The physician-patient relationship remains the core of medical practice. Adherence to clinical practice guidelines has been shown to improve prognosis1–11; nevertheless, only a fraction of the recommendations are supported by undisputed evidence.12–16 Moreover, the complexity of the individual patient and the organization of medical practice have led to substantial individual, institutional, and intercountry variability.9,17–41 Significant efforts have been made to evaluate quality of care, particularly in cardiology, including the definition and identification of metrics in selected populations,42–62 public reporting,62–75 and the development of systems that enhance adherence to recommendations (eg, accreditations, payment of performance reports,75–83 and quality measures aimed at improving outcomes, including benchmarking.84–98

Overall, the process of quality measurement, benchmarking, quality report enhancement, and auditing is more advanced in the US than in Europe. However, in some European countries, this process is highly developed and is often centralized.92,99–105 One outstanding example is the Society of Cardiothoracic Surgery of Great Britain and Ireland.99 In Spain, the National Health Ministry and some of the autonomous communities have developed diverse reports on numerous standardized quality metrics for cardiology.57–62

Quality is often based on perception. Official and private organizations have voluntarily developed quality standards and benchmarking programs, using opinions or registries that seldom provide reliable information to measure quality. Moreover, attempts to assess the quality and safety of clinical practice have proliferated in recent years, leading to different rating systems that may yield completely different results and ratings for the same hospital during the same time period, thus adding confusion rather than helping to prove their usefulness and leading to doubts about whether quality can actually be measured by existing measures.82–85,89,106–109 However, quality could be either measured throughout the process of organization and delivery of care, or more importantly by the final results of clinical practice: clinical outcomes. Most importantly, benchmarking itself may be associated with a progressive improvement in performance and outcomes,60,89,92,99,110–112 highlighting the importance of standardization of quality measures and the responsibility of scientific societies.

The objective of this document is to identify and standardize quality metrics in hospital cardiology practice. Two groups where clearly differentiated (Figure 1): a) selection of the best and most simplified metrics of the final quality of cardiology practice or outcomes measures (eg, a primary endpoint in a clinical trial), and b) identification of the performance metrics of clinical practice (performance measures) that are known to positively influence desirable outcomes (eg, surrogates in clinical trials).

Figure 1.

Quality marker clusters related to better results in clinical practice (performance measures) and quality of results in clinical practice (outcome measure clusters). The latter are the best way to compare the quality of care in different hospitals/healthcare organizations and to provide information to the public. CCT, cardiac computed tomography; CMR, cardiac magnetic resonance; ECG, electrocardiogram.

(0.54MB).

Beyond that, scientific societies and, in particular, health care authorities should be responsible for the implementation of programs to measure quality, ensure the quality of the data, benchmarking, and certification/accreditation of cardiology services.

This document focuses on quality measures of in-patient cardiovascular care. The quality of outpatient cardiovascular care will be considered in other documents.

This document will be limited to the identification and recommendation of the use of quality metrics. Beyond that, scientific societies and health care authorities should not only be responsible for the implementation of the program that best measures quality but should also ensure the reliability of data through audits and provide reports to the public. Three steps are recommended:

• 1.

  Organization of databases including the necessary information for quality measuremets. Universal participation of hospitals is strongly recommended and the quality of the data must be ascertained through appropriate monitoring and audits for progressive improvement of data quality. If the data used to measure quality are not fully reliable, performance measures will be equivocal and will cause more harm and confusion than benefit. At this time in Spain, a mandatory national health care system database includes core information from all hospital discharge reports101 obtained from the ICD-9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification) codes,113 but the quality of the data may be questioned as there is no quality control. Therefore, the SEC (Sociedad Española de Cardiología) is measuring outcomes through the minimum basic data set of the Ministry of Health registry within the RECALCAR (REsultados de CALidad en CARdiología)40 program, in an attempt to test quality and identify opportunities for improvement. Other institutions, including Health Departments of the autonomous communities, are using the same database for the same purpose, but are employing different parameters. Other countries have similar systems or obtain the information from mandatory or voluntary dedicated databases.83–85,99 This document intends to provide uniformity by standardizing quality markers.

• 2.

  Benchmarking of hospital outcomes and assurance of public, controlled access to data and reports (the latter are a responsibility and decision of the health care authorities).

• 3.

  Initiation of certification/accreditation of hospitals according to their results (responsibility of health care authorities).

The SEC, in cooperation with the SECTCV (Sociedad Española de Cirugía Torácica-Cardiovascular), founded a task force dedicated to identifying and defining quality markers in cardiology. Experts were identified and were invited to cover 8 areas of expertise: clinical cardiology, cardiac imaging, acute cardiac care, interventional cardiology, electrophysiology and arrhythmias, heart failure (HF), cardiac rehabilitation, and cardiac surgery. All European Society of Cardiology (ESC)103 and ACC/AHA (American College of Cardiology/American Heart Association)104 guidelines were reviewed, and recommendations related to quality standards were included in the document. Beyond the guidelines, an informal review was performed of the literature for quality metrics, performance metrics, and quality programs.

The following issues related to quality metrics and evaluation were identified and defined:

• •

  Class of recommendations and levels of evidence.

• •

  Types of hospitals.

• •

  Clusters to assess quality in clinical practice.

• •

  Main markers to assess quality of results in clinical practice (outcomes measures).

• •

  Performance measures associated with better results in clinical practice (performance measures).

Cost and cost-effectiveness have become important components of quality performance54 but were not considered in this document.

The task force was constituted by the SEC and the SECTCV. The document was sent to different cardiology societies, related scientific societies and health care authorities for comment and feedback. The final document was sent to external reviewers before simultaneous publication in Revista Española de Cardiología and Revista Española de Cirugía Cardiovascular. A shorter version will be published in the European Heart Journal.

The costs of convening the task force, arranging meetings, and secretarial assistance were covered by the SEC. All members of the task force volunteered their time and services and received no fees or payment in exchange for their service. No funding from industry was received by the members of the task force or the scientific societies involved in the preparation of this document. Members of the task force reported any possible conflicts of interest.

To grade potential quality markers, the following aspects were considered: a) clinical and practical relevance; b) source and difficulty of obtaining the information; c) difficulty of auditing and ascertaining the information, and d) evidence in the literature. Three levels were established both for class of recommendation and level of evidence, as detailed in Table 1. The classification is based on recommendations in guidelines, the level of evidence attributed in published guidelines, recommendations from regulatory agencies or opinion surveys, and other sources. Mortality and stroke were considered as self-evident.101–104 To avoid confusion with the nomenclature of general clinical practice guidelines, the class of recommendation is graded as grades 1, 2 and 3 instead of I, II, and III.

Hospitals differ in size, organization, volume, and technology. The complexity of cardiology requires the centralization of some technologies and services in selected hospitals for efficiency. High risk and highly complex patients may be transferred to “referral hospitals”, which poses a challenge to the comparison of performance and outcomes without establishing a hierarchy of hospitals with similar resources and patients.40 Most countries within the European Union have reorganized the practice of cardiology by concentrating certain procedures, such as surgery, complex percutaneous interventions, and complex arrhythmia ablations, in fewer hospitals. For quality benchmarking, the task force established 3 types of hospitals defined as low-, intermediate- and high-complexity according to their organization, resources, and the need to transfer patients to other hospitals (Table 2). This classification is arbitrary, may not apply to all countries, and may need refinement in the future. In addition, due to the growing complexity of cardiology practice, routine admission of patients in type I hospitals may not be recommended, except for palliative care or nursing home functions.

Assessing out-of-hospital practice and long-term follow-up presents special difficulties and is not considered in this document.

Quality of care parameters may be grouped in clusters (Table 3), including institution characteristics, available technologies, staffing of the hospital and cardiac unit, organization, certification and accreditation, reputation, and patient opinion.49,58,73 All these clusters may influence outcomes, most are clearly identified in guidelines for clinical practice, and all should be taken into consideration in all hospitals. Some indicate the minimum requirements for accreditation of specific cardiology units such as electrophysiology, interventional cardiology laboratories, and cardiac surgery. Others reflect performance in clinical practice and others are directly related to the measurement of outcomes. Benchmarking of some of these parameters may be difficult, and obtaining the appropriate information may require a dedicated database that is difficult to standardize or complete and is even more difficult to accurately audit. Nevertheless, health care authorities should consider specific requirements for special units and may use some of them for benchmarking but, most importantly, for accreditation. Individual hospitals may monitor selected parameters as measures to identify suboptimal performance and opportunities for improvement.

Of special interest is the organization of safety programs (eg, staff and patient radiation, bleeding, infections, medical errors) and other local programs to improve quality.

Teamwork is always recommended and is mandatory between hospitals that transfer patients on a routine basis.

Mortality constitutes the first and most important metric recommended by this task force to measure the quality of results in clinical practice. The relevance of mortality is self-evident, it remains the most important outcome measure in clinical trials designed to change clinical practice, and is the most powerful evidence to support recommendations in practice guidelines. In many clinical settings, mortality is related to guideline adherence as well as performance measures,103,104,121,122 it is included in different programs that evaluate quality of care,6–9,40,99 and it can certainly be audited (class of recommendation 1, level of evidence A). Mortality may be classified as all-cause mortality, cardiovascular mortality, or other cause-related mortality. All-cause mortality during the index hospitalization is the metric recommended by this task force, as different causes of mortality need adjudication for uniformity, which will not be possible except in dedicated registries. Ideally, mortality at a predefined follow-up (eg, 30 days after index hospitalization) is preferred instead of hospital mortality, but this may be difficult or impossible to ascertain except in well-organized dedicated registries. Mortality should be measured in uniform groups of patients and requires corrections for casemix complexity. Another caveat with mortality is that as a measure it requires a relatively large number of patients and may be statistically misleading or misinterpreted in low-complexity hospitals. In such cases, mortality may be measured through longer time periods and presented per year, but there is no excuse to avoid measuring mortality in cardiology patients.

The length of hospital stay and readmission rates constitute the second metric recommended by this task force. Hospitalization reflects quality of care, impacts health care cost, is commonly used in quality programs,115–122 and is also included in many quality control databases. On the other hand, length of stay may not be as reliable as an outcome metric to compare the results of practice in different countries/areas where hospitalization may be driven not only by medical reasons but also by administrative and social determinants. In addition, rehospitalization may depend on other conditions or comorbidities, which are always difficult to properly determine. For this reason, hospitalization is recommended as a quality metric only when hospitals participate in a prospective, dedicated registry, where criteria for admission and discharge are predefined, or the cluster of hospitals is uniform. Ideally, hospitalization should be measured in a predetermined time period (eg, 30 days), but if reliable measurements are not possible, length of hospital stay is preferred and recommended. The task force also recommends measuring unplanned readmission for any cause to any acute care hospital within 30 days of hospital discharge (class of recommendation 2, level of evidence B).

In-hospital or post-discharge myocardial infarction is one of the components of the main outcomes in clinical trials and registries in patients with ischemic heart disease. However, it may be a poor metric for outcomes due to the difficulties of standardizing the diagnosis in large populations, in particular during the first few days after hospital admission for acute coronary syndromes (ACS),115–123 and should only be used in dedicated, prospective, controlled registries (class of recommendation 2, level of evidence B).

Disabling stroke is self-relevant, is related to iatrogeny, percutaneous coronary intervention (PCI), surgery, and the use of antithrombotic therapy. Stroke is a metric included in registries and some quality programs.10,79,114 However, minor forms of stroke are difficult to diagnose without the routine use of imaging techniques, there are no reliable scales for stroke risk in different clinical settings, and this metric may represent a confounding factor for benchmarking if not centrally adjudicated.124–129 Stroke is a most important component for outcomes in clinical trials but inappropriate evaluation may lead to inaccurate representation of hospital performance and may potentially have serious unintended consequences; accordingly, stroke is only recommended as a quality measure when considering well organized, controlled, and audited registries (class of recommendation 2, level of evidence B).

Safety parameters such as major bleeding, medical errors, infections, cardiac tamponade during percutaneous interventions, and other relevant clinical complications of clinical practice should be considered in quality performance reports. Again, the complexity of achieving uniform diagnosis and reporting in a large number of hospitals precludes the use of safety parameters for benchmarking of quality except when data are prospectively obtained in dedicated, controlled registries. Nevertheless, major bleeding, stroke, infections, medical errors, cardiac tamponade, and other safety parameters should be recorded locally to identify opportunities for improvement (class of recommendation 2, level of evidence B). Safety measures in quality programs will be addressed in detail in another publication.

Comparisons should be made only between similar hospitals and in selected, well-defined, high-risk specific populations with prognosis known to be dependent on overall cardiology treatment (diagnosis-related groups [DRGs]).40,58,130–138 Diagnosis-related groups relatively homogenize diagnosis and procedures, but are divided into too many groups, sometimes arbitrarily. Extreme high-risk and low-prevalence groups of patients that may only be admitted by some highly selected hospitals (such as patients with endocarditis, trauma, and those with complications of noncardiac surgery) should be excluded from analysis rather than corrected for risk.136,137 Sometimes this information is not well reflected in registries or databases, highlighting the importance of dedicated databases for the measurement of quality outcomes.137 Furthermore, diagnosis at admission may be imprecise (eg, endocarditis) or even worse, not included in the ICD-9-CM codes113 (eg, prehospital cardiac arrest admitted unconscious to the hospital). Exclusion of these DRGs could provide more uniform and reliable groups for benchmarking. Widespread introduction of the ICD-10 (Tenth Revision of the International Classification of Diseases) codes will improve the classification of patients considering more contemporary diagnoses. Table 5 shows the recommended DRGs to assess overall quality of results in clinical practice and the recommended reference values.40,58,70,72,75,99,131–133,139–159

Some corrections are needed for risk adjustment. Table 6 summarizes the advantages and disadvantages of the most common strategies for risk adjustment. At a minimum, corrections should be made for age and sex. The use of specific and validated risk scores will provide further refinement and make the metrics more reliable for benchmarking. Whenever possible, the use of simplified risk scores validated in clinical practice is strongly recommended.160–172 Nevertheless, some are too complex and difficult to assess in large populations, such as some important parameters (eg, biological markers) that may not be routinely used in every hospital and will not be available in every patient. This may be the case with HF.173–175 In such circumstances, it is recommended to use adjusted models, such as that published by the Institute for Clinical Evaluative Sciences of Ontario, Canada,176 which considers common risk factors usually present in clinical risk scales (age, sex, shock, diabetes mellitus with complications, congestive HF, malignant tumor, cerebrovascular disease, pulmonary edema, acute renal failure, and chronic renal failure). In addition to patient demographics and clinical variables, hierarchical models of risk adjustment (multilevel models)176–180 take into consideration specific effects at the “hospital” level. One problem is that the Institute for Clinical Evaluative Sciences of Ontario adjustment model is not universally used, making it difficult to benchmark against other countries/systems. Furthermore, the reliability of this correction has not been fully validated in some specific clinical settings (ACS, stable coronary artery disease, bleeding, surgery, and other invasive procedures) and has not been universally accepted. Hence, whenever possible, more specific risk scores should be used that have been validated in clinical practice and recommended in guidelines. These include the GRACE (Global Registry of Acute Coronary Events) or TIMI (Thrombolysis In Myocardial Infarction) risk scores for ACS,163,165 EuroSCORE II risk score,166,167,172 and others.171,173–175

Table 1 of the supplementary material indicates the population selection and adjustments to compare outcomes among different hospitals. The recommended ICD-9-CM codes are listed in Table 2 of the supplementary material.

More complex adjustments allow the calculation of other indexes, such as the risk-standardized mortality ratio (the ratio of predicted mortality, which considers, on an individual basis, the functionality of the hospital treating the patient) to expected mortality (which considers a standard functionality according to the average of all the hospitals), multiplied by the crude mortality rate40,176; however, these metrics may be more difficult to understand by nonexpert observers (for whom the metrics and benchmarking are intended) and the lack of universal standardization makes benchmarking unreliable.

Universal standardization for risk correction should be a priority of scientific societies committed to improving the reliability of benchmarking in quality of care.

Quality of care audits highlighting performance measures and outcomes are of interest to physicians and medical personnel, healthcare authorities, and the general public. Therefore, reporting of quality measures for outcomes should be transparent and available to all stakeholders. Use of the Internet for benchmark reporting is recommended but should be overseen by health care authorities or scientific societies.

Rates (eg, crude and risk adjusted) should be chosen over other forms of reporting (eg, odds ratios, predicted mortality) as they are better understood and preferred for benchmarking.178–184 The use of terms such as first, best, last, and worst, in benchmark reporting is discouraged. Averaged values or recommended target values should be included for reference. Table 7 summarizes different types of reporting results for benchmarking. Simple data are preferred for clarity.

Graphic representation is preferred over tables for clarity. Graphs for clustering should include numbers in different hospitals or hospital clusters, as well as graphs for trends through different time periods. Median values and a possible reference value (eg, target value recommended in guidelines) should also be included as a reference target for outcomes for a particular measure. Figure 2 illustrates 30-day mortality rates in different European countries.32

Figure 2.

Graphic reporting of metrics for benchmarking between different hospital clusters. Extracted with permission from the Organisation for Economic Co-operation and Development. Different 30-day mortality rates after admission for acute myocardial infarction, 2009 (or nearest year).32 The suggested reference mortality rate is also indicated. EU, European Union. Reproduced from Health at a Glance: Europe 201232. Reference values from Steg et al.115

(0.28MB).

Tables may include detailed information but such information may be confusing or distract from the main objective of benchmarking. Tables should be complemented with figures that include main outcomes, preferably with actual values in percentage format, in addition to ratios and other information (Figure 3).183

Figure 3.

Graphic reporting of differences in hospital mortality after surgical procedures (noncardiac) between European countries in the 7 days of the study. Adjusted odds ratios graph and table including detailed data. AVR, aortic valve replacement. Reproduced from the Society for Cardiothoracic Surgery in Great Britain & Ireland.183

(0.96MB).

Trends in outcomes through different time periods are encouraged to illustrate the progress of a particular marker in a hospital or hospital cluster (self-benchmarking). This type of representation is illustrated in Figures 4182 and 5.138,179182.138,179 Combined reporting of metrics may illustrate a possible relationship between changes in treatment strategies and outcomes (Figure 6).184

Figure 4.

Trends in outcomes for in-hospital mortality after first-time aortic valve replacement. 95%CI, 95% confidence interval; AVR, aortic valve replacement; IQR, interquartile range; OR, odds ratio; UK, United Kingdom. Reproduced with permission from Pearse et al.182

(0.31MB).

Figure 5.

Benchmarking between different hospitals (single site and cluster of hospitals in a country (Spain) and complete cohort (world) showing temporal trends in door-to-balloon time in hospitals with primary percutaneous coronary intervention facilities. Global Registry of Acute Coronary Events. Adapted from Spanish benchmark reports, Fox et al.138.

(0.1MB).

Figure 6.

Combined reporting of metrics illustrating the change in the use of effective treatments in acute myocardial infarction and mortality. Berlin registry. A: medications and reperfusion therapy. B: hospital mortality for ST-segment elevation myocardial infarction and non—ST-segment elevation myocardial infarction. ACE, angiotensin converting enzyme; ARB, angiotensin receptor blockers; ASA, acetylsalicylic acid; PPCI, primary percutaneous coronary intervention; STEMI, ST-segment elevation myocardial infarction. Adapted from Röehnisch et al.184

(0.31MB).

Some quality markers are recommended for the accreditation of cardiology units of all hospitals (eg, staffing, technology, volumes); others are aimed at controlling internal quality or identifying problems and opportunities for improvement and are recommended for all hospitals.54,181,185–255 Arguably, the most relevant recommendations are the use of local protocols for diagnosis and treatment, based on ESC/AHA or country-specific guidelines and approved by the hospital.103,104,196 Teamwork with internal medicine and other related specialties, with special reference to primary care, should be a priority.186–195Table 8 shows selected metrics and Table 3 of the supplementary material provides a more detailed description of clinical cardiology and hospital-related markers.

Cardiac imaging constitutes the core for diagnosis in cardiology and its rapid development in recent years, as well as its complexity, requires specific training and teamwork with other specialists. Technology should be available in all hospitals, in-house. or in referral hospitals. Transthoracic echocardiography performed by well-trained cardiologists is recommended in all patients and in all hospitals. For more complex techniques requiring specific training, accreditation, and certification is strongly recommended and teamwork may be useful with radiologists (nuclear imaging, cardiac computed tomography, cardiovascular magnetic resonance). Accreditation of imaging laboratories by the ESC or other official accreditation agencies is recommended, particularly in hospitals classified as type II and III. Quality controls include accreditation, low interobserver variability, and prompt systematic reports. Protocols aimed at reducing radiation doses in CT scans and the systematic reporting of the total radiation dose are recommended in every case256–292Table 9 shows selected metrics in cardiac imaging.

Acute cardiac care requires teamwork with out-of-hospital professionals, emergency departments, and internal medicine and intensive care physicians that follow well-defined protocols for common cardiac conditions such as acute myocardial infarction and ACS. Protocols that follow the guidelines must be developed, approved, and implemented in all cases. Patients with ST-segment elevation myocardial infarction (STEMI) should be immediately referred to hospitals with available primary PCI. Well-trained nurses are of the utmost importance in emergency departments, medical wards in type II and III hospitals, and intensive care units. A dedicated intensive care cardiology unit is strongly recommended in type III hospitals, whereas in lower volume hospitals, a general intensive care unit should have specific protocols to transfer patients with STEMI, cardiogenic shock, and other conditions according to prespecified protocols. In hospitals admitting patients requiring intensive cardiology care, the presence of at least 1 cardiologist certified in acute cardiac care is strongly recommended.115–118,145,199,224,255,290–293

Outcomes include mortality related to STEMI and ACS (Table 5). Local safety controls should focus on antithrombotic complications. Table 10 shows selected performance metrics to improve outcomes in acute cardiac care.

The results of PCI are highly dependent on the expertise and training of interventional cardiologists, as well as on the volume of procedures performed at each hospital and by individual interventional cardiologists. Fellow-in-training activity may have a negative impact on both outcomes and therefore local laws and regulations must be strictly followed. This may have legal implications. Accreditation should be considered in all cases. In general, complex cases should only be treated in hospitals with cardiac surgery support.319 Low-volume, highly-complex interventions (transcatheter aortic valve implantation, closure of left atrial appendage and foramen ovale, valvular and adult congenital heart disease interventions) should be permitted only in selected type III hospitals with specific training and accreditation. Adherence to local protocols based on guidelines and heart team decisions for nonurgent interventions should be considered in all cases.115–118,139,140,294–349

Outcome metrics include STEMI and ACS mortality, as well as transcatheter aortic valve implantation mortality and elective PCI mortality. The main safety control is focused on bleeding, renal failure, stroke and vascular complications requiring surgery or extended length of stay (Table 11).

Interventional treatment of complex arrhythmias requires accreditation of both the laboratory and interventional cardiologists. Indications for catheter ablation and other techniques including cardiac resynchronization therapy and implantable cardioverter-defibrillator implantation are rapidly changing. Ablation procedures in some arrhythmias (eg, atrial fibrillation) are increasing rapidly without proper evidence of benefit in clinical trials. In all cases, the indication should be established after a heart team approach that adheres to the guideline recommendations. Again, accreditation of units and staff is crucial for outcomes and proper legislation should regulate the activity and level of responsibility of fellows in training (Table 12).150–152,331,350–373

Outcomes targets should include complex electrophysiological procedures and device implantation mortality. Safety should focus on complications requiring surgery, transfusions, or prolongation of hospitalization.

Complex invasive electrophysiological procedures may be defined as procedures performed by < 50% of the country's laboratories, including150,350,351,354 ventricular tachycardia catheter ablation, atrial fibrillation catheter ablation, left atrial tachycardia/flutter ablation, percutaneous/surgical epicardial procedures, and referred procedures after failure in other centers.

Noncomplex invasive electrophysiological procedures include catheter ablation of the different substrates in regular paroxysmal supraventricular tachycardia, common atrial flutter, and atrio-ventricular junction nodal ablation.

Diagnosis and treatment of HF are rapidly changing and increasing in complexity, and adherence to guidelines is likely to ensure better outcomes including survival. Many patients require treatment before hospital admission. Most present with comorbidities that require specific treatment and cardiac care must be continued after discharge of the patient from the hospital in all cases. Teamwork as opposed to admitting patients in cardiology or internal medicine is crucial and strongly recommended. Some type of HF unit is strongly recommended in all hospitals. Outcomes include mortality and hospital readmissions. The recommendations in Table 13 apply to all hospitals unless stated otherwise.11,60,67,119,120,156–158,214,246,374–378

Cardiac rehabilitation is more than controlled exercise training. The main objective should be patient education for long-term changes related to lifestyle, adherence to medical treatment for the specific condition, and use of appropriate secondary prevention strategies. In many cases cardiac rehabilitation is neglected, especially for long-term secondary prevention. Cardiac rehabilitation units or programs should be implemented to offer all patients appropriate counselling and follow-up for secondary prevention. Teamwork, especially with general physicians, is essential (Table 14).115,379–405

Quality controls should include access to rehabilitation programs for all patients with ischemic heart disease and adherence to guidelines during long-term follow-up.

Cardiac surgery is closely related to clinical cardiology and team work between both specialties is, without exception, essential. Interstingly, quality controls in cardiac surgery have been implemented in many hospitals in some countries in the past few years. Hospital volumes and the training and expertise of surgeons, anesthesiologists, nurses, and referring cardiologists have a strong impact on outcomes (Table 15).161,406–421

Outcomes are relatively easy to measure and should focus on mortality and length of hospital stay in prevalent, well-defined surgical procedures such as staged, first-time coronary artery bypass grafting, and aortic and mitral valve surgery.

Databases currently used for benchmarking may not have the appropriate quality and all derived information will be misleading, causing a substantial negative impact on scientific and public opinion. Audited prospective mandatory reports would arguably be the best way to capture simple, but at the same time, essential/core information. Dedicated data registries (eg, transcatheter aortic valve implantation, STEMI, PCI, arrhythmia ablation registries) may include more detailed and specific information, but their validation will depend on the universal inclusion of patients and the quality of the audits. However, even in prospective registries, some patients could be missed. This could be even more likely in the sickest patients or those who die soon after admission, illustrating the need for serious and detailed audits.422 Retrospective data collection may yield a different type of information. Voluntary registries including a selected number of patients may not represent true values for benchmarking.

International Classification of Diseases codes are universally accepted but do not clearly allow the identification of DRGs perceived to be of the upmost importance in modern cardiology and must be periodically adapted to properly capture changes in clinical practice. One example is the lack of specific codes for STEMI and other types of myocardial infarction,113 a diagnose that is currently included in most quality control programs; another example is the lack of appropriate codes to differentiate a simple episode of ventricular fibrillation resolved with an electric shock from a complex cardiac arrest in a patient admitted unconscious to the hospital. Future editions (ICD-10 and subsequent) should include the appropriate coding required for quality assessment standards in contemporary clinical practice.

Diagnosis itself may not be as reliable as desirable. Diagnosis of HF leads to a significant number of false positive and negative interpretations (typically, hospitalization for HF is difficult to adjudicate in clinical trials).40,173–176,413,414 The same is true for relevant, high prevalent conditions that require central adjudication in major clinical trials to overcome the different interpretation of clinical data at the local level. These include stroke, myocardial infarction, major bleeding, and cardiovascular mortality, among other conditions.422