Tackling Inaccurate Sample Size Estimation using Risk of Bias

Abstract

Dear Editor, The majority of the clinical research is carried out on a sample of participants with a specific characteristic rather than the entire population. As a result, estimating the appropriate sample size for any research is critical to ensure adequate power to detect statistical significance. We frequently extract data from previous similar research publications to calculate sample size. When there are several research publications available in the literature from which the sample size for a particular study can be estimated, the researchers will be at a loss as to which one to choose. In such cases, the researcher may choose a reference article at random, or by identifying a recent publication, and/or by selecting a reference article that results to erroneous sample size estimation. There is no appropriate guidance for selecting the reference article, so the resulting sample size can be overestimated or underestimated. Sometimes, researchers purposefully choose a reference study to calculate a smaller sample size for their research, and such manipulations go unnoticed. Occasionally, a researcher may unintentionally select a highly biased reference article without critically evaluating it. These practices can result in inaccurate sample size estimates, undermining the scientific validity of research evidence. Considering this dilemma when selecting a suitable reference article for sample size estimation, it is appropriate to use the risk of bias assessment on all relevant articles and choose the one with the lowest risk of bias. There are already several risks of bias scales available for various research designs. This technique can address both unintentional and intentional mistakes in selecting the reference article for sample size. In research publications, the disclosure of the risk of bias assessment for sample size reference articles improves transparency and the internal and external validity of research. Considering the real potential of this strategy of introducing the risk of bias in selecting appropriate reference articles for sample size estimation should also be incorporated in the research reporting guidelines such as CONSORT guidelines. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.