This latest herb reduces blood pressure. This latest drug stops osteoporosis. This superfood prevents cancer. All are backed up by scientific studies. Every day we are presented with these kind of headlines in the news. This Ted-Ed video provides some very nice basic information about how to evaluate scientific studies in the news.
It starts out by pointing out that studies on animals or cells only point the way for further research. These studies provide hypotheses that can be tried on humans. Only human studies can tell you about the effects on humans. The video discusses the gold standard for scientific studies as a randomized clinical trial. Many experts would add that the trial should be double blinded so that neither the participants nor the researchers know which group is which.
It points out that for drugs and medical devices the FDA usually requires two randomized controlled trials for approval. However because this is not practical in all circumstances, epidemiological studies are sometimes needed. Epidemiological studies compare two groups that differ concerning the variable that is being studied.
However an epidemiological study has inherent flaws. It can only show a correlation and not causation. For this reason multiple studies are required. The strongest evidence of causation is found by satisfying Hill's criteria. See the post here for a discussion of Hill's criteria. To read the original article by Hill click here.
The video discusses that there are many confounding variables that may coexist in epidemiological studies and give some good examples. Many times the studies will try to correct for the confounding variables with statistical methods. Even so, to satisfy Hill's criteria multiple studies by multiple researchers in multiple locations under multiple circumstances are needed.
To seriously evaluate an epidemiological study, Hill's criteria need to be satisfied. However as a good way to evaluate all the scientific studies presented in the news on a daily basis, this video gives you good guidelines.