Diet, Lifestyle, and Longevity—The Next Steps?

Nearly 50 years ago, Keys^{1 - 2} recognized the enormously divergent rates of heart disease around the world, even after adjusting for differences in age. Although coronary disease was and remains the leading cause of death in the United States and many developed and developing countries, it was almost nonexistent in the traditional cultures of Crete and Japan.¹ Rates of cancer at various sites also differ enormously—up to 100-fold—in different populations.³ The rapid changes in rates of many of these diseases over time and studies that show increases in chronic disease rates among migrants from traditional to Westernized cultures demonstrate that relatively swift changes in disease rates cannot be attributed solely to genetic differences between populations.⁴ Instead, they are likely due to differences in lifestyle, with dietary factors and physical activity the leading candidates.

In this issue of JAMA, Knoops and colleagues⁵ from the Netherlands, France, Spain, and Italy show that in European men and women aged 70 through 90 years adherence to a Mediterranean diet pattern, moderate alcohol consumption, nonsmoking status, and physical activity each were associated with a lower rate of all-cause mortality. Taken together, the combination was associated with a mortality rate of about one third that of those with none or only one of these protective factors. These healthful behaviors are not extreme: for example, the physical activity criterion could be met by half an hour of walking daily.

Each of these factors was associated with lower mortality rates and combinations were particularly powerful. This work adds to a considerable body of previous studies with remarkably similar conclusions. For example, Trichopoulou et al⁶ observed a comparable lower total mortality rate associated with a Mediterranean diet and moderate alcohol consumption. Our group found that adherence to similar healthful lifestyle practices was associated with an 83% reduction in the rate of coronary disease,⁷ a 91% reduction in diabetes in women,⁸ and a 71% reduction in colon cancer in men.⁹

The study by Knoops et al⁵ is unique in its focus on older persons; however, given the higher underlying death rates in older people, application of these results to younger populations would almost surely show greater benefit. In this European study, only participants who remained free of chronic disease into their 70s and 80s were included for analysis. Even in this highly selected population, adherence to a healthful lifestyle prolonged life.

Effects of dietary interventions also have been studied in randomized trials.^{10 - 13} In the Lyon Diet Heart Study,¹⁰ individuals with established coronary disease showed reductions of 79% in heart disease after just a few years of following the experimental Mediterranean-type diet. Although the results may seem simply too good to be true, given the 20-fold or more differences in coronary rates across countries, such results for dietary change are entirely plausible. The PREMIER group found that an intervention to increase activity, reduce weight, and implement the Dietary Approaches to Stop Hypertension (DASH) diet reduced blood pressure after 6 months.¹²

Longer-term effects of diet, such as for colon cancer, may be less amenable to interventions in studies of primary prevention. Giovannucci et al¹⁴ found that high intake of folate for more than 15 years was associated with lower risk of colon cancer, but a successful clinical trial of that duration is difficult to envision. Likewise, long-term trials of alcohol consumption, trans fat intake, carbohydrate quantity and quality, or physical activity for cardiovascular disease, cancer, and other clinical outcomes seem unlikely.

Nevertheless, further randomized trials examining components of the healthful lifestyle would be useful, especially for short-term effects, either on clinical outcomes, biochemical markers of risk, or intermediate end points. In this issue of JAMA, Esposito and colleagues¹⁵ report the results of their investigation exploring possible mechanisms underlying a dietary intervention. The authors randomized 180 patients (99 men, 81 women) with metabolic syndrome to a Mediterranean-style diet (instructions about increasing daily consumption of whole grains, vegetables, fruits, nuts, and olive oil; n = 90) vs a cardiac-prudent diet with fat intake less than 30% (n = 90). Physical activity increased equally in both groups. After 2 years, body weight decreased more in the intervention group than in the control group, but even after controlling for weight loss, inflammatory markers and insulin resistance declined more in the intervention than in the control group, while endothelial function improved. Only 40 patients in the intervention group still had metabolic syndrome after 2 years compared with 78 patients on the control diet. These results suggest a plausible mechanism for the beneficial effects of the Mediterranean diet and provide substantial support for the observations of Knoops et al.

Additional intervention trials using intermediate end points would be useful. For example, although no long-term trial of alcohol consumption has been conducted, many short trials have reported a wide range of apparently beneficial changes in biochemical markers of cardiovascular risk.¹⁶ Trials of diet and alcohol with intermediate end points, such as ultrasound or magnetic resonance imaging measures of atherosclerotic changes, are quite feasible with the caveat that much of the effect may not be captured with a limited intermediate end point compared with clinical outcomes. For example, fish oil may reduce cardiovascular mortality by reducing arrhythmias, an effect that would not be observed in a study of arterial intima-media thickness.

Because of the limitations in feasibility for randomized trials to address the long-term effects of lifestyle on clinical outcomes, clinicians and researchers must rely mainly on a combination of epidemiologic investigations and short-term trials with intermediate end points. Continued refinement of observational studies is thus essential. Knoops et al used a 10-year follow-up of 2339 persons, with a single assessment at baseline. To study longer-term effects among younger individuals with reasonable precision, a far larger number of participants are required, and repeated assessments are needed to reflect long-term exposure and take into account changes over time. The value of long durations of follow-up was illustrated dramatically in the recent 50-year follow-up of the British Doctors Cohort.¹⁷ That study demonstrated that the adverse effects of persistent smoking had been substantially underestimated in studies that relied on shorter durations of follow-up. A similar phenomenon has been documented for dietary intake. In a 14-year follow-up study of dietary fat intake and coronary heart disease events among women from the Nurses' Health Study, the positive association between trans fat and saturated fat and the inverse association for monounsaturated and polyunsaturated fat was generally stronger when we used a cumulative measure of diet constructed from repeated dietary assessments compared with a simple baseline dietary measurement.¹⁸

Assessment of diet requires further refinement. Knoops et al assessed diet using 8 broad components obtained from interview-administered diet histories that focused on diet only over the previous 2 to 4 weeks. The authors recognize the substantial potential for misclassification of this approach, which likely markedly underestimates the true impact of a healthful diet. Large studies with repeated comprehensive dietary assessments, coupled when possible with biochemical assays of nutrients, will define dietary intakes with greater accuracy, and permit better quantification of intake of nutrients and foods. The simple index using, for example, saturated vs unsaturated fats, will give way to an assessment of specific groups of fatty acids. Higher intake of fruits and vegetables seems to be healthful, but surely some vegetables are preferable to others. Knoops et al were compelled by the data collection methods to include potatoes with vegetables, although substantial evidence suggests that intake of foods with a high-glycemic index, like potatoes, especially in the context of inactivity and overweight, does not provide the health effects conveyed by other vegetables.^{19 - 20} A more precisely defined healthful diet likely would be associated with even lower mortality rates.

Assessment of physical activity also requires further refinement. Knoops et al used a physical activity questionnaire designed for the elderly to assess moderate leisure-time activities. As with some other healthful lifestyles, no long-term trials have proved the benefits of physical activity on mortality, but the observational evidence is overwhelmingly consistent.²¹ However, general activity questionnaires cannot assess activity with complete accuracy and often do not capture sedentary behavior as a separate risk domain. Daily hours of watching television or videos or playing computer games may not be important health determinants in an elderly European population, but time spent at these activities continues to increase at an alarming rate among children and is already high among middle-aged adults in developed and developing nations. Growing evidence supports an independent link between these sedentary behaviors and risk of obesity, chronic disease, and mortality.^{22 - 23}

Large cohorts also permit an assessment of specific causes of morbidity and mortality. Although improved mortality rates can be an important motivation for maintaining a healthful lifestyle, 60% to 75% of first coronary events, for example, are not fatal, but can lead to great economic burden and substantial premature disability. Also, the effect of lifestyle patterns on general quality of life²⁴ and common nonfatal conditions, such as cognitive decline, that influence the quality of life²⁵ should be evaluated. Moreover, a benefit for less common specific causes of death also might easily be missed when assessing total mortality. Thus, the results for mortality from the study by Knoops et al may greatly underestimate the total health benefits of a maintaining a healthful lifestyle.

Although understanding of the relation of lifestyle and health outcomes will continue to be refined, information available now is sufficient to take action. Knoop et al have identified a simple set of lifestyle practices that can reduce the mortality rate among elderly individuals by nearly two thirds. Esposito and colleagues provide evidence of the possible mechanisms for such effects. Both studies are supported by prior data. As a society, the United States spends billions on chronic disease treatments and interventions for risk factors. Although these are useful and important, a fraction of that investment to promote healthful lifestyles for primary prevention among individuals at all ages would yield greater benefit.