Assessing spin bias in abstracts of systematic reviews from Revista médica de Chile between 2017-2021

Session Type
Assessment of the certainty of evidence
Flores N1, Briceño F1, Cabrera C1, Villagran S1, Grandi D1, Marambio B1, Morales D1, Riva N1, Bracchiglione J2
1School of Medicine Universidad de Valparaíso, Chile
2Interdisciplinary Centre for Health Studies Universidad de Valparaíso, Chile

Background: Revista Médica de Chile (RevMedChile) is the oldest monthly journal of Chilean health science, which is responsible for publishing original articles related to internal medicine and its derived subspecialties. It is the Chilean journal of health science with the highest indexes h5 (26) and m5 (40) in the year 2021 according to SCImago Journal Rank (SJR). Systematic reviews (SRs) are a key component in the evidence ecosystem. An absence of coherence between findings of a SR and its interpretation may lead to inaccurate recommendations due to spin bias, defined as a distortion of the results that mislead readers towards a more favourable (or unfavourable) conclusion. To our knowledge, no assessment of spin bias has been conducted regarding local evidence.
Objectives: Assess the frequency of spin bias in SRs published in Rev Med Chile between 2017 and 2021.
Methods: We retrieved every article published in RevMedChile during 2017-2021 and made a full text review to identify the first author, country, theme, methodology, conflict of interest and founding, all of them declared on the articles. Only SRs were included for this study. To assess the presence of spin bias, a masked duplicate fashion data extraction was performed by two reviewers. The data extraction consisted of two phases, the first aimed to characterize the methodological quality of each study by applying the 16 item AMSTAR-2 tool. The second phase aimed to identify the presence of the nine most severe types of spin (Table 1) within the abstract of the included articles. After the data extraction reviewers will be unmasked and discuss their results in order to reach consensus. If a discrepancy is found, a third reviewer will be consulted. Preliminary
Results: We found 1056 articles of which 22 systematic reviews were included. The data extraction is currently ongoing and it will be presented at the colloquium.
Conclusions: Full results will be presented at the colloquium. Spin bias may be an important issue to consider when interpreting conclusions of SR authors.
Patient, public and/or healthcare consumer involvement: Patients, the public and/or healthcare consumers were not involved in this study.

Tabla 1. Spin Bias.png