Abstract 10200: Drug-Eluting Stents versus Bare Metal Stents for Medicare Beneficiaries with Acute Coronary Syndromes: Instrumental Variables Compared to Propensity Scores
Context: Controversy persists regarding the safety and efficacy of drug-eluting coronary stents (DES). Observational studies have demonstrated a survival benefit from DES not found in trials, but observational estimates may be undermined by bias from unobserved confounding characteristics that affect stent choice. The rapid diffusion of DES may serve as an instrumental variable (IV) that reduces this bias.
Objectives: To evaluate the safety and efficacy of DES in clinical practice using IV, and to compare IV Results to those produced by propensity-based matching, inverse propensity weighting, and multivariate model risk adjustment.
Design / Methods: Retrospective cohort study involving 74,882 fee-for-service Medicare beneficiaries identified from May 2003 through February 2004 using claims data. Patients were aged 66 years and older and treated with DES or bare-metal stents (BMS) for acute coronary syndrome. Outcomes included mortality and repeat revascularization within 36 months after the initial intervention. Logistic regression was used for all analyses.
Results: DES utilization increased from 30.1% in May 2003 to 59.4% in February 2004; DES recipients were younger, less likely to be admitted for MI, and had fewer comorbidities than BMS recipients. DES utilization was associated with lower rates of all-cause mortality when estimated with conventional techniques [OR 0.76, CI: (0.73, 0.80)) with inverse propensity weighting] but not with IV models (OR 0.98, CI: (0.79, 1.19)]. IV models indicate a greater reduction in repeat revascularization rates [OR 0.64, CI: (0.54, 0.74)] than do conventional approaches [OR 0.89, CI: (0.86, 0.92)]. Propensity-based matching and multivariate adjustment produced estimates comparable to inverse propensity weighting.
Conclusions: The increased utilization of DES among Medicare beneficiaries is associated with reduced rates of repeat revascularization but no change in mortality when estimated by IV analysis, consistent with evidence from trials in populations at lower risk of adverse clinical outcomes than our cohort. Research conducted with conventional techniques such as propensity scores can severely under- or over-estimate treatment effects; IVs offer important correction for selection bias.
- Drug eluting stents
- Epidemiologic methods
- Acute coronary syndromes
- Health policy
- Health services research
- © 2010 by American Heart Association, Inc.