Causal Inference for Better Understanding Clinical Trials Results: Reconciling Discrepant Comparative Evidence from Two Major Cardiovascular Safety Trials of Urate-Lowering Therapy

点击翻译按钮获取中文摘要

Project Summary / Abstract Clinical trials are the gold standard of causal inference. However, analyses typically employed in clinical trials cannot answer nuanced questions involving mediating mechanisms. As a result, debate continues regarding the cardiovascular implications of gout medications, such as urate-lowering therapy and, increasingly, colchicine, despite the two major cardiovascular trials among gout patients: Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidities (CARES) and Febuxostat versus Allopurinol Streamlined Trial (FAST). CARES showed an increase in deaths among febuxostat users. However, it had a high loss to follow-up rate. On the other hand, some of the effect estimates in FAST favored febuxostat. However, there were differential discontinuation of urate-lowering therapy and differential use of colchicine in FAST. These idiosyncrasies have led to lingering questions. 1. Did the increased use of colchicine (cardio-protective in non-gout trials) among the febuxostat group affect FAST results in favor of febuxostat? 2. Did higher drug discontinuation in the febuxostat group in FAST and higher loss to follow-up in the allopurinol group in CARES bias their results? In this proposed study, we will use advanced causal inference methods to address these questions via parallel analyses of FAST and CARES. In Aim 1, we will first conduct a comparative effectiveness study of colchicine prophylaxis compared to no prophylaxis or non-steroidal anti-inflammatory drug prophylaxis in FAST and CARES. We will then conduct causal mediation analysis to quantify the extent to which colchicine use explains the protective point estimates favoring febuxostat in FAST. In Aim 2, We will estimate the "per-protocol effect" of febuxostat compared to allopurinol on cardiovascular outcomes. This approach estimates the effect of these medications under a causal scenario of full adherence and follow-up, reducing this significant interpretability difficulty of these trials. Our methodologically innovative project is expected to help to reconcile the discrepant results from FAST and CARES. PI will also develop a valuable collaboration network and preliminary data demonstrating the usefulness of causal inference methods in clinical trial analysis. We will further disseminate these advanced methods via online and offline methodology tutorials.