A Direction for Revitalizing Quality Improvement in American Cardiovascular Care

Avedis Donabedian, the founder of modern quality improvement, recognized the importance of valid and accurate data several decades ago.¹ Nevertheless, we are still struggling with how to integrate and regulate data to drive quality improvement. This task is critically important. In the United States (US), cardiovascular disease still accounts for about one third of all deaths, more than 800,000 Americans per year.² The cardiovascular death rate in the United States is substantially higher than other countries, including Canada, Australia, and Guatemala.³ Since technology and medications in the US are not inferior to these other countries, the higher death rate in the US suggests that factors including public health and quality of care – rather than the technological sophistication of care – are associated with increased, preventable mortality.

In that context, focusing on innovative approaches to both improve the quality of care for patients with heart disease and those at risk of developing heart disease seems appealing. Improving quality delivered to American cardiology patients may provide substantial benefits in terms of health outcomes. The growth of electronic health records (EHRs), clinical quality registries such as the National Cardiovascular Data Registry (NCDR) and the Society of Thoracic Surgeons (STS) database, new validated decision tools, and measures of health status provide opportunities to improve quality. Despite that apparent opportunity, in recent years, improvements in mortality from cardiovascular illness has slowed.

That is why this new American Heart Association scientific statement⁴ on the learning health care system and cardiovascular care is so welcome. This statement provides a path forward on how to translate our increasing data resources into care innovations that can meaningfully improve our patients’ lives. As noted in the scientific statement, cardiovascular care leads in the development of clinical registries. Those registries have already been associated with quality improvement on key process measures. Registries therefore are an important place to start in terms of developing a learning health care system. The authors propose to integrate registry data collection into standard work flow and link these data to other data resources. As a practical matter, the most prominent limitations of registries are that they do not include patients not in the registries and can impose substantial data collection burdens. So these are sensible steps that could expand the early successes of clinical registries. Since cardiology registries have generally been established earlier then registries in other areas, lessons from cardiology registries will help generate knowledge about how to operate registries more broadly. The authors have formulated additional recommendations about how to build on the progress from clinical registries, integrating additional data sources such as semi-structured data from the EHR derived through natural language processing and patient-reported outcomes data.

Critically, the authors also propose reforms in the ways that research oversight occurs regarding quality improvement. Currently, there is substantial variation between institutions with respect to the oversight of this type of work. Often, data analyses with the primary purpose of improving the quality of care locally is exempt from review from institutional review boards (IRB). That exemption is often still true if the results are later published in medical journals or disseminated in other ways. Bringing this type of work under the authority of traditional IRBs, on the other hand, would likely slow quality improvement, and seems unwise. The sort of oversight – and intensity of oversight – needed when new devices or drugs are tested on patients seems different than what is needed for a learning health care system. Risks to patients in a learning health care system should focus more on data privacy, rather than direct harms from the testing of novel devices or drugs. This issue is likely to become more relevant, as insights from local quality improvement programs become increasingly important to other institutions. In this scientific statement, the authors propose new forms of regulation, more sensitive to the critical need to generate insights at operational speed.

Many physicians and other care providers have bemoaned the administrative burden associated with the EHR. As a clinical cardiologist myself, I experience that burden often. Furthermore, the aggregation of data in clinical registries imposes overhead costs on institutions. Even the collection of patient reported outcomes data sometimes imposes substantial burdens on patients. In this sense, many of the costs of a learning healthcare system have already been imposed, both by providers and patients.

Instead of protesting these burdens, it behooves us to start thinking more deeply about the payoffs of this data infrastructure. The burden of data collection is the cost we have already paid. The tremendous value of this scientific statement, therefore, is that it points a path forward to start realizing the concomitant potential benefits of all these data in terms of improving quality and improving the lives of patients with heart disease.

So much of the excess mortality attributable to cardiovascular illness in the US is related to insufficient public health measures and risk factors. But much of that excess mortality also is related to a quality of care that is too often uneven. We must start understanding better how to improve physician risk-estimation for individual patients and personalized decision making, improve the safety of procedures, and work with patients to advance adherence to effective therapies. Understanding so many of these questions would be facilitated by a learning health care system – a system with still under-realized potential. This scientific statement, therefore, represents a remarkable step forward that will help us improve the lives of our patients.

Donabedian became interested in quality measurement when studying epidemiology and health services administration at the Harvard School of Public Health, after immigrating from Lebanon where he previously had directed the health service at the American University of Beirut.1 As a clinician who must have witnessed firsthand the shortcomings of practical care delivery before struggling with how to improve health care delivery, I am sure that he would have been gratified to read this timely and essential scientific statement.