Another Step in the Journey to Combat Diagnostic Errors in Hospitalized Patients

By Nayla Ahmed, MBBS, and Alexander Niven, MD

Dr. Ahmed is with the Division of Pulmonary & Critical Care Medicine, Mayo Clinic, Rochester, MN. Dr. Niven is Consultant, Division of Pulmonary/Critical Care Medicine, Mayo Clinic, Rochester, MN.

SYNOPSIS: Diagnostic errors in hospitalized patients who died or were transferred to the intensive care unit were common in this large retrospective, multicenter trial and frequently associated with patient harm. Problems with clinical assessment and ordering and interpreting tests were the most prevalent diagnostic process faults, with sepsis the diagnosis most associated with diagnostic error.

SOURCE: Auerbach AD, et al. Diagnostic errors in hospitalized adults who died or were transferred to intensive care. JAMA 2024;184:164-173.

It has been more than two decades since the Institute of Medicine report “To Err is Human” first identified medical error as a leading cause of death in the United States.¹ Unfortunately, the journey to zero harm has proven longer and more challenging than anticipated. Despite significant improvements in adverse event rates over the past decade, a study by the National Patient Safety Foundation in 2015 found limited improvement in many problem areas. A National Academies report published the same year concluded that “most people will experience at least one diagnostic error in their lifetime, sometimes with devastating consequences.”^2-4

Auerbach et al performed a retrospective cohort study of adult patients who died or were transferred to the intensive care unit (ICU) in collaboration with 29 academic medical centers participating in the Hospital Medicine ReEngineering Network (HOMERUN). Using a Viziant database, a random sample of 24,591 patients was identified who met these criteria after their second hospital day, selected to reduce the impact of mistriage on admission. The authors excluded cases where ICU transfer was performed for administrative reasons (e.g., medication desensitization), comfort or hospice care only, following an out-of-hospital cardiac arrest, or for whom the medical record or diagnosis codes were unavailable. Two physicians trained in error adjudication systematically reviewed up to 100 medical records from each participating site using a modified version of the Safer-Dx algorithm and the Diagnostic Error Evaluation and Research (DEER) framework. Cases with errors were reviewed for harm using the National Coordinating Council for Medication Error Reporting and Prevention Scale, and disagreements were adjudicated by a third physician until concurrence was achieved.

The investigators reviewed a population of 2,428 patients who were predominantly white, mean age 63.9 years, a slight male majority, and had multiple common medical comorbidities. Most were admitted through the emergency department, and Medicare was the primary payer. Twenty-three percent (550 patients) experienced a diagnostic error, which was associated with temporary or permanent harm or death in 77.1% of this group. Among the 1,863 patients who died, diagnostic error was judged to have contributed to death in 6.6% of cases overall; in patients who died and had a diagnostic error, 29.4% of the cases involved the error contributing to that person’s death. The most prevalent diagnostic process faults were problems with assessment or testing. Of note, there were wide variations in diagnostic error rates between sites, but these primary failure modes persisted in a predetermined secondary subset analysis of patients at sites that fell in the middle 50% of error distribution. Although there were a substantial number of cases where errors were associated with difficulties with healthcare access, incorrect triage at presentation, or failures or delays in seeking care, these factors were not statistically significant in multivariate models when adjusted for patient sociodemographic factors and comorbidities. Approximately half the errors were in patients with the diagnosis of sepsis, followed by myocardial infarction and stroke.

COMMENTARY

The practice of medicine is complex, high-risk, and unpredictable. Although our field is regularly compared to other high reliability systems, including the airline and nuclear power industries, the case vignettes that Auerbach et al offer in support of their findings graphically illustrate the very different and real challenges that we face. While the rates of diagnostic error and their impact reported here are higher compared to prior large studies, they underscore a consistent theme: diagnostic errors remain unacceptably high and are associated with significant patient harm.^5-7 It is worth noting that the inpatient care examined in this study predated the COVID-19 pandemic, following which substantial increases in healthcare-associated infections and patient care events were documented in 2020 due to healthcare system strain.⁸

Individual and system factors contributing to medical error are challenging to distinguish given their intertwined roles within the diagnostic process. While there is no single universal framework for diagnostic reasoning, this complex process typically involves some combination of Type 1 (heuristic or instinctive) and Type 2 (systematic or analytical) reasoning. One commonly cited approach is the “Diagnostic Reasoning Cycle,” which includes an initial assessment, hypothesis generation, diagnostic testing and interpretation, finalizing a diagnosis and treatment plan, and re-evaluation based on new data or complications. Cognitive biases, such as anchoring bias, availability bias, or confirmation bias, can unwittingly contribute to diagnostic errors or delays at any of these stages.^9-11 This study is important because it identifies problems with patient assessment and testing as primary drivers of medical error, things that are under greater direct control of a physician and their healthcare team. Using metacognitive skills to monitor bias, leveraging the power of collaborative decision-making, and ensuring ready availability of educational resources are a few strategies that may help inform decision-making.

While Auerbach et al’s conclusions are appropriately limited to the individual provider due to use of medical record abstraction only in this study, it is important to emphasize other contributing factors to diagnostic error that our daily clinical experiences can elaborate. Healthcare teams currently struggle under a staggering administrative burden, and although electronic health records (EHRs) may improve healthcare quality, their implementation and demonstrated benefit to date have been variable.¹² Removing non-germane cognitive load and improving EHR workflow integration may allow healthcare teams more time at the bedside, where they can reflect and refine initial diagnostic assessments and monitor treatment responses more closely, thereby reducing the frequency of the failure modes seen in this study. As inpatient acuity and complexity increases, effective care delivery demands a distributed cognitive network, a well-coordinated healthcare team working in an “intelligent clinical environment” adapted to accommodate the limits of human cognition. In addition to closing staffing shortfalls created by the COVID-19 pandemic, a more systematic approach that leverages existing telemonitoring services is needed, using artificial intelligence and lean technology to better identify patients at risk for clinical deterioration early, streamline inefficient care delivery processes, and provide teams with feedback when care plans do not comply with best practice guidelines.¹³ While checklists have been well-proven to reduce errors of omission and to improve both care processes and patient-centered outcomes, implementation of this important tool in many acute care settings also remains inconsistent.¹⁴

In summary, this study informs important areas for further focus in our journey to zero harm. It also illustrates that our inpatient setting is becoming an “ICU without walls,” and the interface between our hospitalist and critical care practices represents an important area for ongoing quality improvement. The magnitude of the challenges described in this study also underscores the importance of our ongoing efforts to eliminate diagnostic errors and therapeutic harm, a journey that unfortunately is far from over.

REFERENCES

1. Institute of Medicine Committee on Quality of Health Care in America. Kohn LT, Corrigan LT, Donaldson MS, eds. To Err Is Human: Building a Safer Health System. National Academies Press; 1999.
2. Eldridge N, et al. Trends in adverse event rates in hospitalized patients, 2010-2019. JAMA 2022;328:173-183.
3. National Patient Safety Foundation. Free from Harm: Accelerating Patient Safety Improvement Fifteen Years after “To Err Is Human.” National Patient Safety Foundation; 2015.
4. National Academies of Sciences, Engineering, and Medicine; Institute of Medicine; Board on Health Care Services; Committee on Diagnostic Error in Health Care; Balogh EP, Miller BT, Ball JR, eds. Improving Diagnosis in Healthcare. 2015. http://www.nationalacademies.org/hmd/Reports/2015/Improving-Diagnosis-in-Healthcare.aspx
5. Bates DW, et al. The safety of inpatient health care. N Engl J Med 2023;388:142-153.
6. Gunderson CG, et al. Prevalence of harmful diagnostic errors in hospitalized adults: A systematic review and meta-analysis. BMJ Qual Saf 2020;29:1008-1018.
7. Newman-Toker DE, et al. Burden of serious harms from diagnostic error in the USA. BMJ Qual Saf 2024;33:109-120.
8. Fleisher LA, et al. Health care safety during the pandemic and beyond — building a system that ensures resilience. N Engl J Med 2022;386:609-611.
9. Furlan L, et al. Choosing wisely in clinical practice: Embracing critical thinking, striving for safer care. J Intern Med 2022;291:397-407.
10. Croskerry P. A universal model of diagnostic reasoning. Acad Med 2009;84:1022-1028.
11. Committee on Diagnostic Error in Health Care; Board on Health Care Services; Institute of Medicine; The National Academies of Sciences, Engineering, and Medicine; Balogh EP, Miller BT, Ball JR, eds. The diagnostic process. In: Improving Diagnosis in Health Care. National Academies Press; 2015. https://www.ncbi.nlm.nih.gov/books/NBK338593
12. Campanella P, et al. The impact of electronic health records on healthcare quality: A systematic review and meta-analysis. Eur J Public Health 2016;26:60-64.
13. Subramanian S, et al. Tele-critical care: An update from the Society of Critical Care Medicine Tele-ICU Committee. Crit Care Med 2020;48:553-561.
14. Vukoja M, et al; Checklist for Early Recognition and Treatment of Acute Illness and Injury (CERTAIN) Investigators of the SCCM Discovery Network. Checklist for early recognition and treatment of acute illness and injury: An exploratory multicenter international quality-improvement study in the ICUs with variable resources. Crit Care Med 2021;49:e598-e612.