Evidence-based Urology: When Is a Study or Meta-analysis Big Enough?

Abstract

Both designing a clinical study and drawing conclusions from studies about the effect of treatments require justification of the sample size via judicious choice of the endpoint and power. Such an a priori power calculation limits the chances of mistakenly claiming a lack of clinical significance if no statistically significant difference between two treatments can be detected. Calculating the sample size for a new trial furthermore requires assessment of existing evidence to determine if its results will contribute to an updated meta-analysis. To rate the power of a meta-analysis, determination of the optimal information size is suggested, following the same principles as for calculating the power of a single trial. Even though these sample size considerations might seem like an additional hurdle, they are necessary to conduct clinical studies and meta-analysis that provide the optimal benefit for both clinicians and patients. PATIENT SUMMARY: Before conducting a clinical trial, researchers should determine how many patients need to be included to detect a difference between treatments. Even though a study might find a statistical difference between treatments, this does not necessarily mean that the difference is relevant for clinical use. These analyses before starting a study are important for producing worthwhile usable evidence for clinicians and patients.