Dataset Information

Bayesian propensity scores for high-dimensional causal inference: A comparison of drug-eluting to bare-metal coronary stents.

ABSTRACT: High-dimensional data provide many potential confounders that may bolster the plausibility of the ignorability assumption in causal inference problems. Propensity score methods are powerful causal inference tools, which are popular in health care research and are particularly useful for high-dimensional data. Recent interest has surrounded a Bayesian treatment of propensity scores in order to flexibly model the treatment assignment mechanism and summarize posterior quantities while incorporating variance from the treatment model. We discuss methods for Bayesian propensity score analysis of binary treatments, focusing on modern methods for high-dimensional Bayesian regression and the propagation of uncertainty. We introduce a novel and simple estimator for the average treatment effect that capitalizes on conjugacy of the beta and binomial distributions. Through simulations, we show the utility of horseshoe priors and Bayesian additive regression trees paired with our new estimator, while demonstrating the importance of including variance from the treatment regression model. An application to cardiac stent data with almost 500 confounders and 9000 patients illustrates approaches and facilitates comparison with existing alternatives. As measured by a falsifiability endpoint, we improved confounder adjustment compared with past observational research of the same problem.

SUBMITTER: Spertus JV

PROVIDER: S-EPMC6397046 | biostudies-literature | 2018 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Bayesian propensity scores for high-dimensional causal inference: A comparison of drug-eluting to bare-metal coronary stents.

Spertus Jacob V JV Normand Sharon-Lise T ST

Biometrical journal. Biometrische Zeitschrift 20180423 4

High-dimensional data provide many potential confounders that may bolster the plausibility of the ignorability assumption in causal inference problems. Propensity score methods are powerful causal inference tools, which are popular in health care research and are particularly useful for high-dimensional data. Recent interest has surrounded a Bayesian treatment of propensity scores in order to flexibly model the treatment assignment mechanism and summarize posterior quantities while incorporating ...[more]

PMID: 29682785

Similar Datasets

Project description:ObjectivesTo assess patients' perspective about factors associated with stent choice.BackgroundDrug eluting stents (DES) markedly reduce the risk of repeat percutaneous coronary intervention (PCI), but necessitate a longer duration of dual anti-platelet therapy (DAPT) as compared with bare metal stents (BMS). Thus, understanding patients' perspective about factors associated with stent choice is paramount.MethodsPatients undergoing angiography rated, on a 10-point scale, the importance (1 = not important, 10 = most important) of avoiding repeat revascularization and avoiding the following potential DAPT drawbacks: bleeding/bruising, more pills/day, medication costs and delaying elective surgery. The factor, or group of factors, that was rated highest by each patient was identified.ResultsAmong 311 patients, repeat revascularization was the single most important consideration to 14.4% of patients, while 20.6% considered avoiding one of the DAPT drawbacks as most important. Most patients (65%) considered avoiding at least one DAPT drawback as important as avoiding repeat revascularization. In no subgroup of patients did more than a quarter of patients prefer avoiding repeat revascularization above all other concerns. Among patients undergoing PCI, more than three quarters received a DES, regardless of their stated preferences (DES use among those most valuing DES benefits, avoiding DAPT drawbacks, or both equally were 78.7%, 86.2%, and 85.6%, respectively, P = 0.56).ConclusionMost patients reported that avoiding DAPT drawbacks was as important as avoiding repeat revascularization. Eliciting patient preferences regarding stent type can enhance shared decision-making and allow physicians to better tailor stent choice to patients' goals and values.Trial registrationDeveloping and Testing a Personalized Evidence-based Shared Decision-making Tool for Stent Selection (DECIDE-PCI). ClinicalTrials.gov Identifier: NCT02046902. URL: https://clinicaltrials.gov/ct2/show/NCT02046902 © 2017 Wiley Periodicals, Inc.

Project description:ObjectiveTo compare the efficacy and safety of biodegradable polymer drug eluting stents with those of bare metal stents and durable polymer drug eluting stents.DesignMixed treatment comparison meta-analysis of 258,544 patient years of follow-up from randomized trials.Data sources and study selectionPubMed, Embase, and Central were searched for randomized trials comparing any of the Food and Drug Administration approved durable polymer drug eluting stents (sirolimus eluting, paclitaxel eluting, cobalt chromium everolimus eluting, platinum chromium everolimus eluting, zotarolimus eluting-Endeavor, and zotarolimus eluting-Resolute) or biodegradable polymer drug eluting stents, with each other or against bare metal stents.OutcomesLong term efficacy (target vessel revascularization, target lesion revascularization) and safety (death, myocardial infarction, stent thrombosis). Landmark analysis at more than one year was evaluated to assess the potential late benefit of biodegradable polymer drug eluting stents.ResultsFrom 126 randomized trials and 258,544 patient years of follow-up, for long term efficacy (target vessel revascularization), biodegradable polymer drug eluting stents were superior to paclitaxel eluting stents (rate ratio 0.66, 95% credibility interval 0.57 to 0.78) and zotarolimus eluting stent-Endeavor (0.69, 0.56 to 0.84) but not to newer generation durable polymer drug eluting stents (for example: 1.03, 0.89 to 1.21 versus cobalt chromium everolimus eluting stents). Similarly, biodegradable polymer drug eluting stents were superior to paclitaxel eluting stents (rate ratio 0.61, 0.37 to 0.89) but inferior to cobalt chromium everolimus eluting stents (2.04, 1.27 to 3.35) for long term safety (definite stent thrombosis). In the landmark analysis after one year, biodegradable polymer drug eluting stents were superior to sirolimus eluting stents for definite stent thrombosis (rate ratio 0.29, 0.10 to 0.82) but were associated with increased mortality compared with cobalt chromium everolimus eluting stents (1.52, 1.02 to 2.22). Overall, among all stent types, the newer generation durable polymer drug eluting stents (zotarolimus eluting stent-Resolute, cobalt chromium everolimus eluting stents, and platinum chromium everolimus eluting stents) were the most efficacious (lowest target vessel revascularization rate) stents, and cobalt chromium everolimus eluting stents were the safest with significant reductions in definite stent thrombosis (rate ratio 0.35, 0.21 to 0.53), myocardial infarction (0.65, 0.55 to 0.75), and death (0.72, 0.58 to 0.90) compared with bare metal stents.ConclusionsBiodegradable polymer drug eluting stents are superior to first generation durable polymer drug eluting stents but not to newer generation durable polymer stents in reducing target vessel revascularization. Newer generation durable polymer stents, and especially cobalt chromium everolimus eluting stents, have the best combination of efficacy and safety. The utility of biodegradable polymer stents in the context of excellent clinical outcomes with newer generation durable polymer stents needs to be proven.

Project description:BackgroundFew studies have examined the efficacy of drug-eluting stents (DES) for reducing aortocoronary saphenous vein bypass graft (SVG) failure compared with bare-metal stents (BMS) in patients undergoing stenting of de-novo SVG lesions. We assessed the risks and benefits of the use of DES versus BMS in de-novo SVG lesions.MethodsPatients were recruited to our double-blind, randomised controlled trial from 25 US Department of Veterans Affairs centres. Eligible participants were aged at least 18 years and had at least one significant de-novo SVG lesion (50-99% stenosis of a 2·25-4·5 mm diameter SVG) requiring percutaneous coronary intervention with intent to use embolic protection devices. Enrolled patients were randomly assigned, in a 1:1 ratio, by phone randomisation system to receive a DES or BMS. Randomisation was stratified by presence or absence of diabetes and number of target SVG lesions requiring percutaneous coronary intervention (one or two or more) within each participating site by use of an adaptive scheme intended to balance the two stent type groups on marginal totals for the stratification factors. Patients, referring physicians, study coordinators, and outcome assessors were masked to group allocation. The primary endpoint was the 12-month incidence of target vessel failure, defined as the composite of cardiac death, target vessel myocardial infarction, or target vessel revascularisation. The DIVA trial is registered with ClinicalTrials.gov, number NCT01121224.FindingsBetween Jan 1, 2012, and Dec 31, 2015, 599 patients were randomly assigned to the stent groups, and the data for 597 patients were used. The patients' mean age was 68·6 (SD 7·6) years, and 595 (>99%) patients were men. The two stent groups were similar for most baseline characteristics. At 12 months, the incidence of target vessel failure was 17% (51 of 292) in the DES group versus 19% (58 of 305) in the BMS group (adjusted hazard ratio 0·92, 95% CI 0·63-1·34, p=0·70). Between-group differences in the components of the primary endpoint, serious adverse events, or stent thrombosis were not significant. Enrolment was stopped before the revised target sample size of 762 patients was reached.InterpretationIn patients undergoing stenting of de-novo SVG lesions, no significant differences in outcomes between those receiving DES and BMS during 12 months of follow-up were found. The study results have important economic implications in countries with high DES prices such as the USA, because they suggest that the lower-cost BMS can be used in SVG lesions without compromising either safety or efficacy.FundingUS Department of Veterans Affairs Cooperative Studies Program.

Dataset Information

Bayesian propensity scores for high-dimensional causal inference: A comparison of drug-eluting to bare-metal coronary stents.

Publications

Bayesian propensity scores for high-dimensional causal inference: A comparison of drug-eluting to bare-metal coronary stents.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets