Registries and surveys in acute coronary syndrome

The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal or of the European Society of Cardiology. The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal or of the European Society of Cardiology. Randomized controlled trials provide the most secure evidence for the impact of a specific treatment in a defined patient population. However, to study such treatments, the defined patient population often excludes individuals at higher risk, those with co-morbidity, and those with greater risk of complications of the drug or interventional therapy. Hence, unless the trial is very large and heterogeneous, it tends to reflect an ‘ideal’ study set rather than the diversity of clinical practice. Registries conducted in the same centres, and at the same time as clinical trials, have revealed substantially higher co-morbidity and risks of death among those excluded from trials.1 Although it is critically important to define what can be achieved with specific treatments, clinical trials cannot define how such treatments are applied in the diversity of practice. In contrast, registries and surveys, such as the second Euro Heart Survey (EHS II) on ACS,2 have the potential to define the ‘gaps’ between evidence and practice. Epidemiological studies provide invaluable sources of information about the behaviour of specific populations and their outcomes.3 However, their design is such that they do not provide insights into the spectrum of treatment options and their impact. Thus, randomized trials, epidemiological studies, and appropriately designed registries and surveys provide complimentary sets of information to allow the clinician to judge the characteristics of the population in question, what may be achievable, and what is actually achieved in clinical practice. The advances in the management of acute coronary syndrome have, by and large, produced incrementally smaller treatment benefits. Thus, differences between certain treatment options may be very modest in comparison with the impact of receiving a therapy or no therapy. For example, the differences between modern thrombolytic drugs are very modest when compared with the impact of reperfusion or no reperfusion. Yet, about a third of patients receive no reperfusion.2,4 The outcome of untreated patients may only be revealed in registries and surveys. On the basis of an editorial by Alpert5 in 2000, key functions of appropriately designed registries and surveys can be defined. Registries and surveys can provide data on an entire condition, for example, the spectrum of acute coronary syndrome, rather than a subset of ‘ideal’ patients. They provide a cross-sectional view of the clinical and demographic aspects of the disease condition. They can define the extent to which a particular disease condition is treated according to national and international guidelines. The potential clinical gain achieved by treating a higher proportion of eligible patients with evidence-based therapy may be substantially greater than the difference in outcome between agents with a similar mode of action.6 Conversely, the risks of over-treatment have not been fully recognized.7 Registries and surveys can collect data rapidly and efficiently, allowing an analysis of a disease condition over a particular chronological interval. Their costs are modest compared with those of phase III clinical trials (the difference can be two orders of magnitude). Continuous registries or repeat sampling can provide a dynamic estimate of the changing patterns of disease presentation, changing patterns of therapy, and changing outcome. Registry and survey data allow clinicians to compare their own practice with that of larger national or international reference populations. This provides an important stimulus for improvements in quality and consistency of practice. They provide a reference standard for event rates in clinical practice and can raise hypotheses for testing in randomized trials. In contrast, assessing the impact of a specific treatment may not be reliable because of the influence of unmeasured variables on choice of therapy. Registries can provide morbidity and mortality analyses, risk prediction, and resource utilization calculations for particular disease entities. In contrast, similar calculations from trial data need to recognize the impact of exclusion criteria (for example, the elderly, those with renal dysfunction or other co-morbidity, and the impact of gender and ethnicity). Risk models and risk calculations have the potential to improve clinical decision making, especially in complex heterogeneous conditions such as ACS. Robust national and international data allow healthcare administrators to compare disease-specific management and outcomes by region and by comparison with international data. Registries and surveys can provide robust educational tools for trainees in cardiology, for students, and for continuing medical education among established cardiologists. In EHS II, the authors analysed information from 6385 patients from 190 medical centres in 32 countries.2 The data were compared with the first EHS, but only 34 (18%) of the centres participated in both surveys. The participants in both tended to be affiliated to academic institutions with more invasive facilities and hence they may not reflect the diversity practice across all centres. Nevertheless, the data are encouraging in that a higher proportion of those with ST elevation received reperfusion therapy (64 vs. 56%) and overall there was an increase in the level of adherence to guidelines for the treatment of ACS. The authors interpret the lower mortality in EHS II as associated with improvements in management and treatment. While this may be true, less than one-fifth of centres participated in both surveys, and this does not allow a truly robust analysis of outcomes by time. Further, the voluntary nature of the survey, the lack of a systematic quality control and audit, and the uncertainty about whether included patients were truly representative of the whole spectrum mean that the interpretations must be viewed with caution. Just as for clinical trials, quality standards are important to ensure that observational data in registries and surveys are representative and that the interpretations are reliable. On the basis of the original proposals by Alpert5 and developments in registries and surveys since then, the following may be proposed as appropriate quality standards. It is critically important that the patients sampled in observational studies are representative of the disease condition. Authors need to present the sampling methods to ensure representativeness. For example, if sampling is mainly in cardiac care units, this may reflect only part of the spectrum of ACS.1,2 Power for primary and subsidiary analyses: is the study appropriately powered to provide reliable estimates of geographic differences? Modest numbers of patients in some countries or regions may not provide reliable regional data. Further, the participating centres may not reflect the spectrum of hospitals admitting patients with the condition. Thus, comprehensive national data are preferable.8,9 How are participating centres identified and are they representative of the spectrum of clinical facilities assessing patients with the condition? Some national registries8,9 involve virtually all cardiac units in a country, whereas others are huge in scale and attempt to ensure representativeness by virtue of their size.10 An alternative approach sets out to identify all patients within defined geographic regions.3,4,6 The need for consistent and centrally adjudicated case definitions and outcome definitions and the need for appropriate training of participating sites. Studies that rely on local interpretations and untrained staff are more difficult to interpret, as unmeasured variables may influence reported data. For example, there may be substantial inconsistency in the criteria for diagnosis of non-ST elevation MI. The criteria may be dependent on varying clinical opinions, different troponin assays with varying precision, and varying decision limits. An independent scientific steering committee and independent data analysis are important to avoid real or perceived bias. Studies need to present the measures taken to avoid reporting bias. Studies need to meet the standards required for good clinical practice and appropriate ethical review. Funding support and the role of the sponsor or sponsors need to be declared. No study, to date, fully meets all of these quality standards. However, by evaluating the strengths and weaknesses of each individual study against these standards, readers should be able to make reliable interpretations. The EHS II on ACS has made a significant contribution. It provides the most comprehensive information, to date, across Europe and the Mediterranean Basin, and the comparisons with the first EHS do provide evidence of encouraging trends. Nevertheless, there are concerns about representativeness of the included patients and the differences in patient characteristics between the first and second EHS. These suggest that there was greater reliance on recruitment in cardiac care units in the latter, and practice may differ in such settings. Nevertheless, the EHS II on acute coronary syndrome has provided valuable insights into the management of ACS and has identified important gaps between evidence and practice. It may provide a valuable stimulus to improve the quality and consistency of patient treatment. Despite the advances in the management of ACS, there remain substantial early and longer term risks, especially among the elderly, those with major co-morbidity, and those with cardiac complications including heart failure and renal dysfunction. Much remains to be done. Conflict of interest: The author is a Primary Investigator or co-investigator in trials and in registries and an author of manuscripts derived from these studies. No commercial conflict of interest.