Narrative bias (‘spin’) is common in randomised trials and systematic reviews of cannabinoids for pain

Andrew Moore, Paige Karadag, Emma Fisher, Geert Crombez, Sebastian Straube, Christopher Eccleston

Research output: Contribution to journalArticlepeer-review

1 Citation (SciVal)


We define narrative bias as a tendency to interpret information as part of a larger story or pattern, regardless of whether the facts support the full narrative. Narrative bias in title and abstract means that results reported in the title and abstract of a paper are done so in a way that could distort their interpretation and mislead readers who had not read the whole paper. Narrative bias is often referred to as ‘spin’. It is prevalent in abstracts of scientific papers and is impactful because abstracts are often the only part of a paper read. We found no extant narrative bias instrument suitable for exploring both efficacy and safety statements in randomized trials and systematic reviews of pain. We constructed a 6-point instrument with clear instructions and tested it on randomised trials and systematic reviews of cannabinoids and cannabis-based medicines for pain, with updated searches to April 2021. The instrument detected moderate or severe narrative bias in the title and abstract of 24% (8 of 34) of RCTs and 17% (11 of 64) of systematic reviews; narrative bias for efficacy and safety occurred equally. There was no significant or meaningful association between narrative bias and study characteristics in correlation or cluster analyses. Bias was always in favour of the experimental cannabinoid or cannabis-based medicine. Put simply, reading title and abstract only could give an incorrect impression of efficacy or safety in about 1 in 5 papers reporting on these products.
Original languageEnglish
Early online date11 Jan 2024
Publication statusE-pub ahead of print - 11 Jan 2024


Dive into the research topics of 'Narrative bias (‘spin’) is common in randomised trials and systematic reviews of cannabinoids for pain'. Together they form a unique fingerprint.

Cite this