To better understand and improve the quality of cardiovascular care, the profession, led by the American College of Cardiology (ACC) and American Heart Association (AHA), has created an infrastructure to advance QA/QI. These efforts include the development of data standards, evidence-based clinical guidelines, performance measures, and appropriate use criteria (AUC). Each of these tools serves a distinct yet complementary role in improving care (Fig. 3–5) and is described in detail in the following sections.
Overview of data standards, guidelines, performance measures, and appropriate use criteria.
To measure and improve care, one first needs to know both how and what to measure. It is critical to have standardized data definitions that enable the reproducible collection of data across different hospitals and settings. To create the foundation for clear, explicit data capture, the ACC/AHA Clinical Data Standards were developed to serve as a foundation for implementing and evaluating the other ACC/AHA quality tools. 48 These data standards are a set of standardized definitions of particular conditions and treatments that can and should be applied in both QA/QI activities and, importantly, clinical trials. Inclusion in clinical trials is particularly important to support both comparability across studies and their incorporation into guidelines, performance measures and clinical care. In particular, standardized definitions support the consistent definition of symptoms, comorbidities, and outcomes in many areas of CVD (eg, acute coronary syndromes, congestive heart failure, PCI). 48 The more these data standards are used in clinical trials, observational registries, and QA/QI efforts, the greater the ability will be to translate the emerging knowledge from clinical research to clinical care.
Clinical Practice Guidelines
To distill the rapidly expanding body of cardiovascular literature, professional agencies, such as the AHA and ACC, have commissioned expert committees to synthesize the available evidence into clinical practice guidelines. 49,50,51 These guidelines are an evidence-based application of published studies, ideally large-scale clinical trials, within a particular area of CVD. When substantial randomized clinical trial data are lacking, smaller clinical trials or expert panel consensus are used to recommend clinical care in specific circumstances. 50 Clinical practice guidelines are written in the spirit of suggesting diagnostic or therapeutic interventions that appear to be effective (or not) for patients in most circumstances. They are the primary activity through which the rapidly evolving literature is synthesized for practicing clinicians and form the foundation for other quality tools, such as performance measures and AUC. 49,50,51
The first clinical practice guideline was developed in 1984, 52 when overutilization of pacemaker implantation led governmental regulators to ask the ACC and AHA to evaluate the available evidence and develop recommendations for practice. Since then, there has been an exponential growth in the work performed by the ACC/AHA Task Force on Practice Guidelines, 49,50,51 which now currently monitors almost all areas of cardiovascular care. 53 To remain current, the guidelines undergo periodic revisions and updates, as needed by emerging science. Importantly, the development of guidelines has shifted from procedure-related guidelines to disease-based guidelines, where the population of patients is more reflective of clinical care rather than being restricted to the subset of patients referred for a particular procedure. Although these guidelines are intended to assist providers in clinical decision making, they do so by describing broad principles and generally acceptable approaches for a particular disease condition. 49,50,51
The creation of guidelines requires writing committees to systematically review the medical literature and to assess the strength of evidence for particular treatment strategies. This necessitates ranking the types of research from which knowledge is generated. Randomized controlled trials are given the highest weight. When these are not available, other study designs, including preintervention and postintervention studies, observational registries, and clinical experience are used. To transparently communicate the strength of a recommendation and the evidence on which it is generated, a class recommendation (Class I = strongly indicated, Class IIA = probably indicated, Class IIB = possibly indicated, or Class III = not indicated) and strength of the evidence (Level A evidence [data derived from multiple randomized trials] through Level C [data derived from expert opinion, case studies, or standard of care]) are provided. 51
Despite being evidenced-based, there are important limitations in the development of clinical guidelines. Although summarizing all of the available evidence is a requirement, it results in lengthy and difficult-to-read documents, sometimes longer than 400 pages. To address this, the ACC/AHA also publishes pocket guidelines, executive summaries, and web-based applications that are distillations of the key recommendations without the justification for those recommendations. Another limitation of guidelines is that they tend to become outdated as a result of an inherent time lag in the process of developing these documents. The ACC/AHA now releases timely focused updates to minimize the delay between the generation of new evidence and its incorporation into practice. Another important criticism is that the current method used to rank the strength of the evidence is focused on clinical trials that show some benefit, regardless of whether or not the amount of benefit is clinically important (or the benefit may be seen in a surrogate outcome, rather than a clinically meaningful outcome). 54 Importantly, the summary of clinical trials, which report the average benefit across the entire population, fail to emphasize the heterogeneity of treatment benefit, whereby some patients may benefit greatly and others do not. Ongoing efforts, using Bayesian analyses and other modalities that can exploit the heterogeneity of treatment benefit, are being explored to improve the quality of the guidelines development process. 54 Notwithstanding these limitations, clinical guidelines are an important resource that serve as the foundation for all other QI efforts in professional cardiology.
At times, the evidence supporting (or for avoiding) a particular diagnostic or therapeutic action is so strong that failure to perform such actions jeopardizes patients’ outcomes. Performance measures represent that subset of the guidelines for which the strongest evidence exists and for which their routine use (or avoidance) is felt to be an important advance to elevating quality. 55,56,57 Creating performance measures entails a distinct methodology from that of guidelines creation 55,57 and, as such, is undertaken by a separate ACC/AHA Task Force writing committee.
Performance measures are often constructed as a set of measures that quantify a range of health-care processes and outcomes (Fig. 3–6) and are designed to identify multiple points in the continuum of care for which clinical inertia—the failure to implement or titrate recommended therapies—can occur. 58,59 Once the relevant domains are identified, those guideline recommendations with the strongest evidence and highest correlation with clinically meaningful outcomes are selected for performance measure creation. Constructing process-of-care performance measures requires the explicit articulation of the denominator of eligible patients, what constitutes compliance with the measure, over what period of care such compliance is needed, what source of data will be used to define these characteristics, and how the measure will be analyzed and reported. 57 Once prototypical measures are proposed, their feasibility, interpretability, and actionability need to be established. In contrast to performance measures for processes of care, outcomes of care can also serve as performance measures if they fulfill established criteria and have the capacity to be risk-adjusted for patient characteristics present prior to the initial delivery of care. 60
Example of care dimensions for ambulatory care: the care continuum. Reproduced with permission from Spertus JA, Eagle KA, Krumholz HM, et al: American College of Cardiology and American Heart Association methodology for the selection and creation of performance measures for quantifying the quality of cardiovascular care, J Am Coll Cardiol. 2005 Apr 5;45(7):1147-1156. 57
Various organizations are involved in the development of performance measures. Scientific bodies such as the ACC and the AHA are involved in the science behind these performance measures. However, organizations such as the Centers for Medicare & Medicaid Services, the Joint Commission, the National Quality Forum, the National Committee for Quality Assurance, and the Ambulatory Quality Care Alliance play key roles in either developing performance measures or adjudicating their value for national QI efforts. Established performance measures are now assuming an important role in public accountability and pay-for-performance initiatives. 61,62 These are rapidly evolving initiatives and warrant the attention of clinicians in their recommendation, interpretation, and application. In addition to the more traditional clinician-focused conceptualization of performance measures, patient participation as part of performance measures is encouraged to improve patients’ outcomes, including health status. 63 Similarly, challenging the historical exclusions of resource utilization and cost considerations in guideline developments and performance measures (although often implicitly considered), newer documents have started to consider cost in measure selection. 64
Over the past several decades, the United States has witnessed a substantial increase in the use of diagnostic testing and therapeutic procedures in cardiovascular care. For example, the National Heart, Lung, and Blood Institute estimated that more than 1.3 million coronary angiography procedures were performed in 2008, an increase of almost 350% since 1979. 65 However, this increase was not uniform and had marked regional variations, such as that documented by the Dartmouth Atlas. 66
Given the nation’s increasing concerns about the escalating costs of health care, there was a pressing need to understand such regional variability. This has led to the conceptualization of appropriateness, which includes underuse (the failure to provide services from which the patient would benefit), misuse (performing procedures in the correct patients but doing so in a manner that results in harm; see Performance Measures), and overuse (where tests and procedures may not be needed or even be harmful to the patient). Inappropriate use of tests and therapies clearly accelerates costs to both patients and society and exposes patients to potential harms.
To address the demand for clearer insights into the appropriateness of care, the ACC, along with the AHA and other professional organizations, created a framework to begin evaluating the selection of patients for specific diagnostic tests and therapeutic interventions. In 2005, the ACC published their methodology for creating appropriate use criteria in cardiovascular imaging. 67 They defined a procedure as appropriate if “the expected incremental information, combined with clinical judgment, exceeds the expected negative consequences by a sufficiently wide margin for a specific indication that the procedure is generally considered acceptable care and a reasonable approach for the indication.” 67
The AUC differ from clinical guidelines and performance measures in several important ways. First, they estimate the relative benefits and harms of a procedure or a test for a specific indication. This is done by first creating prototypical patient scenarios commonly encountered in clinical practice. Using guidelines, clinical evidence, and contemporary practices as a guide, an expert committee then evaluates the strength of the indication for that procedure or test. After an adaptation of the RAND Delphi approach, 68 a multistep process is pursued that engages the multidisciplinary expert panel to individually and collectively rate the appropriateness of a test. Ratings range from inappropriate (1–3), to uncertain (4–6), to appropriate (7–9). 67 The mean value of the collective responses and the degree of agreement with these ratings are then reported and form the initial AUC for that procedure or test. This methodology is continually updated as experience accrues, 69 which changes the nomenclature and expands the AUC framework for all cardiovascular technologies and procedures.
The AUC thus help identify what specific tests and procedures to perform and when and how often and, as such, have the potential to more transparently document the clinical rationale for performing tests and procedures. They overcome the limitation of the Dartmouth Atlas by providing a framework against which to judge observed variations. Since 2005, published AUC documents have assessed many areas of cardiovascular medicine, such as echocardiography (both transthoracic and transesophageal, as well as pediatric cardiology), 70,71 stress echocardiography, 72 cardiac computed tomography 73 and cardiac magnetic resonance imaging, 74 single-photon emission computed tomography myocardial perfusion imaging, 75 joint criteria for utilization of cardiovascular imaging, 76 utilization of imaging in heart failure, 77 multimodality detection and risk assessment of stable ischemic heart disease, 78 diagnostic catheterization, 79 coronary revascularization, 80 and peripheral vascular ultrasound and physiological testing. 81 Future directions for these efforts are to explicitly contrast the relative appropriateness of alternative diagnostic modalities for specific clinical indications.
Adoption of AUC in routine practice has resulted in marked improvement in certain areas of cardiovascular care, such as coronary revascularization. A study examining the appropriateness of PCI showed that between July 2009 and September 2010, 98.6% of PCIs were classified as appropriate, 0.3% uncertain, and 1.1% inappropriate for acute indications (AMI and high-risk unstable angina). 82 However, for nonacute indications, 50.4% were classified as appropriate, 38.0% uncertain, and 11.6% inappropriate. Similarly, although there was minimal variation in the proportion of inappropriate PCI across hospitals for acute indications, there was substantial variation for nonacute indications (median hospital rate, 10.8%; interquartile range, 6.0%–16.7%). 82 In a separate study, inappropriate PCIs for nonacute indications were more common in men, whites, and patients who had private insurance, suggesting overuse in traditionally privileged groups. 83 Recently, the trends in appropriateness over time have been reported. This showed that since the publication of the original AUC for coronary revascularization in 2009 there were significant reductions in the volume of nonacute PCI, particularly among those classified as inappropriate nonacute PCIs (26.2%–13.3%). 28 Despite this, hospital-level variation of inappropriate PCIs persisted (median, 12.6%; interquartile range, 5.9%–22.9%) in 2014. 28
Although AUCs are a promising method for the profession to guide the more cost-effective use of expensive technology, there are potential challenges with the AUC. First, developing AUC is limited by rigorous scientific evidence for a number of common clinical scenarios. Thus, validation of these criteria—establishing that those with more appropriate indications obtain more benefit from the procedure than those with less appropriate indications—is important. Although this has been shown for the coronary revascularization AUCs, 84,85 it has not yet been assessed for the other AUC ratings. Second, AUC are a general guide to clinical care and cannot possibly address the many extenuating factors that might lead to very rational decisions in real-world practice. This has an important implication on the interpretation of AUC results in clinical practice. In the use of performance measures, inclusion and exclusion criteria for the denominator often mean that the expected goal of compliance with performance measures is 100%. This will not be the case with AUC. There will be patients who are rated as uncertain or inappropriate, using the coarse definitions with which the AUC were created, who would clearly benefit from a procedure. However, it is unlikely that such patients will differ markedly by practice. Thus, those practices that are outliers, as compared with others, warrant careful evaluation of their treatment practices. For example, finding that a practice in which less than 50% of patients are deemed appropriate will raise concerns of overuse, whereas a practice in which 100% of patients are considered appropriate may suggest underuse of the procedure in patients who might benefit. More experience with the use of AUC will improve their use and interpretation as the country struggles to maximize the value of cardiovascular care.