OUP user menu

★ editor's choice ★

Factors related to heart rupture in acute coronary syndromes in the Global Registry of Acute Coronary Events

José López-Sendón, Enrique P. Gurfinkel, Esteban Lopez de Sa, Giancarlo Agnelli, Joel M. Gore, Phillippe Gabriel Steg, Kim A. Eagle, Jose Ruiz Cantador, Gordon Fitzgerald, Christopher B. Granger,
DOI: http://dx.doi.org/10.1093/eurheartj/ehq061 1449-1456 First published online: 15 March 2010


Aims To determine the incidence and factors associated with heart rupture (HR) in acute coronary syndrome (ACS) patients.

Methods and results Among 60 198 patients, 273 (0.45%) had HR (free wall rupture, n = 118; ventricular septal rupture, n = 155). Incidence was 0.9% for ST-segment elevation myocardial infarction (STEMI), 0.17% for non-STEMI, and 0.25% for unstable angina. Hospital mortality was 58 vs. 4.5% in patients without HR (P < 0.001). The incidence was lower in STEMI patients with primary percutaneous coronary intervention (PCI) than in those without (0.7 vs. 1.1%; P = 0.01), but primary PCI was not independently related to HR in adjusted analysis (P = 0.20). Independent variables associated with HR included: ST-segment elevation (STE)/left bundle branch block; ST-segment deviation; female sex; previous stroke; positive initial cardiac biomarkers; older age; higher heart rate; systolic blood pressure/30 mmHg decrease. Conversely, previous MI and the use of low-molecular-weight heparin and beta-blockers during first 24 h were identified as protective factors for HR.

Conclusion The incidence of HR is low in patients with ACS, although its incidence is probably underestimated. Heart rupture occurs more frequently in ACS with STE and is associated with high hospital mortality. A number of variables are independently related to HR.

  • Heart rupture
  • Acute coronary syndromes
  • Reperfusion therapy
  • Beta-blockers
  • Primary percutaneous coronary intervention

See page 1433 for the editorial comment on this article (doi:10.1093/eurheartj/ehq089)


Heart rupture (HR) is a major lethal complication of acute myocardial infarction (MI), with an incidence as high as 6% in the pre-reperfusion era,14 accounting for up to 30% of hospital mortality. Widespread availability of reperfusion therapies quickly improved outcomes of patients with acute MI, with a dramatic decrease in mortality, mechanical complications in general, and HR.3,58 Modern therapies focus on reperfusion strategies and systematic use of medications, but the contemporary prevalence of HR complications after MI is not well known. Most publications focus on HR after transmural MI, with little or no information available for this complication in other types of acute coronary syndromes (ACS).

The objective of this study was to determine the incidence of, and factors associated with, HR in patients with ACS enrolled in the Global Registry of Acute Coronary Events (GRACE).


GRACE is a multinational co-operative effort involving 117 hospitals in 14 countries (in Europe, North and South America, Australia, and New Zealand) and was designed to reflect an unbiased, representative population of ACS patients. Hospital clusters were chosen based on local demographic characteristics and hospital facilities to ensure a representative sample of patients with ACS from each country and of hospital systems of different sizes and treatment and diagnostic capabilities. Full details of the methods have been published.9 Briefly, patients had to be ≥18 years of age, be admitted for ACS as a presumptive diagnosis (i.e. have symptoms consistent with acute myocardial ischaemia), and have at least one of the following: electrocardiographic changes consistent with ACS, serial increases in serum cardiac biomarkers, and/or documentation of coronary artery disease. The qualifying ACS must not have been precipitated by significant non-cardiovascular comorbidity, such as acute anaemia, hyperthyroidism, trauma, or surgery.

At each enrolling hospital, study investigators worked with their ethics or institutional review board to obtain appropriate approval to participate.

This study was based on a cohort of 60 198 patients enrolled from January 2000 to December 2007. Patients transferred from other hospitals not participating in GRACE were included in the analysis.

Clinical variables

Information on patient demographic characteristics, medical history, timing, and occurrence of acute coronary symptoms, clinical characteristics, electrocardiographic findings, treatment approaches, and in-hospital outcomes were collected through completion of a standardized case report form. Pre-established criteria for the diagnosis of acute MI, including ST-segment elevation (STE) MI (STEMI), non-STEMI (NSTEMI), and unstable angina, were used. The diagnosis of HR was based on clinical, echocardiographic, and anatomic findings (post-mortem, surgery) according to the investigator criteria, and was specified as ventricular free wall rupture or interventricular septal rupture. No protocol for diagnosis of HR was pre-specified and the incidence of HR in this registry may thus be underestimated. Rupture of papillary muscles was not noted specifically and was not considered for this study. Treatments were decided by physicians at each site, with recommendations to follow the ACS guidelines that evolved during the time of the study.

Statistical analysis

Continuous variables are reported as medians with 25th and 75th percentiles. Differences in the distribution of selected characteristics between patient groups were examined using χ2 tests for categorical variables. Differences in continuous variables between study groups were analysed using the Wilcoxon rank-sum test.

Predictors of HR were determined by logistic regression. Candidates included GRACE risk-score10 variables (age, Killip class, systolic blood pressure, ST-segment deviation, cardiac arrest at presentation, serum creatinine, cardiac biomarkers, and heart rate), hospital cluster, type of ACS (STEMI, NSTEMI, or unstable angina), and clinical variables with P < 0.05 in univariate analysis comparing the demographic data, medical history, presenting features, and hospital treatments in patients with and without HR. Several models were explored, considering also the GRACE hospital cluster, timing of reperfusion therapies, medications received during the first 24 h, and previously reported factors related to HR.7

The relationship between HR and use and timing of reperfusion therapies was investigated in patients with STE in the first 24 h, with an indication for reperfusion therapy according to guidelines.11,12

SAS statistical software version 9.1 (SAS Institute Inc., Cary, NC, USA) was used for all analyses.


Incidence of heart rupture

Of the 60 198 patients enrolled, 273 (0.45%) developed HR: 118 (0.2%) free wall ventricular rupture and 155 (0.26%) ventricular septal rupture. The incidence was 0.9% in patients with a final diagnosis of STEMI and lower in patients with NSTEMI (0.17%), or unstable angina (0.25%).

The incidence of HR varied across the 18 hospital clusters, from 0 to 1.4% of patients included in the registry (mean 0.45%; range 0.0–1.38; P < 0.0001). The C-index was 0.65 for the regression model including only hospital cluster.


Hospital mortality rate was 58% in HR patients vs. 4.5% in those without (P < 0.001), representing 5.6% of all hospital deaths. Mortality was higher in patients with free wall ventricular rupture (80%) than in patients with ventricular septal rupture (41%). In septal rupture, mortality was higher in patients with cardiogenic shock than in patients without (100 vs. 38%, respectively), whereas in patients with free wall rupture, mortality was similar in patients with and without shock (85 vs. 79%, respectively).

Factors associated with heart rupture

Table 1 shows the demographics, medical history, presenting features, and hospital treatments in patients with and without HR. Patients with HR were typically older and women; history of stroke occurred more frequently, whereas previous history of chronic ischaemic heart disease, smoking, hyperlipidaemia, and history of revascularization procedures occurred less frequently in HR patients.

View this table:
Table 1

Demographics, medical history, presenting features, and hospital treatments in patients with and without heart rupture

No heart rupture (n = 59 925)Heart rupture (n = 273)P-value
Agea, years66 (56, 76)74 (64, 81)<0.001
Women, %3349<0.001
Weighta, kg77 (67, 88)72 (62, 84)<0.001
Medical history, %
 Smoker (current/former)57470.001
 Myocardial infarction3015<0.001
 Transient ischaemic attack/stroke8.4140.004
 Angiogram diagnostic for coronary artery disease3314<0.001
 Percutaneous coronary intervention187.4<0.001
 Coronary artery bypass graft134.8<0.001
Clinical presentation
 Cardiac arrest at presentation, %
 ST-elevation/left bundle branch block on index electrocardiogram, %3769<0.001
 Positive initial cardiac biomarkers, %4765<0.001
 Pulsea, b.p.m.76 (65, 90)80 (70, 100)<0.001
 Systolic blood pressure, mmHg140 (120, 160)130 (106, 153)<0.001
 Initial serum creatininea, mg/dL1.02 (0.90, 1.26)1.09 (0.89, 1.40)0.25
 Killip class I, %83680.001
 GRACE risk scorea129 (105, 155)161 (130, 158)<0.001
Treatment in first 24 h
 Angiotensin-converting enzyme inhibitor, %4840<0.01
 Aspirin, %89840.01
 Thienopyridine, %4230<0.001
 Beta-blocker, %7352<0.001
  IV only (n = 1978), %3.44.6
  Oral only (n = 35 653), %6141
  Both (n = 4880), %8.36.4
 Calcium antagonist, %15110.07
 Statin, %5439<0.001
 Unfractionated heparin, %37380.61
 Low-molecular-weight heparin, %5037<0.001
 Glycoprotein IIb/IIIa inhibitor, %18130.02
 Primary percutaneous intervention, %15190.11
 Primary percutaneous intervention in ST-segment elevation myocardial infarction, %38270.01
 Lytic any time (all patients), %1224<0.001
 Lytic within 24 h, %1020<0.001
 Lytic any time in ST-segment elevation myocardial infarction, %28320.28
 Lytic within 24 h in ST-segment elevation myocardial infarction, %24270.51
 Length of hospital stay, daysa5 (3, 9)4 (1, 8)<0.001
  • aMedian (interquartile range).

At admission, HR patients presented more frequently with STE and elevated cardiac biomarkers; heart rate, creatinine values, and GRACE risk score were higher, whereas systolic blood pressure was lower and Killip class I at admission occurred less frequently in HR patients. Recommended medications during the first 24 h after the index event (aspirin, thienopyridines, angiotensin-converting enzyme inhibitors, beta-blockers, statins) were used less frequently in HR patients.

Thrombolytic treatment was used more frequently in HR patients. The incidence of HR was lower in patients with STEMI with primary percutaneous coronary intervention (PCI) than in those without (0.7 vs. 1.1%; P = 0.01), but primary PCI was not independently related to HR in adjusted analysis (P = 0.21). The incidence of HR was similar in patients with STE-ACS receiving or not receiving thrombolytics during the first 24 h (1.1 vs. 0.9%; P = 0.4). Table 2 shows the percentage of patients with STE treated with thrombolytics. Time delay from onset of symptoms to thrombolytic administration was longer in HR patients and the incidence of HR increased progressively with the delay in thrombolytic administration (Figure 1), but neither thrombolytic use nor delay in administration was independently related to HR.

View this table:
Table 2

Patients with ST-segment elevation treated with thrombolytics; time delay from onset of symptoms to thrombolytic administration

Delay, hNo heart rupture (n = 6152)Heart rupture (n = 58)P-value
Lytic 0–2, %
Lytic 2–4, %35.034.5
Lytic 4–6, %13.417.2
Lytic 6–12, %9.713.8
Lytic 12–24, %3.53.5
Lytic >24, %3.28.6
Time from symptom onset to lyticb2.7 (1.7, 4.7)3.8 (2.2, 6.5)0.01
  • aMedian (interquartile range).

  • bCochran–Mantel–Haenszel test for linear trend, to compare time delay to lytics and heart rupture.

Figure 1

Patients with ST-segment elevation at presentation. Incidence of heart rupture according to reperfusion strategy and time delay to lytics (P = 0.02, unadjusted for linear trend in heart rupture with lytic delay). However, the use of lytics was not an independent predictor for heart rupture in the multivariable analysis (odds ratio 1.18, 95% CI 0.841.66, P = 0.34). PCI, percutaneous coronary intervention.

Multivariable analysis

In the model including all variables with significant differences (P ≤ 0.05) between groups (n = 54 164 patients with complete data), the following were associated independently with HR (Figure 2): age per 10-year increase; female sex; previous stroke; previous MI; STE/LBBB; ST-segment deviation; positive initial cardiac biomarkers; heart rate per 30 b.p.m. increase; systolic blood pressure per 30 mmHg; low-molecular-weight heparin; and beta-blockers during the first 24 h. The inclusion of hospital cluster made little difference to the independent variables (C-indices 0.77 and 0.75, respectively, in models with and without hospital cluster included). Factors that were not statistically significant (P > 0.05) included: history of coronary artery disease; fibrinolytics in first 24 h (no lytics, lytics 0–2 h after symptom onset, lytics >2 h after symptom onset); primary PCI within 12 h; elevated creatinine; past or current smoking; previous PCI or coronary artery bypass graft; history of dyslipidaemia; Killip class; and history of angina.

Figure 2

Factors related to heart rupture in the multiple logistic regression model. Eleven factors in final model and estimates are adjusted for hospital cluster; 54 164 patients with 239 heart ruptures. C-index, 0.75. B blockers, β-blockers; BP, blood pressure; LBBB, left bundle branch block; LMWH, low-molecular-weight heparin; MI, myocardial infarction.

Other logistic regression models

In the logistic regression model including all variables with significant differences between groups except medications received in hospital (n = 54 223 patients with complete data), the following were independent predictors of HR (C-indices 0.76 and 0.74, respectively, in models with and without hospital cluster included): female sex (OR 1.58; 95% CI 1.21–2.07); age per 10-year increase (OR 1.36; 95% CI 1.22–1.52); medical history of coronary artery disease (OR 0.67; 95% CI 0.45–1.00); chronic oral beta-blockers (OR 0.68; 95% CI 0.47–0.98); increasing Killip class (OR 1.20; 95% CI 1.01–1.43); new STE/LBBB in initial electrocardiogram (OR 2.17; 95% CI 1.55–3.05); ST-segment deviation in initial electrocardiogram (OR 1.69; 95% CI 1.14–2.49); positive initial cardiac biomarkers (OR 1.35; 95% CI 1.03–1.78); heart rate per 30 b.p.m. increase (OR 1.26; 95% CI 1.07–1.48); systolic blood pressure per 30 mmHg decrease (OR 1.18; 95% CI 1.08–1.28).

In the logistic regression model including only previously reported risk factors for HR7 (n = 51 875 patients), the following were independent predictors of HR (C-index 0.71): increasing Killip class (OR 1.26; 95% CI 1.04–1.52); ST-segment deviation (OR 3.47; 95% CI 2.47–4.85); positive initial cardiac biomarkers (OR 1.43; 95% CI 1.08–1.89); age per 10-year increase (OR 1.36; 95% CI 1.23–1.52); and systolic blood pressure per 30 mmHg decrease (OR 1.21; 95% CI 1.10–1.32).


This is the first international study to explore the incidence of HR in a large, unselected population including the full spectrum of patients with ACS. Heart rupture occurred more frequently in patients with STEMI (0.9%) than in patients with other forms of ACS (NSTEMI, 0.17%; unstable angina, 0.25%). The true incidence was probably underestimated; although the overall rate of HR was 0.5%, in some hospital clusters it was as high as 1.4%, while in others no cases of HR were reported. Awareness, special interest in this pathology, routine immediate echocardiographic evaluation of patients with hypotension or shock, and careful screening of patients with electromechanical dissociation may influence the relative incidence of HR in different hospitals. Ventricular septal rupture, easier to diagnose but with a much lower prevalence in post-mortem studies,35,13 was reported in about the same number of patients as free wall ventricular rupture (0.26 and 0.20%), underlining the importance of awareness when identifying patients presenting with HR.14

Heart rupture after STEMI was the focus of attention in most of the studies, and its incidence decreased with the use of modern therapies. In some reports during the pre-reperfusion era, HR occurred in as many as 6% of all cases admitted for transmural MI3 and decreased progressively through the years.3,58 Figueras et al.,3 in an overview of HR in a single hospital during a 30-year period, reported an initial prevalence of 6% before 1982, with a progressive decline to 3.2% during 2001–6, despite a trend towards increasing age (one of the most powerful risk factors for HR). With the widespread use of thrombolytics, PCI, and modern therapies, the prevalence of HR decreased dramatically, and most of the contemporary studies, including large registries and clinical trials, report an incidence around 1%, similar to that found in the GRACE registry,5,7,15,16 although, again, awareness of and a systematic attitude towards diagnosis increase the incidence of HR; in some series of patients reported from single institutions with a high interest in mechanical complications, the incidence of HR in STEMI varied from 2 to 3%8,1719 while in large multicentre studies and clinical trials, the number of patients presenting with HR was lower. Moreno et al.,8 in a series of 1375 consecutive cases treated with primary PCI or thrombolysis in a single institution, reported an incidence of 2.5%, considering only cases with a direct diagnosis of left ventricular free wall rupture.


Despite advances in diagnostic procedures and surgical techniques, hospital mortality remains high in patients with left ventricular free wall rupture (80%) or ventricular septal rupture (41%), representing 5.6% of all hospital deaths in the GRACE registry. Similar or higher mortality rates have been observed in other contemporary studies.2023 Heart rupture, in particular left ventricular wall rupture, is considered to be a hopeless complication after MI, although the survival rate is much higher than was expected some years ago.

Factors related to heart rupture

Heart rupture occurred more frequently in women, in older patients, and in those without a history of MI; HR patients had STE and positive cardiac biomarkers more frequently, and a higher heart rate, lower blood pressure, and higher GRACE risk score at admission than patients without HR. These factors have also been reported previously.3,58,15,1719,24 HR patients were treated less frequently with beta-blockers, aspirin, statins, and angiotensin-converting enzyme inhibitors before the ACS episode, possibly reflecting a lower incidence of chronic ischaemic disease, but during the first day after admission to the hospital, HR patients also received fewer medications recommended in the current guidelines (aspirin, statins, thienopyridines, low-molecular-weight heparin, beta-blockers) than patients without HR.

In the multivariable analysis, age, female sex, previous stroke, ST-segment deviation, heart rate, and blood pressure at presentation, and positive initial cardiac biomarkers were independently related to a higher rate of HR during hospitalization for ACS. Conversely, a history of MI and the early use of beta-blockers during the first 24 h after admission and low-molecular-weight heparin were associated with a lower probability of HR. The GRACE risk score is associated with heart rupture in an unadjusted model (OR 1.20, 95% CI 1.16–1.23, c-index 0.69). However, it is superfluous and is not statistically significant in the final 11-factor model (c-index = 0.75), which includes five factors contained in the GRACE risk score. While the GRACE risk score is associated with heart rupture, it was developed to predict hospital death (c-index = 0.8310), and should probably not be used to predict in-hospital heart rupture.

Of all the factors related to heart rupture, age is probably the most relevant. Older age was invariably reported in all trials as the leading risk factor for HR. In some reports focusing specifically on age, the incidence of HR among patients aged >70 years was found to be as high as 10%.25 In the Gruppo Italiano per lo Studio della Sopravvivenza nell'Infarto Miocardico (GISSI-2) trial, HR found post mortem increased progressively with age, from 19% among those aged ≤60 years to 58% among those aged 61–70, and to 86% among those aged >70.26 The powerful influence of age on HR could explain some of the differences between clinical trials that included patients of different age ranges.

Heart rupture is related to a complete and maintained coronary artery occlusion without collateral circulation.27,28 This explains why HR occurs more frequently in first myocardial infarcts with STE, and represents an opportunity for reperfusion therapy. The possible role of reperfusion therapy to prevent or facilitate ventricular rupture has been the focus of interest in many studies and the role of thrombolytic therapy has always been controversial. Early thrombolysis and reperfusion of the infarct-related artery might play a protective role, whereas late reperfusion may facilitate intramyocardial haemorrhage and rupture. Thrombolytic administration may lead to an inadequate myocardial reperfusion and reperfusion injury, resulting in HR.29 In some studies, administration of thrombolytics has been related to an increase in HR,3032 while in others, HR was observed less frequently.33,34 Other than in a few trials,6 multivariable analysis failed to identify the use of lytics as an independent risk factor for HR3,17,19,3537—an observation repeated in our present study. The presence of other risk factors for HR, in particular older age and delay to reperfusion, seems to play an important role. Figure 1 shows the steady increase in HR with delay in lytic therapy. An elegant meta-analysis from the Duke University examined the relationship between the risk of HR and time delay to thrombolysis in randomized clinical trials.35 The odds ratio for HR comparing treated patients with control patients demonstrated correlation with time to treatment, with a decrease in risk of rupture in the first 7 h of infarct evolution and a significant increase after 17 h. The analysis of data from the GISSI trial confirmed the relationship between time to thrombolytic therapy and risk of HR independently,38 but the hypothesis was not confirmed in the Late Assessment of Thrombolytic Efficacy (LATE) trial.5 Age may also play an important role and late thrombolysis in older patients yields an extremely high risk for HR.25,38

Reperfusion with primary PCI seems to provide protection against HR, independent of other factors.3,8,17,18,37 Primary angioplasty achieves the restoration of coronary patency more frequently than thrombolysis and the risk of bleeding is lower—two important factors to explain why mechanical complications occur less frequently with PCI than with thrombolysis.17 However, failed reperfusion is associated with an increased risk.17,19 Nakatani et al.,17 in a series of 2209 patients with STEMI treated with PCI, found that the risk ratio for mechanical complications increased in the failed reperfusion group compared with that in the successful reperfusion group, even when PCI was performed 12–24 h after symptom onset. In our present study, the incidence of HR was lower in STEMI patients treated with primary PCI than in those without (0.7 vs. 1.1%; P = 0.01) but the procedure was not independently related to HR in adjusted analysis. The use of other clinical variables protecting against HR could explain these findings.

Early use of recommended medical therapies was higher in patients without HR (Table 1). Beta-blockers and low-molecular-weight heparin were identified as independent factors related to lower risk of HR (Figure 2). In previous studies, HR occurred less frequently in patients receiving nitrates,39 beta-blockers,40 and statins,15 highlighting the importance of using appropriate medical treatment in patients with acute MI. In the GRACE registry, improvement in the use of recommended therapies was associated with a progressive decline in mortality,41 and although this relationship was not explored for the effect on HR, the early use of beta-blockers (intravenous or oral) was clearly related to a lower risk of HR. We have not explored, however, the relative role of the beta-blocker route of administration.

Study limitations

As an observational study, GRACE is subject to certain inherent limitations and potential biases, including collection of non-randomized data, missing or incomplete information, and potential confounding by drug indication or other unmeasured covariates that must be considered when interpreting the results. Heart rupture was not included as a major objective of GRACE, there were no rigorous diagnostic criteria, its diagnosis was not emphasized, and the significant variation in HR prevalence between participating hospitals may well reflect different levels of awareness of the problem. Most importantly, the exact timing of HR was not recorded on the GRACE case report forms; some of the patients may present with HR at admission14,42 and that will be an extra confounding factor for some clinical variables related to rupture e.g. blood pressure or Killip class at admission or early treatment, including medical treatments such as beta-blockers, low-molecular-weight heparin, and reperfusion therapies, including thrombolytic treatment and percutaneous coronary revascularization. Some factors associated previously with HR, such as the presence of collateral circulation, dynamic electrocardiogram changes, and clinical manifestations after admission, were not explored in this study. In addition, new biomarkers possibly related to a predisposition to HR—factor XIII,43 content of insoluble and soluble collagen, expression and activity of matrix metalloproteinases and density of inflammatory cells44,45—may provide further information related to the risk of HR and may modify the relative value of other clinical risk factors.


In the reperfusion era, HR is an uncommon complication in ACS, although the real incidence is probably underestimated in this retrospective, observational study, with significant variations of incidence between hospital clusters. Heart rupture occurs more frequently, but not exclusively, in patients with ACS with ST elevation. Female sex, age, previous stroke, ST elevation, positive cardiac biomarkers, and heart rate at presentation were independently related to HR. History of MI and early use of beta-blockers and low-molecular-weight heparin were associated with a lower incidence.


GRACE is supported by a grant from sanofi-aventis to the Center for Outcomes Research, University of Massachusetts Medical School.

Conflict of interest: sanofi-aventis had no involvement in the collection, analysis, and interpretation of data, in the writing of the manuscript, or in the decision to submit the paper for publication. The design, conduct, and interpretation of the GRACE data are undertaken by an independent steering committee. J.L.-S., E.P.G., E.L.S., G.A., J.M.G. received reseach grant (modest) from sanofi-aventis. P.G.S. received research grant from sanofi-aventis. Speakers bureau (all modest) from Boehringer-Ingelheim, BMS, GSK, Medtronic, Nycomed, sanofi-aventis, Servier, The Medicines Company; Consulting/advisory board (all modest): Astellas, AstraZeneca, Bayer, Boehringer-Ingelheim, BMS, Endotis, GSK, Medtronic, MSD, Nycomed, sanofi-aventis, Servier, The Medicines Company; Stockholding: none. K.A.E. received research grant from Biosite, Bristol-Myers Squibb, Blue Cross Blue Shield of Michigan, Hewlett Foundation, Mardigian Fund, Pfizer, sanofi-aventis, Varbedian Fund; Consultant/Advisory Board: NIH NHLBI, Pfizer, sanofi-aventis, Robert Wood Johnson Foundation. C.B.G.: Alexion, Astra Zeneca, Boehringer Ingelheim, Bristol Myers Squibb, decode Genetics, Genentech, GlaxoSmithKline, Novartis, Proctor and Gamble, Sanofi-aventis, The Medicines Company, INO Therapeutics, Medicure, Proctor and Gamble.


The authors thank the physicians and nurses who are participating in GRACE. Sophie Rushton-Smith, PhD, provided editorial support on the final version of this manuscript and was funded by sanofi-aventis. To find out more about GRACE, visit the website at: www.outcomes.org/grace.


View Abstract