Determination of sample size for research projects
Research protocols designed to address issues in human health usually require involvement of participants in the data collection and study phase. This involvement may be direct in that participants would actively participate, for example, by completing a questionnaire or by actively participating in the trial of a new device to potentially decrease heat stress. Indirect involvement could occur when medical records are to be accessed as part of a protocol
where any participant interaction is passive and the individuals may not be aware that their records have
been used for a study. In all cases the investigators must consider the number of individuals required to participate to ensure that a meaningful outcome is possible. Both scientific and ethical issues must be considered. The number of participants chosen must provide a statistical basis to test the hypothesis and provide an outcome based on set criteria. An HREC would consider, among other issues, whether there was under or over sampling and whether the study, and the research aims, had been conducted earlier.
An ongoing observation has been that determination of sample size for protocols has not always been rigorously addressed. The accepted approach is to conduct a power calculation based on predicted or known variability expected in a population.
Associated with determination of sample size is choice of an appropriate population and sample for a study
relevant to the actual population within Defence who will directly benefit from the outcomes. Care must be exercised to ensure that results are valid as intended.
This presentation will consider both population choice and matching and provide a practical description and solution for sample size calculation for investigators who may not have ready access to a statistician. Individuals within Defence often do not have the statistical knowledge or access to statisticians to complete the framework for a project. The goal is to establish a support system with tools and educational resources to support investigators. This will also
provide encouragement for new investigators who identify a health problem with the confidence to consider finding a solution.
