Inter-regional differences and outcome in unstable angina

European Heart Journal (2000) 21, 1433 1439 doi:10.1053/euhj.1999.1983, available online at http://www.idealibrary.com on Inter-regional differences and outcome in unstable angina Analysis of the International ESSENCE trial K. A. A. Fox 1, S. Goodman 2, F. Bigonzi 3, V. Le Louer 3, M. Cohen 4 and The ESSENCE Trial Investigators 1 Department of Cardiology, The Royal Infirmary, Edinburgh, U.K.; 2 The Terrence Donnelly Heart Centre, Division of Cardiology, St Michael s Hospital, University of Toronto, Toronto, Canada; 3 Rhône-Poulenc Rorer, Cardiovascular Research Department, Antony Cedex, France; 4 Division of Cardiology, MCP-Hahnemann School of Medicine, Philadelphia, U.S.A. Aims Worldwide there is a large variation in outcome (death, myocardial infarction and recurrent myocardial infarction) in patients with unstable angina or non-q wave myocardial infarction. These variations may be explained by differences in characteristics of the presenting patients. Here we describe differences in patient presentation, treatment protocols and outcome and we investigate their relationship using data from the ESSENCE (Efficacy and Safety of Subcutaneous Enoxaparin in Non-Q-wave Coronary Events) trial. Methods A total of 2981 patients from six countries which enrolled more than 100 patients were included in the present analysis: United States, Canada, Argentina, France, the Netherlands and United Kingdom. Logistic regression analysis was performed to determine the effect of baseline characteristics on regional outcome. Results At day 30, the lowest triple end-point rate, irrespective of study drug treatment, was noted in the Netherlands (18 9%) and the highest in Argentina (30 5%). A model including the variables age 65 years, prior angina, diabetes, prior aspirin use, ECG changes at baseline and diagnosis of non-q wave myocardial infarction and dummy variables for Argentina and France resulted in concordance of about 60%. Conclusions Inter-regional differences in outcome in unstable angina and non-q wave myocardial infarction patients can reasonably well be explained by differences in patient characteristics. However, other so far unidentified variables present in Argentina and France also contributed to differences in outcome and their effect warrants further investigation. (Eur Heart J 2000; 21: 1433 1439) 2000 The European Society of Cardiology Key Words: Unstable angina, non-q wave myocardial infarction, ESSENCE, low molecular weight heparin, enoxaparin, regional differences. See page 1397 for the Editorial comment on this article Introduction Patients with unstable angina or suspected non-q wave myocardial infarction are at high risk of death, myocardial infarction or ischaemia, despite currently available drug therapies [1 5]. Worldwide, there is a large variation in the outcome of these patients [6]. This variation may in part be explained by differences in characteristics of the Revision submitted 25 September 1999, and accepted 6 October 1999. Correspondence: Professor Keith A. A. Fox, Department of Cardiology, The Royal Infirmary, Edinburgh EH3 9YW, U.K. 0195-668X/00/211433+07 $35.00/0 presenting patients. However, a patient s outcome is also likely to be influenced by their clinical treatment [1 5]. Clinical practice is influenced by a variety of factors including the underlying incidence of coronary heart disease in the local population, the availability of resources and the current perceptions within the medical community of available treatments [6,7]. Although a number of reports have considered this problem [6 10], the impact of regional differences on the prognosis and outcome of cardiology patients remains uncertain. To consider this problem fully, it is necessary to undertake a country by country analysis of relevant data collected through an international cardiology trial. One such suitable source is the ESSENCE trial [1]. 2000 The European Society of Cardiology

1434 K. A. A. Fox et al. The ESSENCE trial (Efficacy and Safety of Subcutaneous Enoxaparin in Non-Q-wave Coronary Events) demonstrated that a low molecular weight heparin (enoxaparin) is more effective than unfractionated heparin in reducing the incidence of ischaemic events in unstable angina and non-q wave myocardial infarction patients. Overall, at day 30, 19 8% patients treated with enoxaparin vs 23 3% of patients treated with unfractionated heparin were at risk of death, myocardial infarction or recurrent angina [1]. This overall treatment effect was maintained across all participating countries, but the overall incidence rates differed substantially in the participating countries. Although an identical trial protocol was followed in all 176 participating centres across 10 countries in North America, South America and Europe, the use of adjunctive medical treatments and interventional cardiac procedures was at the discretion of the local responsible clinician [11]. Therefore, the ESSENCE data provides an ideal opportunity to investigate the effect of interregional differences such as these on the outcome of the coronary heart disease patient. The main aims of the present investigation are to describe differences in baseline characteristics, clinical management and outcome (end-point of death, myocardial infarction or recurrent angina), across the different countries participating in the ESSENCE study, and to determine whether differences in outcome can be explained by differences in patient baseline characteristics and clinical management practices. Methods Patients and populations The design, treatment protocol and overall findings of the ESSENCE study have been published previously [1,11]. In brief, ESSENCE was an international prospective, randomized, double-blind, double-dummy, parallel-group multicentre study, comparing the use of low-molecular-weight heparin (enoxaparin) with unfractionated heparin in the treatment of unstable coronary artery disease. Patients were eligible for inclusion in the study if they had recent onset rest angina and underlying ischaemic heart disease. Worldwide, 3171 patients were enrolled in 176 hospitals in 10 countries. Six of the 10 countries enrolled more than 100 patients: United States (n=936), Canada (n=1259), Argentina (n=256), France (n=133), the Netherlands (n=206) and United Kingdom (n=191). In total, these six countries enrolled 2981 patients, accounting for 94% of the entire ESSENCE trial population. Data from these six countries have been used in the regional analysis reported here. For this analysis, outcome was assessed using a composite triple end-point of death, myocardial infarction (or reinfarction) or recurrent angina at day 30. Recurrent angina was defined as either angina that was associated with electrocardiographic changes, or angina that prompted a revascularization procedure or angina that resulted in rehospitalization. Statistical analysis Baseline characteristics of study patients were summarized for each country as means for continuous variables and frequencies for categorical variables. Similarly, clinical practices and means of hospital days were evaluated per country. 95% CI were calculated. To investigate the potential effect of baseline characteristics on regional differences in outcome at day 30, we performed an exploratory logistic regression analysis. The following baseline characteristics and clinical predictor variables were included as dichotomous variables: age 65, male gender, prior myocardial infarction, prior angina, diabetes, hypertension, prior aspirin use, ECG changes at baseline, baseline diagnosis of unstable angina, diagnosis of non-q wave myocardial infarction. In addition, dummy variables for each country were added to the logistic regression model. Treatment was not included in any of the models. Only patients with complete data for each of the baseline characteristics listed above were included. This resulted in the exclusion of 30 (1%) patients of the randomized patient population. The multivariable model was first fitted to the heparin patient population (n=1462) and then applied to the enoxaparin patient population (n=1489), and finally to the entire population to provide more exact estimates of measures (n=2951). Stepwise logistic modelling was carried out to identify the subset of variables (main effects) that together were most significant in predicting the end-point at 30 days. Variables were excluded from the model if their level of significance exceeded P=0 10. A measure of concordance (based on rank calculations between the observed responses and predicted probabilities, C value) were calculated to describe the quality of the adjustment. The predicted rates of triple end-points were then compared to the observed rate of triple end-points per country. All statistical analyses were performed using SAS software, version 6.11. Results Regional variation in patient characteristics Baseline characteristics varied between the six geographical populations studied in the regional analysis (Table 1). Although more males than females were recruited into the ESSENCE trial in each of the six regions, the highest proportion of male recruits were noted in the Netherlands and Argentina (74 8% and 74 6% respectively) with the lowest proportion in Canada (63 5% male). Patients were oldest and lightest in France (mean age 69; mean weight 73 3 kg) and

Differences and outcome in unstable angina 1435 Table 1 Patient baseline characteristics in the ESSENCE trial Baseline parameters ARG FRA UK CAN USA NL Number of patients 256 133 191 1259 936 206 Demographic Sex ratio (% male) 74 6 69 9 71 2 63 5 65 3 74 8 Age (mean in years SD) 62 0 9 7 68 6 11 2 63 6 10 2 63 9 12 0 63 3 11 6 62 7 10 1 Weight (mean in kg SD) 78 6 13 9 73 3 13 9 77 5 15 5 77 4 16 1 83 2 17 5 78 5 11 6 History of underlying coronary artery disease Prior MI 42 6 38 3 54 5 48 1 45 8 43 2 Prior CABG (%) 20 7 8 3 15 7 15 6 27 2 24 3 Prior PTCA (%) 16 4 19 5 8 9 17 6 29 8 25 2 Prior angina (%) 54 3 63 9 72 8 73 8 67 0 59 2 Prior aspirin user (%) 73 0 45 1 56 0 62 7 63 1 61 7 Prior beta-blocker user (%) 65 2 39 1 33 0 40 9 35 3 45 1 Cardiac risk factors Family history (%) 15 6 34 6 50 3 46 9 42 5 43 7 Current smoker (%) 17 6 15 8 21 5 23 4 26 1 32 5 Hypertension (%) 64 1 53 4 30 4 47 2 68 3 39 3 Hypercholesterolaemia (%) 63 3 39 8 19 9 42 9 48 6 33 0 Diabetes (%) 17 2 15 0 14 7 21 7 30 2 9 2 heaviest in the United States (mean weight 83 2 kg). The youngest patients were from Argentina (mean age 62). There were considerable differences in the frequency of prior manifestations of coronary artery disease. For instance, the highest rates of prior myocardial infarction were noted in the United Kingdom (54 5%) and the lowest in France (38 3%), whilst the highest frequency of prior angina was seen in Canada (73 8%) closely followed by United Kingdom (72 8%), with the lowest frequency found in Argentina (54 3%). There was a noticeably low rate of prior coronary artery bypass graft (CABG) in France (8 3%) and prior percutaneous coronary artery angioplasty (PTCA) in the United Kingdom (8 9%). Documented cardiac risk factors also varied markedly between countries. For example, the highest frequency of family history of ischaemic heart disease was in the United Kingdom (50 3%) and the lowest in Argentina (15 6%). Conversely, the lowest frequencies of hypertension and hypercholesterolaemia were seen in United Kingdom (30 4% and 19 9% respectively) with the highest in Argentina (64 1% and 63 3% respectively). Smoking prevalence varied from 32 5% in the Netherlands to 15 8% in France. Regional variation in clinical presentation There was some variation in clinical presentation between the six geographical populations studied in the regional analysis (Fig. 1). The proportion of patients who presented with unstable angina compared to those with non-q wave myocardial infarction was broadly similar in all the participating countries, with the exception of Argentina which had the lowest prevalence of unstable angina coupled with the highest prevalence of non-q wave myocardial infarction. There were significantly fewer patients with baseline ischaemic ECG changes in North America compared to Argentina and the European countries. Regional variation in clinical management Although the ESSENCE protocol provided an algorithm for non-trial antianginal therapy, the use of cardiac medications was not dictated by the study protocol. Consequently, adjunctive medical procedures and interventional cardiac procedures varied throughout the participating countries (Table 2 and Fig. 2). During the first 48 h, Argentinean patients received the highest percentage of nitrates and beta-blockers and the lowest percentage of calcium channel-blockers. Conversely, patients in the United Kingdom received the lowest percentage of nitrates and beta-blockers and the highest percentage of calcium channel-blockers. The overall, combined duration of use of ESSENCE trial medication (enoxaparin+aspirin, unfractionated heparin+aspirin) differed between countries, ranging from a mean of 2 4 days in the United States to 4 6 days in France, with a mean of around 3 days in the other countries. During the 30 day study period the duration of hospitalization differed substantially across countries (Table 2). The shortest hospital and ICU stays were observed in the United States and United Kingdom and the longest in France. Over the 30 day study period, 50% of all ESSENCE trial patients underwent angiography and 30% underwent revascularization via CABG (13%) or percutaneous means (PTCA, 17%) [5]. The use of these cardiac procedures differed markedly among the six countries (Fig. 2). Angiography was most commonly used in France and the U.S.A., with the lowest application

1436 K. A. A. Fox et al. 90 80 70 60 50 % 40 30 20 10 0 ARG (n = 256) FRA (n = 133) UK (n = 191) CAN (n = 1259) USA (n = 936) NL (n = 206) Figure 1 Clinical presentation of patients by country. =unstable angina; =non-q wave myocardial infarction; =ECG changes. Table 2 Clinical practice by country in the ESSENCE trial ARG (n=256) FRA (n=133) UK (n=191) CAN (n=1259) USA (n=936) NL (n=206) Cardiac medication at 48 h (%) Nitrates 96 9 93 2 80 1 86 9 89 5 90 8 Beta-blocker 87 5 66 9 53 4 69 7 67 6 71 4 Ca-channel blocker 24 2 54 1 57 6 53 1 50 9 49 0 Hospital days (mean) 10 0 12 1 7 5 9 8 6 7 10 4 [95 CI] [9 1 10 8] [10 9 13 3] [6 7 8 4] [9 4 10 2] [6 4 7 0] [9 5 11 4] ICU days (mean) 5 5 6 6 2 2 3 0 2 8 3 8 [95% CI] [5 0 6 0] [5 8 7 5] [1 8 2 6] [2 8 3 2] [2 5 3 0] [3 4 4 3] in United Kingdom. The highest rates of revascularization were seen in Argentina and the lowest rates in United Kingdom. In all six countries, CABG was less frequent than PTCA. The mean time to revascularization varied from 5 7 days in Argentina to 14 1 days in United Kingdom. Regional outcome at 30 days The incidence of triple end-point consisting of death, myocardial infarction or recurrent angina varied across the six different regions (Fig. 3). At day 30, the lowest triple end-point rate, irrespective of study drug treatment, was noted in the Netherlands (18 9%) followed by the U.S.A. (20 1%), Canada (20 4%), U.K. (21 5%), France (28 6%) and Argentina (30 5%). A country vs treatment interaction variable was not significant (P=0 56). Death or myocardial infarction occurred in 38/256 (14 8%) in Argentina, 14/133 (10 5%) in France, 62/936 (6 6%) in the U.S.A., 73/1259 (5 8%) in Canada, 11/191 (5 8%) in the U.K. and 10/206 (4 9%) in the Netherlands. Multivariable analysis findings The following baseline factors proved significant in predicting outcome in the heparin using study population (n=1462): age 65 years, prior angina, diabetes, prior aspirin use, ECG changes at baseline and diagnosis of non-q wave myocardial infarction. In addition, two dummy variables for Argentina and France significantly contributed to the model. The concordance of the model equalled 61 9%. Application of this model to the population of enoxaparin using patients (n=1489) and all patients (n=2951), resulted in a concordance of 58 5% and 59 8%, respectively. The expected average event rates per country calculated with this model compared well with the actual average event rates observed in each country using data on all patients (Table 3). Application of the logistic regression model to the double end-point of death or myocardial infarction revealed very similar levels of predictive power (63%). As residence in Argentina and France emerged in the final model as important predictors of clinical outcome, we applied the initial regression model independently for both countries (concordance 62 1% and 69 1%,

Differences and outcome in unstable angina 1437 80 70 60 50 % 40 30 20 10 0 ARG (n = 256) FRA (n = 133) UK (n = 191) CAN (n = 1259) USA (n = 936) NL (n = 206) Figure 2 Cardiac procedures post-randomization. =angiography; =CABG; =PTCA. 40 35 30 5 28 6 30 21 5 25 20 4 20 1 18 9 % 20 15 10 5 0 ARG (n = 256) FRA (n = 133) UK (n = 191) CAN (n = 1259) USA (n = 936) Figure 3 Incidence of death, myocardial infarction, and recurrent angina at 30 days. NL (n = 206) Table 3 Predicted versus observed outcome probability ARG (n=256) FRA (n=133) UK (n=191) CAN (n=1259) USA (n=936) NL (n=206) Variables in model age 65 years; history of angina, diabetes, and aspirin use; ECG changes at baseline; diagnosis of non-q wave MI; Argentina and France Average predicted probability of end-point 30% 28% 22% 20% 20% 20% Actual end-point rate at 30 days 30 5% 28 6% 21 5% 20 4% 20 1% 18 9% respectively). In both cases, the variable ECG changes at baseline was the single most important independent predictor in the model of end-point rates. The contribution of each country was re-tested by comparing the final model to a model without the country factors but including ECG changes at baseline. Both France and Argentina proved highly significant (P=0 04 and P<0 001, respectively), suggesting that, in these countries, factors other than ECG changes alone contribute significantly to differences in patient outcome. One key observed difference was the increased frequency of revascularization in Argentina and France and it is

1438 K. A. A. Fox et al. possible that this may be associated with a higher frequency of peri-procedure events. A similar finding has been observed in the OASIS Registry [6]. Discussion As expected, there was a large variation in the baseline characteristics, clinical management and outcome of coronary heart disease patients enrolled in the international ESSENCE trial. Particularly apparent were the high prevalence of primary risk factors for coronary heart disease such as hypertension, hypercholesterolaemia and diabetes in US patients as compared to patients from other countries, and the high use of prior revascularization treatments in the US. Clinical management practices also varied between countries. For instance, patients in Argentina were prescribed 50% fewer calcium channel blockers than patients in other participating countries. Also, patients in the U.K. underwent 50% fewer angiography and revascularization procedures than patients in other countries. The length of hospital stay differed considerably between countries with the largest difference observed between ICU days in France (6 6) and ICU days in the US (2 8). Differences in outcomes between countries were less pronounced with the largest difference observed between the Netherlands at 18 9% and Argentina at 30 5% events at 30 days. Our regression analysis revealed that these differences could be explained reasonably well by between-country differences in patient baseline characteristics. The baseline conditions independently predictive of outcome at 30 days were: age over 65 years, history of angina, diabetes and of prior aspirin use, ECG changes at baseline and diagnosis of non-q wave myocardial infarction. Interestingly the addition of two dummy variables for Argentina and France was found to be significant. The results for France and Argentina could be partially explained by ECG changes at baseline, but this variable did not account for the total contribution of these countries to the variation in outcome. Other factors including the higher frequency of revascularization may account for these findings [6]. In addition to baseline patient characteristics, other so far non-identified characteristics play an important role in predicting cross-cultural differences in the ESSENCE trial. Since the use of resources was not dictated by the ESSENCE protocol, regional differences in interventions and outcome may partially reflect the available facilities at different participating centres. Other reasons may include differences in local standard practices, the availability of cardiac investigational facilities, cost constraints within different health care systems, and patient preferences [6 10]. Indeed, the regional differences in clinical management is apparent despite the fact that the data were collected in a clinical trial setting with well defined inclusion/exclusion criteria and treatment protocol. The results reflect treatment practice in the participating centres and not necessarily in the respective countries, in entirety. We were not able to investigate the effect of regional management practices such as revascularization or treatment duration because our end-point could be clinically linked to frequency of revascularization post-randomization. Hence, a clear cause and effect relation cannot be established by such an analysis. However, the regression analysis did demonstrate that a treatment country interaction variable was not significant. This suggests that the effect of treatment did not differ between countries. Previous studies in myocardial infarction have reported differences in countries assessment, treatment and outcome [8]. Comparisons of myocardial infarction patients in the U.S.A. and other countries have shown significantly lower mortality for patients treated in the U.S.A.; a difference which is associated with the greater use of diagnostic and therapeutic intervention procedures in the U.S.A. [8,10]. However, in patients with unstable angina or myocardial infarction without initial ST elevation, higher rates of invasive and revascularization procedures were found to be associated with lower rates of refractory or recurrent angina, but not with a reduction in cardiovascular death or myocardial infarction and an increased stroke rate. In fact a higher rate of death and myocardial infarction was seen [11]. It is important to recognise that this report is based solely on international comparisons within the boundaries of a randomized controlled trial and is not a true epidemiological study. The findings cannot be interpreted as a representative reflection of practice in the six countries because the participating centres in each individual country were not necessarily representative of hospital practice in that country. Thus, although the present study examines the relationship between regional characteristics, management and outcome for patients with unstable coronary artery disease, it is not necessarily representative of the entire practice in the respective countries. Conclusion This analysis of the ESSENCE trial data by six different countries has shown that there are differences in the 30 day outcome (death, myocardial infarction or recurrent angina) of patients with coronary artery disease, irrespective of trial drug therapy. The differences in regional outcome can reasonably well be predicted by differences in regional baseline characteristics, specifically; age over 65, prior angina, diabetes, prior aspirin use, ECG changes and a diagnosis of non-q wave myocardial infarction. Clinical management practices have been shown to vary by country and the specific relationship between outcome and increased use of revascularization procedures warrants further study. Other so far unidentified variables, particularly present in Argentina and France, also contributed to differences in outcome and their effect needs to be investigated. The authors wish to thank all participants of the ESSENCE trial. The ESSENCE trial was supported by a grant from Rhône- Poulenc Rorer, Collegeville, Pa., U.S.A. and Paris, France.

Differences and outcome in unstable angina 1439 References [1] Cohen M, Demers C, Gurfinkel EP et al. A comparison of low-molecular-weight heparin with unfractionated heparin for unstable coronary artery disease. N Engl J Med 1997; 337: 447 52. [2] The Global Use of Strategies To Open occluded arteries (GUSTO) IIb investigators. A comparison of recombinant hirudin with heparin for the treatment of acute coronary syndromes. N Engl J Med 1996; 29: 775 82. [3] The Platelet Receptor Inhibition in Ischemic Syndrome Management (PRISM) Study Investigators. A comparison of aspirin plus tirofiban with aspirin plus heparin for unstable angina. N Engl J Med 1998; 338: 1498 505. [4] The Platelet Receptor Inhibition for Ischaemic Syndrome Management in Patients Limited by Unstable Signs and Symptoms (PRISM-PLUS) Trial Investigators. Inhibition of the platelet glycoprotein IIb/IIIa receptor with tirofiban in unstable angina and non-q-wave myocardial infarction. N Engl J Med 1998; 338: 1488 97. [5] The Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin Therapy (PURSUIT) Trial Investigators. Inhibition of platelet glycoprotein IIb/IIIa with eptifibatide in patients with acute coronary syndromes without persistent ST-segment elevation. N Engl J Med 1998; 339: 436 43. [6] Yusuf S, Flather M, Pogue J et al. Variations between countries in invasive cardiac procedures and outcomes in patients with suspected unstable angina or myocardial infarction without initial ST elevation. Lancet 1998; 352: 507 14. [7] Windecker S, Meyer BJ, Bonzel T et al. Interventional cardiology in Europe 1994. Eur Heart J 1998; 19: 40 54. [8] Holmes DR, Califf RM, Van de Werf F et al. Difference in countries use of resources and clinical outcome for patients with cardiogenic shock after myocardial infarction: results from the GUSTO trial. Lancet 1997; 349: 75 8. [9] Van de Werf F, Topol EJ, Lee KL et al. Variations in patient management and outcomes for acute myocardial infarction in the United States and other countries. JAMA 1995; 273 920: 1586 91. [10] Adams PC, Skinner JS, Cohen M et al. Acute Coronary Syndromes in the United States and United Kingdom: A comparison of approaches. Clin Cardiol 1998; 21: 348 52. [11] Cohen M, Blaber R, Demers C et al. The ESSENCE trial: efficacy and safety of subcutaneous enoxaparin in unstable angina and non-q-wave MI: a double-blind, randomised, parallel-group multicenter study comparing enoxaparin and intravenous unfractionated heparin. J Thromb Thrombolysis 1997; 4: 271 4.