Age, alcohol consumption, and all-cause mortality

Age, alcohol consumption, and all-cause mortality

Age, Alcohol Consumption, and All-cause Mortality VOLKER ARNDT, MD, MPH, DIETRICH ROTHENBACHER, MD, MPH, REINHARD KRAULEDAT, MD, ULRICH DANIEL, MD, AN...

71KB Sizes 2 Downloads 77 Views

Age, Alcohol Consumption, and All-cause Mortality VOLKER ARNDT, MD, MPH, DIETRICH ROTHENBACHER, MD, MPH, REINHARD KRAULEDAT, MD, ULRICH DANIEL, MD, AND HERMANN BRENNER, MD, MPH

PURPOSE: Numerous epidemiological studies on middle aged men suggest that moderate alcohol consumption lowers all-cause mortality. Uncertainty exists whether this finding also applies to younger adults. METHODS: The authors studied the age specific association between alcohol consumption and allcause mortality in a cohort of 19,943 male employees in the German construction industry (age range 25–64 years) who underwent an occupational health examination between 1986 and 1992 and were followed with respect to vital status over an average period of 10 years. Information on alcohol consumption at baseline examination relied on self report and strongly correlated with biological markers of alcohol consumption. Covariates considered in the proportional hazard regression analysis included age, nationality, smoking status, and comorbidity. RESULTS: Age specific analysis revealed that the J/U-shaped association between alcohol consumption and all-cause mortality is limited to men aged 35 years and above. In contrast, a linear positive dose– response relationship was observed for men aged 25 to 34 years (p-trend ⫽ 0.02). CONCLUSIONS: Recommendations regarding alcohol consumption and potential health benefits should not be generalized to all ages. Ann Epidemiol 2004;14:750–753. 쑕 2004 Elsevier Inc. All rights reserved. KEY WORDS:

Age Factors, Alcohol Drinking, Dose–Response Relationship, Mortality.

INTRODUCTION A J- or U-shaped association between alcohol consumption and all-cause mortality has been reported in numerous epidemiological studies (1). The particular shape of the association is thought to reflect a combination of protective effects seen primarily for coronary heart disease and increased risk of other diseases (such as cardiomyopathy, hypertension, liver disease, certain types of cancer, and violent deaths). However, existing studies rely almost exclusively on middle aged men, and it is questionable whether potential health benefits of moderate alcohol consumption can be generalized to all age groups. Results from a recent meta-regression analysis suggest that the older the persons at baseline the more pronounced the protective effect (2). However, this study was restricted to the age groups 45 years and above. Protective effects of moderate alcohol consumption were also observed for younger (30–59 years) and older (60–79 years) participants of the large American Cancer Society II study,

From the German Centre for Research on Ageing, Department of Epidemiology, Heidelberg, Germany (V.A., D.R., R.K., H.B.); and Occupational Health Service, Workmen’s Compensation Board for Construction Workers Wu¨rttemberg, Bo¨blingen, Germany (U.D.). Address correspondence to: Volker Arndt, M.D., M.P.H., German Centre for Research on Ageing (DZFA), Department of Epidemiology, Bergheimer Strasse 20, D-69115 Heidelberg, Germany. Tel.: ⫹49-6221548144; Fax: ⫹49-6221-548142. E-mail: [email protected] This study was supported by the Association of the Workmen’s Compensation Board for Construction Workers, Germany. Received November 17, 2003; accepted January 26, 2004. 쑕 2004 Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010

but the cardioprotective effects appeared to be stronger among the elderly (3). Conversely, a linear increase in allcause mortality with greater alcohol consumption was found in a study of young Swedish military recruits (18–19 years) (4–5). In contrast, age did not modify the U-shaped association between alcohol consumption and all-cause mortality in a Danish study among 30- to 70-year-old men and women (6). Finally, a recent ecological study from the UK indicated that the U-shaped association might be limited to men aged over 35 years and women aged over 65 years, whereas a positive linear dose–response was reported for younger ages (7). Although the majority of these studies support the hypothesis that age modifies the association between alcohol consumption and mortality, uncertainty persists whether a potential beneficial effect of alcohol drinking also exists for younger adults. We therefore studied the age specific relative risk for all cause mortality according to alcohol consumption in a large cohort of young and middle aged men in Germany, which is one of the countries with the highest per capita alcohol consumption in the world.

METHODS A cohort study was set up with 19,943 male employees from the German construction industry (age range 25–64 years) who underwent routine occupational health exams between 1986 and 1992 and were followed over an average period 1047-2797/04/$–see front matter doi:10.1016/j.annepidem.2004.01.004

AEP Vol. 14, No. 10 November 2004: 750–753

Arndt et al. AGE, ALCOHOL AND MORTALITY

of 10 years. Details of the data collection procedure have previously been reported in earlier studies among employees recruited between 1986 and 1988 (8–9). Briefly, health exams included a standardized physical examination, functional measurements (such as electrocardiograms, spirometry, audiometry) as well as laboratory measures, which were taken according to a standardized protocol. In addition, occupational and life style factors were recorded. Quantitative information on alcohol consumption was reported by 17,279 study members (86.6%) at baseline. Average daily amount of ethanol was calculated from frequency, type of beverage, and amount of alcohol consumption and was categorized into none; occasional but not daily; daily up to 30 g; 31 to 60 g/day; 61 to 90 g/day; and over 90 g/day. In the absence of a national death index in Germany, residents’ registration offices were contacted to obtain most recent information on vital status of all cohort members between October 1998 and February 2000. Study members who moved to a foreign country (n = 719; 3.6%) or with unknown place of residence (n = 159; 0.8%) were censored with date of last known residence. No follow-up information

751

was available for 116 subjects of the baseline population (0.6%). Overall and age specific relative risks of dying depending on level of daily alcohol consumption were calculated using Cox’s proportional hazards model. Baseline variables considered as covariates in the multivariate analysis included age, nationality, smoking status, and comorbidity. All analyses were restricted to cohort members with quantitative information on alcohol consumption. Study members without quantitative data of alcohol consumption did not differ from the rest of the study population with respect to baseline characteristics nor with respect to overall survival. In addition, 47 abstainers with a history of alcoholic psychosis (ICD-9: 291), alcohol dependence syndrome (ICD-9: 303), or alcohol abuse (ICD-9: 305.0) were excluded from survival analysis. RESULTS Characteristics of the study participants at baseline are shown in Table 1. Smoking status and proportion of foreign

TABLE 1. Characteristics of the study population by self-reported alcohol consumption at baseline and overall results from follow-up Self reported alcohol consumption

Baseline characteristics Age at baseline: mean (SD)* Proportion of - foreign workers - current smokers Blood parameters (median)* GGT (U/l) ALAT (U/l) ASAT (U/l) Baseline prevalence of† Gastrointestinal disorders (ICD-9: 520–579) Liver disorders (ICD-9: 570–573) Hypertension (ICD-9: 401–405) Diabetes mellitus (ICD-9: 250) Alcohol disorders (ICD-9: 291, 305, 305.0) Follow-up Completeness Number of deaths Person years (PY) Crude death rate (per 100,000 PY)

None n ⫽ 1835

Occasional n ⫽ 7495

1–30 g/day n ⫽ 1349

31–60 g/day n ⫽ 3223

61–90 g/day n ⫽ 1655

⬎ 90 g/day n ⫽ 1722

Total n ⫽ 17,279

43.6 (10.5)

40.0 (11.1)

42.9 (11.0)

44.3 (10.4)

45.1 (9.9)

45.6 (9.3)

42.5 (10.9)

51.5% 46.8%

19.5% 53.3%

22.9% 51.7%

23.9% 58.8%

22.5% 62.5%

23.7% 67.7%

24.7 55.8

13.0 13.0 10.0

16.0 14.0 11.0

18.0 14.0 11.0

22.0 15.0 11.0

28.0 17.0 13.0

38.0 19.0 14.0

19.0 15.0 11.0

22.7%

18.6%

19.6%

25.4%

36.2%

49.3%

25.1%

4.5%

6.8%

7.6%

13.1%

22.8%

35.7%

12.2%

16.9%

18.0%

21.5%

25.0%

30.6%

32.3%

22.1%

5.0%

4.2%

4.5%

4.3%

5.5%

7.0%

4.8%

2.6%

1.3%

1.1%

2.0%

3.5%

7.8%

2.4%

98.4% 73 15,397 474

99.4% 187 68,954 271

99.6% 43 12,395 347

99.6% 128 30,363 422

99.8% 105 15,614 672

99.4% 162 15,896 1019

99.4% 698 158,619 440

*SD, standard deviation; GGT, gammaglutamyl-transferase; ASAT, aspartate aminotransferase; ALAT, alanine aminotransferase. † Diagnosis recorded by physicians in the context of the occupational health exam.

752 Arndt et al.

AEP Vol. 14, No. 10 November 2004: 750–753

AGE, ALCOHOL AND MORTALITY

workers varied according to level of alcohol consumption. In particular, workers of foreign nationality, who represented approximately one fourth of the study population, were overrepresented among non-drinkers due to a substantial proportion of Moslem migrants from Turkey and former Yugoslavia. A monotonic positive dose–response relationship was observed between self reported alcohol consumption and blood parameters known to be related to alcohol consumption. Also, baseline prevalence of liver and other disorders known be associated with alcohol consumption sharply increased with increasing categories of alcohol consumption. During follow-up, 698 study participants died. There was a strong J-shaped association between alcohol consumption and all-cause mortality in the total cohort (Table 2). Mortality was 1.52 times higher among non-drinkers and 2.31 times higher among heavy drinkers (⬎ 90 g alcohol per day) compared with occasional drinkers. The nadir of lowest mortality was broad and ranged from occasional drinking up to consumption of 60 grams of alcohol per day. Age specific analyses revealed that the association between alcohol consumption and mortality was J/U-shaped for all but the youngest age group. Among the group of 25- to 34year-old men, mortality increased already at low levels of daily drinking (p-trend = 0.02), whereas no excess mortality was found for non-drinkers (relative risk = 0.91). Further controlling for baseline disorders which could influence drinking behavior and mortality (such as liver diseases, cancer, ischemic heart disease, diabetes mellitus) did not substantially change the findings.

DISCUSSION Our study indicates that the beneficial effect of moderate drinking with respect to all-cause mortality, which has been reported by numerous epidemiological studies, may be limited to ages over 35 years. Furthermore, this study suggests that daily consumption of even moderate amounts of alcohol below the age of 35 years may increase the risk of dying. The difference between young and old men presumably relates to the cause of death (10). Although the limited numbers of deaths among the youngest age group in our study did not allow for a detailed analysis of cause specific mortality by alcohol consumption, we found that the predominant causes of death among younger construction workers were intentional or unintentional injury. Injuries accounted for 31% of all deaths among the youngest and for 6% of all deaths among the oldest age group. In contrast, the proportion of cardiovascular diseases as cause of death increased from 15% (25–34 years old) to 32% (55–64 years old). Taken together, these patterns are consistent with observations that the cardioprotective effect of moderate alcohol consumption, which is predominantly seen among older people (2, 3), may be outnumbered among young adults by an increased risk of injuries (11). Our conclusions are also supported by findings from a recent ecological study (7), but further results from other individual-level epidemiologic studies are necessary given that our study included male construction workers, who represent a population with a high risk of fatal accidents, and that the confidence intervals for the age specific estimates

TABLE 2. Age specific relative all-cause mortality (95% confidence limits) according to alcohol consumption (abstainers with history of alcohol disease have been excluded) Age at baseline examination All ages (25–64 years) (n ⫽ 17,135; d ⫽ 693)

25–34 years (n ⫽ 5332; d ⫽ 45)

35–44 years (n ⫽ 3566; d ⫽ 85)

45–54 years (n ⫽ 5520; d ⫽ 324)

55–64 years (n ⫽ 2717; d ⫽ 239)

Alcohol consumption

RR

(95% CI)

RR

(95% CI)

RR

(95% CI)

RR

(95% CI)

RR

(95% CI)

None (n ⫽ 1793; d ⫽ 68) Occasional (n ⫽ 7454; d ⫽ 187) 1–30 g/day (n ⫽ 1344; d ⫽ 43) 31–60 g/day (n ⫽ 3210; d ⫽ 128) 61–90 g/day (n ⫽ 1651; d ⫽ 105) ⬎ 90 g/day (n ⫽ 1713; d ⫽ 162)

0.91

(0.26–3.16)

1.63

(0.75–3.55)

1.42

(0.91–2.22)

1.76

(1.07–2.89)

1.52

(1.13–2.08)

1.00

referent

1.00

referent

1.00

referent

1.00

referent

1.00

referent

1.24

(0.37–4.17)

0.71

(0.25–2.02)

1.13

(0.69–1.87)

1.09

(0.63–1.86)

1.05

(0.75–1.46)

1.75

(0.79–3.89)

0.64

(0.31–1.32)

1.10

(0.78–1.56)

1.26

(0.87–1.83)

1.13

(0.90–1.42)

2.06

(0.76–5.61)

0.99

(0.46–2.13)

1.99

(1.42–2.79)

1.31

(0.84–2.03)

1.63

(1.28–2.08)

2.46

(0.88–6.83)

2.58

(1.47–4.51)

2.07

(1.51–2.84)

2.54

(1.74–3.72)

2.31

(1.86–2.87)

n, numbers at risk; d, numbers of deceased; RR, relative adjusted for age (in years with a linear and quadratic term), nationality (German, Turkish, Italian, Yugoslavian, other), smoking (never, former, 1–10 cigarettes/day, 11–20 cigarettes/day, ⬎ 20 cigarettes/day, other); CI, confidence interval.

AEP Vol. 14, No. 10 November 2004: 750–753

of relative risk were wide. Strengths of the study include the good correlation of the alcohol data with specific biological markers, the wide range of alcohol consumption studied (with large numbers of none and heavy drinkers), and the control for potential confounders (including smoking and preexisting disorders). Although based on small number of deaths among younger men, our finding is biologically plausible, and it has important public health implications. While the message of beneficial effects of moderate alcohol consumption among middle aged and older adults is clearly an important one, care has to be taken that studies reporting beneficial health effects of alcohol might encourage excessive drinking. This is of specific concern for adolescents and young adults. Our study indicates that recommendations regarding alcohol consumption and the potential health benefits should not be generalized to all ages. REFERENCES 1. Poikolainen K. Alcohol and mortality: A review. J Clin Epidemiol. 1995;48: 455–465. 2. Gmel G, Gutjahr E, Rehm J. How stable is the risk curve between alcohol and all-cause mortality and what influences the shape? A precisionweighted hierarchical meta-analysis. Eur J Epidemiol. 2003;18:631–642.

Arndt et al. AGE, ALCOHOL AND MORTALITY

753

3. Thun MJ, Peto R, Lopez AD, Monaco JH, Henley SJ, Heath CW Jr, et al. Alcohol consumption and mortality among middle-aged and elderly US adults. N Engl J Med. 1997;337:1705–1714. 4. Andreasson S, Allebeck P, Romelsjo A. Alcohol and mortality among young men: Longitudinal study of Swedish conscripts. BMJ. 1988;296: 1021–1025. 5. Romelsjo A, Leifman A. Association between alcohol consumption and mortality, myocardial infarction, and stroke in 25-year follow-up of 49 618 young Swedish men. BMJ. 1999;319:821–822. 6. Gronbaek M, Deis A, Sorensen TI, Becker U, Borch-Johnsen K, Muller C, et al. Influence of sex, age, body mass index, and smoking on alcohol intake and mortality. BMJ. 1994;308:302–306. 7. White IR, Altmann DR, Nanchahal K. Alcohol consumption and mortality: Modeling risks for men and women at different ages. BMJ. 2002;325: 191–194. 8. Arndt V, Rothenbacher D, Brenner H, Fraisse E, Zschenderlein B, Daniel U, et al. Older workers in the construction industry: Results of a routine health examination and a five-year follow up. Occup Environ Med. 1996;53:686–691. 9. Brenner H, Arndt V, Rothenbacher D, Schuberth S, Fraisse E, Fliedner TM. Body weight, pre-existing disease, and all-cause mortality in a cohort of male employees in the German construction industry. J Clin Epidemiol. 1997;50:1099–1106. 10. Anderson P. Alcohol consumption and all-cause mortality. Addiction. 1995;90:481–484. 11. Britton A, McPherson K. Mortality in England and Wales attributable to current alcohol consumption. J Epidemiol Community Health. 2001;55: 383–388.