Propensity score methods in rare disease: A demonstration using observational data in systemic lupus erythematosus

Abstract

Observational studies allow researchers to understand the natural history of rheumatic conditions, risk factors for disease development, and factors affecting important disease-related outcomes, and to estimate treatment effect from real-world data. However, this design carries a risk of confounding bias. A propensity score (PS) is a balancing score that aims to minimize the difference between study groups and consequently potential confounding effects. The score can be applied in 1 of 4 methods in observational research: matching, stratification, adjustment, and inverse probability weighting. Systemic lupus erythematosus (SLE) is a rare disease characterized by a relatively small sample size and/or low event rates. In this article, we review the PS methods. We demonstrate application of the PS methods to achieve study group balance in a rare disease using an example of risk of infection in SLE patients with hypogammaglobulinemia.