Customize your JAMA Network experience by selecting one or more topics from the list below.
Get the latest research based on your areas of interest.
Åkesson A, Weismayer C, Newby PK, Wolk A. Combined Effect of Low-Risk Dietary and Lifestyle Behaviors in Primary Prevention of Myocardial Infarction in Women. Arch Intern Med. 2007;167(19):2122–2127. doi:10.1001/archinte.167.19.2122
Limited data are available on the benefit of combining healthy dietary and lifestyle behaviors in the prevention of myocardial infarction (MI) in women.
We used factor analysis to identify a low-risk behavior–based dietary pattern in 24 444 postmenopausal women from the population-based prospective Swedish Mammography Cohort who were free of diagnosed cancer, cardiovascular disease, and diabetes mellitus at baseline (September 15, 1997). We also defined 3 low-risk lifestyle factors: nonsmoking, waist-hip ratio less than the 75th percentile (< 0.85), and being physically active (at least 40 minutes of daily walking or bicycling and 1 hour of weekly exercise).
During 6.2 years (151 434 person-years) of follow-up, we ascertained 308 cases of primary MI. Two major identified dietary patterns, “healthy” and “alcohol,” were significantly associated with decreased risk of MI. The low-risk diet (high scores for the healthy dietary pattern) characterized by a high intake of vegetables, fruit, whole grains, fish, and legumes, in combination with moderate alcohol consumption (≥ 5 g of alcohol per day), along with the 3 low-risk lifestyle behaviors, was associated with 92% decreased risk (95% confidence interval, 72%-98%) compared with findings in women without any low-risk diet and lifestyle factors. This combination of healthy behaviors, present in 5%, may prevent 77% of MIs in the study population.
Most MIs in women may be preventable by consuming a healthy diet and moderate amounts of alcohol, being physically active, not smoking, and maintaining a healthy weight.
Coronary heart disease (CHD) is the most important cause of death and disability in women.1,2 Despite a lower incidence in women, CHD-related mortality3 and the percentage of sudden deaths from CHD without previous symptoms2 is higher and the trend of decline in incidence is slower4 than in men. Thus, CHD risk factor characterization and prevention in women need improvement.1 Regardless of predisposing factors or proved benefits of pharmacologic therapies, diet and lifestyle largely influence morbidity and mortality in CHD.1,5-9 There is little information on the benefit achieved with a combination of several healthy lifestyle behaviors, but available data indicate a large reduction in CHD incidence6 and mortality.7
The dietary pattern concept, similar to the diet-based intervention trials, reflects the actual eating behavior in the population and, thus, mirrors the joint nutritional exposures more closely than analyses of individual foods or single nutrients.10,11 A few existing studies indicate that dietary patterns, such as those identified using factor analysis, may predict the risk of CHD.12,13 However, a more comprehensive approach to CHD prevention can be achieved by combining the dietary pattern method of identifying a low-risk diet with other major low-risk modifiable lifestyle factors.
We examined the benefit of a combined healthy diet and lifestyle including low-risk dietary behavior (identified by factor analysis) and 3 major low-risk lifestyle factors, that is, avoiding smoking and abdominal adiposity and being physically active, on the risk of primary myocardial infarction (MI) in a population-based prospective cohort of Swedish women. We also estimated the burden of CHD that could potentially be avoided if all women adhered to this healthy lifestyle.
In 1997, women born between 1914 and 1948, living in 2 counties in central Sweden, who participated in a survey in 1987 (74% of the source population) and were still alive, received a follow-up questionnaire on diet and lifestyle (70% participation rate). From the 1997 cohort of 38 984 women who responded to the questionnaire, we excluded those with previous ischemic heart disease, cardiovascular diseases, and malignant neoplasms before baseline in 1997, identified by computerized linkage to the Swedish Hospital Discharge Registry and the Swedish Cancer Registry, and those with self-reported diabetes mellitus (n = 4913). We also excluded women with implausible values for energy intake (ie, 3 SDs from the mean value for logarithmically transformed energy; n = 407) and those who answered less than 75% of the food frequency questions (n = 9220). Thus, the study population included 24 444 women, aged 48 to 83 years, at the start of follow-up. The study was approved by the Regional Ethical Review Board at Karolinska Institutet, Stockholm, Sweden.
The Food Frequency Questionnaire contained questions about 96 commonly eaten foods, primarily assessed using predefined frequency consumption categories.14 For energy and nutrient calculations, we used age-specific portion sizes based on mean values of 5922 days of weighed food records kept by 213 randomly selected women. We adjusted nutrients for total energy intake (1700 kcal/d [to convert to kilojoules, multiply by 4.186], mean in the study population) using the residual method.
We used factor analysis to derive dietary patterns empirically. Factor analysis reduces dietary data to a few composite factors that describe the eating pattern in the population.10 The validity of Food Frequency Questionnaire–based patterns compared with food record–based ones, assessed by Spearman correlation coefficient (r), varied between 0.50 and 0.86 (for the major pattern, healthy, r = 0.59).15 The 1-year reproducibility of the patterns varied between r = 0.63 and r = 0.73 (for the healthy pattern, r = 0.63)15 and the stability during 4 to 6 years varied between r = 0.54 and r = 0.76 (for the healthy pattern, r = 0.63).16 We performed exploratory factor analysis to extract patterns that we then confirmed using confirmatory factor analysis.14 To avert subjective influences in food grouping,10 we included all individual food items in the exploratory factor analysis (Intercooled Stata, version 8.2; STATA software; StataCorp LP, College Station, Texas). We considered eigenvalues greater than 1.0, interpretability of factors, and number of items and their frequency to decide how many factors to extract from the data and confirm. We included items with factor loadings of 0.20 or greater from exploratory analysis to test specific factor structures using confirmatory factor analysis (LISREL software; Scientific Software International, Inc, Lincolnwood, Illinois); the goodness-of-fit index was high (0.93 for the model including all patterns). Factor scores were calculated for each individual for each pattern by weighting the standardized intakes of the food items by their factor loadings and summing for all items. The scores of each dietary pattern were categorized into quintiles.
We obtained self-reported information on level of educational achievement, family history of MI in parents or siblings before the age of 60 years, presence of high cholesterol levels and hypertension, use of hormone therapy and aspirin, history of smoking, and waist and hip circumferences. Physical activity was assessed with a validated questionnaire17 that included questions on time spent walking or bicycling (6 predefined duration categories, as follows: hardly ever, < 20 minutes, 20-39 minutes, 40-59 minutes, 1-1½ hours, or > 1½ hours per day) and leisure time exercise (5 predefined duration categories, as follows: < 1 hour, 1 hour, 2-3 hours, 4-5 hours, or > 5 hours per week).
We considered smoking status, abdominal adiposity, and level of physical activity as the major modifiable nondietary risk factors to be included in the a priori definition of a low-risk practice.6,18-20 Thus, women who never smoked or had stopped smoking (at least 1 year), who had a waist-hip ratio less than the 75th percentile (< 0.85), and who walked or bicycled for at least 40 minutes per day and exercised at least 1 hour per week composed the low-risk group. Women with missing information for any of the 3 lifestyle behaviors were placed in the high-risk category to give the most conservative estimate. However, excluding these women did not appreciably change the results.
The cohort was linked to the Swedish Hospital Discharge Registry and the Cause of Death Registry, considered more than 99% complete, and all incident cases of nonfatal and fatal MI (International Statistical Classification of Diseases, 10th Revision code I21) were ascertained from baseline (1997) through December 2003. The diagnostic criteria were as follows: (1) specified changes in blood levels of troponin occurring 2 times or more in addition to either specified symptoms or specified electrocardiographic changes (eg, new pathologic Q-wave or pathologic ST-segment elevation or inversion), (2) specified symptoms and ST-segment elevation with no further possibility of characterization, or (3) autopsy findings showing myocardial necrosis or coronary thrombosis of an age compatible with the time of disease onset. A thorough validation of the registries for 1987 and 1995 revealed high sensitivity (94%) and positive predictive value (86%) for MI in comparison with data from other countries.21 Dates of death were ascertained through the Swedish Death Registry at Statistics Sweden, Stockholm.
Univariate association was assessed using the Pearson correlation coefficient (r). We estimated relative risk and 95% confidence intervals using Cox proportional hazards regression models. Women were censored at date of the first event of MI, death, or end of follow-up, whichever occurred first. Multivariate associations between dietary patterns and risk of MI were simultaneously adjusted for all dietary patterns, age (continuous), educational achievement (< 9, 9-12, and > 12 years), family history of MI in parents or siblings before age 60 years (yes/no), high cholesterol levels (yes/no), presence of hypertension (yes/no), use of hormone therapy (never/ever), use of aspirin (yes/no), waist to hip ratio (quartiles), smoking history (current vs never and former), physical activity (≥ 40 min/d of walking or bicycling and exercise ≥ 1 h/wk; yes/no), and energy intake (continuous). Tests of linear trend were conducted by including the median dietary pattern score for each quintile as a continuous variable. Assuming there was a causal relation between risk factors and MI, we calculated the population-attributable risk to estimate the percentage of cases that potentially could be prevented had all women adhered to low-risk diet and lifestyle behaviors.
During a mean of 6.2 years (151 434 person-years) of follow-up, we ascertained 308 incident cases of primary MI. Fifty-one of these cases were fatal.
We derived 4 major dietary patterns: “healthy” (vegetables, fruit, and legumes), “Western/Swedish” (red meat, processed meat, poultry, rice, pasta, eggs, fried potatoes, and fish), “alcohol” (wine, liquor, beer, and some snacks), and “sweets” (sweet baked goods, candy, chocolate, jam, and ice cream).14 Characteristics of the healthy dietary pattern are given in Table 1. There were no major differences in nondietary risk factor distributions with increasing healthy pattern scores, with the exception that women in the top quintile were less likely to be smokers, were more likely to have attended college, and were more physically active compared with those in the lowest quintile. For dietary characteristics, women in the highest quintile of the healthy pattern had an almost 4-fold higher weekly consumption of vegetables and fruits, 3-fold higher consumption of legumes, 70% higher consumption of fish, more than 2-fold intake of vitamin C, and about 50% higher intake of dietary fiber and folate compared with those in the lowest quintile. This variation in food consumption was reflected as direct correlations between the healthy pattern score and the intakes of dietary fiber (r = 0.54), folate (r = 0.67), vitamin E (r = 0.45), and vitamin C (r = 0.60) and as inverse correlations with total fat (r = −0.29) and saturated fat (r = −0.32) (for all, P < .001).
Table 2 gives the estimates of the relative risk of MI for each of the 5 modifiable lifestyle factors considered in further analyses of the low-risk behavior and the percentage of the cohort in each risk category. Each lifestyle factor was inversely and, after adjustment for the other elements of the low-risk profile, independently associated with the risk of coronary events. The healthy dietary pattern and the alcohol dietary pattern scores were statistically significantly associated with the risk of primary MI. In the full multivariate-adjusted model, the lowest quintile of the healthy dietary pattern was associated with 71% increased risk compared with the highest quintile (Pfor trend = .004) (Table 2). In the lowest quintile of the alcohol dietary pattern, the relative risk of MI was 1.64 (95% confidence interval, 1.09-2.47) compared with the highest quintile (Pfor trend = .002). There were no associations between the Western/Swedish pattern (Pfor trend = 0.99) or the sweets pattern (Pfor trend = .14) and risk of MI. Exclusion of 51 fatal cases did not appreciably change the estimates for any of the 4 patterns (data not shown).
We defined a low-risk dietary behavior based on high scores for the healthy dietary pattern (quintiles 3-5). For simplicity, we replaced the alcohol dietary pattern with alcohol consumption as a discrete variable. The low-risk alcohol group comprised those who consumed a moderate amount (5 g) of alcohol per day or more6; the median intake in this group was 10 g/d (Table 2). We did not define an upper limit for alcohol consumption because few women consumed high amounts of alcohol; less than 0.3% reported drinking more than 45 g/d. The combined low-risk diet and the moderate alcohol group was associated with 57% (95% confidence interval, 37%-70%) reduction in the risk of MI compared with the reference group (those with both low healthy dietary pattern score [quintiles 1-2] and low alcohol consumption [< 5 g/d of alcohol]). This combined low-risk dietary and alcohol group, which included 29% of the study population, was characterized by a median daily consumption of 4.1 servings of vegetables, 2 servings of fruits and berries, and 3.4 servings of whole grains, and weekly consumption of 2.5 servings of fish, 0.5 servings of legumes, and 2.5 servings of alcoholic beverages (8.7 g/d of alcohol).
In the final analysis, we investigated the combined effect of the low-risk factors and created a comprehensive low-risk profile by combining the low-risk diet and moderate alcohol consumption with the 3 independent low-risk nondietary modifiable lifestyle factors (no smoking, high physical activity, and low abdominal adiposity) (Table 3). The final comprehensive low-risk profile, including all 5 low-risk factors, was associated with 92% (95% confidence interval, 72%-98%) lower risk of MI compared with the high-risk group (women who smoked and had abdominal adiposity, were less physically active, and had low scores for the healthy pattern and low alcohol consumption). The population-attributable risk in the study population, in which only 5% of the women fulfilled the criteria for the comprehensive low-risk behavior, was 77%, which suggests that more than 3 of 4 of the coronary events could potentially be averted if all women would change their behavior to the low-risk profile.
In this prospective cohort of postmenopausal women, we observed that a low-risk dietary behavior characterized by high consumption of vegetables, fruits, whole grains, fish, and legumes and moderate consumption of alcoholic beverages was associated with a statistically significant 57% risk reduction of primary MI. The women combining the low-risk diet and moderate alcohol consumption with such low-risk lifestyle behaviors as not smoking, being physically active, and avoiding overweight had a 92% lower risk. The combined benefit of diet, lifestyle, and healthy body weight may prevent more than 3 of 4 cases of MI in our study population.
Dietary pattern analysis is a robust technique inasmuch as all 4 dietary patterns have been reproduced across populations.10,14 Our patterns showed good validity and were stable over a period corresponding to the duration of our follow-up.15,16 Several constituents of vegetables, legumes, fruit, and whole grains, that is, dietary fiber, antioxidant vitamins, minerals, and phytonutrients, which characterized the healthy dietary pattern, have been linked to reduced risk of CHD.5 We observed a somewhat stronger inverse association between the healthy dietary pattern and risk of MI (43% reduced risk in the highest quintile) than that observed for the corresponding quintile of the “prudent” dietary pattern and risk of CHD in a US cohort of women (24% reduced risk).13 About one-third reduction in primary CHD incidence is also reported after interventions with cholesterol-reducing drugs (statins).22 Our results of alcohol, either as a dietary pattern or as a discrete variable, agreed with previous findings on alcohol and CHD incidence23,24 and atherosclerosis progression25 in women who consumed similar amounts of alcohol.
Results based on the dietary pattern approach, which resembles the total diet approach often used in dietary intervention trials and reflects actual eating behavior in the population,10,11,26 may be easy to communicate as dietary recommendations in CHD prevention. This approach is in contrast to previous studies of diets constructed on the basis of scores of specific nutrients and dietary factors or foods considered protective against CHD.6,7,9
The estimated population-attributable risk for CHD incidence resulting from the behavior-based definition of low-risk diet combined with other low-risk lifestyle factors indicates that much of the burden of CHD could be reduced by changes in modifiable lifestyle behaviors. Our results agree well with population-attributable risk in a population of women in the United States when analyses were based on a combined a priori defined low-risk diet (a composite diet score), moderate alcohol consumption, daily exercise, and absence of smoking and overweight.6 Furthermore, our results are similar to those obtained for CHD-related mortality in a small study in European women and men with a high Mediterranean diet score, moderate alcohol consumption, high physical activity, and absence of smoking.7 The strength of association of the 5 low-risk behaviors in our study is comparable to that observed for CHD mortality in young women with a clinical definition of a low-risk profile based on favorable levels of blood pressure and serum cholesterol and absence of diabetes mellitus, smoking, and overweight.27,28
The strengths of our study include a prospective population-based design and the availability of detailed data on diet as well as potential risk factors for CHD. The prospective design prevents recall bias, and the practically complete follow-up of the study population through linkages to various population-based registers minimized the concern that our findings were affected by differential loss to follow-up.
Several potential limitations may be associated with this analysis. Measurement error in self-reports is inevitable, which can lead to misclassification of exposure. However, misclassification in our study is most likely nondifferential because of the prospective design, which may lead to attenuation of the true association. Although our cohort was large, the study is limited by the low number of cases in the low-risk group, leading to somewhat imprecise estimates. We also had too few fatal cases to evaluate associations between the low-risk diet and lifestyle and CHD mortality. That the incidence showed a clear graded decline with increasing adherence to a low-risk practice supported the consistency of our findings, and we had adequate power to detect statistically significant associations. Generalizability of our results might be somewhat limited by exclusion of many women with missing information on frequency of consumption. Nevertheless, our results suggest that important steps can be taken to significantly reduce the risk of primary nonfatal CHD, although precaution is needed when deriving more stringent recommendations based on the results. Randomized trials are clearly preeminent to establish causality between both diet and lifestyle and CHD risk, but because long-term randomized trials are difficult to perform for multiple risk factors,29 the combined low-risk behavior may never be able to be tested in primary prevention trials. Therefore, prevention should build on best available information.
Our study findings indicate that healthy dietary behaviors are present in the population. These dietary behaviors together with a healthy lifestyle and body weight may prevent most MI events.
Correspondence: Agneta Åkesson, PhD, MPH, Division of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Box 210, 17177 Stockholm, Sweden (Agneta.Akesson@ki.se).
Accepted for Publication: June 18, 2007.
Author Contributions: Dr Åkesson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Åkesson, Weismayer, and Wolk. Acquisition of data: Wolk. Analysis and interpretation of data: Åkesson, Weismayer, Newby, and Wolk. Drafting of the manuscript: Åkesson. Critical revision of the manuscript for important intellectual content: Åkesson, Weismayer, Newby, and Wolk. Statistical analysis: Åkesson, Weismayer, Newby, and Wolk. Obtained funding: Åkesson and Wolk. Administrative, technical, and material support: Wolk. Study supervision: Wolk.
Financial Disclosure: None reported.
Funding/Support: This study was supported by research grants from the Center for Health Care Sciences, Karolinska Institutet; the Swedish Research Council/Medicine and Longitudinal Studies; and the Swedish Council for Working Life and Social Research.
Role of the Sponsors: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.
Create a personal account or sign in to: