aPatients withdrew consent after randomization, before further study procedures.bPatients were withdrawn from the study by the investigators within 1 hour because of absence of acute coronary syndrome (ACS) (n = 6) or immediate transfer to another hospital. No data were obtained in these individuals.
Median glucose values per time point after admission. At 6, 12, 24, and 36 hours, there was a significant difference between treatment arms (all P < .001). To convert glucose levels to micromoles per liter, multiply by 0.0555. Horizontal lines indicate medians; upper and lower limits of the boxes, 25th and 75th percentiles; and whiskers, 5th and 95th percentiles.
Infarct size was dichotomized by dividing patients above (large infarction) or below (small infarction) the median troponin T values determined 72 hours after admission (hsTropT72) value. In 10 patients the hsTropT72 value was missing; therefore, they were not included in this analysis. To convert glucose levels to micromoles per liter, multiply by 0.0555. NSTEMI indicates non–ST-segment elevation myocardial infarction; STEMI, ST-segment elevation MI.
eAppendix. Additional details for the BIOMArCS-2 Glucose trial
eTable 1a. Glucose change and outcome by randomization arm
eTable 1b. Infarct size in patients with or without persistently elevated glucose levels (=140 mg/dL) 6 hours after symptom onset
eTable 1c. Association between glucose (change) in patients with or without a clinical event (death or repeat myocardial infarction)
eFigure 1. Cumulative infarct size
eFigure 2. Treatment effects
de Mulder M, Umans VA, Cornel JH, van der Zant FM, Stam F, Oemrawsingh RM, Akkerhuis KM, Boersma E. Intensive Glucose Regulation in Hyperglycemic Acute Coronary SyndromeResults of the Randomized BIOMarker Study to Identify the Acute Risk of a Coronary Syndrome–2 (BIOMArCS-2) Glucose Trial. JAMA Intern Med. 2013;173(20):1896-1904. doi:10.1001/jamainternmed.2013.10074
Elevated plasma glucose levels in patients with acute coronary syndrome (ACS) on hospital admission are associated with increased mortality. Clinical trials of glucose regulation have provided inconsistent results with respect to cardiovascular outcomes, perhaps because target glucose levels have been suboptimal.
To study the effectiveness and safety of intensive glucose management in patients with ACS who have hyperglycemia, aiming at strict blood glucose normalization.
Design, Setting, and Participants
Single-center, prospective, open-label, randomized clinical trial in a large teaching hospital. Patients with ACS with an admission plasma glucose level of 140 to 288 mg/dL were eligible for inclusion and enrolled from July 23, 2008, to February 8, 2012. Patients with insulin-dependent diabetes mellitus were excluded. Informed consent was obtained from 294 patients, who were randomized. Of these, 93.6% received percutaneous coronary intervention (PCI).
Intensive glucose management strategy, aiming at a plasma glucose level of 85 to 110 mg/dL by using intravenous insulin, or to conventional expectative glucose management.
Main Outcomes and Measures
End points were assessed according to the intention-to-treat principle. The primary end point was high-sensitivity troponin T value 72 hours after admission (hsTropT72); secondary end points, area under the curve of creatine kinase, myocardial band (AUC–CK-MB), release and myocardial perfusion scintigraphy findings at 6 weeks’ follow-up.
In the intensive management arm, median hsTropT72 was 1197 ng/L (25th and 75th percentiles of distribution, 541-2296 ng/L) vs 1354 ng/L (530-3057 ng/L) in the conventional arm (P = .41). Median AUC–CK-MB was 2372 U/L (1242-5004 U/L) vs 3171 U/L (1620-5337 U/L) (P = .18). The difference in median extent of myocardial injury measured by myocardial perfusion scintigraphy was not significant (2% vs 4%) (P = .07). Severe hypoglycemia (<50 mg/dL) was rare and occurred in 13 patients. Before discharge, death or a spontaneous second myocardial infarction occurred in 8 patients (5.7%) vs 1 (0.7%) (P = .04).
Conclusions and Relevance
Intensive glucose regulation did not reduce infarct size in hyperglycemic patients with ACS treated with PCI, and was associated with harm. Future studies should focus on patients with ACS who have persistently elevated blood glucose after PCI, and should evaluate alternative strategies for optimizing glycemia.
www.trialregister.nl Identifier: NTR1205
Elevated plasma glucose levels on hospital admission are common among patients admitted with acute coronary syndrome (ACS). More than 40% of patients who develop myocardial infarction (MI) have an admission plasma glucose level greater than or equal to 140 mg/dL,1- 3 the clinically relevant threshold of impaired glucose tolerance.4 (To convert glucose to millimoles per liter, multiply by 0.0555.) Admission plasma glucose has been recognized as an independent determinant of adverse outcomes in patients with and without established diabetes mellitus in both the thrombolytic and percutaneous coronary intervention (PCI) eras.2,3,5- 8 In patients with MI, elevated glucose results in a prothrombotic state, modulates the inflammatory response and oxidative stress, and leads to microvascular dysfunction and no reflow.9- 11 These mechanisms may explain the association between elevated plasma glucose and adverse reactions in patients with MI. However, it is unclear whether elevated plasma glucose contributes to myocardial injury and infarction or is a marker of disease severity.
Earlier studies attempted to regulate hyperglycemia in patients with MI through a metabolic approach at a cellular level, using a combined insulin-glucose infusion. In the landmark Diabetes Mellitus Insulin-Glucose Infusion in Acute Myocardial Infarction (DIGAMI-1) trial,12 which studied patients with diabetes mellitus or hyperglycemia who experienced an MI, lower 1-year mortality was observed after insulin-glucose infusion compared with control treatment (18.6% vs 26.1% deaths). This finding was not confirmed in the DIGAMI-2,13 Clinical Trial of Reviparin and Metabolic Modulation in Acute Myocardial Infarction Treatment and Evaluation—Estudios Clinicos Latino America/Organization for the Assessment of Strategies for Ischemic Syndromes–6 (CREATE-ECLA/OASIS-6),14 Glucose Insulin Potassium Study (GIPS),15 and Immediate Myocardial Metabolic Enhancement During Initial Assessment and Treatment in Emergency Care (IMMEDIATE)16 trials, perhaps because of inadequate plasma glucose level reduction and extensive fluid overload. The recent Hyperglycemia: Intensive Insulin Infusion in Infarction (HI-5) trial17 explored a strategy that directly influenced plasma glucose levels with insulin-dextrose infusion. The incidence of a second MI and heart failure was reduced by active treatment, but the incidence of mortality was not. More important, 24-hour glucose levels were similar in both trial arms and exceeded the normal range. The potential effects of insulin also were investigated in an outpatient setting. The recent Outcome Reduction With an Initial Glargine Intervention (ORIGIN) trial18 tested whether the provision of sufficient basal insulin could reduce cardiovascular events but failed to demonstrate such a relationship after a 6-year follow-up period.
We designed the BIOMarker Study to Identify the Acute Risk of a Coronary Syndrome–2 (BIOMArCS-2) Glucose trial to study the hypothesis that treatment of hyperglycemia aiming at strict glucose control within normal levels will limit infarct size in patients with ACS. In patients allocated to intensive glucose management, we aimed to obtain a rapid reduction of blood glucose to normoglycemic levels by using an intensive insulin protocol while avoiding hypoglycemia.
The BIOMArCS-2 Glucose trial was designed as a single-center (Medical Center Alkmaar, Alkmaar, the Netherlands) open-label, randomized clinical trial to investigate the effectiveness and safety of intensive glucose level control and regulation in patients with ACS and hyperglycemia.
The design of the BIOMArCS-2 Glucose trial has been published,19 and additional details are given in the Supplement. Briefly, patients with ACS whose admission plasma glucose level was between 140 and 288 mg/dL were eligible for inclusion. Acute coronary syndrome was defined as typical ischemic chest pain with either ST-segment elevation or elevated biomarkers of myocardial necrosis (creatine kinase, myocardial band [CK-MB] >16 U/L or cardiac troponin I >0.45 ng/mL, the upper limits of normal in our center; conversion to micrograms per liter for both laboratory values is 1:1). The main exclusion criteria were the use of subcutaneous insulin, a creatinine level greater than 2.5 mg/dL (for conversion to micromoles per liter, multiply by 88.4), mechanical ventilation, and a previously known left ventricular ejection fraction less than 30%. Eligible patients were randomized to a strategy of intensive glucose level control for 48 hours with intravenous insulin or to conventional glucose management. The design of the BIOMArCS-2 Glucose trial was approved by the Medical Ethics Committee Noord Holland, and patients were treated in accordance with the guidelines of the European Society of Cardiology.20,21
Patients were randomized by means of a computer program in a 1:1 ratio to either intensive or conventional glucose management in addition to standard medical care. In patients randomized to the strategy of intensive glucose level control, glucose levels were regulated according to a nurse-driven protocol using intravenous insulin aspart (NovoRapid; NovoNordisk). Details of the protocol have been published.22 Perfusor adjustments were made to reach normoglycemic values while avoiding hypoglycemia. We targeted glucose levels between 85 and 110 mg/dL from 6 am to 10:59 pm and 85 to 139 mg/dL from 11 pm to 6 am. Measurements were repeated every hour, and perfusor adjustments were performed until glucose levels appeared within the target range. Glucose levels then were measured every 3 hours. Additional subcutaneous insulin (standard dose of 6 IU or additional if required) was given near meals. Two hours after each meal, the glucose level was measured. Further details are provided in the Supplement.
In patients randomized to conventional expectative glucose management, glucose levels were measured at 6, 12, 24, 36, and 72 hours after the onset of ACS symptoms. Patients with previously diagnosed non–insulin-dependent diabetes mellitus continued their oral hypoglycemic drugs. Insulin treatment was not started if glucose levels remained below 288 mg/dL at all time points. Patients were crossed over to the intensive glucose regulation group as soon as a single glucose level exceeded that level.
Final infarct size was estimated by 3 methods. First, high-sensitivity troponin T values were determined 72 hours after admission (hsTropT72), which was defined as the sample closest to 72 hours within the 48- to 96-hour time window after admission.
Second, the area under the CK-MB curve (AUC–CK-MB) was calculated by the linear trapezoidal method. Blood samples were obtained on admission and 6, 12, 24, 36, and 72 hours after the onset of ACS symptoms. Missing baseline and 72-hour CK-MB values were set to zero. The log normal function was used to estimate CK-MB values in case of missing values at intermediate time points.23 The AUC–CK-MB was not determined if there were 2 or more missing intermediate samples (29 patients).
The third method was based on rest-gated myocardial perfusion scintigraphy (MPS) using technetium 99m-myoview single-photon emission computed tomography (SPECT). The MPS-SPECT imaging took place at 6 ± 1 week after the index event. Technical details have been described previously.19 We determined the extent of myocardial injury, as well as left ventricular ejection fraction. We realize that MPS-SPECT measurements might also reflect previous myocardial damage, since patients with previous MI were not excluded. However, because BIOMArCS-2 Glucose is a randomized trial, we can assume that the 2 groups were similar with regard to prior injury at entry into the study.
We chose hsTropT72 as the primary end point of our trial. Previous studies24- 27 in patients with ACS demonstrated that the accumulation of cardiac troponin reaches a plateau in this time window, whereas a single measurement in this time window correlated well with final infarct size as determined by cardiac magnetic resonance imaging.
Secondary end points included AUC–CK-MB during the first 72 hours, left ventricular ejection fraction and infarct size as determined by MPS-SPECT at 6 weeks, all-cause death, a second nonfatal MI, and their composite at discharge. Recurrent MI was defined as a second elevation of troponin or CK-MB above the upper limits of normal and/or electrocardiogram showing new ST-segment elevation greater than 1.0 mm. Clinical end points were adjudicated by 2 independent cardiologists who were blinded to treatment strategies. A decision was made by a third cardiologist (V.A.U. or J.H.C.) in case of disagreement.
An independent statistician who was blinded to the allocated treatment performed all data analyses. We refer to the statistical analysis plan28 for details on the analyses. The plan was published before the primary end point was determined.
Categorical data are presented as numbers and proportions, and continuous data are presented as median values with the 25th and 75th percentiles of the distribution or mean (1 SD). The χ2 or Fisher exact tests were used to study differences in categorical data, as appropriate. The distribution of continuous data was tested with the Kolmogorov-Smirnov test. Subsequently, t or Mann-Whitney tests were applied to study between-sample differences.
Data on all 294 randomized patients were analyzed according to the allocated treatment strategy. However, 14 patients had to be excluded from the analysis of the primary end point because of incomplete data or because they withdrew consent (Figure 1). The main analyses of treatment effect were based on a comparison of medians (hsTropT72-, AUC–CK-MB–, and MPS-SPECT–based end points). Secondary analyses used comparisons of frequencies (binary classification of patients according to their hsTropT72 [below or above the median] and clinical end points) and linear regression models (to adjust for baseline data to increase the precision of effect estimates).
Stratified analyses are presented according to sex, age, Killip classification, infarct localization, diabetic state, admission glucose level, and infarct type. All statistical tests used a 2-tailed, 2-sided significance level of α = .05. The analyses were performed using commercial software (SPSS, version 20; IBM) and R, version 2.14 (Vienna University of Economics and Business [www.r-project.org]).
Between July 23, 2008, and February 8, 2012, a total of 1773 patients with ACS were admitted. Plasma glucose levels exceeded 140 mg/dL in 929 patients (52.4%), who were then assessed for eligibility. Agreement to participate, signed informed consent, and randomization were achieved in 294 patients. Seven patients were withdrawn from the study by the investigators within 1 hour because of absence of ACS (n = 6) or immediate transfer to another hospital. No data were obtained in these individuals. Another 7 patients withdrew consent within 72 hours after randomization. Thus, a total of 280 patients were available for analysis, and 140 of these were randomized to intensive glucose management (Figure 1). Two hundred fifty patients underwent MPS-SPECT imaging. No patients were lost for the follow-up of clinical end points.
At baseline, patients in the control group were more likely to have a history of PCI (P = .01) and prior MI (P = .053) and tended to be more likely to have hypercholesterolemia (P = .07). Characteristics of the 2 groups were otherwise similar (Table 1). There were 218 men (77.8%) and 62 women (22.2%); the median age was 65 (56-74) years. Two hundred twenty-nine patients (81.8%) were admitted with ST-elevation MI (STEMI), of whom 45.1% had an anterior MI. Admission plasma glucose was 140 to 179 mg/dL in 188 patients (67.1%) and 180 to 288 mg/dL or higher in 92 patients (32.9%), the median admission plasma glucose was 166 (151-187) mg/dL, and 27 patients (9.6%) had previously diagnosed non–insulin-dependent diabetes mellitus.
Treatment characteristics were similar in both treatment arms (Table 2). During admission, 227 patients (99.1%) with STEMI underwent primary PCI. The time from symptom onset to admission was 120 (75-195) minutes in patients randomized to intensive glucose management and 105 (75-180) minutes in those randomized to conventional management. The time from admission to the first balloon inflation was 35 (25-50) minutes and 30 (20-50) minutes, respectively. In the patients with non-STEMI (NSTEMI), in the intensive and conventional treatment groups, respectively, 26 patients (86.6%) and 21 patients (100%) underwent catheterization during admission, with 16 procedures (61.5%) and 14 procedures (66.7%) performed within 48 hours, and PCI was performed in 68.6% of patients with NSTEMI. Overall, 262 patients (93.6%) in the sample had a PCI performed during admission, 253 of which were within 72 hours after admission.
The 140 patients in the intensive treatment arm received intravenous insulin for a median of 47 (43-48) hours. They had 29 (25-32) glucose measurements. A median glucose level of 112 (97-130) mg/dL was reached at 24 hours. Severe hypoglycemia (<50 mg/dL) was rare and occurred in 16 measurements (0.41%) in 13 patients. All episodes of hypoglycemia could be corrected with oral glucose. A severe episode of hypoglycemia occurred in 1 of the patients with a second in-hospital MI.
The time from symptom onset to the start of insulin therapy was 5.0 (3.9-7.7) hours, and the admission-to-insulin time was 2.4 (1.8-3.5) hours. In patients with STEMI, insulin was started earlier than in those with NSTEMI: 2.3 (1.7-3.0) vs 4.3 (2.5-8.2) hours after admission. Six hours after symptom onset, 40.1% of glucose levels were within the target range; this percentage had increased to 59.8% at 24 hours. Median glucose values in patients randomized to intensive glucose management were significantly lower than in those randomized to conventional management (Figure 2). However, neither the change in glucose levels nor the time to normalization was related to the hsTropT72 level, regardless of allocated therapy (Supplement [eTable]).
There were no significant differences in enzymatic infarct size between the 2 treatment groups (Supplement [eFigures 1 and 2]). Patients randomized to intensive glucose management had a median hsTropT72 of 1197 (541-2296) ng/L compared with 1354 (530-3057) ng/L in patients treated conventionally (P = .41).
We observed no significant difference between the 2 randomized groups in the percentage of patients with an hsTropT72 below the overall median value (1302 ng/L). Analysis of prespecified subgroups also did not reveal significant differences (Figure 3).
Finally, we applied linear regression with admission hsTropT and randomly allocated treatment as determinants of hsTropT72. Also in this analysis, hsTropT72 was not influenced by allocated treatment (P = .70). Further adjustment by adding age and sex as end point determinants confirmed these findings.
The median AUC–CK-MB during the first 72 hours was 2372 (1242-5004) U/L in the intensive glucose management arm, and the median value in the conventional treatment group was 3171 (1620-5337) U/L (P = .18). The differences in median extent of MPS-SPECT myocardial injury at 6 weeks (2% vs 4%; P = .07) and left ventricular function (59% vs 57%; P = .33) were similar in both trial arms. Despite randomization, there were relevant differences in the prevalence of prior PCI and MI between the study arms; this may have influenced the measurements of myocardial injury as measured by MPS-SPECT. Therefore, we applied linear regression with prior MI, prior PCI, randomly allocated treatment arm, age, and sex as determinants of myocardial injury. There was still no significant difference in MPS-SPECT infarct extent between treatment arms.
Patients were admitted for a median of 3.6 (2.9-4.9) days. During this period, 4 patients (2.9%) who were randomized to intensive glucose management died compared with 1 patient (0.7%) in the conventional arm (P = .37). A spontaneous second MI occurred in 5 patients (3.6%) receiving intensive glucose management compared with none in the conventional treatment group (P = .06). The composite end point of death or a second spontaneous MI occurred in 8 patients (5.7%) vs 1 patient (0.7%), respectively (P = .04).
In the BIOMArCS-2 Glucose trial, rapid and (near) normalization of blood glucose levels was achieved in patients with ACS and hyperglycemia by applying an early intensive glucose management strategy. The well-controlled glucose levels, however, were not accompanied by a reduction in enzymatic or scintigraphic infarct size, whereas the incidence of the composite of death or a second MI was increased.
Management of ACS has undergone several important changes in recent decades. In particular, primary PCI is now considered the first treatment option in patients with STEMI, whereas early coronary angiography and subsequent PCI are also applied in most patients with NSTEMI. In addition, antiplatelet strategies have evolved, and dual antiplatelet therapy is common.
Furthermore, out-of-hospital networks have been developed for paramedics to initiate medical therapy while transporting patients directly to the interventional clinics, thereby reducing symptom onset–to-balloon times. Such expeditious acute ACS management has been achieved in the current trial, which should be considered when interpreting the presented data.
The BIOMArCS-2 Glucose trial was successful in enrolling high-risk patients with hyperglycemia and STEMI (82%) or NSTEMI (18%), with 50% having multivessel coronary artery disease. Coronary angiography was performed in 100% vs 92% of cases, and primary PCI was accomplished in almost all patients with STEMI after a median of 30 minutes following presentation. The favorable outcome of this infarction-limiting concept of optimal ACS care has been reported recently,29,30 and may explain why our intervention designed to normalize glucose levels did not achieve any further benefits.
The BIOMArCS-2 Glucose trial achieved rapid normalization of plasma glucose levels in the insulin-treated patients. Glucose levels were significantly lower in patients receiving intensive treatment than in controls at all measured time points between 6 and 72 hours. This illustrates the effectiveness of glucose management in our study; compared with other studies aiming at strict blood glucose control, lower mean blood glucose levels were reached with even fewer hypoglycemic episodes.13,17 However, this treatment effect was not translated to a smaller enzymatic infarct size.
There are several possible reasons why rapid and adequate glucose level control was not accompanied by a reduction in enzymatic infarct size. First, the timing of insulin therapy might have been suboptimal. Intensive glucose management often was delayed until after the primary PCI was conducted. As a result of this early reperfusion, part of the metabolic stress of the earlier occlusion is resolved and several patients might have had a near-normal glucose level at the start of the insulin infusion. It is conceivable that the effect of insulin is dependent on the reperfusion-induced glucose regulation, and those with persistently elevated glucose levels after PCI might still benefit.
Second, patients were enrolled at a median of 4.1 (3.0-6.5) hours after symptom onset. Their evolving MI was terminated as a result of the early PCI. Consequently, the post-PCI evolvement of the MI might have been too small (at least enzymatically) to be influenced by the intensive glucose regulation that was started after the procedure. Preprocedural (ie, during transportation to the intervention center) reduction of glucose levels may have greater impact.
Third, given the wide variation in glucose levels that we observed until 6 hours after presentation, glucose levels appear to remain unstable early after MI onset. Furthermore, patients with persistently elevated glucose levels had somewhat larger infarct sizes than did those who obtained normal values regardless of the allocated treatment. Therefore, glucose lowering and stabilization may be more important early after symptom onset.31
Finally, it is possible that elevated plasma glucose is a marker of severity of injury and does not have a causal role in extent of myocardial infarction.
We opted for hsTropT72 as the primary end point and sought supportive evidence by the classic measures—AUC–CK-MB and MPS-SPECT—as secondary end points. One might question whether hsTropT72 was sensitive and specific enough to quantify final MI size. We chose this end point because prior studies demonstrated that troponin levels within 48 to 96 hours correlate well with final infarct size determined by cardiac magnetic resonance imaging.24- 27,32 Still, an in-depth analysis of the evolvement of cardiac troponin during the first 72 hours (and its relationship with infarct size as determined by MPS-SPECT) is warranted. As a result of the expedited management, the infarct extent in our patients with ACS treated with PCI was relatively small. Future trials in similar populations might consider positron emission tomography or magnetic resonance imaging instead of cardiac enzymes and SPECT to study infarct size. The better spatial resolution of magnetic resonance imaging, compared with MPS, may be more appropriate to detect small differences.
Hypoglycemia as a result of too-strict glucose regulation bears an important risk as has been demonstrated by a J-shaped relationship between glucose level and outcome.33 Furthermore, in the NICE SUGAR (Normoglycemia in Intensive Care Evaluation–Survival Using Glucose Algorithm Regulation) trial, patients in the intensive care unit who were receiving intensive glucose regulation had increased mortality compared with those receiving conventional glucose regulation. This increased mortality has been attributed to hypoglycemia.34 Although the incidence of severe hypoglycemia in the present trial was low (0.4% vs 6.8% in NICE SUGAR), the incidence of the composite end point of death or a second spontaneous MI was significantly increased in the patients randomized to intensive glucose management. The mechanism underlying this increased risk needs to be studied. In the meantime, this observation emphasizes that the potential benefit of intensive glucose regulation needs to be carefully weighed against the demonstrated patient risk.
The BIOMArCS-2 Glucose is a single-center study, which might limit the generalizability of its results. Furthermore, we excluded patients with insulin-dependent diabetes and patients who required mechanical ventilation. Also, the outcome of glucose management in the current BIOMArCS 2 Glucose trial and in the IMMEDIATE trial should be considered in the context of PCI delivered in a timely fashion at an experienced intervention center. The short symptom-to-balloon times and urgent PCI as appropriate in patients with NSTEMI may have contributed to limited infarct sizes.
The BIOMArCS-2 glucose trial failed to demonstrate that intensive glucose regulation is associated with a reduction in enzymatic or scintigraphic infarct size in hyperglycemic, troponin-positive patients with ACS who are undergoing PCI. In fact, we found evidence that intensive glucose regulation is associated with harm. Thus, based on our data, intensive insulin therapy is not a recommended practice.
Fifty percent of patients with ACS present with an elevated blood glucose level, which is related to adverse clinical outcomes during longer term follow-up, even after initial successful PCI treatment. Therefore, further research in this field is warranted. We believe that future studies should focus on patients with ACS who have persistently elevated blood glucose levels after PCI, and should evaluate alternative strategies for optimizing glycemia. Also, in view of the results of our trial, the relationship between (intensive) blood glucose management by intravenous insulin infusion in the first hours after ACS onset and the increased risk of an early second MI needs further exploration. In the meantime, a strategy of strict, but not too strict, glucose control, as suggested by the European Society of Cardiology guidelines,21 seems to be the best practical approach.
Accepted for Publication: June 21, 2013.
Corresponding Author: Victor A. Umans, MD, PhD, Department of Cardiology, Medical Center Alkmaar, Wilhelminalaan 12, 1815 JD Alkmaar, The Netherlands (email@example.com).
Published Online: September 9, 2013. doi:10.1001/jamainternmed.2013.10074.
Author Contributions: Drs de Mulder and Boersma had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: de Mulder, Umans, Cornel, Oemrawsingh, Akkerhuis, Boersma.
Acquisition of data: de Mulder, Cornel, van der Zant, Stam.
Analysis and interpretation of data: de Mulder, Umans, Cornel, Stam, Boersma.
Drafting of the manuscript: de Mulder, Umans, Boersma.
Critical revision of the manuscript for important intellectual content: All authors.
Obtained funding: Boersma.
Administrative, technical, or material support: de Mulder, Umans, van der Zant, Stam.
Study supervision: Umans, Cornel, Stam, Boersma.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was funded by grant No. FI 0610 from the Foreest Medical School, Alkmaar, the Netherlands, and grant No. 07101 from the Netherlands Heart Institute, Interuniversitair Cardiologisch Instituut Nederland.
Role of the Sponsors: The sponsors had no influence on design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.