Time to Take Health Delivery Research Seriously

Biomedical science has made amazing life-saving discoveries. The science of “omics” offers hope for further improvements in preventing, prognosticating, and personalizing care. Each year, researchers publish approximately 18 000 active clinical trials,¹ yet a gap remains between the development and publication of this new knowledge and better patient outcomes. While clinical translational science awards strive to improve population health, their predominant focus has been on translating basic research findings to humans. Although this Commentary focuses on preventable harm within the health care system, the principles discussed likely apply to optimizing population health using a broader set of public health interventions.

Preventable harms from health care–acquired infections, deep venous thrombosis, and pulmonary embolus have been estimated to account for 200 000 deaths annually in the United States,² and thousands more deaths are associated with teamwork failures and diagnostic errors.^{3 - 4} There is limited evidence that outcomes are improving⁵ despite abundant activities at federal and state levels and within local hospitals to improve the quality of care.

The current approach to biomedical research, including funding priorities and promotion criteria, may partially explain the gap between biomedical advances and health care outcomes. Too often, research ends once researchers and physicians develop and publish new knowledge. For instance, the main finding after decades of federally funded research to reduce mortality in acute lung injury patients was lung protective ventilation. A minority of these patients receive this therapy.⁶ Despite individual physicians having significant accountability for their patients, researchers and physicians have been slow to embrace the science of health care delivery—a science that takes a systems view to improve population outcomes. Most academic medical centers have many researchers studying omics, but are largely devoid of human factors and systems engineers, sociologists, psychologists, or health services researchers who could lead health delivery research (also called knowledge translation research or implementation science). This imbalance is likely driven by the limited absolute and relative funding for health delivery research.

The need for health delivery research is paramount. Patients receive half of the recommended therapies, health care cost increases are not sustainable, preventable patient harms are increasingly visible, hospitals and other care delivery organizations are increasingly financially accountable for patient outcomes, and an explosion in basic and clinical research makes it impossible for physicians to remember all of the latest therapies. For instance, basic research has revealed that breast, colon, and prostate cancers are not 1 disease but dozens. These medical advances offer more accurate prognostic information and therapies that optimize patient outcomes. Nevertheless, such advances also pose significant risks for diagnostic and therapeutic errors given that clinicians are expected to remember an exploding number of omic diagnostic and treatment combinations. Without health delivery research, patients will predictably not realize the full potential of these scientific advances and health care costs may further increase.

Several changes in how health care delivery considers biomedical research could support improvements in patient outcomes. First, researchers should recognize that patient health is the ultimate goal, understanding that health delivery research links clinical research with improved health. This will require adding social scientists, human factors, and systems engineers to the research pool and expanding the capacity to conduct mixed (quantitative and qualitative) methods research to understand whether an intervention works and why. Health delivery research seeks to ensure that patients receive recommended therapies, which involves changing clinician behavior, removing barriers so physicians can comply with best practices, improving teamwork and culture, measuring and providing feedback about performance, and creating incentives (social norms, economics, public policy) for clinicians and patients to encourage the desired behaviors. This research must coexist in the real world of health care delivery, where the context in which an intervention is implemented influences its success. It draws on the expertise of social scientists—experts who are rare in academic medical centers. This research will also require that funding agencies accept multifaceted interventions seeking to optimize patient outcomes, particularly when each component of the intervention is inexpensive, low risk, and low value. When effective interventions are known, the reductionist view of requiring independent evaluations for each component would waste research dollars. Thus, funding agencies and promotion committees must embrace health delivery research.

Second, health care science needs to conduct feedback (outcome-optimizing) research in addition to feed-forward (hypothesis-testing) research. Most biomedical research is hypothesis-testing research, wherein investigators attempt to identify the independent effect of each intervention (eg, drug, or a change in payment policy) on patient outcomes. While necessary to develop new knowledge, hypothesis-testing research is painstakingly slow and insufficient to maximally improve patient outcomes. When the science regarding how to measure and improve patient outcomes is adequately mature, efforts can switch from hypothesis-testing to outcome-optimizing research. Outcome-optimizing research is a systematic effort that starts with the desired human health outcome and works backward, implementing and coordinating multiple interventions by diverse groups to maximally improve outcomes and evaluating their impact on patients, setting research agendas to fill in knowledge gaps.

For example, central line–associated bloodstream infections are associated with approximately 31 000 deaths each year. Decades of research have established evidence for appropriate measurement and practices to prevent these infections. Researchers prioritized these practices, coupled them with multifaceted interventions to ensure clinicians used these practices, and demonstrated the ability to substantially reduce infection rates.⁷ In a federally funded, coordinated national effort, researchers and clinicians kept score with measures they deemed valid, drew upon best practices from clinical research to develop a multifaceted intervention, and created incentives to reduce infection rates.⁸ The multifaceted intervention intended to reduce infections quickly in a patient population, without estimating the relative contribution of individual components of the intervention. This approach appears to be effective. The Centers for Disease Control and Prevention reported a 58% reduction in central line–associated bloodstream infections in intensive care units between 2001 and 2009.⁹ Moreover, data from 22 states implementing the program substantially reduced infections, and mortality among patients in Michigan intensive care units was reduced.¹⁰ While this approach may result in rapid improvements in patient outcomes, stimulating innovation to further reduce this and other preventable harms, it cannot identify the independent contribution of any specific intervention.

Health care funders and researchers can supplement hypothesis-testing research with outcome-optimizing research when there is a valid measure for a patient outcome that can be accurately and feasibly collected from a large number of health systems, when evidence-based interventions to improve the outcome are sufficiently known, and when pilot tests demonstrate that a multifaceted intervention improves patient outcomes. Policy makers can support the creation of national mechanisms to monitor performance and share evidence-based practices while encouraging local provider organizations to innovate, coordinating social, economic, and regulatory pressures to change behavior to optimize outcomes.⁸

The benefits from investing in biomedical and clinical research fall far short of their full health potential because limited funding and attention go into health delivery and outcome-optimizing research—research that can improve health care quality by aligning stakeholders in an important and unified purpose with a valid and measurable goal. Greater investments in health delivery research, for more balanced portfolio and enhanced absolute funding, could produce improved outcomes and reduce costs of care. If patients are to receive the full benefit of medical advances, health care has to take health delivery research seriously and conduct research to improve population health.