In addition to the factors of concern to the investigator in clinical studies, other factors can also affect the clinical outcome of patients. These factors are called confounding factors. Multivariate analysis is often used to control for the influence of confounding factors to explore whether or how strongly the main factors are associated with the outcome. However, this approach has the premise that the effects of factors that researchers care about should be the same intensity among different confounders. But in practice, it is often encountered that there is an interaction between different factors. The interaction exists in any study design and will also exist in randomized controlled trials, and this effect cannot be changed by study design. By exploring the interaction, we can more accurately locate the patients with the strongest response to a certain treatment measure or exposure factor, clarify the true effect of intervention or exposure, and accurately locate the patient population of prevention and intervention.