How to Read a Study

Go to the source

New findings about health are published in scientific journals almost every day, and those articles often become news stories. NCCIH’s Know the Science guide suggests going to the original study rather than relying on a summary of it — while acknowledging that finding, understanding and interpreting a journal article can be hard. Knowing the parts of an article makes it easier.

The parts of a journal article

The labeled parts of a scientific journal article, as described in NCCIH’s Know the Science guide (Key Words is the sixth label).

Start with the Methods: what kind of study was it?

The Methods section tells you the type of research, the design of the study, its size, and how the researchers tried to limit bias. NIH describes a well-designed randomized controlled trial — in which participants are assigned by chance to a treatment group or a control group — as the gold standard for showing that a treatment works, because it can reveal a cause-and-effect relationship. But researchers cannot always use that design. They cannot ethically ask people to start a risky behavior, or assign families to live in different places. So they often turn to observational studies, which watch and analyze a group’s health or behavior. These can find associations and generate ideas to test later, but they cannot prove cause and effect.

The two main types of clinical research. Sources: ClinicalTrials.gov, NIH.

Questions MedlinePlus suggests asking

• Was the study in animals or in people? Results in animals may not hold in humans.
• Were the participants similar to your family in ways that matter, such as age or sex?
• How big was the study? Larger studies generally give more reliable results; small ones usually need confirming.
• How long did it last? A study has to run long enough to show longer-term benefits or risks.
• What type of study was it, where was it done, and what side effects were reported — and how common were they?
• How do the results compare with earlier studies? Conflicting results mean more research is needed.
• Who paid for it, and could the funder — or whoever is reporting it — benefit from a particular result?

Reading the Results and the Discussion

In the Results, NCCIH points to two phrases worth understanding. “Statistically significant” means the difference between groups is real and not likely due to chance. “Clinically significant” is about the size of the effect — how much difference it makes to patients. A result can be statistically significant yet so small that it is not useful in practice. The Results section is also where you look for potential conflicts of interest: most journal articles disclose relevant financial relationships. The Discussion gives the authors’ own explanation and opinions about what the results mean; because those conclusions are the authors’ own, other experts may not fully agree.

One study is a piece, not the picture

A single study rarely provides a final answer. Repeating a study with the same methods but different volunteers and investigators is how researchers learn whether results are reliable. NIH also notes that results from clinical trials are usually published in peer-reviewed journals, where experts check the analysis and conclusions before publication, and that it usually takes many studies before the way doctors prevent or treat an illness changes.