• Iowa Women’s Health Study. This is a study of 41,836 postmenopausal Iowa women who were between the ages of fifty-five and sixty-nine in 1986. It was designed to examine the effect of several dietary and other lifestyle patterns on the development of cancer.
• Mexican Teachers’ Cohort. This study is following more than 115,000 female teachers living in Mexico, enrolled in the late 2000s, to investigate the effects of socioeconomic status, reproductive history, lifestyle, and dietary factors on the development of chronic diseases and mental illness.
• Multiethnic Cohort Study of Diet and Cancer. This is an ambitious study begun in 1993 that includes 215,000 men and women representing five different ethnic groups: whites, African Americans, Japanese Americans, Latinos, and Native Hawaiians.
• NIH-AARP Diet and Health Study. A joint project between the National Cancer Institute and the AARP, this cohort was started in 1995 to investigate relationships between diet, lifestyle, and cancer.
• Nurses’ Health Study/Nurses’ Health Study II. These studies have been following the health and wellbeing of more than 200,000 female nurses since 1976 (see “Praise for Nurses and Health Professionals” on page 33).
• Shanghai Women’s and Men’s Studies. These cohorts consist of over 130,000 women and men living in Shanghai, China, who were between the ages of forty and seventy-five in 1986 and 1989. They focus on diet-related, environmental, and genetic factors that may cause cancer.
Studies in the U.S. that focus on racially and ethnically diverse populations will offer important information for all Americans. Those under way in Asia, and Mexico will provide valuable information on a wider range of dietary patterns. Africa and South America are still blank pages when it comes to diet and health, because large cohort studies haven’t yet been launched.
Case-Control Studies
In this type of study, researchers gather information from a group of people who have developed a particular disease (the cases) and a similar group of people who are free of that disease (the controls). They then compare the two groups for differences in diet, exercise, or whatever variable they are interested in. Case-control studies are effective tools when the variable is clear-cut—say, cigarette smoking or occupation. They don’t work as well for diet, when only small differences are likely to be seen from person to person. Case-control studies are also more prone to error and bias than cohort studies.
Because case-control studies can be done quickly and inexpensively, they supplied the evidence for many of the early recommendations about diet and health. As information emerges from cohort studies, though, we are finding that the conclusions from case-control studies were, not surprisingly, often off the mark.
Controlled Feeding Studies
These are a kind of short-term randomized trial done with volunteers, sometimes living in special clinic wards, who eat specially prepared meals. The controlled conditions make it possible to see how different foods or nutrients affect changes in blood cholesterol or other biochemical markers. But these studies are too small and don’t go on long enough to measure the effect on disease risks. Nor can they measure how real diets affect people living in the far messier and less controlled real world.
Praise for Nurses and Health Professionals
* * *
Back in 1976, Dr. Frank Speizer at the Channing Laboratory of Brigham and Women’s Hospital and the Harvard School of Public Health started the Nurses’ Health Study. Its initial aim was to investigate the potential long-term consequences of oral contraceptives, which were then being taken by millions of women. Nurses were chosen as the study population because of their knowledge about health and their ability to provide complete and accurate information about various diseases, thanks to their nursing education. The research team signed up 121,700 female registered nurses between the ages of thirty and fifty-five. Since then, the aims of the Nurses’ Health Study have broadened to look at the effects of diet and other lifestyle factors on cancer, cardiovascular disease, osteoporosis, mental health, and other conditions.
The participants complete follow-up questionnaires every two years to update information on a variety of health risk factors, and they complete diet questionnaires every four years.
Former secretary of the U.S. Department of Health and Human Services Donna Shalala called the Nurses’ Health Study “one of the most significant studies ever conducted on the health of women.” To recognize the fortieth anniversary of this study, the American Journal of Public Health devoted a whole issue to recount its many contributions.5
More studies are under way. The Nurses’ Health Study II, started in 1989, includes 116,000 younger nurses. In addition, 15,000 of the children of these nurses are taking part in the Growing Up Today Study. The Nurses’ Health Study 3 is now enrolling women and men and is also focusing on diet and lifestyle factors at younger ages; this study is being conducted entirely online.
Since the nurses’ studies originally included only women, several colleagues and I started the Health Professionals Follow-Up Study in 1986 to examine the effects of diet on chronic disease in men. It initially included 51,529 male dentists, pharmacists, optometrists, osteopathic physicians, podiatrists, and veterinarians.
These dedicated nurses, their children, and male health professionals have made huge contributions to our understanding of the connections between diet and health. This book reflects their time and effort.
Ecological Studies
Much of the motivation for research on diet and health, and some of the early clues about what might be important, have come from studies that compare diets and disease rates in various geographical areas. One of the seminal ecological studies was the Seven Countries Study, conducted by Dr. Ancel Keys and colleagues in the 1960s. These investigators enrolled about 1,000 men in fourteen different areas in seven countries and followed them for a decade to document their rates of heart attacks. They documented about a tenfold difference in rates of heart disease, with the lowest being on the Greek island of Crete and in Japan. Keys and colleagues also showed that, among the fourteen areas, there was a correlation between intake of saturated fat and heart disease rates.6
At the same time, other scientists were showing that men who migrated from areas like Japan, where heart disease rates were low, to the United States, where they were high, were more likely to develop heart disease than men who stayed put. These findings were profoundly important because they clearly showed that the high heart disease rates of the U.S. were not due to genetic factors and were not inevitable.
The central weakness of ecological studies is that many factors other than diet often differ between geographic regions. In the Seven Countries Study it wasn’t possible to conclude that saturated fat was the key cause of heart disease. Clearly more research was needed, but evidence from these ecological studies provided the impetus to look at diet because, in principle, all populations might have been able to achieve the low rates of heart disease seen in Crete even without sophisticated medicine.
In parallel with the work of Keys and colleagues, other scientists were conducting ecological studies of breast and other major cancers. Similar findings emerged: large differences in rates from country to country, an increase in the breast cancer rate seen with migration to the U.S., and strong correlations with dietary factors.
Mendelian Randomization Studies
This approach, named after Gregor Mendel, the nineteenth-century monk known as the father of genetics, is a newcomer to study designs. It takes advantage of new technologies to identify DNA variations in almost every one of our 30,000 genes. If a large epidemiologic study links a genetic variant that’s involved in metabolizing a specific dietary factor with a particular disease risk, it makes a strong case for a cause-effect relationship between the dietary factor and the disease.
Systematic Reviews, Meta-analyses, and Pooled Analyses
When many studies have been done on a particular topic—say, the effect of alcohol on cardiovascular health—it can
be helpful to take a step back and look at all of them together. A systematic review combs through the medical literature to identify all the relevant studies and then offers conclusions based on them. A meta-analysis statistically combines the published results from a systematic review to provide an overall “bottom line.”
One problem with meta-analyses is that they gather data only from published studies and so can’t capture information from “negative” studies, which tend not to get published. Another is that almost anyone with a computer and Internet connection can do a meta-analysis. But to be done well for a complicated topic like diet and health, deep knowledge of the topic is also required. For example, investigators who conducted a headline-grabbing meta-analysis concluding that replacing saturated fat with unsaturated fat had no benefit for heart disease risk7 were clearly unfamiliar with the published literature, the design of the studies they included, and even the definition of the dietary variables that they used.
In a pooled analysis, investigators contribute raw data, both published and unpublished, and analyze it altogether. This allows for more complete and detailed analysis because the raw data are used rather than just data from published studies. Pooled analyses also have their limits, as they are only as strong as the studies included. To combine studies, it is usually necessary to use only the variables that were included in all of the studies, such as just a single baseline assessment of diet.
DECIPHERING MEDICAL NEWS
Careful journalists try to put new research into perspective. But it’s impossible to cram that kind of context into thirty seconds of air time or 250 words, so you often end up with little more than sound bites or headlines. Other than mastering the fine points of nutrition research, here are a few tips that can help you know what nutrition news is worth paying attention to:
• Studies done on people. How foods, nutrients, and food additives affect mice, dogs, and monkeys is an important thread in the fabric of nutrition research. But they may have completely different effects on people. Animal studies can pave the way for future research but are rarely the basis for changing your diet.
• Studies done in the real world. Diet studies done in hospitals or special research centers have given us important information about how the body responds to different nutrients and foods. But they don’t look directly at disease risk, only intermediate markers of disease, so they can’t reliably predict the consequences of different eating habits or strategies on what really matters: your health.
• Studies that look at diseases, not markers for them. Because it takes so long for chronic diseases to develop, many studies use intermediate markers like narrowing of the heart’s arteries or changes in bone density as stand-ins. These changes don’t necessarily translate into real diseases, though. Pay more attention to research that has looked at real health problems like broken bones or heart attacks.
• Large studies. In science, the play of chance is a real problem. The larger the study, the smaller the possibility that chance alone explains potentially important differences between two groups. Larger studies are also more likely to spot important connections that would be missed in smaller ones.
• Weight of evidence. The most persuasive evidence that an effect is real comes from a number of studies done by different researchers at different times using different methods and involving different groups of people. This is a bit like a court of law, in which multiple pieces of evidence are considered and weighed to determine whether someone is guilty with a high level of certainty. (The courtroom is an example of a situation in which important decisions, some of them literally matters of life or death, are made without randomized trials.) In diet and health, when data from randomized trials aren’t available or feasible, the best evidence often comes when a link is seen between a dietary factor and disease in multiple well-designed cohort studies and controlled feeding studies. As described in chapter five, this is how trans fat was “convicted” for increasing the risk of heart disease.
A good example of consistent evidence is the link between moderate alcohol use and reduced risk of heart disease. Possible beneficial effects of alcohol have been suspected for more than two thousand years. In the late 1700s, William Heberden, the British physician who first described the chest pain known today as angina, wrote that “wine and spiritous liquors—afford considerable relief from angina.”8
Sporadic reports appeared throughout the twentieth century suggesting that drinking alcohol prevented clogged arteries, but they were often balanced by reports of the detrimental effects of heavy drinking. Since 1974, though, dozens of case-control and cohort studies from different geographic regions with different alcoholic beverages have shown that people who have one or two alcoholic drinks a day are less likely to have a heart attack or die from heart disease than nondrinkers or heavy drinkers.9 This relation persists even after the results have been statistically adjusted for smoking, exercise, and other variables that could differ between drinkers and nondrinkers. These observations have been further bolstered by evidence from laboratory, animal, and controlled feeding studies in humans showing that alcohol increases levels of protective HDL cholesterol and also makes blood less likely to clot, both of which would be expected to protect against heart disease. Using a Mendelian randomization approach, a genetic variant involved in metabolizing alcohol was shown to be associated with heart attack risk, and only in those consuming alcohol.10
This body of evidence points to a firm conclusion that drinking moderate amounts of alcohol reduces the risk of heart disease. A randomized trial just getting under way as this book goes to press should offer even more information about the benefits and risks of drinking alcohol.
Regardless of the results from all of these different streams of evidence, any decision about drinking should take into account alcohol’s full range of risks and benefits (see chapter nine).
PUTTING IT INTO PRACTICE
Given the flood of information from nutrition research, I suggest that you not make big changes in what or how you eat based on a single study. If a result is on the right track, other studies will show the same thing. And it won’t matter much in the long run whether you make a change today (like taking a vitamin or increasing the amount of monounsaturated fat in your diet) or six months from now.
In fact, Mark Twain’s cynical, laconic view of health information is as good today as it was one hundred years ago: “Be careful about reading health books. You may die of a misprint.”
In the following chapters, I describe the building blocks of evidence that support the key conclusions of this book and can make an important difference in your well-being.
CHAPTER FOUR
* * *
Healthy Weight
MY AIM IN THIS BOOK is to offer straightforward, no-nonsense advice on health and nutrition based on the best information available. I’ll start right here. If your weight is in the “healthy” range, keep it there (see Figure 7). If you are overweight, change your diet and exercise pattern so you won’t add any more pounds and ideally will lose some. This isn’t a new idea, and it certainly won’t land me a spot as the next diet guru on The Dr. Oz Show. But the number that stares up at you from the bathroom scale is one of the most important measures of your future health. Keeping that number in the healthy range is more important for long-term health than the types and amounts of antioxidants in your food or the ratio of fats to carbohydrates.
The amount of food you eat is fundamentally important to whether you gain or lose weight. That will be the focus of this chapter. But the types of food you eat—the quality of your diet—influences how much you eat, so I will focus on the quality of what you eat, not just the amount. I hope you’ll be relieved to know that the same diet that works for maximum health also helps control weight.
Weight sits like a spider at the center of an intricate, tangled web of health and disease. Three related aspects of weight—how much you weigh in relation to your height, your waist size, and how much weight you gai
n after your early twenties—strongly influence your chances of having or dying from a heart attack, stroke, or other type of cardiovascular disease; of developing high blood pressure, high cholesterol, or diabetes; of being diagnosed with postmenopausal breast cancer or cancer of the prostate, endometrium, colon, pancreas, esophagus, or kidney; of having arthritis; of being infertile or having trouble getting an erection; of developing gallstones or cataracts; of snoring or suffering from sleep apnea; of developing adult-onset asthma; and more. As shown in Figure 4, weight is directly linked with a variety of diseases in the Nurses’ Health Study. These data indicate that with increasing body mass index—a measure that includes both weight and height—the risks of heart disease, high blood pressure, gallstones, and type 2 diabetes all steadily increase, even among those in the healthy weight category. Above a body mass index of 30, which is the boundary between overweight and obesity, the risks continue to increase. Similar trends are seen among men in the Health Professionals Follow-Up Study.
Figure 7. Weight and Disease. Among women in the Nurses’ Health Study, the chances of developing any of four common conditions increases with increasing body mass index.
Given the importance of weight in staying healthy, no mention of weight in the USDA Food Guide Pyramid, MyPyramid, or MyPlate for two decades was a serious omission. And weight recommendations in the current Dietary Guidelines for Americans are set too high for many people and may mislead some into thinking that substantial weight gain within the “healthy” weight category is perfectly fine (see page 43). As the evidence shows, it’s not.
Eat, Drink, and Be Healthy Page 5