20/20--Alcohol and Age-Related Macular Degeneration
20/20--Alcohol and Age-Related Macular Degeneration
The Melbourne Collaborative Cohort Study (MCCS) is a volunteer-based prospective cohort study of 41,514 people of white European descent. Almost all (99.3%) participants were aged 40–69 years at baseline (1990–1994), with approximately equal proportions of participants across the 3 age decades. Follow-up of this cohort occurred during 2003–2007 when participants were aged 48–86 years. The MCCS was approved by the human research and ethics committees of the Cancer Council Victoria and Royal Victorian Eye and Ear Hospital, Australia.
Participants underwent a structured face-to-face interview where they were asked if they had ever drunk at least 12 alcoholic drinks in a year in their lifetime. Those who answered "no" were considered lifetime abstainers. Participants who answered "yes" were asked about their current average quantity (number of glasses of wine, cans or bottles of beer, nips of spirit) and frequency of alcohol intake. These are henceforth referred to as the "beverage-specific frequency-quantity questions." A can or bottle of regular beer constituted 15 g of alcohol, a glass of wine constituted 15 g of alcohol, and a nip (30 mL) of spirit constituted 10 g of alcohol (Australian food composition tables). The total alcohol intake in grams per day from wine, beer, and spirits was then computed from this information. Participants who were not lifetime abstainers but did not consume alcohol at baseline were classified as former drinkers.
In addition, participants were asked about the intake of alcoholic beverages on each day during the week before the interview (this method will be referred to as the "diary"). All analyses except those for pattern of consumption are based on the beverage-specific frequency-quantity questions. Analyses of pattern of consumption are based upon the diary because the frequency-quantity questions did not ask about the frequency of consumption of all alcoholic beverages combined; that is, for example, if twice weekly wine consumption occurred on the same days as twice weekly beer consumption.
Comprehensive questionnaires regarding lifestyle, dietary intakes, and health conditions were completed at baseline. Participants were classified according to their smoking habits as never smokers, former smokers, and current smokers.
Food intake in the year before baseline was estimated by using a 121-item food frequency questionnaire specifically developed for the MCCS. The energy derived from food was calculated by using standard sex-specific portion sizes from Australian food composition tables.
Detection of age-related macular degeneration and other fundus pathology was conducted at follow-up by using digital photographs centered on the macula and optic disc, as described previously. Grading and quality-control procedures of the photographs have been described in detail elsewhere. Early AMD was defined as the presence of from 63- to 125-μm drusen with the presence of hyper/hypopigmentation or ≥125-μm drusen, with or without the presence of hyper/hypopigmentation. Late AMD was defined as evidence of choroidal neovascularization, geographic atrophy (≥175 µm of hypopigmentation with visible choroidal vessels), or a disciform scar. The fundoscopic photos were interpreted blindly with respect to baseline traits.
Total alcohol intake was categorized into lifetime abstainers, former drinkers, current drinkers consuming 1–19 g/day, and current drinkers consuming 20 or more g/day. An additional variable consisting of the categories—lifetime abstainers, former drinkers, those consuming 1–19 g/day, 20–39 g/day, 40–59 g/day, and ≥60 g per day—was used to explore higher intakes of alcohol consumption.
Former drinkers may have changed their dietary habits and lifestyle. Therefore, they were included in the analysis as a distinct group (separate from lifetime abstainers). Wine, spirit, and beer consumptions were categorized similarly to total alcohol. Associations for each beverage were evaluated by including consumption of wine, beer, and spirits in separate models as they were highly collinear. The reference group in each analysis consisted of lifetime abstainers.
The total alcohol intake during the week before interview was highly correlated with the average daily intake from the frequency-quantity questions (Spearman's correlation coefficient = 0.87). Additionally, the pattern of consumption (number of days/week) was also highly correlated with the quantity of alcohol consumption (Spearman's correlation coefficient = 0.83), where those with higher intakes were also more likely to drink more frequently. Therefore, to examine weekly patterns of consumption, the participants were stratified into quantities of consumption of 1–139 g, 140–279 g, and ≥280 g/week, and the odds for those drinking on 1–3 days or 4–6 days a week were compared with those drinking every day within each stratum. Participants with no intake of alcohol during the week before baseline were excluded from the analysis of pattern of consumption.
Chi-squared, t tests, and multivariable logistic regression were used to compare baseline demographics and other factors associated with inclusion or exclusion from the analysis.
Multivariable logistic regression was also used to calculate odds ratios for alcohol consumption at baseline, with the presence or absence of AMD at follow-up as the outcome. A directed acyclic graph was used to select variables to include in the models. Under the assumptions made in this graph, to estimate the effect of alcohol intake on AMD, the following variables need to be included in the logistic regression model: age, sex, country of birth, physical activity, smoking status, and dietary energy intake. Although the waist/hip ratio is associated with AMD in this cohort, it was not included in the regression model as we consider it to be on the causal pathway between alcohol and AMD. Smoking was grouped into the categories current, past, or never, and country of birth was grouped into the countries Australia/United Kingdom, Italy, and Greece. Four ordered categories combining frequency and intensity were used to group physical activity.
Effect modification by sex, smoking, and country of birth was assessed by fitting interaction terms between each of these variables and alcohol intake, and it was tested by using the likelihood ratio test. Linear associations between alcohol intake and the odds of AMD were investigated by comparing regression models with alcohol intake as a categorical variable and a pseudo-continuous variable by using the likelihood ratio test. All statistical analyses were performed by using Stata, version 10, statistical software (StataCorp LP, College Station, Texas).
Participation at the MCCS follow-up (2003–2007) was 67.2% (Figure 1). A total of 13,612 participants did not participate in follow-up due to death, illness, refusal, leaving Victoria or Australia, or unknown reasons. A further 5,483 were unable to be photographed because of the lack of photographic facilities at some of the clinics. Ophthalmic data from 1,119 participants were excluded from the analysis because the photographs were not able to be graded (Figure 1). The mean time between baseline when dietary and alcohol data were collected and the time of eye photography was 11.5 years (standard deviation, 1.4 years; range, 8.6–16.4). Participants were excluded if they reported extreme energy intakes (n = 324), that is, less than the 1st percentile and more than the 99th percentile of the total MCCS cohort of 41,501. The total number of participants included in the final analysis was 20,963.
(Enlarge Image)
Figure 1.
Flow chart of participants in the Melbourne Collaborative Cohort Study (MCCS), Australia, 1990–2007. AMD, age-related macular degeneration.
Materials and Methods
Study Population
The Melbourne Collaborative Cohort Study (MCCS) is a volunteer-based prospective cohort study of 41,514 people of white European descent. Almost all (99.3%) participants were aged 40–69 years at baseline (1990–1994), with approximately equal proportions of participants across the 3 age decades. Follow-up of this cohort occurred during 2003–2007 when participants were aged 48–86 years. The MCCS was approved by the human research and ethics committees of the Cancer Council Victoria and Royal Victorian Eye and Ear Hospital, Australia.
Baseline Assessment of Alcohol Consumption, Smoking, and Diet
Participants underwent a structured face-to-face interview where they were asked if they had ever drunk at least 12 alcoholic drinks in a year in their lifetime. Those who answered "no" were considered lifetime abstainers. Participants who answered "yes" were asked about their current average quantity (number of glasses of wine, cans or bottles of beer, nips of spirit) and frequency of alcohol intake. These are henceforth referred to as the "beverage-specific frequency-quantity questions." A can or bottle of regular beer constituted 15 g of alcohol, a glass of wine constituted 15 g of alcohol, and a nip (30 mL) of spirit constituted 10 g of alcohol (Australian food composition tables). The total alcohol intake in grams per day from wine, beer, and spirits was then computed from this information. Participants who were not lifetime abstainers but did not consume alcohol at baseline were classified as former drinkers.
In addition, participants were asked about the intake of alcoholic beverages on each day during the week before the interview (this method will be referred to as the "diary"). All analyses except those for pattern of consumption are based on the beverage-specific frequency-quantity questions. Analyses of pattern of consumption are based upon the diary because the frequency-quantity questions did not ask about the frequency of consumption of all alcoholic beverages combined; that is, for example, if twice weekly wine consumption occurred on the same days as twice weekly beer consumption.
Comprehensive questionnaires regarding lifestyle, dietary intakes, and health conditions were completed at baseline. Participants were classified according to their smoking habits as never smokers, former smokers, and current smokers.
Food intake in the year before baseline was estimated by using a 121-item food frequency questionnaire specifically developed for the MCCS. The energy derived from food was calculated by using standard sex-specific portion sizes from Australian food composition tables.
AMD Detection
Detection of age-related macular degeneration and other fundus pathology was conducted at follow-up by using digital photographs centered on the macula and optic disc, as described previously. Grading and quality-control procedures of the photographs have been described in detail elsewhere. Early AMD was defined as the presence of from 63- to 125-μm drusen with the presence of hyper/hypopigmentation or ≥125-μm drusen, with or without the presence of hyper/hypopigmentation. Late AMD was defined as evidence of choroidal neovascularization, geographic atrophy (≥175 µm of hypopigmentation with visible choroidal vessels), or a disciform scar. The fundoscopic photos were interpreted blindly with respect to baseline traits.
Statistical Analysis
Total alcohol intake was categorized into lifetime abstainers, former drinkers, current drinkers consuming 1–19 g/day, and current drinkers consuming 20 or more g/day. An additional variable consisting of the categories—lifetime abstainers, former drinkers, those consuming 1–19 g/day, 20–39 g/day, 40–59 g/day, and ≥60 g per day—was used to explore higher intakes of alcohol consumption.
Former drinkers may have changed their dietary habits and lifestyle. Therefore, they were included in the analysis as a distinct group (separate from lifetime abstainers). Wine, spirit, and beer consumptions were categorized similarly to total alcohol. Associations for each beverage were evaluated by including consumption of wine, beer, and spirits in separate models as they were highly collinear. The reference group in each analysis consisted of lifetime abstainers.
The total alcohol intake during the week before interview was highly correlated with the average daily intake from the frequency-quantity questions (Spearman's correlation coefficient = 0.87). Additionally, the pattern of consumption (number of days/week) was also highly correlated with the quantity of alcohol consumption (Spearman's correlation coefficient = 0.83), where those with higher intakes were also more likely to drink more frequently. Therefore, to examine weekly patterns of consumption, the participants were stratified into quantities of consumption of 1–139 g, 140–279 g, and ≥280 g/week, and the odds for those drinking on 1–3 days or 4–6 days a week were compared with those drinking every day within each stratum. Participants with no intake of alcohol during the week before baseline were excluded from the analysis of pattern of consumption.
Chi-squared, t tests, and multivariable logistic regression were used to compare baseline demographics and other factors associated with inclusion or exclusion from the analysis.
Multivariable logistic regression was also used to calculate odds ratios for alcohol consumption at baseline, with the presence or absence of AMD at follow-up as the outcome. A directed acyclic graph was used to select variables to include in the models. Under the assumptions made in this graph, to estimate the effect of alcohol intake on AMD, the following variables need to be included in the logistic regression model: age, sex, country of birth, physical activity, smoking status, and dietary energy intake. Although the waist/hip ratio is associated with AMD in this cohort, it was not included in the regression model as we consider it to be on the causal pathway between alcohol and AMD. Smoking was grouped into the categories current, past, or never, and country of birth was grouped into the countries Australia/United Kingdom, Italy, and Greece. Four ordered categories combining frequency and intensity were used to group physical activity.
Effect modification by sex, smoking, and country of birth was assessed by fitting interaction terms between each of these variables and alcohol intake, and it was tested by using the likelihood ratio test. Linear associations between alcohol intake and the odds of AMD were investigated by comparing regression models with alcohol intake as a categorical variable and a pseudo-continuous variable by using the likelihood ratio test. All statistical analyses were performed by using Stata, version 10, statistical software (StataCorp LP, College Station, Texas).
Study Size
Participation at the MCCS follow-up (2003–2007) was 67.2% (Figure 1). A total of 13,612 participants did not participate in follow-up due to death, illness, refusal, leaving Victoria or Australia, or unknown reasons. A further 5,483 were unable to be photographed because of the lack of photographic facilities at some of the clinics. Ophthalmic data from 1,119 participants were excluded from the analysis because the photographs were not able to be graded (Figure 1). The mean time between baseline when dietary and alcohol data were collected and the time of eye photography was 11.5 years (standard deviation, 1.4 years; range, 8.6–16.4). Participants were excluded if they reported extreme energy intakes (n = 324), that is, less than the 1st percentile and more than the 99th percentile of the total MCCS cohort of 41,501. The total number of participants included in the final analysis was 20,963.
(Enlarge Image)
Figure 1.
Flow chart of participants in the Melbourne Collaborative Cohort Study (MCCS), Australia, 1990–2007. AMD, age-related macular degeneration.
