Abstract
Falzer (2018) effectively describes the application of naturalistic decision making (NDM) to the best practices regimen. We agree on the increasing challenges posed in health care regarding clinical decision making under uncertain conditions, as magnified by the burden of exponentially expanding clinical knowledge, the timeliness and comprehensiveness of available data, as well as care and choice complexity. However, we see no inherent conflict between engineering for complex systems and the “best practices regimen.”
As human factors practitioners and clinicians working to tackle diagnostic error, as well as many other clinical decision-making challenges, we fully recognize the criticality of a systems approach. If human factors engineering has taught us anything, it is that we do not redesign humans; we redesign the systems within which humans work. Human factors engineering focuses on industrial engineering, cognitive psychology, information processing, and perception to maximize the use of human strengths and accommodate human limitations. Falzer’s call to identify and address limitations in the design of complex interactive systems with a focus on how systems operate in actual practice is important and aligned with the human factors and systems safety approach. The gradual shift toward evidence-based medicine can and should be designed to take a systems approach with a better understanding of the decision-making practices of health care providers and the resulting impact on service delivery. It is why current efforts to produce value-based care encompass multifarious initiatives and programs that run the gamut of health care improvement, ranging from training and education to performance improvement and research. As noted by Falzer, the impact of case-specific factors (arguably in addition to environmental, resource, and nontechnical challenges) does not exclude suboptimal decision making and resistance to change. Providers face challenging clinical cases in resource-constrained and complex environments. As the complexity of a system increases, the ability of the human to fully understand the intricacies of a system diminishes without a countervailing force.
The case study presented by Falzer highlights the challenge of effectively implementing evidence-based guidelines. Despite advances in evidence-based medicine, there remains a gap between the evidence-based knowledge developed through research and the systematic application of that research in the clinical setting. Given an insufficient focus on the effective translation of evidence-based knowledge to frontline clinical practice and the sheer number of clinical practice guidelines promulgated by many diverse authoritative bodies, it is not surprising that uptake by frontline clinicians is low. Our ability to produce clinical care guidelines is currently far more advanced than our ability to translate and implement them into clinical care, even acknowledging that guideline making is prone to the overstatement of the evidence base. For example, take evidence-based guidelines integrated as clinical decision support (CDS). After more than three decades of evidence-based guideline development and despite the development of more sophisticated CDS platforms, such as SMART on Fast Healthcare Interoperability Resources (Mandl et al., 2012), in practice, most CDSs in clinical use have been either stand-alone systems or small components embedded within electronic health record (EHR) or physician order entry systems that ignore critical and complex factors that clinicians face when making care decisions. To truly facilitate successful support of clinical decision making, a range of contextual factors must be taken into account, such as clinical workflow, competing goals, and administrative incentives.
That evidence-based guidelines face barriers at the point of dissemination and implementation, requiring deliberate assessment of usability during integration into practice, does not preclude the goals of the “best practices regimen.” Any clinician would agree that “conflating the general effectiveness with case-specific appropriateness” is an error, but it is also the case that when there is a systemwide practice that is in conflict with well-established evidence, there is an ethical obligation to measurably change practice in the aggregate. As a simple example, consider the use of albuterol (a bronchodilator medication), steroids, and chest x-rays for acute bronchiolitis, an illness characterized by obstruction of the small airways of the lungs from viral infections in infants. Well-meaning physicians have been using all three interventions for years, despite an abundance of evidence indicating that, for typical infants with bronchiolitis, there is no clinical benefit, and all three have some risk of harm (Ralston et al., 2014). A quality improvement collaborative was able to decrease the use of all three interventions in aggregate at multiple sites, saving many infants from unnecessary and potentially harmful interventions (Ralston et al., 2016). Note that the collaborative did not target zero utilization of albuterol, steroids, or chest x-rays. All three may be appropriately used in certain clinical contexts, which an expert provider may recognize as being suggestive of asthma or pneumonia—clinical conditions that may be the true diagnosis for infants presenting with apparent bronchiolitis. Yet the aggregate reduction of unnecessary care without any unintended consequences is, without a doubt, an improvement in clinical care.
The problems with overly simplistic approaches to decision making and health information technology extend well beyond guideline implementations. A wealth of research demonstrates the inefficiencies associated with a variety of health care interventions: alerts that are ignored >90% of the time, increased rates of errors, adverse events, and mortality are all outstanding concerns. Entire technology systems have been abandoned because of problems with poor reliability, credibility, and consistency, as well as problematic user interface design. Effective solutions indeed must satisfy a number of constraints arising from, for instance, clinical needs, clinician preferences, social interactions, cognitive limitations, and health care policy; however, we do not fully understand to what extent safety problems for patients or inefficiencies for providers can be attributed to the mistaken use of guidelines as mandates for conformance, rather than to the other sociotechnical challenges prevalent in our health care system today.
NDM focuses on articulating and understanding the complex characteristics of decision making (e.g., ill-structured problems and uncertain, dynamic environments; information-rich environments where situational cues change rapidly; cognitive processing that proceeds in iteration). Falzer promotes expertise to contend with complex situations, enhancing skills in evaluating and aggregating evidence and instituting programs to develop teamwork and collaboration. Researchers use NDM methods in richly grounded, clinically informed ways to accurately evaluate the entire interactive system. We believe the future of the best practices regimen is in the translation of evidence to the frontline provider informed by the theories of NDM and human factors engineering, supporting providers of all levels of expertise in coping with the accelerating growth of knowledge while allowing clinical experts to operate to the top of their capacity.
In addition to novel methodological approaches, emerging technologies can bridge the gap to support evidence-based practice. Rapid advancements in research, scientific discovery, and health information technology are generating innovative decision-making strategies and technologies to support clinician behavior. For example, there are synthesized evidence databases, such as ePocrates, UpTo-Date, DynaMed, EvidenceCare, and Visual DX. Furthermore, researchers are implementing CDS that works not only for the clinician but for the patient and family as well (Fiks, 2011). Solutions must be collaboratively developed across stakeholders to address key challenges and accelerate the availability of reliable information resources that are smoothly and affordably incorporated with patient-centered, clinician-friendly workflows. To fully leverage and extend the EHR, health care systems need to draw on the expertise of content vendors, where clinical knowledge can be packaged concisely and delivered efficiently to the point of care. This will require participation from all stakeholders: EHR vendors, health care providers, developers of practice guidelines, and researchers. Clinical decision making in modern, patient-centered, health information technology–enabled environments requires us to reimagine and rethink decision making.
Footnotes
Kristen Miller, DrPH, CPPS, is a senior research scientist at the National Center for Human Factors in Healthcare at MedStar Health. She is a clinically oriented human factors researcher focusing on medical decision making, informatics, and the assessment of medical interventions, with an emphasis on health information technology, usability, and patient safety. She applies systems thinking and human factors engineering to support the delivery of high-quality care and, most notably, is dedicated to optimizing clinical decision making through new technology that reduces the consequence of disease and understanding the mechanisms by which providers receive, understand, and respond to technology. Her training spans three public health degrees, a postdoctorate with the Department of Veterans Affairs, and experience with multiple health care systems. Her research portfolio includes federally funded work from the National Institutes of Health and the National Science Foundation, evaluating clinical decision making and clinical decision support.
Naveen Muthu, MD, is a pediatrician in hospital medicine and an informatics researcher at the Children’s Hospital of Philadelphia (CHOP). He established and now leads the Cognitive Informatics Group in the CHOP Department of Biomedical and Health Informatics. He attended medical school at the University of Missouri before completing residency in internal medicine and pediatrics at Georgetown University Hospital and a fellowship in clinical informatics at the CHOP. He is now board certified in pediatrics, internal medicine, and clinical informatics. His current research focuses on the application of cognitive informatics to study the work system interactions and safety of pediatric health information technology, develop novel pediatric clinical decision support tools, and evaluate clinical decision support interventions throughout the health information technology life cycle.
Raj M. Ratwani, PhD, is the center director and scientific director of MedStar Health’s National Center for Human Factors in Healthcare and an assistant professor of emergency medicine at the Georgetown University School of Medicine. He is an expert in applying human factors theories to health care to improve safety, efficiency, and quality. His areas of focus include workflow analysis, task interruption and resumption, and health information technology usability and safety. His work has been sponsored by several federal agencies and foundations, and his research has been published in high-impact journals. He has also focused on optimizing federal policies to improve electronic health records. He has testified in front of the U.S. Senate Health, Education, Labor and Pensions Committee and currently serves on the 21st Century Cures Act Health Information Technology Advisory Committee. He holds a PhD in human factors and applied psychology.