Effects of Early Treatment With Statins on Short-term Clinical Outcomes in Acute Coronary Syndromes
A Meta-analysis of Randomized Controlled Trials
- Matthias Briel, MD;
- Gregory G. Schwartz, MD, PhD;
- Peter L. Thompson, MD, FRACP;
- James A. de Lemos, MD;
- Michael A. Blazing, MD;
- Gerrit-Anne van Es, PhD;
- Meral Kayikçioğlu, MD;
- Hans-Richard Arntz, MD;
- Frank R. den Hartog, MD;
- Nic J. G. M. Veeger, MSc;
- Furio Colivicchi, MD;
- Jocelyn Dupuis, MD, PhD;
- Shinya Okazaki, MD;
- R. Scott Wright, MD;
- Heiner C. Bucher, MD, MPH;
- Alain J. Nordmann, MD, MSc
- Author Affiliations: Basel Institute for Clinical Epidemiology, University Hospital Basel, Basel, Switzerland (Drs Briel, Bucher, and Nordmann); VA Medical Center and University of Colorado Health Sciences Center, Denver (Dr Schwartz); Sir Charles Gairdner Hospital, University of Western Australia, Perth (Dr Thompson); University of Texas Southwestern Medical School, Dallas (Dr de Lemos); Duke Clinical Research Institute, Durham, NC (Dr Blazing); Cardialysis BV, Rotterdam, the Netherlands (Dr van Es); Ege University School of Medicine, Izmir, Turkey (Dr Kayikçioğlu); Charité Campus Benjamin Franklin, Berlin, Germany (Dr Arntz); Gelderse Vallei Hospital, Ede, the Netherlands (Dr den Hartog); University Medical Center Groningen, Groningen, the Netherlands (Mr Veeger); S. Filippo Neri Hospital, Rome, Italy (Dr Colivicchi); Montreal Heart Institute, Montreal, Quebec (Dr Dupuis); Juntendo University School of Medicine, Tokyo, Japan (Dr Okazaki); and the Mayo Clinic, Rochester, Minn (Dr Wright).
- Corresponding Author: Heiner C. Bucher, MD, MPH, Basel Institute for Clinical Epidemiology, University Hospital Basel, Hebelstrasse 10, CH-4031 Basel, Switzerland (hbucher{at}uhbs.ch).
Abstract
Context The short-term effects of early treatment with statins in patients after the onset of acute coronary syndromes (ACS) for the outcomes of death, myocardial infarction (MI), and stroke are unclear.
Objective To evaluate relevant outcomes of patients from randomized controlled trials comparing early statin therapy with placebo or usual care at 1 and 4 months following ACS.
Data Sources and Study Selection Systematic search of electronic databases (MEDLINE, EMBASE, PASCAL, Cochrane Central Register) from their inception to August 2005, which was supplemented by contact with experts in the field. Two reviewers independently determined the eligibility of randomized controlled trials that compared treatment with statins with a control, were initiated within 14 days after onset of ACS, and had a minimal follow-up of 30 days. Trials with cerivastatin were only included in a sensitivity analysis.
Data Extraction Information on baseline characteristics of included trials and patients, reported methodological quality, lipid levels, and clinical outcome was independently extracted by 2 investigators. Investigators from each included trial contributed additional data if necessary.
Data Synthesis Twelve trials involving 13 024 patients with ACS were included in the meta-analysis. The risk ratios for the combined end point of death, MI, and stroke for patients treated with early statin therapy compared with control therapy were 0.93 (95% confidence interval [CI], 0.80-1.09; P = .39) at 1 month and 0.93 (95% CI, 0.81-1.07; P = .30) at 4 months following ACS. There were no statistically significant risk reductions from statins for total death, total MI, total stroke, cardiovascular death, fatal or nonfatal MI, or revascularization procedures (percutaneous coronary intervention or coronary artery bypass graft surgery). Sensitivity analyses with restriction to trials of high quality or with additional data from a large trial using cerivastatin indicated summary risk ratios even closer to 1.
Conclusion Based on available evidence, initiation of statin therapy within 14 days following onset of ACS does not reduce death, MI, or stroke up to 4 months.
- KEYWORDS:
- ACUTE CORONARY SYNDROMES
- ANGINA, UNSTABLE
- CYCLOOXYGENASE INHIBITORS
- META-ANALYSIS
- MYOCARDIAL INFARCTION
- OUTCOME ASSESSMENT (HEALTH CARE)
- STROKE
Numerous clinical trials and meta-analyses show that long-term therapy with statins reduces the risk of myocardial infarction (MI), stroke, and overall mortality in patients at varying risks for cardiovascular disease.1-4 These trials typically excluded patients with recent acute coronary syndromes (ACS). The early period following onset of ACS represents a critical stage of coronary heart disease with a high risk for recurrent events and death due to vessel occlusions from vulnerable coronary plaques.5 Therefore, strategies to stabilize vulnerable coronary plaques during this high-risk period are arguably paramount.
There is growing evidence that statins, beyond their low-density lipoprotein (LDL) cholesterol-lowering effects, reduce vascular inflammation, improve endothelial function, and decrease thrombus formation.6 All these mechanisms are expected to have favorable effects on ACS. Several observational studies7-10 but not all11 suggest a large reduction in mortality in patients with ACS treated with statins prior to or at hospital discharge. Randomized controlled trials (RCTs) in patients with ACS indicate that statins may reduce composite end points that include recurrent angina, coronary revascularization procedures, and rehospitalization.12-15 The latter end points, however, may be less reliable because they depend to a greater extent on clinical judgment and local practices. None of the RCTs have individually shown a clear benefit on death and MI.
The purpose of this meta-analysis of RCTs was to investigate whether early use of statins within 14 days following the onset of ACS reduces relevant clinical end points of cardiovascular morbidity and overall mortality at 1 and 4 months.
METHODS
Eligibility Criteria
Eligible trials had to fulfill the following criteria to be included in this systematic review: (1) RCT design comparing statin therapy with placebo or usual care in patients with ACS (MI or unstable angina); (2) initiation of statin therapy within 14 days following the onset of ACS; and (3) follow-up of at least 30 days. Trials that compared 2 different statins or investigated patients with prior heart transplantation were excluded.15 Trials using cerivastatin16 were only considered for sensitivity analysis because this compound was withdrawn from the market in 2001.17
Data Sources and Study Search
To identify relevant trials, the electronic databases MEDLINE, EMBASE, PASCAL (all from their inception to August 2005) and the Cochrane Central Register of Controlled Trials (Cochrane Library 2005, issue 2) were searched using the terms pravastatin, atorvastatin, fluvastatin, simvastatin, lovastatin, cerivastatin, rosuvastatin and acute coronary syndrome* as text words and hydroxymethylglutaryl-CoA reductase inhibitors, myocardial ischemia, myocardial infarction, unstable angina, and coronary arteriosclerosis as Medical Subject Headings. The search was restricted to articles indexed as a clinical trial (publication type) or those that included the words “random” or “placebo” in their titles or abstracts. Language restrictions were not imposed. Reference lists of identified articles, recently published editorials, and reviews on the topic for further eligible trials also were searched. Authors of included primary trials contributed additional data relevant for the purpose of this analysis. We were unable to reach investigators from 1 trial.18
Selection and Quality Assessment
Two authors (M.B. and A.J.N) independently assessed trial eligibility and quality. The quality of the trials was assessed according to concealment of treatment allocation; blinding of patients, caregivers, or clinical outcome assessors; and the proportion of patients with complete clinical follow-up.19 We considered treatment allocation to be concealed if a central independent randomization facility, the use of numbered sealed opaque envelopes, or a central pharmacy, which prepared and distributed numbered containers, was mentioned in the RCT.
End Points and Data Extraction
The primary end point was the combined outcome of nonfatal MI, nonfatal stroke, and total death. Secondary individual end points were total death, total MI, total stroke, cardiovascular death, fatal MI, nonfatal MI, revascularization procedures (coronary artery bypass graft surgery, angioplasty), and unstable angina (recurrent myocardial ischemia requiring emergency hospitalization). We aimed to assess all end points at 1 and 4 months of follow-up. Adverse events (rhabdomyolysis, creatine kinase levels >10 times upper limit of normal and liver aminotransferase levels >3 times upper limit of normal) were recorded at the end of follow-up for each trial.
Two authors (M.B. and A.J.N.) independently extracted in duplicate all trial data and the additional data provided by the original trial investigators. End points and adverse events were considered irrespective of their putative relationship with the treatment.
Statistical Analysis
We pooled treatment effects and calculated risk ratios (RRs) for all end points in the treatment and control groups by using a random-effects model.20 We also calculated odds ratios21 but because of the minimal differences in estimates prefer to report only RRs. The presence of publication bias was investigated by using funnel plots.22 Heterogeneity was tested using the Cochran Q test and inconsistency (the percentage of total variance across studies that is due to heterogeneity rather than chance) of treatment effects was measured across the primary and secondary end points.23-24
Sensitivity analyses were prespecified. The treatment effects were examined according to quality components (concealed treatment allocation, blinding of patients and caregivers, blinded outcome assessment), time to initiation of statins, and the type of statin. One post hoc sensitivity analysis was conducted by including unpublished data from a trial using cerivastatin.16 For statistical data analysis, Stata software version 9.0 (StataCorp, College Station, Tex) was used. P<.05 was set as the level of significance.
RESULTS
Seventeen RCTs were identified that compared early statin therapy with placebo or usual care in individuals with ACS (Figure 1). Five trials were excluded from the meta-analysis: 2 reported a follow-up shorter than 30 days,25-26 the investigator was unable to provide outcome data according to the prespecified time points in 1 trial,27 and another trial was still ongoing at the time of our analysis.28 One trial using cerivastatin was prematurely stopped because the drug was withdrawn from the market.16 Data from this trial for the 4.5 months of follow-up were included only in the sensitivity analysis.
Figure 1. Selection of Studies
RCT indicates randomized controlled trial.
The 12 trials included in the meta-analysis investigated 4 different statins in a total of 13 024 individuals with ACS: pravastatin (6 trials),12, 18, 29-32 atorvastatin (3 trials),13, 33-34 fluvastatin (2 trials),14, 35 and simvastatin (1 trial)36 (Table 1). In accordance with our eligibility criteria, only the subgroup of patients with unstable angina was included from the Lescol Intervention Prevention Study (LIPS).14 Only the data from the placebo comparison during the first 4 months of follow-up in the A to Z trial36 were used in this analysis. The analysis for publication bias indicated no evidence of bias for any of the end points. The methodological quality of included trials is summarized in Table 1.
Table 1. General Characteristics of Included Trials
Study Population
The reported mean age of the participants in the trials ranged from 53 to 69 years (Table 2). Each trial enrolled mostly men (Table 3). Prevalence of individual cardiovascular risk factors and cointerventions for an index event such as fibrinolysis therapy or percutaneous coronary intervention (PCI) varied considerably among the included trials.
Table 2. Baseline Characteristics, Type and Treatment of Index Event
Table 3. Baseline Characteristics and Cardiovascular Risk Factors
Lipid-Lowering Effects
The average weighted mean baseline LDL cholesterol level of included participants was 123 mg/dL (3.2 mmol/L) (range, 112-178 mg/dL [2.9-4.6 mmol/L]) (Table 4). The mean reduction in LDL cholesterol ranged from −15% to −53% and the mean reduction in total cholesterol ranged from −9% to −37%. There were higher reductions in the trials using higher drug doses and/or more potent drugs. The effects on levels of high-density lipoprotein cholesterol and triglycerides were less pronounced and inconsistent among trials.
Table 4. Lipid Values at Baseline and Changes During Follow-up
Combined Primary Outcome
During the first month following onset of ACS, 301 (4.7%) of 6464 individuals in the early statin groups experienced death, MI, or stroke compared with 324 (5.0%) of 6421 individuals in the control groups (RR, 0.93; 95% confidence interval [CI], 0.80-1.09; P = .39). At 4 months of follow-up, there were 356 (7.5%) of 4756 individuals in the early statin groups with a primary end point event compared with 381 (8.1%) of 4713 individuals in the control groups (RR, 0.93; 95% CI, 0.81-1.07; P = .30) (Figure 2). We found no evidence for relevant heterogeneity among trials at both follow-up time points. In sensitivity analyses, summary estimates of the primary end point at 1 and 4 months suggested smaller risk reductions for trials with higher methodological quality compared with trials with lower methodological quality (Table 5). Different treatment effects were not found for trials with initiation of statin therapy within 3 days vs up to 14 days after ACS or for trials using different types of statins (Table 6).
Figure 2. Risk Ratios for the Combined Primary End Point of Death, Nonfatal Myocardial Infarction, and Nonfatal Stroke
At 1 month, LAMIL and ESTABLISH are not presented due to the absence of end point events. Cochran Q test for heterogeneity
yielded P = .48. Inconsistency measure is 0% (95% uncertainty interval, 0%-62%). At 4 months, RECIFE and PACT were excluded due to
follow-up of only 1 month and 1.5 months, respectively. Cochran Q test for heterogeneity yielded P = .64. Inconsistency measure is 0% (95% uncertainty interval, 0%-62%). CI indicates confidence interval; RR, risk ratio.
See Table 1 footnote for expansions of trial names. The sizes of the data markers relate to study sample size and the inverse of the
SE of each study.
*Three months follow-up data used.
†Six months follow-up data used.
Table 5. Sensitivity Analysis of Quality Components for the Composite End Point, Total Death, and Unstable Angina at 4 Months
Table 6. Composite End Point, Total Death, and Cardiovascular Death in Patients With Acute Coronary Syndromes
Individual Secondary Outcomes
There were no statistically significant risk reductions from early statin therapy for total death, total MI, total stroke, cardiovascular death, fatal or nonfatal MI, or revascularization procedures (PCI or coronary artery bypass graft surgery) at 1 and 4 months of follow-up or for unstable angina at 1 month of follow-up (Figure 3). At 4 months after the onset of ACS, unstable angina was reduced: 206 (4.8%) of 4268 individuals experienced unstable angina in the early statin groups compared with 256 (6.0%) of 4238 in the control groups (RR, 0.80; 95% CI, 0.64-1.00; P = .05).
Figure 3. Summary Risk Ratios for Secondary End Points
Trials without events for a specific end point were excluded from the corresponding analysis (different totals for different end points). At 4 months, the Reduction of Cholesterol in Ischemia and Function of the Endothelium (RECIFE) study and the Pravastatin in Acute Coronary Treatment (PACT) study were excluded due to follow-up of only 1.5 months and 1 month, respectively. Cochran Q test was used for heterogeneity. CABG indicates coronary artery bypass graft; CI, confidence interval; I2, the percentage of total variance across studies that is due to heterogeneity rather than chance; MI, myocardial infarction; PCI, percutaneous coronary intervention; RR, risk ratio; UI, uncertainty interval. The sizes of the data markers relate to study sample size and the inverse of the SE of each study.
The heterogeneity among treatment effects was low except for unstable angina at 4 months, in which moderate heterogeneity was found. This may be due to differences in the definition of the end point of unstable angina among trials (Table 7). Absolute numbers of revascularization procedures varied among trials not only because of different trial sizes but also because of different trial criteria for revascularization.
Table 7. Fatal MI, Nonfatal MI, Total Stroke, Revascularization, and Unstable Angina in Patients With Acute Coronary Syndromes*
In a sensitivity analysis of the secondary end points, trials of adequate methodological quality were associated with smaller risk reductions than trials lacking quality components (Table 5). However, statistically significant risk reductions from statins at 4 months were found for unstable angina in trials that reported concealed allocation and in trials with blinded outcome assessment.
When we additionally included in a sensitivity analysis data from 3605 patients with ACS with only 4.5 months of follow-up in the Prevention of Ischemic Events by Early Treatment of Cerivastatin Study (PRINCESS), the RRs were 0.95 (95% CI, 0.78-1.17; P = .66) for total death, 0.91 (95% CI, 0.78-1.05; P = .19) for total MI, 0.80 (95% CI, 0.53-1.20; P = .28) for total stroke, and 0.81 (95% CI, 0.69-0.96; P = .02) for unstable angina.
Adverse Events
Among all included trials, there were 9 individuals (0.1%) receiving statin therapy vs 4 individuals (0.06%) receiving placebo who developed myopathy (increase of creatine kinase levels >10 times the upper limit of normal). Three individuals (0.05%) with creatine kinase elevations while receiving statin therapy met the definition of rhabdomyolysis (creatine kinase level >10 000 U/L). All 9 cases with myopathy while receiving statins occurred after the first month of treatment and only among patients treated with high-dose simvastatin (80 mg/d).36 None of these patients experienced a fatal outcome. Among all trials, there were 75 individuals (1.1%) treated with statins vs 28 individuals (0.4%) in control groups whose liver aminotransferase levels increased to higher than 3 times the upper limit of normal.
COMMENT
This systematic review of RCTs in patients after the onset of ACS investigated whether early initiation of statin therapy compared with placebo or usual care improves patients' outcomes in the short-term. The results of this meta-analysis failed to demonstrate a reduction in the composite primary end point (death, MI, or stroke) for patients treated early with statins at 1 and 4 months following ACS. Of the secondary individual end points, evidence for a reduced risk at 4 months following the onset of ACS was found for unstable angina only.
Strengths and Weaknesses
An extensive literature search was conducted to retrieve all relevant eligible trials and investigators of the primary trials were contacted to provide additional information and collaboration. This collaboration with experts in the field should have minimized the potential for publication bias. In addition, formal testing found little evidence for such a bias.
One small trial with 151 randomized individuals was not included because the original investigators failed to clarify outcome events.27 One trial that planned to enroll 1000 individuals was still ongoing at the time of our analysis.28 Two other trials including 3468 patients only had a follow-up of 1 month29 and 1.5 months.32 As a consequence, the power of our analysis, especially at 4 months, was compromised.
We cannot rule out a small beneficial short-term effect with the early use of statins after the onset of ACS on total mortality, MI, and stroke and that this meta-analysis may have lacked the power to rule out this effect. The Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering (MIRACL) trial comparing 80 mg/d of atorvastatin with placebo for 4 months suggested a 16% reduction of a primary composite end point that included recurrent myocardial ischemia with rehospitalization.13 In a post hoc calculation, we estimated the power of our meta-analysis to be 73% to detect a 16% difference in our primary combined outcome of death, MI, and stroke at 4 months of follow-up (1-sided α of .05).37 Thus, a type II error is possible but not likely. To rule out effects of a 10% risk reduction or less on our primary end point, more than 34 000 patients with ACS would be needed in an RCT.38 In addition, our sensitivity analyses indicated even smaller treatment effects when restricting the analysis to trials of adequate methodological quality, or when we additionally included secondary end point data from a large, prematurely terminated trial using cerivastatin in 3605 patients.16
As expected, statins lowered LDL cholesterol levels more efficiently than placebo or usual care, and there were larger reductions in LDL cholesterol in trials using higher doses of statins. However, available data precluded adequate exploration of an association between clinical outcomes and lipid-lowering potency of different statin types and doses. Because only 2 trials measured ultrasensitive C-reactive protein,13, 36 we were not able to systematically investigate an association of early statin therapy with a reduction in inflammatory parameters. Finally, this systematic review cannot address the benefit of the early use of statins after onset of ACS in patients undergoing early PCI of culprit lesions because only a minority of patients in the included trials underwent PCI.
Comparison With Other Studies
Our findings contrast with results from some recently published observational studies8-10 that suggest a lower risk of mortality for early statin therapy within 1 month following onset of ACS (odds ratios as low as 0.4). Results from these observational studies, however, may be prone to bias due to survivor treatment selection,39 competing medical issues,40 or differences of unknown confounders between comparison groups.41 Another large observational study found no benefit of early initiation of statin therapy in a propensity- and covariate-adjusted analysis; this study may have better captured important confounders.11 Our meta-analysis of RCTs demonstrates that observational studies with insufficient control of confounders greatly overestimate the magnitude of effect from early statin therapy after onset of ACS.
On first sight, our findings might appear to contrast with results from the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE IT) trial that randomized patients with ACS to 80 mg/d of atorvastatin or 40 mg/d of pravastatin.15 In fact, however, there is no obvious discordance. In PROVE IT, Kaplan-Meier curves of the primary composite end point appear to diverge as early as 30 days after ACS in favor of patients treated with atorvastatin but the difference did not reach statistical significance until 6 months. It is important to note that the primary composite end point of PROVE IT comprised not only death, MI, and stroke but also recurrent unstable angina requiring rehospitalization and revascularization. If the more limited composite of death, MI, and stroke is considered, there was no significant difference between the 2 treatment groups in PROVE IT. Unstable angina and revascularization were the most frequent events in PROVE IT and appeared to have driven the primary composite end point. Similarly, our meta-analysis indicates that statins reduce the risk of unstable angina following onset of ACS. Although end points such as unstable angina depend at least in part on clinicians' judgment or action and therefore may be less reliable,42 our finding of a risk reduction for unstable angina of 19% (95% CI, 1%-44%) at 4 months in trials with blinded outcome assessment supports the validity of this result.
Implications for Clinicians, Researchers, and Policy Makers
Statins impact lipid profiles within days,25 and in vitro studies show immediate inhibition of smooth muscle cell proliferation and stimulation of reendothelialization by statins.43 These effects seem to translate into a reduction of unstable angina pectoris at 4 months following onset of ACS but not into a reduction of death, MI, or stroke.
In our meta-analysis, only end point events that occurred during the period of randomized treatment were considered. It is likely that the beneficial effects of statins are cumulative. In most of the landmark trials of statins in patients with chronic coronary heart disease, a benefit of treatment was not evident until 1 to 2 years after randomization.2, 44 Similarly, there appeared to be a delayed benefit of more intensive statin treatment compared with less intensive statin treatment in the late phase of the A to Z trial.36 Therefore, some of the benefit of statin treatment in the period up to 4 months after ACS may only become manifest after 4 months. This systematic review confirms that early treatment with statins after the onset of ACS can in general be considered safe even when high doses of statins are used. However, physicians and patients should pay close attention to muscle-related symptoms, especially when maximum available doses—in particular of simvastatin—are administered.
Concern exists that when administered in clinical practice, long-term adherence to statins among patients with recent onset of ACS is poor.45 Evidence from a small RCT and from observational studies suggests better adherence to statins when therapy is started while the patient is still in the hospital and shortly after the onset of ACS.46-47 Together with our finding of a reduced risk for unstable angina at 4 months, these arguments provide a basis to recommend initiation of statin therapy prior to hospital discharge in all patients with ACS to achieve LDL cholesterol levels of at least less than 100 mg/dL48 and preferably less than 70 mg/dL.49 Our results support the safety of current guidelines48, 50 and may help to improve prevailing practices of insufficient lipid-lowering therapy in patients with ACS.51
CONCLUSIONS
Based on available evidence, early statin therapy is not associated with a relevant reduction of death, MI, or stroke during the first 4 months following ACS. However, early statin therapy may reduce the occurrence of unstable angina at 4 months after the onset of ACS while serious adverse events associated with early initiation of statins are rare.
Author Contributions: Drs Briel, Bucher, and Nordmann had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Briel, Schwartz, Bucher, Nordmann.
Acquisition of data: Briel, Schwartz, Thompson, de Lemos, Blazing, van Es, Kayikçioğlu, Arntz, den Hartog, Veeger, Colivicchi, Dupuis, Okazaki, Wright, Bucher, Nordmann.
Analysis and interpretation of data: Briel, Schwartz, Bucher, Nordmann.
Drafting of the manuscript: Briel, Schwartz, Bucher, Nordmann.
Critical revision of the manuscript for important intellectual content: Briel, Schwartz, Thompson, de Lemos, Blazing, van Es, Kayikçioğlu, Arntz, den Hartog, Veeger, Colivicchi, Dupuis, Okazaki, Wright, Bucher, Nordmann.
Statistical analysis: Briel, Nordmann.
Obtained funding: Briel, Bucher, Nordmann.
Administrative, technical, or material support: Briel, Schwartz, de Lemos, Wright, Bucher, Nordmann.
Study supervision: Briel, Bucher, Nordmann.
Financial Disclosures: Dr Schwartz has reported receiving research support from Pfizer. Dr de Lemos has reported receiving honoraria from Merck, Pfizer, Merck/Schering, Bristol-Myers Squibb and grant support from Merck and Pfizer. Dr Blazing has reported receiving honoraria from Merck and Pfizer. Dr Bucher has reported receiving grant support from Merck and Bristol-Myers Squibb.
Funding/Support: The study was supported in part by Novartis in Basel, Switzerland. Drs Briel, Bucher, and Nordmann are supported by Santésuisse (Solothurn, Switzerland) and the Gottfried and Julia Bangerter-Rhyner-Foundation (Berne, Switzerland).
Role of the Sponsor: Novartis, Santésuisse, and the Gottfried and Julia Bangerter-Rhyner-Foundation had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the study; or in the preparation, review, or approval of the manuscript.
Acknowledgment: We are grateful to Peter Wolf, PhD (medical library at the University Hospital Basel, Basel, Switzerland), for his help with the literature search. We thank Catherine May (Sir Charles Gairdner Hospital, Perth, Australia) for her assistance with data extraction, and we thank Eugene Braunwald, MD (Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, Mass), for reviewing the manuscript. None of these persons received compensation for their work on the manuscript.
