Meta-Analysis
Definition
Statistical pooling of results from several studies into one estimate. Its reliability is capped by the quality of what it pools and by which studies were published in the first place, so a pooled figure is not automatically a better one.
What this means for research sourcing
Understanding evidence quality helps researchers prioritize compounds. Volta's research library summarizes the literature behind stocked compounds.
Related Terms in Evidence & Research
Cochrane Review
A systematic review published by Cochrane, an international network with published methodology and conflict-of-interest rules. Cochrane reviews are generally treated as among the more rigorous syntheses available.
Publication Bias
The tendency for positive results to be published while null and negative ones are not, which leaves the visible literature skewed toward effects. In a field with few large trials it is the single largest distortion.
Peer Review
Assessment of a manuscript by independent researchers before a journal publishes it. It catches some errors and does not verify that the underlying data exist or are correct.
Retraction
A journal's formal withdrawal of a published paper for error or misconduct. Retracted papers keep accruing citations for years, so a claim can stay in circulation long after its source was withdrawn.
Expression of Concern
A notice a journal attaches to a paper when it has credible doubts about its integrity but has not finished investigating. It is a warning short of retraction and is easy to miss on a citation.
ClinicalTrials.gov
The US National Library of Medicine's public registry of clinical studies. Comparing a registered protocol against the published paper is how a switched endpoint is caught.
NCT Number
The identifier every study registered on ClinicalTrials.gov carries, in the form NCT followed by eight digits. It resolves to the protocol, the enrolment status and, where posted, the results.
PubMed
The US National Library of Medicine's free index of biomedical literature, covering more than 30 million citations. A reference that resolves to a PubMed, PMC or DOI record can be checked; one that does not, cannot.
