt et al. 2001; Seddon et al. 2003) as risk factors for AMD. Sex, dyslipidaemia and diabetes have also been associated with AMD in some, but not all studies (Hyman et al. 2000; van Leeuwen et al. 2003; Clemons et al. 2005; Klein et al. 2007; Cackett et al. 2008; Tan et al. 2008; Connell et al. 2009; Chakravarthy et al. 2010; Adams et al. 2011; Rudnicka et al. 2012). A few studies have shown a protective effect of physical activity (Knudtson et al. 2006; Mares et al. 2011). Lately, an increasing body of evidence has shown that genetic susceptibility plays an important role in the development of AMD (Connell et al. 2009). Still, the aetiology is not fully understood and further studies on modifiable risk factors are warranted. Risk factors for AMD in populations living in the high north are scarcely described in the literature. The Norwegian population share common ancestry with western Europeans, such as the populations of Beaver Dam and Rotterdam, but the living conditions are different. The climate in Tromsø is subarctic with large variation in seasonal sunlight exposure. We have previously published prevalence data from this population (Erke et al. 2012). The aim of this study was to examine the relationship between traditional cardiovascular risk factors and AMD in a cross-sectional study among elderly subjects living above the Arctic Circle in Tromsø, Norway.
Material and Methods Study population
The Tromsø Study is a prospective, population-based cohort study, with participants from a homogeneous, Cau-
casian population (additional information available at: http://www.tromso study.com). Its initial focus was cardiovascular disease, eventually including a spectrum of lifestyle-related and chronic diseases. The design includes repeated health surveys, and the 6th survey was conducted in 2007/2008. All residents in the municipality of Tromsø aged 60– 87 years and random samples from the group aged 30–59 years were invited to the first of two visits in the 6th survey. Preselected participants who attended the 1st visit were invited to more extensive examinations in a 2nd visit. The groups invited for the 2nd visit were as follows: all subjects aged 50–62 years or aged 75–84 years (n = 7657), a 20% random sample aged 63–74 years (n = 942) and subjects, if not already included in the two groups above, who had attended the 2nd visit of the 4th survey (n = 2885). The participation rate in the 2nd visit was 92%, with a total of 7307 subjects participating. The study population in this paper included 3025 participants aged ≥65 years from the 2nd visit of the 6th Tromsø Study. Detailed sampling procedure is described elsewhere (Eggen et al. 2013). The 6th survey of the Tromsø Study and cross-sectional studies originating from it were approved by the Regional Committee for Medical and Health Research Ethics and the Data Inspectorate. The study followed the tenets of the Helsinki Declaration. All participants have given written, informed consent. Cardiovascular risk factors
Methods for the Tromsø Study have been described (Jacobsen et al. 2012). Self-reported questionnaires were answered before and after the examination addressing life style habits, health and use of medication. A standardized physical examination was performed by trained personnel. Weight, height and circumference of waist and hip were measured. Blood pressure was recorded with a fully automatic device (Dinamap Pro care 300 Monitor, GE Healthcare, Norway). The average of the last two of three recordings was used. Non-fasting blood samples were analysed for glycosylated haemoglobin A1c (HbA1c) and serum lipids (total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL)
cholesterol and triglycerides) by enzymatic colorimetric methods at the Department of Clinical Chemistry, University Hospital of North Norway. Body mass index (BMI) was calculated as weight in kilograms divided by the square of the height in metres and categorized as underweight 6.5% or current use of insulin or antidiabetic tablets. We defined cardiovascular disease as heart attack or stroke. The Tromsø Study runs a continuously updated and validated end-point registry of cardiovascular events. All first time cardiovascular events are recorded, and data are confirmed by hospital records and data from the Cause of Death registry. Data from this registry were linked to the study sample for identification and verification of subjects with cardiovascular disease. Data on self-reported cardiovascular disease were available but not reported as it did not provide additional information. All other variables were used as recorded in the Tromsø Study database (available at: http://tromsoundersokelsen.uit.no/ tromso/). Eye examinations
Corrected Snellen visual acuity and refractive error were obtained by trained personnel. Digital retinal photography was performed in pharmacological mydriasis. Photographs were graded for AMD based on the features described in the international classifi-
663
Acta Ophthalmologica 2014
cation system (Bird et al. 1995). The grading was standardized and masked from all clinical information, sex and age. Detailed procedure has been described elsewhere (Bertelsen et al. 2012; Erke et al. 2012). Briefly, images were graded for the presence of macular drusen, drusen size and late AMD features (choroidal neovascularization and/or geographical atrophy). Pigment abnormalities were not graded. The most severe features present within 3 mm from the fovea determined the predominant phenotype of that eye. Individuals with asymmetric phenotype between eyes were classified according to the more severe phenotype of the two eyes. Phenotypes were categorized for statistical analysis as (i) normal or drusen ≤125 lm (large drusen), (ii) drusen >125 lm or (iii) late AMD. Visual impairment was defined as Snellen ≤0.32 in the better-seeing eye. Statistical analysis and data management
We used independent sample t-tests and chi-squared tests, as appropriate, to assess differences between participants with and without gradable photographs and in univariable analyses between large drusen or late AMD, respectively, and the risk factors. Late AMD cases were combined with large drusen for the analysis of associations between large drusen and the risk factors. We performed a stepwise backward selection procedure with late AMD as the outcome in a multivariable logistic regression model including sex and all variables with p-values < 0.25 from univariable analyses. The variable with highest p-value was removed for each step, except for sex. Next, variables were singly added to the reduced model and the model assessed by Hosmer–Lemeshow chisquared tests and 2 log-likelihood tests. Interactions were assessed between smoking and systolic blood pressure; and between smoking and BMI; and between smoking and age; further between sex and all variables listed in Tables 2 through 4 except waist circumference. Significant interaction terms were evaluated multiplicatively by 2 log-likelihood test. The same multivariate model was applied to associations with large drusen. We calculated odds ratios (ORs) with 95% confidence intervals (CIs) describing the relationship between late AMD
664
or large drusen and the variables. The assumptions of logistic regression were tested. All data were analysed in STATA/SE 12.0 (StataCorp LP, TX, USA).
Results Subjects
Among the 3025 participants, 372 subjects had no photographs and 22 subjects had ungradable photographs from both eyes. Included in the study sample were 2631 participants with photographs gradable for AMD. Females constituted 57% of the sample overall. In the age group 65–69 years, they constituted 62%, in the age group 70– 74 years 52%, in the age group 75–79 years 56% and in the age group 80–87 years 61%. The crude prevalence of late AMD was 3.5% and 10.9% among participants aged 80 years and older. Large drusen were found in 24% of the sample, excluding late AMD subjects (Erke et al. 2012). Participants not providing photographs were older (73.8 versus 72.3 years, p < 0.0001) and had lower visual acuity (Snellen ≤0.32 in 6.4% versus 2.2%, p < 0.01), had higher frequency of diabetes (13.8% versus 9.6%, p < 0.01) and cardiovascular disease (18.5% versus 13.4%, p < 0.01) and were less physically active (24.0% versus 34.5%, p < 0.001) than participants with gradable photographs. No difference between these two groups was observed with respect to education, blood pressure, BMI, daily smoking and serum lipids (data not shown). Participant characteristics are presented in Table 1. Among those with gradable photographs, men and women differed with respect to age (p < 0.01), smoking habits (p < 0.001), systolic blood pressure (p < 0.0001), physical activity (p < 0.001), serum cholesterol (p < 0.0001) and cardiovascular disease (p < 0.001). Large drusen
Adjusted ORs for large drusen (>125 lm) are shown in Table 2. Two models were used. Model 1 includes adjustment for age and sex, and model 2 includes in addition smoking and systolic blood pressure. Older age and number of pack years smoked showed significant positive relationships with
large drusen in model 1. We also found exercising for duration of 30 min or more compared with less when subjects were physically active, to be significantly protective for having large drusen. These associations remained significant with additional adjustments in model 2 (OR 1.10, 95% CI 1.09–1.12 for age, OR 1.04, 95% CI 1.01–1.09 per 5 pack years and OR 0.77, 95% CI 0.60–0.98 for duration of exercise), also when adjusting further for visual impairment. In model 2, alcohol consumption was a protective factor when consuming monthly (OR 0.78, CI 95% 0.62–0.98) or weekly (OR 0.73, 95% CI 0.54–0.99) compared with no alcohol intake. Further, HDL cholesterol and triglycerides were borderline significantly associated with large drusen in the multivariable analyses (p = 0.07). When we applied the Bonferroni method to the results to allow for multiple comparisons, only age remained significantly related to large drusen. Late AMD
Adjusted ORs for late AMD are presented in Tables 3 and 4, each with two different models. Model 1 included only adjustment for age. Model 2 also included adjustment for smoking and systolic blood pressure. In both models, we observed a significant interaction between sex and age. The p-value was 0.023 in model 2 and inclusion of the interaction term altered the log likelihood (p = 0.022). Three other interaction terms including the variable sex were also significant in model 2: the interaction between sex and use of lipidlowering drugs (p = 0.038), sex and serum triglyceride levels (p = 0.044), and sex and exercise duration of 30 min (p = 0.017). Consequently, and because some cardiovascular risk factors have been shown to act differently in women and men, results for late AMD were presented for women in Table 3 and men in Table 4. Besides older age, current daily smoking was a strong risk factor for both women and men in the multivariable analyses. Odds ratios were 4.06 (95% CI 1.69–9.76) for women and 3.59 (95% CI 1.17–11.04) for men. The association with former smoking was not significant for men and borderline significant for women (p = 0.06). Higher systolic blood pressure (OR 1.06, 95% CI 1.01–1.12 per 5 mmHg
Acta Ophthalmologica 2014
Table 1. Participant characteristics by sex. The Tromsø Study 2007–2008. Participants with non-missing data N Age-related maculopathy status, worse eye Drusen >125 lm 2631 Late AMD 2631 Visual acuity ≤0.32, Snellen 2486 Age, years 2631 Daily smoking 2568 Never Former Current Body mass index, kg/m2, 2629 continuous Body mass index, kg/m2 2629 Normal (20.0–24.9) Underweight (88 cm in 2560 women and >102 cm in men Systolic blood pressure, mmHg 2614 Diastolic blood pressure, mmHg 2614 Serum lipids, mM Total cholesterol 2603 HDL-C 2603 LDL-C 2603 Triglycerides 2603 Ever used lipid-lowering drugs 2546 Physically active† 2478 Diabetes* 2605 Verified stroke 2631 Verified myocardial infarction 2631 Verified cardiovascular event 2631 (stroke/myocardial infarction)
Women N = 1512
Men N = 1119
378 48 36 72.2
(25%) (3.2%) (2.5%) (5.6)
257 44 19 72.5
(23%) (3.9%) (1.8%) (5.1)
704 556 212 27.1
(48%) (38%) (14%) (4.5)
268 686 142 27.0
(24%) (63%) (13%) (3.5)
0.41
479 49 623 360 917
(32%) (3%) (41%) (24%) (62%)
316 12 591 199 435
(28%) (1%) (53%) (18%) (40%)
-
151 (25) 76 (11)
147 (22) 80 (10)
Cardiovascular risk factors associated with agerelated macular degeneration: the Tromsø Study Maja G Erke,1,2 Geir Bertelsen,1,2 Tunde Peto,3 Anne K Sjølie,4,5 Haakon Lindekleiv2 and Inger Njølstad2 1
Department of Ophthalmology and Neurosurgery, University Hospital of North Norway, Tromsø, Norway Research Group of Epidemiology of Chronic Diseases, Department of Community Medicine, Faculty of Health Sciences, University of Tromsø, Tromsø, Norway 3 NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, Head of Reading Centre, London, UK 4 Department of Ophthalmology, Odense University Hospital, Odense, Denmark 5 Brain and Circulation Research Group, Department of Clinical Medicine, Faculty of Health Sciences, University of Tromsø, Tromsø, Norway 2
ABSTRACT. Purpose: To examine associations between cardiovascular risk factors and agerelated macular degeneration (AMD). Methods: A population-based, cross-sectional study of Caucasians aged 65–87 years was conducted in Norway in 2007/2008. Retinal photographs were graded for AMD. Multivariable logistic regression analyses were performed based on questionnaires addressing habits of smoking, alcohol consumption, physical activity, health and medication; and physical examination comprising anthropometric measurements, blood pressure and blood sampling. Cardiovascular disease status was obtained from a validated end-point registry. Results: Gradable photographs were available for 2631 participants, of whom 92 (3.5%) subjects had late AMD. In the multivariable analysis of late AMD, significant interactions were found between sex and the variables age, triglyceride level, use of lipid-lowering drugs and physical exercise. Current daily smoking was significantly related to late AMD in both sexes (odds ratio (OR) 4.06, 95% confidence interval (CI) 1.69–9.76 and OR 3.59, 95% CI 1.17–11.04, women and men, respectively) compared with never smokers. Higher number of pack years was associated with the presence of large drusen (>125 lm) (OR 1.04, 95% CI 1.01– 1.09 per 5 years). Higher systolic blood pressure (OR 1.06, 95% CI 1.01–1.12 per 5 mmHg), overweight (OR 2.87, 95% CI 1.13–7.29) and obesity (OR 2.92, 95% CI 1.06–8.03), physical exercise duration (OR 0.41, 95% 0.18–0.96 for 30 min or more compared with less) and frequency (OR 0.46, 95% CI 0.23–0.92 for weekly or more often compared to less) were associated with late AMD in women only. Conclusions: Smoking was strongly associated with AMD, in line with results from other populations. Also, late AMD was related to higher systolic blood pressure, physical inactivity, overweight and obesity in women. Key words: epidemiology – age-related macular degeneration – smoking – risk factors – physical activity – hypertension – body mass index – sex difference
Acta Ophthalmol. 2014: 92: 662–669 ª 2014 Acta Ophthalmologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
doi: 10.1111/aos.12346
662
Introduction Among elderly people, age-related macular degeneration (AMD) is a major threat to vision worldwide (Resnikoff et al. 2004; Rudnicka et al. 2012). In Denmark, blindness due to AMD has declined in the years 2000 through 2010 (Bloch et al. 2012). The Danes share genetic origin with and are comparable to Norwegians socioculturally. The authors suggested that the new anti-angiogenic treatments for neovascular AMD could be partly responsible for the decline. These treatments have, however, imposed a considerable economic and logistic burden on the healthcare system due to high numbers of patients needing frequent intravitreal injections (Day et al. 2011). Currently, no available treatment exists for the dry end stage of AMD, geographical atrophy. These aspects combined with increasing longevity and ageing populations are urgent incentives for detecting potential modifiable risk factors to reduce the future pressure of this blinding disease (Lindekleiv & Erke 2012). Due to the similarities between drusen deposition and the development of atherosclerosis (Curcio et al. 2011), traditional cardiovascular risk factors have been suggested as risk factors for AMD. Persons with AMD are reported to have increased risk of stroke and
Acta Ophthalmologica 2014
coronary heart disease in some studies (Tan et al. 2008; Hu et al. 2010; Wieberdink et al. 2011), but previous epidemiologic studies evaluating risk factors for cardiovascular disease and AMD have not yielded consistent relationships or strong associations (Connell et al. 2009). The exception is the strong evidence for increasing age and smoking as factors common for both diseases (Connell et al. 2009; Chakravarthy et al. 2010; Rudnicka et al. 2012). There are some evidences for high blood pressure (Age-related Eye Disease Study Group 2000; Klein et al. 2003; van Leeuwen et al. 2003) and obesity (Delcourt et al. 2001; Seddon et al. 2003) as risk factors for AMD. Sex, dyslipidaemia and diabetes have also been associated with AMD in some, but not all studies (Hyman et al. 2000; van Leeuwen et al. 2003; Clemons et al. 2005; Klein et al. 2007; Cackett et al. 2008; Tan et al. 2008; Connell et al. 2009; Chakravarthy et al. 2010; Adams et al. 2011; Rudnicka et al. 2012). A few studies have shown a protective effect of physical activity (Knudtson et al. 2006; Mares et al. 2011). Lately, an increasing body of evidence has shown that genetic susceptibility plays an important role in the development of AMD (Connell et al. 2009). Still, the aetiology is not fully understood and further studies on modifiable risk factors are warranted. Risk factors for AMD in populations living in the high north are scarcely described in the literature. The Norwegian population share common ancestry with western Europeans, such as the populations of Beaver Dam and Rotterdam, but the living conditions are different. The climate in Tromsø is subarctic with large variation in seasonal sunlight exposure. We have previously published prevalence data from this population (Erke et al. 2012). The aim of this study was to examine the relationship between traditional cardiovascular risk factors and AMD in a cross-sectional study among elderly subjects living above the Arctic Circle in Tromsø, Norway.
Material and Methods Study population
The Tromsø Study is a prospective, population-based cohort study, with participants from a homogeneous, Cau-
casian population (additional information available at: http://www.tromso study.com). Its initial focus was cardiovascular disease, eventually including a spectrum of lifestyle-related and chronic diseases. The design includes repeated health surveys, and the 6th survey was conducted in 2007/2008. All residents in the municipality of Tromsø aged 60– 87 years and random samples from the group aged 30–59 years were invited to the first of two visits in the 6th survey. Preselected participants who attended the 1st visit were invited to more extensive examinations in a 2nd visit. The groups invited for the 2nd visit were as follows: all subjects aged 50–62 years or aged 75–84 years (n = 7657), a 20% random sample aged 63–74 years (n = 942) and subjects, if not already included in the two groups above, who had attended the 2nd visit of the 4th survey (n = 2885). The participation rate in the 2nd visit was 92%, with a total of 7307 subjects participating. The study population in this paper included 3025 participants aged ≥65 years from the 2nd visit of the 6th Tromsø Study. Detailed sampling procedure is described elsewhere (Eggen et al. 2013). The 6th survey of the Tromsø Study and cross-sectional studies originating from it were approved by the Regional Committee for Medical and Health Research Ethics and the Data Inspectorate. The study followed the tenets of the Helsinki Declaration. All participants have given written, informed consent. Cardiovascular risk factors
Methods for the Tromsø Study have been described (Jacobsen et al. 2012). Self-reported questionnaires were answered before and after the examination addressing life style habits, health and use of medication. A standardized physical examination was performed by trained personnel. Weight, height and circumference of waist and hip were measured. Blood pressure was recorded with a fully automatic device (Dinamap Pro care 300 Monitor, GE Healthcare, Norway). The average of the last two of three recordings was used. Non-fasting blood samples were analysed for glycosylated haemoglobin A1c (HbA1c) and serum lipids (total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL)
cholesterol and triglycerides) by enzymatic colorimetric methods at the Department of Clinical Chemistry, University Hospital of North Norway. Body mass index (BMI) was calculated as weight in kilograms divided by the square of the height in metres and categorized as underweight 6.5% or current use of insulin or antidiabetic tablets. We defined cardiovascular disease as heart attack or stroke. The Tromsø Study runs a continuously updated and validated end-point registry of cardiovascular events. All first time cardiovascular events are recorded, and data are confirmed by hospital records and data from the Cause of Death registry. Data from this registry were linked to the study sample for identification and verification of subjects with cardiovascular disease. Data on self-reported cardiovascular disease were available but not reported as it did not provide additional information. All other variables were used as recorded in the Tromsø Study database (available at: http://tromsoundersokelsen.uit.no/ tromso/). Eye examinations
Corrected Snellen visual acuity and refractive error were obtained by trained personnel. Digital retinal photography was performed in pharmacological mydriasis. Photographs were graded for AMD based on the features described in the international classifi-
663
Acta Ophthalmologica 2014
cation system (Bird et al. 1995). The grading was standardized and masked from all clinical information, sex and age. Detailed procedure has been described elsewhere (Bertelsen et al. 2012; Erke et al. 2012). Briefly, images were graded for the presence of macular drusen, drusen size and late AMD features (choroidal neovascularization and/or geographical atrophy). Pigment abnormalities were not graded. The most severe features present within 3 mm from the fovea determined the predominant phenotype of that eye. Individuals with asymmetric phenotype between eyes were classified according to the more severe phenotype of the two eyes. Phenotypes were categorized for statistical analysis as (i) normal or drusen ≤125 lm (large drusen), (ii) drusen >125 lm or (iii) late AMD. Visual impairment was defined as Snellen ≤0.32 in the better-seeing eye. Statistical analysis and data management
We used independent sample t-tests and chi-squared tests, as appropriate, to assess differences between participants with and without gradable photographs and in univariable analyses between large drusen or late AMD, respectively, and the risk factors. Late AMD cases were combined with large drusen for the analysis of associations between large drusen and the risk factors. We performed a stepwise backward selection procedure with late AMD as the outcome in a multivariable logistic regression model including sex and all variables with p-values < 0.25 from univariable analyses. The variable with highest p-value was removed for each step, except for sex. Next, variables were singly added to the reduced model and the model assessed by Hosmer–Lemeshow chisquared tests and 2 log-likelihood tests. Interactions were assessed between smoking and systolic blood pressure; and between smoking and BMI; and between smoking and age; further between sex and all variables listed in Tables 2 through 4 except waist circumference. Significant interaction terms were evaluated multiplicatively by 2 log-likelihood test. The same multivariate model was applied to associations with large drusen. We calculated odds ratios (ORs) with 95% confidence intervals (CIs) describing the relationship between late AMD
664
or large drusen and the variables. The assumptions of logistic regression were tested. All data were analysed in STATA/SE 12.0 (StataCorp LP, TX, USA).
Results Subjects
Among the 3025 participants, 372 subjects had no photographs and 22 subjects had ungradable photographs from both eyes. Included in the study sample were 2631 participants with photographs gradable for AMD. Females constituted 57% of the sample overall. In the age group 65–69 years, they constituted 62%, in the age group 70– 74 years 52%, in the age group 75–79 years 56% and in the age group 80–87 years 61%. The crude prevalence of late AMD was 3.5% and 10.9% among participants aged 80 years and older. Large drusen were found in 24% of the sample, excluding late AMD subjects (Erke et al. 2012). Participants not providing photographs were older (73.8 versus 72.3 years, p < 0.0001) and had lower visual acuity (Snellen ≤0.32 in 6.4% versus 2.2%, p < 0.01), had higher frequency of diabetes (13.8% versus 9.6%, p < 0.01) and cardiovascular disease (18.5% versus 13.4%, p < 0.01) and were less physically active (24.0% versus 34.5%, p < 0.001) than participants with gradable photographs. No difference between these two groups was observed with respect to education, blood pressure, BMI, daily smoking and serum lipids (data not shown). Participant characteristics are presented in Table 1. Among those with gradable photographs, men and women differed with respect to age (p < 0.01), smoking habits (p < 0.001), systolic blood pressure (p < 0.0001), physical activity (p < 0.001), serum cholesterol (p < 0.0001) and cardiovascular disease (p < 0.001). Large drusen
Adjusted ORs for large drusen (>125 lm) are shown in Table 2. Two models were used. Model 1 includes adjustment for age and sex, and model 2 includes in addition smoking and systolic blood pressure. Older age and number of pack years smoked showed significant positive relationships with
large drusen in model 1. We also found exercising for duration of 30 min or more compared with less when subjects were physically active, to be significantly protective for having large drusen. These associations remained significant with additional adjustments in model 2 (OR 1.10, 95% CI 1.09–1.12 for age, OR 1.04, 95% CI 1.01–1.09 per 5 pack years and OR 0.77, 95% CI 0.60–0.98 for duration of exercise), also when adjusting further for visual impairment. In model 2, alcohol consumption was a protective factor when consuming monthly (OR 0.78, CI 95% 0.62–0.98) or weekly (OR 0.73, 95% CI 0.54–0.99) compared with no alcohol intake. Further, HDL cholesterol and triglycerides were borderline significantly associated with large drusen in the multivariable analyses (p = 0.07). When we applied the Bonferroni method to the results to allow for multiple comparisons, only age remained significantly related to large drusen. Late AMD
Adjusted ORs for late AMD are presented in Tables 3 and 4, each with two different models. Model 1 included only adjustment for age. Model 2 also included adjustment for smoking and systolic blood pressure. In both models, we observed a significant interaction between sex and age. The p-value was 0.023 in model 2 and inclusion of the interaction term altered the log likelihood (p = 0.022). Three other interaction terms including the variable sex were also significant in model 2: the interaction between sex and use of lipidlowering drugs (p = 0.038), sex and serum triglyceride levels (p = 0.044), and sex and exercise duration of 30 min (p = 0.017). Consequently, and because some cardiovascular risk factors have been shown to act differently in women and men, results for late AMD were presented for women in Table 3 and men in Table 4. Besides older age, current daily smoking was a strong risk factor for both women and men in the multivariable analyses. Odds ratios were 4.06 (95% CI 1.69–9.76) for women and 3.59 (95% CI 1.17–11.04) for men. The association with former smoking was not significant for men and borderline significant for women (p = 0.06). Higher systolic blood pressure (OR 1.06, 95% CI 1.01–1.12 per 5 mmHg
Acta Ophthalmologica 2014
Table 1. Participant characteristics by sex. The Tromsø Study 2007–2008. Participants with non-missing data N Age-related maculopathy status, worse eye Drusen >125 lm 2631 Late AMD 2631 Visual acuity ≤0.32, Snellen 2486 Age, years 2631 Daily smoking 2568 Never Former Current Body mass index, kg/m2, 2629 continuous Body mass index, kg/m2 2629 Normal (20.0–24.9) Underweight (88 cm in 2560 women and >102 cm in men Systolic blood pressure, mmHg 2614 Diastolic blood pressure, mmHg 2614 Serum lipids, mM Total cholesterol 2603 HDL-C 2603 LDL-C 2603 Triglycerides 2603 Ever used lipid-lowering drugs 2546 Physically active† 2478 Diabetes* 2605 Verified stroke 2631 Verified myocardial infarction 2631 Verified cardiovascular event 2631 (stroke/myocardial infarction)
Women N = 1512
Men N = 1119
378 48 36 72.2
(25%) (3.2%) (2.5%) (5.6)
257 44 19 72.5
(23%) (3.9%) (1.8%) (5.1)
704 556 212 27.1
(48%) (38%) (14%) (4.5)
268 686 142 27.0
(24%) (63%) (13%) (3.5)
0.41
479 49 623 360 917
(32%) (3%) (41%) (24%) (62%)
316 12 591 199 435
(28%) (1%) (53%) (18%) (40%)
-
151 (25) 76 (11)
147 (22) 80 (10)
