Find an evidence term quickly

Absolute risk The actual chance of an outcome occurring in a defined group over a defined period. Absolute risk reduction The arithmetic difference in event rates between two groups. Actionability Whether evidence is strong and relevant enough to support a practical decision or behaviour change. Adjustment A statistical attempt to account for other measured factors. Adverse event An unwanted health event occurring during a study or treatment period. Allocation concealment Protecting the randomisation process so group assignment cannot be predicted or manipulated. Animal model Research using animals to explore disease biology or treatment effects before human evidence exists. Animal study Research performed in animals rather than humans. Association A statistical relationship between an exposure and an outcome without proving cause.

Baseline risk The starting risk of an outcome before an intervention, exposure or comparison is considered. Bias A systematic error that can push study results away from the truth. Biomarker A measurable biological signal used to indicate a process, exposure, disease or response. Blinding Keeping participants, clinicians or assessors unaware of group assignment to reduce bias. Breakthrough A media term often used for a finding that may or may not be practice-changing.

Case-control study An observational study comparing people with an outcome to people without it, looking back for exposures. Case series A descriptive report of several patients or events without a formal comparison group. Causation A relationship where one factor directly contributes to producing an outcome. Caveat A limitation or caution that changes how confidently a result should be interpreted. Caveats Important limitations or qualifications that affect how a finding should be interpreted. Cell study Research performed in cells grown or tested outside the body. Cherry-picking Selecting evidence that supports a claim while ignoring less favourable evidence. Claim audit The comparison between what a public article implies and what the evidence directly supports. Clinical endpoint A real-world patient outcome such as symptoms, disease events, function, or survival. Clinical significance Whether a result is large and relevant enough to matter to patients or practice. Cohort study An observational study following people over time to compare outcomes by exposure or characteristic. Composite endpoint An endpoint that combines several outcomes into one measure. Confidence interval A range of values that expresses statistical uncertainty around an estimate. Conflict of interest A financial, professional or personal interest that could influence study design, interpretation or reporting. Confounder A third factor that can distort the apparent relationship between an exposure and an outcome. Confounding When another factor partly or fully explains an apparent association. Control group The comparison group used to judge what would happen without the tested exposure or intervention. Correlation A relationship where two variables tend to change together. Cross-sectional study A study measuring exposure and outcome at a single point in time.

deHype grade A summary judgement of how well a public claim matches the evidence. Directness How closely the evidence matches the claim, population, intervention and outcome being discussed. Dose-response relationship A pattern where greater exposure is linked to greater or lesser effect.

Effect size The magnitude of a difference, association or treatment effect. Endpoint A measured outcome used to judge the effect of an exposure or intervention. Evidence grade A concise deHype judgement on how strongly the source chain supports the public claim. Evidence report A deHype page that compares a public claim with the underlying evidence. Evidence strength How much confidence the available evidence can reasonably support. Evidence trail The sequence of sources and evidence used to assess a claim. External validity The extent to which findings apply to real-world populations and settings. Extrapolation Applying a finding beyond the population, setting, dose, or outcome actually studied.

Funding source The organisation or body that paid for, sponsored or supported a study.

Generalisability How well study findings apply beyond the specific people, settings and conditions studied.

Hazard ratio A time-to-event measure comparing the rate at which an outcome occurs between groups over follow-up. Headline claim The main public-facing claim made or implied by an article title or summary. Heterogeneity Variation between studies, participants, interventions, or results. Hype Presentation that makes evidence sound more certain, dramatic, or useful than it is.

Incidence The number of new cases occurring in a population over time. Intention-to-treat analysis An analysis that keeps participants in their originally assigned groups. Internal validity How well a study design and conduct support a trustworthy answer for its own participants. In vitro Research performed outside a living organism, often in cells or test tubes. In vitro study Research performed outside a living organism, often in cells, tissue or laboratory systems. In vivo Research performed inside a living organism.

Lead-time bias The appearance of longer survival because disease is detected earlier, not because death is delayed.

Mechanism A proposed biological explanation for how something might work. Mechanistic evidence Evidence about how something might work biologically, rather than whether it improves outcomes. Mechanistic study Research exploring how a biological process might work. Meta-analysis A statistical synthesis combining results from multiple studies. Mouse model A study system using mice to investigate biology, disease or treatment ideas. Multiple comparisons The problem that testing many outcomes or groups increases the chance of a false-positive result.

Narrative review A review that summarises literature without fully systematic search and selection methods. Non-inferiority trial A trial designed to test whether a new intervention is not unacceptably worse than a comparator. Number needed to harm The number of people exposed to an intervention for one additional person to be harmed. Number needed to treat The number of people who need an intervention for one additional person to benefit.

Observational study A study where researchers observe exposures and outcomes without assigning the intervention. Odds ratio A comparison of odds between groups, often used in case-control studies and logistic regression. Overclaim A claim that goes beyond what the evidence can support.

Patient-important outcome An outcome that directly matters to patients, such as symptoms, function, admission or survival. Peer review Editorial assessment by other experts before or after publication. Per-protocol analysis An analysis including only participants who sufficiently followed the study protocol. Phase 1 trial An early human trial mainly assessing safety, dosing and tolerability. Phase 2 trial A mid-stage trial exploring whether an intervention shows enough activity to justify larger testing. Phase 3 trial A larger confirmatory trial designed to test effectiveness and safety for clinical use. Pilot study A small early study used to test feasibility or refine methods. Placebo An inactive comparison designed to look or feel like the intervention being tested. Placebo-controlled trial A trial where one group receives an inactive comparison treatment. Post hoc analysis An analysis decided after looking at the data rather than specified in advance. Preliminary finding An early result that needs confirmation before strong conclusions are drawn. Preprint A research manuscript shared publicly before completion of formal peer review. Press release A promotional or explanatory summary issued by a university, journal, company or organisation. Prevalence How common a condition is in a population at a given time. Protocol A pre-specified plan for how a study will be conducted and analysed. Publication bias Distortion caused when positive or striking studies are more likely to be published than negative studies. P value A statistical measure of how compatible the observed data are with a no-effect hypothesis.

Randomisation Assigning participants to groups by chance to reduce systematic differences between groups. Randomised controlled trial A study assigning participants to intervention and comparison groups by chance. Recall bias Bias caused when people remember past exposures or events inaccurately or differently between groups. Regression analysis A statistical method used to estimate relationships while accounting for other variables. Relative risk A comparison of outcome risk between two groups, usually expressed as a ratio or percentage change. Relative risk reduction The percentage reduction in risk compared with another group. Replication Repeating a finding in an independent study or dataset. Reporting bias Distortion caused when some measured outcomes or analyses are selectively reported. Request cap A limit on how many checks or requests a user can make in a given period. Retraction A formal withdrawal of a published paper because its findings should no longer be relied on. Reverse causation When the apparent outcome may actually influence the exposure, reversing the assumed direction. Risk ratio A measure showing how many times more or less likely an outcome is in one group than another.

Safer interpretation A more cautious version of a claim that better fits the evidence. Sample size The number of participants, samples or observations included in a study. Selection bias Bias caused by who enters, remains in or is analysed in a study. Source chain The path from a public claim back through media, press release and original evidence. Statistical power The ability of a study to detect an effect of a specified size if it truly exists. Statistical significance A threshold-based label suggesting a result is unlikely under a specified no-effect model. Subgroup analysis An analysis testing whether results differ within a smaller defined group. Surrogate marker A measurement used as a stand-in for a direct patient-important outcome. Systematic review A structured review using pre-specified methods to identify and appraise relevant studies.

Translational gap The distance between early research findings and real-world clinical use.

Umbrella review A review of systematic reviews, often used to map a broad evidence area.