Patrick Bélisle

A hierarchical Bayesian meta-analysis of randomised clinical trials of drug-eluting stents

BACKGROUND Drug-eluting stents (DES) are associated with lower restenosis rates than bare-metal stents (BMS), but the benefits and safety of the new devices have not been systematically quantified across different trials. We undertook a meta-analysis of randomised trials comparing BMS and stents eluting sirolimus or paclitaxel. METHODS A systematic literature search aimed to identify all randomised clinical trials with 6-12 months of clinical follow-up. Results were pooled by a hierarchical Bayesian random-effects model with prespecified stratification for drug and the presence of carrier polymer. The primary outcomes examined were rates of death, myocardial infarction, target-lesion revascularisation, major adverse cardiac events (death, myocardial infarction, and target-vessel revascularisation), and angiographic restenosis. FINDINGS We identified 11 eligible trials involving 5103 patients. The pooled mortality rates were low for both DES and BMS with no evidence of any difference between them (odds ratio 1.11 [95% credible interval 0.61-2.06]). Pooled rates of myocardial infarction showed no between-group difference (0.92 [0.65-1.25]). The rate of major adverse cardiac events was 7.8% with DES and 16.4% with BMS (0.42 [0.32-0.53]), and the angiographic restenosis rates were also lower for DES (8.9% vs 29.3%; 0.18 [0.06-0.40]). The pooled rates of major adverse cardiac events for each DES type and the respective BMS were: for sirolimus, 6.8% versus 21.0% (0.28 [0.17-0.41]); for polymer-based paclitaxel 8.7% versus 16.7% (0.47 [0.25-0.71]); and for non-polymer-based paclitaxel 7.7% versus 9.5% (0.64 [0.42-1.00]). We did not observe higher rates of edge restenosis, stent thrombosis, or late incomplete stent apposition with DES, although the credible intervals were wide. INTERPRETATION Sirolimus-eluting and polymeric paclitaxel-eluting stents are effective at decreasing rates of angiographic restenosis and major adverse cardiac events compared with BMS. However, there is no evidence that they affect mortality or myocardial-infarction rates. They also appear to be safe in the short to medium term, although definitive conclusions are not possible. Larger studies with longer follow-up are needed to define better the role of these new devices.

Outcomes of total hip and knee replacement: Preoperative functional status predicts outcomes at six months after surgery

OBJECTIVE To determine whether patients with knee or hip osteoarthritis (OA) who have worse physical function preoperatively achieve a postoperative status that is similar to that of patients with better preoperative function. METHODS This study surveyed an observational cohort of 379 consecutive patients with definite OA who were without other inflammatory joint diseases and were undergoing either total hip or knee replacement in a US (Boston) and a Canadian (Montreal) referral center. Questionnaires on health status (the Short Form 36 and Western Ontario and McMaster Universities Osteoarthritis Index) were administered preoperatively and at 3 and 6 months postoperatively. Physical function and pain due to OA were deemed the most significant outcomes to study. RESULTS Two hundred twenty-two patients returned their questionnaires. Patients in the 2 centers were comparable in age, sex, time to surgery, and proportion of hip/knee surgery. The Boston group had more education, lower comorbidity, and more cemented knee prostheses. Patients undergoing hip or knee replacement in Montreal had lower preoperative physical function and more pain than their Boston counterparts. In patients with lower preoperative physical function, function and pain were not improved postoperatively to the level achieved by those with higher preoperative function. This was most striking in patients undergoing total knee replacement. CONCLUSION Surgery performed later in the natural history of functional decline due to OA of the knee, and possibly of the hip, results in worse postoperative functional status.

Pharmacotherapies for smoking cessation: a meta-analysis of randomized controlled trials

Background: Many placebo-controlled trials have demonstrated the efficacy of individual pharmacotherapies approved for smoking cessation. However, few direct or indirect comparisons of such interventions have been conducted. We performed a meta-analysis to compare the treatment effects of 7 approved pharmacologic interventions for smoking cessation. Methods: We searched the US Centers for Disease Control and Preventions Tobacco Information and Prevention database as well as MEDLINE, EMBASE and the Cochrane Library for published reports of placebo-controlled, double-blind randomized controlled trials of pharmacotherapies for smoking cessation. We included studies that reported biochemically validated measures of abstinence at 6 and 12 months. We used a hierarchical Bayesian random-effects model to summarize the results for each intervention. Results: We identified 70 published reports of 69 trials involving a total of 32 908 patients. Six of the 7 pharmacotherapies studied were found to be more efficacious than placebo: varenicline (odds ratio [OR] 2.41, 95% credible interval [CrI] 1.91–3.12), nicotine nasal spray (OR 2.37, 95% CrI 1.12–5.13), bupropion (OR 2.07, 95% CrI 1.73–2.55), transdermal nicotine (OR 2.07, 95% CrI 1.69–2.62), nicotine tablet (OR 2.06, 95% CrI 1.12–5.13) and nicotine gum (OR 1.71, 95% CrI 1.35–2.21). Similar results were obtained regardless of which measure of abstinence was used. Although the point estimate favoured nicotine inhaler over placebo (OR 2.17), these results were not conclusive because the credible interval included unity (95% CrI 0.95–5.43). When all 7 interventions were included in the same model, all were more efficacious than placebo. In our analysis of data from the varenicline trials that included bupropion control arms, we found that varenicline was superior to bupropion (OR 2.18, 95% CrI 1.09–4.08). Interpretation: Varenicline, bupropion and the 5 nicotine replacement therapies were all more efficacious than placebo at promoting smoking abstinence at 6 and 12 months.

Evidence for use of coronary stents: A hierarchical bayesian meta-analysis

Context Although expensive, coronary stents are routinely used in angioplasties. However, are they better than balloon angioplasty? Contribution This meta-analysis of 29 randomized trials found that routine stenting reduced restenosis rates as compared with provisional stenting but did not affect rates of mortality or myocardial infarction or the need for bypass surgery. Routinely using stents instead of angioplasty with bailout stents for complications of or unsatisfactory results from angioplasties (provisional stenting) reduced the need for repeated revascularization procedures by at most 5 per 100 treated patients. Implications The benefits of routine stents versus angioplasty with provisional stents are modest. Cautions Effects may vary depending on the lesions and vessels that are treated. Trials did not test new drug-eluting stents. The Editors Percutaneous transluminal coronary angioplasty (PTCA) is a common intervention that is used primarily to reduce the symptoms of angina pectoris; it has no discernible benefit for reducing rate of myocardial infarction or death when compared with other treatments (1). An important limitation of PTCA has been the occurrence of restenosis. Coronary stenting is a percutaneous technique involving the intraluminal introduction of metal scaffolding. Coronary stenting was introduced in 1989 to treat the acute complications of PTCA (2) but is now routinely used for most angioplasties. The elective stent era began with the publication in 1994 of two randomized clinical trials showing a reduced rate of restenosis with coronary stenting compared with ordinary PTCA (3, 4). Subsequently, the use of stents has increased exponentially; some consensus panels endorsed this clinical enthusiasm for coronary stenting even before a large body of high-quality evidence was available (5). Recently, more randomized clinical trials comparing coronary stenting to ordinary angioplasty have been published. However, these trials have often had small sample sizes with low event rates; their focus, therefore, has been on composite outcomes. As a consequence, editorialists, citing selected studies, have arrived at opposite conclusions on the role of elective stenting in interventional cardiology (6, 7). A systematic overview may better quantify the benefits of coronary stenting and provide meaningful insights into the separate clinical end points. Two studies have qualitatively evaluated the benefits of coronary stents (8, 9), but only one study, which was published in a nonclinical journal, has quantitatively reviewed this subject (10, 11). Because of the widespread use and cost of stents (an estimated

Comparison of transradial and femoral approaches for percutaneous coronary interventions: A systematic review and hierarchical Bayesian meta-analysis

1.6 billion in 2002 for the United States market alone [12]), an updated quantitative synthesis of the risks and benefits of this technology seems appropriate. Our paper adds three important elements to the assessment of stents. First, we update the earlier quantitative assessment by including several recent trials. Second, we use a more sophisticated statistical analysis that considers variability among studies. Finally, we provide a bias-adjusted result, which allows estimation of an upper limit of any possible bias. Methods Study Group We attempted to identify all randomized studies published before 30 June 2002 that compared PTCA with routine coronary stenting (Appendix Table). We searched the PubMed database by using the key words angioplasty and stent. The search was limited to clinical trials; review articles were excluded. Language of publication was not restricted. We identified a total of 578 articles. Trials of stenting in acute myocardial infarction and comparisons with other percutaneous, medical, or surgical techniques, as well as comparative studies of different stent models were excluded. A total of 29 randomized trials (3, 4, 1339) were identified. All selected trials had at least 6 months of follow-up and reported the meaningful clinical end points of death, myocardial infarction, and repeated angioplasty of the target lesion. Two of the authors independently abstracted all data and resolved differences by consensus agreement. We used a multifaceted approach to validate the search process. Using the criteria stated earlier, MEDLINE retrieved 2277 references but no additional studies. Finally, we hand-searched several prominent general medical and cardiology journals (The Lancet, British Medical Journal, The New England Journal of Medicine, Journal of the American Medical Association, Annals of Internal Medicine, Circulation, Journal of the American College of Cardiology, American Journal of Cardiology, and Heart), all references from the original articles, and recent review articles (810). No additional studies were identified. Statistical Analysis All analyses were intention-to-treat, so that patients requiring crossover were assessed in their originally assigned group. Rates of events are reported at 6 months, unless stated otherwise. It is unlikely, as implied by a fixed-effects meta-analysis model, that the effects of coronary stenting in each trial will be identical because of differences in trial methods, patients, and investigators. Therefore, we used a Bayesian hierarchical random-effects model (40, 41) to synthesize the results. In this model, the individual data in each trial and for each outcome are assumed to follow a binomial distribution; outcome probabilities are allowed to vary between the stent and PTCA groups within each study, and these parameters also vary among studies. The logarithm of the odds ratios among studies varies according to a normal distribution. While this simple method models variability among trials, it does not attempt to explain any observed differences. The benefits of stents may depend on the anatomic lesion, the rate of crossover stenting, and the type of stent. Therefore, we added another level to our hierarchical model that included these explanatory variables. We allowed the logarithm of the odds ratio in each study to depend on a linear model that included the stated three variables. The coefficients of this linear model on the log odds scale were then interpreted as multiplicative factors on the original odds ratio scale. For example, an odds ratio multiplicative factor of 1.2 implies a change of 20% in the odds ratio. Thus, the effect is larger if the original odds ratio is greater than the null value of 1 or smaller if the original odds ratio is less than 1. Conversely, although an odds ratio multiplicative factor of 0.8 also implies a 20% change in the odds ratio, in this case, the effect is larger if the original odds ratio is less than 1 or smaller if the original odds ratio is greater than 1. Another possible explanation for the observed differences is the possible bias introduced because the studies were not blinded (42). Therefore, we produced adjusted estimates by assuming that the rates of repeated PTCA, given the occurrence of restenosis, should be similar in the stent and PTCA groups. We estimated the proportion of patients with angiographic restenosis in the PTCA group who then underwent repeated PTCA and assumed that this rate should also apply to the stent group if the trials were blinded. This assumes that the entire effect of stents on repeated PTCA is due to a reduction in restenosis and not to other factors, such as an unwillingness among interventional cardiologists to attempt repeated PTCA in a patient with a stent and angiographic restenosis. This revised estimate probably provides an upper bound for the bias adjustment; the true odds ratio for repeated PTCA probably is somewhere between this upper bound and the original unadjusted estimate. We estimated marginal posterior densities for all unknown parameters in our models by using the Gibbs sampler via BUGS software, version 0.6 for UNIX (Medical Research Council Biostatistics Unit, Cambridge, United Kingdom). All results are reported as posterior means with 95% equal-tailed credible intervals (CrIs). Credible intervals are the Bayesian analogue to confidence intervals. Role of the Funding Sources The funding sources had no role in the choice of topic; collection, analysis, or interpretation of the data; or in the decision to submit the manuscript for publication. Results The search protocol identified 29 randomized studies involving 9918 patients that compared standard PTCA to routine coronary stenting (Appendix Table). The studies examined patients with stable and unstable angina (Table 1) as well as various types of lesions: 15 studies in large native vessels (>3 mm), 5 studies in small native vessels (<3 mm), 7 studies of occluded vessels, and 1 study each of restenosed native arteries and bypass grafts. In addition, rates of crossover were studied (Table 2). Table 1. Patient Characteristics Table 2. Lesion and Stent Characteristics Methodologic quality of the studies was generally satisfactory. Although the randomization process was not always fully described, accountability for patients was excellent; almost no patients were lost to follow-up. The nature of coronary stenting does not permit blinding of investigators or patients. As will be discussed later, this inability to blind investigators may have influenced the outcome of repeated angioplasty. Few patients died in either group (65 [1.2%] for the PTCA group; 39 [0.8%] for the stent group). The odds ratio for death was 0.69 (95% CrI, 0.43 to 1.05). Figure 1 shows the combined death or myocardial infarction outcomes. Overall, the rates of death and myocardial infarction were similar in the two groups (odds ratio, 0.90 [CrI, 0.72 to 1.11]). Despite the large number of randomly assigned patients, the small number of events leads to a relatively wide credible interval. This implies that a 28% relative reduction or even an 11% increase in the combined death and myocardial infarction event rate with stenting cannot be excluded. In absolute terms, the effect was a difference of 0.38% (CrI, 0.81% to 0.05%) f

Behavioural interventions for smoking cessation : a meta-analysis of randomized controlled trials

BACKGROUND Despite lower risks of access site-related complications with transradial approach (TRA), its clinical benefit for percutaneous coronary intervention (PCI) is uncertain. We conducted a systematic review and meta-analysis of clinical studies comparing TRA and transfemoral approach (TFA) for PCI. METHODS Randomized trials and observational studies (1993-2011) comparing TRA with TFA for PCI with reports of ischemic and bleeding outcomes were included. Crude and adjusted (for age and sex) odds ratios (OR) were estimated by a hierarchical Bayesian random-effects model with prespecified stratification for observational and randomized designs. The primary outcomes were rates of death, combined incidence of death or myocardial infarction, bleeding, and transfusions, early (≤ 30 days) and late after PCI. RESULTS We collected data from 76 studies (15 randomized, 61 observational) involving a total of 761,919 patients. Compared with TFA, TRA was associated with a 78% reduction in bleeding (OR 0.22, 95% credible interval [CrI] 0.16-0.29) and 80% in transfusions (OR 0.20, 95% CrI 0.11-0.32). These findings were consistent in both randomized and observational studies. Early after PCI, there was a 44% reduction of mortality with TRA (OR 0.56, 95% CrI 0.45-0.67), although the effect was mainly due to observational studies (OR 0.52, 95% CrI 0.40-0.63, adjusted OR 0.49 [95% CrI 0.37-0.60]), with an OR of 0.80 (95% CrI 0.49-1.23) in randomized trials. CONCLUSION Our results combining observational and randomized studies show that PCI performed by TRA is associated with substantially less risks of bleeding and transfusions compared with TFA. Benefit on the incidence of death or combined death or myocardial infarction is found in observational studies but remains inconclusive in randomized trials.

BAYESIAN AND MIXED BAYESIAN/LIKELIHOOD CRITERIA FOR SAMPLE SIZE DETERMINATION

AIMS Widely varying estimates of treatment effects have been reported in randomized controlled trials (RCTs) investigating the efficacy of behavioural interventions for smoking cessation. Previous meta-analyses investigating behavioural interventions have important limitations and do not include recently published RCTs. We undertook a meta-analysis of RCTs to synthesize the treatment effects of four behavioural interventions, including minimal clinical intervention (brief advice from a healthcare worker), and intensive interventions, including individual, group, and telephone counselling. METHODS AND RESULTS We searched the CDC Tobacco Information and Prevention, Cochrane Library, EMBASE, Medline, and PsycINFO databases. We included only RCTs that reported biochemically validated smoking cessation outcomes at 6 and/or 12 months after the target quit date. Outcomes were aggregated using hierarchical Bayesian random-effects models. We identified 50 RCTs, which randomized n = 26 927 patients (minimal clinical intervention: 9 RCTs, n = 6456; individual counselling: 23 RCTs, n = 8646; group counselling: 12 RCTs, n = 3600; telephone counselling: 10 RCTs, n = 8225). The estimated mean treatment effects were minimal clinical intervention [odds ratio (OR) 1.50, 95% credible interval (CrI) 0.84-2.78], individual counselling (OR 1.49, 95% CrI 1.08-2.07), group counselling (OR 1.76, 95% CrI 1.11-2.93), and telephone counselling (OR 1.58, 95% CrI 1.15-2.29). CONCLUSION Intensive behavioural interventions result in substantial increases in smoking abstinence compared with control. Although minimal clinical intervention may increase smoking abstinence, there is insufficient evidence to draw strong conclusions regarding its efficacy.

Bayesian sample size determination for normal means and differences between normal means

Sample size estimation is a major component of the design of virtually every experiment in medicine. Prudent use of the available prior information is a crucial element of experimental planning. Most sample size formulae in current use employ this information only in the form of point estimates, even though it is usually more accurately expressed as a distribution over a range of values. In this paper, we review several Bayesian and mixed Bayesian/likelihood approaches to sample size calculations based on lengths and coverages of posterior credible intervals. We apply these approaches to the design of an experiment to estimate the difference between two binomial proportions, and we compare results to those derived from standard formulae. Consideration of several criteria can contribute to selection of a final sample size.

Adjunctive Thrombectomy for Acute Myocardial Infarction A Bayesian Meta-Analysis

SUMMARY Several criteria for Bayesian sample size determination based on lengths and coverages of posterior credible intervals have recently appeared in the literature. Some but not all of these have been applied to estimating sample sizes for normal distributions. In this paper, these criteria are applied to find sample sizes for single normal means and differences between two normal means, both when the variances are known and when they are unknown. Fully Bayesian approaches as well as mixed Bayesian-likelihood approaches are considered. In the case of the difference between two normal means, situations with equal and unequal variances of the two distributions are considered. In addition to the rule that assumes equally sized groups, optimal solutions are determined which allow the sizes drawn from the two populations to differ while minimizing their sum. Exact closed form solutions are available for many of the situations, whereas numerical algorithms are described for others.

Estimating the prevalence of polymyositis and dermatomyositis from administrative data: age, sex and regional differences

Background—In available trials and meta-analyses, adjunctive thrombectomy in acute myocardial infarction (MI) improves markers of myocardial reperfusion but has limited effects on clinical outcomes. Thrombectomy devices simply aspirate thrombus or mechanically fragment it before aspiration. Simple aspiration thrombectomy may offer a distinct advantage. Methods and Results—We identified 21 eligible trials (16 that used a simple aspiration thrombectomy device) involving 4299 patients with ST-segment elevation MI randomized to reperfusion therapy by primary percutaneous coronary intervention with or without thrombectomy. By using Bayesian meta-analysis methods, we found that thrombectomy yielded substantially less no-reflow (odds ratio [OR], 0.39; 95% credible interval [CrI], 0.18 to 0.69), more ST-segment resolution ≥50% (OR, 2.22; 95% CrI, 1.60 to 3.23), and more thrombolysis in myocardial infarction/myocardial perfusion grade 3 (OR, 2.50; 95% CrI, 1.48 to 4.41). There was no evidence for a decrease in death (OR, 0.94; 95% CrI, 0.47 to 1.80), death, recurrent MI, or stroke (OR, 1.07; 95% CrI, 0.63 to 1.92) with thrombectomy. Restriction of the analysis to trials that used simple aspiration thrombectomy devices did not yield substantially different results, except for a positive effect on postprocedure thrombolysis in myocardial infarction grade 3 flow (OR, 1.49; 95% CrI, 1.14 to 1.99). Conclusions—In this Bayesian meta-analysis, adjunctive thrombectomy improves early markers of reperfusion but does not substantially effect 30-day post-MI mortality, reinfarction, and stroke. The use of aspiration thrombectomy devices is not associated with a reduction in post-MI clinical outcomes. Thrombectomy is one of the rare effective preventive measures against no-reflow.