Glossary
The words that decide whether a study can be trusted, in plain English. On every page the first use of each term is underlined; tap it for the definition.
- randomised controlled trial
- A study where people are assigned by chance to the treatment or a comparison, so differences can be trusted. The strongest kind of test.
- randomised
- Assigned by chance to treatment or comparison group, which removes bias in who gets what.
- placebo-controlled
- Compared against a dummy treatment, so the effect of simply being treated can be separated from the drug.
- placebo
- A dummy treatment with no active ingredient, used for comparison.
- double-blind
- Neither the patients nor the researchers know who got the real drug until the end.
- open-label
- Everyone knows who is getting the drug. Results are easier to bias.
- uncontrolled
- No comparison group, so you cannot tell what the drug did versus what would have happened anyway.
- retrospective
- Looked back at records after the fact, rather than planned in advance. Weaker than a planned trial.
- case series
- A description of a few patients who were treated, with no comparison group.
- cohort
- A group of people followed over time.
- systematic review
- A study that gathers every published trial on a question and weighs them together.
- meta-analysis
- A statistical pooling of several trials into one combined result.
- surrogate endpoint
- A lab marker (like a hormone level) measured instead of a real outcome (like fewer fractures). It may not translate.
- surrogate marker
- A lab marker measured instead of a real outcome. It may not translate to a benefit people feel.
- endpoint
- The outcome a trial set out to measure.
- primary endpoint
- The main outcome a trial set out to measure; the one that decides whether it succeeded.
- preclinical
- Studied in animals or cells only, not yet in people.
- in vitro
- In cells or test tubes, not in a living body.
- phase 1
- First small human trial, mainly checking safety.
- phase 2
- Medium-sized trial checking whether the drug seems to work and at what dose.
- phase 3
- Large trial, usually the last step before approval.
- 503A
- The part of US law that lets a pharmacy make a drug to order for one patient. A substance has to be on an FDA list to qualify.
- 503B
- The part of US law covering outsourcing facilities that make compounded drugs in bulk under FDA oversight.
- compounding
- A pharmacy making a drug to order instead of a factory. Legal only for substances the FDA allows.
- compounded
- Made to order by a pharmacy rather than a factory.
- bulks list
- The FDA list of substances pharmacies may use to make drugs to order.
- boxed warning
- The FDA's strongest label warning, in a black box at the top.
- indication
- The condition a drug is officially approved to treat.
- off-label
- Prescribed for something other than the approved use. Legal for doctors, but untested for that use.
- research chemical
- Sold online with a 'not for human use' label to avoid drug rules. Purity and contents are unchecked.
- agonist
- A substance that switches a receptor on.
- antagonist
- A substance that blocks a receptor.
- analog
- A lab-made version of a natural molecule, changed slightly to last longer or work differently.
- half-life
- How long it takes for half the drug to leave the body.
- bioavailability
- How much of a dose actually reaches the bloodstream.
- subcutaneous
- Injected under the skin.
- efficacy
- Whether a treatment works under trial conditions.
- statistically significant
- Unlikely to be due to chance. Says nothing about whether the effect is big enough to matter.
- confidence interval
- The range the true effect probably falls in.
- WADA
- The World Anti-Doping Agency, which bans substances in sport.