A Research Agenda for Diagnostic Excellence in Critical Care Medicine

INTRODUCTION

Diagnosing critically ill patients in the intensive care unit (ICU) is difficult. To arrive at an accurate and timely diagnosis, the intensivist must make sense of continuous data streams in the context of prior diagnostic labels in order to determine or confirm the underlying etiology of the patient’s critical illness, while simultaneously searching for evidence of impending physiologic disaster.¹ Critically ill patients often require life support including mechanical ventilation and sedation, which make data gathering and interpretation more challenging.² The intensivist must incorporate evolving scientific progress in the definitions and diagnosis of commonly-encountered syndromes and reconcile divergent opinions among ICU team members and consultants.³ Clinicians must perform all of these tasks within the pressured ICU environment, which is known to increase the risk of clinician fatigue, emotional stress, and burnout.⁴ As a result, diagnostic errors in the ICU are common and cause serious harm.^5–9

Authors of the National Academies of Sciences, Engineering, and Medicine (NASEM) landmark report, Improving Diagnosis in Health Care, concluded that there is an urgent need for research on the diagnostic process and diagnostic error in medicine and listed high-yield areas for investigation.¹⁰ Diagnostic safety experts have echoed this need for rigorous research.^11,12 Given this call and growing institutional and government support for diagnostic error research,^13,14 the time is ripe for investigators to broaden our understanding of diagnosis in critically ill patients. We present a research agenda geared towards diagnostic excellence in critical care medicine, advocating for a balanced strategy of continued biomedical discovery to improve the diagnosis of individual diseases while also addressing the complex care delivery systems underpinning the diagnostic process in critical care.

GAPS IN DIAGNOSIS RESEARCH IN CRITICAL CARE

The preceding decade brought about increasing knowledge of diagnostic errors in critical care. However, research on the diagnostic process and the development of effective interventions to improve diagnosis among critically ill patients remain underdeveloped. Autopsy studies remain a dominant source of data on ICU misdiagnoses despite autopsy rates below 50%.^5,6,15 There is little information available on the impact of diagnostic errors on critically ill patients, their families, and the ICU team, even though we know that misdiagnoses can contribute to death and disability,^5,6,16 have high financial costs resulting in billions of dollars in payouts for malpractice claims,^17,18 can be psychologically devastating to families,¹⁹ and can negatively impact physicians’ clinical practice, careers, and well-being.^20,21

Because current avenues of research funding largely adopt a disease-focused approach, most research on diagnosis in the ICU is disease-oriented. However, many diagnostic errors are likely underpinned by common diagnostic process missteps that require disease-agnostic approaches to address. It is impossible to fully understand how and why diagnostic errors occur in critically ill patients without studying the diagnostic process as a whole.¹¹ The fundamental components of diagnosis are rooted in, and affected by, the sociotechnical work environment of the ICU (Figure 1).^1,22 We need to optimize the cognitive work of diagnosis as it occurs within the ICU work system and construct a conceptual model of critical care diagnosis to maximize the impact of disease-oriented diagnostic research and develop effective interventions towards diagnostic excellence.

Figure 1. Disease-Specific and Sociotechnical Aspects of the Diagnostic Process.

Sociotechnical factors (work system) underpin the diagnostic process and must be considered alongside disease-specific considerations in order to improve the diagnosis of critical illness.

Illustration by Ani Rofiqah, https://thenounproject.com/

A PROPOSED RESEARCH AGENDA

Defining the Diagnostic Process and the Burden of Diagnostic Error

The Critical Care Diagnostic Process

Improving diagnosis and preventing diagnostic error is only possible with a deep understanding of the diagnostic process, including its strengths and vulnerabilities. The diagnostic process is defined by NASEM as a complex, patient-centered, collaborative activity that involves information gathering and clinical reasoning with the goal of explaining the patient’s health problem. NASEM’s model of the diagnostic process describes a patient seeking care for a health problem, clinicians undergoing an iterative process of information gathering and information integration/interpretation before arriving at a working diagnosis, treatment, response to treatment (serving as further information to confirm or revise the working diagnosis), and patient outcomes. The diagnostic process occurs over time and within the context of a larger health care work system, which in turn influences the process in many ways.²² However, this model does not fully capture the complexity and challenges involved in critical care diagnosis given time and resource constraints and the severity of illness encountered unique to the critical care setting (Figure 2).

Figure 2. Complexity and Challenges in Critical Care Diagnosis.

Adapted with permission of the American Thoracic Society. Copyright © 2021 American Thoracic Society. All rights reserved. Bergl PA, et al. Diagnostic Error in the Critically III: Defining the Problem and Exploring Next Steps to Advance Intensive Care Unit Safety. Ann Am Thorac Soc. 2018;15(8):903-907. Annals of the American Thoracic Society is an official journal of the American Thoracic Society. Readers are encouraged to read the entire article for the correct context at https://www.atsjournals.org/doi/full/10.1513/AnnalsATS.201801-068PS. The authors, editors, and The American Thoracic Society are not responsible for errors or omissions in adaptations.

Further research is needed to create a conceptual model of the diagnostic process specific to critical care. Each specific aspect should be studied to better elucidate the barriers and facilitators of timely and accurate diagnosis (Table 1). Some examples of important areas for future investigation include the following:

Table 1.

Research Directions to Understand the Diagnostic Process and Epidemiology of Diagnostic Error in Critical Care

Potential Areas and Methods of Research
Understanding the Diagnostic Process
Transitions to higher level of care	*Referral communication between referring clinicians and intensivists
	*Information transfer across clinical settings
	*Effect of prior diagnostic labels on ICU diagnosis
	*Professional relationships’ impact on the diagnostic process
	*Institutional characteristics’ impact on the diagnostic process
Information gathering, integration, and interpretation	*Patient and family factors affecting information gathering
	*Time pressure effects on information gathering and processing
	*Effects of clinician cognitive load on information processing
	*Contributions of subspecialty consultants to diagnosis
	*Effects of clinician task-switching on diagnostic outcomes
	*ICU team dynamics’ contributions to diagnostic outcomes
Communication of the diagnosis	*Communication of the diagnosis to the patient and family
	*Communication relevant to the diagnostic process, including handoffs, clinical notes, electronic messages
Impact of the ICU environment	*Contributions of the ICU built environment to the diagnostic process
	*Safety climate and interpersonal relationships’ impact on diagnosis
Epidemiology of Diagnostic Errors
Burden of diagnostic error	*Interventions to improve autopsy rates in the ICU
	*Use of novel imaging technology to perform “virtual autopsies”
	*Standard and validated chart review tools to determine diagnostic error
	*Electronic trigger tools for identification of medical records with high likelihood of diagnostic error
Factors contributing to diagnostic error	*Survey of ICU clinicians’ perceptions of why diagnostic errors occur
	*Qualitative interviews of patients/families and ICU clinicians
	*Clinical vignette studies presenting critical care scenarios to clinicians
	*Simulation studies to investigate team dynamics in diagnosis

ICU - intensive care unit

Higher level of care transitions.

Unlike transitions of care to home or a long-term care facility, transitions to higher levels of acute care are relatively understudied, despite work showing that patients often have diagnostic discrepancies between frontline and tertiary care settings.^23,24 Emerging research shows that characteristics of referral communication may affect the diagnosis on admission of patients to the ICU.²⁵ Further work is needed to determine how information transfer, electronic health records, professional relationships, and institutional characteristics between referring clinicians and receiving ICU teams affect diagnosis.

Critical care clinicians’ cognitive load.

High cognitive load is a known risk factor for diagnostic error.²⁶ Clinicians practicing in the ICU perform many cognitive tasks under time pressure³ and make hundreds of significant medical decisions per day.²⁷ Research using validated tools to measure information load²⁸ and simulation to mimic critical scenarios in the ICU²⁹ are needed to quantify the cognitive load of clinicians especially during stressful situations. Simulation may also be helpful to determine the occurrence and causes of cognitive errors in diagnosis when assessing critically ill patients.³⁰

Critical care team communication.

Communication is at the center of much research evaluating teamwork in critical care settings.^31,32 However, little work has focused on relating the characteristics of communication with changes in the diagnostic process or diagnostic outcomes among critically ill patients. Information is conveyed in various ways (verbal vs. written, in-person vs. through electronic or mobile messaging) and should be investigated for their impact on diagnosis.^33–35 Communication during handoffs between clinicians is especially ripe for further study. Handoff practices among ICU attending physicians are heterogeneous, which are perceived by physicians to be related to inappropriate care and serious adverse events,³⁶ though the quality of handoffs has not been linked to diagnostic error.

ICU environment’s impact on diagnostic decision-making.

The critical care environment encompasses both the built physical environment of the ICU and the professional climate within which clinicians practice. In the NASEM report, the environment was emphasized as an important factor in the diagnostic process,³⁷ thus research is needed to investigate its impact on diagnosis in critical care. In the ICU, qualitative work shows that open, connected, and visible physical spaces encourage “macro-cognitive interactions” among team members.³⁸ In a randomized simulation trial, unprofessional communication, specifically rude behavior during operating room-to-ICU handoff, resulted in failure of the ICU team to detect and overcome diagnostic error.³⁹

Diagnostic Error Epidemiology in Critical Care

Most of our knowledge of the epidemiology of diagnostic errors in critical care come from autopsy studies. Although autopsies remain a useful way to identify missed diagnoses and innovations such as the virtual autopsy⁴⁰ have increased autopsy rates beyond traditional post-mortem exams, they remain limited to patients who die. Autopsies clearly reveal the ways in which diagnostic errors can be fatal but are less useful for studying misdiagnoses which cause harm short of death. Scientific advances in improving diagnosis will require knowledge beyond autopsies as the main source of information regarding diagnostic error in critical care (Table 1).

Retrospective cohort studies performed via medical record reviews are more representative of the general ICU population;^7–9,41 however, selection of study samples and the process of chart review have been variable. The relatively recent development of validated record review tools such as Singh et al.’s Revised Safer Dx instrument,⁴² which has been adapted for use in both adult and pediatric critical care settings,^7,8 should lead to more standardized reviews in future work, allowing for valid comparisons across studies. Electronic trigger tools, which mine large amounts of data to identify signals indicative of possible diagnostic error, are also a recent innovation and helps focus manual chart review efforts towards patients at high risk for error.⁴³

Aside from the burden of error, factors associated with misdiagnosis also need to be identified. In addition to autopsies and chart reviews, other innovative quantitative and qualitative approaches should be used to collect richer data.⁴⁴ Even simple surveys of ICU clinicians have revealed perceived threats to accurate and timely diagnosis.⁴⁵ Qualitative observations using ethnography with audio and video review can also reveal clinician behaviors and interactions which help or hinder diagnosis.⁴⁶

Identifying and Implementing Solutions to Improve Diagnosis

Advancing Science to Better Diagnose Critical Illness

Accurate diagnosis ultimately depends on advances in biomedical science. As much as there are preventable diagnostic errors due to failures in the diagnostic process leading to missed or delayed diagnosis, there are also unavoidable diagnostic errors caused by inadequate scientific knowledge of illness or underdeveloped methods to diagnose disease (Figure 3).⁴⁷

Figure 3. Unavoidable Diagnostic Errors.

Failures in the diagnostic process leading to failure to identify a known disease is a preventable diagnostic error. Sometimes, clinicians fail to identify a disease because of inadequate scientific knowledge, which then causes unavoidable diagnostic errors. With scientific advances over time, diagnostic errors in this category should decrease.

Reprinted with permission from De Gruyter. Newman-Toker DE. A unified conceptual model for diagnostic errors: underdiagnosis, overdiagnosis, and misdiagnosis. Diagnosis. 2014;1:43-48

The collective efforts of investigators over the decades have pushed the limits of critical care science forward, making previously unknown illness or ill-defined conditions recognizable to intensivists. Sepsis is the quintessential example of a disease -- and diagnostic process -- seemingly in constant flux as clinicians’ understanding of the disease has evolved to keep pace with scientific discovery. Previously seen as a combination of infection and systemic inflammatory response syndrome (SIRS),⁴⁸ sepsis is now widely regarded as life-threatening organ dysfunction caused by a dysregulated host response to infection.⁴⁹ New methods of detecting sepsis in the ICU have also emerged such as the use of procalcitonin levels,⁵⁰ adding to intensivists’ arsenal of diagnostic tools. Further, there may be many phenotypes of sepsis, a finding which can lead to useful sub-categories of diagnosis that can help target treatment.⁵¹

Without losing sight of the sociotechnical components of diagnosis in critical care, we need to continue disease-specific efforts that will improve the diagnosis of unique pathophysiology in the ICU. This work includes basic and translational research into the role of biomarkers and genomics for precision diagnosis in the ICU^52,53 and clinical research on diagnostic tests and imaging to improve bedside assessment such as the use of point-of-care ultrasound.⁵⁴ Similar to the evolving definitions of sepsis, we need to continue to fine-tune diagnostic criteria for other critical illness such as acute respiratory distress syndrome,^55,56 develop scoring systems to accurately and reliably identify conditions relevant to ICU care such as delirium,⁵⁷ and recognize and better define emerging ICU-related conditions such as post-intensive care syndrome or “chronic critical illness”.⁵⁸

Leveraging Health Information Technology

Diagnosis in critical care is vulnerable to human limitations. Clinicians have fallible memory, are prone to cognitive bias,²⁶ and communicate ineffectively.⁵⁹ Such limitations are compounded by common challenges in the ICU environment including high workloads, constant distraction, and communication barriers.⁶⁰ Health information technology (HIT) can help clinicians overcome or minimize these constraints⁶¹ and should be a strong focus of research to improve diagnosis in critical care (Table 2).

Table 2.

Research Directions to Leverage Health Information Technology to Improve Diagnosis in Critical Care

	Potential Areas of Research
Electronic Health Records
Improving clinical documentation to support diagnosis	*Revise how problem lists are created and maintained
	*Optimize how clinical reasoning is documented by ICU team members in clinical notes
	*Develop natural language processing techniques to allow free text entry while retaining the ability to extract needed data for billing, quality improvement, and research
Improving data organization and presentation to clinicians	*Optimize clinical summarization functions of the EHR
	*Develop graphical data visualizations optimized to ease clinicians’ cognitive load
Interoperability and data standards	*Increase interoperability of EHR systems within and across organizations for better information flow
Feedback functions	*Develop electronic tools to deliver automated clinician feedback on patients’ changing diagnoses
Telemedicine
Configuration of telemedicine care model	*Determine optimal characteristics of critical care providers, recipient ICUs, patient populations to maximize benefits to diagnosis
	*Develop implementation structures and workflows to best support diagnosis
Integration of other HIT interventions into telemedicine	*Investigate how AI and machine learning can be used to improve diagnosis, leveraging the large volumes of data generated by telemedicine
Clinical Decision Support Systems
Traditional CDSS *	*Re-design traditional CDSSs to improve utility in assisting with diagnosis, improve usability, and prevent alert fatigue among clinicians
Integration of other HIT interventions with CDSS	*Study the effects of embedding clinical care pathways into CDSSs on the diagnostic process
	*Investigate the impact of AI and machine learning-powered CDSSs on diagnostic outcomes
CDSS and ICU workflow	*Optimize the manner in which CDSS are incorporated into ICU workflows
Artificial Intelligence and Machine Learning
Development and training of AI systems	*Identify specific diagnostic scenarios most appropriate for interventions using AI and machine learning
	*Determine appropriate ICU datasets for training AI systems tailored to specific diagnostic goals
	*Identify and correct for healthcare disparities that may be included in training datasets
Validation of diagnostic output	*Validate AI systems’ diagnostic output under realistic or real-world ICU conditions
AI and ICU workflow	*Study how AI systems can be deployed most effectively within ICU workflows
	*Optimize the balance between decision support and clinician autonomy in diagnostic decision-making

ICU - intensive care unit, EHR - electronic health record, HIT - health information technology, AI - artificial intelligence, CDSS - clinical decision support systems

^*Traditional CDSS includes simple alerts, reminders, order sets, medical calculators, and care summary dashboards

Because HIT applications can be quite broad, it may be helpful for investigators to map potential interventions to a conceptual model of critical care diagnosis in order to specify which aspects of the diagnostic process are specifically being targeted.⁶¹ Some innovations such as the electronic health record (EHR) can have multiple applications and can assist clinicians in many aspects of diagnosis. Here we outline specific HIT interventions applied to improve diagnosis in the ICU and potential future areas of investigation for each:

Electronic health records.

The EHR has transformed clinical practice in ways both desired and unintended.⁶² Future directions in research on how the EHR can further improve diagnosis in critical care thus involve not only developing new applications but also preventing or mitigating the unintended consequences of EHR use. Pioneers in diagnostic safety have suggested many ways in which electronic clinical documentation can be improved to support diagnosis,^34,62 which in turn need to be studied for applicability to the critical care setting. These include revising how problem lists are created/updated and ensuring that clinical reasoning is documented in the medical record. Natural language processing should allow extraction of structured data elements from free text documentation for billing, quality improvement, and clinical research.⁶³ Research is also needed to determine optimal ways of summarizing and presenting information to clinicians to ease cognitive load and minimize bias in data interpretation.⁶⁴ Because critical care patients are almost always referred to the ICU from a different clinical setting, efforts to increase interoperability of EHRs between organizations should be a priority to ensure smooth transfer of critical information between institutions.⁶⁵ The EHR’s ability to provide feedback on diagnostic performance is also understudied—there is an opportunity to develop automated feedback systems so that clinicians can be informed of patients’ changing diagnoses even after transfer out of the ICU.⁶⁶

Telemedicine in critical care.

Telemedicine, the use of medical information exchanged using electronic communication, is rapidly becoming an established care delivery model in critical care since it extends ICU expertise.⁶⁷ Telemedicine is associated with lower ICU and hospital mortality.^68,69 However, little work has been performed to determine the impact of telemedicine on diagnosis. There is literature showing that implementation of a tele-ICU is associated with improved teamwork and safety climate,⁷⁰ which can potentially translate into better diagnosis; however, optimal characteristics of critical care providers, recipient ICUs, patient populations, and implementation structures/workflows have not been identified.^68,69 Telemedicine will likely be a fixture of ICU care in the future, therefore dedicated research is needed to delineate how this care model affects the diagnostic process. Implementing other types of HIT applications such as clinical decision support and artificial intelligence (AI) onto a telemedicine platform is also a promising new area of investigation.⁷¹

Clinical decision support.

A clinical decision support system (CDSS) is any electronic system designed to aid directly in clinical decision-making, wherein characteristics of individual patients are used to generate patient-specific recommendations presented to clinicians for consideration.⁷² CDSSs can assist critical care clinicians in a wide range of diagnostic activities, from the simple selection of diagnostic tests to support for resolving diagnostic dilemmas.⁷³ Most CDSSs are now embedded within and deployed using the EHR. Traditional examples include alerts, reminders, order sets, medical calculators, and care summary dashboards. Over time, CDSSs have included information retrieval tools (“infobuttons”) available alongside clinical information to aid clinicians in the search and retrieval of patient- and context-specific knowledge.⁷⁴ CDSSs have also incorporated clinical care pathways, which are structured multi-disciplinary plans of care operationalizing evidence-based guidelines for criteria-based implementation of standard diagnosis and management.⁷⁵ Although several systematic reviews have shown evidence of improved implementation of health care processes (specifically, ordering appropriate diagnostic tests) in diverse clinical settings using CDSS,^72,75,76 there is still minimal evidence of impact on diagnostic outcomes especially among critically ill patients. In addition to improving diagnostic algorithms and the medical knowledge databases that underpin CDSSs, future directions should also include investigations into improving clinician acceptance and devising optimal ways to integrate these systems into ICU clinical workflows.⁷⁷

Artificial intelligence and machine learning.

The advanced analytic methods of artificial intelligence (AI) systems and machine learning techniques (a subset of AI) —where computers process large amounts of data to learn from examples, rather than being preprogrammed with rules based on human inputs—have the potential to fundamentally change medical practice.⁷⁸ In the data-intensive environment of the ICU, such systems can assist clinicians in efficiently processing large volumes of information to make expeditious and accurate diagnosis. The most practical impact of AI on critical care diagnosis will likely be through machine learning-powered CDSSs designed to be used interactively by clinicians at the bedside. Early studies have shown promising results related to the prediction of physiologic instability and early detection of conditions such as sepsis, acute respiratory distress syndrome, pulmonary embolism, and acute kidney injury.^79,80 However, a recent systematic review showed sparse evidence of association between machine learning-based CDSSs and diagnostic performance due to low sample size, unclear risk of study bias, lack of consideration of human factors, and lack of studies evaluating these systems under real-world conditions.⁸¹ Thus, although these innovative techniques hold immense promise, future research will need to address specific barriers to progress in this field. For instance, machine learning models depend on “gold standards” in training data with which to compare subsequent previously unseen data. This aspect of machine learning presents a problem in critical care since much of ICU practice is highly subjective, multiple diagnostic pathways are often reasonable, and few clinical decisions are unequivocally “right”.⁷⁹ AI systems also have difficulty drawing inferences from limited data, making them less useful for identifying rare or unusual conditions.⁸² Furthermore, datasets may contain data reflecting healthcare disparities in the provision of systematically worse care for vulnerable groups, which AI systems can then erroneously “learn”.⁸³ Researchers must thus judiciously choose particular aspects of critical care diagnosis that are most suited to AI support and select appropriate training datasets while taking precautions to protect against unintended bias. Aside from overcoming these inherent issues in AI system training and function, investigators must also study how best to incorporate these tools into the ICU, ensuring that clinicians trust in the performance of the system without encouraging decision-making passivity and preventing alert fatigue.⁸⁴

Finally, thoughtful implementation of HIT into complex clinical workflows and environments is just as important as the programmed functions of the HIT application itself. Poorly implemented initiatives using HIT can be more harmful than not using HIT at all.⁸⁵ For successful implementation and maximum benefit, investigators should strive to embed human factors engineering in the user-centered design of HIT interventions⁸⁶ in order to create usable and sustainable systems. A useful framework to consider is Singh and Sittig’s Health IT Safety Framework,⁸⁷ which provides a guide to considering the many sociotechnical dimensions of implementing HIT into complex healthcare systems.

Improving Team Cognition and Teamwork

One of the top recommendations of the NASEM report was to improve teamwork in the diagnostic process.⁸⁸ This is a welcome departure from the classic thinking that physicians are solely responsible for diagnosis. The literature is replete with the known benefits of teamwork in medicine⁸⁹ and specifically in the ICU,⁹⁰ where teamwork has improved both clinical processes and patient outcomes. Despite this awareness, there are few studies investigating how to optimize teamwork to improve critical care diagnosis. We are only beginning to understand the role of nurses,⁹¹ allied medical professionals,⁹² and subspecialty consultants in diagnosis,⁹³ but further research is needed to determine how to maximize their contributions given each member’s specific role in critical care. Early studies have also revealed the importance of developing shared mental models across ICU team members in order to deliver appropriate care.^94,95 Researchers need to build on this work to better understand how different ways of communicating affect mental model creation across an interprofessional group^33,35,36 and leverage known principles and methods in human factors and cognitive psychology to support interventions that will help teams quickly achieve a shared patient understanding. In the ICU, certain team tasks are high yield for improving diagnosis. For example, daily team rounds seem to be an obvious locus of day-to-day ICU team collaboration and decision-making⁹⁶ while critical patient events wherein multiple team members are helping troubleshoot at the bedside present situations where urgent team decisions and actions need to be made.⁹⁷ Researchers may do well to focus on these low-hanging fruit of ICU scenarios ripe for study to improve team diagnosis. Finally, in addition to determining the effects of teamwork interventions on patient outcomes such as mortality, investigators need to shift their attention to diagnosis-relevant outcomes of teamwork, such as accuracy and timeliness of diagnosis and the occurrence of diagnosis-related harm.

Including Patients and Families in the Diagnostic Process

Patient-centered care is defined as care that is respectful of and responsive to individual patients’ and families’ preferences, needs, and values.⁹⁸ Major professional critical care organizations have endorsed this approach for the past decade or more, suggesting that patient and family involvement can profoundly influence clinical decisions and patient outcomes in the ICU.⁹⁹ Although most ICUs have incorporated patient-centered and family-centered care,^99,100 we have yet to explore how shared decision making in diagnosis can be effectively integrated into this model. The concept of patient-centered diagnosis is relatively new, and it is distinguished from shared decision-making for treatment by (1.) greater uncertainty in the diagnostic process, (2.) the patient’s and clinician’s tolerance for uncertainty, (3.) benefits and harms of diagnostic tests that are more difficult to quantify, and (4.) the more iterative nature of diagnostic decision-making.¹⁰¹ Research in this area should focus on how to incorporate shared diagnostic decision-making with families during common activities in the ICU such as family-centered rounds and family conferences. Work is needed to determine optimal ways for clinicians to convey information on the benefits and risks of testing in light of critical illness and eliciting patients’ and families’ preferences and risk tolerance.¹⁰¹

SUMMARY AND CONCLUSIONS

Research to improve diagnosis in critical care medicine has accelerated with increasing awareness of the burden and harms of diagnostic error among critically ill patients. However, much work remains to fully elucidate the diagnostic process in critical care, which is fundamental to understanding how diagnostic errors occur in the ICU. Interdisciplinary research is needed to investigate the many potential interventions to improve diagnostic outcomes and prevent diagnostic-error related harm in this population. In order to make significant progress towards diagnostic excellence, we need to adopt a balanced strategy of continued biomedical discovery while addressing the complex care delivery systems underpinning the diagnosis of critical illness.

Key Points:

• Research to improve diagnosis in critical care medicine has accelerated with increasing awareness of the burden and harms of diagnostic error among critically ill patients.

• Much work remains to fully elucidate the diagnostic process in critical care, which is fundamental to understanding how diagnostic errors occur in the ICU.

• To achieve diagnostic excellence, interdisciplinary research is needed, adopting a balanced strategy of continued biomedical discovery while addressing the complex care delivery systems underpinning the diagnosis of critical illness.

Synopsis:

Diagnosing critically ill patients in the intensive care unit (ICU) is difficult. As a result, diagnostic errors in the ICU are common and have been shown to cause harm. Research to improve diagnosis in critical care medicine has accelerated in past years. However, much work remains to fully elucidate the diagnostic process in critical care. To achieve diagnostic excellence, interdisciplinary research is needed, adopting a balanced strategy of continued biomedical discovery while addressing the complex care delivery systems underpinning the diagnosis of critical illness.

CLINICS CARE POINTS.

• Research to improve diagnosis in critical care medicine has accelerated with increasing awareness of the burden and harms of diagnostic error among critically ill patients.

• Much work remains to fully elucidate the diagnostic process in critical care, which is fundamental to understanding how diagnostic errors occur in the ICU.

• To achieve diagnostic excellence, interdisciplinary research is needed, adopting a balanced strategy of continued biomedical discovery while addressing the complex care delivery systems underpinning the diagnosis of critical illness.