Author + information
- Javed Butler, MD, MPH* ( and )
- Andreas Kalogeropoulos, MD
- ↵*Reprint requests and correspondence:
Dr. Javed Butler, Cardiology Division, Emory University Hospital, 1365 Clifton Road NE, Suite AT430, Atlanta, Georgia 30322
A pessimist sees the difficulty in every opportunity; an optimist sees the opportunity in every difficulty.
—Sir Winston Churchill (1)
The U.S. health care system is under unprecedented pressure for reform. Health care costs continue to rise and are deemed unsustainable. The population is aging, and health care needs are projected to increase. To add to the challenge, the global economy is going through a difficult period. However, situations like these beget the best of creativity and enthusiasm. It is not surprising that multiple options of how to re-engineer health care delivery are being discussed. Although the alternatives are many and disparate, a common element is the focus on improving health care quality. Widespread evidence of suboptimal overall care and the variations in care provided across both the hospital types and the geographical regions is difficult to dispute (2).
To effectively manage quality, there are 3 important prerequisites. First, quality must be defined in a measurable matrix. Second, there should be vehicles to measure the quality matrix. Third, the root causes for the numerical facts should be understood. Without adequately addressing all 3 facets, effective intervention development and implementation is difficult. To move the quality notion from a metaphysical to a measurable paradigm, quality measures for various diseases are being defined. The American College of Cardiology and the American Heart Association have defined heart failure performance measures that include warfarin use in patients with chronic or recurrent atrial fibrillation (3).
To measure and manage these indicators, various forms of registries have come into existence. In this issue of the Journal, Piccini et al. (4), using one such registry, the American Heart Association's Get With The Guidelines (GWTG) Heart Failure Registry, provide insights into the use of anticoagulation for atrial fibrillation in heart failure patients. The results are both important and sobering. The investigators report that warfarin use was suboptimal, that it did not improve over time, that there was substantial regional variation in use, and interestingly, that warfarin use was inversely related to the CHADS2(congestive heart failure, hypertension, age, diabetes, prior stroke or transient ischemic attack) risk score. This study underscores the potential of large registries for improving health care, because individual hospital data are unlikely to give this detailed insight. It is possible, and very likely, that this study will lead to development of quality improvement interventions that will hopefully improve on the current trends.
Are all of the conclusions of this study justified? Registries are snapshots of real life and should be interpreted as such. Interpreting registry data with approaches and terms stemming mainly from clinical trials and cohort studies is unlikely to provide insights that will help improve quality of care. In their study, Piccini et al. (4) encountered a situation commonly observed in registries, that is, aggressiveness of treatment decreases as risk increases. This seems paradoxical and alarming at first. However, in most such instances, only risk is measured formally, for example, with the CHADS2score for atrial fibrillation. Unfortunately, the physical, mental, and social vulnerabilities, all strongly associated with risk, are usually not systematically recorded. The median age of patients with atrial fibrillation in this study was 79 years, yet common geriatric problems such as cognitive impairment, fall risk, dizziness, visual disturbances, and so on, and dependency in activities of daily living were not recorded. This probably reflects the fact that these conditions are not part of the traditional medical model and thus are not common targets in the current system of health care. In line with the traditional model, eligibility for warfarin treatment was defined by lack of formal contraindications. However, up to 50% of individuals ages 65 years or older have 1 or more geriatric conditions, and a sizable proportion are dependent on others for their extended needs (5). These and other forms of vulnerability may reduce an older person's ability to participate in their care; this is especially important for oral anticoagulation and may have affected the physician's decision to start warfarin. Thus, before concluding that physicians failed to start warfarin in 1 of 3 eligible patients, it might be prudent to consider and ideally to record these real-life issues. Indeed, the fact that despite regular feedback there was no trend in improvement over 3.5 years is telling us something!
So how do we measure health care delivery and quality data? In this respect, the last decade has seen a rapid proliferation of registries with rapid growth in both numbers and scope. These include registries operated by voluntary health and patient advocacy organizations, for example, GWTG registries by the American Heart Association; professional organizations such as the National Cardiovascular Data Registry by the American College of Cardiology; and the pharmaceutical industry, for example, the Acute Decompensated Heart Failure National Registry. The impact of these registries in understanding the disease process, contemporary epidemiology, racial and sex disparities, and treatment trends, and their impact on influencing guidelines, are too numerous to cite. Most importantly, they are great tools for managing evidence-based medicine and improve quality of care.
Registries have evolved over time. Starting with primarily individual hospital-based operations to later regional and then national initiatives, registries evolved from paper-based tools requiring manual data entry to sophisticated software tools that can communicate with electronic health records. From an intermittent individual look at the data, these have matured to generate periodic comparative reports. Advanced features such as decision support tools, real-time reporting and benchmarking, and customization for inquiry and reporting are featured in several registries already, for example, the GWTG registry. The evolution, as impressive as it has been, is not perfect. Despite their success, we believe that the future evolution of registries needs to consider the following.
Developing smarter registries
With the proposed governmental support in the American Recovery and Reinvestment Act, the use of electronic health records will likely increase. This may directly or indirectly increase the scope and participation in registries. However, these registries must go beyond mere data collection tools and move from discovering numerical truths to understanding the root causes behind the numbers. Can we have device registries with real-time data collection to understand the rationale for medical decision making? It is impossible to design practically feasible registries with an exhaustive list of inquiries; however, point-of-care data collection to capture medical decision rationale is certainly possible. Without understanding the “why” behind the trends, the implicit assumption is that variation is related to simple explanations such as not remembering to prescribe, but the reality is more complicated. Health care quality literature is perpetually plagued by lack of a true denominator, that is, how many patients were truly eligible for certain treatment options. Reasons why physicians do not follow guidelines are complicated. Without understanding the root causes, designing effective interventions is difficult. The current study is one example. Do we just take it at face value that physicians use less anticoagulation when the risk of stroke is higher? Perhaps we should give more credit to the physicians. At least part of these trends may be explainable by data not captured in registries. Thus, registries that go beyond the usual fact determination to understanding reasons behind decisions will be immensely useful.
There is an aversive reaction to mandating anything in medicine. However, as we restructure health care and build in more incentives for quality improvement, participation in registries may increase. In certain cases, for example, left ventricular assist devices, the very nature of the therapy precludes enrollment of large numbers of patients in randomized trials before approval. Perhaps mandating participation in certain registries is justifiable because this will be the only way to learn appropriate patient selection and management strategies.
Develop a business case for registries
Development and implementation of an infrastructure for registries can be expensive. Participating in registries at the hospital or practice level may not be very costly and usually requires nominal participation fees and salary support for personnel to submit data. In the current health care environment, adding any costs will be scrutinized and likely rejected. However, the return on investment in terms of real-time data, opportunities for intervention implementation, and financial incentives for quality improvement may make it attractive to participate in registries. For the up-front cost of initiating registries, those that are funded by specific external sources may not have the leverage to be conducted in the most beneficial and unbiased manner. Thus, a business case for registries needs to be developed. These may include cost shifting from other less productive medical endeavors; cost sharing between providers, payers, or vendors; or other alternatives. Ideally, we may assign a value system to registries based on their need and benefit; however, the current economic environment will likely not support such a notion.
Electronic health records and registries
Although currently it may not be costly to participate in any individual registry, costs add up when participating in multiple registries. As the push for improving quality of care accelerates, it is likely that registries will be developed for multiple disease states. Personnel support for data extraction and submission for all such initiatives can be expensive. Thus investment in electronic health records in which the database may be programmed to extract data for various measures for different disease states is likely to provide the best option.
Registries as marketing tools
Without an independent business model, registries may often be funded by entities with a vested interest. Caution is needed to not have the commercial aspects guide the content and conduct of registries, which may lead to losses in both the data content and credibility.
Registry and clinical trial databases in many cases are not easily accessible. Considering the costs and effort needed to collect these data, one can understand such a dynamic. However, in certain cases this can lead to unnecessary obstacles in analyzing the data that could be used for public good. Rapid analysis and dissemination of information should be encouraged. One may even argue that at the time of individuals consenting to allow access to their data for trials and registries, or when obtaining a waiver of consent by the institutional review boards, it is not spelled out explicitly that data access will be limited. A common concern cited is to ensure the database and analysis credibility, which might not be possible with widespread data distribution. However, this is manageable. There are examples of National Institutes of Health-sponsored studies that can be accessed by any investigator, but there are safeguards placed against misuse, including analytic plan and manuscript reviews by specific committees before journal submission. Similarly, complying with certain privacy rules is another hurdle. Any attempt that increases rapid data dissemination for public good should be encouraged.
Registries are not substitutes for randomized control trials
One must caution regarding the use of registry data to generate evidence. Medical literature is full of association studies from databases that do not hold up in randomized trials. There are no statistical tools that can control for unmeasured confounding. Registries are great for managing evidence-based medicine developed by randomized clinical trials, but not for replacing them.
With appropriate safeguards, widening horizons, utilization of electronic and web capabilities, and government support for electronic health records, and by creating sound business models, the use and usefulness of registries will continue to evolve and expand. In the meantime, Piccini et al. (4) should be congratulated on their work and we should go back to the drawing board to figure out the root causes of suboptimal utilization and how to optimize anticoagulation in heart failure patients with atrial fibrillation.
Dr. Butler serves as the Deputy Chief Science Advisor for the American Heart Association. All perspectives in this editorial are the authors' views and do not necessarily reflect either the American Heart Association or Emory University's standpoint.
↵* Editorials published in the Journal of the American College of Cardiologyreflect the views of the authors and do not necessarily represent the views of JACCor the American College of Cardiology.
- American College of Cardiology Foundation
- ↵Winston Churchill quotes. Thinkexist.com http://thinkexist.com/quotation/a_pessimist_sees_the_difficulty_in_every/15269.html. Accessed August 20, 2009.
- Bonow R.O.,
- Bennett S.,
- Casey D.E. Jr..,
- et al.
- Piccini J.P.,
- Hernandez A.F.,
- Zhao X.,
- et al.