Accounting for the mortality benefit of drug-eluting stents in percutaneous coronary intervention: a comparison of methods in a retrospective cohort study
1 Cardiology Division, GRB800, Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA
2 Division of Research, Kaiser Permanente of Northern California, 2000 Broadway Street, 3rdFloor, Oakland, CA 94612, USA
3 Division of Biostatistics, Department of Epidemiology and Biostatistics, University of California, San Francisco, 185 Berry Street, Suite 5700, San Francisco, CA 94107, USA
BMC Medicine 2011, 9:78 doi:10.1186/1741-7015-9-78Published: 24 June 2011
Drug-eluting stents (DES) reduce rates of restenosis compared with bare metal stents (BMS). A number of observational studies have also found lower rates of mortality and non-fatal myocardial infarction with DES compared with BMS, findings not observed in randomized clinical trials. In order to explore reasons for this discrepancy, we compared outcomes after percutaneous coronary intervention (PCI) with DES or BMS by multiple statistical methods.
We compared short-term rates of all-cause mortality and myocardial infarction for patients undergoing PCI with DES or BMS using propensity-score adjustment, propensity-score matching, and a stent-era comparison in a large, integrated health system between 1998 and 2007. For the propensity-score adjustment and stent era comparisons, we used multivariable logistic regression to assess the association of stent type with outcomes. We used McNemar's Chi-square test to compare outcomes for propensity-score matching.
Between 1998 and 2007, 35,438 PCIs with stenting were performed among health plan members (53.9% DES and 46.1% BMS). After propensity-score adjustment, DES was associated with significantly lower rates of death at 30 days (OR 0.49, 95% CI 0.39 - 0.63, P < 0.001) and one year (OR 0.58, 95% CI 0.49 - 0.68, P < 0.001), and a lower rate of myocardial infarction at one year (OR 0.72, 95% CI 0.59 - 0.87, P < 0.001). Thirty day and one year mortality were also lower with DES after propensity-score matching. However, a stent era comparison, which eliminates potential confounding by indication, showed no difference in death or myocardial infarction for DES and BMS, similar to results from randomized trials.
Although propensity-score methods suggested a mortality benefit with DES, consistent with prior observational studies, a stent era comparison failed to support this conclusion. Unobserved factors influencing stent selection in observational studies likely account for the observed mortality benefit of DES not seen in randomized clinical trials.