Learn to Read Health Claims Before They Read You

How to evaluate health information, statistics, headlines, and research quality for yourself.

Why This Page Exists

A headline can technically be true and still be misleading. Health literacy is the skill of reading past the headline to what a study actually measured, in whom, and how confidently. These are the tools used to write every evidence record on this site, laid out so a reader can apply them independently.

Relative Risk vs. Absolute Risk

A treatment that "cuts risk in half" sounds dramatic. If the underlying risk was 1 in 10,000 and drops to 1 in 20,000, the relative risk reduction is 50 percent, but the absolute risk reduction is 0.005 percentage points. Both numbers are accurate; only one tells you what actually changes for an individual. Look for the absolute numbers before deciding how meaningful a result is.

Correlation vs. Causation

Two things moving together does not prove one causes the other. People who take a certain supplement might also exercise more, eat differently, or have better healthcare access, any of which could explain a health difference. Randomized controlled trials exist specifically to separate a variable from the confounders that surround it in observational data.

Observational Studies vs. Randomized Trials

Observational studies (cohort and case-control designs) follow people as they live their lives and look for patterns. They are essential for questions that cannot ethically or practically be randomized, such as long-term diet patterns, but they cannot fully rule out confounding. Randomized controlled trials assign participants to groups by chance, which is the strongest tool available for establishing that an intervention itself caused an outcome, within the scope of what the trial measured.

Sample Size and Confidence Intervals

A study of 40 people and a study of 40,000 people can report the same percentage effect with very different reliability. A confidence interval shows the range the true effect probably falls within; a wide interval (say, a risk reduction of somewhere between 2 percent and 45 percent) signals more uncertainty than a narrow one.

Screening Statistics

A screening test's sensitivity (how often it correctly flags disease) and specificity (how often it correctly clears healthy people) both matter, and for rare conditions even an accurate-sounding test can produce more false positives than true positives in absolute terms. This is why screening guidelines weigh age, risk factors, and test accuracy together rather than recommending universal testing for everything.

Diagnostic Accuracy

No diagnostic test is perfect. Understanding a test's real-world accuracy, and what a positive or negative result actually changes about the probability of disease, prevents both false reassurance and unnecessary alarm.

Conflicts of Interest

Who funded a study, and who stands to profit from its conclusions, is relevant context. Industry funding does not automatically invalidate a study, but it is a detail worth checking, especially for supplement and consumer-product research.

Medical Guidelines

Clinical guidelines from recognized professional organizations synthesize large bodies of evidence and are generally more authoritative than any single study. Guidelines are also revised as evidence changes, so checking the guideline's publication date matters.

Put It Into Practice

The Evidence Library applies every one of these tools to real evidence records. The Research Explained hub walks through specific studies and guidelines in more depth.