Vegan for Life - Jack Norris [9]
Another type of ecological study is the migration study, which looks at what happens to the health of people when they relocate and acquire the food and lifestyle habits of their adopted homeland. These kinds of studies can help show whether risk for certain diseases is related more to genetics or lifestyle.
Ecological studies are riddled with problems because there are many factors that affect health outcomes and these can’t be completely controlled for in the analysis of the data. Additionally, individual food intakes can only be roughly estimated.
Better Evidence: Epidemiologic Research
Epidemiologic studies can establish that two factors occur together but not that one causes the other. They are prone to confounding variables, which means that there might be unidentified issues that cause two factors to be associated. For example, if researchers find out that people with low fruit intakes are more likely to get cancer, it seems logical that fruit is protective against this disease. But what if those who don’t eat fruit also don’t exercise? It’s difficult to establish whether it is lack of fruit or lack of exercise or a combination of both that raises the risk.
Ecological studies, discussed above, are the weakest type of epidemiological studies. The following three types of epidemiology provide stronger evidence.
• Retrospective studies compare past eating habits between people with and without a particular disease. For example, if people with heart disease are more likely to have eaten a diet high in saturated fat, we might conclude that saturated fat has something to do with heart disease. The main drawback of these studies for nutrition research is that people’s memories of their previous diet can be faulty, especially if it has changed over the years.
• Cross-sectional studies compare eating habits and disease rates in groups of people at one moment in time. One problem is that people who have recently become ill may have recently changed their diet.
• Prospective (also called cohort) studies follow large numbers of people who are (usually) healthy when the study begins. As the population is followed, eating patterns of those who eventually get a disease are compared to those who do not. These studies require a lot of subjects—numbering in the tens of thousands—and take place over a long period of time, but they carry the most weight among epidemiologists
Best Evidence: Clinical Trials
The randomized controlled trial (RCT) is the gold standard in nutrition research. It’s the most credible type of study because it randomly assigns people to different groups and then controls what they eat. Ideally, the study is double-blinded; that is, the subjects don’t know whether they are in the test group or the control group. And when the researchers collect the data, they don’t know which group it came from until all the data has been collected and analyzed. The effects of different supplements or foods on disease markers, like cholesterol or bone density, can be studied this way. These studies can be very powerful, and ideally, everything we want to know about nutrition would be tested through RCTs. Unfortunately, they are expensive and complex, which is why they are often smaller in size and shorter in duration than is ideal.
OTHER CONSIDERATIONS
A Word about Statistics
Statistical analyses are always performed to eliminate the probability that different outcomes occurred by random chance. Generally, a finding is statistically significant if there is less than a 5 percent chance that it occurred by chance. When studies are small in size, it becomes difficult to show statistical significance. Even if there appears to be an effect of a treatment or differences between groups, if the differences and treatment are not statistically significant, scientists conclude that there was no effect.
One way to make good use of the data from smaller studies is to do a meta-analysis. This is a statistical analysis of a large number of studies for the purpose of integrating the findings. It is often done to compensate