Strategies for multivariable analysis in exploratory research: an example of risk factors for chronic kidney disease

Abstract

Researchers often want to identify and explore the relationships between multiple risk factors and an outcome of interests. This kind of research requires multiple regression analysis, due to the multiple risk factors of interests. Stepwise regression is a strategy commonly used in selecting multiple risk factors in a multivariable analysis. Unfortunately, it can lead to inadequate use of information and sometimes offers little in terms of the implications of interventions. In reality, different analysis strategies, using various plausible frameworks, are required for such research. Researchers need to consider the effects of potential confounders and the relationships between risk factors themselves in the multivariable analysis. This can help to disentangle the mechanisms by which these risk factors associate with the outcome of interests, and provide interpretable and meaningful results that help to inform for an evidence-based policy and interventions.