Discrepancies between clinical and autopsy diagnoses: a comparison of university, community, and private autopsy practices

Although it is known that autopsies often disclose unexpected findings, few studies have been published that address the effect of institutional setting, selection bias, and length of hospitalization. Records of medical autopsies from 3 institutional settings were studied for discrepancies between clinical and autopsy findings. The settings were university hospital (n = 85); community hospital (n = 146); and private autopsy (n = 60), which were referred from various community hospitals and paid for by family members. The same prosector performed the autopsies in the community and private settings. The overall rate of major discrepancy that involved the cause of death was 17.2%. Factors that increased the likelihood of missed diagnoses were private setting (P = .0005), community setting (P = .02), and short hospital stay before death (P = .02). Additional major findings were present in 28.5% of autopsies. Length of hospital stay before death, institution, and selection bias all affect the rate of major unexpected findings in hospital-based autopsies.     

The autopsy. The ultimate audit

Four hundred twenty-eight autopsy cases were reviewed at a departmental conference over a 4 1/2-year period. All diagnostic test data were discussed before the results of the autopsies were known, and the participants tried to extract the maximum diagnostic information from the clinical findings and the test data. Fifty-two percent of the autopsy cases failed to provide any additional major insight; premortem clinical and laboratory findings were sufficient for diagnosis. In the remaining 48%, however, the autopsies contributed to the improvement of medical care.     

Factors affecting autopsy rates, autopsy request rates, and autopsy findings at a large academic medical center.

Autopsy rates continue to decline in the United States. To assess the impact of various objective factors (time of death, day of death, age at death, patient gender, clinical service, and length of hospital stay) on the autopsy request rate, autopsy rate, successful request rate, and percentage of cases in which the autopsy examination added to or altered the clinical assessment of the patient, we prospectively studied all hospital deaths at a major academic medical center for the 3-year period from 1996 through 1998. The autopsy rate decreases significantly with patient age, both because of a decreasing request rate and because family members are less likely to grant permission. An autopsy is less likely to be requested for deaths in the emergency department or on general surgery services and most likely to be requested for fetal, medicine, cardiothoracic surgery, and pediatric deaths. Families more commonly grant permission for autopsy on fetal deaths, pediatric deaths, and emergency department deaths. Forty percent of autopsies reveal significant information about the patient's death beyond what was known premortem. This is least frequent among the fetal deaths, but relatively constant for adults of all ages. Patients who die in the emergency department are most likely to have significant unexpected findings at autopsy. Increasing the request rate for adult patients who die in the emergency department and on the medicine services will result in the greatest increase in information learned from autopsy.     

Concordance between premortem and postmortem diagnosis in the autopsy: results of a 10-year study in a tertiary care center

We describe the concordance between clinical diagnoses and autopsy findings in a tertiary care center. Clinical autopsy is a useful tool for the evaluation of accuracy of the clinical diagnoses. However, autopsy rates have declined around the world recently. We randomly evaluated 500 of 2,211 autopsies performed in the department of pathology of a tertiary care hospital during a 10-year period. We computed the sensitivity, specificity, predictive values, and concordance scores between premortem and postmortem diagnoses. The autopsy diagnoses were used as the "gold standard." Four-hundred twenty-two (84.4%) of the autopsies met inclusion criteria. Diseases of the respiratory tract were diagnosed in 44.1% (186) of all autopsy reports reviewed. The higher sensitivity for diagnosis was observed in congenital anomalies (87.5%), while the higher specificity was observed in diagnosis of complications of pregnancy, childbirth, and the puerperium (98.98%). The higher concordance between premortem and postmortem diagnosis was observed with the diagnoses of neoplasms (kappa = 0.76), and for the group of complications of pregnancy, childbirth, and the puerperium (kappa = 0.76). A clinical diagnosis successfully addressed the cause of death in 40% of the cases Low values for concordance between autopsy reports and clinical diagnoses were present in most of the autopsies reports reviewed. We encourage physicians to continue considering the autopsy as an important tool that extends our understanding of diseases.     

Medical autopsy: whose gain is it? An audit

Medical audit is essential in assessing the efficacy of health care delivery system. Though autopsy services are generally looked upon indifferently and with sceptism by the clinicians, it can form an important part of the medical audit system. The aims of this study were to audit autopsies of deaths within 24 hours of hospital admission by: 1) Comparing premortem and postmortem diagnosis; 2) Comparing postmortem gross diagnosis with postmortem histopathologic diagnosis; 3) Whether deaths could be certified based on clinical judgement and autopsies avoided. The study sample was 99 autopsies. In 45% autopsies, clinical impression did not match the final cause of death. In 14.2% autopsies, final cause of death could have been given by the clinician based on his clinical judgement. In 54.5% autopsies, there was agreement between premortem and postmortem diagnosis. In 67.6% autopsies, gross findings matched with the histopathologic findings.     

Unexpected autopsy findings arising from postmortem ocular examination

BACKGROUND: A full autopsy at our institution includes removal of the eyes for pathologic examination. To our knowledge, the rate of ophthalmic findings at autopsy has not been documented previously. DESIGN: We retrospectively reviewed 277 consecutive autopsies conducted between 1995 and 1999 in which the eyes were removed for examination. Ophthalmic findings were placed in the following categories: I, major findings included those that contributed to the patient's death, would have changed patient management, and/or may have important medical implications for close relatives; II, expected findings after ophthalmologic surgery and minor findings that may have eventually required treatment; and III, incidental findings. RESULTS: Major findings (category I) were found in 32% of autopsies. Minor findings (category II) and incidental findings (category III) were documented in 62% and 34% of autopsies, respectively. Only 14% of autopsies revealed no ophthalmologic diagnoses. CONCLUSION: In our series, postmortem ocular examination revealed a number of important findings, including several heritable and rare conditions. Eighty-six percent of autopsies disclosed at least one pathologic ophthalmologic finding, approximately one third of which demonstrated findings significant enough to have likely required management. Diabetic retinopathy was the most frequent major finding. Malignant melanoma of the choroid was the most commonly found intraocular neoplasm. Chronic uveitis was the most common minor finding. We conclude that important, often unexpected ophthalmic findings are frequently encountered at autopsy, underscoring the relevance of routine postmortem examination of the eyes.     

The autopsy as a performance measurement tool--diagnostic discrepancies and unresolved clinical questions: a College of American Pathologists Q-Probes study of 2479 autopsies from 248 institutions

OBJECTIVE: To develop a multi-institutional reference database for quality improvement purposes using the autopsy to define clinical diagnostic discrepancies and resolve clinical questions. DESIGN: Using the College of American Pathologists Q-Probes format, institutions prospectively assessed a maximum of 15 consecutive autopsies each, excluding forensic cases and stillborn infants, conducted over a 6-month period. They documented answers to clinical questions provided at autopsy and classified unexpected disease diagnoses according to a standardized system. SETTING AND PARTICIPANTS: Hospital-based autopsies performed at 248 institutions participating in the 1993 College of American Pathologists Q-Probes Quality Improvement Program. MAIN OUTCOME MEASURES: Percentages of clinical questions resolved by the autopsy and percentage of autopsies with unexpected findings of graded clinical impact. RESULTS: In the aggregate database of 6427 questions from 2479 autopsies, overall 93.0% were answered by the autopsy. The 3 most common question categories were (1) identify pathology to account for clinical signs or symptoms (28.0%); (2) establish the cause of death (21.0%); and (3) confirm a clinical diagnosis (19.0%). At least one major unexpected disease finding that contributed to the patient's death was discovered in 39.7% of the total number of autopsies. There were no differences in the percentages of autopsies with these major unexpected findings when the data were stratified by institutional demographics or decedent characteristics. CONCLUSION: This multi-institutional study underscores the clinical relevance of postmortem examination in current medical practice by consistently providing answers to unresolved clinical questions and frequently revealing major unexpected findings that contributed to the patient's death. It is our strong belief that this postmortem-derived clinicopathologic information is a key indicator of effectiveness of care. Integration of this information into institutional quality improvement programs will improve system processes and clinician performance.     

Diagnostic yield from 231 autopsies in a community hospital

Autopsy reports of 231 adult autopsies were examined for clinically unexpected diagnoses. Analysis of data showed that 97 of 188 autopsy diagnoses were clinically unexpected. The diagnosis of pulmonary embolism was clinically the most frequently missed diagnosis relative to its actual occurrence. Also noted was that the proportion of cases with unexpected diagnoses increased from 1983 to 1987.     

The autopsy in a tertiary teaching hospital in Brazil

A high autopsy rate allows accurate epidemiological studies and quality control of medical care. This study aims to analyze all autopsies performed in a university teaching hospital in Brazil during 52 consecutive wk. The following data were retrieved from individual autopsy records: gender, age, time of death (hr, day, and month), and the main cause of death. There were 1419 autopsies (79% adults and 21% pediatrics, 60% male and 40% female). Those performed during working days summed up to 67.5%, the remaining 32.5% were performed during weekends or holidays. Autopsies were more frequent during the nocturnal than diurnal period (52.6% vs 47.4%, respectively). The causes of death distributed among the ICD-10 categories were: cardiovascular diseases 21.3%, infectious diseases 19.2%, neoplasms 12.8%, perinatal conditions 10.8%, respiratory diseases 6.6%, gastrointestinal diseases 6.0%, congenital anomalies 4.7%, CNS diseases 3.8%, genitourinary diseases 1.8%, and others 13.0%. There was coexistence of the diseases typical of both industrialized and developing countries, indicating the epidemiological transition in our country. Our data indicate that staff supervision of undergraduate medical students and residents is adequate. Along with a brief discussion of the historical, cultural, and legal factors that allow a high autopsy rate, this study reinforces the importance of the autopsy in a tertiary teaching hospital.     

Declining autopsy rate in a French hospital: physician's attitudes to the autopsy and use of autopsy material in research publications

CONTEXT: Autopsy rates have been declining throughout the world, although preservation of the autopsy is considered a fundamental principle of medical care. In France, the 1994 bioethics law requires physicians to inform relatives before performing an autopsy. OBJECTIVE: To analyze the following factors that potentially influence hospital autopsy rates: legal constraints, autopsy reporting times, opinions of physicians requesting autopsies and pathologists regarding the usefulness of autopsy in patient care, and use of autopsy material in research publications. DESIGN: Record of the annual numbers of deaths and autopsies during a 10-year period (1988-1997). Record of the delays for transmission of final autopsy report to the requesting physician. Questionnaire analyzing the possible factors influencing autopsy rate. Categorization of articles published by pathologists according to the use of autopsy material. SETTING: A 1000-bed, university teaching hospital in the Paris, France, area. PARTICIPANTS: Questionnaire addressed to physicians, head nurses, and mortuary staff. RESULTS: A total of 1454 autopsies were reviewed. The autopsy rate declined from 15.4% in 1988 to 3.7% in 1997. This decline was marked after 1994 and tended to be slower for neurologic indications than for other indications. The final report had not been communicated within 180 days in 620 (42.6%) of 1454 autopsies. Fifty-five of 105 respondents considered that the bioethics law was one cause of the recent decrease of autopsy rate. Considering the contribution of autopsy to medical research, 94 (81%) of 116 articles dealing with central nervous system but only 28 (6%) of 464 articles dealing with other organs used autopsy-derived material. CONCLUSIONS: The 1994 bioethics law seems to contribute to the decline of autopsy. Inadequate delays for communicating autopsy results are frequent. Except for neuropathologists, autopsy is a minor source of research material.     

Clinical utility of autopsy after hematopoietic stem cell transplantation.

Autopsy is the gold standard for establishing the cause of death. We present results of the largest retrospective review of complete autopsies of subjects after hematopoietic stem cell transplantation to better define the role of the autopsy in discovering a missed diagnosis. We reviewed the medical chart and autopsy records of 111 patients who had undergone hematopoietic stem cell transplantation from July 1986 to June 2003 from a single center. We compared the cause of death as charted by the clinical team with data obtained from postmortem chart review and autopsy reports. Of 29 (26%) cases when the premortem and postmortem major diagnoses did not agree, only 4 (4%) autopsy records provided data that might have led to the initiation of new treatments, and none of these diagnoses would be missed today with more sensitive and specific diagnostics and improved supportive care. Although autopsies after transplantation can be important educational, research, and epidemiologic tools and provide an emotional benefit to patient's families, in our series they rarely provided missed diagnoses that would alter the management of subsequent patients. Improvements in noninvasive tests for relapse or occult infections may further erode the role of autopsies in discovering missed diagnoses.     

A retrospective study of the accuracy between clinical and autopsy cause of death in the University of Malaya Medical Centre

It is well known that diagnostic accuracy of the clinical cause of death has not improved despite advances in diagnostic techniques. We aimed to investigate the accuracy of the clinical cause of death compared with the autopsy cause of death and to see if the Coroner's autopsy can play a role in clinical audit. Our study population consisted of all autopsies where the deceased was hospitalised or resuscitated at the Accident and Emergency Unit of the University of Malaya Medical Centre before death, performed during the period July 1998 to June 2000. The cases were subdivided according to natural and unnatural causes of deaths. Natural deaths were further subdivided in relation to the main organ systems involved while unnatural deaths were subcategorised into trauma, poisoning and burns. The rate of agreement between clinical and autopsy cause of death was further compared with duration of survival in the hospital. Of 132 autopsies included in this study, 115 were Coroner's autopsies. 78% of cases showed agreement between clinical and autopsy cause of death. The agreement rate in Coroner's cases was 80.0%. For natural and unnatural causes, the agreement rate was 56.7% and 84.3% respectively. There were 6 cases (4.5%) where an initial accurate diagnosis might have altered the prognosis of the deceased. In general, the rate of agreement increased with duration of survival of patients. However, this was no longer observed after a survival of more than 28 days. Our findings agree with other similar studies. The diagnostic accuracy of cause of death has not improved despite the modernisation in medical technology. The autopsy still plays an important role in clinical audit and medical education.     

Autopsy and the internal medicine department.

The accuracy of major clinical diagnoses and causes of death was estimated from the analysis of clinical diagnoses and autopsy findings of patients deceased at the internal medicine department of a large university hospital in two different periods (1965-1967 and 1985-1987). The autopsy cases from the second period were subdivided in two groups i.e. less and more than seven days of hospital stay prior death. From 100 randomly selected autopsy cases in each period there was no statistically significant variability in clinico-pathological concordance. In three groups studied (1965-1967, 1985-1987 < seven days and 1985-1987 < seven days hospital stay) the major clinical diagnosis was confirmed in 85, 94 and 86% respectively. A decrease of nearly 27% in the number of autopsies performed is shown between analysed periods. Clinico-pathological conferences are regularly held weekly, this being one of the reasons for a stable and extremely high clinico-pathological concordance.     

Utility of autopsy in uncovering unexpected neuropathology

Autopsy rates have declined in the last several decades for a variety of reasons. The purpose of this study is to compare autopsy neuropathologic findings from 2 periods to assess the prevalence of unexpected neuropathologic findings and unexpected neuropathologic diagnoses determined to be the major cause of death. Retrospective review of autopsies with examination of the central nervous system was performed in 2007 to 2008 (n = 289) and 1984 to 1985 (n = 328). Unexpected neuropathologic diagnoses were found at autopsy in 42.4% of cases from 1984 to 1985 vs 38.8% of cases from 2007 to 2008. The neuropathology was felt to contribute to the cause of death in 22% of cases from 1984 to 1985 vs 19.7% of cases from 2007 to 2008. Unexpected neuropathologic findings were the cause of death in 5.2% of cases from 1984 to 1985 vs 3.1% of cases from 2007 to 2008. These findings underscore the continued use of brain and spinal cord examination at autopsy despite advances in "modern" medicine.     

Utility of Autopsy among Pediatric Allogeneic Hematopoietic Stem Cell Transplant Recipients: One Last Chance to Learn?

Autopsy may confirm clinical diagnoses or identify conditions that were not suspected prior to a patient's death. Previous studies evaluating the utility of autopsy in hematopoietic stem cell transplant (HSCT) recipients yielded conflicting results.We conducted a retrospective cohort study of children (<18 years of age) undergoing allogeneic HSCT at Duke University who died of any cause between January 1, 1995, and December 31, 2016. We evaluated associations between patient characteristics and autopsy performance using chi-square or Fisher exact tests. We reviewed autopsy reports to determine the concordance between preautopsy causes of death and pathological diagnoses identified on autopsy. We classified unexpected diagnoses on autopsy using criteria developed by Goldman et al. We evaluated for temporal changes in the autopsy consent rate and the frequency of unexpected diagnoses on autopsy using Cochran-Armitage tests.During the 22-year study period, 475 patients died and had data available on autopsy performance, and 130 (27%) of these patients underwent autopsy. The autopsy consent rate declined over time (P < .0001), with autopsies being performed for 40% of deaths in 1995 to 1999 and 17% of deaths in 2009 to 2016. White patients were more likely to undergo autopsy than nonwhite patients (P?=?.03). There were no associations between autopsy performance and patient age, sex, HSCT indication, or HSCT donor. Unexpected diagnoses were identified in 31 (24%) autopsies. The proportion of autopsies with an unexpected diagnosis did not change during the study period (P?=?.45). However, infectious diagnoses that would have led to a change in management were more frequently identified on autopsies in 1995 to 2003 than in 2004 to 2016 (20% versus 0%; P?=?.001).The autopsy consent rate for pediatric HSCT recipients at our institution has declined substantially over the past several decades. The utility of autopsy in this patient population remains high despite a reduction in the identification of unexpected infections.     

Malignant neoplasms: discordance between clinical diagnoses and autopsy findings in 3,118 cases

BACKGROUND: During the past few decades, hospital autopsy rates have steadily declined throughout the Western world. This decline is mainly attributed to the introduction of advanced diagnostic techniques. Despite technological developments, discrepancy rates between clinical diagnoses and autopsy findings remain high. Few studies have addressed discrepancy rates exclusively with regard to malignant neoplasms. In the present study, we reviewed the records of 3,118 autopsies performed at Mayo Clinic during a 6-year period (1994-1999) and identified clinically undiagnosed malignancies found at autopsy and clinically diagnosed cancers not confirmed at postmortem examination. MATERIALS AND METHODS: Autopsy protocols, provisional and final anatomic diagnoses, and data from the Mayo Autopsy Pathology Quality Assurance program were reviewed in an attempt to identify discrepancies between clinical diagnoses and autopsy findings regarding malignant neoplasms. RESULTS: In 3,118 autopsies performed at Mayo Clinic between 1994 and 1999, a malignant tumor was identified in 768 cases (25%). In 128 of 3,118 cases (4.1%), the malignancy was not diagnosed clinically. In 14 of 3,118 cases (0.45%), autopsy failed to confirm a clinically diagnosed cancer. A review of the literature is presented. CONCLUSIONS: Autopsy remains an effective tool for the confirmation and refutation of clinical diagnostic findings regarding malignant neoplasms.     

Postmortem findings and opportunistic infections in HIV-positive patients from a public hospital in Peru

There is a paucity of HIV autopsy data from South America and none that document the postmortem findings in patients with HIV/AIDS in Peru. The purpose of this autopsy study was to determine the spectrum of opportunistic infections and the causes of mortality in HIV-positive patients at a public hospital in Lima. Clinico-epidemiological information regarding HIV infection in Peru is also reviewed. Sixteen HIV-related hospital postmortems, performed between 1999 and 2004, were included in this retrospective analysis. The primary cause of death was established in 12 patients: one died of neoplasia and 11 of infectious diseases, including 3 from pulmonary infection, 7 from disseminated infection, and 2 from central nervous system infection (one case had dual pathology). Opportunistic infections were identified in 14 cases, comprising cytomegalovirus, histoplasmosis, cryptococcosis, toxoplasmosis, Pneumocystis pneumonia, aspergillosis, tuberculosis, varicella zoster virus, and cryptosporidiosis. Fourteen patients had at least one AIDS-related disease that had been neither clinically suspected nor diagnosed premortem. Moreover, 82% of the diagnoses considered to be of important clinical significance had not been suspected antemortem. The spectrum and frequency of certain opportunistic infections differed from other South American autopsy studies, highlighting the importance of performing HIV/AIDS postmortems in resource-limited countries where locally specific disease patterns may be observed.     

Autopsy result utilization: a College of American Pathologists Q-probes study of 256 laboratories.

OBJECTIVES: To document the level of involvement and communication with nonpathology clinical personnel regarding autopsies and to document the destination of autopsy reports. DESIGN: The College of American Pathologists Q-Probes format was used to collect information on 15 consecutively performed autopsies per institution or for 6 months, whichever occurred first. The following information was recorded for each autopsy: decedent's age, hospital service, length of hospital stay, whether organs were donated, who was present at autopsy, methods of communicating preliminary and final autopsy results, special techniques used to arrive at a preliminary diagnosis, activities for which the autopsy was used, and destination of final report. PARTICIPANTS: Two hundred fifty-six laboratories collected information on 2755 autopsies. RESULTS: The aggregate autopsy rate was 12.4% (median 8.5%). Nonpathology clinical personnel attended 35.8% of all autopsies. A clinical physician was more likely to attend an autopsy if the patient was from a surgical service. Three primary methods were used to communicate preliminary autopsy results, namely, written reports (82.5%), telephone calls (50.6%), and meetings (11.5%). The primary care physician was sent the autopsy report in 91.1% of cases. Approximately half of the autopsy cases were used in both pathology departmental and extradepartmental activities. Aggregate autopsy data were distributed in the majority of cases to various departmental chairpersons and institutional quality assurance committees. CONCLUSIONS: This study provides a comparative multiinstitutional database for the utilization of autopsy results by clinicians and clinical departments. Although autopsy rates are low, autopsy results are routinely being used for hospital quality assurance activities and for educational purposes.     

A prospective study of 1152 hospital autopsies: II. Analysis of inaccuracies in clinical diagnoses and their significance

A further analysis of 1152 hospital autopsies provides data on inaccuracies of specific diagnoses; there are many examples of overdiagnosis and underdiagnosis. All were encountered in a routine hospital autopsy service and their frequency confirms the importance of the hospital autopsy in medical audit. A knowledge of the misdiagnoses which recur frequently could provide guidance in the selection of cases for autopsy.     

Opportunistic infections in adults with acquired immunodeficiency syndrome: a comparison of clinical and autopsy findings.

BACKGROUND AND PURPOSE: Many opportunistic infections causing death in acquired immunodeficiency syndrome (AIDS) patients are often not diagnosed prior to death. The objective of this study was to compare the premortem and postmortem diagnoses of opportunistic infections and tumors among 15 AIDS patients treated in a hospital in southern Taiwan. METHODS: Total autopsy (brain, chest and abdominal cavity) was performed in 2 patients, and partial autopsy in 13. RESULTS: Pneumocystis carinii pneumonia, candidiasis, lymphoma, Kaposi's sarcoma, toxoplasmosis and salmonellosis were more commonly diagnosed before death than at autopsy. By contrast, cytomegalovirus (CMV) infections and herpes simplex virus or varicella-zoster virus infections were more frequently diagnosed at postmortem examinations than prior to death. CONCLUSIONS: In conclusion, this study found substantial discrepancies between autopsy findings and premortem clinical diagnoses in AIDS patients, especially for CMV infection.