Sample Size

In quasi-experimental studies, decisions about sample size often consider both the primary outcome, usually fatigue intensity, and the potential for valuable covariate analyses. At minimum, the sample size should be large enough to allow meaningful analysis of the fatigue prevalence and intensity variables. If there is any information about the expected effect size and standard deviation, this can be used to estimate a sample size that would be large enough to demonstrate a clinical, significantly pre-post, difference. This analysis often requires inferences about the anticipated effects, and consultation with a biostatistician is prudent.

The sample size of the study group and comparison groups also can be large enough to allow a variety of interesting covariate analyses. Some of these analyses relate specifically to the fatigue assessment. If this assessment has included multidimensional data, such as information about cognitive impairment, sleep and depressed mood, it is possible to control for these phenomena when determining the impact of the intervention on fatigue intensity, or to analyze the effect of the intervention on the different dimensions. If the sample size is large enough, and information about potential etiologies and relevant comorbidities is carefully collected, it may be possible to explore the extent to which various other elements, such as low hemoglobin levels or physical deconditioning, influence the effect of the intervention on fatigue. Multivariate statistical models require numerous subjects for each variable assessed (e.g. 10 or more). Consultation with a biostatistician while designing the study can clarify the types of analyses that might be undertaken and, in turn, determine the optimal sample size.