A practical guide to causal inference in healthcare operations using single-world intervention graphs (SWIGs)

• 成果类型: Article; Early Access
• 署名作者: Cochran, Amy Louise; Alvarez-Avendano, Sebastian Alejandro; Acosta-Perez, Fernando; Kocher, Keith Eric; Patterson, Brian William; Zayas-Caban, Gabriel
• 署名单位: University of Wisconsin System; University of Wisconsin Madison; University of Wisconsin System; University of Wisconsin Madison; University of Wisconsin System; University of Wisconsin Madison; University of Michigan System; University of Michigan; University of Wisconsin System; University of Wisconsin Madison
• 刊物名称: PRODUCTION AND OPERATIONS MANAGEMENT
• ISSN/ISSBN: 1059-1478
• DOI: 10.1177/10591478261483354
• 发表日期: 2026
• 关键词: sensitivity-analysis; identification; IMPACT; BIAS

摘要

Observational studies in operations management (OM) increasingly guide managerial, clinical, and policy decisions in healthcare. To strengthen their rigor, empirical OM research has turned toward causal inference. However, reliably attributing specific effects to interventions using observational data remains challenging. This tutorial describes a causal inference approach for healthcare OM, centered on single-world intervention graphs, which unify the potential outcomes and do-calculus frameworks. We emphasize constructing precise causal questions and determining when causal effects can be identified from observed variables. We present a detailed case study examining whether a longer treatment time in the emergency department can reduce unnecessary admissions without adversely affecting downstream patient outcomes. The example shows what applying the proposed framework looks like in practice, from defining causal questions and encoding assumptions graphically to walking through the identification process, thereby illustrating how structured causal reasoning can inform healthcare OM.