A novel high dimensional approach to assess heterogeneous treatment effect in claims data

Area of Science:

• Pharmacovigilance and Pharmacoepidemiology
• Biostatistics and Health Informatics
• Cardiovascular Disease Research

Background:

• Detecting heterogeneous treatment effects (HTE) is crucial for personalized medicine.
• Standard high-dimensional propensity score (hdPS) methods have limitations in capturing complex patient characteristics.
• Novel high-dimensional approaches are needed to improve HTE detection.

Purpose of the Study:

• To compare a novel high-dimensional approach with the standard hdPS method for HTE detection.
• To identify patient subgroups with differential treatment effects for sodium-glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists.
• To assess the 2-year risk difference in hospitalized heart failure.

Main Methods:

• Utilized an iterative causal forest (iCF) subgrouping algorithm.
• Analyzed a Medicare cohort of 8,075 SGLT2 inhibitors and 7,313 GLP-1 RAs users.
• Employed a novel high-dimensional approach (1 ordinal variable/code) and standard hdPS (3 binary variables/code).
• Extracted top 200 prevalent diagnosis, procedure, and prescription codes.
• Assessed subgroup-specific conditional average treatment effects (CATEs) using inverse-probability treatment weighting.

Main Results:

• The overall population showed a -0.4% risk difference (aRD) in heart failure hospitalization.
• The novel high-dimensional approach identified patients with ≥2 loop diuretic prescriptions as a subgroup with the largest CATE (aRD: -2.6%).
• The standard hdPS method identified patients with chronic kidney disease (aRD: -1.7%).
• Sensitivity analyses confirmed the novel approach's superior accuracy in identifying expected HTE subgroups.

Conclusions:

• The novel high-dimensional approach demonstrates superior performance in detecting HTE compared to standard hdPS.
• This method accurately identifies patient subgroups, such as those on loop diuretics, with significant differential treatment effects.
• Findings align with prior clinical evidence and support the use of advanced methods for personalized treatment strategies in cardiovascular disease.