Have you ever heard of Mr. Safer? In the 1990s, he wanted to understand how the French could eat more high-fat foods yet have lower rates of heart disease than Americans. He argued that having one glass of wine with a meal a day could help a person have better health, as that was what he claimed the French were doing. In a study from 1997, they found that those who had at least one alcoholic beverage a day were 30-40% less likely to die from heart disease. However, what the study did not acknowledge is that people who could afford that glass of wine could also afford better healthcare. Therefore, this study did not provide clear evidence on how healthy it is to drink wine. This indicates that studies related to human health are often unreliable, because they face many issues that question their validity, stemming from confounding variables, untruthful participants, or publication bias.
The first issue with any study is confounding variables. To conclude a valid answer, researchers must not only focus on the independent and dependent variables, but they also must take into account the different conditions and characteristics of the study’s participants. There are many important health-related factors that need to be considered, such as personal wealth, habitat (e.g. city/countryside), eating habits, age, physical activity, occupation, sex, previous health issues, family history etc. Another issue is that individuals are untrustworthy. People tend to lie to appear better or to avoid judgment, specifically with sensitive topics. A review article published by Tourangeau and Yan (2007), that compares and analyses different studies on sensitive topics such as substance abuse, sexual behaviours, voting and income, showed that in some studies about these topics, up to 50% of people lied about their answers. Undoubtedly, there were many studies that showed little to no incorrect data (lies), yet just the fact that there were some studies on which people lied, shows that results around sensitive topics can be quite untrustworthy. The last issue is publication bias. Essentially, researchers are more likely to not publish a study if the results they got are uninteresting or negate their hypothesis i.e. prove them wrong. The non-publication of these studies, the mistaken exclusion of some factors and the false results from untruthful participants all skew the results of individual studies, which is what creates the illusion that a certain result of many studies is correct. Given these limitations, can health studies even be reliable?
Yes and no. While these issues are very real and apparent in many studies, they can also be addressed and are being addressed. By making studies very specific and/or including some of the previously mentioned confounding variables, the studies’ value grows. Instead of trying to link very general problems to some general common factor for all humans everywhere, studies should be more focused. It’s important to remember, that correlation doesn’t always mean causation. If a researcher wants to conclude a relationship between two variables, they must first thoroughly examine and specify it. Furthermore, it is easier to get truthful answers from participants, if they feel safe, not judged or not pressured to answer a certain way or by using tactics and methods, such as the three-card method or using forgiving wording, to get those truthful results from the participants. Ultimately, people want to have better quality of research, so they try to combat these issues. The Declaration of Helsinki, an internationally created document full of ethical guidelines regarding medical research on human participants, states that researchers have an ethical duty to make the results of their studies publicly available, no matter if they are positive, null or negative, while ensuring participant anonymity. The document itself is not legally binding, but many countries require their researchers to comply with the ethical guidelines set by it.
In conclusion, not all health-related studies are inherently wrong. There are many studies that are able to eliminate as much of the previously mentioned errors as possible, which makes them reliable and valid. It is just important to know that these issues exist, and therefore to not treat every published health-related study or meta-analysis as the truth, as the results could be incorrect.
