Research glossary
Plain-language definitions for the terms used in the evidence reviews.
- Active comparator
- A credible intervention used as the comparison, rather than waiting or receiving no added intervention.
- Behavioural activation
- A structured approach that increases meaningful activity and addresses patterns of withdrawal; its tested treatment form is more than an activity reminder.
- CBT-I
- CBT adapted for insomnia, with sleep-specific components and assessment.
- Cognitive restructuring
- Examining an interpretation and considering evidence and alternative perspectives.
- Confidence interval
- An interval expressing statistical uncertainty around an estimate under the analysis assumptions.
- Effect size
- A quantitative description of a difference or association; it is not automatically a recovery percentage.
- ERP
- Exposure and response prevention, a defined behavioural treatment used in OCD care.
- GRADE
- A framework for judging certainty in a body of evidence for a specific outcome, comparison, and population.
- Guided iCBT
- Structured internet CBT accompanied by human support; the type and amount of guidance vary.
- Heterogeneity
- Variation in study findings or methods that affects how a pooled result should be interpreted.
- Indirectness
- A mismatch between available evidence and the population, intervention, comparison, or outcome of interest.
- Intention to treat
- An analysis approach that preserves randomized assignment; its handling of missing data still requires scrutiny.
- Mediation
- An analysis of whether a change in a proposed intermediate process helps account for an observed relationship; mediation alone does not prove causation.
- Meta-analysis
- A statistical synthesis of sufficiently compatible study estimates; its reliability depends on the included studies and methods.
- Network meta-analysis
- A synthesis combining direct and indirect comparisons across a network of interventions, with additional comparability assumptions.
- Noninferiority
- A prespecified comparison assessing whether an intervention is not unacceptably worse than another by a justified margin; no significant difference alone is insufficient.
- Remission
- Meeting a study-defined low-symptom or diagnostic threshold, which differs from average symptom reduction.
- Risk of bias
- The possibility that study design, conduct, analysis, or reporting systematically distorts a result.
- Unguided self-help
- An intervention used without the human guidance provided in a supported treatment programme.
- Waitlist control
- A comparison group waiting for the intervention; it does not provide the same test as another credible treatment.