Alcohol Consumption and the Prevalence of Diabetic Retinopathy

Scot E. Moss, MA, Ronald Klein, MD, MPH, Barbara E. K. Klein, MD, MPH

Objective: To determine if alcohol consumption is associated with the prevalence of diabetic retinopathy.

Participants: This study surveyed a population-based sample (n = 121 0) of younger­onset diabetic persons (diagnosed before age 30 years and taking insulin) and a stratified random sample (n = 1780) of older-onset diabetic persons (diagnosed after age 30 years). Baseline and 4-year follow-up examinations completed by 996 and 891 (730 by persons age 21 or older) younger-onset persons, respectively, and 1370 and 987 older­onset persons, respectively. Data analyzed are from the 4-year follow-up examination. Questionnaires concerning consumption were completed at follow-up.

Main Outcome Measure: Diabetic retinopathy as determined from stereographic fundus photography.

Results: After controlling for known risk factors in the adult younger-onset group, average alcohol consumption, as determined by questionnaire, was inversely associated with prevalence of proliferative diabetic retinopathy (PDR), odds ratio, 0.49; 95% con­fidence interval, 0.27 to 0.92. The trend was similar for recent consumption, odds ratio, 0.63; confidence interval, 0.37 to 1.09. In the older-onset groups taking or not taking insulin, average or recent alcohol consumption or usage history were not significantly associated with the prevalence of any retinopathy or PDR.

Conclusions: Alcohol consumption does not appear to increase the risk of reti­nopathy and may have a beneficial effect in younger-onset persons, although further study is needed. Ophthalmology 1 992;99:926-932

If alcohol consumption was found to be a risk factor for diabetic retinopathy, it would be an important develop­ment as it is subject to intervention. To date, investigations

of alcohol as a risk factor for diabetic retinopathy have been few and conflicting. Young et al, 1 in a prospective study of 296 diabetic men free of retinopathy at initial examination, found heavy drinkers to develop 2.25 and 3.5 times more retinopathy and severe retinopathy, re­spectively, than did nondrinkers and moderate drinkers. In contrast, Kingsley, et at,2 in a cross-sectional analysis of 546 insulin-dependent diabetic persons, noted a higher prevalence of ever drinking in persons without than with severe retinopathy. However, they reported no measure ofthe quantity of alcohol consumed. Thus, in this study, we undertook to investigate the relationship of alcohol consumption history and quantity consumed to the prev­alence and severity of diabetic retinopathy in persons par­ticipating in the Wisconsin Epidemiologic Study of Dia­betic Retinopathy (WESDR).

Originally received: October 28, 1991. Revision accepted: January 23, 1992.

From the Department of Ophthalmology, University of Wisconsin, Madison.

Presented at the Association for Research in Vision and Ophthalmology Annual Meeting, Sarasota, May 1991.

Supported by the National Eye Institute/NIH grant EY03083 (R. Klein). Glycosylated hemoglobin determinations were performed in the Core Laboratory of the Clinical Nutrition Center with support from U. S. Public Health Service grant P30 AM AG 26659 (Dr. E. Shrago).

Reprint requests to Dr. Ronald Klein, Department of Ophthalmology, Clinical Science Center, 6QO Highland Ave, Madison, WI 53792.

926

Moss et al · Alcohol Consumption and Diabetic Retinopathy

Methods

Case identification methods and descriptions of the pop­ulation appear in previous reports of the WESDR.3

•4

Briefly, the study area is comprised of 11 counties in southern Wisconsin. During the period of July 1, 1979, to June 30, 1980, lists of diabetic patients were kept by 452 of the 457 primary care physicians in the area. In this manner, 10,135 diabetic persons were identified. A sample of2990 ofthese persons were invited to participate in the baseline examination in the period 1980 to 1982. The sample was composed of two groups. The first group con­sisted of the entire population of insulin-taking persons diagnosed before 30 years of age (n = 1210), current age 2 to 78 years, and is referred to as the younger-onset group. The second group consisted of a probability sample, strat­ified by duration of diabetes, of persons diagnosed at 30 years of age or older, current age 30 to 96 years, regardless of insulin-taking status, but meeting other eligibility cri­teria (n = 1780), and is referred to as the older-onset group. The specifics of the sampling strategy and exact propor­tions appear in an earlier report,4 but were approximately 25% for eligible persons ofO to 4 or 5 to 14 years' duration and 100% for persons with 15 or more years of diabetes. Eighty-two percent (996) of the younger-onset group and 77% (1370) of the older-onset group participated in the baseline examination. Younger-onset nonparticipants had a shorter duration of diabetes than participants. There were no differences in age at diagnosis, current age, or blood pressure. Older-onset nonparticipants were older at diagnosis of diabetes, had a longer duration, and were currently older than participants. There was no difference in blood pressure. Surviving members of the cohort were invited to participate in a 4-year follow-up examination in the period 1984 to 1986. The younger-onset group was restricted to those persons who were at least 21 years of age at the time of the follow-up examination.

Both the baseline and follow-up examinations followed a similar protocol. Pertinent parts of the examination in­cluded explaining the nature of the procedures and ob­taining informed consent, measuring blood pressure, 5 di­lating the pupils, administering a medical history ques­tionnaire, performing an ophthalmoscopic examination, taking stereoscopic color fundus photographs of seven standard fields and a nonstereoscopic red reflex photo­graph for each eye,6 determining urine protein level using a reagent strip (Labstix, Ames Division of Miles Labo­ratories Inc, Elkhart, IN), and determining blood glucose and glycosylated hemoglobin levels.

To determine severity of retinopathy, all fundus pho­tographs were graded using a modification of the Airlie House classification scheme as further adapted for the WESDR follow-up examination.7

•8 This scheme specifies

nine levels of retinopathy. For tabular analyses, the level in the worse eye was used and was grouped into four cat­egories: no retinopathy (level 10), mild nonproliferative retinopathy (levels 21 and 31 ), moderate non proliferative retinopathy (levels 41 and 51), and proliferative diabetic retinopathy (PDR) (levels 60+ ). For multivariate analyses,

the scale was collapsed into the absence or presence of any retinopathy and the absence or presence of prolifer­ative retinopathy.

At the follow-up examination, information on alcohol consumption was obtained while the medical history questionnaire was being completed. Average weekly amounts of beer, wine, and distilled spirits for the previous year were obtained. Average daily consumption of ab­solute alcohol, in ounces, was calculated as (0.04 X 12 X A + 0.15 X 4 X B + 0.45 X 1.5 X C)/7 where A, B, and C were the average weekly consumption of 12-oz cans or bottles of beer, 4-oz glasses of wine, and 1.5-oz drinks of distilled spirits, respectively. This variable is subsequently referred to as average consumption. The conversion factors were obtained from a prior survey of alcohol consumption.9 Information about alcohol con­sumption during the week before the follow-up exami­nation also was obtained. This was summarized as oz/day as above. This variable is subsequently referred to as recent consumption. For tabular analyses of these vari­ables, an abstainer had consumed no alcohol, a light drinker had consumed 0.01 to 0.21 oz/day, a moderate drinker had consumed 0.22 to 0.99 oz/day, and a heavier drinker consumed 1.00 or more oz/day.9 In addition, a person was defined as a nondrinker if he had never con­sumed alcoholic beverages, as an exdrinker if he had con­sumed alcoholic beverages in the past but not in the pre­vious year, and as a current drinker if he had consumed alcoholic beverages in the previous year. This variable is referred to as drinking history.

Current age is defined as the age at the time of the follow-up examination. Age at diagnosis is defined as the age at the time the diagnosis of diabetes was first recorded by a physician on the patient's chart or hospital record. The duration of diabetes is the difference between age at diagnosis and age at follow-up. The systolic and diastolic blood pressures are the means of two systolic and diastolic blood pressure measurements, respectively. Glycosylated hemoglobin was measured on a rinsed sample using a microcolumn technique. 10

The current study is cross-sectional in design and based on data collected at the time of the follow-up examination between 1984 and 1986. Statistical Analysis System 11 was used for calculating prevalence proportions, chi-square statistics, and Mantel-Haenszel tests for trend. 12 Multi­variate analyses of risk factors for prevalence of retinop­athy were performed by logistic regression. 13

Results

There were 891 (89.5%) younger-onset persons who par­ticipated in the follow-up examination, of whom 730 were at least 21 years of age. Sixty-four (6.4%) younger-onset persons had died before the follow-up examination, and the remaining 41 ( 4.1%) could not be located, permitted an interview only, or refused. 8 Nonparticipants were older, had had diabetes longer, and had higher systolic blood pressure, greater body mass, more proteinuria, poorer vi-

927

Ophthalmology Volume 99, Number 6, June 1992

Table 1. Prevalence and Severity of Diabetic Retinopathy by Alcohol Consumption in Younger- and Older-onset Diabetic Persons

Y ounger·onset

Alcohol Severity

Consumption Number* None (o/o) Mild (o/o) Moderate (o/o) PDRt (o/o)

Average Abstainer 123 6.5 27.6 22.8 43.1 Light 398 8.3 35.2 22.6 33.9 Moderate 170 10.0 32.9 31.2 25.9 Heavier 35 2.9 34.3 37.1 25.7 Total 726 8.1 33.3 25.3 33.2 Tests for trendt p = 0.48~ p = 0.06§

Recent Abstainer 393 6.4 33.3 22.1 38.2 Light 148 10.1 31.8 26.4 31.8 Moderate 146 11.6 35.6 29.5 23.3 Heavier 42 4.8 31.0 35.7 28.6 Total 729 8.1 33.3 25.2 33.3 Tests for trendt p = 0.96~ p = 0.08§

History Nondrinker 27 3.7 40.7 14.8 40.7 Exdrinker 96 7.3 24.0 25.0 43.8 Current drinker 606 8.4 34.5 25.7 31.4 Total 729 8.1 33.3 25.2 33.3 Chi-square tests p = 0.65~ p = 0.04§

Older-onset Using Insulin

Alcohol Severity

Consumption Number* None (o/o) Mild (o/o) Moderate (o/o) PDRt (o/o)

Average Abstainer 242 23.6 34.7 24.0 17.8 Light 219 17.8 40.6 26.0 15.5 Moderate 66 19.7 40.9 24.2 15.2 Heavier 18 11.1 33.3 27.8 27.8 Total 545 20.4 37.8 25.0 16.9 Tests for trendt p = 0.29~ p = 0.32§

Recent Abstainer 390 20.8 37.2 24.4 17.7 Light 80 18.8 41.2 25.0 15.0 Moderate 57 22.8 40.4 24.6 12.3 Heavier 18 11.1 27.8 38.9 22.2 Total 545 20.4 37.8 25.0 16.9 Tests for trendt p = 0.42~ p = 0.88§

History

Nondrinker 64 23.4 34.4 26.6 15.6 Exdrinker 178 23.6 34.8 23.0 18.5 Current drinker 304 17.8 40.1 25.7 16.4 Total 546 20.3 37.7 24.9 17.0 Chi-square tests p = 0.25~ p = 0.80§

928

Moss et al · Alcohol Consumption and Diabetic Retinopathy

Alcohol Consumption Number*

Average

Abstainer 174 Light 182 Moderate 48 Heavier 19 Total 423 Tests for trend'f

Recent

Abstainer 313 Light 45 Moderate 47 Heavier 18 Total 423 Tests for trend'f

History

Nondrinker 40 Exdrinker 134 Current drinker 249 Total 423 Chi-square tests

*Variations in total sample sizes are due to missing data.

t PDR =proliferative diabetic retinopathy.

t Based on the Mantel-Haenszel test for trend.

§ Test for trend for prevalence of PDR.

1f Test for trend for prevalence of any retinopathy.

Table 1.

None (o/o)

50.0 51.6 60.4 63.2 52.5

51.1 51.1 61.7 55.6 52.5

47.5 50.7 54.2 52.5

II Test for trend for prevalence of moderate retinopathy or PDR.

sual acuity, and more severe retinopathy than participants. In the older-onset group, 987 (72.0%) participated in the follow-up; 557 of the 987 participants were using insulin and 430 were not. Three hundred forty (24.8%) older­onset persons had died before follow-up and the remaining 43 (3.1%) had moved, could not be located, permitted an interview only, or refused. 14 Nonparticipants were older, had had diabetes longer, and had less education, higher glycosylated hemoglobin, higher systolic blood pressure, lower diastolic blood pressure, lower body mass, more proteinuria, poorer visual acuity, and more severe reti­nopathy than participants.

Table l presents the prevalence and severity of diabetic retinopathy by average and recent alcohol consumption and drinking history for the younger- and older-onset groups. There is some evidence for a trend toward lower prevalence of proliferative diabetic retinopathy (PDR) at higher levels of average (P = 0.06) and recent (P = 0.08)

(continued)

Older-onset not Using Insulin

(%) Severity

Mild (o/o) Moderate (o/o) PDRt (%)

37.4 8.6 4.0 31.3 16.5 0.5 27.1 6.2 6.2 31.6 0 5.3 33.3 11.3 2.8

p = 0.20~ p = 0.2411

34.8 11.5 2.6 35.6 13.3 0 21.3 12.8 4.3 33.3 0 11.1 33.3 11.3 2.8

p = 0.49~ p = 0.8511

42.5 7.5 2.5 35.8 9.0 4.5 30.5 13.3 2.0 33.3 11.3 2.8

p = 0.65~ p = 0.6511

alcohol consumption in the younger-onset group. For av­erage consumption, the prevalence of PDR decreased from 43.1% in abstainers to 25.7% in heavier drinkers. With respect to recent consumption, the prevalence of PDR decreased from 38.2% in abstainers to 23.3% in moderate drinkers, whereafter it increased to 28.6% in heavier drinkers. The analysis of drinking history shows exdrinkers to have the highest prevalence of PDR. How­ever, it is not significantly different from nondrinkers. These analyses were repeated separately for men and women. Although women reported drinking less, the as­sociations between average and recent alcohol consump­tion and retinopathy were similar for men and women. Male but not female exdrinkers had a higher prevalence of PDR. In the older-onset group, whether using insulin or not, there were no statistically significant associations between alcohol consumption and retinopathy either in the total group (Table I) or in men and women separately.

929

Ophthalmology Volume 99, Number 6, June 1992

Any relationships between alcohol consumption and retinopathy may be confounded by other risk factors for retinopathy such as duration of diabetes, blood pressure, and glycosylated hemoglobin. To investigate this possi­bility, multiple logistic regression was used to control for known risk factors for the prevalence of retinopathy. Sev­eral end points were examined. In the younger-onset and older-onset insulin-using groups, the absence or presence of any retinopathy and the absence or presence of PDR were the dependent variables. In the older-onset group not using insulin, the absence or presence of any retinop­athy and the absence or presence of moderate nonprolif­erative retinopathy or worse were used. Duration of di­abetes, glycosylated hemoglobin, and proteinuria were included in every model for all three groups. 3•

4 In addition, current age, diastolic blood pressure, and sex were in­cluded in models of the younger-onset group,3 age at di­agnosis and systolic blood pressure in the older-onset group using insulin, 4 and systolic blood pressure and body mass in the older-onset group not using insulin.4 Average

or recent alcohol consumption, as continuous variables, or drinking history were then added to the models to de­termine their significance in explaining the prevalence of retinopathy. Drinking history was recorded as two indi­cator variables to denote exdrinkers and current drinkers. In addition, the univariate analyses suggested square root transformations of alcohol consumption would give a better fit in the logistic model. These also were included and were generally more significant than the untrans­formed values. Thus, only the results using the trans­formed values are reported here. Table 2 summarizes the results. In the younger-onset group, average alcohol con­sumption is inversely associated with the prevalence of PDR (P = 0.03). Each unit increase in the square root of average alcohol consumption is associated with 51% lower odds of having PDR. Interaction effects between average alcohol consumption and the other independent variables also were examined and none were found to be significant. Recent consumption is also inversely associated with the prevalence of PDR but is not statistically significant

Table 2. Odds Ratios and 95% Confidence Intervals for the Presence of any Retinopathy, Proliferative Diabetic Retinopathy, or Moderate Retinopathy or Worse Associated with Alcohol Consumption

after Controlling for Other Independent Risk Factors in Multiple Logistic Regression Models*

Group

Younger-onset

Older-onset using insulin

Older-onset not using insulin

Dependent Variable

Any retinopathy

PDR

Any retinopathy

PDR

Any retinopathy

Moderate retinopathy or worse

PDR = proliferative diabetic retinopathy.

Alcohol Consumption Variable

Root (average [oz/d]) Root (recent [oz/d]) Root (average [oz/d]) Root (recent [oz/d]) History: ex-drinkers

current drinkers Root (average [oz/d]) Root (recent [oz/d]) History: ex-drinkers

current drinkers Root (average [oz/d]) Root (recent [oz/d]) History: ex-drinkers

current drinkers Root (average [oz/d]) Root (recent [oz/d]) History: ex-drinkers

current drinkers Root (average [oz/d]) Root (recent [oz/d]) History: ex-drinkers

current drinkers

Odds Ratio 9 5% Confidence Interval

0.88 0.32, 2.42 0.51 0.22, 1.19 0.49 0.27, 0.92 0.63 0.37, 1.09

1.47 0.46, 4.70 1.01 0.35, 2.89 1.09 0.48, 2.48 0.91 0.41, 2.01 0.99 0.46, 2.15 1.30 0.62, 2.76

0.74 0.33, 1.69 0.65 0.29, 1.47 1.33 0.51, 3.51

1.17 0.46, 3.00 0.73 0.40, 1.31 0.78 0.45, 1.35 0.69 0.29, 1.63 0.51 0.23, 1.17 0.77 0.31, 1.88 1.16 0.55, 2.46 1.24 0.32, 4.79 1.38 0.38, 5.02

• Other variables included duration of diabetes, current age, glycosylated hemoglobin, diastolic blood pressure, proteinuria, and sex for the younger­onset group; duration of diabetes, age at diagnosis, glycosylated hemoglobin, systolic blood pressure, and proteinuria for the older-onset group using insulin; and duration of diabetes, glycosylated hemoglobin, systolic blood pressure, proteinuria, and body mass for the older-onset group not using insulin.

930

Moss et al · Alcohol Consumption and Diabetic Retinopathy

(P = 0.1 0). Drinking history could not be included in the younger-onset model for prevalence of any retinopathy because of the small number of people without retinop­athy. Otherwise, no alcohol consumption variable was statistically significant for predicting the presence of any retinopathy in the younger-onset group or any end point in either older-onset group (smallest, P = 0.19).

Discussion

Self-reported alcohol consumption data obtained by sur­vey have been shown to be reliable and valid. 15 In addition, the distribution of alcohol consumption in our population is comparable to national estimates. 16 In our combined group of younger- and older-onset people, there were 32%, 4 7%, 17%, and 4% of abstainers, light drinkers, moderate drinkers, and heavier drinkers, respectively, compared with 35%, 35%, 22%, and 8%, respectively, in national estimates for 1985.16 While consumption is somewhat less in our diabetic population, this is to be expected in a group with a chronic disease that may have a substantial proportion of people in poor health.

Studies of the effect of alcohol consumption on dia­betic retinopathy are few. 1

•2 The prospective study by

Young et al1 showed alcohol consumption to be a risk factor for the development of diabetic retinopathy. However, the cohort was restricted to a clinic cohort of men of unspecified type of diabetes. Thus, the results may not be representative of a broader diabetic popu­lation. Moreover, retinopathy was detected by ophthal­moscopy, a technique that may be insensitive to the presence of retinopathy. 17- 19

In contrast, as in our younger-onset group, Kingsley et af found a weak beneficial effect for alcohol consump­tion in a cross-sectional study of insulin-dependent dia­betes. However, there also are difficulties with this ap­proach that apply to both studies. Biases may be intro­duced due to prior history. For instance, if heavy drinkers with severe retinopathy had died before the examination or were too ill to be examined, the strength of a relation­ship between the two would be lessened. Indeed, it has been shown that diabetic persons with retinopathy or who are heavy drinkers have a greater mortality risk.9•

20 There also is the potential that modification of the risk factor occurred after development of retinopathy; that is, diabetic persons may reduce alcohol consumption or stop drinking entirely after they begin to develop complications whether caused by alcohol or not. Alternatively, Wallace and Colsher1 have shown that persons who do not currently drink may have been heavy drinkers in the past and may be at increased risk of morbidity. Either of these alter­natives may explain why we found exdrinkers to have the highest prevalence ofPDR in the univariate analyses and odds ratios greater than unity in the multivariate analyses.

However, it remains possible that alcohol consumption exerts a beneficial effect on retinopathy through several mechanisms. It has been shown that moderate alcohol consumption increases the level of high-density lipopro-

tein cholesterol.22•23 Furthermore, Klein et al24 have shown

an inverse relationship between high-density lipoprotein cholesterol and the prevalence of retinopathy in younger­onset persons. Also, alcohol has been associated with a decrease in platelet aggregability,25

•26 which in turn may

be beneficial in preventing progression of retinopathy.27

Finally, alcohol appears to lower the level of fibrino­gen. 28

•29 Higher levels of fibrinogen have been associated

with more severe retinopathy. 30

References

1. Young RJ, McCulloch DK, Prescott RJ, Clarke BF. Alcohol: another risk factor for diabetic retinopathy? Br Med J 1984;288: I 035-7.

2. Kingsley LA, Dorman JS, Doft BH, et al. An epidemiologic approach to the study of retinopathy: the Pittsburgh diabetic morbidity and retinopathy studies. Diabetes Res Clin Prac 1988;4:99-109.

3. Klein R, Klein BEK, Moss SE, et al. The Wisconsin Epi­demiologic Study of Diabetic Retinopathy. II. Prevalence and risk of diabetic retinopathy when age at diagnosis is less than 30 years. Arch Ophthalmol 1984;102:520-6.

4. Klein R, Klein BEK, Moss SE, et al. The Wisconsin Epi­demiologic Study of Diabetic Retinopathy. III. Prevalence and risk of diabetic retinopathy when age at diagnosis is 30 or more years. Arch Ophthalmol 1984;102:527-32.

5. Hypertension Detection and Follow-up Program Cooper­ative Group. The hypertension detection and follow-up program. Prev Med 1976;5:207-15.

6. Early Treatment of Diabetic Retinopathy Study (ETDRS). Manual of Operations. May 1, 1985. Available from: Na­tional Technical Information Service, (Accession No. PB85-223006-XAB/ AS).

7. Diabetic Retinopathy Study Research Group. Report 7. A modification of the Airlie House classification of diabetic retinopathy. Invest Ophthalmol Vis Sci 1981 ;21 :210-26.

8. Klein R, Klein BEK, Moss SE, et al. The Wisconsin Epi­demiologic Study of Diabetic Retinopathy. IX. Four-year incidence and progression of diabetic retinopathy when age at diagnosis is less than 30 years. Arch Ophthalmol1989;107: 237-43.

9. Malin H, Wilson RW, Williams GD. 1983 NHIS alcohol/ health practices supplement: preliminary findings. In: Pro­ceedings of the 1985 Public Health Conference on Records and Statistics. Hyattsville, MD: Public Health Service, 1986: 490-5 (DHHS Pub No. (PHS) 86-1214).

10. Moss SE, Klein R, Klein BEK, et al. Methodologic consid­erations in measuring glycosylated hemoglobin in epide­miologic studies. J Clin Epidemiol 1988;41:645-9.

11. SAS Institute Inc. SAS User's Guide: Statistics. Version 5 ed. Cary, NC: The Institute, 1985:403-32.

12. Mantel N. Chi-square tests with one degree of freedom; ex­tensions of the Mantel-Haenszel procedure. JAm Stat Assoc 1963;58:690-700.

13. Walker SH, Duncan DB. Estimation of the probability of an event as a function of several independent variables. Biometrika 1967;54:167-79.

14. Klein R, Klein BEK, Moss SE, et al. The Wisconsin Epi­demiologic Study of Diabetic Retinopathy. X. Four-year incidence and progression of diabetic retinopathy when age at diagnosis is 30 years or more. Arch Ophthalmol 1989; 107: 244-9.

931

Ophthalmology Volume 99, Number 6, June 1992

15. Williams GO, Aitken SS, Malin H. Reliability of self-re­ported alcohol consumption in a general population survey. J Stud Alcohol 1985;46:223-7.

16. National Center for Health Statistics. Health, United States, 1989. Hyattsville, MD: Public Health Service, 1990:168 (DHHS Pub No. (PHS) 90-1232).

17. Palm berg P, Smith M, Waltman S, et al. The natural history of retinopathy in insulin-dependent juvenile-onset diabetes. Ophthalmology 1981 ;88:613-8.

18. Sussman EJ, Tsiaras WG, Soper KA. Diagnosis of diabetic eye disease. JAMA 1982;247:3231-4.

19. Moss SE, Klein R, Kessler SO, Richie KA. Comparison between ophthalmoscopy and fundus photography in de­termining severity of diabetic retinopathy. Ophthalmology 1985;92:62-7.

20. Klein R, Moss SE, Klein BEK, DeMets DL. Relation of ocular and systemic factors to survival in diabetes. Arch Int Med 1989;149:266-72.

21. Wallace RB, Colsher PL. Enhancing the utility of quantity­frequency measures of alcohol consumption with assess­ments of problem drinking in a population study: a meth­odologic note. Ann Epidemiol 1990;1:157-65.

22. Castelli WP, Doyle JT, Gordon T, et al. Alcohol and blood lipids. The cooperative lipoprotein phenotyping study. Lancet 1977;2:153-5.

23. Ernst N, Fisher M, Smith W, et al. The association of plasma

932

high-density lipoprotein cholesterol with dietary intake and alcohol consumption. The Lipid Research Clinics Program Prevalence Study. Circulation 1980;62(4 Pt. 2):1V41-52.

24. Klein BEK, Moss SE, Klein R, Surawicz TS. The Wisconsin Epidemiologic Study of Diabetic Retinopathy. XIII. Rela­tionship of serum cholesterol to retinopathy and hard ex­udate. Ophthalmology 1991;98:1261-5.

25. Wautier JL, Kadeva H, Deschamps JF, et al. Inhibition de l'agregation plaquettaire apres ingestion d'alcool. Presse Med 1988;10:2040.

26. Jakubowski JA, Vaillancourt R, Deykin D. Interaction of ethanol, prostacyclin, and aspirin in determining human platelet reactivity in vitro. Arteriosclerosis 1988;8:436-41.

27. The TIMAD Study Group. Ticlopidine treatment reduces the progression of nonproliferative diabetic retinopathy. Arch Ophthalmol 1990;108:1577-83.

28. Meade TW, North WRJ, Chakrabarti R, et al. Population­based distributions of haemostatic variables. Br Med Bull 1977;33:283-8.

29. Meade TW, Chakrabarti R, Haines AP, et al. Characteristics affecting fibrinolytic activity and plasma fibrinogen concen­trations. Br Med J 1979; 1:153-6.

30. Kostraba JN, Klein R, Dorman JS, et al. The epidemiology of diabetes complications study. IV. Correlates of diabetic background and proliferative retinopathy. Am J Epidemiol 1991;133:381-91.