Julia ZubkoHow to read medical research papers like a Pro
Once upon a time, a woman had a question about how to get stronger. As she was living in a digitally connected world, she relied on Google search engine to find the answer to her problem. She types, "How to get stronger"? Many excellent blog articles come up. They are about weightlifting techniques for young men in a gym environment. In addition, the papers are full of various dietary supplementation advice. However, as she is a woman, she is unsure if the recommendations suit her needs. The Google experience is confusing. She needs information specific to her situation. Of course, she could have reached for a better adviser than Google and consulted a fitness industry professional, a personal trainer or an online coach. Unfortunately, not everyone can afford such help. So, let me show you how you can educate yourself without searching Google.
#1 Where you can find free articles.
The best way to find medical research is to find it yourself. The most common place for it is Pubmed. It is a web-based portal that lists all published research papers in various medical journals around the world. However, sometimes Pubmed appears too daunting to navigate. For example, the collection contains studies in rats and humans, and it is impossible to separate them before reading. So, if you are only interested in human studies, try another great website examine.com.
On examine.com, you type a keyword of your interest, and the website will point you out to the published articles covering your topic. If an article is found on Pubmed is not available for free, contact the author directly or ask your local library.
#2 Be sure that the outcome of the study is relevant to you.
All articles contain an abstract describing the study's highlights: the background of the issue, design of the trial, results of the research and conclusion. Read the abstract first. If the presented information looks interesting, go to the study's Results Section and find Table 1. This table describes the demographics of the population involved in the study. Does this look like you? Are you an Asian male of 50 to 60 years of age? If the gender, age, race do not look like yours, there is no point to read the article as the study's outcome will not be relevant to you. The same principle applies to any groundbreaking scientific sensation published in the media: always go back to the original paper and check Table 1.
#3 Study designs you can trust.
Human studies are based on various designs. The two main types are either real studies or systematic reviews of the previously published research papers. The actual studies can be done in a clinical environment or through interviews or questionnaires. The best type of clinical research study is a randomized control study. By randomizing both the control placebo group and the group of participants exposed to treatment, medication, environmental factors etc., we balance other factors causing the results described in the study.
The worse type of questionnaire-based study is a multi-comparison study. One example would be a food consumption frequency questionnaire when an author attempts to find an association between eating a particular food and developing a specific disease. Why is such a design bad? Firstly, we should not trust any data obtained through recollection of events. Our memory can fail us. Secondly, a valid study works on only one principle outcome or hypothesis, which must be identified before the study starts. For example, a hypothesis could be that exposure to A leads to an outcome B. In the food consumption studies, if a primary outcome is negative, in other words, no relationship found, the author usually focuses on the secondary hypothesis. Such an approach makes the study's conclusion confusing and the findings accidental. For example, a paper may say that the authors did not find any relationship between eating food A and the risk of cardiovascular disease. Still, they found the relationship between consuming food B and the risk of cancer. We cannot trust this conclusion.
#4 How to spot a fraud or a dodgy conclusion.
The whole system of medical research publishing is built on trust. Any omission of data is a fraud. Some apparent characteristic of a fraudulent paper is when an article is published by only one group of scientists or in an unknown journal and with a big statement. Big claims require big evidence. Sometimes a study is underpowered, meaning it does not contain enough data points or people to draw a meaningful conclusion. You can spot an underpowered study if it includes even minimum clinically important differences.
And one more tip. Whenever you are reading a study, pay attention to the type of risk the authors are talking about. Most of the time, they give you the relative risk figure: we cut the death rate in half. It sounds more impressive than the absolute risk: we reduced the death rate from 1% to 0.5%.
Happy reading!
