Making the Case for Negative Control Outcome Studies in a Real-World Evidence (RWE) Setting

OBJECTIVES: In real-world settings, unmeasured confounding variables can be a potent threat to study validity. Before embarking on a comparative safety or effectiveness analysis using real-world data (RWD), researchers should assess if patients starting a course of treatment can be validly compared to patients starting a different medication.

METHODS: In this session, we will discuss negative control outcome (NCO) studies. These studies are designed to assess comparability of potential treatment groups. A valid NCO should have 1) no biological relationship with the treatment of interest and 2) share common sources of bias between the treatments and outcomes of interest. Any observed association between the treatment and NCO would represent uncontrolled confounding bias. We will discuss how conducting an NCO study before a comparative safety or effectiveness study can aid in determining if the comparative study can be conducted with minimal bias, thus avoiding unnecessary studies or studies yielding invalid results.

RESULTS: The resulting NCO effect measures can range in size and precision in which case interpreting the set of results may not be straightforward. We discuss various approaches for synthesizing this information.

CONCLUSIONS: Should residual bias be detected, it may be attributable to insufficient sample size or selective prescribing of new therapies. This can be due to a treatment’s lack of uptake in the marketplace and the channeling of new drugs to unique patient populations, such as those with the most severe forms of the disease. We will review ways to set up a system to monitor comparability over time and refresh results with new data as it is available. Often with more data and a longer time on the market, confidence in comparability between treatment groups increases, at which point a comparative effectiveness study is possible.