ArticlesDiagnostic errors in three medical eras: a necropsy study
Introduction
The diagnostic process is a complex interaction of cognitive skills and technical procedures in conditions of uncertainty.1 Diagnostic procedures have been improved over the past 30 years. No data, however, show decreases in diagnostic errors in unselected patients, with necropsy as the gold standard for diagnosis. Postmortem examination is the only way to assess overall clinical diagnostic performance, since clinicians can diagnose only diseases for which they have been looking.2 Two studies comparing clinical diagnoses with necropsy findings in different medical eras showed no improvement in diagnostic accuracy over time in unselected patients.3, 4 Diagnostic accuracy depends on sensitivity and specificity of the diagnostic process. The use of diagnostic accuracy as a sole criterion for judging discrepancies between clinical and necropsy diagnoses may obscure what occurs during the diagnostic process.5 To improve clinical practice, it is important to know whether diagnostic errors occur because of lack of sensitivity or specificity of the diagnostic work-up. We assessed the changes in overall accuracy over 20 years, as well as accuracy, sensitivity, and specificity for the three major categories of diseases, to gain more insight into the process of clinical diagnosis.
Section snippets
Selection of cases
We analysed retrospectively the medical and necropsy records of 300 randomly selected adult patients admitted to the Medical Clinic at the University Hospital, Zurich, Switzerland-100 from each of 1972, 1982, and 1992. 1972 was chosen because the oldest complete set of case notes was available from this year. Random sampling was done with a random-number table6 from the list of patients who died in the Medical Clinic and underwent necropsy.
The role of the University Hospital Zurich, and Medical
Results
The characteristics of patients were similar in each year. The number of patients with previous admissions or with two or more admissions was, however, slightly lower among patients in 1982 (table 2). The necropsy rate was 94·0% in 1972 and 89·2% in 1982 and 1992. Cardiovascular diseases were the largest diagnostic group, followed by neoplastic diseases (table 1). Infectious diseases were the third largest diagnostic group of the study patients in 1992, of whom six had HIV-1 infection.
The
Discussion
The frequency of major diagnostic errors have been reduced significantly since the early 1970s. This improvement was due mainly to a decline in the frequency of errors in the diagnosis of cardiovascular diseases
Two previous studies showed an unchanged rate of major discrepancies between clinical and postmortem diagnosis.3, 4 In those two studies, the rate of necropsy dropped from 80% to lower than 40% in the last year of study. This reduction could have masked improved clinical performance
References (19)
- et al.
Cognitive errors in diagnosis: instantiation, classification, and consequences
Am J Med
(1989) - et al.
Prevalence of acute pulmonary embolism among patients in a general hospital and at autopsy
Chest
(1995) Teaching problem-solving-how are we doing?
N Engl J Med
(1995)- et al.
The autopsy: its role in the evaluation of patient care
J Gen Intern Med
(1989) - et al.
The value of the autopsy in three medical eras
N Engl J Med
(1983) - et al.
Misdiagnosis at a university hospital in 4 medical eras: report on 400 cases
Medicine
(1996) Is necropsy a valid monitor of clinical diagnosis performance?
BMJ
(1991)Wissenschaftliche Tabellen Geigy, Teilband Statistik, 8
(1980)- et al.
The emerging role of “hospitalists” in the American health care system
N Engl J Med
(1996)
Cited by (197)
Diagnostic algorithm allows for a scientifically robust and reliable retrospective diagnosis using textual evidence from mid-19th century Basel, Switzerland
2024, International Journal of PaleopathologyAnalysis of the major clinical-pathological discrepancies in autopsies of adults in a tertiary referral hospital between 2008-2017
2021, Revista Espanola de PatologiaDecision analysis and reinforcement learning in surgical decision-making
2020, Surgery (United States)Quality of care and maternal mortality in a tertiary-level hospital in Mozambique: a retrospective study of clinicopathological discrepancies
2020, The Lancet Global HealthCitation Excerpt :Another possible limitation of the study is that, although we met three of the four conditions proposed for complete diagnostic autopsy to monitor clinical diagnosis performance,24 we did not assess the error of the autopsy itself, which would have required a specific study. Finally, a possible limitation could be the disagreement rate of more than 20% in error assignment in this study, although this figure is not dissimilar from previously reported data.25 Most actions and programmes focused on maternal and neonatal mortality reduction rely on imprecise information on the actual causes of maternal mortality.
A Practical Guide to Assess the Reproducibility of Echocardiographic Measurements
2019, Journal of the American Society of Echocardiography