Cardiac Resynchronization Therapy in Clinical Practice: Some Interesting News but Methodological Advances are Needed

This article refers to ‘CRT Survey II: a European Society of Cardiology survey of cardiac resynchronisation therapy in 11 088 patients—who is doing what to whom and how’ by K. Dickstein et al., published in this issue on pages xxx. The outcomes of patients with heart failure and reduced ejection fraction (HFrEF) have improved substantially over the last two decades due to the introduction in clinical practice of drugs and devices shown to be effective in well-conducted large-scale clinical trials.1, 2 Part of the improvement may be attributable to cardiac resynchronization therapy (CRT), shown to reduce mortality and hospitalization rates in a selected population of patients with HFrEF.3 In this issue of the Journal, Dickstein et al.4 report on a survey conducted in 42 countries, all members of the European Society of Cardiology (ESC), with the aim of re-assessing ‘who is doing what to whom and how’ with respect to CRT. Information was collected on 11 088 patients from 288 CRT implanting centres. The results are, to some extent, expected: electrophysiologists were the most frequent implanters; CRT with defibrillator devices were implanted in 70% of cases, mainly in men with reduced ejection fraction (<35%), left bundle branch block (LBBB) and prolonged QRS (>150 ms in 67% of patients). The procedure was safe, with short-term complications observed in only a very small number of cases. The overall picture of the survey, conducted on a voluntary basis in specialized centres implanting CRT, substantially shows good adherence to the current ESC guideline recommendations. This observation is remarkable, considering that recent reports such as those of the Swedish Heart Failure5 and of the US Get With The Guidelines Heart Failure registries6 had led to the conclusion that CRT was largely underused.7 As can be expected when clinical practice is evaluated, the survey of Dickstein et al. shows some deviations from guidelines: 8% of patients showed a QRS <120 ms and about one out of four patients presented with atrial fibrillation or a non-LBBB QRS morphology. As in other studies on the same topic, the paper does not report the level of ‘optimal medical therapy’ before CRT implantation that is instead strongly recommended by the guidelines before taking into consideration this therapeutic intervention. The major concern regarding this survey is the representativeness of the population enrolled across the national networks of enrolling centres, which appears at least suboptimal. As stated, the study was conducted on a voluntary basis, with very variable rates of recruitment per centre and per country; for instance, very large countries, such as Russia, included a number of patients similar to that of smaller countries, such as Latvia, or Macedonia FYR. This is not a bias peculiar to the survey by Dickstein et al., but can be generalized to most international observational studies conducted by specialists in the field of cardiovascular diseases (and not only), which suffer from generally weak central study coordination and the very limited available resources. Why is network representativeness so important? Because observational studies should not be simple academic exercises but should represent a descriptive source of current medical practice forming the basis on which public health decisions should be made. Although this applies more to regional/national research than to multi-country studies, this principle should be taken as a constant guide for many methodological reasons, including the possibility of providing credible inter-country benchmarking. How can the representativeness of such types of studies be improved? There are several possible solutions. First, the North European model. Northern European countries, in particular Sweden1 and Denmark,2 pioneered the use of registries through the systematic collection of data from national networks of hospitals and health practices, with the constant support of public resources. The main characteristics of these registries are the longitudinal long-term (decades) planning, the simplicity of the data set (with consequent compatibility with routine clinical activity of physicians), the use of shared definitions of variables (enhancing data accuracy), the limitation of missing data (increasing representativeness), a reasonable central data quality control (reducing errors) and, importantly, low cost. The application of this model implies that governments value public health as a real priority and create digital operational structures to collect the data; it also implies that physicians (guided by authoritative medico-scientific societies) do a good job in interpreting and using the collected information. A second possible solution implies a systematic centralized, potentially universal, electronic health recording (EHR) system,8 sharable across countries, allowing the collection of all daily activities in hospitals and/or practices, including common, simple clinical data sets to be linked to the routine collection of administrative data (i.e. hospitalization records, prescriptions, survival status, outpatient procedures). The aim would be to shape in real time both the evolving clinical needs and the initiatives undergoing to satisfy these needs, by generating big data flows to be interpreted by scientists, and used by politicians and public health system managers. In essence, to create a learning health system supported by a developed evidence-generation system, as recently pointed out by Califf et al.9 The essential key concept of the ‘learning health system’ is to integrate research with clinical practice through interoperable EHRs, registries, claims data and administrative data.10 Importantly, however, the EHR system is not a ‘served dish’. Definitions and semantics of patient and healthcare information may vary a lot, as well as the conditions for interaction among different EHR systems, and incomplete data on particular topics may be expected. In short, high quality data, uniformity and interoperability are objectives yet to be reached in most ongoing EHR systems; this is not surprising—and should not be discouraging—in an extremely complex commitment still under implementation.11 The observational research coming from registries should be at the centre of the cycle of health quality, as pointed out in a recent joint ACC/AHA/STS statement.12 An important step forward to implement an efficient digital health system in clinical practice, that assures both good practice and science, is the ‘Roadmap for innovation in the era of digital health, big data, and precision health” statement.13 This statement proposes an advanced model of pragmatic integration of clinical and digital health-based information on the so-called ‘Innovation Collaborative’ ultimately aimed at improving individual medical care. The ESC, a European federation of National Societies of Cardiology from 56 different countries, fully independent in the area of public health, may play a role as cultural link and in promoting digital health, through a number of ongoing deliverables.14 An initiative started in 2015 called ‘Atlas’ is a good example.15 Atlas collects (and periodically updates) descriptive and quantitative data on the economy, demography, health status, health system and policies for each ESC member country. In addition, information regarding specific aspects of cardiac care is collected. Preliminary results show that there are significant within- and across-country variations and inequalities in financial support, organization, accessibility, delivery, quality, and effectiveness of cardiac care among the ESC countries.15 This database should probably be examined when a new registry or survey is planned in order to select and invite an appropriate number and type of hospitals to contribute to data collection. In this way, it is more likely that a balanced group of centres, in terms of dimension, location, technological level of sites can be ensured, achieving the best possible representativeness of participating countries. The Atlas database also offers the opportunity to enrich the clinical registry data coming from the ESC EURObservational Research Programme with the socio-economic background of each participating country. In conclusion, the paper by Dickstein et al.4 provides valuable information on the use of a relatively new technology that can improve the outcomes of patients with heart failure. As in other observational studies, however, representativeness of participating centres and countries is questionable. In this sense, we propose methodological advances to overcome this relevant issue, in order to provide more reliable and comparable data to the stakeholders that deal with heart failure or other cardiovascular conditions. Conflict of interest: A.P.M. and L.T. have no conflict of interest to disclose with respect to the present manuscript. Outside the present work: A.P.M. received honoraria for participation in committees of studies sponsored by Bayer, Novartis and Fresenius; L.T. is trial committee member and member of the speaker bureau for Servier, and trial committee member for Boston Scientific, Medtronic, Cardiorentis, CVIE Therapeutics, ZS Pharma, St Jude Medical.