13 Research methods, statistics, critical review and evidence-based practice13.4 Systematic reviews
13.4.1 Primary research and evidence retrieval
2 min read
Written by Resident Doctor Committee · Published
New in the 2027 syllabus
The College added this topic for exams from January 2027.
Types of research
Term | Meaning | Example |
|---|---|---|
Primary research | An original study of a research question | Trial, cohort or interview study |
Secondary research | Brings together or analyses existing research | Systematic review |
Meta-analysis | Combines results across studies statistically | Pooled estimate of a treatment effect |
Primary research can analyse existing individual records. It does not always need new measurements from participants.
A systematic review uses explicit methods to find, select and assess evidence. It can be done without any statistical pooling.
Meta-analysis is only appropriate when the studies and their effect measures are similar enough for a combined result to mean something.
Searching for studies systematically
Eligibility criteria set out the population, intervention or exposure, comparison, outcomes and eligible study designs. Fix them before the results can influence which studies are chosen.
Choice of databases follows the question. Psychiatric research may need biomedical and psychological databases as well as trial registers.
Controlled vocabulary, such as Medical Subject Headings, is used alongside free-text terms. Together they catch indexed concepts and records that are recently added or worded differently.
Boolean searching joins synonyms with OR and different concepts with AND. Unnecessary exclusions can remove relevant studies.
From records to data
Keep a record at each stage so that someone else could repeat the work.
Stage | What to record |
|---|---|
Search | Database, platform, date, full search strategy and limits |
Deduplication | Duplicate records removed and how |
Screening | Eligibility decisions and how disagreements were resolved |
Full-text review | Reasons for exclusions |
Extraction | Study methods, participants, outcomes, estimates and their uncertainty |
Reports and studies are different things. Several articles can describe one trial, so link them before counting participants or pooling results.
Data extraction uses a piloted form, clear outcome definitions and a way of checking the data. Note where in the source each item came from so it can be verified.
Missing information should be sought from protocols, registers, supplementary material or the investigators, where practical. An outcome missing from a report is not necessarily missing from the study.
Reproducibility depends on keeping the actual searches and selection decisions, not just naming the databases.