Designing Analyses by Sex or Gender, Race, and Ethnicity in NIH-defined Phase 3 Clinical Trials


NIH-defined phase 3 clinical trials are required to conduct and report analyses by sex or gender, race, and ethnicity for each primary outcome. These analyses, referred to as “valid analyses” are unbiased assessments that on average, yield the correct estimate of the difference in outcomes between two groups of participants. One can think of valid analysis as a stratified analysis that explores how well an intervention works among sex or gender, racial and ethnic groups.

Investigators may ask, “how do I conduct an ‘unbiased assessment’ as stated in the definition of valid analysis?” Bias can be reduced in several ways. For example, bias in the evaluation can be reduced by using objective measures, and blinding study staff to the treatment assignment of participants. Statistical bias can be reduced by adjusting for potential confounders. When it comes to comparing the intervention effects, one way to achieve this includes reporting intervention effects and their confidence intervals separately for each sex or gender, racial, and ethnic group. It is important to note that simply adjusting the primary analysis for sex or gender, race, or ethnicity is generally insufficient.

Still have questions? Review the resources found on the Analyses by Sex or Gender, Race and Ethnicity for NIH-defined Phase III Clinical Trials (Valid Analysis) website or check with your Program Officer.