Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Water, Other Fluids, and Fatal Coronary Heart Disease: The Adventist Health Study

Water, Other Fluids, and Fatal Coronary Heart Disease: The Adventist Health Study Abstract Whole blood viscosity, plasma viscosity, hematocrit, and fibrinogen are considered independent risk factors for coronary heart disease and can be elevated by dehydration. The associations between fatal coronary heart disease and intake of water and fluids other than water were examined among the 8,280 male and 12,017 female participants aged 38–100 years who were without heart disease, stroke, or diabetes at baseline in 1976 in the Adventist Health Study, a prospective cohort study. A total of 246 fatal coronary heart disease events occurred during the 6-year follow-up. High daily intakes of water (five or more glasses) compared with low (two or fewer glasses) were associated with a relative risk in men of 0.46 (95% confidence interval (CI): 0.28, 0.75; p trend = 0.001) and, in women, of 0.59 (95% CI: 0.36, 0.97). A high versus low intake of fluids other than water was associated with a relative risk of 2.47 (95% CI: 1.04, 5.88) in women and of 1.46 (95% CI: 0.7, 3.03) in men. All associations remained virtually unchanged in multivariate analysis adjusting for age, smoking, hypertension, body mass index, education, and (in women only) hormone replacement therapy. Fluid intake as a putative coronary heart disease risk factor may deserve further consideration in other populations or using other study designs. blood viscosity, coronary disease, dehydration, fluids and secretions, hemorheology, men, water, women CI, confidence interval Whole blood viscosity (1–3), plasma viscosity (4), fi-brinogen (1–5), and hematocrit (2–6) levels are positively correlated with coronary heart disease. Even in the high “normal” range, they have been considered independent risk factors (2, 4–7). In some reports, the magnitudes of their effects are comparable with those of smoking, diastolic blood pressure, and low density serum cholesterol (2, 4, 6). These hemorheologic factors are found to be elevated years before manifestation of acute ischemic events (4) and are implicated in the early stages (6), acceleration (8), and extent (9) of atherosclerosis, as well as prognosis in patients with arteriosclerotic disease (5, 10). They are also associated with hypertension (11) and intermittent claudication (12). Additionally, a higher hematocrit level has been associated with tachycardia (13), infarct size (14), reduced oxygen transport (15), and reduced myocardial perfusion (5). Elevation of hemorheologic factors can result from chronic hypohydration (16, 17). They also respond acutely to circadian changes in hydration (18, 19), normal daily activities (20), and use of medications such as diuretics (21). Moreover, an increased risk of coronary heart disease has been associated with circadian increases in hematocrit and fibrinogen (22). This may result in part from an increased risk of coronary thrombosis (4, 21). Because of these reported relations, the authors prospectively examined the associations between fluid intake and risk of fatal coronary heart disease, using data from the Adventist Health Study, a large cohort study. The effects of plain water and fluids other than water were examined separately because both composition and volume may affect a fluid's impact on acute and chronic hydration (23, 24). MATERIALS AND METHODS Study design and participants The Adventist Health Study is a cohort study designed to test the effects of lifestyle on the risk of coronary heart disease, cancer, and all-cause mortality. Details have been described elsewhere (25). Briefly, in 1976, persons aged 25 years or more living in California Seventh-day Adventist households were sent a detailed lifestyle questionnaire that was completed by 34,192 White, non-Hispanic persons (75 percent response rate). Analyses are confined to the 8,280 males and 12,017 females who reported no physician-diagnosed heart disease, stroke, or diabetes and were 38 years of age or more at baseline (because no fatal coronary heart disease events occurred in younger subjects). The lifestyle questionnaire included information on traditional and dietary risk factors of coronary heart disease and intake of all types of beverages. Intake of water was reported as the number of glasses consumed daily (less than one, from one to two, from three to four, from five to six, and more than six), whereas intake of other fluids was reported as the frequency of intake per day, week, or month. The cohort was followed for 6 years. Follow-up data for coronary heart disease events were complete for 97 percent of the participants (25). The criteria for diagnosing fatal coronary heart disease in this study have been described in detail elsewhere (26). Briefly, fatal coronary heart disease defined as either “definite fatal myocardial infarction” or “other definite fatal coronary heart disease” required death within 30 days of a myocardial infarction confirmed by hospital records, electrocardiogram and cardiac enzymes, or fresh myocardial infarction recorded at autopsy. Deaths were found by computer-assisted linkage with the California death certificate files, the National Death Index, use of church records, contact with relatives, and, when available, autopsy reports. Statistical analysis For analysis, intake of all types of fluids was recoded to monthly intakes. A validity study on a subset of 147 participants of the Adventist Health Study found that the average serving size of fluids other than water was close to one 8-ounce (240 ml) glass (range, 0.94–1.28 glasses) (26). Alcoholic beverages were not included as part of the total fluid intake because they were consumed by less than 11 percent of this special population and then in very small amounts. Participants with a total fluid intake of less than once per day (n = 55) or with a daily milk intake greater than seven times daily (n = 206) were considered outliers and excluded from analysis. However, including them in a sensitivity analysis of the final model did not significantly change the results. The use of three exposure categories allowed for analysis of dose-response relations. For water, these categories were obtained by collapsing the two lowest and the two highest intake levels. Similar categories for intake of fluids other than water were formed using average portion size data. This formed low intake comparison categories that were biologically reasonable with sufficient numbers of events to provide stability in multivariate analysis. The Statistical Package for the Social Sciences (SPSS for Windows: Advanced Statistics Release 7.0, 1995, and 9.0.0, 1998; SPSS, Inc., Chicago, Illinois) was used for all analyses, including gender-specific Cox proportional hazard models (27). Subjects with missing values in any of the variables included in a model were excluded from the analysis. The assumption of proportional hazards was not violated as tested by the log-minus-log survival plots (27). In addition to the age-adjusted model (model 1), eight other models were examined. Model 2 adjusted for age and several conventional coronary heart disease risk factors (smoking, hypertension, educational level, body mass index, and, in women, estrogen replacement therapy). Models 3–8, in addition, adjusted for other variables one at a time: fluid other than that being tested, energy intake, exercise, and dietary factors found to be related to coronary heart disease in the Adventist Health Study (nuts, type of bread, and, in men, meat intake) (26, 28, 29), respectively. This kept the number of variables in a particular model in reasonable proportion to the number of events (27). In model 9, weight replaced body mass index as a variable in model 2. The consistency of the association between water intake and risk of fatal coronary heart disease was further explored in age- and sex-adjusted models applied to subgroups of the population defined by dichotomization of 11 different coronary heart disease risk factors. RESULTS During follow-up, 246 fatal coronary heart disease events (128 in men, 118 in women) were reported. Baseline characteristics of the study population are shown in table 1. The subjects were well educated, 37 percent of men and 19 percent of women being college graduates. Overall, the population adhered to a lifestyle that is commonly held to be healthy, as assessed by their diet and level of exercise. Very few subjects drank alcoholic beverages or were current smokers, and these were usually non-Adventists living in Adventist households (8.7 percent of the total Adventist Health Study population). The mean daily intake of beverages among the study population and the general population is shown in table 2. Water was the fluid consumed in greatest amounts. Compared with the national averages (30), the Adventist Health Study population drank more water, milk, and fruit juices and less coffee, tea, and carbonated and alcoholic beverages. TABLE 1. Selected baseline characteristics of study participants 38 years of age or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study Selected characteristic  Men (n = 8,280)  Women (n = 12,017)  Age (mean years (SD*))  57.6 (12.5)  59.2 (12.9)  Hypertensive (%)†  17  24  Smoking (%)       Never smoked  61  86   Smoked in the past  33  12   Currently smokes  6  2  Body mass index (tertiles)‡       Low  <23.4  <21.8   Medium  23.4–25.9  21.8–25.2   High  >25.9  >25.2  Education level (%)       High school  34  39   Some college  30  43   College graduate  37  19  Exercise level (%)§       None to low  31  39   Moderate  23  20   High  46  41  Energy intake (kcal quintiles)       1  <2,100  <1,588   2  2,100–<2,268  1,588–<1,743   3  2,268–<2,401  1,743–<1,876   4  2,401–<2,562  1,876–<2,028   5  ≥2,562  ≥2,028  Weight (lb¶ quintiles)       1  <150  <120   2  150–<165  120–<132   3  165–<178  132–<145   4  178–<195  145–<160   5  ≥195  ≥160  Menopausal females (%)  Not applicable  72  Current estrogen users among menopausal females (%)  Not applicable  30  Vegetarian status (%)       Eats meat <1 time/week  48  56  Nut intake (%)       Never  5  6   ≤2 times/week  53  52   ≥3 times/week  42  42  Type of bread (%)       Whole wheat only  64  62   Mixed  23  29   White only  13  9  Alcohol intake (%)       None  84  93  Selected characteristic  Men (n = 8,280)  Women (n = 12,017)  Age (mean years (SD*))  57.6 (12.5)  59.2 (12.9)  Hypertensive (%)†  17  24  Smoking (%)       Never smoked  61  86   Smoked in the past  33  12   Currently smokes  6  2  Body mass index (tertiles)‡       Low  <23.4  <21.8   Medium  23.4–25.9  21.8–25.2   High  >25.9  >25.2  Education level (%)       High school  34  39   Some college  30  43   College graduate  37  19  Exercise level (%)§       None to low  31  39   Moderate  23  20   High  46  41  Energy intake (kcal quintiles)       1  <2,100  <1,588   2  2,100–<2,268  1,588–<1,743   3  2,268–<2,401  1,743–<1,876   4  2,401–<2,562  1,876–<2,028   5  ≥2,562  ≥2,028  Weight (lb¶ quintiles)       1  <150  <120   2  150–<165  120–<132   3  165–<178  132–<145   4  178–<195  145–<160   5  ≥195  ≥160  Menopausal females (%)  Not applicable  72  Current estrogen users among menopausal females (%)  Not applicable  30  Vegetarian status (%)       Eats meat <1 time/week  48  56  Nut intake (%)       Never  5  6   ≤2 times/week  53  52   ≥3 times/week  42  42  Type of bread (%)       Whole wheat only  64  62   Mixed  23  29   White only  13  9  Alcohol intake (%)       None  84  93  * SD, standard deviation. † Self-reported, physician diagnosed. ‡ Body mass index = self-reported weight (kg)/height (m)2. § Occupational and leisure exercise index. ¶ One pound = 0.454 kg. View Large TABLE 2. Daily fluid intake in 8-oz (240 ml) servings per day of the US population aged 20 years or more in 1977–1978* and of study participants aged 38 years or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study Fluids  Study participants   US population*   Men (n = 8,280)   Women (n = 12,017)   Men + women (n = 14,273)   Mean (SD†)  Range  % missing  Mean (SD)  Range  % missing  Mean (SD)  99th percentile  Water  4.7 (2.2)  0–>6  1.7  4.4 (2.1)  0–>6  1.8  2.9 (2.3)  10.6  Fluids other than water (nonalcoholic)  3.9 (2.2)  0–19    3.6 (2.1)  0–16.1    3.8    Coffee  0.7 (1.5)  0–7  4.4  0.4 (1.0)  0–7  7.0  1.6 (1.9)  7.9  Hot chocolate  0.1 (0.2)  0–7  6.3  0.1 (0.2)  0–4.5  8.6  Included with milk  Black tea  0.1 (0.5)  0–7  6.6  0.1 (0.4)  0–7  9.0  0.6 (1.1)  4.5  Other hot drinks‡  0.4 (0.8)  0–7  4.7  0.6 (1.0)  0–7  5.9  0.0 (0.1)  0.2  Milk  1.7 (1.3)  0–7  0.9  1.7 (1.3)  0–7  1.3  0.7 (0.8)  4.4  Juice, unsweetened  0.5 (0.5)  0–2  3.7  0.5 (0.5)  0–2  4.7  0.3 (0.3)§  2.1§  Juice, sweetened  0.1 (0.2)  0–2  5.3  0.07 (0.2)  0–2  8.2  Fruit drinks  0.1 (0.3)  0–2  5.4  0.1 (0.3)  0–2  7.2  0.1 (0.4)  1.6  Carbonated soft drinks (noncaffeinated)¶  0.1 (0.2)  0–2  3.3  0.1 (0.2)  0–2  4.7  0.4 (0.8) (regular)#  3.4  Carbonated soft drinks (caffeinated)¶  0.1 (0.3)  0–2  3.3  0.1 (0.3)  0–2  5.4  0.1 (0.5) (low calorie)#  2.2  Alcoholic beverages  0.08 (0.34)  0–4  3.1  0.02 (0.14)  0–4  4.3  0.3 (0.9)  5.1  Fluids  Study participants   US population*   Men (n = 8,280)   Women (n = 12,017)   Men + women (n = 14,273)   Mean (SD†)  Range  % missing  Mean (SD)  Range  % missing  Mean (SD)  99th percentile  Water  4.7 (2.2)  0–>6  1.7  4.4 (2.1)  0–>6  1.8  2.9 (2.3)  10.6  Fluids other than water (nonalcoholic)  3.9 (2.2)  0–19    3.6 (2.1)  0–16.1    3.8    Coffee  0.7 (1.5)  0–7  4.4  0.4 (1.0)  0–7  7.0  1.6 (1.9)  7.9  Hot chocolate  0.1 (0.2)  0–7  6.3  0.1 (0.2)  0–4.5  8.6  Included with milk  Black tea  0.1 (0.5)  0–7  6.6  0.1 (0.4)  0–7  9.0  0.6 (1.1)  4.5  Other hot drinks‡  0.4 (0.8)  0–7  4.7  0.6 (1.0)  0–7  5.9  0.0 (0.1)  0.2  Milk  1.7 (1.3)  0–7  0.9  1.7 (1.3)  0–7  1.3  0.7 (0.8)  4.4  Juice, unsweetened  0.5 (0.5)  0–2  3.7  0.5 (0.5)  0–2  4.7  0.3 (0.3)§  2.1§  Juice, sweetened  0.1 (0.2)  0–2  5.3  0.07 (0.2)  0–2  8.2  Fruit drinks  0.1 (0.3)  0–2  5.4  0.1 (0.3)  0–2  7.2  0.1 (0.4)  1.6  Carbonated soft drinks (noncaffeinated)¶  0.1 (0.2)  0–2  3.3  0.1 (0.2)  0–2  4.7  0.4 (0.8) (regular)#  3.4  Carbonated soft drinks (caffeinated)¶  0.1 (0.3)  0–2  3.3  0.1 (0.3)  0–2  5.4  0.1 (0.5) (low calorie)#  2.2  Alcoholic beverages  0.08 (0.34)  0–4  3.1  0.02 (0.14)  0–4  4.3  0.3 (0.9)  5.1  * 1977–1978 Nationwide Food Consumption Survey of the United States (A. G. Ershow and K. P. Cantor. Tables 62 and 63. In: Total water and tap water intake in the United States: population-based estimates of quantities and sources. Bethesda, MD: National Cancer Institute, 1989). † SD, standard deviation. ‡ Including postum, herbal tea (except black tea), decaffeinated coffee, etc. § The Adventist Health Study differentiated between “juice, unsweetened” and “juice, sweetened,” but the US population study did not. ¶ Regular + low calorie. # Caffeinated + noncaffeinated. View Large Water intake and risk of coronary heart disease The clearest and most consistent association with fatal coronary heart disease was found with water intake (table 3). Among men, univariate analysis showed a dose-response relation (p < 0.001). Compared with those drinking two or fewer glasses of water daily (low), subjects drinking from three to four glasses (medium) and five or more glasses (high) had relative risks of 0.65 (95 percent confidence interval (CI): 0.40, 1.05) and 0.46 (95 percent CI: 0.28, 0.75), respectively. Among women, the relative risks of drinking medium and high levels of water were 0.54 (95 percent CI: 0.32, 0.90) and 0.59 (95 percent CI: 0.36, 0.97), respectively. The associations remained virtually unchanged when adjusting for traditional risk factors as well as for fluids other than water, energy intake, diet, exercise, and when weight replaced body mass index in model 2. Therefore, we have included the results for only models 1–4. TABLE 3. Relative risk of fatal coronary heart disease according to water intake among study participants 38 years of age or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study   8-oz (240 ml) glasses of water/day  Model 1*,†   Model 2*,‡   Model 3*,§   Model 4*,¶   RR#  95% CI#  RR  95% CI  RR  95% CI  RR  95% CI  Men (n = 8,280)  ≤2  1.00    1.00    1.00    1.00      3–4  0.65  0.40, 1.05  0.60  0.36, 0.99  0.64  0.36, 1.16  0.67  0.40, 1.13    ≥5  0.46  0.28, 0.75  0.38  0.23, 0.64  0.33  0.17, 0.62  0.39  0.22, 0.67      Trend    Trend    Trend    Trend        p = 0.001    p = 0.0002    p = 0.0003    p = 0.0003    Women (n = 12,017)  ≤2  1.00    1.00    1.00    1.00      3–4  0.54  0.32, 0.90  0.57  0.32, 1.00  0.44  0.21, 0.91  0.41  0.20, 0.85    ≥5  0.59  0.36, 0.97  0.61  0.35, 1.06  0.57  0.29, 1.11  0.52  0.27, 1.03      Trend    Trend    Trend    Trend        p = 0.08    p = 0.1    p = 0.17    p = 0.17      8-oz (240 ml) glasses of water/day  Model 1*,†   Model 2*,‡   Model 3*,§   Model 4*,¶   RR#  95% CI#  RR  95% CI  RR  95% CI  RR  95% CI  Men (n = 8,280)  ≤2  1.00    1.00    1.00    1.00      3–4  0.65  0.40, 1.05  0.60  0.36, 0.99  0.64  0.36, 1.16  0.67  0.40, 1.13    ≥5  0.46  0.28, 0.75  0.38  0.23, 0.64  0.33  0.17, 0.62  0.39  0.22, 0.67      Trend    Trend    Trend    Trend        p = 0.001    p = 0.0002    p = 0.0003    p = 0.0003    Women (n = 12,017)  ≤2  1.00    1.00    1.00    1.00      3–4  0.54  0.32, 0.90  0.57  0.32, 1.00  0.44  0.21, 0.91  0.41  0.20, 0.85    ≥5  0.59  0.36, 0.97  0.61  0.35, 1.06  0.57  0.29, 1.11  0.52  0.27, 1.03      Trend    Trend    Trend    Trend        p = 0.08    p = 0.1    p = 0.17    p = 0.17    * Excluding subjects with unknown values for model variables. † Adjusted for age. ‡ Adjusted for traditional coronary heart disease risk factors age, smoking, hypertension, body mass index, and education level for men, plus estrogen replacement therapy in menopausal women. § Adjusted for traditional coronary heart disease risk factors plus intake of fluids other than water. ¶ Adjusted for energy in addition to traditional coronary heart disease risk factors. # RR, relative risk; CI, confidence interval. View Large The negative association between risk of fatal coronary heart disease and water intake was seen consistently when further tested within subgroups defined by 11 coronary heart disease risk factors. The relative risks then ranged from 0.34 to 0.73 for the highest level of water intake (table 4). TABLE 4. Relative risk of fatal coronary heart disease in dichotomized groupings of 10 coronary heart disease risk factors, among study men and women 38 years of age or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study Risk factor for coronary heart disease  High vs. low water intake   Risk factor for coronary heart disease  High vs. low water intake   RR*,†  95% CI*  RR†  95% CI  Gender‡      Exercise       Women  0.60  0.36, 0.98   High  0.60  0.34, 1.07   Men  0.46  0.28, 0.75   Low  0.55  0.35, 0.87  Age, years§      Meat consumption       <80  0.68  0.43, 1.06   <1/week  0.46  0.28, 0.75   ≥80  0.59  0.34, 1.02   ≥1/week  0.54  0.34, 0.96  Smoker      Bread       Never  0.45  0.30, 0.69   Whole wheat  0.38  0.25, 0.60   Ever  0.61  0.33, 1.16   Not whole wheat only  0.73  0.42, 1.28  Blood pressure      Nut intake       Normotensive  0.52  0.33, 0.81   >1, 2/week  0.45  0.19, 1.04   Hypertensive  0.50  0.30, 0.85   ≤1, 2/week  0.54  0.36, 0.80  Body mass index¶      Intake of fluids other than water       ≤23.7  0.34  0.14, 0.83   <3 times/day  0.60  0.24, 1.52   >23.7  0.52  0.32, 0.85   ≥3 times/day  0.38  0.23, 0.62  Education             >High school  0.56  0.31, 1.02         ≤High school  0.50  0.33, 0.77        Risk factor for coronary heart disease  High vs. low water intake   Risk factor for coronary heart disease  High vs. low water intake   RR*,†  95% CI*  RR†  95% CI  Gender‡      Exercise       Women  0.60  0.36, 0.98   High  0.60  0.34, 1.07   Men  0.46  0.28, 0.75   Low  0.55  0.35, 0.87  Age, years§      Meat consumption       <80  0.68  0.43, 1.06   <1/week  0.46  0.28, 0.75   ≥80  0.59  0.34, 1.02   ≥1/week  0.54  0.34, 0.96  Smoker      Bread       Never  0.45  0.30, 0.69   Whole wheat  0.38  0.25, 0.60   Ever  0.61  0.33, 1.16   Not whole wheat only  0.73  0.42, 1.28  Blood pressure      Nut intake       Normotensive  0.52  0.33, 0.81   >1, 2/week  0.45  0.19, 1.04   Hypertensive  0.50  0.30, 0.85   ≤1, 2/week  0.54  0.36, 0.80  Body mass index¶      Intake of fluids other than water       ≤23.7  0.34  0.14, 0.83   <3 times/day  0.60  0.24, 1.52   >23.7  0.52  0.32, 0.85   ≥3 times/day  0.38  0.23, 0.62  Education             >High school  0.56  0.31, 1.02         ≤High school  0.50  0.33, 0.77        * RR, relative risk; CI, confidence interval. † Relative risk comparing those with intake of five or more 8-oz (240 ml) glasses of water daily versus two or fewer, excluding unknowns and adjusted for age and sex except ‡ and §. RR = 1.00 for intake of two or fewer glasses of water daily. Results for medium level of water intake were omitted to conserve space. ‡ Adjusted for age only. § Adjusted for sex only. ¶ Body mass index = weight (kg)/height (m)2. View Large Fluids other than water and risk of coronary heart disease In univariate analysis, intake of fluids other than water was associated with increased risk of fatal coronary heart disease (table 5). The association was statistically significant in women who drank five or more compared with two or fewer servings daily, with a relative risk of 2.47 (95 percent CI: 1.04, 5.88). When adjusting for traditional risk factors, as well as for water and caloric intake, exercise, diet, and replacing body mass index with weight, the point estimates remained virtually unchanged and significance was retained. However, the confidence intervals around these estimates were quite wide. TABLE 5. Relative risk of fatal coronary heart disease according to intake of “fluids other than water” among study participants 38 years of age or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study   8-oz (240 ml) glasses of “fluids other than water”/ day  Model 1*,†   Model 2*,‡   Model 3*,§   Model 4*,¶     RR#  95% CI#  RR  95% CI  RR  95% CI  RR  95% CI  Men (n = 8,280)  ≤2  1.00    1.00    1.00    1.00      >2–<5  1.44  0.75, 2.76  1.34  0.75, 2.76  1.29  0.65, 2.52  1.46  0.73, 2.92    ≥5  1.46  0.70, 3.03  1.07  0.48, 2.35  1.00  0.45, 2.22  1.34  0.59, 3.04      Trend    Trend    Trend    Trend        p = 0.4    p = 0.9    p = 0.9    p = 0.7    Women (n = 12,017)  ≤2  1.00    1.00    1.00    1.00      >2–<5  2.02  0.90, 4.50  2.14  0.83, 5.50  2.01  0.78, 5.18  2.25  0.87, 5.80    ≥5  2.47  1.04, 5.88  2.98  1.10, 8.07  2.79  1.03, 7.62  3.32  1.18, 9.30      Trend    Trend    Trend    Trend        p = 0.06    p = 0.03    p = 0.04    p = 0.02      8-oz (240 ml) glasses of “fluids other than water”/ day  Model 1*,†   Model 2*,‡   Model 3*,§   Model 4*,¶     RR#  95% CI#  RR  95% CI  RR  95% CI  RR  95% CI  Men (n = 8,280)  ≤2  1.00    1.00    1.00    1.00      >2–<5  1.44  0.75, 2.76  1.34  0.75, 2.76  1.29  0.65, 2.52  1.46  0.73, 2.92    ≥5  1.46  0.70, 3.03  1.07  0.48, 2.35  1.00  0.45, 2.22  1.34  0.59, 3.04      Trend    Trend    Trend    Trend        p = 0.4    p = 0.9    p = 0.9    p = 0.7    Women (n = 12,017)  ≤2  1.00    1.00    1.00    1.00      >2–<5  2.02  0.90, 4.50  2.14  0.83, 5.50  2.01  0.78, 5.18  2.25  0.87, 5.80    ≥5  2.47  1.04, 5.88  2.98  1.10, 8.07  2.79  1.03, 7.62  3.32  1.18, 9.30      Trend    Trend    Trend    Trend        p = 0.06    p = 0.03    p = 0.04    p = 0.02    * Excluding subjects with unknown values for model variables. † Adjusted for age. ‡ Adjusted for traditional coronary heart disease risk factors age, smoking, hypertension, body mass index, and education level for men, plus estrogen replacement therapy in menopausal women. § Adjusted for intake of water in addition to traditional coronary heart disease risk factors. ¶ Adjusted for energy in addition to traditional coronary heart disease risk factors. # RR, relative risk; CI, confidence interval. View Large We were unable to determine the degree to which juices or sugared drinks (including soda) might individually contribute to the increased risk of fatal coronary heart disease in women, because too few of this study population consumed any of these beverages individually more than once a day. After adjustment for water intake, the association with intake of milk (omitting soy milk) was close to the null and with caffeinated beverages (coffee, tea, and caffeinated sodas), positive but not statistically significant. DISCUSSION We have found a strong negative multivariate association between intake of water and risk of fatal coronary heart disease and, in contrast, a positive association between intake of fluids other than water and risk. These results were consistent between the sexes, although the negative association with water was stronger in men and the positive association with fluids other than water was stronger in women. Moreover, formal tests of the differences between the opposite-signed linear trends of water and fluids other than water were significant (p < 0.01) in both sexes independently. Our study has strengths and limitations. As an observational study, potential confounding always needs to be considered. Could fluid consumption be a marker of other risk factors not included in the model? Few candidate confounding variables are obvious. Those subjects who drink more water may be more health conscious in other less clearly defined ways. Drinking more water may be a marker of higher physical activity or those with higher energy intake. However, a two-tailed Pearson's correlation test showed no correlation between water and energy intakes (r = 0.08, p < 0.01 and r = 0.06, p < 0.01 for males and females, respectively) and, when exercise and energy consumption are included in the multivariate model, the effect of water consumption remained. Diabetics, who may drink more fluids, are excluded from these analyses. That the negative association with water intake was relatively consistent across many subgroups of the population makes confounding less likely as an explanation for this effect. The study population is large and relatively well educated, includes men and women with a broad range of ages, and provides detailed baseline information on the intake of fluids of different kinds. Data were missing for males and females, respectively, on between 1.7 and 6.6 percent and on between 1.3 and 9.0 percent of the various items included in the measurements of fluid intake. If the associations between fluid intake and coronary heart disease were markedly different for those who did not provide all the relevant data, then associations in the remainder may differ from those in the whole population. However, we have no reason to suspect this. The results can strictly be applied only to the California Adventist population. Moreover, because they drink more water and less caffeinated and alcoholic beverages than the US averages, it could be that, unlike some other findings in this population, this specific finding may apply only to this population alone. However, California Adventists probably do not differ biochemically or physiologically from others. If these results describe causal associations, they will probably also be found in the general population. The traditional risk factors for coronary heart disease operate with their usual force in California Adventists (31). Associations between coronary heart disease and certain foods (nuts and meats), which were first described in this population, have now been confirmed in several other populations (32–34). An important question is whether there are known mechanisms that might underlie these results if the associations are causal. There is a small literature strongly suggesting that several factors related to blood viscosity are adversely affected by hypohydration (16–21, 35–39) and that these rheologic factors predict the risk of coronary heart disease (4, 6, 8, 9). Whole blood viscosity is determined mainly by hematocrit and fibrinogen levels, both of which have frequently been associated with coronary heart disease events (1–9). It is interesting that, although percentage differences in blood viscosity-associated factors between those who do and do not experience coronary heart disease events are often statistically significant, they are quite small, often in the range of 2–7 percent (2, 6, 9, 18, 19, 21, 40, 41). Yet these are the kind of changes seen in mild, nonclinical hypohydration (16–20), within the presumed normal range. There is a well-described circadian rhythm of hemorheologic factors that can be quite marked (18, 41), perhaps due to continued non-reabsorbed glomerular filtration at night in the absence of fluid intake (19, 42). One can speculate that relatively elevated blood viscosity in the morning could be one factor accounting for the well-known increased risk of coronary heart disease events at this time of the day (43). Why may coronary heart disease risk be increased with a higher intake of fluids other than water? Several mechanisms can be postulated. Caffeinated beverages are mild diuretics and thus may raise blood viscosity (44). High energy drinks such as juices and regular sodas have osmolalities between 556 and 836 mOsm/kg (45). Their consumption causes a net movement of fluid from the vascular system into the intestinal lumen, resulting in a rapid elevation in blood viscosity after consumption (24). Perfusing the duodenum with a glucose solution that has an osmolality of 456 mOsm/kg, which is lower than that of juices and regular soda, was observed to reduce plasma volume by 3.3 percent within 105 minutes (46). Further, serum triglyceride levels may be raised by drinks containing high concentrations of sugar, including fruit juices (47–50). In subjects drinking 2.7 glasses (640 ml) of grape juice and three glasses (750 ml) of orange juice daily, triglyceride levels were seen to increase by 50 percent (51) and 30 percent (52), respectively. There is growing evidence that postprandial and fasting elevations of plasma triglycerides are independent coronary heart disease risk factors (53–58), even within the high normal range (59). Elevated triglyceride levels have been positively associated with coagulation factor VII (60), plasminogen activator inhibitor-1 (57), and thrombotic factors X (60) and IX (57) but inversely associated with antithrombin III (58) and plasma fibrinolytic activity (57). We hypothesize that higher intake of diuretic and high energy beverages results in more frequent and larger exposure to conditions that increase the risk of thrombosis and atherosclerosis. The magnitudes of the apparent effects associated with greater intake of both types of fluid appeared to differ between the two sexes. Although these differences could easily be explained by chance, it may be relevant that women typically have lower blood viscosity than do men because of lower hematocrit levels, erythrocyte rigidity, and aggregability at all shear rates (7). They may also experience less reduction in plasma volume with dehydration (61) and regulate their body temperatures at lower sweat rates (62). Reduced water intake in women, therefore, may not produce high levels of blood viscosity as readily as in men. On the other hand, because men have larger blood volumes than do women, the same quantities of hyperosmolar fluids or caffeinated beverages (fluids other than water) would produce a lower proportionate decrease in intravascular volume and subsequently a smaller increase in blood viscosity. In summary, we report intriguing associations between the intake of fluids and the risk of coronary heart disease that are not obviously explained by confounding. Further research in other populations, possibly including experimental study designs, is necessary to decide whether the associations are causal. Correspondence to Dr. Jacqueline Chan, Adventist Health Studies 2, School of Public Health, Loma Linda University, 24785 Stewart St., Room 203, Loma Linda, CA 92350 (e-mail: jchan@sph.llu.edu). Supported in part by National Institutes of Health grant 5-R01-HL-26210. REFERENCES 1. Becker RC. The role of blood viscosity in the development and progression of coronary artery disease. Cleve Clin J Med  1993; 60: 353–8. Google Scholar 2. Lowe GD, Lee AJ, Rumley A, et al. Blood viscosity and risk of cardiovascular events: the Edinburgh Artery Study. Br J Haematol  1997; 96: 168–73. Google Scholar 3. Lee AJ, Mowbray PI, Lowe GD, et al. Blood viscosity and elevated carotid intima-media thickness in men and women. Circulation  1998; 97: 1467–73. Google Scholar 4. Koenig W, Ernst E. The possible role of hemorheology in atherothrombogenesis. Atherosclerosis  1992; 94: 93–107. Google Scholar 5. Becker RC. Seminars in thrombosis, thrombolysis, and vascular biology. Part 5. Cellular-rheology and plasma viscosity. Biorheology  1991; 79: 265–70. Google Scholar 6. Ernst E. Hematocrit and cardiovascular risk. J Intern Med  1995; 237: 527–8. Google Scholar 7. De Simone G, Devereux RB, Shu C, et al. Relation of blood viscosity to demographic and physiologic variables and to cardiovascular risk factors in apparently normal adults. Circulation  1990; 81: 107–17. Google Scholar 8. Erikssen G, Thaulow E, Sandvik L, et al. Haematocrit: a predictor of cardiovascular mortality? J Intern Med  1993; 234: 493–9. Google Scholar 9. Lowe GD, Drummond MM, Lorimer AR, et al. Relation between extent of coronary artery disease and blood viscosity. Br Med J  1980; 280: 673–4. Google Scholar 10. Resch KL, Ernst E, Matrai A, et al. Can rheologic variables be of prognostic relevance in arteriosclerotic diseases? Angiology  1991; 42: 963–70. Google Scholar 11. Smith WC, Lowe GD, Lee AJ, et al. Rheological determinants of blood pressure in a Scottish adult population. J Hypertens  1992; 10: 467–72. Google Scholar 12. Kannel WB, McGee DL. Update on some epidemiologic features of intermittent claudication: the Framingham Study. J Am Geriatr Soc  1985; 33: 13–18. Google Scholar 13. Julius S, Palatini P, Nesbitt SD. Tachycardia: an important determinant of coronary risk in hypertension. J Hypertens Suppl  1998; 16: S9–15. Google Scholar 14. Harrison MJ, Pollock S, Kendall BE, et al. Effect of hematocrit on carotid stenosis and cerebral infarction. Lancet  1981; 2: 114–15. Google Scholar 15. Finch CA, Lenfant C. Oxygen transport in man. N Engl J Med  1972; 286: 407–15. Google Scholar 16. Dvilansky A, Bar-Am J, Nathan I, et al. Hematologic values in healthy older people in the Negev area. Isr J Med Sci  1979; 15: 821–5. Google Scholar 17. Kristal-Boneh E, Glusman JG, Chaemovitz C, et al. Improved thermoregulation caused by forced water intake in desert dwellers. Eur J Appl Physiol  1988; 57: 220–4. Google Scholar 18. Seaman GV, Engel R, Swank RL, et al. Circadian periodicity in some physicochemical parameters of circulating blood. Nature  1965; 207: 833–5. Google Scholar 19. Kurabayashi H, Kubota K, Tamura J, et al. A glass of water at midnight for possible prevention of cerebral infarction. Stroke  1991; 22: 1326–7. Google Scholar 20. Vandewalle H, Lacombe C, Lelièvre JC, et al. Blood viscosity after a 1-hour submaximal exercise with and without drinking. Int J Sports Med  1988; 9: 104–7. Google Scholar 21. Yasaka M, Yamaguchi T, Oitia J, et al. Clinical features of recurrent embolization in acute cardioembolic stroke. Strong disposing factors: low plasma levels of antithrombin II. Stroke  1993; 24: 1681–5. Google Scholar 22. Toiler GH, Brezinski D, Schafer AI, et al. Concurrent morning increase in platelet aggregability and the risk of myocardial infarction and sudden cardiac death. N Engl J Med  1987; 316: 1514–18. Google Scholar 23. Blanchard J, Sawers SJ. Relationship between urine flow rate and renal clearance of caffeine in man. J Clin Pharmacol  1983; 23: 134–8. Google Scholar 24. Maughan RJ, Leiper JB. Limitations to fluid replacement during exercise. Can J Appl Physiol  1999; 24: 173–87. Google Scholar 25. Beeson WL, Mills PK, Phillips RL, et al. Chronic disease among Seventh-day Adventists, a low risk group. Rationale, methodology, and description of the population. Cancer  1989; 64: 570–81. Google Scholar 26. Fraser GE, Sabaté J, Beeson WL, et al. A possible protective effect of nut consumption on risk of coronary heart disease. Arch Intern Med  1992; 152: 1416–24. Google Scholar 27. Cox DR, Oakes D. Chapter 7. In: Analysis of survival data. New York, NY: Chapman and Hall, 1984. Google Scholar 28. Fraser GE. Diet and coronary heart disease: beyond dietary fats and low density lipoprotein cholesterol. Am J Clin Nutr  1994; 59(suppl): 1117S–23S. Google Scholar 29. Snowdon DA, Phillips RL, Fraser GE. Meat consumption and fatal ischemic heart disease. Prev Med  1984; 13: 490–500. Google Scholar 30. Ershow AG, Cantor KP. Total water and tap water intake in the United States: population-based estimates of quantities and sources. Prepared under National Cancer Institute order #263-MD-810264 with the Life Sciences Research Office. Bethesda, MD: Federation of American Societies for Experimental Biology, 1989. Google Scholar 31. Fraser GE, Strahan M, Sabate J, et al. Effects of traditional coronary risk factors on rates of incident coronary events in a low risk population. The Adventist Health Study. Circulation  1992; 86: 406–13. Google Scholar 32. Fraser GE. Nut consumption, lipids, and risk of a coronary event. Clin Cardiol  1999; 22(suppl): III-11–15. Google Scholar 33. Sabaté J. Nut consumption, vegetarian diets, ischemic heart disease risk, and all-cause mortality: evidence from epidemiologic studies. Am J Clin Nutr  1999; 70(suppl): 500S–3S. Google Scholar 34. Key TJ, Fraser GE, Thorogood M, et al. Mortality in vegetarians and nonvegetarians: detailed findings from a collaborative analysis of 5 prospective studies. Am J Clin Nutr  1999; 70(suppl): 516S–24S. Google Scholar 35. Rolls BJ, Wood RJ, Rolls ET, et al. Thirst following water deprivation in humans. Am J Physiol  1980; 239: R476–82. Google Scholar 36. Dintenfass L. The cause of death: blood hyperviscosity? (Letter). J R Soc Med  1987; 80: 536–7. Google Scholar 37. Pearson TC. Apparent polycythaemia. Blood Rev  1991; 5: 205–13. Google Scholar 38. Beijering RJ, Gips CH, Huizenga JR, et al. Whole blood and plasma water in health and disease: longitudinal and transverse observations and correlations with several different hematological and clinicochemical parameters. Clin Chim Acta  1997; 258: 59–68. Google Scholar 39. Warren JL, Bacon WE, Harris T, et al. Burden and outcomes associated with dehydration among U.S. elderly, 1991. Am J Public Health  1994; 84: 1265–9. Google Scholar 40. Koenig W, Sund M, Filipiak B, et al. Plasma viscosity and the risk of coronary heart disease: results from the MONICA-Augsburg Cohort Study, 1984 to 1992. Arterioscler Thromb Vasc Biol  1998; 18: 768–72. Google Scholar 41. Kubota K, Sakurae T, Tamura J, et al. Is the change in hematocrit and blood viscosity a factor triggering cerebral and myocardial infarction? Stroke  1987; 18: 812–13. Google Scholar 42. Wood JH. Is the change in hematocrit and blood viscosity a factor triggering cerebral and myocardial infarction? (Letter). Stroke  1987; 18: 813. Google Scholar 43. Cohen MC, Muller JE. Onset of acute myocardial infarction—circadian variation and triggers. Cardiovasc Res  1992; 26: 831–8. Google Scholar 44. Neuhauser-Berthold, Beine S, Verwied SC, et al. Coffee consumption and total body water homeostasis as measured by fluid balance and bioelectrical impedance analysis. Ann Nutr Metab  1997; 41: 29–36. Google Scholar 45. Bell SJ, Anderson FL, Bistrian BR, et al. Osmolality of beverages commonly provided on clear and full liquid menu. Nutr Clin Pract  1987; 2: 241–4. Google Scholar 46. Shi X, Summers RW, Schedl HP, et al. Effects of carbohydrate type and concentration and solution osmolality on water absorption. Med Sci Sports Exerc  1995; 27: 1607–15. Google Scholar 47. Kuo PT, Bassett DR. Dietary sugar in the production of hypertriglyceridemia. Ann Intern Med  1965; 62: 1199–212. Google Scholar 48. Parks EJ, Hellerstein MK. Carbohydrate-induced hypertriacylglycerolemia: historical perspective and review of biological mechanisms. Am J Clin Nutr  2000; 71: 412–33. Google Scholar 49. Hudgins LC, Hellerstein MK, Seidman CE, et al. Relationship between carbohydrate-induced hypertriglyceridemia and fatty acid synthesis in lean and obese subjects. J Lipid Res  2000; 41: 595–604. Google Scholar 50. Parks EJ, Krauss RM, Christiansen MP, et al. Effects of a low-fat, high-carbohydrate diet on VLDL-triglyceride assembly, production, and clearance. J Clin Invest  1999; 104: 1087–96. Google Scholar 51. Stein JH, Keevil JG, Wiebe DA, et al. Purple grape juice improves endothelial function and reduces the susceptibility of LDL cholesterol to oxidation in patients with coronary artery disease. Circulation  1999; 100: 1050–5. Google Scholar 52. Kurowska EM, Spence JD, Jordan J, et al. HDL-cholesterol-raising effect of orange juice in subjects with hypercholesterolemia. Am J Clin Nutr  2000; 72: 1095–100. Google Scholar 53. Cullen P. Evidence that triglycerides are an independent coronary heart disease risk factor. Am J Cardiol  2000; 86: 943–9. Google Scholar 54. Gaziano JM. Triglycerides and coronary risk. Curr Cardiol Rep  1999; 1: 125–30. Google Scholar 55. Patsch JR, Miesenbock G, Hopferwieser T, et al. Relation of triglyceride metabolism and coronary artery disease. Studies in the postprandial state. Arterioscler Thromb  1992; 12: 1336–45. Google Scholar 56. Austin MA. Epidemiology of hypertriglyceridemia and cardiovascular disease. Am J Cardiol  1999; 83: 13F–16F. Google Scholar 57. Miller GJ. Lipoproteins and the haemostatic system in atherothrombotic disorders. Baillieres Clin Haematol  1994; 7: 713–32. Google Scholar 58. Zajtchuk R, Zajtchuk J. Relationship of triglyceride levels to thrombosis in patients with coronary artery disease. Ann Thorac Surg  1983; 35: 274–6. Google Scholar 59. Kreisberg RA. Hypertriglyceridemia and coronary heart disease. Clin Rev  2000;spring: 29–32. Google Scholar 60. de Sousa JC, Soria C, Ayrault-Jarrier M, et al. Association between coagulation factors VII and X with triglyceride rich lipoproteins. J Clin Pathol  1988; 41: 940–4. Google Scholar 61. Byrd R, Stewart L, Torranin C, et al. Sex differences in response to hypohydration. J Sports Med Phys Fitness  1977; 17: 65–8. Google Scholar 62. Rocker L, Kirsch KA, Stoboy H, et al. Influence of heat stress on plasma volume and intravascular proteins in sedentary females. Eur J Appl Physiol  1977; 36: 187–92. Google Scholar http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png American Journal of Epidemiology Oxford University Press

Water, Other Fluids, and Fatal Coronary Heart Disease: The Adventist Health Study

Download PDF
9 pages

Loading next page...
 
/lp/oxford-university-press/water-other-fluids-and-fatal-coronary-heart-disease-the-adventist-xLfuP50O1D

References (67)

Publisher
Oxford University Press
ISSN
0002-9262
eISSN
1476-6256
DOI
10.1093/aje/155.9.827
Publisher site
See Article on Publisher Site

Abstract

Abstract Whole blood viscosity, plasma viscosity, hematocrit, and fibrinogen are considered independent risk factors for coronary heart disease and can be elevated by dehydration. The associations between fatal coronary heart disease and intake of water and fluids other than water were examined among the 8,280 male and 12,017 female participants aged 38–100 years who were without heart disease, stroke, or diabetes at baseline in 1976 in the Adventist Health Study, a prospective cohort study. A total of 246 fatal coronary heart disease events occurred during the 6-year follow-up. High daily intakes of water (five or more glasses) compared with low (two or fewer glasses) were associated with a relative risk in men of 0.46 (95% confidence interval (CI): 0.28, 0.75; p trend = 0.001) and, in women, of 0.59 (95% CI: 0.36, 0.97). A high versus low intake of fluids other than water was associated with a relative risk of 2.47 (95% CI: 1.04, 5.88) in women and of 1.46 (95% CI: 0.7, 3.03) in men. All associations remained virtually unchanged in multivariate analysis adjusting for age, smoking, hypertension, body mass index, education, and (in women only) hormone replacement therapy. Fluid intake as a putative coronary heart disease risk factor may deserve further consideration in other populations or using other study designs. blood viscosity, coronary disease, dehydration, fluids and secretions, hemorheology, men, water, women CI, confidence interval Whole blood viscosity (1–3), plasma viscosity (4), fi-brinogen (1–5), and hematocrit (2–6) levels are positively correlated with coronary heart disease. Even in the high “normal” range, they have been considered independent risk factors (2, 4–7). In some reports, the magnitudes of their effects are comparable with those of smoking, diastolic blood pressure, and low density serum cholesterol (2, 4, 6). These hemorheologic factors are found to be elevated years before manifestation of acute ischemic events (4) and are implicated in the early stages (6), acceleration (8), and extent (9) of atherosclerosis, as well as prognosis in patients with arteriosclerotic disease (5, 10). They are also associated with hypertension (11) and intermittent claudication (12). Additionally, a higher hematocrit level has been associated with tachycardia (13), infarct size (14), reduced oxygen transport (15), and reduced myocardial perfusion (5). Elevation of hemorheologic factors can result from chronic hypohydration (16, 17). They also respond acutely to circadian changes in hydration (18, 19), normal daily activities (20), and use of medications such as diuretics (21). Moreover, an increased risk of coronary heart disease has been associated with circadian increases in hematocrit and fibrinogen (22). This may result in part from an increased risk of coronary thrombosis (4, 21). Because of these reported relations, the authors prospectively examined the associations between fluid intake and risk of fatal coronary heart disease, using data from the Adventist Health Study, a large cohort study. The effects of plain water and fluids other than water were examined separately because both composition and volume may affect a fluid's impact on acute and chronic hydration (23, 24). MATERIALS AND METHODS Study design and participants The Adventist Health Study is a cohort study designed to test the effects of lifestyle on the risk of coronary heart disease, cancer, and all-cause mortality. Details have been described elsewhere (25). Briefly, in 1976, persons aged 25 years or more living in California Seventh-day Adventist households were sent a detailed lifestyle questionnaire that was completed by 34,192 White, non-Hispanic persons (75 percent response rate). Analyses are confined to the 8,280 males and 12,017 females who reported no physician-diagnosed heart disease, stroke, or diabetes and were 38 years of age or more at baseline (because no fatal coronary heart disease events occurred in younger subjects). The lifestyle questionnaire included information on traditional and dietary risk factors of coronary heart disease and intake of all types of beverages. Intake of water was reported as the number of glasses consumed daily (less than one, from one to two, from three to four, from five to six, and more than six), whereas intake of other fluids was reported as the frequency of intake per day, week, or month. The cohort was followed for 6 years. Follow-up data for coronary heart disease events were complete for 97 percent of the participants (25). The criteria for diagnosing fatal coronary heart disease in this study have been described in detail elsewhere (26). Briefly, fatal coronary heart disease defined as either “definite fatal myocardial infarction” or “other definite fatal coronary heart disease” required death within 30 days of a myocardial infarction confirmed by hospital records, electrocardiogram and cardiac enzymes, or fresh myocardial infarction recorded at autopsy. Deaths were found by computer-assisted linkage with the California death certificate files, the National Death Index, use of church records, contact with relatives, and, when available, autopsy reports. Statistical analysis For analysis, intake of all types of fluids was recoded to monthly intakes. A validity study on a subset of 147 participants of the Adventist Health Study found that the average serving size of fluids other than water was close to one 8-ounce (240 ml) glass (range, 0.94–1.28 glasses) (26). Alcoholic beverages were not included as part of the total fluid intake because they were consumed by less than 11 percent of this special population and then in very small amounts. Participants with a total fluid intake of less than once per day (n = 55) or with a daily milk intake greater than seven times daily (n = 206) were considered outliers and excluded from analysis. However, including them in a sensitivity analysis of the final model did not significantly change the results. The use of three exposure categories allowed for analysis of dose-response relations. For water, these categories were obtained by collapsing the two lowest and the two highest intake levels. Similar categories for intake of fluids other than water were formed using average portion size data. This formed low intake comparison categories that were biologically reasonable with sufficient numbers of events to provide stability in multivariate analysis. The Statistical Package for the Social Sciences (SPSS for Windows: Advanced Statistics Release 7.0, 1995, and 9.0.0, 1998; SPSS, Inc., Chicago, Illinois) was used for all analyses, including gender-specific Cox proportional hazard models (27). Subjects with missing values in any of the variables included in a model were excluded from the analysis. The assumption of proportional hazards was not violated as tested by the log-minus-log survival plots (27). In addition to the age-adjusted model (model 1), eight other models were examined. Model 2 adjusted for age and several conventional coronary heart disease risk factors (smoking, hypertension, educational level, body mass index, and, in women, estrogen replacement therapy). Models 3–8, in addition, adjusted for other variables one at a time: fluid other than that being tested, energy intake, exercise, and dietary factors found to be related to coronary heart disease in the Adventist Health Study (nuts, type of bread, and, in men, meat intake) (26, 28, 29), respectively. This kept the number of variables in a particular model in reasonable proportion to the number of events (27). In model 9, weight replaced body mass index as a variable in model 2. The consistency of the association between water intake and risk of fatal coronary heart disease was further explored in age- and sex-adjusted models applied to subgroups of the population defined by dichotomization of 11 different coronary heart disease risk factors. RESULTS During follow-up, 246 fatal coronary heart disease events (128 in men, 118 in women) were reported. Baseline characteristics of the study population are shown in table 1. The subjects were well educated, 37 percent of men and 19 percent of women being college graduates. Overall, the population adhered to a lifestyle that is commonly held to be healthy, as assessed by their diet and level of exercise. Very few subjects drank alcoholic beverages or were current smokers, and these were usually non-Adventists living in Adventist households (8.7 percent of the total Adventist Health Study population). The mean daily intake of beverages among the study population and the general population is shown in table 2. Water was the fluid consumed in greatest amounts. Compared with the national averages (30), the Adventist Health Study population drank more water, milk, and fruit juices and less coffee, tea, and carbonated and alcoholic beverages. TABLE 1. Selected baseline characteristics of study participants 38 years of age or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study Selected characteristic  Men (n = 8,280)  Women (n = 12,017)  Age (mean years (SD*))  57.6 (12.5)  59.2 (12.9)  Hypertensive (%)†  17  24  Smoking (%)       Never smoked  61  86   Smoked in the past  33  12   Currently smokes  6  2  Body mass index (tertiles)‡       Low  <23.4  <21.8   Medium  23.4–25.9  21.8–25.2   High  >25.9  >25.2  Education level (%)       High school  34  39   Some college  30  43   College graduate  37  19  Exercise level (%)§       None to low  31  39   Moderate  23  20   High  46  41  Energy intake (kcal quintiles)       1  <2,100  <1,588   2  2,100–<2,268  1,588–<1,743   3  2,268–<2,401  1,743–<1,876   4  2,401–<2,562  1,876–<2,028   5  ≥2,562  ≥2,028  Weight (lb¶ quintiles)       1  <150  <120   2  150–<165  120–<132   3  165–<178  132–<145   4  178–<195  145–<160   5  ≥195  ≥160  Menopausal females (%)  Not applicable  72  Current estrogen users among menopausal females (%)  Not applicable  30  Vegetarian status (%)       Eats meat <1 time/week  48  56  Nut intake (%)       Never  5  6   ≤2 times/week  53  52   ≥3 times/week  42  42  Type of bread (%)       Whole wheat only  64  62   Mixed  23  29   White only  13  9  Alcohol intake (%)       None  84  93  Selected characteristic  Men (n = 8,280)  Women (n = 12,017)  Age (mean years (SD*))  57.6 (12.5)  59.2 (12.9)  Hypertensive (%)†  17  24  Smoking (%)       Never smoked  61  86   Smoked in the past  33  12   Currently smokes  6  2  Body mass index (tertiles)‡       Low  <23.4  <21.8   Medium  23.4–25.9  21.8–25.2   High  >25.9  >25.2  Education level (%)       High school  34  39   Some college  30  43   College graduate  37  19  Exercise level (%)§       None to low  31  39   Moderate  23  20   High  46  41  Energy intake (kcal quintiles)       1  <2,100  <1,588   2  2,100–<2,268  1,588–<1,743   3  2,268–<2,401  1,743–<1,876   4  2,401–<2,562  1,876–<2,028   5  ≥2,562  ≥2,028  Weight (lb¶ quintiles)       1  <150  <120   2  150–<165  120–<132   3  165–<178  132–<145   4  178–<195  145–<160   5  ≥195  ≥160  Menopausal females (%)  Not applicable  72  Current estrogen users among menopausal females (%)  Not applicable  30  Vegetarian status (%)       Eats meat <1 time/week  48  56  Nut intake (%)       Never  5  6   ≤2 times/week  53  52   ≥3 times/week  42  42  Type of bread (%)       Whole wheat only  64  62   Mixed  23  29   White only  13  9  Alcohol intake (%)       None  84  93  * SD, standard deviation. † Self-reported, physician diagnosed. ‡ Body mass index = self-reported weight (kg)/height (m)2. § Occupational and leisure exercise index. ¶ One pound = 0.454 kg. View Large TABLE 2. Daily fluid intake in 8-oz (240 ml) servings per day of the US population aged 20 years or more in 1977–1978* and of study participants aged 38 years or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study Fluids  Study participants   US population*   Men (n = 8,280)   Women (n = 12,017)   Men + women (n = 14,273)   Mean (SD†)  Range  % missing  Mean (SD)  Range  % missing  Mean (SD)  99th percentile  Water  4.7 (2.2)  0–>6  1.7  4.4 (2.1)  0–>6  1.8  2.9 (2.3)  10.6  Fluids other than water (nonalcoholic)  3.9 (2.2)  0–19    3.6 (2.1)  0–16.1    3.8    Coffee  0.7 (1.5)  0–7  4.4  0.4 (1.0)  0–7  7.0  1.6 (1.9)  7.9  Hot chocolate  0.1 (0.2)  0–7  6.3  0.1 (0.2)  0–4.5  8.6  Included with milk  Black tea  0.1 (0.5)  0–7  6.6  0.1 (0.4)  0–7  9.0  0.6 (1.1)  4.5  Other hot drinks‡  0.4 (0.8)  0–7  4.7  0.6 (1.0)  0–7  5.9  0.0 (0.1)  0.2  Milk  1.7 (1.3)  0–7  0.9  1.7 (1.3)  0–7  1.3  0.7 (0.8)  4.4  Juice, unsweetened  0.5 (0.5)  0–2  3.7  0.5 (0.5)  0–2  4.7  0.3 (0.3)§  2.1§  Juice, sweetened  0.1 (0.2)  0–2  5.3  0.07 (0.2)  0–2  8.2  Fruit drinks  0.1 (0.3)  0–2  5.4  0.1 (0.3)  0–2  7.2  0.1 (0.4)  1.6  Carbonated soft drinks (noncaffeinated)¶  0.1 (0.2)  0–2  3.3  0.1 (0.2)  0–2  4.7  0.4 (0.8) (regular)#  3.4  Carbonated soft drinks (caffeinated)¶  0.1 (0.3)  0–2  3.3  0.1 (0.3)  0–2  5.4  0.1 (0.5) (low calorie)#  2.2  Alcoholic beverages  0.08 (0.34)  0–4  3.1  0.02 (0.14)  0–4  4.3  0.3 (0.9)  5.1  Fluids  Study participants   US population*   Men (n = 8,280)   Women (n = 12,017)   Men + women (n = 14,273)   Mean (SD†)  Range  % missing  Mean (SD)  Range  % missing  Mean (SD)  99th percentile  Water  4.7 (2.2)  0–>6  1.7  4.4 (2.1)  0–>6  1.8  2.9 (2.3)  10.6  Fluids other than water (nonalcoholic)  3.9 (2.2)  0–19    3.6 (2.1)  0–16.1    3.8    Coffee  0.7 (1.5)  0–7  4.4  0.4 (1.0)  0–7  7.0  1.6 (1.9)  7.9  Hot chocolate  0.1 (0.2)  0–7  6.3  0.1 (0.2)  0–4.5  8.6  Included with milk  Black tea  0.1 (0.5)  0–7  6.6  0.1 (0.4)  0–7  9.0  0.6 (1.1)  4.5  Other hot drinks‡  0.4 (0.8)  0–7  4.7  0.6 (1.0)  0–7  5.9  0.0 (0.1)  0.2  Milk  1.7 (1.3)  0–7  0.9  1.7 (1.3)  0–7  1.3  0.7 (0.8)  4.4  Juice, unsweetened  0.5 (0.5)  0–2  3.7  0.5 (0.5)  0–2  4.7  0.3 (0.3)§  2.1§  Juice, sweetened  0.1 (0.2)  0–2  5.3  0.07 (0.2)  0–2  8.2  Fruit drinks  0.1 (0.3)  0–2  5.4  0.1 (0.3)  0–2  7.2  0.1 (0.4)  1.6  Carbonated soft drinks (noncaffeinated)¶  0.1 (0.2)  0–2  3.3  0.1 (0.2)  0–2  4.7  0.4 (0.8) (regular)#  3.4  Carbonated soft drinks (caffeinated)¶  0.1 (0.3)  0–2  3.3  0.1 (0.3)  0–2  5.4  0.1 (0.5) (low calorie)#  2.2  Alcoholic beverages  0.08 (0.34)  0–4  3.1  0.02 (0.14)  0–4  4.3  0.3 (0.9)  5.1  * 1977–1978 Nationwide Food Consumption Survey of the United States (A. G. Ershow and K. P. Cantor. Tables 62 and 63. In: Total water and tap water intake in the United States: population-based estimates of quantities and sources. Bethesda, MD: National Cancer Institute, 1989). † SD, standard deviation. ‡ Including postum, herbal tea (except black tea), decaffeinated coffee, etc. § The Adventist Health Study differentiated between “juice, unsweetened” and “juice, sweetened,” but the US population study did not. ¶ Regular + low calorie. # Caffeinated + noncaffeinated. View Large Water intake and risk of coronary heart disease The clearest and most consistent association with fatal coronary heart disease was found with water intake (table 3). Among men, univariate analysis showed a dose-response relation (p < 0.001). Compared with those drinking two or fewer glasses of water daily (low), subjects drinking from three to four glasses (medium) and five or more glasses (high) had relative risks of 0.65 (95 percent confidence interval (CI): 0.40, 1.05) and 0.46 (95 percent CI: 0.28, 0.75), respectively. Among women, the relative risks of drinking medium and high levels of water were 0.54 (95 percent CI: 0.32, 0.90) and 0.59 (95 percent CI: 0.36, 0.97), respectively. The associations remained virtually unchanged when adjusting for traditional risk factors as well as for fluids other than water, energy intake, diet, exercise, and when weight replaced body mass index in model 2. Therefore, we have included the results for only models 1–4. TABLE 3. Relative risk of fatal coronary heart disease according to water intake among study participants 38 years of age or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study   8-oz (240 ml) glasses of water/day  Model 1*,†   Model 2*,‡   Model 3*,§   Model 4*,¶   RR#  95% CI#  RR  95% CI  RR  95% CI  RR  95% CI  Men (n = 8,280)  ≤2  1.00    1.00    1.00    1.00      3–4  0.65  0.40, 1.05  0.60  0.36, 0.99  0.64  0.36, 1.16  0.67  0.40, 1.13    ≥5  0.46  0.28, 0.75  0.38  0.23, 0.64  0.33  0.17, 0.62  0.39  0.22, 0.67      Trend    Trend    Trend    Trend        p = 0.001    p = 0.0002    p = 0.0003    p = 0.0003    Women (n = 12,017)  ≤2  1.00    1.00    1.00    1.00      3–4  0.54  0.32, 0.90  0.57  0.32, 1.00  0.44  0.21, 0.91  0.41  0.20, 0.85    ≥5  0.59  0.36, 0.97  0.61  0.35, 1.06  0.57  0.29, 1.11  0.52  0.27, 1.03      Trend    Trend    Trend    Trend        p = 0.08    p = 0.1    p = 0.17    p = 0.17      8-oz (240 ml) glasses of water/day  Model 1*,†   Model 2*,‡   Model 3*,§   Model 4*,¶   RR#  95% CI#  RR  95% CI  RR  95% CI  RR  95% CI  Men (n = 8,280)  ≤2  1.00    1.00    1.00    1.00      3–4  0.65  0.40, 1.05  0.60  0.36, 0.99  0.64  0.36, 1.16  0.67  0.40, 1.13    ≥5  0.46  0.28, 0.75  0.38  0.23, 0.64  0.33  0.17, 0.62  0.39  0.22, 0.67      Trend    Trend    Trend    Trend        p = 0.001    p = 0.0002    p = 0.0003    p = 0.0003    Women (n = 12,017)  ≤2  1.00    1.00    1.00    1.00      3–4  0.54  0.32, 0.90  0.57  0.32, 1.00  0.44  0.21, 0.91  0.41  0.20, 0.85    ≥5  0.59  0.36, 0.97  0.61  0.35, 1.06  0.57  0.29, 1.11  0.52  0.27, 1.03      Trend    Trend    Trend    Trend        p = 0.08    p = 0.1    p = 0.17    p = 0.17    * Excluding subjects with unknown values for model variables. † Adjusted for age. ‡ Adjusted for traditional coronary heart disease risk factors age, smoking, hypertension, body mass index, and education level for men, plus estrogen replacement therapy in menopausal women. § Adjusted for traditional coronary heart disease risk factors plus intake of fluids other than water. ¶ Adjusted for energy in addition to traditional coronary heart disease risk factors. # RR, relative risk; CI, confidence interval. View Large The negative association between risk of fatal coronary heart disease and water intake was seen consistently when further tested within subgroups defined by 11 coronary heart disease risk factors. The relative risks then ranged from 0.34 to 0.73 for the highest level of water intake (table 4). TABLE 4. Relative risk of fatal coronary heart disease in dichotomized groupings of 10 coronary heart disease risk factors, among study men and women 38 years of age or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study Risk factor for coronary heart disease  High vs. low water intake   Risk factor for coronary heart disease  High vs. low water intake   RR*,†  95% CI*  RR†  95% CI  Gender‡      Exercise       Women  0.60  0.36, 0.98   High  0.60  0.34, 1.07   Men  0.46  0.28, 0.75   Low  0.55  0.35, 0.87  Age, years§      Meat consumption       <80  0.68  0.43, 1.06   <1/week  0.46  0.28, 0.75   ≥80  0.59  0.34, 1.02   ≥1/week  0.54  0.34, 0.96  Smoker      Bread       Never  0.45  0.30, 0.69   Whole wheat  0.38  0.25, 0.60   Ever  0.61  0.33, 1.16   Not whole wheat only  0.73  0.42, 1.28  Blood pressure      Nut intake       Normotensive  0.52  0.33, 0.81   >1, 2/week  0.45  0.19, 1.04   Hypertensive  0.50  0.30, 0.85   ≤1, 2/week  0.54  0.36, 0.80  Body mass index¶      Intake of fluids other than water       ≤23.7  0.34  0.14, 0.83   <3 times/day  0.60  0.24, 1.52   >23.7  0.52  0.32, 0.85   ≥3 times/day  0.38  0.23, 0.62  Education             >High school  0.56  0.31, 1.02         ≤High school  0.50  0.33, 0.77        Risk factor for coronary heart disease  High vs. low water intake   Risk factor for coronary heart disease  High vs. low water intake   RR*,†  95% CI*  RR†  95% CI  Gender‡      Exercise       Women  0.60  0.36, 0.98   High  0.60  0.34, 1.07   Men  0.46  0.28, 0.75   Low  0.55  0.35, 0.87  Age, years§      Meat consumption       <80  0.68  0.43, 1.06   <1/week  0.46  0.28, 0.75   ≥80  0.59  0.34, 1.02   ≥1/week  0.54  0.34, 0.96  Smoker      Bread       Never  0.45  0.30, 0.69   Whole wheat  0.38  0.25, 0.60   Ever  0.61  0.33, 1.16   Not whole wheat only  0.73  0.42, 1.28  Blood pressure      Nut intake       Normotensive  0.52  0.33, 0.81   >1, 2/week  0.45  0.19, 1.04   Hypertensive  0.50  0.30, 0.85   ≤1, 2/week  0.54  0.36, 0.80  Body mass index¶      Intake of fluids other than water       ≤23.7  0.34  0.14, 0.83   <3 times/day  0.60  0.24, 1.52   >23.7  0.52  0.32, 0.85   ≥3 times/day  0.38  0.23, 0.62  Education             >High school  0.56  0.31, 1.02         ≤High school  0.50  0.33, 0.77        * RR, relative risk; CI, confidence interval. † Relative risk comparing those with intake of five or more 8-oz (240 ml) glasses of water daily versus two or fewer, excluding unknowns and adjusted for age and sex except ‡ and §. RR = 1.00 for intake of two or fewer glasses of water daily. Results for medium level of water intake were omitted to conserve space. ‡ Adjusted for age only. § Adjusted for sex only. ¶ Body mass index = weight (kg)/height (m)2. View Large Fluids other than water and risk of coronary heart disease In univariate analysis, intake of fluids other than water was associated with increased risk of fatal coronary heart disease (table 5). The association was statistically significant in women who drank five or more compared with two or fewer servings daily, with a relative risk of 2.47 (95 percent CI: 1.04, 5.88). When adjusting for traditional risk factors, as well as for water and caloric intake, exercise, diet, and replacing body mass index with weight, the point estimates remained virtually unchanged and significance was retained. However, the confidence intervals around these estimates were quite wide. TABLE 5. Relative risk of fatal coronary heart disease according to intake of “fluids other than water” among study participants 38 years of age or more who reported no physician-diagnosed heart disease, stroke, or diabetes at baseline (1976–1977), Adventist Health Study   8-oz (240 ml) glasses of “fluids other than water”/ day  Model 1*,†   Model 2*,‡   Model 3*,§   Model 4*,¶     RR#  95% CI#  RR  95% CI  RR  95% CI  RR  95% CI  Men (n = 8,280)  ≤2  1.00    1.00    1.00    1.00      >2–<5  1.44  0.75, 2.76  1.34  0.75, 2.76  1.29  0.65, 2.52  1.46  0.73, 2.92    ≥5  1.46  0.70, 3.03  1.07  0.48, 2.35  1.00  0.45, 2.22  1.34  0.59, 3.04      Trend    Trend    Trend    Trend        p = 0.4    p = 0.9    p = 0.9    p = 0.7    Women (n = 12,017)  ≤2  1.00    1.00    1.00    1.00      >2–<5  2.02  0.90, 4.50  2.14  0.83, 5.50  2.01  0.78, 5.18  2.25  0.87, 5.80    ≥5  2.47  1.04, 5.88  2.98  1.10, 8.07  2.79  1.03, 7.62  3.32  1.18, 9.30      Trend    Trend    Trend    Trend        p = 0.06    p = 0.03    p = 0.04    p = 0.02      8-oz (240 ml) glasses of “fluids other than water”/ day  Model 1*,†   Model 2*,‡   Model 3*,§   Model 4*,¶     RR#  95% CI#  RR  95% CI  RR  95% CI  RR  95% CI  Men (n = 8,280)  ≤2  1.00    1.00    1.00    1.00      >2–<5  1.44  0.75, 2.76  1.34  0.75, 2.76  1.29  0.65, 2.52  1.46  0.73, 2.92    ≥5  1.46  0.70, 3.03  1.07  0.48, 2.35  1.00  0.45, 2.22  1.34  0.59, 3.04      Trend    Trend    Trend    Trend        p = 0.4    p = 0.9    p = 0.9    p = 0.7    Women (n = 12,017)  ≤2  1.00    1.00    1.00    1.00      >2–<5  2.02  0.90, 4.50  2.14  0.83, 5.50  2.01  0.78, 5.18  2.25  0.87, 5.80    ≥5  2.47  1.04, 5.88  2.98  1.10, 8.07  2.79  1.03, 7.62  3.32  1.18, 9.30      Trend    Trend    Trend    Trend        p = 0.06    p = 0.03    p = 0.04    p = 0.02    * Excluding subjects with unknown values for model variables. † Adjusted for age. ‡ Adjusted for traditional coronary heart disease risk factors age, smoking, hypertension, body mass index, and education level for men, plus estrogen replacement therapy in menopausal women. § Adjusted for intake of water in addition to traditional coronary heart disease risk factors. ¶ Adjusted for energy in addition to traditional coronary heart disease risk factors. # RR, relative risk; CI, confidence interval. View Large We were unable to determine the degree to which juices or sugared drinks (including soda) might individually contribute to the increased risk of fatal coronary heart disease in women, because too few of this study population consumed any of these beverages individually more than once a day. After adjustment for water intake, the association with intake of milk (omitting soy milk) was close to the null and with caffeinated beverages (coffee, tea, and caffeinated sodas), positive but not statistically significant. DISCUSSION We have found a strong negative multivariate association between intake of water and risk of fatal coronary heart disease and, in contrast, a positive association between intake of fluids other than water and risk. These results were consistent between the sexes, although the negative association with water was stronger in men and the positive association with fluids other than water was stronger in women. Moreover, formal tests of the differences between the opposite-signed linear trends of water and fluids other than water were significant (p < 0.01) in both sexes independently. Our study has strengths and limitations. As an observational study, potential confounding always needs to be considered. Could fluid consumption be a marker of other risk factors not included in the model? Few candidate confounding variables are obvious. Those subjects who drink more water may be more health conscious in other less clearly defined ways. Drinking more water may be a marker of higher physical activity or those with higher energy intake. However, a two-tailed Pearson's correlation test showed no correlation between water and energy intakes (r = 0.08, p < 0.01 and r = 0.06, p < 0.01 for males and females, respectively) and, when exercise and energy consumption are included in the multivariate model, the effect of water consumption remained. Diabetics, who may drink more fluids, are excluded from these analyses. That the negative association with water intake was relatively consistent across many subgroups of the population makes confounding less likely as an explanation for this effect. The study population is large and relatively well educated, includes men and women with a broad range of ages, and provides detailed baseline information on the intake of fluids of different kinds. Data were missing for males and females, respectively, on between 1.7 and 6.6 percent and on between 1.3 and 9.0 percent of the various items included in the measurements of fluid intake. If the associations between fluid intake and coronary heart disease were markedly different for those who did not provide all the relevant data, then associations in the remainder may differ from those in the whole population. However, we have no reason to suspect this. The results can strictly be applied only to the California Adventist population. Moreover, because they drink more water and less caffeinated and alcoholic beverages than the US averages, it could be that, unlike some other findings in this population, this specific finding may apply only to this population alone. However, California Adventists probably do not differ biochemically or physiologically from others. If these results describe causal associations, they will probably also be found in the general population. The traditional risk factors for coronary heart disease operate with their usual force in California Adventists (31). Associations between coronary heart disease and certain foods (nuts and meats), which were first described in this population, have now been confirmed in several other populations (32–34). An important question is whether there are known mechanisms that might underlie these results if the associations are causal. There is a small literature strongly suggesting that several factors related to blood viscosity are adversely affected by hypohydration (16–21, 35–39) and that these rheologic factors predict the risk of coronary heart disease (4, 6, 8, 9). Whole blood viscosity is determined mainly by hematocrit and fibrinogen levels, both of which have frequently been associated with coronary heart disease events (1–9). It is interesting that, although percentage differences in blood viscosity-associated factors between those who do and do not experience coronary heart disease events are often statistically significant, they are quite small, often in the range of 2–7 percent (2, 6, 9, 18, 19, 21, 40, 41). Yet these are the kind of changes seen in mild, nonclinical hypohydration (16–20), within the presumed normal range. There is a well-described circadian rhythm of hemorheologic factors that can be quite marked (18, 41), perhaps due to continued non-reabsorbed glomerular filtration at night in the absence of fluid intake (19, 42). One can speculate that relatively elevated blood viscosity in the morning could be one factor accounting for the well-known increased risk of coronary heart disease events at this time of the day (43). Why may coronary heart disease risk be increased with a higher intake of fluids other than water? Several mechanisms can be postulated. Caffeinated beverages are mild diuretics and thus may raise blood viscosity (44). High energy drinks such as juices and regular sodas have osmolalities between 556 and 836 mOsm/kg (45). Their consumption causes a net movement of fluid from the vascular system into the intestinal lumen, resulting in a rapid elevation in blood viscosity after consumption (24). Perfusing the duodenum with a glucose solution that has an osmolality of 456 mOsm/kg, which is lower than that of juices and regular soda, was observed to reduce plasma volume by 3.3 percent within 105 minutes (46). Further, serum triglyceride levels may be raised by drinks containing high concentrations of sugar, including fruit juices (47–50). In subjects drinking 2.7 glasses (640 ml) of grape juice and three glasses (750 ml) of orange juice daily, triglyceride levels were seen to increase by 50 percent (51) and 30 percent (52), respectively. There is growing evidence that postprandial and fasting elevations of plasma triglycerides are independent coronary heart disease risk factors (53–58), even within the high normal range (59). Elevated triglyceride levels have been positively associated with coagulation factor VII (60), plasminogen activator inhibitor-1 (57), and thrombotic factors X (60) and IX (57) but inversely associated with antithrombin III (58) and plasma fibrinolytic activity (57). We hypothesize that higher intake of diuretic and high energy beverages results in more frequent and larger exposure to conditions that increase the risk of thrombosis and atherosclerosis. The magnitudes of the apparent effects associated with greater intake of both types of fluid appeared to differ between the two sexes. Although these differences could easily be explained by chance, it may be relevant that women typically have lower blood viscosity than do men because of lower hematocrit levels, erythrocyte rigidity, and aggregability at all shear rates (7). They may also experience less reduction in plasma volume with dehydration (61) and regulate their body temperatures at lower sweat rates (62). Reduced water intake in women, therefore, may not produce high levels of blood viscosity as readily as in men. On the other hand, because men have larger blood volumes than do women, the same quantities of hyperosmolar fluids or caffeinated beverages (fluids other than water) would produce a lower proportionate decrease in intravascular volume and subsequently a smaller increase in blood viscosity. In summary, we report intriguing associations between the intake of fluids and the risk of coronary heart disease that are not obviously explained by confounding. Further research in other populations, possibly including experimental study designs, is necessary to decide whether the associations are causal. Correspondence to Dr. Jacqueline Chan, Adventist Health Studies 2, School of Public Health, Loma Linda University, 24785 Stewart St., Room 203, Loma Linda, CA 92350 (e-mail: jchan@sph.llu.edu). Supported in part by National Institutes of Health grant 5-R01-HL-26210. REFERENCES 1. Becker RC. The role of blood viscosity in the development and progression of coronary artery disease. Cleve Clin J Med  1993; 60: 353–8. Google Scholar 2. Lowe GD, Lee AJ, Rumley A, et al. Blood viscosity and risk of cardiovascular events: the Edinburgh Artery Study. Br J Haematol  1997; 96: 168–73. Google Scholar 3. Lee AJ, Mowbray PI, Lowe GD, et al. Blood viscosity and elevated carotid intima-media thickness in men and women. Circulation  1998; 97: 1467–73. Google Scholar 4. Koenig W, Ernst E. The possible role of hemorheology in atherothrombogenesis. Atherosclerosis  1992; 94: 93–107. Google Scholar 5. Becker RC. Seminars in thrombosis, thrombolysis, and vascular biology. Part 5. Cellular-rheology and plasma viscosity. Biorheology  1991; 79: 265–70. Google Scholar 6. Ernst E. Hematocrit and cardiovascular risk. J Intern Med  1995; 237: 527–8. Google Scholar 7. De Simone G, Devereux RB, Shu C, et al. Relation of blood viscosity to demographic and physiologic variables and to cardiovascular risk factors in apparently normal adults. Circulation  1990; 81: 107–17. Google Scholar 8. Erikssen G, Thaulow E, Sandvik L, et al. Haematocrit: a predictor of cardiovascular mortality? J Intern Med  1993; 234: 493–9. Google Scholar 9. Lowe GD, Drummond MM, Lorimer AR, et al. Relation between extent of coronary artery disease and blood viscosity. Br Med J  1980; 280: 673–4. Google Scholar 10. Resch KL, Ernst E, Matrai A, et al. Can rheologic variables be of prognostic relevance in arteriosclerotic diseases? Angiology  1991; 42: 963–70. Google Scholar 11. Smith WC, Lowe GD, Lee AJ, et al. Rheological determinants of blood pressure in a Scottish adult population. J Hypertens  1992; 10: 467–72. Google Scholar 12. Kannel WB, McGee DL. Update on some epidemiologic features of intermittent claudication: the Framingham Study. J Am Geriatr Soc  1985; 33: 13–18. Google Scholar 13. Julius S, Palatini P, Nesbitt SD. Tachycardia: an important determinant of coronary risk in hypertension. J Hypertens Suppl  1998; 16: S9–15. Google Scholar 14. Harrison MJ, Pollock S, Kendall BE, et al. Effect of hematocrit on carotid stenosis and cerebral infarction. Lancet  1981; 2: 114–15. Google Scholar 15. Finch CA, Lenfant C. Oxygen transport in man. N Engl J Med  1972; 286: 407–15. Google Scholar 16. Dvilansky A, Bar-Am J, Nathan I, et al. Hematologic values in healthy older people in the Negev area. Isr J Med Sci  1979; 15: 821–5. Google Scholar 17. Kristal-Boneh E, Glusman JG, Chaemovitz C, et al. Improved thermoregulation caused by forced water intake in desert dwellers. Eur J Appl Physiol  1988; 57: 220–4. Google Scholar 18. Seaman GV, Engel R, Swank RL, et al. Circadian periodicity in some physicochemical parameters of circulating blood. Nature  1965; 207: 833–5. Google Scholar 19. Kurabayashi H, Kubota K, Tamura J, et al. A glass of water at midnight for possible prevention of cerebral infarction. Stroke  1991; 22: 1326–7. Google Scholar 20. Vandewalle H, Lacombe C, Lelièvre JC, et al. Blood viscosity after a 1-hour submaximal exercise with and without drinking. Int J Sports Med  1988; 9: 104–7. Google Scholar 21. Yasaka M, Yamaguchi T, Oitia J, et al. Clinical features of recurrent embolization in acute cardioembolic stroke. Strong disposing factors: low plasma levels of antithrombin II. Stroke  1993; 24: 1681–5. Google Scholar 22. Toiler GH, Brezinski D, Schafer AI, et al. Concurrent morning increase in platelet aggregability and the risk of myocardial infarction and sudden cardiac death. N Engl J Med  1987; 316: 1514–18. Google Scholar 23. Blanchard J, Sawers SJ. Relationship between urine flow rate and renal clearance of caffeine in man. J Clin Pharmacol  1983; 23: 134–8. Google Scholar 24. Maughan RJ, Leiper JB. Limitations to fluid replacement during exercise. Can J Appl Physiol  1999; 24: 173–87. Google Scholar 25. Beeson WL, Mills PK, Phillips RL, et al. Chronic disease among Seventh-day Adventists, a low risk group. Rationale, methodology, and description of the population. Cancer  1989; 64: 570–81. Google Scholar 26. Fraser GE, Sabaté J, Beeson WL, et al. A possible protective effect of nut consumption on risk of coronary heart disease. Arch Intern Med  1992; 152: 1416–24. Google Scholar 27. Cox DR, Oakes D. Chapter 7. In: Analysis of survival data. New York, NY: Chapman and Hall, 1984. Google Scholar 28. Fraser GE. Diet and coronary heart disease: beyond dietary fats and low density lipoprotein cholesterol. Am J Clin Nutr  1994; 59(suppl): 1117S–23S. Google Scholar 29. Snowdon DA, Phillips RL, Fraser GE. Meat consumption and fatal ischemic heart disease. Prev Med  1984; 13: 490–500. Google Scholar 30. Ershow AG, Cantor KP. Total water and tap water intake in the United States: population-based estimates of quantities and sources. Prepared under National Cancer Institute order #263-MD-810264 with the Life Sciences Research Office. Bethesda, MD: Federation of American Societies for Experimental Biology, 1989. Google Scholar 31. Fraser GE, Strahan M, Sabate J, et al. Effects of traditional coronary risk factors on rates of incident coronary events in a low risk population. The Adventist Health Study. Circulation  1992; 86: 406–13. Google Scholar 32. Fraser GE. Nut consumption, lipids, and risk of a coronary event. Clin Cardiol  1999; 22(suppl): III-11–15. Google Scholar 33. Sabaté J. Nut consumption, vegetarian diets, ischemic heart disease risk, and all-cause mortality: evidence from epidemiologic studies. Am J Clin Nutr  1999; 70(suppl): 500S–3S. Google Scholar 34. Key TJ, Fraser GE, Thorogood M, et al. Mortality in vegetarians and nonvegetarians: detailed findings from a collaborative analysis of 5 prospective studies. Am J Clin Nutr  1999; 70(suppl): 516S–24S. Google Scholar 35. Rolls BJ, Wood RJ, Rolls ET, et al. Thirst following water deprivation in humans. Am J Physiol  1980; 239: R476–82. Google Scholar 36. Dintenfass L. The cause of death: blood hyperviscosity? (Letter). J R Soc Med  1987; 80: 536–7. Google Scholar 37. Pearson TC. Apparent polycythaemia. Blood Rev  1991; 5: 205–13. Google Scholar 38. Beijering RJ, Gips CH, Huizenga JR, et al. Whole blood and plasma water in health and disease: longitudinal and transverse observations and correlations with several different hematological and clinicochemical parameters. Clin Chim Acta  1997; 258: 59–68. Google Scholar 39. Warren JL, Bacon WE, Harris T, et al. Burden and outcomes associated with dehydration among U.S. elderly, 1991. Am J Public Health  1994; 84: 1265–9. Google Scholar 40. Koenig W, Sund M, Filipiak B, et al. Plasma viscosity and the risk of coronary heart disease: results from the MONICA-Augsburg Cohort Study, 1984 to 1992. Arterioscler Thromb Vasc Biol  1998; 18: 768–72. Google Scholar 41. Kubota K, Sakurae T, Tamura J, et al. Is the change in hematocrit and blood viscosity a factor triggering cerebral and myocardial infarction? Stroke  1987; 18: 812–13. Google Scholar 42. Wood JH. Is the change in hematocrit and blood viscosity a factor triggering cerebral and myocardial infarction? (Letter). Stroke  1987; 18: 813. Google Scholar 43. Cohen MC, Muller JE. Onset of acute myocardial infarction—circadian variation and triggers. Cardiovasc Res  1992; 26: 831–8. Google Scholar 44. Neuhauser-Berthold, Beine S, Verwied SC, et al. Coffee consumption and total body water homeostasis as measured by fluid balance and bioelectrical impedance analysis. Ann Nutr Metab  1997; 41: 29–36. Google Scholar 45. Bell SJ, Anderson FL, Bistrian BR, et al. Osmolality of beverages commonly provided on clear and full liquid menu. Nutr Clin Pract  1987; 2: 241–4. Google Scholar 46. Shi X, Summers RW, Schedl HP, et al. Effects of carbohydrate type and concentration and solution osmolality on water absorption. Med Sci Sports Exerc  1995; 27: 1607–15. Google Scholar 47. Kuo PT, Bassett DR. Dietary sugar in the production of hypertriglyceridemia. Ann Intern Med  1965; 62: 1199–212. Google Scholar 48. Parks EJ, Hellerstein MK. Carbohydrate-induced hypertriacylglycerolemia: historical perspective and review of biological mechanisms. Am J Clin Nutr  2000; 71: 412–33. Google Scholar 49. Hudgins LC, Hellerstein MK, Seidman CE, et al. Relationship between carbohydrate-induced hypertriglyceridemia and fatty acid synthesis in lean and obese subjects. J Lipid Res  2000; 41: 595–604. Google Scholar 50. Parks EJ, Krauss RM, Christiansen MP, et al. Effects of a low-fat, high-carbohydrate diet on VLDL-triglyceride assembly, production, and clearance. J Clin Invest  1999; 104: 1087–96. Google Scholar 51. Stein JH, Keevil JG, Wiebe DA, et al. Purple grape juice improves endothelial function and reduces the susceptibility of LDL cholesterol to oxidation in patients with coronary artery disease. Circulation  1999; 100: 1050–5. Google Scholar 52. Kurowska EM, Spence JD, Jordan J, et al. HDL-cholesterol-raising effect of orange juice in subjects with hypercholesterolemia. Am J Clin Nutr  2000; 72: 1095–100. Google Scholar 53. Cullen P. Evidence that triglycerides are an independent coronary heart disease risk factor. Am J Cardiol  2000; 86: 943–9. Google Scholar 54. Gaziano JM. Triglycerides and coronary risk. Curr Cardiol Rep  1999; 1: 125–30. Google Scholar 55. Patsch JR, Miesenbock G, Hopferwieser T, et al. Relation of triglyceride metabolism and coronary artery disease. Studies in the postprandial state. Arterioscler Thromb  1992; 12: 1336–45. Google Scholar 56. Austin MA. Epidemiology of hypertriglyceridemia and cardiovascular disease. Am J Cardiol  1999; 83: 13F–16F. Google Scholar 57. Miller GJ. Lipoproteins and the haemostatic system in atherothrombotic disorders. Baillieres Clin Haematol  1994; 7: 713–32. Google Scholar 58. Zajtchuk R, Zajtchuk J. Relationship of triglyceride levels to thrombosis in patients with coronary artery disease. Ann Thorac Surg  1983; 35: 274–6. Google Scholar 59. Kreisberg RA. Hypertriglyceridemia and coronary heart disease. Clin Rev  2000;spring: 29–32. Google Scholar 60. de Sousa JC, Soria C, Ayrault-Jarrier M, et al. Association between coagulation factors VII and X with triglyceride rich lipoproteins. J Clin Pathol  1988; 41: 940–4. Google Scholar 61. Byrd R, Stewart L, Torranin C, et al. Sex differences in response to hypohydration. J Sports Med Phys Fitness  1977; 17: 65–8. Google Scholar 62. Rocker L, Kirsch KA, Stoboy H, et al. Influence of heat stress on plasma volume and intravascular proteins in sedentary females. Eur J Appl Physiol  1977; 36: 187–92. Google Scholar

Journal

American Journal of Epidemiology – Oxford University Press

Published: May 1, 2002

There are no references for this article.