Several epidemiological studies have reported lower overall mortality rates in light-to-moderate drinkers compared with those not consuming alcohol, while overall mortality seems to increase with higher levels of alcohol consumption. 1–3 The lower cardiovascular disease mortality observed in drinkers of up to several drinks per day compared with non-drinkers 2–6 partly explains the beneficial effect of alcohol. Nevertheless, when mortality from cardiovascular disease was excluded, mortality from other causes was observed to be lower in those consuming up to 3 drinks per day compared with non-drinkers. 1,3
Another important cause of death is chronic obstructive pulmonary disease (COPD), defined as chronic bronchitis and emphysema. There are indications for a beneficial effect of light-to-moderate alcohol consumption on pulmonary function 7 and 25-year incidence of asthma and COPD. 8 Furthermore, a beneficial effect of alcohol consumption on the prevalence and extent of emphysema, determined by autopsy, was suggested. 9 Heavy alcohol consumption is thought to have deleterious effects on the lungs, mainly on the basis of studies in alcoholics. 10–12
To our knowledge the relation between alcohol consumption and long-term COPD mortality has not been studied before. Based on present knowledge, we hypothesized a U-shaped relation. To test this hypothesis we used data on alcohol consumption around 1970 and 20-year COPD mortality in 2,953 middle-aged men from Finland, Italy, and the Netherlands gathered in the Seven Countries Study. Concurrently, we investigated whether cross-sectional data on pulmonary function and alcohol consumption in the three countries indicated a similar effect of alcohol consumption on COPD.
Subjects and Study Design
Around 1960 sixteen population samples of men 40–59 years of age from seven countries were enrolled and examined at baseline for the Seven Countries Study. 13 The vital status of the participants was recorded during 30 years of follow-up and the men were re-examined at 5 and 10 years after baseline. In Table 1 the data collected on the Finnish, Italian, and Dutch cohorts in the three examination rounds have been summarized. For all other cohorts data on alcohol consumption and/or pulmonary function was not available.
In Finland, one cohort was situated in rural east Finland, close to the Russian border, and the other in a rural area in the west of Finland. The Italian cohorts comprised men living in Montegiorgio and Crevalcore, two rural villages in central and northern Italy, respectively. The fifth cohort comprised men from Zutphen, a small commercial town in the east of the Netherlands.
Complete data on alcohol consumption around 1970 (for Montegiorgio alcohol consumption in 1965 was used as a proxy for alcohol consumption in 1970), COPD-mortality between 1970 and 1990 and potential confounders were available for 2,953 men in the three countries. Complete data on alcohol consumption, pulmonary function, and potential confounders were available for 1,248 Finnish men in 1969 and for, respectively, 1,386 and 691 men in Italy and the Netherlands in 1965.
Food intake, including consumption of alcoholic beverages, was estimated using the cross-check dietary history method. This method provides information about the usual food consumption pattern 6 to 12 months preceding the interview. 14 All interviews were conducted by extensively trained dietitians and nutritionists. Although the dietary history method was adapted to the local situation in each specific country, the methodology was comparable. The nutrient intake, including alcohol, was assessed using computerized versions of the local food tables for the three different countries. 15–18
The underlying cause of death of those who died during follow-up was established centrally by one investigator (A. Menotti). He reviewed information from clinical records, from family doctors, specialists, relatives, and from other useful sources, collected by local investigators. Usually the official cause of death from the death certificate was not considered or only used as a preliminary indication. Primary mortality was coded according to the 8th revision of the International Classification of Diseases (COPD = ICD 491–493).
Pulmonary function was measured by spirometry. Equipment and protocols differed between the countries, but were identical among cohorts in a country, as described earlier. 19 The level of forced expiratory volume (FEV) can, therefore, not be compared between countries. Nevertheless, comparisons among men within the three separate countries are valid. In short, forced expiratory volume in 0.75 seconds (FEV0.75) was measured in Finland and Italy. In both countries a nose clip was used, but the Finnish subjects were measured in a sitting and the Italian subjects in a standing position. In the Dutch cohort of Zutphen, forced expiratory volume in 1 second (FEV1) was measured with subjects sitting in an upright position. In all countries the FEV0.75 or FEV1 was established in three attempts.
Information on age, height, weight, and smoking was collected in a standardized way. 13 Body mass index (BMI) was calculated (weight/height2). Pack-years were calculated as the product of the number of years smoked and the number of packs of cigarettes smoked per day. 19 A package of cigarettes was assumed to contain 25 cigarettes.
To assess the longitudinal relation between baseline alcohol consumption (around 1970) and 20-year COPD mortality, we used GAIM-software 20 to fit a Cox proportional hazards model, fitting alcohol and pack-years of smoking as cubic smoothing splines with 8 degrees of freedom. We calculated relative risks from the smoothed coefficients by taking the exponent after subtracting the value of the smoothed coefficient for the reference group (the non-drinkers).
Subsequently, we categorized the number of alcoholic drinks consumed into 5 levels: none, ≤1 per week (occasional), >1 per week and ≤3 per day (light), >3 and ≤9 per day, and > 9 per day. We assumed that a drink contained 10 grams of alcohol. We then used the Cox Proportional Hazard Model (SAS procedure PHREG) to determine the relative risk of COPD mortality in the different categories of alcohol consumption, again with the non-drinkers as a reference.
In analyses concerning COPD mortality, we considered age, BMI, energy intake, cigarette smoking, and country as potential confounders. To adjust for smoking we entered pack-years of smoking into the model. Alternative ways to adjust for smoking did not alter the results in a relevant way. In the model used for smoothing the mortality data, country was included as a categorical variable. In the subsequent analyses adjustment for country was conducted by calculating a pooled relative risk, using the strata option of the PHREG option (SAS), which allows baseline hazards to vary between countries.
Although we used the term COPD, defined as chronic bronchitis and emphysema, asthma does contribute to the mortality rates. Considering the overlap in clinical features 21 and the fact that in our study chronic bronchitis or emphysema was often noted as the secondary cause of death in the small number of cases where asthma was reported as the underlying cause of death, we decided against excluding asthma cases from the analysis.
Owing to the small numbers of COPD deaths in the separate countries, we could not stratify by country. We could, however, evaluate the relation between alcohol consumption and pulmonary function in separate countries. In Finland and the Netherlands alcohol consumption was generally low (mean <1 drink per day, more than 95% with ≤3 drinks per day) and in Italy generally high (mean >8 drinks per day, more than 85% with >3 drinks per day). Therefore, none, occasional, and light alcohol consumption was studied in relation to pulmonary function in Finland and the Netherlands and moderate to heavy alcohol consumption (>3 drinks per day) was studied in relation to pulmonary function in Italy. We used the statistical package S-plus (2000) to create smoothed spline-curves. We used a model with FEV0.75/HT2 or FEV1/HT2 as the dependent and age (years) as an independent variable. 22 Potential confounders considered in these analyses were: BMI, energy intake, and cigarette smoking. Adjustment for cigarette smoking was performed by entering smoking status (two categorical variables), number of years smoked, and average number of cigarettes smoked per day into the model. In Finland and Italy associations were adjusted for cohort. For ease of interpretation the results are presented age-adjusted and on the original scale (FEV1 or FEV0.75 in ml) for a man of average height (1.70 meters).
In all analyses energy intake was defined as the energy derived from fats, carbohydrates, and proteins, and not from alcohol. To study whether the effect of alcohol was independent of that of other dietary factors potentially associated with COPD (ie vitamin C, vitamin E, β-carotene, fruits, vegetables, and fish), we added these variables to the adjusted models. Analyses stratified by smoking status (non-smoker, former smoker, or current smoker) were not possible owing to small numbers in several cells.
Alcohol Consumption around 1970 and 20-Year Chronic Obstructive Pulmonary Disease Mortality
During 20 years of follow-up 1,729 men or 58.6% of the study population died. Seventy-three (2.5%) of the men died from COPD, with rates varying from 2.0 to 3.4% between the three countries (Table 2). At baseline 50% of the study population reported to smoke cigarettes, 29% had smoked cigarettes in the past and 21% had never smoked cigarettes. In former and current smokers, the average number of pack-years smoked was 20 and 25, respectively. The 20-year COPD mortality rate was 0.6, 1.7, and 2.2 per 1,000 person-years in never smokers, former smokers, and current smokers, respectively. After adjustment for age, BMI, and energy intake, pack-years of smoking was positively associated with COPD mortality (RR = 1.08 per 5 pack-years (95% CI = 1.01–1.16). Alcohol consumption was positively associated with smoking, with the average number of pack-years smoked increasing from 14.7 in non-drinkers to 22.1 in those consuming >9 drinks per day or with 1.9 pack-years per category of alcohol consumption (95% CI = 1.3–2.5)
Figure 1 gives both the smoothed curve and a bar graph of the adjusted relative risk (RR) of dying from COPD for those consuming alcohol compared with non-drinkers at baseline. In those consuming more than 9 drinks per day (not shown in Figure 1), the RR was 1.14 (95% CI = 0.33–3.89). We observed a U-shaped curve, with the lowest risk for those with a light alcohol consumption (≥1.4 and ≤ 30 gm per day).
Compared with the non-drinkers and occasional drinkers, the relative risk of 20-year COPD mortality was 0.60 (95% CI = 0.33–1.09) in the light drinkers and 1.25 (95% CI = 0.47–3.31) in those with a higher alcohol consumption, after adjustment for all potential confounders. Additional adjustment for the effects of intake of antioxidant vitamins, vegetables, or fish had little effect. Adjustment for fruit intake, however, changed the observed RRs to 0.68 (95% CI = 0.36–1.26) and 1.58 (95% CI = 0.54–4.64), respectively. The energy-adjusted fruit consumption was on average 98, 186, and 202 gm per day for the three compared groups. Finally, after exclusion of deaths in the first 3 years of follow-up, to evaluate the potential effect of subjects with advanced COPD at baseline refraining from drinking, the RRs were 0.59 (95% CI = 0.32–1.08) and 1.18 (95% CI = 0.42–3.30).
Alcohol and Pulmonary Function at Baseline
We studied the relation between alcohol consumption and pulmonary function in 1,186 Finnish men (examined in 1969) and 667 Dutch men (examined in 1965) consuming up to 3 alcoholic drinks per day, and in 1,183 Italian men (examined in 1965) consuming more than 3 drinks per day. The average alcohol consumption was 4.2, 5.3, and 98.8 gm per day in Finland, the Netherlands, and Italy, respectively. The Finnish men were on average 5 years older than the Dutch and Italian men (Table 3).
Cigarette smoking (per pack-year) was inversely associated with the FEV0.75 in Finland (β = −5.3 ml, 95% CL = −7.6, −3.0) and Italy (β = −5.8 ml, 95% CL = −8.2, −3.4) and with the FEV1 in the Netherlands (β = −5.6 ml, 95% CL = −8.9, −2.4). Smoking was positively associated with alcohol consumption: the average number of pack-years smoked was 18.8 and 16.9 in the non-drinkers and 23.7 and 20.8 in light drinkers in Finland and the Netherlands, respectively. In Italy those consuming more than 3 and less than 9 drinks per day smoked on average 13.2 pack-years and those with higher alcohol consumption on average 17.8 pack-years.
Table 4 gives the mean level of FEV1 or FEV0.75 for the different categories of alcohol consumption. The smoothed spline curves in Figure 2 indicate that in Finland and the Netherlands pulmonary function was higher in occasional and light drinkers (>0 and ≤30 gm per day) compared with non-drinkers. In Finland the observed difference in FEV0.75 was 75 ml (95% CL = −2, 151) and in the Netherlands the observed difference in FEV1 was 93 ml (95% CI = 0–186). In Italy, very heavy drinkers had a lower FEV0.75 than moderate-to-heavy drinkers (>3 and ≤12 drinks per day) (Figure 2). The observed difference in FEV0.75 was 99 ml (95% CI = 9–186).
Finally, additional adjustment for intake of other dietary factors (antioxidant vitamins, fruits, vegetables, or fish) or only including the men 50–64 years of age (the age-range included in all three countries) did not alter by much the association between alcohol and pulmonary function in the Netherlands or Italy. In Finland, however, adjustment for fruit intake reduced the observed effect from 75 to 66 ml (95% CL = −16, 147). Excluding those aged 64 to 69 years also reduced the effect in Finland, to 57 ml (95% CL = −26, 141).
We observed a U-shaped curve between baseline alcohol consumption and 20-year COPD mortality in middle-aged men from Finland, Italy, and the Netherlands, with the lowest risk in those with a light alcohol consumption (>1 drink per week and ≤ 3 drinks per day). Compared with the non-drinkers and occasional drinkers, the relative risk of 20-year COPD mortality was 0.60 (95% CI = 0.33–1.09) in the light drinkers and 1.25 (95% CI = 0.47–3.31) in those with a higher alcohol consumption, after adjustment for potential confounders. In the two countries with a generally low alcohol consumption (Finland and the Netherlands), pulmonary function was higher in occasional and light (up to 3 drinks per day) compared with non-drinkers. The observed differences were 75 ml in FEV0.75 (95% CL = −2, 151) in Finland and 93 ml in FEV1 (95% CI = 0–186) in the Netherlands. In Italy with a generally high alcohol consumption, the FEV0.75 was 99 ml (95% CI = 9–189) lower in very heavy drinkers (>12 drinks per day) compared with moderate-to-heavy drinkers (>3 and ≤ 12 drinks per day).
To our knowledge no prior longitudinal study has focused on alcohol consumption and COPD mortality. Thun et al.3 reported no consistent association between death from the wider category of all respiratory causes and alcohol consumption. Unfortunately the data were not shown. In a study by Doll et al., 1 13-year mortality from all respiratory causes in male British doctors seemed to be lower in those consuming 1 to 21 drinks per week (1.3 to 1.7 per 1000 men) compared with non-drinkers (2.0 per 1000 men) and seemed to increase at higher levels of alcohol consumption (2.5 to 3.5 per 1000 men). This is well in accordance with our findings.
In the Dutch cohort a lower risk of 25-year incidence of asthma and COPD in men with light-to-moderate alcohol consumption compared with non-drinkers was reported earlier. 8 Also in other studies indications for a beneficial effect of alcohol consumption on COPD related outcomes were observed. 7,9 In two large epidemiologic studies, no cross-sectional association between light-to-moderate alcohol consumption and airway obstruction was observed. 7,23 In these studies, however, the results were not shown for non-drinkers and occasional drinkers separately.
Our data on COPD mortality did not show a detrimental effect of heavy alcohol consumption. In those consuming more than 9 drinks per day, the COPD mortality rate was comparable with that in the non-drinkers. Studies in alcoholics, showing a high prevalence of obstructive lung disease, have suggested a detrimental effect of heavy alcohol consumption on the lungs. 10–12 It is possible, however, that we were unable to detect an increase in COPD mortality risk in subjects with very high intakes (eg >12 drinks per day), as suggested by the pulmonary function data in Italy, owing to the relatively small number of observations in this range of alcohol consumption.
Whether the beneficial effect of light alcohol consumption compared with not drinking alcohol regularly is caused by a direct effect of alcohol or by a confounding effect of ill-health among ex-drinkers 24 or by differences in other characteristics between the men in the different categories of alcohol consumption, remains to be clarified. A direct protective effect of alcohol is possible, since various inhibitory effects of alcohol on inflammatory cells have been described. Inflammation of the peripheral airways, mainly initiated by inhaled oxidants and free radicals, is one of the main processes in the pathogenesis of COPD. Alcohol has been observed to inhibit neutrophil delivery to inflammatory sites 25,26 in a dose-dependent manner, 25 possibly by diminishing the required adherence of these cells to the endothelium. 26 Both chronic alcoholism and mild intoxication have furthermore been observed to decrease the production of pro-inflammatory substances called eicosanoids (eg prostaglandins and leukotrienes) by activated inflammatory cells. 27,28 Theoretically it is also possible that subjects with advanced COPD at baseline refrained from drinking alcohol. The result would be a higher COPD mortality among non-drinkers in the first years of follow-up. To evaluate this possibility, we repeated the analysis after excluding deaths in the first 3 years of follow-up. The difference in COPD mortality between non-drinkers and occasional drinkers and light drinkers remained virtually unchanged.
Cigarette smoking, the main risk factor for COPD, is positively associated with alcohol consumption, 2–4 as we found in the present study. Therefore, residual confounding by smoking may partly obscure the positive effect of light alcohol consumption and may lead to over-estimation of the negative effect of heavier alcohol consumption on COPD. Unfortunately the numbers of never smokers in the study populations were too small to limit the analyses to this subgroup.
Since we were interested in the shape of the relation between alcohol and the studied outcomes, we used smoothing techniques. The exact shape of the curves, however, varied somewhat with the chosen technique (eg Loess vs. spline) and the degree of smoothing applied. Furthermore, it is clear that the reliability of a smoothed curve depends on the number of observations on which a (part of the) curve is based. We are therefore reluctant to draw too detailed conclusions on the shape of the presented curves. The peak in COPD mortality in occasional drinkers (>0, ≤1.4 gm per day) in Figure 1 is, for instance, caused by the fact that four of the eight cases among occasional drinkers had an alcohol consumption below 0.2 gm per day. This finding may well be coincidence.
The observed effect of alcohol on the FEV1 and the FEV0.75, 75 to 99 ml, is roughly equivalent to the effect of an increase in age of 3 years and to smoking 1 pack of cigarettes per day during 14 to 17 years. 29 For a high intake of fruits, fish, vitamin C, vitamin E, and β-carotene, similar beneficial effects on pulmonary function have been observed. 30 In the final models we adjusted for intake of these dietary factors and observed a small confounding effect of fruit on the association between alcohol consumption and COPD mortality and on the association between alcohol and pulmonary function in Finland. Residual confounding by fruit or the other dietary factors due to misclassification of dietary intake cannot be excluded. Nevertheless, the effect of alcohol seems to be largely independent of that of the other dietary factors and the effect of the different dietary factors on COPD may involve an important public health impact.
In conclusion, we observed a U-shaped curve between alcohol consumption and 20-year COPD mortality in middle-aged men that was supported by cross-sectional data on alcohol and pulmonary function. The lowest COPD mortality, and the highest level of pulmonary function, was observed in men with low alcohol consumption (up to 3 drinks per day).
We thank the many people who were involved in this study, including the men who took part in the surveys and the organisers of the fieldwork in the three countries, Alessandro Menotti and Simona Giampaoli in Italy, Aulikki Nissinen in Finland, and Daan Kromhout in the Netherlands. We also thank Hendriek C. Boshuizen for her contribution to the data-analyses and the preparation of the smoothed curves.
1. Doll R, Peto R, Hall E, Wheatley K, Gray R. Mortality in relation to consumption of alcohol: 13 years’ observations on male British doctors. BMJ 1994; 309: 911–918.
2. Fuchs CS, Stampfer MJ, Colditz GA, Giovanucci EL, Manson JE, Kawachi I, Hunter DJ, Hankinson SE, Hennekes CH, Rosner B, and others. Alcohol consumption and mortality among women. N Engl J Med 1995;332:1245–1250.
3. Thun MJ, Peto R, Lopez AD, Monac JH, Henley SJ, Heath CW, Doll R. Alcohol consumption and mortality among middle-aged and elderly U.S. adults. N Engl J Med 1997; 337: 1705–1714.
4. Stampfer MJ, Colditz GA, Willett WC, Speizer FE, Hennekens CH. A prospective study of moderate alcohol consumption and the risk of coronary disease and stroke in women. N Engl J Med 1988; 319: 267–273.
5. Friedman LA, Kimball AW. Coronary heart disease mortality and alcohol consumption in Framingham. Am J Epidemiol 1986; 124: 481–489.
6. Rimm EB, Giovannuchi EL, Willett WC, Colditz GA, Ascherio A, Rosner B, Stampfer MJ. Prospective study of alcohol consumption and risk of coronary disease in men. Lancet 1991; 338: 464–468.
7. Cohen BH, Celentano DD, Chase GA, Diamond EL, Graves CG, Levy DA, Menkes HA, Meyer MB, Permutt S, Tockman MS. Alcohol consumption and airway obstruction. Am Rev Respir Dis 1980; 121: 205–215.
8. Miedema I, Feskens EJM, Heederik D, Kromhout D. Dietary determinants of long-term incidence of chronic non-specific lung disease; The Zutphen study. Am J Epidemiol 1993; 138: 37–45.
9. Pratt PC, Volmer RT. The beneficial effect of alcohol consumption on the prevalence and extent of centrilobular emphysema; a retrospective autopsy analysis. Chest 1984; 85 (3): 372–377.
10. Burch GE, Pasquale NP. Alcoholic lung disease; an hypothesis (editorial). Am Heart J 1967; 73: 147–148.
11. Banner AS. Pulmonary function in chronic alcoholism. Am Rev Respir Dis 1973; 108: 851–857.
12. Emergil C, Sobol BJ, Heymann B, Shibutani K. Pulmonary function in alcoholics. Am J Med 1974; 57: 69–77.
13. Keys A, Aravanis C, Blackburn H. Epidemiological studies related to coronary heart disease: characteristics of men aged 40–59 in seven countries. Acta Med Scand 1967; 460 (suppl): 1–392.
14. Burke BS. The dietary history as a tool in research. J Am Diet Assoc 1947; 23: 1041–1046.
15. Den Hartog C, Van Schaik ThFSM, Dalderup LM, Drion EF, Mulder T. The diet of volunteers participating in a long term epidemiological field survey on coronary heart disease at Zutphen, The Netherlands. Voeding 1965;26:184–208.
16. Hautvast JGAJ. Commissie uniforme codering voedingsenquetes; ontwikkeling van een systeem om gegevens van voedingsenquetes met behulp van de computer te verwerken. (Committee uniform coding of dietary surveys: development of a system to process dietary survey data by computer) Voeding 1975; 36: 356–361.
17. Pekkarinen M. Dietary surveys in connection with coronary heart disease studies in Finland. In: Bazan NG, Paoletti R, Iacono JM, eds. New Trends in Nutrition, Lipid Research, and Cardiovascular Diseases. New York: Alan R Liss, Inc., (Current Topics in Nutrition and Disease, Vol 5.), 1981;243–261.
18. Alberti-Fidanza A, Seccareccia F, Torsello S, Fidanza F. Diet of two rural population groups of middle-aged men in Italy. Int J Vitam Nutr Res 1988; 58: 442–451.
19. Tabak C, Smit HA, Rásánen L, Fidanza F, Menotti A, Nissinen A, Feskens EJM, Heederik D, Kromhout D. Dietary factors and pulmonary function: a cross-sectional study in middle-aged men from three European countries. Thorax 1999; 54: 1021–1026.
20. Hastie TJ, Tibshirani RJ. Generalized additive models. 1st edition. London: Chapman & Hall, 1990.
21. Snider GL. Emphysema. The first two centuries-and beyond; a historical overview, with suggestions for future research: part 2. Am Rev Respir Dis 1992; 146: 1615–1622.
22. Grievink L, Smit HA, Ocké MC, van ‘t Veer P, Kromhout D. Dietary intake of antioxidant (pro)-vitamins, respiratory symptoms and pulmonary function: the MORGEN-study. Thorax 1998; 53: 166–171.
23. Sparrow D, Rosner B, Cohen B, Weiss ST. Alcohol consumption and pulmonary function: a cross-sectional and longitudinal study. Am Rev Respir Dis 1983; 127: 735–738.
24. Shaper AG. Alcohol and mortality: a review of prospective studies. Br J Addict 1990; 85: 837–947.
25. Astry CL, Warr GA, Jakab GJ. Impairment of polymorphonuclear leukocyte immigration as a mechanism of alcohol-induced suppression of pulmonary antibacterial defenses. Am Rev Respir Dis 1983; 128: 113–117.
26. MacGregor RR, Safford M, Shalit M. Effect of ethanol on functions required for the delivery of neutrophils to sites of inflammation. J Infect Dis 1988; 157: 682–689.
27. Maxwell WJ, Keating JJ, Hogan FP, Kennedy NP, Keeling PWN. Prostaglandine E 2 and leukotriene B4 synthesis by peripheral leucocytes in alcoholics. Gut 1989; 30: 1270–1274.
28. Rettori V, McCann SM. Role of nitric oxide and alcohol on gonadotropin release in vitro
and in vivo
. Ann NY Acad Sci 1998; 840: 185–193.
29. Dockery DW, Speizer FE, Ferris BG, Ware JH, Louis TA, Spiro III A. Cumulative and reversible effects of lifetime smoking on simple tests of lung function in adults. Am Rev Respir Dis 1988; 137: 286–292.
30. Smit HA, Grievink L, Tabak C. Dietary influences on chronic obstructive lung disease and asthma: a review of the epidemiological evidence. Proc Nutr Soc 1999; 58: 309–319.
Keywords:© 2001 Lippincott Williams & Wilkins, Inc.
alcohol consumption,; chronic obstructive pulmonary disease mortality,; pulmonary function,; gender,; international comparisons