Evaluating the Performance of the Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines (CRUSADE) Bleeding Score in a Contemporary Spanish Cohort of Patients With Non–ST-Segment Elevation Acute Myocardial Infarction
Background— The Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines (CRUSADE) model provides a risk score that predicts the likelihood of major bleeding in patients hospitalized for non–ST-elevation acute myocardial infarction. The aim of the present work was to evaluate the performance of this model in a contemporary cohort of patients hospitalized for non–ST-elevation acute myocardial infarction in Spain.
Methods and Results— The study subjects were 782 consecutive patients admitted to our center between February 2004 and June 2009 with non–ST-elevation acute myocardial infarction. For each patient, we calculated the CRUSADE risk score and evaluated its discrimination and calibration by the C statistic and the Hosmer-Lemeshow goodness-of-fit test, respectively. The performance of the CRUSADE risk score was evaluated for the patient population as a whole and for groups of patients treated with or without ≥2 antithrombotic medications and who underwent cardiac catheterization or not. The median CRUSADE score was 30 points (range, 18 to 45). A total of 657 patients (84%) were treated with ≥2 antithrombotic, of whom 609 (92.7%) underwent cardiac catheterization. The overall incidence of major bleeding was 9.5%. This incidence increased with the risk category: very low, 1.5%; low, 4.3%; moderate, 7.8%; high, 11.8%; and very high, 28.9% (P<0.001). For the patients as a whole, for the groups treated with or without ≥2 antithrombotics, and for the subgroup treated with ≥2 antithrombotics who did or did not undergo cardiac catheterization, the CRUSADE score showed adequate calibration and excellent discriminatory capacity (Hosmer-Lemeshow P>0.3 and C values of 0.82, 0.80, 0.70, and 0.80, respectively). However, it showed little capacity to discriminate bleeding risk in patients treated with ≥2 antithrombotics who did not undergo cardiac catheterization (C=0.56).
Conclusions— The CRUSADE risk score was generally validated and found to be useful in a Spanish cohort of patients treated with or without ≥2 antithrombotics and in those treated with or without ≥2 antithrombotics who underwent cardiac catheterization. More studies are needed to clarify the validity of the CRUSADE score in the subgroup treated with ≥2 antithrombotics who do not undergo cardiac catheterization.
Received November 23, 2009; accepted March 24, 2010.
Hemorrhagic complications are the most common nonischemic complications encountered in patients with acute coronary syndrome (ACS). The frequency of major hemorrhaging oscillates between 2% and 9% across the spectrum of ACS without ST-segment elevation, largely depending on the definition and the type of treatment used, particularly the dose of antithrombotic agents prescribed and the invasive procedures undertaken.1–3 In ACS, major hemorrhaging is associated with a number of important risks such as death and (re)infarction.2,4,5 Determining the net benefit of aggressive treatment based on the administration of multiple antithrombotic therapies and invasive procedures in patients with ACS and estimating a priori the likelihood of hemorrhagic complications in relation to the treatment provided can be complicated by the lack of quantitative tools able to suggest which treatment strategy might be the most appropriate. Recently, the Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines (CRUSADE) study researchers1 developed a logistic model to estimate the patient’s likelihood of having an in-hospital major bleeding event. The CRUSADE bleeding score considers baseline patient characteristics (female sex, history of diabetes, prior vascular disease), admission clinical variables (heart rate, systolic blood pressure, signs of congestive heart failure), and admission laboratory values (hematocrit, calculated creatinine clearance). This score was found to perform consistently across the postadmission treatment subgroups (eg, invasive care, use of antiplatelet and/or anticoagulants) in patients with non–ST-elevation acute myocardial infarction (NSTEMI). The development of the CRUSADE score involved >80 000 patients recruited between 2001 and 2006 in >400 hospitals across the United States. However, given the differences in patient characteristics and medical assistance (ie, different health systems) resulting from geographic location,6,7 when such a model is to be used outside the environment in which it was created, it first needs to be validated for its new context; only then can users be sure that the scores provided are not misleading.8,9
Clinical Perspective on p 2426
The aim of the present work was to assess the CRUSADE model in a contemporary Spanish cohort of patients with NSTEMI largely treated with a strategy of invasive coronary angiography and revascularization, checking its calibration and discriminatory capacity to determine whether it correctly predicts the likelihood of major bleeding.
Although this was a retrospective study, the demographic, clinical, and angiographic data, as well as data relating to management and in-hospital complications, were collected prospectively and recorded on a computer database for ACS patients admitted to our institution. To comply with the definition of major bleeding adopted by the CRUSADE registry, the lowest hematocrit values available for each patient were collected retrospectively. The study subjects were all patients consecutively admitted to our hospital with a definitive diagnosis of NSTEMI between February 2004 and June 2009. The initial cohort was made up of 855 patients. Those who died within 48 hours (n=6; 0.7%; none died of hemorrhage) were excluded. Twenty-seven patients (3.2%) were excluded because they were already being treated with oral anticoagulant drugs, a treatment strategy for which the CRUSADE model has not been tested. Forty of the 822 remaining patients (4.9%) were excluded because no CRUSADE scores or no information on major bleeding during hospitalization was available. The final cohort was therefore composed of 782 patients. They were treated according to the criteria of their attending physicians, following the assistance protocol for ACS approved by our unit, which is inspired by the clinical guidelines of the European Society of Cardiology.10,11 The study was approved by the Clinical Research Ethics Committee of our hospital.
Variables and Definitions
NSTEMI was defined as the presence of suggestive symptoms, including a serum cardiac troponin I concentration of ≥0.2 ng/dL measured with the Dimension and Flex kits (Siemens Healthcare Diagnostics Inc, Newark, NJ) and/or dynamic changes in the ST segment (depression ≥1 mm or elevation lasting <20 minutes) in ≥2 contiguous leads. Major bleeding was defined in line with the definition used in the CRUSADE study,1 ie, intracranial bleeding, documented retroperitoneal bleeding, a fall in hematocrit of ≥12% (baseline to nadir), or any red blood cell transfusion in which baseline hematocrit was ≥28% or <28% with clinically documented bleeding. In patients who underwent revascularization surgery, only major bleeding events that occurred before the intervention were taken into account. Renal clearance of creatinine was calculated with the Cockcroft-Gaut formula.12 The definition of prior vascular disease was also that adopted by the CRUSADE registry (prior stroke and/or peripheral artery disease).
Quantitative variables were expressed as medians and interquartile ranges (IQRs); qualitative variables were expressed as frequencies and percentages. The Mann-Whitney U test was used to compare quantitative variables; the χ2 or Fisher exact test was used as required to compare qualitative variables. The Wilcoxon signed-rank test was used to compare the median values recorded for the present subjects with those recorded for the CRUSADE cohort.
Calculation of the CRUSADE Score
The final CRUSADE score was calculated for each patient from the corresponding scores for the 8 prognostic variables it involves (see Appendix A in the online-only Data Supplement). Five categories of risk of major bleeding during hospitalization were established from the cutoff points and intervals defined by the CRUSADE researchers: very low, ≤20 points; low, 21 to 30; moderate, 31 to 40; high, 41 to 50; and very high, >50.
The calibration of the model was assessed with the Hosmer-Lemeshow goodness-of-fit test.13 This test is commonly used to validate models that have just been developed, but it is equally useful for validating (using an external database) existing logistic models such as the CRUSADE model. This test determines how closely the predicted event rate approximates the observed event rate over a range of scores. A significant value of P indicates a lack of fit.
The component variables of the model and the total risk score were entered into separate logistic regression models to test their association with major bleeding. The Hosmer-Lemeshow statistic obtained from regression modeling was used as an indicator of goodness of fit for the ability of the total score to predict major bleeding.
The discriminatory capacity of the CRUSADE score, a measure of its ability to distinguish between patients who will and who will not suffer in-hospital major bleeding, was determined by calculating the C statistic, which is equivalent to the area under the receiver-operating characteristic curve.14 A model with a C statistic >0.70 is generally considered to have acceptable discriminatory capacity. Both the calibration of the model and its discriminatory capacity were assessed with respect to the entire patient population, the risk categories established, the patients receiving or not receiving ≥2 antithrombotic agents (antiplatelets such as aspirin or clopidogrel, anticoagulants, or glycoprotein IIb/IIIa inhibitors), and those treated with ≥2 antithrombotic agents who did and did not undergo cardiac catheterization. After the predictive accuracy of the model in the total sample was analyzed, its performance was assessed including the 27 patients previously on oral anticoagulants therapy and those in the following subgroups: age ≥75 versus <75 years and sex.
Significance was set at P<0.05. All calculations were made with SPSS version 15.0 software (SPSS Inc, Chicago, Ill) and the BSDA and STATS libraries in the R Free software version 18.104.22.168
Table 1 shows the differences between the present patients and those used to develop the original CRUSADE model. The values of 3 of the 8 variables involved in the final score differed significantly between the present and the CRUSADE populations: sex (26.7% compared with 39.8% women; P<0.001), heart rate (77 bpm [IQR, 66 to 90 bpm] compared with 83 bpm [IQR, 70 to 98 bpm]; P<0.001), and systolic blood pressure at admission (136 mm Hg [IQR, 120 to 154 mm Hg] compared with 144 mm Hg [IQR, 124 to 165 mm Hg]; P<0.001). The values and distribution of the remaining prognostic variables included in the CRUSADE score were similar.
Distribution of Patients in the Different Risk Categories
The median CRUSADE score calculated from the scores for the 8 variables was 30 points (IQR, 18 to 45; range, 1 to 86 points). The distribution of patients across 3 of the 5 risk categories was similar to that observed in the CRUSADE study. However, in the present study, 33.6% and 18.2% of patients belonged to the very low and very high categories, respectively, compared with 28% and 21.8% in CRUSADE (Table 1).
Patient Management and In-Hospital Events
Patient management and the in-hospital events recorded are given in Table 1. In total, 657 patients (84%) were treated with ≥2 antithrombotics. The proportion of cardiac catheterization was higher than that observed in the CRUSADE study (92.3% compared with 89.0%; P=0.004). However, the in-hospital revascularizations were similar to those recorded in CRUSADE (71.6% compared with 73.4%; P=0.27).
Forty patients (5.1%) died in hospital. The mortality rate of patients who suffered major bleeding was higher than that of patients who suffered no such event (17.6% compared with 3.8%; P<0.001). The same was true for the rates of heart failure (25.5% compared with 4.1%; P<0.001) and cardiogenic shock (5.4% compared with 1%; P<0.001). Thirty-five women (16.7%) suffered a major bleeding event compared with 39 men (6.8%; P<0.001).
Incidence and Causes of Major Bleeding
The overall incidence of major bleeding was 9.5%. The incidence of bleeding showed a consistent gradient of risk across the 5 risk categories (Figure 1). The patients treated with ≥2 antithrombotics (657, 84%) had an incidence of major bleeding of 9.7% compared with 8% in those treated with <2 antithrombotics. The incidence of major bleeding stratified by risk category and treatment group (with or without ≥2 antithrombotics) was incremental (Appendix B in the online-only Data Supplement). Cardiac catheterization was performed in 609 patients (92.3%) treated with ≥2 antithrombotics, of whom 60 (9.9%) suffered major bleeding. In addition, the rate of major bleeding in patients treated with ≥2 antithrombotics who underwent cardiac catheterization increased with increasing risk category (1.4%, 3.8%, 9.2%, 14.4%, and 30.7% for the very low, low, moderate, high, and very high risk categories, respectively; P<0.01 for all comparisons). Four of the 48 patients (8.3%) treated with ≥2 antithrombotics who did not undergo an invasive approach suffered major bleeding (all in the very high risk category).
The proportions seen for the components defining major bleeding (nonexclusive) according to the CRUSADE study were 41.2%, 52.7%, 17.6%, 1.4%, and 2.7%, respectively, for a fall in hematocrit of ≥12%, any red blood cell transfusion in which the baseline hematocrit was ≥28%, any in which the baseline hematocrit was <28% in the presence of clinically documented bleeding, and intracranial and retroperitoneal hemorrhage.
Calibration and Discrimination
The calibration of the model was excellent for the patients as a whole, for the groups treated with or without ≥2 antithrombotics, in those treated with ≥2 antithrombotics who underwent cardiac catheterization, in both sexes, and regardless of whether the patients were ≥75 years of age (Table 2). In the subgroup of patients treated with ≥2 antithrombotics who did not undergo cardiac catheterization (n=48, 6.1%), the observed rate of major bleeding was of 8.33% (4 of 48 patients) with a predicted rate of 5.56%. The major bleeding predicted by the model closely approximated the observed values in 4 of the 5 risk categories. Major bleeding was underestimated in the patients in the very high risk category, but the number was small (2.7%; Figure 2). The model showed adequate discriminatory capacity for the patients as a whole (C=0.82; 95% confidence interval [CI], 0.770 to 0.867; Figure 3). It also showed good discriminatory power for the groups treated with antithrombotics (C=0.80; 95% CI, 0.749 to 0.854 for those treated with ≥2 antithrombotics; and C=0.70; 95% CI, 0.524 to 0.875 for the group receiving <2 antithrombotics). For the subgroup of patients treated with ≥2 antithrombotics who underwent cardiac catheterization, the C value was 0.80 (95% CI, 0.742 to 0.852). The model also had good discriminatory capacity in both sexes (C=0.77; 95% CI, 0.688 to 0.842 for men; and C=0.75; 95% CI, 0.669 to 0.823 for women; Table 2). However, for the subgroup receiving ≥2 antithrombotics who did not undergo an invasive approach, the discriminatory capacity of the model was poor (C=0.56; 95% CI, 0.210 to 0.915).
Of the 27 patients excluded because they were already treated with oral anticoagulant, 4 (14.8%) suffered major bleeding. At admission, the median value of the international normalized ratio was 2.3 (IQR, 2.0 to 3.5). When the model was rerun with these 27 patients included, the main result was almost unchanged, with a C statistic of 0.79 (95% CI, 0.753 to 0.851).
Table 3 shows the differences between the odds ratios for major bleeding reported in CRUSADE and the respective odds ratios in our study using a multivariable model with the CRUSADE component variables. In our validation cohort, every component variable of the CRUSADE risk score except systolic blood pressure ≥180 mm Hg at admission was independently associated with major bleeding.
In the present study, a validation of the CRUSADE score for predicting major bleeding in a contemporary cohort of Spanish patients with NSTEMI is reported. In general, its results show that the major bleeding rate predicted by the models closely approximated the observed rate. Similarly, the model had excellent discriminatory capacity overall. Moreover, the results were consistent when the study population was stratified according to advanced age and sex, considering that the rate of some of these population characteristics differed from those in CRUSADE cohort.
Given the importance of female sex in major bleeding, as reported in CRUSADE and earlier studies,1,4,16 the validation analysis of the model was performed in both men and women. Despite the smaller proportion of women in the present study compared with CRUSADE, neither the discriminatory capacity nor the calibration of the model was affected when stratified by sex. Similarly, in agreement with other contemporary Spanish ACS registries,17 the average age in our population was higher than in the CRUSADE cohort. Despite this difference and considering that this variable was not retained in the final CRUSADE model, the present study shows that one can robustly predict major bleeding in NSTEMI patients regardless of whether they are ≥75 years of age. Although not reported, it is likely that renal dysfunction, a condition highly correlated with older age, could account for much of the age-related bleeding risk. In this study and regardless of whether this hypothesis is completely or partially true, the accuracy of the model was not reduced in these 2 subgroups of patients.
The discriminatory performance in the subgroup of patients treated with ≥2 antithrombotics who did not undergo cardiac catheterization was poor. This finding may be related to the low rate of patients with this condition (n=48). Perhaps for the same reason in the CRUSADE study, the discriminatory capacity of the score for patients of this particular subgroup was also rather modest (C=0.68). However, in the remaining subgroups of patients, relative to the antithrombotic treatment and catheterization management, the discriminatory capacity of the model seen in the present work agrees with that obtained in the original CRUSADE study.1 With regard to oral anticoagulation, although the discriminatory accuracy of the model was unchanged when it was rerun in a sensitivity analysis including the patients on this therapy (C=0.79 compared to 0.82; P=0.18), this result should be taken with caution given the small number of patients in this subgroup.
The variable systolic blood pressure ≥180 mm Hg at admission was the only one not associated with major bleeding in the multivariate analysis. The adequate control of systolic blood pressure in the present cohort during the postadmission period may have limited the association between higher systolic blood pressure and risk of major bleeding, but no information on this point was recorded.
Patient Management and In-Hospital Events
Although compared with the CRUSADE cohort our population had a higher rate of patients treated with ≥2 antithrombotic drugs and a higher rate of cardiac catheterizations, the rate of major bleeding in our study was very similar to that in CRUSADE (9.5% versus 9.4%). Major bleeding event rate was higher among those patients on ≥2 antithrombotic drugs (9.75% versus 8%), similar to that reported in CRUSADE (8.2% versus 6.9%), and it was especially high among those on ≥2 antithrombotic drugs who underwent cardiac catheterization (9.9%), with a rate different from that reported in CRUSADE for this subgroup (6.9%). In any case, the observation that the incidence of major bleeding increased with the CRUSADE score was consistent for the patient population as a whole and across all the groups, in agreement with observations by the CRUSADE researchers.
As expected, adverse cardiovascular events correlated well with major bleeding events. Thus, the rate of in-hospital heart failure and cardiogenic shock was greater among those who suffered major bleeding than among those who did not, similar to that observed in CRUSADE.1 In addition, in-hospital mortality was almost 5 times higher among patients who suffered major bleeding, a finding also similar to that observed in CRUSADE.1 Although the incidence of in-hospital death in our cohort doubled that recorded in CRUSADE (5.1% compared with 2.7%), this value agrees with that recorded in other registries.17,18 The higher age of those patients with high comorbidity rates such as diabetes and peripheral vascular disease might explain, at least in part, the higher in-hospital mortality in the present study.
The performance of the model was good in most of the analyses undertaken despite some differences between the original derivation cohort of CRUSADE and the present population. This is not surprising considering that the CRUSADE score was developed from a cohort of “real-life” patients from hundreds of hospitals, thus reflecting different practice patterns, which makes its generalizability considerably high and consequently supports the validity of the CRUSADE score in many different contexts. In addition, most of variables involved in the CRUSADE model have clearly been shown to be powerful independent bleeding risk factors in many different scenarios,4,11,19 which was also the case in the present population, except for systolic blood pressure ≥180 mm Hg at admission.
The widespread use of aggressive management in patients with NSTEMI in recent years, involving multiple antithrombotic therapies and invasive procedures, has been accompanied by a significant increase in the incidence of bleeding, sometimes leading to serious complications and even death. The use of the CRUSADE score in this scenario can accurately identify those patients with a high likelihood of major bleeding, which may help physicians balance the benefits and risks of using these therapeutic strategies and thus select the most appropriate antithrombotic regimen. Furthermore, accurate assessment of the baseline bleeding risk with the CRUSADE score may be useful in the selection of both the optimal choice of antiplatelet treatment and those strategies that appear to reduce the risk of bleeding in NSTE ACS patients.2,20–22 The CRUSADE score could be also useful if new anticoagulant therapies such as bivalirudin and fondaparinux are considered in this setting. Although these anticoagulants have demonstrated an antiischemic efficacy similar to that of standard therapy but with lower bleeding rates,23–25 the experience accumulated with them is by far much less important than with heparin, and their cost-effectiveness is still open to question.26–29 Thus, before the widespread use of these new agents in NSTEMI patients, the CRUSADE score can be a useful tool to properly select those patients at risk of bleeding who presumably may benefit more from these therapies. Ideally, a consideration of bleeding risk/clinical benefit balance should be included in clinical recommendations and guidelines for the treatment of ACS. The CRUSADE score might prove to be a good candidate instrument for assessing such a balance.
The sample size is a critical limitation in this study, particularly with respect to the validity of the model in the subgroup treated with ≥2 antithrombotic drugs who did not undergo cardiac catheterization. A study with a larger sample size and more registered events (major bleeding) would provide more power. Another limitation of the study is its single-center nature. Only patients admitted to our unit, which is equipped to perform coronary angiography and coronary revascularizations, were included; the applicability of the present results should therefore be viewed with caution in centers with other types of populations and medical facilities. However, single-center studies offer the advantage of evaluating homogeneous populations and care processes, unlike multicenter studies, which often differ in the availability of their logistical resources and management habits. Finally, one of the limitations of any risk score is that although it may discriminate well at the group level, it may not necessarily discriminate well at the individual level.30
The CRUSADE risk score for the prediction of in-hospital major bleeding in patients hospitalized for NSTEMI was assessed and generally validated in this study. In our population, it would appear to be good practice to include the CRUSADE score in reports on patients hospitalized for NSTEMI. However, more studies are needed to clarify the validity of the CRUSADE score in the subgroup treated with ≥2 antithrombotic drugs who do not undergo cardiac catheterization.
We greatly appreciate the editing support of Gaietá Permenyer-Miralda, MD, PhD, in the preparation of this article.
Subherwal S, Bach RG, Chen AY, Gage BF, Rao SV, Newby LK, Wang TY, Gibler WB, Ohman EM, Roe MT, Pollack CV Jr., Peterson ED, Alexander K. Baseline risk of major bleeding in non-ST-segment-elevation myocardial infarction: the CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines) bleeding score. Circulation. 2009; 119: 1873–1882.
Rao SV, Eikelboom JA, Granger CB, Harrington RA, Califf RM, Bassand JP. Bleeding and blood transfusion issues in patients with non-ST-segment elevation acute coronary syndromes. Eur Heart J. 2007; 28: 1193–1204.
Eikelboom JW, Mehta SR, Anand SS, Xie C, Fox KA, Yusuf S. Adverse impact of bleeding on prognosis in patients with acute coronary syndromes. Circulation. 2006; 114: 774–782.
Moscucci M, Fox KA, Cannon CP, Klein W, López-Sendón J, Montalescot G, White K, Goldberg RJ. Predictors of major bleeding in acute coronary syndromes: the Global Registry of Acute Coronary Events (GRACE). Eur Heart J. 2003; 24: 1815–1823.
Fox KA, Goodman SG, Klein W, Brieger D, Steg PG, Dabbous O, Avezum A. Management of acute coronary syndromes: variations in practice and outcome: findings from the Global Registry of Acute Coronary Events (GRACE). Eur Heart J. 2002; 23: 1177–1189.
Fox KA, Goodman SG, Anderson FA Jr, Granger CB, Moscucci M, Flather MD, Spencer F, Budaj A, Dabbous OH, Gore JM, for the GRACE Investigators. From guidelines to clinical practice: the impact of hospital and geographical characteristics on temporal trends in the management of acute coronary syndromes: the Global Registry of Acute Coronary Events (GRACE). Eur Heart J. 2003; 24: 1414–1424.
Brindle P, Beswick A, Fahey T, Ebrahim S. Accuracy and impact of risk assessment in the primary prevention of cardiovascular disease: a systematic review. Heart. 2006; 92: 1752–1759.
Bertrand ME, Simoons ML, Fox KA, Wallentin LC, Hamm CW, McFadden E, De Feyter PJ, Specchia G, Ruzyllo W, for the Task Force on the Management of Acute Coronary Syndromes of the European Society of Cardiology. Management of acute coronary syndromes in patients presenting without persistent ST-segment elevation. Eur Heart J. 2002; 23: 1809–1840.
Bassand JP, Hamm CW, Ardissino D, Boersma E, Budaj A, Fernandez-Aviles F. Fox KA, Hasdai D, Ohman EM, Wallentin L, Wijns W. Guidelines for the diagnosis and treatment of non-ST-segment elevation acute coronary syndromes: the Task Force for the Diagnosis and Treatment of Non-ST-Segment Elevation Acute Coronary Syndromes of the European Society of Cardiology. Eur Heart J. 2007; 28: 1598–1660.
Lemeshow S, Hosmer D. A review of goodness of fit statistic for use in the development of logistic regression models. Am J Epidemiol. 1982; 115: 92–106.
R Development Core Team (2009). R: A Language and Environment for Stadistical Computing. Vienna, Austria: R Foundation for Statistical Computing. Available at: http//www.R-project.org. Accessed November 1, 2009.
Nikolsky E, Mehran R, Dangas G, Fahy M, Na Y, Pocock SJ, Lincoff AM, Stone GW. Development and validation of a prognostic risk score for major bleeding in patients undergoing percutaneous coronary intervention via the femoral approach. Eur Heart J. 2007; 28: 1936–1945.
Ferreira-González I, Permanyer-Miralda G, Marrugat J, Heras M, Cuñat J, Civeira E, Arós F, Rodríguez JJ, Sánchez PL, Bueno H, for the MASCARA study research team. MASCARA (Manejo del Síndrome Coronario Agudo, Registro Actualizado) study: general findings. Rev Esp Cardiol. 2008; 61: 803–816.
Goldberg RJ, Currie K, White K, Brieger D, Steg PG, Goodman SG, Dabbous O, Fox KA, Gore JM. Six-month outcomes in a multinational registry of patients hospitalized with an acute coronary syndrome (the Global Registry of Acute Coronary Events [GRACE]). Am J Cardiol. 2004; 93: 288–293.
Anderson JL, Adams CD, Antman EM, Bridges CR, Califf RM, Casey DE Jr, Chavey WE II, Fesmire FM, Hochman JS, Levin TN, Lincoff AM, Peterson ED, Theroux P, Wenger NK, Wright RS, Smith SC Jr, Jacobs AK, Halperin JL, Hunt SA, Krumholz HM, Kushner FG, Lytle BW, Nishimura R, Ornato JP, Page RL, Riegel B. ACC/AHA 2007 guidelines for the management of patients with unstable angina/non-ST-elevation myocardial infarction. Circulation. 2007; 116: e148–e304.
Rao SV, Ou FS, Wang TY, Roe MT, Brindis R, Rumsfeld JS, Peterson ED. Trends in the prevalence and outcomes of radial and femoral approaches to percutaneous coronary intervention: a report from the National Cardiovascular Data Registry. J Am Coll Cardiol Cardiovasc Interv. 2008; 1: 379–386.
Stone GW, McLaurin BT, Cox DA, Bertrand ME, Lincoff AM, Moses JW, White HD, Pocock SJ, Ware JH, Feit F, Colombo A, Aylward PE, Cequier AR, Darius H, Desmet W, Ebrahimi R, Hamon M, Rasmussen LH, Rupprecht HJ, Hoekstra J, Mehran R, Ohman EM. Bivalirudin for patients with acute coronary syndromes. N Engl J Med. 2006; 355: 2203–2216.
Pinto DS, Stone GW, Shi C, Dunn ES, Reynolds MR, York M, Walczak J, Berezin RH, Mehran R, McLaurin BT, Cox DA, Ohman EM, Lincoff AM, Cohen DJ, for the ACUITY (Acute Catheterization and Urgent Intervention Triage Strategy) Investigators. Economic evaluation of bivalirudin with or without glycoprotein IIb/IIIa inhibition versus heparin with routine glycoprotein IIb/IIIa inhibition for early invasive management of acute coronary syndromes. J Am Coll Cardiol. 2008; 52: 1758–1768.
Sculpher MJ, Lozano-Ortega G, Sambrook J, Palmer S, Ormanidhi O, Bakhai A, Flather M, Steg PG, Mehta SR, Weintraub W. Fondaparinux versus enoxaparin in non-ST-elevation acute coronary syndromes: short-term cost and long-term cost-effectiveness using data from the Fifth Organization to Assess Strategies in Acute Ischemic Syndromes Investigators (OASIS-5) trial. Am Heart J. 2009; 157: 845–852.
The current guidelines for the management of patients with unstable angina/non–ST-elevation myocardial infarction stress the importance of balancing antithrombotic and interventional therapies with therapeutic risk and urge special attention in groups at high bleeding risk, including women and the elderly. However, determining the net benefit of aggressive treatment in this setting can be complicated because of the lack of quantitative tools able to suggest which treatment strategy might be the most appropriate. The Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the American College of Cardiology/American Heart Association Guidelines (CRUSADE) score, which involves variables obtained upfront, was found to perform well in the test population. Testing risk scores in a completely independent data set provides a rigorous test of their utility and should be undertaken before its widespread use is recommended. Using data from >700 patients with non–ST-elevation acute myocardial infarction admitted to a single Spanish center, we found that this model performs well for most patients except those treated with >2 antithrombotics who do not undergo cardiac catheterization. The CRUSADE score had excellent capacity in discriminating between high- and low-risk patients (C statistic, 0.82). Moreover, we have established that the CRUSADE risk model maintained its performance even when the study population was stratified by subgroups of age (≥75 versus <75 years) (C statistic, 0.80 and 0.81, respectively). The major bleeding predicted by the model closely approximated that observed in this study. Validation of risk models within the population to which they are to be applied is essential. Only then can physicians and healthcare providers be reassured of the performance of the models and subsequently their applicability for risk stratification.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/cgi/content/full/CIRCULATIONAHA.109.925594/DC1.