Outcomes Research in Cardiovascular Disease
Report of the NHLBI Working Group

January, 2004

I. Objective

The objective of the National Heart, Lung, and Blood Institute (NHLBI) Working Group on Outcomes Research in Cardiovascular Disease (CVD) is to advise NHLBI on opportunities in Outcomes Research related to CVD, and to provide guidance on priorities in allocating resources for Institute-sponsored and investigator-initiated CVD Outcomes Research.

Meeting Agenda

II. Background

The definition of CV Outcomes Research.  For the purposes of this Working Group, "Outcomes Research" is defined as applied clinical research that generates knowledge to improve clinical decision-making and health care delivery to optimize patient outcomes. It is the study of the delivery and consequences of health care on outcomes from the perspective of patients, providers, and the health care system. Outcomes Research focuses particularly on providing evidence to assess and promote the effectiveness of therapeutic interventions, enhancing the migration of best practice to clinical practice, elevating the quality of care, and supporting the optimal allocation of resources for all patients. The research has direct, proximate, and immediate relationships to patient care and health-care delivery. Research addresses not only patterns of care, but also elucidates factors contributing to those patterns, and determines their consequences.  The research also is directed toward developing strategies to address current deficiencies in knowledge or implementation directly relevant to clinical practice. 

Outcomes Research is not defined by a particular methodology, but rather its objective – the study and eventual improvement of the end results of health care. Outcomes Research can employ methodology of non-experimental (observational) designs, as well as quasi-experimental or clinical trial approaches to test interventions, whichever is most suited to the specific research question. Outcomes Research draws from the disciplines of clinical epidemiology, statistics, health services research, psychology, sociology, anthropology, economics and management sciences.

The need for CV Outcomes Research. Despite the accelerating pace of basic and traditional clinical research in CVD, clinicians and policy-makers often lack knowledge about how best to promote CV health and health care delivery in clinical practice. Despite all our advances we lack basic information about contemporary patterns of care. We have a limited array of valid tools to quantify and improve clinical performance. We often define efficacy but commonly do not know the effectiveness of different clinical strategies in actual practice as well as the impediments to achieving the true potential of the best interventions. We need information concerning the impact of clinical and policy strategies on a broad range of patient outcomes, not just mortality and events. We need to know how to best involve patients in decision making and the best approaches to translating evidence into clinical practice so that outstanding quality of care is universally available.

Outcomes Research generates knowledge about the effectiveness of clinical strategies; the health and health care experiences of patients (including measures of health status); quality of care; the balance of costs and effectiveness; and studies of clinical decision-making. Outcomes Research can be descriptive and prescriptive, focusing on current deficiencies in knowledge or interventions to improve clinical practice. Thus, Outcomes Research aims to inform patients, providers, health care systems, payers, and policy makers of the effects of health care and provide data to support the development of evidence-based strategies (at the patient, provider, and system levels) that will best serve the health needs of the nation. CV Outcomes Research applies these aims and approaches specifically to the prevention or treatment of CVD.

CV Outcomes Research can support national initiatives to improve health care. The goals of CV Outcomes Research are closely aligned with a number of existing federal initiatives.

• Healthy People 2010 sets forth ambitious goals to reduce CVD risk, improve CVD outcomes, and eliminate health disparities. Attaining these goals will require effective programs that actively support the integration of existing knowledge and discoveries into clinical practice.
• The Agency for Healthcare Research and Quality, the Centers for Disease Control and Prevention, the Centers for Medicare & Medicaid Services, the Department of Veterans Affairs, and the Department of Defense are all seeking to promote the best "real-world" strategies for the prevention and treatment of disease; to determine the success of health care strategies by assessing patient-centered outcomes; to measure adherence to "best practices" (usually representing those embodied in clinical practice guidelines); and to develop interventions to improve the performance of the health care system. Success in these endeavors will depend on research that can identify best clinical practices, evaluate different clinical strategies, identify wasteful resource utilization, explore the involvement of patients in decision-making and study the effects of clinical and policy-related decisions related to CV health.
• The National Academy of Sciences’ Institute of Medicine (IOM) has published a blueprint for improving health care delivery in a document entitled, ‘Crossing The Quality Chasm: A New Health System For The 21^st Century.’ This report identified six core goals for the future of American health care: safety, effectiveness, equity, efficiency, timeliness and patient-centeredness. These domains, which are not currently a focus of traditional CV clinical research, are important topics for investigation and there is an urgent need for research into how best to attain the IOM’s goals in patients with CVD.
• Many federal organizations, including the NHLBI, have strategic plans to reduce health disparities and the newly created National Center for Minority Health and Health Disparities is championing national efforts to identify and remediate disparities in CV care. Central to the elimination of health disparities is research that investigates where disparities exist, assesses their consequences, and explores the causes of differences in outcomes between racial, ethnic and vulnerable socioeconomic groups and develops and evaluates the effectiveness of interventions to eliminate these disparities.

The Working Group believes that the NHLBI can take a leading role in fostering research to support these and other efforts related to CVD care and outcomes, ensuring that health care provided at the patient, provider, and health system level is supported by a strong scientific foundation built on contemporary, patient-centered, clinician-relevant information.

III. Methods Used to Generate this Report

Working Group members, whose backgrounds cover a broad spectrum of experience within Outcomes Research, considered a range of possible research areas, current gaps in knowledge, and the expected CV trends in the nation. The group first reviewed ongoing programs and funded projects within the NHLBI. Related initiatives and priorities set by other National Institutes of Health and federal agencies were also considered. The group had conference calls and an in-person meeting to reach consensus about the research recommendation priorities.

The Working Group sought to identify high-priority areas in CV outcomes research that deserve support and to provide examples of important areas for future investigation. The objective was not to provide a comprehensive list of topics in CV outcomes research, but rather to identify research directions that the Working Group believes require the most immediate action, with particular attention given to the stated missions of the NHLBI and the NIH. The examples listed below are not intended to exclude the consideration of other meritorious research areas or investigator-identified topic areas.

Having described specific priority research domains and specific topics, the Working Group concluded that directed programmatic commitments are needed to support the growth and development of CVD Outcomes Research.

IV. Working Group Recommendations

The Working Group identified several priority areas, including the development of national surveillance projects for high prevalence CV conditions, the promotion of patient-centered care, definition and implementation of best practice, and expansion of studies that consider the cost of interventions in the context of their effectiveness. Within each of these topics, the Working Group described examples of initiatives that could serve the Institute and the public. In addition, the Group identified several areas, including promoting the use of existing data, augmenting the methodology of Outcomes Research, training of new investigators, and increasing study section support, that are critical to the success of CV Outcomes Research.

Top Tier Priorities

National Surveillance of CV Care and Outcomes

Background

Large-scale NHLBI-funded population-based epidemiological studies, such as the Framingham Heart Study and the Cardiovascular Health Study, have helped define the prevalence, determinants, incidence, and consequences of CVD. However, there are no analogous national resources directed to assess the quality of CV care and health outcomes of community-based populations with or at risk for CVD. For example, the Worcester Heart Study has provided important insights, but is geographically limited and does not include long-term outcomes aside from mortality. The Atherosclerosis Risk in Communities (ARIC) surveillance identifies myocardial infarction (MI) and coronary heart disease (CHD) death in four communities (Minneapolis suburbs; Forsyth County, NC; Washington County, MD; and Jackson, MS) by review of hospital discharge diagnosis codes, medical record abstraction, and death certificate reviews. It has no follow-up component except for linking, when appropriate, of events for patients occurring within 30 days of one another. The Minnesota Heart Survey identifies MI and CHD mortality in the Minneapolis/St. Paul area by review of hospital discharge diagnosis codes, medical record abstraction, and death certificate reviews. Non-fatal cases are not followed longitudinally. This study also has an out-of-hospital sudden cardiac death autopsy component. It does not involve any patient interviews.

Other agencies also have databases and projects that are relevant to surveillance, but they also have their limitations. The administrative databases of the Agency for Healthcare Research and Quality provide information about hospitalizations, but lack clinical detail and long-term outcomes. The Centers for Disease Control and Prevention conduct national surveys of risk factors and behaviors, but do not focus on clinical CV conditions. Similarly, the surveys by the Agency for Healthcare Research and Quality provide important insights about access and health behaviors, but do not focus on clinical conditions. The National Health and Nutrition Examination Survey (NHANES), conducted by the Centers for Disease Control and Prevention’s National Center for Health Statistics (NCHS), collects information concerning general population health and diet, but does not address CV-specific quality of care and a wide range of patient outcomes.

Thus, despite rapid changes in CV knowledge, efficacious treatments, and technical capabilities, we have no established national capacity to assess how technologies and treatments are used or to measure their performance when implemented in real world conditions. The generalizability of randomized controlled trials and other efficacy studies is rarely formally assessed but often presumed. Answers to questions concerning population and patient-level impacts of health care across a broad spectrum of outcomes, including mortality, events, quality of life, and cost, are not available. For example, how will drug-eluting stents be adopted and perform in practice? What determines the adoption of internal defibrillators, are there disparities in their use, and what is their impact on patients’ lives? Similarly, the identification of problems with medications or devices often comes through the fortuitous reporting of individual cases rather than an existing systematic process of evaluation.

Given the substantial commitment of resources to improving CV health and health care, a national CV surveillance system could provide the infrastructure to develop new insights into improving practice and target future efforts to improve care and outcomes. The Centers for Medicare & Medicaid Services’ (CMS) Cooperative Cardiovascular Project (CCP) was a limited example of such a system. The CCP was developed as a large, one time cross-sectional evaluation of acute myocardial infarction care in elderly patients. This clinically rich database has supported many studies addressing the clinical management of acute myocardial infarction in the elderly. The CCP was able to produce such research because it generated a database that included a representative national sample and detailed medical record abstraction of information about processes of care and clinical events, supplemented by additional physician and hospital data. Despite the novel value of CCP, the study had notable limitations, including its limited time frame (1994-96), the lack of data on quality of life or functional status, the omission of patients younger than 65 years of age, and the absence of information on outpatient treatment patterns.

Other groups have also sought to provide data on CV care and outcomes. The Veterans Administration has studied the relationship between patterns of health care delivery and outcomes through its Quality Enhancement Research Initiative (QUERI) groups. For example, the heart failure QUERI has identified a national cohort of veterans with heart failure using administrative data, and efforts are underway to link this resource to clinical and outcomes data. Professional medical societies have also initiated CV procedure registries, including the Society for Thoracic Surgeons’ national database for cardiac surgery and the American College of Cardiology’s National CV Disease Registry for percutaneous coronary interventions. Finally, there are clinical registries funded by pharmaceutical and medical device companies. These condition- or procedure-specific registries have added substantially to the medical literature, but most of these current efforts are limited in that they focus on the treatment of hospitalized patients, frequently lack follow-up information, do not have information on post-hospitalization outcomes or utilization of care, lack reliable assessment of detailed patient-centered outcomes, and are often restricted to limited geographic areas, age groups, or other selected patient populations. These also collect information only from hospitals that volunteer to participate. In addition, health services researchers often have limited access to these proprietary commercial databases and their use is aligned with company interests.

Another example of such a proposed surveillance system has been developed in cancer via the NCI Cancer Care Outcomes Research and Surveillance (CanCORS) Consortium (http://healthservices.cancer.gov/cancors/). The CanCORS Consortium is collecting detailed primary data from medical records and surveys of patients and providers to understand variations in the treatment and outcomes of lung cancer and colorectal cancer in 5 states and regions, 5 large health plans, and the Veterans Administration system.

Recommendation

Establish a national surveillance program for CV care and outcomes. We recommend a series of national or multi-regional clinical cohort studies that would bridge the gap between evidence and practice and provide information on practice, care and outcomes. These studies should be undertaken for several key CVD conditions, designed to describe, with longitudinal data and adequate clinical detail, population-based patterns of care across the US, and their short- and long-term consequences. We recommend that initial priority conditions focus on acute coronary syndromes, heart failure, and atrial fibrillation. We also recommend that selected CV procedures (e.g., percutaneous coronary intervention, coronary artery bypass surgery, and implantable cardioverter-defibrillators) be studied based on their high prevalence, impact on patients and the health care system, and variations in clinical strategies and outcomes.

These studies will provide descriptions of current practice patterns and their associations with patient outcomes, support the development of performance measures for quality, allow the identification of ‘best practices’ from which quality improvement initiatives may be developed, stimulate discussion about best practice, generate evidence about the effect of national health care policies, and identify topics for future research. Those that enroll patients by condition (rather than procedures) can also provide the foundation for risk-stratification models and the ability to appreciate which patient subsets benefit most from procedures and other such interventions. Finally, such registries also lay the foundation for the existence of an established network of investigators to describe, refine and test novel strategies that will accelerate the translation of new scientific knowledge into clinical practice.

Because of the urgency in addressing known and suspected health care disparities, we suggest that racial and ethnic minority groups, individuals in poverty, older people and/or other vulnerable subgroups should be well represented, in order to permit appropriate comparisons across subpopulations.

Promoting ‘Patient-Centered’ Care

Background

Recently, the IOM emphasized the need for more ‘patient-centered’ care as a critical step towards improving the quality of American health care. Patient-centered care is that which helps clinicians, "in attending to their patients’ physical and emotional needs, and maintaining or improving their quality of life," and gives patients the opportunity to be "the locus of control" in medical decision-making. Key components of patient-centered care include the assessment of patient-centered health status outcomes (i.e. symptom burden, functional limitation, and health-related quality of life), and the incorporation of patients’ views about treatment into clinical decision-making.

Although clinical research has facilitated tremendous advances in the care of CV patients, surprisingly little research has focused on patient-centered health status and on patient preferences. Studies are needed to improve our understanding of the impact of various CV diseases and therapies on patient health status and to evaluate interventions designed to improve health status outcomes for patients with CVD. Furthermore, studies are needed to assess the utility of health status measurement and elicitation of patient preferences in clinical practice to maximize patient outcomes.

Recommendations

1. Identify determinants of patient-centered health status. We recommend studies to identify the key determinants of patient-centered health status in populations with CVD. These studies will require the collection of longitudinal clinical and health status outcomes. The knowledge generated from these studies can lead to the development of interventions to improve health status for patients with CVD, and to the development of studies focused on the use of health status data to guide clinical decision-making. This recommendation overlaps with the call above for "National Surveillance of CV Care and Outcomes." Thus, as above, we recommend a focus on high prevalence conditions such as acute and chronic ischemic heart disease, heart failure, and atrial fibrillation. In addition, such studies should address high risk and under-served populations such as older persons, women, racial/ethnic minorities, patients in poverty, patients with multiple comorbid conditions (e.g. diabetes, chronic obstructive pulmonary disease, depression), and patients in high-risk settings (e.g. nursing homes).

2. Evaluate patient-centered tools and interventions. We recommend studies to evaluate the utility of patient-centered tools or interventions in clinical care. The measurement and explicit use of patient preferences to promote shared decision-making and the measurement and integration of health status information to guide management of CV disease are promising approaches to increase the appropriateness of interventions and optimize the impact of clinical interventions. For example, CV health status measures may be provided to clinicians to help identify patients at particularly high risk for subsequent adverse outcomes, and who may thus benefit from more intensive management. As another example, potentially modifiable aspects of health status (e.g. depression) may be targeted for intervention. Patient-level interventions, such as patient activation and/or patient training, and interventions aimed at the hospital, health-care system, or community are also important. Where possible, studies should include explicit evaluation of the impact of these interventions on a broad range of patient outcomes (e.g. mortality, morbidity, satisfaction, and health status). Longitudinal studies will be necessary to demonstrate the value of these interventions over time. A focus on high-risk and/or under-served populations should be encouraged. Studies that include a focus on chronic disease management, coordination and integration of care, access to care, and/or transitions between health care environments would be of particular value. Studies are needed in all CV populations, but those with marked health status deficits, such as heart failure, ischemic heart disease, atrial fibrillation, peripheral arterial disease, and cerebrovascular disease, may be of particular interest.

3. Include patient-centered health status outcomes in clinical trials. The NHLBI has been committed to the inclusion of health status outcomes in clinical trials. We recommend NHLBI continue this support and continue considering the inclusion of CV health status measures (i.e. formal measures of symptom burden, functional status, and health-related quality of life) in all relevant CV clinical trials sponsored by the NHLBI. A workshop on integrating health status into CV studies might be an appropriate approach to pursue this recommendation.

Defining "Best Practices"

Background

Many clinical decisions lack a strong evidence base, resulting in uncertainty regarding what constitutes optimal care. Thus, data are needed concerning the comparative effectiveness of different clinical strategies

Also critical to improving care is defining best practice based on the medical literature. This effort requires the impartial, methodologically rigorous synthesis of the research literature into evidence-based guidelines for care. Importantly, this effort should also include rapid integration of new evidence on a "real-time" basis. What is currently lacking is a professionally-recognized core group of experts, such as the Agency for Healthcare Research and Quality Evidence-Based Practice Centers, to provide a rigorous synthesis of the evidence to support the continuous integration of new evidence. The effort could directly support the development and revision of CV clinical practice guidelines, conducted by professional societies such as the American College of Cardiology and the American Heart Association.

Recommendation

1. Evaluate effectiveness of CV treatments and strategies. We propose that the NHLBI enhance its current portfolio of studies to evaluate further the effectiveness of CV treatments and clinical strategies. An effectiveness study is one that is testing an intervention in an applied ("real-world") setting, where the effects of the intervention are affected by both efficacy and adherence/delivery. Studies consistent with this aim may be observational or experimental. Randomized trials have the particular strength of allocating measured and unmeasured factors among the various study groups. Observational study designs can provide important evidence to guide practice where experimental designs are not feasible or forthcoming. They also may have particular utility in quantifying the risks of interventions as they are applied beyond the original efficacy studies.

2. Support evidence-based reviews of CV treatments. We propose the NHLBI provide support for CV Evidence-Based Centers, perhaps in collaboration with AHRQ and the professional societies, which would be specifically charged to provide research and analytic support for the development of empiric, unbiased evidence reviews using accepted evidence synthesis and meta-analytic methods. The Evidence-Based Centers would provide adequate resources to conduct systematic literature searches needed to support scientifically-based national CV care guidelines that are being conducted by professional societies. They would also continuously monitor the published literature and provide ‘real-time’ updates to specific guidelines as needed. The topics identified through the systematic reviews as having insufficient evidence for clinical care recommendations should then be referred back to the NHLBI and disseminated to potential investigators as topics for future research. This effort should be synergistic, not redundant, with current AHRQ efforts.

Implementing "Best Practices"

Background

Numerous studies have documented a large gap between evidence-based care recommendations and the actual care delivered in community practice. These ‘lost opportunities’ have real consequences for patients. Thus, there is a need to generate knowledge about the means to improve physicians’ adoption of optimal therapies in routine CV care. To date, intervention studies designed to increase physician compliance with practice guidelines have often lacked adequate controls and have tended to assess single, rather than multi-modal interventions. Such assessments also have generally focused on hospital-based treatment rather than the outpatient setting. Furthermore, nearly all of these studies have assessed changes in provider behavior over a limited time period, creating uncertainty about the durability of these approaches. Finally, these studies also have been limited in terms of power and scope, with limited patient and provider samples, and have been applied in selected clinical settings and populations. While these early demonstration projects were important first steps for the field, there remains much to do in terms of identifying optimal strategies for implementing best practices.

Recommendations

1. Identify factors associated with provider delivery of effective therapies. We propose a call for observational qualitative and quantitative research to identify strategies to evaluate and promote physicians’ adherence to evidence-based guidelines as well as to characterize key structure and process features of successful implementation of "best practices". These efforts would investigate potential causes for the wide national variation in guideline adherence. Studies may employ existing data sources, utilize the surveillance initiative proposed above, or alternatively gather de novo additional data on structure and/or process variables that may explain variations in treatment. The goals of this research would be to identify successful practices and/or organizational features that can be applied widely. Study of existing ‘leading centers’ could generate novel potential approaches for improving provider adherence that could then be rigorously tested.

2. Test interventions to improve provider delivery of effective therapies. We recommend the continuation and expansion of programs such as NHLBI-RFA HL-01-011 (Trials Assessing Innovative Strategies To Improve Clinical Practice Through Guidelines In Heart, Lung, And Blood Diseases), which was a one-time call for research "to evaluate interventions that represent innovative strategies that can be employed in clinical practice to improve implementation of national, evidence-based clinical practice guidelines." Areas of highest impact would include primary and secondary prevention interventions, acute coronary syndromes, and heart failure. Particular interest should be focused on the transition from inpatient to outpatient settings and longitudinal care. While unique interventions in a particular setting are of interest, priority should be focused on those interventions that can be applied broadly, ideally on a regional or national level.

Patient Involvement in Care

Background

Advances in the prevention and treatment of CVD are often not fully realized because of limited involvement by patients in their own care. Rates of non-adherence with prescribed regimens of pharmacotherapy, exercise, diet, smoking cessation and self-monitoring often range from 30% to 50%. The issue of patient adherence becomes even more salient as management guidelines advocate extensive combination therapy for many conditions. Discontinuities in the long-term management of therapeutic regimens, including the paucity of systems to facilitate feedback to patients regarding their progress in achieving therapeutic goals, also diminish patient adherence.

Previous research has predominantly focused on hospitals and primary care practices. Critical opportunities for improvement in patient adherence exist in home self-management at the transition from outpatient setting to hospital care, and at the transition from hospital care to outpatient care. Maximizing the benefits of pharmacotherapy and lifestyle approaches will require a sustained and substantial commitment to new research initiatives on patient adherence. Future research on patient adherence should focus not only on the patient, but also on the provider and the health care environment.

Recommendations

1. Study approaches to improving patient adherence. We propose calls for research that enhances patients’ adherence to medication regimens, with the consideration of incorporating proven models of behavioral change. Research on patient adherence has tended to be disease-specific, without sufficient attention to individual behavioral factors and cultural factors that may differ among racial, ethnic, and socioeconomic groups that affect the effectiveness of various preventive and treatment regimens. Strategies that mobilize not only the resources and capabilities of individual patients, but the network of family, friends, and others with whom they interact should be developed. As part of this research, there is a need for improved approaches to measuring adherence. These efforts would build on recent NHLBI initiatives that have sought to understand and promote patient adherence. For example, HL-01-005 (Overcoming Barriers To Treatment Adherence In Minorities And Persons Living In Poverty) sought to evaluate interventions in clinical care settings designed to improve adherence to medically prescribed lifestyle and medical regimens but was limited to specific groups. Future initiatives could build on this effort and be less restrictive of the population studied. Also, HL-01-005 required a randomized trial study design, whereas other study designs might be very informative in addressing this topic.

2. Study novel approaches to enhance patient self-management. We propose that the NHLBI foster research on the development and clinical implementation of technologies and systems that enhance patients’ participation in monitoring and management of chronic CV conditions. Devices for self-monitoring parameters such as symptoms, health status, blood pressure, anticoagulation and blood glucose in the home environment are available, yet these devices are not well integrated into treatment practice. Likewise, interactive technologies such as web-based instruction and feedback that can support patient responses to these data are infrequently utilized. Innovative ways of organizing health care personnel and systems of care to empower patients and enhance their involvement in care are also needed. Research support should be directed to evaluate the effectiveness of advanced interactive technologies and systems that incorporate not only instruction but also ongoing monitoring and corrective feedback.

Economic Evaluations

Background

In an era in which health care costs are rapidly increasing, society is demanding that health care offer good value, providing good outcomes at a fair price. Thus, it is no longer sufficient to show that a test predicts outcome, that a therapy is efficacious or that a pattern of practice can be shown to be effective. It is also critically important to show that these services are worth the costs associated with them.

Recommendation

Study cost-effectiveness of clinical strategies. We recommend NHLBI continue including cost-effectiveness analyses in relevant CV clinical trials that it sponsors. Establishing cost-effectiveness in real world settings is equally important to the efficacy evaluations in clinical trials. We recommend that the NHLBI consider fostering investigations of the economic consequences of alternative clinical strategies to reduce risk and diagnose and treat disease in real world settings. These investigations, funded by non-industry sources, are important to guiding reimbursement policies and resource allocation.

V. Additional Recommendations

Promote research using existing databases.

The investment in studies by the NHLBI and other federal agencies has produced several detailed datasets that contain information that can be used to yield important insights even after the specific aims of the project are addressed. The datasets are a national resource and have tremendous potential to yield important insights that can advance clinical science. The NHLBI has done an excellent job making datasets available to investigators. Support for such access to databases should be continued. In addition, we propose that mechanisms be expanded to provide funding support for proposals that make appropriate use of existing data to address important research questions. This mechanism ideally would have a rapid evaluation cycle and be directed specifically for data analysis.

Develop and refine analytic tools in Outcomes Research.

We need to promote the development of advanced analytic tools from other disciplines that can be used in Outcomes Research. Such disciplines include biostatistics, econometrics, and the behavioral sciences. Examples of powerful tools used in these disciplines but generally not supported in CV research efforts are: hierarchical linear models (for nested study designs such as group-randomized trials of physician-level interventions); propensity scores and instrumental variables (for causal inferences from observational evaluation), and structural equation modeling (for the study of unobservable categories underlying observed phenomena such as components of CV quality of care). Basic methodological work also needs to be undertaken to quantify potential differences among providers in their quality of care, to develop and refine tools to assess patient-centered data (i.e. preferences and health status) including strategies to overcome barriers to the use of these data in clinical practice, and to advance the methodology of the handling of missing data. Finally, research methods are needed to enhance appropriate interpretation of data on health status and resource use. If CV researchers are to use these advanced tools, methodological research to further advance the tools should also be supported. To support the interdisciplinary work in this area the NHLBI could consider Program Project grants in Outcomes Research.

Support trainees/young investigators.

Ultimately, CV Outcomes Research will be defined by the investigators who are dedicated to this field. As the field grows and matures, an essential component will be the specific training and career development of young investigators. This training needs to include the development of skills necessary to conduct high-level Outcomes Research and provision of strong mentorship from established outcomes researchers. To this end, we recommend the formation of career development awards (KO8/K23) within NHLBI specifically dedicated to CV Outcomes Research. The review panel for these applications should consist of established outcomes investigators with multidisciplinary representation.

In addition to the standard requirements for KO8/K23 applicants, key components to career development awards in CV Outcomes Research would include:

A. Requirement of a formal educational plan with emphasis on the attainment of an advanced degree (e.g. PhD, MPH) in an appropriate field (e.g. health services research, epidemiology, statistics, public health), unless already attained. Career development awardees would be required to obtain or document prior formal didactic training in topics such as clinical epidemiology, biostatistics, health services research methods, health policy, health economics, and health systems.

B. Strong local and national mentorship including:

• Demonstration of strong local support for the development of the awardee, including commitment of protected clinical time and administrative support.
• Local mentorship in Outcomes Research by one or more investigators who are established independent outcomes researchers (e.g. active funding and a publication track record in the field as defined in this document) and can offer significant time commitment to the development of the awardee.
• National collaborative mentorship with established outcomes researchers in areas that will complement the applicant’s research interests and promote further career development.
• Multidisciplinary mentorship and collaboration.
• An Outcomes Research application by the awardee that addresses one or more important Outcomes Research domains (e.g., the priorities outlined in this document or one or more of the IOM domains of quality), and demonstrates the applicant’s increasing independence as an investigator.
• A detailed career plan from the awardee demonstrating strong commitment to an academic career as an independent outcomes researcher and clear short- and long-term career goals. 

The formation of dedicated career development awards in CV Outcomes Research would directly support many of the key recommendations of the NHLBI Research Training and Career Development report (November 1999), including expansion of existing training programs to address new/emerging areas, the inclusion and emphasis on advanced education of the awardees, a strong emphasis on mentorship plans, and a multidisciplinary approach to research and mentorship.

There is also a need for support for mid-career awards. The NIH has K18 awards (1- 2 years of support for "re-tooling and gaining lab experience" for mid-career investigators who want to broaden or redirect their research).  These awards are usually targeted towards a specific area of NIH interest (e.g. genetics, stem cells, etc.). This might be a good mechanism for the mid-career clinician or scientist who wants to learn Outcomes Research methodology.

We believe that the implementation of career development awards in CV Outcomes Research will help ensure that the NHLBI trains the leaders in this field for the future.

Expand study sections’ expertise in Outcomes Research.

The Working Group recommends efforts be undertaken to ensure that study sections that will review Outcomes Research applications include experts who appreciate the role of Outcomes Research, understand the methods employed, and understand the strengths and weaknesses of different types of research designs employed in this type of work. Ensuring that study sections have appropriate representation of Outcomes Research investigators is crucial to the fair evaluation of applications responsive to the research priorities described in this report.

VI. Conclusions

CV Outcomes Research holds great promise for providing evidence that will guide the evolution of health care delivery and promote a health care system that is safe, effective, efficient, timely, equitable and patient-centered. CV Outcomes Research seeks to discover new knowledge about the prevention, diagnosis and treatment of CVD through a focus on the end results of health care; and seeks to ensure the translation of ‘best practice’ into clinical practice. The focus is on the patient and ‘real world’ settings with the intention of addressing the practical needs of clinicians, clinical policy makers, purchasers and patients/consumers. This document provides a roadmap to support investigations that are directly relevant to clinical care and complement the longer-term investments of the Institute. We believe that a dedicated investment in CV Outcomes Research will directly improve the quality of care delivered in the United States.

Posted: June 2004