Alcohol consumption and blood pressure change: 5-year follow-up study of the association in normotensive workers

(2001) 15, 367 372 2001 Nature Publishing Group All rights reserved 0950-9240/01 $15.00 www.nature.com/jhh ORIGINAL ARTICLE Alcohol consumption and blood pressure change: 5-year follow-up study of the association in normotensive workers Y Okubo, Y Suwazono, E Kobayashi and K Nogawa Department of Hygiene, Chiba University School of Medicine, Japan This study endeavours to clarify the effect of alcohol consumption on changes in the blood pressure of Japanese men. The subjects, who were followed from 1990 to 1995, were 2189 male normotensive steelworkers aged from 40 to 54 years in 1990. Drinking habits were represented by two variables: average daily baseline alcohol consumption, and change in alcohol consumption during the observation period, both derived from self-report questionnaires and interviews. Nine other items were analysed as covariates: systolic and diastolic blood pressure, age, body mass index (BMI), salt intake, physical activity, and levels of gamma glutamyl transpeptidase (GTP), uric acid and plasma glucose in 1990. The association between drinking habits and blood pressure was evaluated using ANCOVA. Subjects who had initiated antihypertensive therapy by 1995 were excluded from this dataset. The results of analysis dis- closed that both daily alcohol consumption and its change were significantly associated with changes in systolic and diastolic blood pressure during the observation period. J-shaped associations of daily alcohol consumption with adjusted changes in systolic and diastolic blood pressure were found. Positive associations were observed between changes in alcohol consumption and adjusted changes in systolic and diastolic blood pressure. In conclusion, daily alcohol consumption was associated non-linearly with changes in blood pressure and with a threshold effect at 18 ml of ethanol per day in these middle-aged Japanese workers. In addition, increasing alcohol consumption was associated with blood pressure elevation, and decreasing consumption was associated with suppression of blood pressure elevation. (2001) 15, 367 372 Keywords: drinking habit; cohort study; linear association; non-linear association; lifestyle Introduction During the last 20 years, excess alcohol consumption has been shown to be associated with elevation in blood pressure and the subsequent development of hypertension. 1 4 However, the precise nature of the association between alcohol consumption and blood pressure remains unclear. Some studies have pointed to a non-linear or J-shaped association, 5 7 but others have suggested this association is linear. 8,9 Therefore, we conducted a cross-sectional study to clarify the relation between drinking habit and blood pressure. In that study, the effect of average monthly alcohol consumption on blood pressure indicated non-linear association in the workers aged 40 44 years and linear association in those 45 54 years in Japanese men (Yasushi Okubo, unpublished data, 2001). This finding is available for education Correspondence: Yasushi Okubo, Department of Hygiene, Chiba University School of Medicine, 1 8 1 Inohana, Chuo-ku, Chiba, Japan 260 8670. E-mail: ookubo med.m.chiba-u.ac.jp Received 25 September 2000; revised and accepted 18 January 2001 to promote moderation or abstinence in health management and promotion. However, we could not confirm the effect of drinking habit on change in blood pressure in our cross-sectional study. Thus, we conducted a 5-year follow-up study to investigate the effects of drinking habit and changes in drinking habit on changes in blood pressure in normotensive middle-aged Japanese workers. Materials and methods Subjects The subjects of this follow-up study were 2999 male Japanese steelworkers aged from 40 to 54 years who showed normal blood pressure (systolic blood pressure 140 mm Hg and diastolic blood pressure 90 mm Hg) in an annual health examination in 1990. Subjects who had been diagnosed with hypertension and had initiated antihypertensive therapy in 1990 or before were excluded. During the observation period, 810 subjects had retired or been transferred to another factory. In addition, we excluded 46 subjects who had initiated antihypertensive therapy by

368 Effects of drinking on blood pressure 1995 from analysis of covariance (ANCOVA). Consequently, we included 2143 cases in our ANCOVA dataset (Figure 1). Blood pressure measurement Blood pressure was measured in a sitting position by a nurse using an automatic sphygmomanometer (BP103 made by Colin Medical Instruments) 10 at annual health examinations in 1990 and 1995. If blood pressure showed abnormal level, we repeated the procedure three times at 5-min intervals. The lowest blood pressure reading was adopted. Drinking habit To evaluate drinking habits, we administered a selfreport questionnaire at the same annual health examination, soliciting the respondents average alcohol consumption during a typical drinking session and their average frequency of drinking sessions. Average daily alcohol consumption in 1990 was the value obtained when the product of the former two items was divided by 30. Change in alcohol consumption during the observation period was simply the difference of daily alcohol consumption in 1990 from that in 1995. A public health nurse confirmed the results of the questionnaire in individual interviews. Daily alcohol consumption was divided into four levels: non-drinker, light drinker; two drinks (18.0 ml of ethanol) per day or less, moderate drinker; two to nine drinks (18.1 45.0 ml of ethanol) per day, and heavy drinker; more than nine drinks (45.0 ml of ethanol) per day. Change in alcohol consumption was classified into three categories: decreased, unchanged, and increased. The number of subjects, classed by age group and drinking habits, is shown in Table 1. Blood test and lifestyle We investigated nine other variables: systolic and diastolic blood pressure, age, body mass index (BMI), salt intake, physical activity, and levels of gamma glutamyl transpeptidase (GTP), uric acid and plasma glucose in 1990. Salt intake was calculated based on self-report data concerning the average content of meals. The level of physical activity was calculated according to subjects reports of their type, duration, and frequency of exercise, as well as calorie consumption at work. Systolic and diastolic blood pressure, BMI, GTP, uric acid and plasma glucose levels were derived from the results of the 1990 annual health examination. Physical activity and levels of GTP, uric acid and plasma glucose were transformed logarithmically in consideration of their distributions. Characteristics of subjects according to age group are shown in Table 2. Statistical analysis In univariate analysis, the prevalence of high blood pressure in 1995 according to age group, daily alcohol consumption, and changes in alcohol consumption was analysed using the Cochran Armitage test as a trend indicator. In multivariate analysis, the associations of drinking habits with changes in systolic and diastolic blood pressure were evaluated using ANCOVA. Each of the changes in systolic and diastolic blood pressure during the observation period was used as a dependent variable. Daily alcohol consumption and changes in alcohol consumption served as the fixed factors among the independent variables. Systolic and diastolic blood pressure, age, BMI, salt intake, physical activity, and the levels of GTP, uric acid and plasma glucose in 1990 were used as covariates. No significant interaction between daily alcohol consumption and changes in alcohol consumption was observed, thus only the main effects of these were used in our analysis. In addition, we estimated the adjusted mean of changes in systolic and diastolic blood pressures for comparison among the categories of each drinking habit variable. Bonferroni s procedure was used for post hoc testing of these multiple comparisons. All statistical analysis was performed using SPSS version 7.5. Figure 1 Subjects profile. Results Comparison between retired or transferred subjects and follow-up subjects No difference in the distribution of drinking habit between retired or transferred subjects and followup subjects was observed (Table 1). No difference in

Effects of drinking on blood pressure Table 1 Number of participants according to age group in 1990 369 Age group Daily alcohol consumption Changes in alcohol consumption (years) Non Light Moderate Heavy Increased Unchanged Decreased Total drinker drinker drinker drinker Follow-up subjects 40 44 89 (0) 217 (3) 291 (3) 150 (2) 240 (1) 278 (3) 229 (4) 747 (8) 45 49 151 (1) 314 (14) 454 (10) 217 (4) 345 (9) 470 (10) 321 (10) 1136 (29) 50 54 41 (2) 75 (0) 130 (5) 60 (2) 94 (3) 127 (3) 85 (3) 306 (9) Total 281 (3) 606 (17) 875 (18) 427 (8) 679 (13) 875 (16) 635 (17) 2189 (46) Retired or transferred subjects 40 44 8 26 24 8 NS 66 45 49 33 78 90 54 NS 255 50 54 97 119 194 79 NS 489 Total 221 419 595 305 NS 810 ( ), The number of the patients who were excluded from ANCOVA. Non-drinker: including abstainer. Light drinker: consuming two drinks (18.0 ml of ethanol) per day or less. Moderate drinker: consuming two to nine drinks (18.1 45.0 ml of ethanol) per day. Heavy drinker: consuming more than nine drinks (45.0 ml of ethanol) per day. Chi-squared test was performed to compare retired or transferred subjects and follow-up subjects according to age group. NS not significant. Table 2 Characteristics of participants at the baseline according to age group in 1990 Follow-up subjects (years) Retired or transferred subjects (years) 40 44 54 49 50 54 40 44 45 49 50 54 Mean s.d. mean s.d. mean s.d. mean s.d. mean s.d. mean s.d. BMI (kg/m 2 ) 23.22 2.62 23.02 2.40 23.21 2.35 23.09 2.56 NS 23.15 2.46 NS 23.32 2.34 NS Systolic blood 122.55 10.66 121.43 11.11 123.89 10.88 122.27 10.6 NS 121.76 11.2 NS 122.63 11.4 NS pressure (mm Hg) Diastolic blood 72.91 7.41 73.09 7.84 75.20 7.39 72.7 7.47 NS 73.49 7.85 NS 73.91 7.79 NS pressure (mm/hg) Salt intake (g/day) 12.44 2.18 12.22 2.22 12.29 2.16 12.44 2.26 NS 12.25 2.33 NS 12.12 2.37 NS GM GSD GM GSD GM GSD GM GSD GM GSD GM GSD Habitual exercise 40.20 3.71 45.93 3.73 45.08 3.87 42.35 3.58 NS 47.31 3.55 NS 41.68 4.15 NS (kcal/day) GTP (IU) 24.53 1.93 25.45 2.00 25.91 2.06 24.23 1.95 NS 26.33 2.11 NS 24.4 2.01 NS Uric acid (mg/dl) 5.51 1.27 5.38 1.26 5.36 1.24 5.47 1.27 NS 5.46 1.26 NS 5.35 1.23 NS Plasma glucose 92.25 1.15 95.05 1.15 95.61 1.16 92.67 1.16 NS 95.03 1.15 NS 96.92 1.16 NS (mg/dl) No. of participants 747 1136 306 66 255 489 t-test was performed to compared retired or transferred subjects and follow-up subjects according to age group. s.d., standard deviation; GM, geometric mean; GSD, geometric standard deviation; NS, not significant. mean of each covariate was observed either (Table 2). Prevalence of high blood pressure in 1995 The prevalence of high blood pressure exceeding normal levels according to age group, daily alcohol consumption, and changes in alcohol consumption is shown in Table 3. High blood pressure was divided in the table into borderline and hypertension. The subjects who had initiated antihypertensive therapy were classified into the hypertension group. The prevalence of high blood pressure had increased with increasing of age group, drinking habit in 1990, and changes in drinking habit, and each tendency was significant. Daily alcohol consumption Adjusted means of changes in systolic and diastolic blood pressure according to daily alcohol consumption and the results of the multiple comparison are shown in Figure 2. No significant difference of adjusted change in either systolic or diastolic blood pressure between non-drinkers and light drinkers was found, above that level, adjusted changes in

Effects of drinking on blood pressure 370 Table 3 Prevalence of abnormal blood pressure in 1995 according to age group in 1990 Abnormal blood pressure Normotension Border Hypertension Total P n % n % n % n % Age group 40 44 years 684 91.57 47 6.29 16 (8) 2.14 63 8.43 45 49 years 1001 88.12 95 8.36 40 (29) 3.52 135 11.88 50 54 years 250 81.70 39 12.75 17 (9) 5.55 56 18.30 *** Daily alcohol consumption Non drinker 263 93.59 12 4.27 6 (3) 2.14 18 6.41 Light drinker 543 89.60 42 6.93 21 (17) 3.47 63 10.40 Moderate drinker 765 87.43 79 9.03 31 (18) 3.54 110 12.57 Heavy drinker 364 85.25 48 11.24 15 (8) 3.51 63 14.75 *** Changes in alcohol consumption Decreased 608 89.54 50 7.36 21 (13) 3.09 71 10.46 Unchanged 784 89.60 64 7.31 27 (16) 3.09 91 10.40 Increased 543 85.51 67 10.55 25 (17) 3.94 92 14.49 * Normotension: systolic blood pressure 140 mm Hg and diastolic blood pressure 90 mm Hg. Border: systolic blood pressure = 140 159 mm Hg and/or diastolic blood pressure = 90 94 mm Hg. Hypertension: systolic blood pressure 160 mm Hg or diastolic blood pressure 95 mm Hg. ( ): The number of the subjects who had under antihypertensive treatment in the hypertensive subjects. Cochran Armitage test were performed to evaluate for trend of prevalence of abnormal blood pressure including border and hypertension. *P 0.05, ***P 0.001. Figure 2 Comparison of changes in systolic and diastolic blood pressure during 5 years according to daily alcohol consumption in 1990 (mean ± s.e.). Light drinker: consuming two drinks (18.0 ml of ethanol) per day or less. Moderate drinker: consuming two to nine drinks (18.1 45.0 ml of ethanol) per day. Heavy drinker: consuming more than nine drinks (45.0 ml of ethanol) per day. Changes in alcohol consumption, age, BMI, SBP in 1990, DBP in 1990, salt intake, habitual exercise, GTP, uric acid and plasma glucose were adjusted using ANCOVA. + P 0.10, *P 0.05, **P 0.01 (Bonferroni s multiple comparison). both systolic and diastolic blood pressure were elevated with increasing of daily alcohol consumption. Therefore, these findings indicate J-shaped association and a threshold effect of consuming 18.0 ml of ethanol per day for systolic and diastolic blood pressure elevation. Changes in alcohol consumption over 5 years Adjusted means of changes in systolic and diastolic blood pressure according to changes in alcohol consumption and the results of the multiple comparison are shown in Figure 3. On the whole, adjusted change in both systolic and diastolic blood pressure was elevated with the increase in alcohol consumption in the model including all subjects. In addition, when we conducted the same analysis in the model including only moderate/heavy drinkers, we observed the same trend. These findings indicated that change in alcohol consumption was positively associated with change in adjusted blood pressure in middle-aged Japanese men.

Effects of drinking on blood pressure 371 Figure 3 Comparison of changes in systolic and diastolic blood pressure during 5 years according to changes in alcohol consumption (mean ± s.e.). Daily alcohol consumption, age, BMI, SBP in 1990, DBP in 1990, salt intake, habitual exercise, GTP, uric acid and plasma glucose were adjusted using ANCOVA. + P 0.10, *P 0.05, **P 0.01 (Bonferoni s multiple comparison). Discussion It is generally accepted that excessive drinking elevates blood pressure. However, uncertainties remain to be elucidated. The first concern is whether changes in blood pressure is affected more by longterm baseline alcohol consumption or by a change in consumption. The second question is whether these associations are linear or non-linear. With regard to the first concern, it is reported that increases in systolic and diastolic blood pressure were significantly greater in heavy drinkers. 1,11 Similarly, Nakanishi et al 12 examined 934 normotensive Japanese for 6 years and suggested that alcohol consumption was independently associated with hypertension. On the other hand, some studies have reported that baseline alcohol consumption is significantly associated with baseline blood pressure, but not with changes in blood pressure during the observation period. 13,14 These conflicting findings suggest that the effects of initial daily alcohol consumption on changes in blood pressure remain unconfirmed. Our study found, nevertheless, that initial daily alcohol consumption significantly and independently associates with changes in systolic and diastolic blood pressure in middle-aged Japanese workers. There have been some studies reporting a significant association between changes in alcohol consumption and changes in blood pressure. 11,14 16 The subjects in these studies consisted of both hypertensive and normotensive participants. In our study, we investigated only normotensive workers and found a significant association. As to the second question, our review of the literature found a study investigating the precise nature of the association between initial alcohol consumption and change in blood pressure. Dyer et al 17 observed an elevation of blood pressure in problem drinkers greater than that in so-called non-problem drinkers. In addition, in non-problem drinkers consuming less than six drinks per day, no significant association was found between alcohol consumption and change in blood pressure. This finding suggests a non-linear association and the existence of a threshold effect. We observed a J-shaped association between alcohol consumption and changes in blood pressure, and a threshold effect at 18.0 ml of ethanol consumption per day. Hence, we suggest that a small amount of alcohol consumption suppresses both systolic and diastolic blood pressure elevation. The above-mentioned studies 11,14 16 and intervention studies 18 23 indicate that an increase in alcohol consumption is associated with a significant increase in blood pressure, whereas decreased alcohol consumption is associated with a significant decrease in blood pressure in the subjects including hypertensive participants. In this study, we found that changes in alcohol consumption linearly associated with changes in blood pressure even in normotensive subjects. The blood pressure of non-drinkers has been observed to be greater than that of light drinkers; suggested reasons for this include an inflow of hypertensive patients who stopped drinking to reduce their blood pressure, the effect of obesity as a confounding factor, and underestimation of drinking habits due to self-report data anomalies. 24 In our study, we examined only normotensive subjects who had no history of hypertension. Therefore, those who stopped drinking because of abnormal blood pressure were not included. In addition, we examined BMI and other factors as confounding factors and confirmed the results of our self-report questionnaire using personal interviews by a public health nurse. Consequently, we are confident of the reliability of this study s data.

372 References Effects of drinking on blood pressure 1 Dyer AR et al. Alcohol consumption, cardiovascular risk factors, and mortality in two Chicago epidemiologic studies. Circulation 1977; 56: 1067 1074. 2 Kozararevic DJ et al. Frequency of alcohol consumption and morbidity and mortality. Lancet 1980; 2: 613 616. 3 Kiyohara Y et al. The impact of alcohol and hypertension on stroke incidence in a general Japanese population. Stroke 1995; 26: 368 372. 4 Iso H et al. Alcohol intake and the risk of cardiovascular disease in middle-aged Japanese men. Stroke 1995; 26: 767 773. 5 Klastky AL, Friedman GD, Armstrong MA. The relationship between alcoholic beverage use and other traits to blood pressure: a new Kaiser Permanente study. Circulation 1986; 73: 628 636. 6 Keil U, Chambless L, Remmers A. Alcohol and blood pressure: results from the Luebeck blood pressure study. Prev Med 1989; 18: 1 10. 7 MacMahon SW, Blacket RB, Macdonald GJ, Hall W. Obesity, alcohol consumption and blood pressure in Australian men and women: The National Heart Foundation of Australia risk factor prevalence study. J Hypertens 1984; 2: 85 91. 8 Klatsky AL, Friedman GD, Siegelaub AB, Gerard MJ. Alcohol consumption and blood pressure. N Engl J Med 1977; 296: 1194 1200. 9 Harburg E, Ozugoren F, Hawthorne VM, Schork MA. Community norms of alcohol usage and blood pressure: Tecumseh, Michigan. Am J Pub Health 1980; 70: 813 820. 10 Sun M, Jones R. Accuracy relationship of blood pressure devices between the AAMI SP 10 standard and the British Hypertension Society protocols. Biomed Instrum Tech 1999; 33: 62 70. 11 DeFrank RS, Jenkins CD, Rose RM. A longitudinal investigation of the relationships among alcohol consumption, psychosocial factors, and blood pressure. Psychosomatic Med 1987; 49: 236 249. 12 Nakanishi N et al. Risk factors for the development of hypertension: a 6-year longitudinal study of middle aged Japanese men. J Hypertens 1998; 16: 753 759. 13 Reed D, McGee D, Yano K. Biological and social correlates of blood pressure among Japanese men in Hawaii. Hypertension 1983; 4: 406 414. 14 Curtis AB et al. Alcohol consumption and change in blood pressure among African Americans. The Pitt County Study. Am J Epidemiol 1997; 146: 727 733. 15 Gordon T, Kannel WB. Drinking and its relation to smoking, BP, blood lipids, and uric acid. Arch Intern Med 1983; 143: 1366 1374. 16 Gordon T, Doyle JT. Alcohol consumption and its relationship to smoking, weight, blood pressure, and blood lipids: The Albany Study. Arch Intern Med 1986; 146: 262 265. 17 Dyer AR et al. Alcohol, cardiovascular risk factors and mortality: the Chicago experience. Circulation 1981; 64 (Suppl 3): 20 27. 18 Malhotra H, Mehta SR, Mathur D, Khandelwal PD. Pressor effects of alcohol in normotensive and hypertensive subjects. Lancet 1985; 7: 584 586. 19 Potter JF, Beevers DG. Pressor effect of alcohol in hypertension. Lancet 1984; 6: 119 122. 20 Puddey IB, Beilin LJ, Vandongen R. Regular alcohol use raises blood pressure in treated hypertensive subjects. A randomized controlled trial. Lancet 1987; 9: 647 651. 21 Parker M, Puddey IB, Beilin LJ, Vandongen R. Two way factorial study of alcohol and salt restriction in treated hypertensive men. Hypertension 1990; 16: 398 406. 22 Maheswaran R, Beevers M, Beevers DG. Effectiveness of advice to reduce alcohol consumption in hypertensive patients. Hypertension 1992; 19: 79 84. 23 Ueshima H et al. Effect of reduced alcohol consumption on blood pressure in untreated hypertensive men. Hypertension 1993; 21: 248 252. 24 MacMahon S. Alcohol consumption and hypertension. Hypertension 1987; 9: 111 121.