Biases Including Confounding and Effect Size Modification in Clinical Studies and Their Mitigations

An association analysis is an integral part of almost every clinical study. To determine the existing relationship of an exposure (e.g., smoking; treatment) with its expected respective outcome (e.g., coronary heart disease (CHD); improvement) accurately, one has to ensure the quality of data. Regardless of the types of association measures depending on various study designs (to be discussed separately), the association/relationship of exposure with an outcome may often get distorted due to the presence of various biases, including confounding and effect size modifications in the collected data. In spite of following each of the major research methodology steps required under a clinical study, to ensure the quality of data, one needs to minimize/nullify each of the numerous probable biases, including confounding and effect size modifications in the data. Hence, the focus of the present write-up is to briefly describe these biases so that researchers can be aware and take care of them in their clinical study.