Relation of diet to cardiovascular disease risk factors in subjects with cardiovascular disease in Australia and New Zealand: analysis of th
the Baker Heart Research Institute, Melbourne, Australia (PJN)
the CSIRO Division of Health Science & Nutrition, Adelaide, Australia (KB)
the Department of Medicine, University of Queensland, Brisbane, Australia (DMC)
the National Health and Medical Research Council Clinical Trials Centre, University of Sydney, Sydney, Australia (RJS, KM, AK, and CP)
the Cardiology Department, Auckland City Hospital, Auckland, New Zealand (HDW)
the Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Australia (AMT)
ABSTRACT
Background: Comparisons of the relation of diet with coronary heart disease (CHD) between countries with similar socioeconomic environments have been few. Patients in Australia and New Zealand (n = 9014) who participated in a large secondary prevention trial had significantly different CHD mortality rates.
Objective: The objective of this study was to ascertain the effects of nutrient consumption on cardiovascular disease risk in patients from the 2 countries.
Design: Nutrient consumption patterns were surveyed in a subgroup of 1077 patients on 3 occasions over 4 y during an intervention trial with a statin.
Results: Within the entire cohort of 9014 patients, the New Zealanders had significantly (40%) more cardiovascular deaths than did the Australians. In the subgroup of 1077 patients, the New Zealanders were found at entry to have eaten significantly more total (69.34 ± 12.35 compared with 66.45 ± 12.9 g/d) and saturated (26.23 ± 8.41 compared with 24.37 ± 7.36 g/d) fat (P < 0.001 for each) and to have significantly (4%) higher concentrations of LDL cholesterol (3.96 ± 0.74 compared with 3.8 ± 0.76 mmol/L; P < 0.001) than did the Australians. At baseline, patients with previous coronary artery bypass grafting had diets that were significantly different from those of patients without previous coronary artery bypass grafting. Relations between nutrients and plasma lipids confirmed the direct effects of saturated fatty acids on LDL cholesterol and of alcohol on plasma triacylglycerol and HDL cholesterol. Dietary counseling throughout the trial led to significant improvements in compliance with guidelines. However, neither the baseline nor the improved 1-y nutrient intakes predicted future changes in cardiovascular events.
Conclusion: Differences in CHD mortality and in LDL-cholesterol concentrations between 2 populations with similar socioeconomic and cultural backgrounds were consistent with the amounts and types of fats eaten.
Key Words: Total fat saturated fat cardiovascular events LDL cholesterol population comparison clinical trial
INTRODUCTION
Differences in patterns of eating or in the consumption of specific nutrients have been found to contribute to observed differences in the prevalence of cardiovascular disease (CVD) between populations. Consumption of saturated fat appeared to correlate strongly in epidemiologic studies beginning with the Seven Countries Study (1), and this correlation was confirmed in subsequent surveys (2, 3). The increasing incidence of coronary heart disease (CHD) among migrant populations also appeared to be associated with changes in dietary habits that included higher intakes of saturated fat (4, 5). Prospective trials among large cohort populations confirmed the importance of the dietary trans and saturated fatty acids in predicting future CVD events (6, 7). From these longitudinal studies has emerged a pattern of eating that included low consumption of animal fats that correlated with less future CVD (8, 9).
Comparisons of the diet-CVD relation in countries whose populations differ only marginally in their socioeconomic and cultural characteristics have not been widely studied. Australia and New Zealand are socioeconomically and culturally similar. Both countries participated in a large placebo-controlled trial of secondary CVD prevention with the -hydroxy--methylglutaryl coenzyme A reductase inhibitor pravastatin (10). More than 9000 persons with previous acute coronary syndrome participated in the Long-Term Intervention with Pravastatin in Ischaemic Disease (LIPID) trial, which lasted 5 y and had a mean 6-y follow-up. Within an overall substantial reduction in CHD events among those treated with pravastatin, there was a significant difference between the findings in subjects from Australia and New Zealand. CHD mortality was significantly higher in those from New Zealand than from Australia, and LDL-cholesterol concentrations were, on average, significantly higher among New Zealanders than Australians (relevant data provided below). The prevalence of other major risk factors did not differ significantly between the 2 groups.
Among subjects recruited into the LIPID trial, 1077 took part in the baseline survey of eating habits that used a detailed food-frequency questionnaire (FFQ). Further surveys were conducted among this subcohort 1 y and 4 y later; the respective response rates were 89% and 84%, so that a progressive dietary history was obtained over 4 y in 849 subjects, approximately one-half of whom were treated with pravastatin. In this report, we describe the dietary characteristics at baseline in Australian and New Zealander participants in relation to the prevalence of major CVD risk factors, the nature of their initial clinical presentation, and the clinical outcome over 5 y.
SUBJECTS AND METHODS
Patients and study design
The design and results of the LIPID trial have been reported elsewhere (10). Patients (n = 9014) had a history of acute myocardial infarction or hospitalization for unstable angina pectoris and a baseline total-cholesterol concentration of 4.0–7.0 mmol/L (155–270 mg/dL). The main finding of the trial was that the risk of death due to CHD was 24% lower in those assigned pravastatin than in the placebo group. Pravastatin in the trial was supplied by Bristol Myer Squibb (Princeton, NJ). After the initial double-blind trial, an open phase was conducted, during which an increasing number of subjects were treated with pravastatin. The rates of cardiovascular events and mortality were compared between the Australian and New Zealander subjects over the study span of 8 y. A full analysis of this comparison is yet to be completed, but the method is similar to that used for the entire cohort (10). In this report, we present the key data in terms of deaths due to either CHD or to any cardiovascular event. These data were adjusted for treatment with pravastatin and for a risk score encompassing conventional risk factors as defined previously (11).
As part of the LIPID trial, patients were given dietary advice at the beginning of the run-in phase and at least every 6 mo at their routine clinic visits throughout follow-up. Advice was administered in a number of ways: via pamphlets and trial newsletters and personally by dietitians, study nurses, or physicians at the LIPID trial centers. Pamphlets supplied by the National Heart Foundation of Australia provided advice regarding foods to avoid (eg, foods high in fat such as fatty meats, full-cream dairy foods, potato chips, and take-away foods), foods to include (ie, fruit, vegetables, fish, and whole grains), guidelines for weight control, guidelines for reducing plasma cholesterol, and low-fat cooking hints. The dietary advice followed the National Heart Foundation's recommended targets of 30% of energy as fat, 55% as carbohydrate, and 15% as protein and of fatty acids derived equally from monounsaturated, polyunsaturated, and saturated fatty acids. Dietary cholesterol was recommended to be <300 mg/d. Except for dietary advice, other aspects of each patient's care were left to the discretion of his or her treating doctor.
A proportion of centers invited their patients to take part in the substudy to assess changes in dietary intake for the duration of the LIPID trial. Patients who agreed to participate were asked to complete self-administered FFQs at 3 time points: during the run-in phase, at 1 y, and at 4 y after randomization. These patients were also provided with feedback regarding success in reaching targets. This feedback was in report form and included suggestions for further dietary substitutions if necessary.
Of the 9014 patients randomly assigned to the LIPID trial, 1314 were approached to take part in the diet substudy, and 1077 agreed to participate. Nine hundred fifty-one (89% of respondents; 73% of those initially approached) and 849 (84% and 65%, respectively) patients completed subsequent questionnaires at years 1 and 4, respectively (Figure 1). A total of 794 patients who survived to year 4 completed all 3 questionnaires. One subject was lost to follow-up.
The baseline characteristics relevant to cardiovascular and nutritional outcomes did not differ significantly between the total cohort and the subgroup. Of the 1077 patients, approximately one-half had been randomly assigned to pravastatin treatment. The baseline characteristics of those assigned to receive placebo and of those assigned to receive pravastatin also did not differ significantly.
Written informed consent was obtained from all participants in the trial. The LIPID trial, including the nutritional evaluation, was approved by the human ethics committees of the participating centers.
Food-frequency questionnaire
The FFQ was a semiquantified questionnaire that was previously validated (12) and subsequently used in Commonwealth Scientific & Industrial Research Organisation Australian Food and Nutrition Surveys between 1988 and 1993. The questionnaire assessed the intake of >170 foods. Participants were asked to record how often they usually ate a specific food. They were provided with standard portion sizes for each food and asked to indicate whether they typically ate more or less than these portions. Response categories for frequency of consumption ranged from "never" or "rarely" to "many times a day." This recording system provided 10 frequency categories from "never" and "rarely" to the number of times a food was eaten daily, then weekly, and then monthly. Information regarding types of foods such as bread, milk, spreads, yogurt, and cooking oil was also obtained. Vitamin and mineral supplement use was not included in the data analyzed.
Dietitians and nurse-managers in each center explained in detail the principles and methods for completing the questionnaires. Incomplete questionnaires were sent back to the participating center for completion by the patient. A questionnaire that was returned without being complete for the second time was excluded from analysis. Nutritionists reviewed the questionnaires with respect to foods and food groups; any items with extremely unrealistic food intakes were reviewed and excluded if no justification for that intake was found in the comments or otherwise.
The robustness of the FFQ was ensured by care in design of the questionnaire, the inclusion of many foods, the setting of an order of questions that allowed ease of answering, the request for information on choices within food groups, qualitative measures of food preparation, and flexibility in serving sizes. All these issues are addressed in this questionnaire (12).
Data analysis
Dietary nutrient intake was calculated by multiplying the frequency of consumption (servings/wk) by the size of the servings consumed and the nutrient composition, with adjustments for type of cooking fats used, meat trimming, type of fats spread, dairy products and bread consumed, and cooking methods and salting habits. Nutrient composition was primarily derived from the Australian NUTTAB92 food-composition database (13), which was supplemented with British food data (14)
The current report presents the nutrient data that are most relevant to CVD risk factors and hence to clinical outcomes. Nutrient intake data are expressed as energy-adjusted grams per day (15).
Assessment of endpoints
An endpoint expanded beyond that used in the LIPID trial was used for the diet substudy to increase the power of the study. The endpoint was a composite of cardiovascular events including coronary artery bypass graft (CABG) or percutaneous coronary intervention, fatal and nonfatal acute myocardial infarction, sudden death, and death due to CVD. Full details about the measurement and assessment of outcomes are provided elsewhere (10).
Statistical analysis
All analyses were performed on an intention-to-treat basis. Exploratory data analysis including the production of histograms was undertaken to assess the distribution of each nutrient and lipid fraction. All variables were shown to be approximately normally distributed. Means and standard deviations have been presented for continuous data. Categorical data have been presented as frequencies and percentages. Differences in baseline characteristics between countries (Australia and New Zealand) and between clinical status (CABG or no CABG) were tested by using the two-sample t test or the chi-square test.
The effects of obesity [body mass index (BMI; in kg/m2): >30] and diabetes on nutrient intake were analyzed with a 3-factor (country, obesity, and diabetes) factorial analysis of variance. Significant (P < 0.05) interactions and main effects for obesity and diabetic status are reported. These analyses were carried out after first ensuring that there were no interactions between country and obesity or diabetes. The strengths of the relations between nutrient and lipid fractions were assessed by using Pearson's correlation coefficients. Correlations were calculated for each country separately and tested for homogeneity. Weighted average correlations between countries were presented.
Differences in mean nutrient intakes over time were analyzed by using repeated-measures analysis of variance with 3 time points. When a significant (P < 0.05) change in nutrient intake was found by using the Greenhouse-Geiser F test, the differences between each time point were examined by using t tests with a Bonferroni correction. Differences in clinical outcomes between countries were examined by using time-to-event analysis involving Cox regression. Time-to-event analyses were performed by using Cox regression, in which the nutrient intake was treated as a categorical variable (quintiles). A test of trend across the quintiles was undertaken. Univariate analysis for each nutrient was conducted.
P values were two-sided. Statistical analyses were performed with SAS for WINDOWS software (version 5.0.2195, Release 8.02; SAS Institute Inc, Cary, NC) and SPSS software (version 11.5; SPSS Inc, Chicago, IL).
RESULTS
Baseline characteristics and nutrient intakes by country
A major aim was to ascertain whether differences existed between the characteristics of the trial patients living in Australia and New Zealand. The differences in the average lipid profile of the 2 populations are shown in Table 1. New Zealanders showed a significantly (4%) higher mean plasma concentration of LDL cholesterol and a significantly lower concentration of HDL cholesterol than did Australians. It may be important that there were no significant differences between the populations in the prevalence of obesity or the number of smokers, although there was a trend for greater incidence of diabetes among Australians than among New Zealandres (9% and 6%, respectively; P = 0.07).
Also shown in Table 1 is macronutrient consumption at the time the lipid measurements were made. New Zealanders consumed significantly more energy (8.5%) and total fat (7.6%) derived from both saturated (7.6%) and monounsaturated (3.5%) fatty acids than did Australians. They also ate 2% less carbohydrate than did Australians. Sodium consumption did not differ significantly between the 2 populations.
Major clinical outcomes for the entire cohort by country
At the end of both the double-blind and open-label (total time: 8 y) phases of the LIPID trial, the incidence of the primary endpoint, CHD mortality, was significantly higher in New Zealand (12.8%) than in Australia (8.6%) (Table 2). This significant difference was maintained after adjustment for baseline risk factors [ie, risk score (11)] and for treatment with pravastatin, ie, a hazard ratio (HR) of 1.42 (95% CI: 1.25, 1.63). Similarly, the risk of death due to CVD, after adjustment, was 40% (95% CI: 23%, 58%) higher in New Zealand than in Australia.
Baseline nutrient intake according to clinical status at entry
We next examined whether macronutrient consumption varied among patients enrolled with different previous manifestations of CHD. It is important to note that the trial included patients who had had acute myocardial infarction or had been hospitalized for unstable angina pectoris between 3 mo and 3 y previously. Other manifestations such as coronary revascularization might have occurred either during this time frame or beforehand, and that might have allowed patients in specific categories or patients who had known about their illness for various lengths of time to make dietary adjustments.
It is interesting that the 342 patients who underwent CABG were found at their initial visit to have a diet that was significantly different from that of the other patients: ie, the CABG patients consumed 3% less total fat, 7.8% less saturated fatty acids, and 7.8% less cholesterol than did the non-CABG patients. The data presented in Table 3 therefore merely define the pattern of nutrient intake of this particular group of patients at entry, which probably reflects these reductions in consumption of meat, eggs, and full-cream dairy products. The CABG patients ate 7.8% more fiber than did the 735 non-CABG patients, and they also had significantly (P < 0.05) lower LDL-cholesterol concentrations—3.74 ± 0.74 compared with 3.86 ± 0.76 mmol/L—than did the non-CABG patients.
Baseline nutrient intake among obese subjects and patients with type 2 diabetes
Also of interest was the pattern of eating in the subjects who were known to be at greater risk because they were also obese or diabetic. Before calculations of significant differences, the country obesity x diabetes interactions were tested and found to be nonsignificant for all nutrients except cholesterol (P = 0.014). The significant data are shown in Table 4. The obese subjects ingested significantly more fat from saturates and monounsaturates, significantly more protein, and significantly less carbohydrate than did the nonobese subjects. The diabetic patients ate significantly more total fat, monounsaturated fat, and protein and significantly less carbohydrate than did the nondiabetic subjects. The obese and diabetic patients therefore ate relatively more animal foods than plant-based foods. The differences shown in Table 4 are independent of country, although, in each category, the differences seen between countries for the entire cohort (Table 1) were also reflected in the 4 subgroups. The differences between countries were therefore the same for diabetic and nondiabetic subjects and for obese and nonobese subjects.
Relations between nutrients and lipid fractions
The correlation coefficients of nutrient intakes at baseline as predictors of plasma lipid concentrations based on continuous nutrient consumption data are shown in Table 5. The following relations were significant, but the correlations were of a low order. Total cholesterol was inversely related to energy and carbohydrate intake and directly related to alcohol intake (which may reflect the effect on HDL cholesterol); LDL cholesterol tended to be directly related to the intake of saturated fatty acid and inversely related to that of energy. Plasma triacylglycerol concentration was significantly related to the intake of alcohol (positively) and fiber (negatively), and HDL cholesterol was inversely related to energy and carbohydrate intakes and positively related to alcohol consumption.
Changes in nutrient intake during trial
We also examined the likelihood that a change in eating habits during the trial as the result of more extensive counseling might also have influenced the predictive value of the patterns of nutrient intake for clinical outcomes. With the exception of monounsaturated fatty acid intake (ie, patients receiving pravastatin consumed 3% less than did those receiving placebo; P = 0.04), there were no nutrient x treatment interactions and therefore the whole group was analyzed together. The changes in nutrient intake over the 4 y are shown in Table 6 by using the data collected at baseline and at years 1 and 4. Significant but modest changes occurred during this time; most occurred within the first year, except for alcohol consumption, which did not change. Moderate reductions occurred in total energy from most sources of macronutrients: total fat, all 3 groups of fatty acids, carbohydrate, and protein. Cholesterol and fiber consumption also decreased. Overall, >80% of subjects achieved the target for cholesterol intake (<250 mg), and 51% and 50% achieved the target for the consumption of saturated fat (<10% of energy) and total fat (<39% of total energy), respectively.
Clinical outcomes related to nutrient intake at baseline
There were no significant relations between clinical outcomes and the pattern of nutrient consumption obtained at baseline (5 y before) and tested as quintiles. There were no significant trends; all values were P > 0.12 (HRs from Cox regression). Because this may have partly reflected the significant dietary changes made, on average, over that period, as shown in Table 6, further analysis of the consumption of nutrients at year 1 in comparison with clinical outcomes was carried out. However, this analysis also failed to show any significant relations.
DISCUSSION
Cardiovascular mortality, nutrient consumption, and plasma lipids by country
The LIPID trial of patients with existing CVD recruited 5958 Australians and 3056 New Zealanders between 1990 and 1992, and clinical outcomes were initially assessed in 1997 (10). The outcomes for the entire cohort—ie, deaths due to CHD and CVD—were significantly worse among New Zealanders, who had a mean HR of 1.42 (95%CI: 1.25, 1.61) than among Australians (Table 2). Although the effects of conventional risk factors and treatment were considered in that analysis and found not to have influenced the differences in mortality, differences in dietary factors in a subgroup of 1077 subjects are explored in this report.
The pattern of macronutrient consumption at baseline did differ significantly between the 2 national populations in several important respects (Table 2). New Zealanders consumed slightly more energy and total fat (even after adjustment for energy), from both saturated and monounsaturated fatty acids than did the Australians; they also ate slightly less carbohydrate. New Zealanders also had a significantly (4%) higher mean LDL-cholesterol concentration than did the Australians, and their intake of saturated fatty acids correlated significantly, albeit weakly, with the LDL-cholesterol concentration on univariate analysis (Table 5). The difference in dietary patterns between the 2 populations might therefore have contributed to the differences in clinical outcomes observed for the entire LIPID trial cohort (10).
It has previously been difficult to show within Western populations a relation between total fat consumption and CVD. This difficulty has been attributed partly to the generally high intake of fat and the counteractive effects of saturated and trans fatty acids on the one hand and of unsaturated fatty acids on the other (9). It is therefore important to note that Australia and New Zealand are socioeconomically and culturally similar. On the other hand, differences between these populations in CVD that appear partly to reflect different dietary habits, especially with respect to the consumption of saturated fat, have been reported frequently (4, 5). Similar conclusions have been drawn in long-term prospective studies from the patterns of eating that were associated with the least and the greatest risk for future CHD events; the consumption of fatty meats, full-cream dairy products, and fried foods comprised the higher-risk pattern (8, 9). Consumption of meat and dairy products has traditionally been higher in New Zealand than in Australia, and that difference may largely explain the differences in fat intake and LDL-cholesterol concentration between the 2 populations. The LIPID trial was conducted against a background of falling CHD mortality rates in both countries. Of the 23% decline between 1982 and 1993 in New Zealand, one-half has been attributed to reductions in major risk factors [ie, 12% of the decline was due to lower serum cholesterol concentrations (16)], although dietary data were not considered. Yet the patients recruited from New Zealand during 1990 and 1992 were still eating foods significantly richer in saturated fatty acids than were the patients in the Australian cohort.
Nutrient intakes at entry into trial and during the subsequent 4 y
The profiles of dietary patterns at baseline and during the first 4 y of the trial showed that one-third of all patients with known CVD were following the key recommendations of the National Heart Foundation of Australia, which resemble those of the American Heart Association (17). The mean (±SE) intake of total fat for the subgroup was 29.8 ± 0.2% of energy, which is comparable to the 35% found in a 1995 national survey of Australian men of similar age to the subjects in the LIPID trial (18).
Both Australians and New Zealanders in the study changed their eating habits in years 1 through 4 (Table 6). At 1 y, >80% of subjects achieved the target for reducing cholesterol intake (10%), and 51% and 50% achieved the targets for reducing fat and total fat intakes (11% and 9%, respectively).
Two additional subgroups were deemed worthy of further investigation: subjects who were obese and subjects who had type 2 diabetes. At that time (1990–1992), both of those groups ate significantly more fat and protein but significantly less carbohydrate than did the nonobese and nondiabetic subjects, respectively (Table 4). In retrospect, this is of concern because higher consumption of saturated fat and cholesterol has been linked to a greater risk of CHD in diabetic as well as nondiabetic subjects (19).
Relations between nutrients and lipid fraction
These relations are consistent with the widely reported effects of nutrients on the 4 major plasma lipid fractions (9). The predictive effects of nutrients either by quintile of intake or as continuous nutrient consumption were highly significant, although the correlations were of a low order (Table 5). Saturated fat intake was directly related to LDL cholesterol, carbohydrate intake was inversely related to HDL cholesterol, and alcohol consumption increased both the plasma triacylglycerol and HDL-cholesterol concentrations. Although anticipated, these findings provide support for the robustness of the questionnaire and the adequacy of the size of the diet-study subgroup.
Clinical outcomes and study limitations
There was no evidence of an effect of nutrient intakes at baseline on clinical outcomes 5 y later in the diet-study subgroup. Dietary changes made mainly in year 1 by at least one-half of all patients, however, included eating less fat, saturated fat, and cholesterol, but an analysis testing the relation of clinical outcomes to year 1 intakes was also negative. The total number of clinical events (ie, 329) was probably inadequate to identify any existing relations. Nevertheless, a Cochrane review did come to a guarded conclusion that a reduced or modified fat intake, if followed for a period of 2 y, results in a small but potentially important reduction in the risk of cardiovascular events (20). Although Australians and New Zealanders differed in CVD mortality, nutrient intakes, and plasma lipids at baseline, the reductions in total fat and saturated fatty acids were not nearly as great in this trial as in dietary intervention trials of primary (21, 22) and secondary (23, 24) prevention in which a large exchange of polyunsaturated for saturated fatty acids led to significant reductions in initial or recurrent nonfatal CHD or subsequent cardiovascular events. On the other hand, more modest changes in diet, such as adoption of a diet based on Mediterranean tradition, when made in the context of other lifestyle changes, have been shown to lower the rate of CHD mortality (25). A further possible diluting factor was the inclusion of pravastatin-treated patients in the analysis, although no change in nutrient intake over time x assigned treatment interaction was found. Events in the placebo group alone were too few. The characteristics of the participants were similar to those of the entire cohort, but only 1077 of the 1314 patients who were approached to take part in the diet study agreed, which raised the possibility of selection bias. Finally, the nutrition data are only as robust as the FFQ method, which relies on the capacity of a person to recollect accurately his or her patterns of intake over time. However, FFQs have been applied extensively and successfully in many prospective trials and case-control investigations of average food and nutrient consumption by individuals.
ACKNOWLEDGMENTS
Statistical analyses were performed at the National Health and Medical Research Centre Clinical Trials Centre, University of Sydney, by CP, AK, and JS. PN, AT, HW, and DC have been members of the LIPID Management Committee, which has supervised various aspects of the trial, since its inception. AMT is the Medical Director of the National Heart Foundation of Australia. None of the authors had a personal or financial conflict of interest.
REFERENCES
Keys A, Menotti A, Karvonen MJ, et al. The diet and 15-year death rate in the Seven Countries Study. Am J Epidemiol 1986;124:903–15.
Shekelle RB, Shryock AM, Paul O, et al. Diet, serum cholesterol, and death from coronary heart disease. The Western Electric Study. N Engl J Med 1981;304:65–70.
Hu BH, Stampfer MJ, Manson JE, et al. Dietary fat intake and the risk of coronary heart disease in women. N Engl J Med 1997;337:1491–9.
Kushi LH, Lew RA, Stare FJ, et al. Diet and 20-year mortality from coronary heart disease. The Ireland-Boston diet-heart study. N Engl J Med 1985;312:811–8.
Kato H, Tillotson J, Nichamen MZ, Rhoads GG, Hamilton HB. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California: serum lipids and diet. Am J Epidemiol 1973;97:372–85.
Hu FB, Stampfer MJ, Manson JE, et al. Dietary saturated fats and their food sources in relation to the risk of coronary heart disease in women. Am J Clin Nutr 1999;70:1001–8.
Siscovick DS, Raghunatahan E, King I, et al. Dietary intake and cell membrane levels of long-chain n–3 polyunsturated fatty acids and the risk of primary cardiac arrest. JAMA 1995;274:1363–7.
Kerver JM, Yang EJ, Bianchi L, Song WO. Dietary patterns associated with risk factors for cardiovascular disease in healthy adults. Am J Clin Nutr 2003;78:1103–10.
Hu FB, Willett W. Optimal diets for prevention of coronary heart disease. JAMA 2002;288:2569–78.
Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and broad range of initial cholesterol levels. The Long-term Intervention with Pravastatin in Ischaemic Disease Study group. N Engl J Med 1998;339:1349–57.
Marschner IC, Colquhoun D, Simes RJ, et al. Long-term risk stratification for survivors of acute coronary syndromes. J Am Coll Cardiol 2001;38:56–63.
Rohan T, Record S, Cook M. Repeatability of estimates of nutrient and energy intakes. The quantitative food frequency approach. Nutr Res 1987;7:15–137.
Lewis J, Holt RJ. Nutrient data table for use in Australia 1991-92. National Food Authority. Canberra, Australia: AGPS, 1991.
Ministry of Agriculture, Fisheries and Food. Food portion sizes. 2nd ed. London, United Kingdom: Her Majesty's Stationery Office, 1993.
Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986;124:17–27.
Capewell S, Beaglehole R, Seddon M, McMurray J. Explanation for the decline in coronary heart disease mortality rates in Auckland, New Zealand, between 1982 and 1993. Circulation 2000;102:1511–6.
American Heart Association. Dietary guidelines for healthy American adults: a statement for physicians and health professionals by the Nutrition Committee, American Heart Association. Circulation 1988;77:721A–4A.
McLennan W, Podger A. A national nutritional survey: nutrient intakes and physical measurements. Australia, 1995. Canberra, Australia: Australian Bureau of Statistics, 1998.
Tanasescu M, Cho E, Manson JE, Hu FB. Dietary fat and cholesterol and the risk of cardiovascular disease among women with type s diabetes. Am J Clin Nutr 2004;79:999–1005.
Hooper L, Summerbell CD, Higgins JPT, et al. Reduced or modified dietary fat for preventing cardiovascular disease. Cochrane Database Syst Rev 2001;(3):CD002137.
Turpeinen O, Karvonen MJ, Pekkarinen M, Miettinen M, Elosuo R, Paavilainen E. Dietary prevention of coronary heart disease: the Finnish mental hospital study. Int J Epidemiol 1979;8:99–117.
Hjermann I, Velve Byre K, Holme I, Leren P. Effect of diet and smoking on the incidence of coronary heart disease. Lancet 1981;2:1303–10.
Leren P. The Oslo diet-heart study. Eleven year report. Circulation 1970;42:935–42.
Dayton S, Pearce ML, Hashimoto S, Dixon WJ, Tomiyasy U. A controlled clinical trial of a diet high in unsaturated fat in preventing complications of atherosclerosis. Circulation 1969;40:S1–63.
Knoops KTB, de Groot LCPGM, Kromhout D, et al. Mediterranean diet, lifestyle factors and 10-year mortality in elderly European men and women. JAMA 2004;292:1433–9., 百拇医药(Paul J Nestel, Katrine Ba)
the CSIRO Division of Health Science & Nutrition, Adelaide, Australia (KB)
the Department of Medicine, University of Queensland, Brisbane, Australia (DMC)
the National Health and Medical Research Council Clinical Trials Centre, University of Sydney, Sydney, Australia (RJS, KM, AK, and CP)
the Cardiology Department, Auckland City Hospital, Auckland, New Zealand (HDW)
the Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Australia (AMT)
ABSTRACT
Background: Comparisons of the relation of diet with coronary heart disease (CHD) between countries with similar socioeconomic environments have been few. Patients in Australia and New Zealand (n = 9014) who participated in a large secondary prevention trial had significantly different CHD mortality rates.
Objective: The objective of this study was to ascertain the effects of nutrient consumption on cardiovascular disease risk in patients from the 2 countries.
Design: Nutrient consumption patterns were surveyed in a subgroup of 1077 patients on 3 occasions over 4 y during an intervention trial with a statin.
Results: Within the entire cohort of 9014 patients, the New Zealanders had significantly (40%) more cardiovascular deaths than did the Australians. In the subgroup of 1077 patients, the New Zealanders were found at entry to have eaten significantly more total (69.34 ± 12.35 compared with 66.45 ± 12.9 g/d) and saturated (26.23 ± 8.41 compared with 24.37 ± 7.36 g/d) fat (P < 0.001 for each) and to have significantly (4%) higher concentrations of LDL cholesterol (3.96 ± 0.74 compared with 3.8 ± 0.76 mmol/L; P < 0.001) than did the Australians. At baseline, patients with previous coronary artery bypass grafting had diets that were significantly different from those of patients without previous coronary artery bypass grafting. Relations between nutrients and plasma lipids confirmed the direct effects of saturated fatty acids on LDL cholesterol and of alcohol on plasma triacylglycerol and HDL cholesterol. Dietary counseling throughout the trial led to significant improvements in compliance with guidelines. However, neither the baseline nor the improved 1-y nutrient intakes predicted future changes in cardiovascular events.
Conclusion: Differences in CHD mortality and in LDL-cholesterol concentrations between 2 populations with similar socioeconomic and cultural backgrounds were consistent with the amounts and types of fats eaten.
Key Words: Total fat saturated fat cardiovascular events LDL cholesterol population comparison clinical trial
INTRODUCTION
Differences in patterns of eating or in the consumption of specific nutrients have been found to contribute to observed differences in the prevalence of cardiovascular disease (CVD) between populations. Consumption of saturated fat appeared to correlate strongly in epidemiologic studies beginning with the Seven Countries Study (1), and this correlation was confirmed in subsequent surveys (2, 3). The increasing incidence of coronary heart disease (CHD) among migrant populations also appeared to be associated with changes in dietary habits that included higher intakes of saturated fat (4, 5). Prospective trials among large cohort populations confirmed the importance of the dietary trans and saturated fatty acids in predicting future CVD events (6, 7). From these longitudinal studies has emerged a pattern of eating that included low consumption of animal fats that correlated with less future CVD (8, 9).
Comparisons of the diet-CVD relation in countries whose populations differ only marginally in their socioeconomic and cultural characteristics have not been widely studied. Australia and New Zealand are socioeconomically and culturally similar. Both countries participated in a large placebo-controlled trial of secondary CVD prevention with the -hydroxy--methylglutaryl coenzyme A reductase inhibitor pravastatin (10). More than 9000 persons with previous acute coronary syndrome participated in the Long-Term Intervention with Pravastatin in Ischaemic Disease (LIPID) trial, which lasted 5 y and had a mean 6-y follow-up. Within an overall substantial reduction in CHD events among those treated with pravastatin, there was a significant difference between the findings in subjects from Australia and New Zealand. CHD mortality was significantly higher in those from New Zealand than from Australia, and LDL-cholesterol concentrations were, on average, significantly higher among New Zealanders than Australians (relevant data provided below). The prevalence of other major risk factors did not differ significantly between the 2 groups.
Among subjects recruited into the LIPID trial, 1077 took part in the baseline survey of eating habits that used a detailed food-frequency questionnaire (FFQ). Further surveys were conducted among this subcohort 1 y and 4 y later; the respective response rates were 89% and 84%, so that a progressive dietary history was obtained over 4 y in 849 subjects, approximately one-half of whom were treated with pravastatin. In this report, we describe the dietary characteristics at baseline in Australian and New Zealander participants in relation to the prevalence of major CVD risk factors, the nature of their initial clinical presentation, and the clinical outcome over 5 y.
SUBJECTS AND METHODS
Patients and study design
The design and results of the LIPID trial have been reported elsewhere (10). Patients (n = 9014) had a history of acute myocardial infarction or hospitalization for unstable angina pectoris and a baseline total-cholesterol concentration of 4.0–7.0 mmol/L (155–270 mg/dL). The main finding of the trial was that the risk of death due to CHD was 24% lower in those assigned pravastatin than in the placebo group. Pravastatin in the trial was supplied by Bristol Myer Squibb (Princeton, NJ). After the initial double-blind trial, an open phase was conducted, during which an increasing number of subjects were treated with pravastatin. The rates of cardiovascular events and mortality were compared between the Australian and New Zealander subjects over the study span of 8 y. A full analysis of this comparison is yet to be completed, but the method is similar to that used for the entire cohort (10). In this report, we present the key data in terms of deaths due to either CHD or to any cardiovascular event. These data were adjusted for treatment with pravastatin and for a risk score encompassing conventional risk factors as defined previously (11).
As part of the LIPID trial, patients were given dietary advice at the beginning of the run-in phase and at least every 6 mo at their routine clinic visits throughout follow-up. Advice was administered in a number of ways: via pamphlets and trial newsletters and personally by dietitians, study nurses, or physicians at the LIPID trial centers. Pamphlets supplied by the National Heart Foundation of Australia provided advice regarding foods to avoid (eg, foods high in fat such as fatty meats, full-cream dairy foods, potato chips, and take-away foods), foods to include (ie, fruit, vegetables, fish, and whole grains), guidelines for weight control, guidelines for reducing plasma cholesterol, and low-fat cooking hints. The dietary advice followed the National Heart Foundation's recommended targets of 30% of energy as fat, 55% as carbohydrate, and 15% as protein and of fatty acids derived equally from monounsaturated, polyunsaturated, and saturated fatty acids. Dietary cholesterol was recommended to be <300 mg/d. Except for dietary advice, other aspects of each patient's care were left to the discretion of his or her treating doctor.
A proportion of centers invited their patients to take part in the substudy to assess changes in dietary intake for the duration of the LIPID trial. Patients who agreed to participate were asked to complete self-administered FFQs at 3 time points: during the run-in phase, at 1 y, and at 4 y after randomization. These patients were also provided with feedback regarding success in reaching targets. This feedback was in report form and included suggestions for further dietary substitutions if necessary.
Of the 9014 patients randomly assigned to the LIPID trial, 1314 were approached to take part in the diet substudy, and 1077 agreed to participate. Nine hundred fifty-one (89% of respondents; 73% of those initially approached) and 849 (84% and 65%, respectively) patients completed subsequent questionnaires at years 1 and 4, respectively (Figure 1). A total of 794 patients who survived to year 4 completed all 3 questionnaires. One subject was lost to follow-up.
The baseline characteristics relevant to cardiovascular and nutritional outcomes did not differ significantly between the total cohort and the subgroup. Of the 1077 patients, approximately one-half had been randomly assigned to pravastatin treatment. The baseline characteristics of those assigned to receive placebo and of those assigned to receive pravastatin also did not differ significantly.
Written informed consent was obtained from all participants in the trial. The LIPID trial, including the nutritional evaluation, was approved by the human ethics committees of the participating centers.
Food-frequency questionnaire
The FFQ was a semiquantified questionnaire that was previously validated (12) and subsequently used in Commonwealth Scientific & Industrial Research Organisation Australian Food and Nutrition Surveys between 1988 and 1993. The questionnaire assessed the intake of >170 foods. Participants were asked to record how often they usually ate a specific food. They were provided with standard portion sizes for each food and asked to indicate whether they typically ate more or less than these portions. Response categories for frequency of consumption ranged from "never" or "rarely" to "many times a day." This recording system provided 10 frequency categories from "never" and "rarely" to the number of times a food was eaten daily, then weekly, and then monthly. Information regarding types of foods such as bread, milk, spreads, yogurt, and cooking oil was also obtained. Vitamin and mineral supplement use was not included in the data analyzed.
Dietitians and nurse-managers in each center explained in detail the principles and methods for completing the questionnaires. Incomplete questionnaires were sent back to the participating center for completion by the patient. A questionnaire that was returned without being complete for the second time was excluded from analysis. Nutritionists reviewed the questionnaires with respect to foods and food groups; any items with extremely unrealistic food intakes were reviewed and excluded if no justification for that intake was found in the comments or otherwise.
The robustness of the FFQ was ensured by care in design of the questionnaire, the inclusion of many foods, the setting of an order of questions that allowed ease of answering, the request for information on choices within food groups, qualitative measures of food preparation, and flexibility in serving sizes. All these issues are addressed in this questionnaire (12).
Data analysis
Dietary nutrient intake was calculated by multiplying the frequency of consumption (servings/wk) by the size of the servings consumed and the nutrient composition, with adjustments for type of cooking fats used, meat trimming, type of fats spread, dairy products and bread consumed, and cooking methods and salting habits. Nutrient composition was primarily derived from the Australian NUTTAB92 food-composition database (13), which was supplemented with British food data (14)
The current report presents the nutrient data that are most relevant to CVD risk factors and hence to clinical outcomes. Nutrient intake data are expressed as energy-adjusted grams per day (15).
Assessment of endpoints
An endpoint expanded beyond that used in the LIPID trial was used for the diet substudy to increase the power of the study. The endpoint was a composite of cardiovascular events including coronary artery bypass graft (CABG) or percutaneous coronary intervention, fatal and nonfatal acute myocardial infarction, sudden death, and death due to CVD. Full details about the measurement and assessment of outcomes are provided elsewhere (10).
Statistical analysis
All analyses were performed on an intention-to-treat basis. Exploratory data analysis including the production of histograms was undertaken to assess the distribution of each nutrient and lipid fraction. All variables were shown to be approximately normally distributed. Means and standard deviations have been presented for continuous data. Categorical data have been presented as frequencies and percentages. Differences in baseline characteristics between countries (Australia and New Zealand) and between clinical status (CABG or no CABG) were tested by using the two-sample t test or the chi-square test.
The effects of obesity [body mass index (BMI; in kg/m2): >30] and diabetes on nutrient intake were analyzed with a 3-factor (country, obesity, and diabetes) factorial analysis of variance. Significant (P < 0.05) interactions and main effects for obesity and diabetic status are reported. These analyses were carried out after first ensuring that there were no interactions between country and obesity or diabetes. The strengths of the relations between nutrient and lipid fractions were assessed by using Pearson's correlation coefficients. Correlations were calculated for each country separately and tested for homogeneity. Weighted average correlations between countries were presented.
Differences in mean nutrient intakes over time were analyzed by using repeated-measures analysis of variance with 3 time points. When a significant (P < 0.05) change in nutrient intake was found by using the Greenhouse-Geiser F test, the differences between each time point were examined by using t tests with a Bonferroni correction. Differences in clinical outcomes between countries were examined by using time-to-event analysis involving Cox regression. Time-to-event analyses were performed by using Cox regression, in which the nutrient intake was treated as a categorical variable (quintiles). A test of trend across the quintiles was undertaken. Univariate analysis for each nutrient was conducted.
P values were two-sided. Statistical analyses were performed with SAS for WINDOWS software (version 5.0.2195, Release 8.02; SAS Institute Inc, Cary, NC) and SPSS software (version 11.5; SPSS Inc, Chicago, IL).
RESULTS
Baseline characteristics and nutrient intakes by country
A major aim was to ascertain whether differences existed between the characteristics of the trial patients living in Australia and New Zealand. The differences in the average lipid profile of the 2 populations are shown in Table 1. New Zealanders showed a significantly (4%) higher mean plasma concentration of LDL cholesterol and a significantly lower concentration of HDL cholesterol than did Australians. It may be important that there were no significant differences between the populations in the prevalence of obesity or the number of smokers, although there was a trend for greater incidence of diabetes among Australians than among New Zealandres (9% and 6%, respectively; P = 0.07).
Also shown in Table 1 is macronutrient consumption at the time the lipid measurements were made. New Zealanders consumed significantly more energy (8.5%) and total fat (7.6%) derived from both saturated (7.6%) and monounsaturated (3.5%) fatty acids than did Australians. They also ate 2% less carbohydrate than did Australians. Sodium consumption did not differ significantly between the 2 populations.
Major clinical outcomes for the entire cohort by country
At the end of both the double-blind and open-label (total time: 8 y) phases of the LIPID trial, the incidence of the primary endpoint, CHD mortality, was significantly higher in New Zealand (12.8%) than in Australia (8.6%) (Table 2). This significant difference was maintained after adjustment for baseline risk factors [ie, risk score (11)] and for treatment with pravastatin, ie, a hazard ratio (HR) of 1.42 (95% CI: 1.25, 1.63). Similarly, the risk of death due to CVD, after adjustment, was 40% (95% CI: 23%, 58%) higher in New Zealand than in Australia.
Baseline nutrient intake according to clinical status at entry
We next examined whether macronutrient consumption varied among patients enrolled with different previous manifestations of CHD. It is important to note that the trial included patients who had had acute myocardial infarction or had been hospitalized for unstable angina pectoris between 3 mo and 3 y previously. Other manifestations such as coronary revascularization might have occurred either during this time frame or beforehand, and that might have allowed patients in specific categories or patients who had known about their illness for various lengths of time to make dietary adjustments.
It is interesting that the 342 patients who underwent CABG were found at their initial visit to have a diet that was significantly different from that of the other patients: ie, the CABG patients consumed 3% less total fat, 7.8% less saturated fatty acids, and 7.8% less cholesterol than did the non-CABG patients. The data presented in Table 3 therefore merely define the pattern of nutrient intake of this particular group of patients at entry, which probably reflects these reductions in consumption of meat, eggs, and full-cream dairy products. The CABG patients ate 7.8% more fiber than did the 735 non-CABG patients, and they also had significantly (P < 0.05) lower LDL-cholesterol concentrations—3.74 ± 0.74 compared with 3.86 ± 0.76 mmol/L—than did the non-CABG patients.
Baseline nutrient intake among obese subjects and patients with type 2 diabetes
Also of interest was the pattern of eating in the subjects who were known to be at greater risk because they were also obese or diabetic. Before calculations of significant differences, the country obesity x diabetes interactions were tested and found to be nonsignificant for all nutrients except cholesterol (P = 0.014). The significant data are shown in Table 4. The obese subjects ingested significantly more fat from saturates and monounsaturates, significantly more protein, and significantly less carbohydrate than did the nonobese subjects. The diabetic patients ate significantly more total fat, monounsaturated fat, and protein and significantly less carbohydrate than did the nondiabetic subjects. The obese and diabetic patients therefore ate relatively more animal foods than plant-based foods. The differences shown in Table 4 are independent of country, although, in each category, the differences seen between countries for the entire cohort (Table 1) were also reflected in the 4 subgroups. The differences between countries were therefore the same for diabetic and nondiabetic subjects and for obese and nonobese subjects.
Relations between nutrients and lipid fractions
The correlation coefficients of nutrient intakes at baseline as predictors of plasma lipid concentrations based on continuous nutrient consumption data are shown in Table 5. The following relations were significant, but the correlations were of a low order. Total cholesterol was inversely related to energy and carbohydrate intake and directly related to alcohol intake (which may reflect the effect on HDL cholesterol); LDL cholesterol tended to be directly related to the intake of saturated fatty acid and inversely related to that of energy. Plasma triacylglycerol concentration was significantly related to the intake of alcohol (positively) and fiber (negatively), and HDL cholesterol was inversely related to energy and carbohydrate intakes and positively related to alcohol consumption.
Changes in nutrient intake during trial
We also examined the likelihood that a change in eating habits during the trial as the result of more extensive counseling might also have influenced the predictive value of the patterns of nutrient intake for clinical outcomes. With the exception of monounsaturated fatty acid intake (ie, patients receiving pravastatin consumed 3% less than did those receiving placebo; P = 0.04), there were no nutrient x treatment interactions and therefore the whole group was analyzed together. The changes in nutrient intake over the 4 y are shown in Table 6 by using the data collected at baseline and at years 1 and 4. Significant but modest changes occurred during this time; most occurred within the first year, except for alcohol consumption, which did not change. Moderate reductions occurred in total energy from most sources of macronutrients: total fat, all 3 groups of fatty acids, carbohydrate, and protein. Cholesterol and fiber consumption also decreased. Overall, >80% of subjects achieved the target for cholesterol intake (<250 mg), and 51% and 50% achieved the target for the consumption of saturated fat (<10% of energy) and total fat (<39% of total energy), respectively.
Clinical outcomes related to nutrient intake at baseline
There were no significant relations between clinical outcomes and the pattern of nutrient consumption obtained at baseline (5 y before) and tested as quintiles. There were no significant trends; all values were P > 0.12 (HRs from Cox regression). Because this may have partly reflected the significant dietary changes made, on average, over that period, as shown in Table 6, further analysis of the consumption of nutrients at year 1 in comparison with clinical outcomes was carried out. However, this analysis also failed to show any significant relations.
DISCUSSION
Cardiovascular mortality, nutrient consumption, and plasma lipids by country
The LIPID trial of patients with existing CVD recruited 5958 Australians and 3056 New Zealanders between 1990 and 1992, and clinical outcomes were initially assessed in 1997 (10). The outcomes for the entire cohort—ie, deaths due to CHD and CVD—were significantly worse among New Zealanders, who had a mean HR of 1.42 (95%CI: 1.25, 1.61) than among Australians (Table 2). Although the effects of conventional risk factors and treatment were considered in that analysis and found not to have influenced the differences in mortality, differences in dietary factors in a subgroup of 1077 subjects are explored in this report.
The pattern of macronutrient consumption at baseline did differ significantly between the 2 national populations in several important respects (Table 2). New Zealanders consumed slightly more energy and total fat (even after adjustment for energy), from both saturated and monounsaturated fatty acids than did the Australians; they also ate slightly less carbohydrate. New Zealanders also had a significantly (4%) higher mean LDL-cholesterol concentration than did the Australians, and their intake of saturated fatty acids correlated significantly, albeit weakly, with the LDL-cholesterol concentration on univariate analysis (Table 5). The difference in dietary patterns between the 2 populations might therefore have contributed to the differences in clinical outcomes observed for the entire LIPID trial cohort (10).
It has previously been difficult to show within Western populations a relation between total fat consumption and CVD. This difficulty has been attributed partly to the generally high intake of fat and the counteractive effects of saturated and trans fatty acids on the one hand and of unsaturated fatty acids on the other (9). It is therefore important to note that Australia and New Zealand are socioeconomically and culturally similar. On the other hand, differences between these populations in CVD that appear partly to reflect different dietary habits, especially with respect to the consumption of saturated fat, have been reported frequently (4, 5). Similar conclusions have been drawn in long-term prospective studies from the patterns of eating that were associated with the least and the greatest risk for future CHD events; the consumption of fatty meats, full-cream dairy products, and fried foods comprised the higher-risk pattern (8, 9). Consumption of meat and dairy products has traditionally been higher in New Zealand than in Australia, and that difference may largely explain the differences in fat intake and LDL-cholesterol concentration between the 2 populations. The LIPID trial was conducted against a background of falling CHD mortality rates in both countries. Of the 23% decline between 1982 and 1993 in New Zealand, one-half has been attributed to reductions in major risk factors [ie, 12% of the decline was due to lower serum cholesterol concentrations (16)], although dietary data were not considered. Yet the patients recruited from New Zealand during 1990 and 1992 were still eating foods significantly richer in saturated fatty acids than were the patients in the Australian cohort.
Nutrient intakes at entry into trial and during the subsequent 4 y
The profiles of dietary patterns at baseline and during the first 4 y of the trial showed that one-third of all patients with known CVD were following the key recommendations of the National Heart Foundation of Australia, which resemble those of the American Heart Association (17). The mean (±SE) intake of total fat for the subgroup was 29.8 ± 0.2% of energy, which is comparable to the 35% found in a 1995 national survey of Australian men of similar age to the subjects in the LIPID trial (18).
Both Australians and New Zealanders in the study changed their eating habits in years 1 through 4 (Table 6). At 1 y, >80% of subjects achieved the target for reducing cholesterol intake (10%), and 51% and 50% achieved the targets for reducing fat and total fat intakes (11% and 9%, respectively).
Two additional subgroups were deemed worthy of further investigation: subjects who were obese and subjects who had type 2 diabetes. At that time (1990–1992), both of those groups ate significantly more fat and protein but significantly less carbohydrate than did the nonobese and nondiabetic subjects, respectively (Table 4). In retrospect, this is of concern because higher consumption of saturated fat and cholesterol has been linked to a greater risk of CHD in diabetic as well as nondiabetic subjects (19).
Relations between nutrients and lipid fraction
These relations are consistent with the widely reported effects of nutrients on the 4 major plasma lipid fractions (9). The predictive effects of nutrients either by quintile of intake or as continuous nutrient consumption were highly significant, although the correlations were of a low order (Table 5). Saturated fat intake was directly related to LDL cholesterol, carbohydrate intake was inversely related to HDL cholesterol, and alcohol consumption increased both the plasma triacylglycerol and HDL-cholesterol concentrations. Although anticipated, these findings provide support for the robustness of the questionnaire and the adequacy of the size of the diet-study subgroup.
Clinical outcomes and study limitations
There was no evidence of an effect of nutrient intakes at baseline on clinical outcomes 5 y later in the diet-study subgroup. Dietary changes made mainly in year 1 by at least one-half of all patients, however, included eating less fat, saturated fat, and cholesterol, but an analysis testing the relation of clinical outcomes to year 1 intakes was also negative. The total number of clinical events (ie, 329) was probably inadequate to identify any existing relations. Nevertheless, a Cochrane review did come to a guarded conclusion that a reduced or modified fat intake, if followed for a period of 2 y, results in a small but potentially important reduction in the risk of cardiovascular events (20). Although Australians and New Zealanders differed in CVD mortality, nutrient intakes, and plasma lipids at baseline, the reductions in total fat and saturated fatty acids were not nearly as great in this trial as in dietary intervention trials of primary (21, 22) and secondary (23, 24) prevention in which a large exchange of polyunsaturated for saturated fatty acids led to significant reductions in initial or recurrent nonfatal CHD or subsequent cardiovascular events. On the other hand, more modest changes in diet, such as adoption of a diet based on Mediterranean tradition, when made in the context of other lifestyle changes, have been shown to lower the rate of CHD mortality (25). A further possible diluting factor was the inclusion of pravastatin-treated patients in the analysis, although no change in nutrient intake over time x assigned treatment interaction was found. Events in the placebo group alone were too few. The characteristics of the participants were similar to those of the entire cohort, but only 1077 of the 1314 patients who were approached to take part in the diet study agreed, which raised the possibility of selection bias. Finally, the nutrition data are only as robust as the FFQ method, which relies on the capacity of a person to recollect accurately his or her patterns of intake over time. However, FFQs have been applied extensively and successfully in many prospective trials and case-control investigations of average food and nutrient consumption by individuals.
ACKNOWLEDGMENTS
Statistical analyses were performed at the National Health and Medical Research Centre Clinical Trials Centre, University of Sydney, by CP, AK, and JS. PN, AT, HW, and DC have been members of the LIPID Management Committee, which has supervised various aspects of the trial, since its inception. AMT is the Medical Director of the National Heart Foundation of Australia. None of the authors had a personal or financial conflict of interest.
REFERENCES
Keys A, Menotti A, Karvonen MJ, et al. The diet and 15-year death rate in the Seven Countries Study. Am J Epidemiol 1986;124:903–15.
Shekelle RB, Shryock AM, Paul O, et al. Diet, serum cholesterol, and death from coronary heart disease. The Western Electric Study. N Engl J Med 1981;304:65–70.
Hu BH, Stampfer MJ, Manson JE, et al. Dietary fat intake and the risk of coronary heart disease in women. N Engl J Med 1997;337:1491–9.
Kushi LH, Lew RA, Stare FJ, et al. Diet and 20-year mortality from coronary heart disease. The Ireland-Boston diet-heart study. N Engl J Med 1985;312:811–8.
Kato H, Tillotson J, Nichamen MZ, Rhoads GG, Hamilton HB. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California: serum lipids and diet. Am J Epidemiol 1973;97:372–85.
Hu FB, Stampfer MJ, Manson JE, et al. Dietary saturated fats and their food sources in relation to the risk of coronary heart disease in women. Am J Clin Nutr 1999;70:1001–8.
Siscovick DS, Raghunatahan E, King I, et al. Dietary intake and cell membrane levels of long-chain n–3 polyunsturated fatty acids and the risk of primary cardiac arrest. JAMA 1995;274:1363–7.
Kerver JM, Yang EJ, Bianchi L, Song WO. Dietary patterns associated with risk factors for cardiovascular disease in healthy adults. Am J Clin Nutr 2003;78:1103–10.
Hu FB, Willett W. Optimal diets for prevention of coronary heart disease. JAMA 2002;288:2569–78.
Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and broad range of initial cholesterol levels. The Long-term Intervention with Pravastatin in Ischaemic Disease Study group. N Engl J Med 1998;339:1349–57.
Marschner IC, Colquhoun D, Simes RJ, et al. Long-term risk stratification for survivors of acute coronary syndromes. J Am Coll Cardiol 2001;38:56–63.
Rohan T, Record S, Cook M. Repeatability of estimates of nutrient and energy intakes. The quantitative food frequency approach. Nutr Res 1987;7:15–137.
Lewis J, Holt RJ. Nutrient data table for use in Australia 1991-92. National Food Authority. Canberra, Australia: AGPS, 1991.
Ministry of Agriculture, Fisheries and Food. Food portion sizes. 2nd ed. London, United Kingdom: Her Majesty's Stationery Office, 1993.
Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986;124:17–27.
Capewell S, Beaglehole R, Seddon M, McMurray J. Explanation for the decline in coronary heart disease mortality rates in Auckland, New Zealand, between 1982 and 1993. Circulation 2000;102:1511–6.
American Heart Association. Dietary guidelines for healthy American adults: a statement for physicians and health professionals by the Nutrition Committee, American Heart Association. Circulation 1988;77:721A–4A.
McLennan W, Podger A. A national nutritional survey: nutrient intakes and physical measurements. Australia, 1995. Canberra, Australia: Australian Bureau of Statistics, 1998.
Tanasescu M, Cho E, Manson JE, Hu FB. Dietary fat and cholesterol and the risk of cardiovascular disease among women with type s diabetes. Am J Clin Nutr 2004;79:999–1005.
Hooper L, Summerbell CD, Higgins JPT, et al. Reduced or modified dietary fat for preventing cardiovascular disease. Cochrane Database Syst Rev 2001;(3):CD002137.
Turpeinen O, Karvonen MJ, Pekkarinen M, Miettinen M, Elosuo R, Paavilainen E. Dietary prevention of coronary heart disease: the Finnish mental hospital study. Int J Epidemiol 1979;8:99–117.
Hjermann I, Velve Byre K, Holme I, Leren P. Effect of diet and smoking on the incidence of coronary heart disease. Lancet 1981;2:1303–10.
Leren P. The Oslo diet-heart study. Eleven year report. Circulation 1970;42:935–42.
Dayton S, Pearce ML, Hashimoto S, Dixon WJ, Tomiyasy U. A controlled clinical trial of a diet high in unsaturated fat in preventing complications of atherosclerosis. Circulation 1969;40:S1–63.
Knoops KTB, de Groot LCPGM, Kromhout D, et al. Mediterranean diet, lifestyle factors and 10-year mortality in elderly European men and women. JAMA 2004;292:1433–9., 百拇医药(Paul J Nestel, Katrine Ba)