In Reply to Mancilla-Galindo et al

To the Editor,

We thank Dr Mancilla-Galindo and his colleagues for their interest in our paper.¹

It is important to note that long before the COVID-19 pandemic it has been recognized that systematic reviews are prone to time-lag bias even at the time of publication because changes in evidence regarding clinical decision-making can occur within a short time frame.² Thus, we agree that a “living systematic review” is a useful tool to provide clinicians with cutting edge evidence, especially during a global pandemic, when the evidence base and the standards of care are constantly shifting. However, there is also place for traditional review articles, subject to rigorous peer review, which can help to establish the evidence-based practice that was available at a certain specific time and thus help with the interpretation of other work published during the same period.

Observational data on a complex behavioral intervention such as awake prone positioning are necessarily fraught with inclusion bias, as exemplified by observational studies that reported a physiologically implausible mortality benefit that was not subsequently demonstrated in a large randomized controlled meta-trial.³ We undoubtedly agree that a meta-analysis of observational data can provide only preliminary evidence while awaiting the definitive results from randomized controlled trials.

Although language limitations are not recommended in the Cochrane Handbook, it is acknowledged that the effects of excluding non-English publications in a given systematic review are difficult to ascertain.⁴ A recent meta-epidemiological study found that none of the effect estimates or conclusions of 59 Cochrane reviews of clinical interventions changed after the exclusion of non-English language studies.⁵ Therefore, inclusion of only English language publications is considered a reliable shortcut, especially when addressing fast-moving research areas.

• Copyright © 2022 by Daedalus Enterprises