Biases

Bias is prevalent in clinical trials and can significantly impair the internal validity of a study if not properly addressed. Critical appraisal of your study is a key step in identifying bias, and one must make every effort to account for it in both study design and statistical analysis. Importantly, bias is not reflected in P-values or confidence intervals—therefore the onus is on the principal investigator and research team to assess for bias, so that interpretation of any statistical measures of significance in the study can be made in confidence. When it comes to reporting bias in your study, transparency is key. While there are limitations to each type of study design, with observational studies and experimental studies being inherently prone to different types of bias, acknowledging the bias in your study upfront and adjusting for it where feasible will enable accurate and informed interpretation of your results. In this chapter, we will review what the different types of bias are, and which study designs are more prone to specific types of bias. Next, we will review the common causes of each type of bias and the impact that each type of bias can have on the interpretation of your results. After outlining the common causes of bias and the risks inherent in introducing bias into your study, the steps that can be taken to reduce each specific type of bias will be covered. Finally, we will review three simple, recently published examples of bias in radiation oncology studies. Using these examples as a template, we will outline the steps involved in assessing for and accounting for bias in both study design and statistical analysis.