Evaluating the Utility of Negative Control Adjustments and Methods of Addressing Unmeasured Confounding in Using Real-World Data

CERSI P.I.s and Collaborators: Zafar Zafari, M.Sc., Ph.D., (Principal Investigator), Susan dosReis, Ph.D. (co-Investigator), Chintal H. Shah, MS, Jeong-eun Park, MPH, Emily Gorman, MLIS, AHIP, University of Maryland Baltimore 

FDA SMEs and Collaborators: Fang Tian, Ph.D., MPH, MHS, Wei Hua, MD, Ph.D., Rita Ouellet-Hellstrom, Ph.D., MPH, Yong Ma, Ph.D., MS 

 

Randomized controlled trials (RCTs) are the gold standard for evaluating the efficacy and safety of a treatment. However, the expensive costs to conduct RCTs coupled with the generalizability of findings of RCTs to real-world settings have made the use of real-world data in studying long-term drug safety and effectiveness research more popular. Nevertheless, a major concern in using real-world data is the presence of unaccounted biases (e.g., selection bias, information bias) and confounding that may threaten the internal validity of the study results.

 

Negative control methods are useful tools to address issues surrounding confounding and other unaccounted biases (e.g., selection bias, information bias), with potential for broad application. In this project, researchers will (1) conduct a comprehensive literature review for the use of negative control methods in epidemiological studies, and (2) evaluate assumptions behind the use of negative controls in these studies. The ultimate goal of this project is to build a methodological framework for evaluating the use of negative control methods in drug safety and effectiveness studies using real-world data.