Using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach, do you usually downgrade because of publication bias if the results come from one or two studies with less than 100 patients in total? some reviews didn't give much attention to the number of studies or participants to assess the publication bias, what are your thoughts about this?
Naturally, small sample size and small number of studies are not eligible for publication bias tests. By the rule of thumb, at least 10 studies should be included for these tests because low statistical power leads to false negative result. By the way, low number of studies may figure funnel plot asymmetry. So in such case, there may not robust in considering publication bias test. In case of GRADE, you should reference previous Cochrane Reviews whether it shoud be upgrade or downgrade from another perspective.
Baraa Daraqel , I will refer you to the commentary of the GRADE toolmakers on upgrading/downgrading based on publication bias concerns: https://www.jclinepi.com/article/S0895-4356(11)00181-8/fulltext.
Studies with small sample size can be included even in the simplest tool for assessing publication bias - the funnel plot. Indeed, these studies often occupy the bottom areas of the funnel because of the relative imprecision of their outcome results. However, the funnel plot cannot reliably rule out other reasons for asymmetry especially when the number of included studies is less than ten, hence other tools, particularly the more quantitative ones, may be utilized.
Anyway, I think the linked document will suffice in helping you with your concern to downgrade the GRADE score based on publication bias suspicion. Whether to downgrade or not because of the few included studies should be guided by the context of the field/discipline - it could be that the endeavors to resolve the research/clinical question are only starting now, hence the publications remain few. It could also be that conducting studies on the question is profoundly challenging. Furthermore, it may also be the case that the search strategy implemented was suboptimal.
Nyan Min Aung I agree with you. But if publication bias assessment is not feasible, could you use the GRADE approach despite the omitted publication bias?
From where I stand, I can use GRADE along with publication bias. However, we should search suitable reference as the above mentioned author. Thank you.
- Badges
- Science topic
- Similar topics
- Psychology