- Email: [email protected]
Discrepancies in initial death certificate diagnoses in sudden unexpected out-of-hospital deaths: the role of cardiovascular autopsy
Cardiovascular Pathology 17 (2008) 178 – 182
Original Article
Discrepancies in initial death certificate diagnoses in sudden unexpected out-of-hospital deaths: the role of cardiovascular autopsy Fabio Tavora a,b,⁎, Clinton Crowder a , Robert Kutys c , Allen Burke a,c a
Department of Pathology, University of Maryland, 22 South Greene Street, Room NBW43, Baltimore, MD 21201, USA b Department of Pathology, The Johns Hopkins University, School of Medicine, Baltimore, MD 21231, USA c CVPath Institute, Inc., 19 Firstfield Road, Gaithersburg, MD 20878, USA Received 31 May 2007; received in revised form 10 July 2007; accepted 24 July 2007
Abstract Background: The accuracy of death certificates issued for out-of-hospital sudden deaths has been questioned. Methods: We retrospectively studied a series of consecutive autopsies performed at two community hospitals. Results: Fifty-four autopsies in which the deaths were sudden and that occurred outside the hospital were retained for study. The indication for autopsy was largely driven by the wishes of family or physician who was uncertain about the diagnosis. The overall discrepancy rate was 52%. The death certificate diagnosis, rendered before autopsy, was coronary artery disease in 44/54 autopsies (81%). At autopsy, coronary artery disease was the cause of death in 26 cases (48%), cardiomyopathy in 10 (18%), ruptured aneurysm in 8 (15%), pulmonary embolism in 7 (13%), and valve disease in 3 (6%). The diagnosis of coronary artery disease on the death certificate was accurate only 50% of the time. The discrepancy rate was lowest in patients with a history of cardiovascular disease (33%) and was 60% in patients with no prior medical history. The accuracy rate of death certificates was under one third in cases of cardiomyopathy, valve disease, ruptured aneurysm, pulmonary embolism, and valve disease. Conclusions: We conclude that in a highly selected group of sudden deaths, in which there was often a question about cause of death, the rate of initial death certificate accuracy is only one half. Furthermore, coronary artery disease as the cause of death is less than 50%, far less than initial death certificate diagnoses would indicate. © 2008 Elsevier Inc. All rights reserved. Keywords: Autopsy discrepancy; Coronary disease; Ruptured aneurysm; Quality control
1. Introduction Studies assessing the validity of death certificates issued on out-of-hospital sudden deaths have shown a high rate of discrepancy between clinical and autopsy findings [1,2]. There are various methods of assessing death certificate accuracy, but none is straightforward or easily verifiable [3]. Methods of assessing death certificate accuracy include retrospective review of medical records and death certificates by experts [4]; retrospective review of random fraction of population-based deaths, including death certificates and coroners' files [1,5]; comparison of causes of death before and after autopsies were performed after filling out the initial
⁎ Corresponding author. Tel.: +1 410 3285555; fax: +1 410 3255508. E-mail address: [email protected] (F. Tavora). 1054-8807/08/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.carpath.2007.07.010
death certificates [6]; and retrospective review of all available information in population-based studies of sudden unexpected death [7]. Some studies [1], but not others [8], demonstrate a high rate of discrepancy when assessing validity of death certificate diagnoses, specifically in cases of sudden death. This rate of discrepancy is especially high if the death occurs within 1 h of onset of symptoms before medical diagnostic procedures can take place [5]. In the state of Maryland, out-of-hospital sudden deaths are handled by emergency services, and death certificates are signed by the deceased's physician, based on prior knowledge of the patient's medical condition. If there is no available physician, the death is referred to the regional medical examiner, an employee of the statewide medical examiner's system. If there are no suspicious circumstances, the body is released to the funeral home, and a death certificate is signed by the medical examiner based on
F. Tavora et al. / Cardiovascular Pathology 17 (2008) 178–182 Table 1 Case characteristics, sudden death patients
Autopsy setting Community hospital Private Gender Male Female Medical history No prior medical history Prior medical history, excluding symptomatic heart disease a History of symptomatic heart disease b Total
n
Age (in years), mean±S.D.
13 41
65±15 65±11
27 27
61±9 69±13 ⁎
15 24
66±12 65±12
15 54
67±13 66±12
a
There were 10 patients with multiple coronary risk factors, including 3 with diabetes; 7 with neoplasia; 3 with prior negative cardiac evaluation, including negative stress echocardiogram; 2 with pulmonary disease (asthma, pulmonary nodule); 1 with prior genitourinary surgery; and 1 with neurologic disorder. b There were 7 patients with known coronary artery disease, 2 with prior cerebrovascular accident, 1 with known amyloid, 1 with known aortic stenosis, 1 with a history of pulmonary embolism and giant cell arteritis, 1 with known abdominal and thoracic aortic aneurysm, 1 with known dilated cardiomyopathy, and 1 with a history of aortic valve replacement. ⁎ P=.01 versus men.
interviews with the family and investigation of the scene. The death certificate allows for an amended diagnosis, made after autopsy, if performed. In out-of-hospital deaths that are turned down by the medical examiner or signed by a physician who had contact with the patient, the family may request an autopsy based on a desire for more specific cause of death, concerns about hereditary disease, litigious concerns, or documentation of death certificate accuracy for insurance purposes. In many cases, the autopsies are referred to private (paid) services because community hospitals are generally unwilling to provide autopsy services unless the deceased was a recent inpatient. The purpose of this study was to report a series of out-ofhospital sudden deaths, in which autopsy was performed based on family request and in which a cardiovascular event was either diagnosed or considered. The autopsy diagnosis was compared to the death certificate diagnosis, and concordance was assessed based on final autopsy diagnosis, degree of clinical evaluation, and age and gender of the patient. 2. Materials and methods Out-of-hospital deaths were referred to two settings for autopsies, occurring between 2001 and 2006: a community hospital and a private autopsy service, which performed all autopsies in a second community hospital, prior to embalming. The cases referred to the community hospital were performed because of the deceased's physician's attending privileges at the institution; in these cases, the deceased's
179
physician signed the death certificate in every instance. The private autopsy service performed autopsies at the family's request, and the death certificate was signed either by a family physician or by the medical examiner, after nonnatural causes of death were excluded by way of investigation of the death scene and interview with family members. Unlike community hospitals, private autopsies required payment by the deceased's family. Cases were included for study if the death occurred outside the hospital or if death was confirmed in the emergency room of a hospital, before the patient was admitted. The time frame encompassed the period 2001– 2006. The cases comprised a subset of autopsies that included inpatient deaths at both institutions. Sudden death was defined as death within 1 h of onset of symptoms. Patients with prior heart disease were included if they were asymptomatic prior to the sudden death. The same prosector with subspecialty expertise in cardiovascular pathology performed all autopsies. In all cases, a minimum of 20 histologic sections was performed, including sections of all viscera. Examination of the brain was performed if there were any neurologic symptoms or if examination of the cardiovascular system did not demonstrate a clear-cut cardiovascular cause. History was available in the community hospital based on medical records of prior admissions and telephonic conversations with the family physician. History was available in the private autopsies based on telephone conversations with the next of kin, who requested and paid for the autopsies. The historical information was gathered at the time of autopsy. Correlation between autopsy findings, death certificate findings, and clinical information was done retrospectively at the time of study, based on archived materials. Discrepancy was defined as the absence of the anatomic cause of death anywhere in the death certificate or if the primary cause of death as recorded on the death certificate was not found at autopsy. The autopsy causes of death were classified into five groups: coronary artery disease, cardiomyopathy, pulmonary embolism, valve disease, and ruptured aneurysm.
3. Results There were 54 autopsies fulfilling the inclusion criteria, including 13 performed at the community hospital and 41 referred for private autopsy. These formed a subset of 158 community adult autopsies and 82 private autopsies, representing 8% and 50%, respectively, of the entire series of autopsies. There were 27 men and 27 women. The mean age of the women was significantly higher than that of the men (Table 1). Fifteen patients had no prior medical history, and 15 patients had a history of symptomatic cardiovascular disease. The remaining 24 patients had prior medical care but no history of symptomatic cardiovascular disease.
180
F. Tavora et al. / Cardiovascular Pathology 17 (2008) 178–182
Table 2 Case distribution by cause of death, age, and gender Cause of death, autopsy
n (%)
Men:Women
Age (in years), mean±S.D.
Coronary artery disease Cardiomyopathy Aneurysm Pulmonary embolism Valve disease Total
26 (48) 10 (18) 8 (15) 7 (13) 3 (6) 54
12:14 6:4 5:3 2:5 2:1 27:27
67±11 56±9 ⁎ 72±9 67±17 58±12 66±12
⁎ P=.005 versus coronary artery disease; P=.003 versus aneurysm.
The most common autopsy cause of death was coronary artery disease in 48%, followed in decreasing incidence by cardiomyopathy, aneurysm, pulmonary embolism, and valve disease (Table 2). There were no significant gender differences, although cardiomyopathy patients were the youngest, significantly younger than coronary and aneurysm patients (Table 2). The majority of patients suffered from presumed instantaneous sudden death. Four of the 54 patients complained of chest pain, shortness of breath, or abdominal discomfort moments before death; 22 patients were found unresponsive but were in their normal state of health while last seen, and 28 were witnessed arrests without apparent symptoms. The majority of coronary deaths had no evidence of prior infarction, and 15 of 26 deaths demonstrated epicardial thrombosis (Table 3). Of 26 hearts in the coronary artery disease groups, 11 demonstrated moderate to severe cardiomegaly, with heart weights N500 g (Table 3). Of the
Table 3 Cardiac findings and gender distribution, coronary artery disease deaths Finding
n
Heart weight Male:Female N500 g
No prior infarct 18 5:13 Acute plaque rupture 7 3:4 Acute plaque rupture with acute infarct 2 1:1 Acute plaque rupture with ruptured 2 0:2 acute infarct Severe coronary atherosclerosis, 4 1:3 no thrombus Severe coronary atherosclerosis, 1 0:1 status post-CABG Nodular calcification with rupture 1 0:1 and thrombus Acute infarct, without thrombus 1 0:1 Prior infarct (healed myocardial infarct) 8 7:1 Plaque rupture, without acute infarct 2 2:0 Plaque rupture with acute infarct 1 1:0 Severe coronary atherosclerosis, 2 2:0 status post-CABG Severe coronary atherosclerosis, 1 1:0 status post-stenting Severe coronary atherosclerosis, 2 1:1 no thrombus CABG, coronary artery bypass graft surgery.
7/18 4/7 0/2 0/2 1/4
Fig. 1. Hypertrophic cardiomyopathy. (A) Gross photograph demonstrates septal hypertrophy and left ventricular outflow tract plaque (arrow). (B) Microscopic section demonstrates myofiber disarray.
0/1 1/1 1/1 4/8 1/2 1/1 1/2 1/1 1/2
15 patients in the entire study with a history of heart disease, only 7 were in the coronary artery disease autopsy group, including 3 with prior coronary artery bypass graft surgery and 1 with prior coronary stenting. Three patients had recent negative cardiac evaluations, including stress testing; interestingly, the initial death certificates indicated atherosclerotic heart disease as the cause of death. Of the noncoronary sudden cardiac deaths, there were 10 cardiomyopathies, 8 ruptured aneurysms, and 3 valve diseases as the cause of sudden cardiac death. The cardiomyopathy group included one patient with amyloidosis, two patients with arrhythmogenic right ventricular
F. Tavora et al. / Cardiovascular Pathology 17 (2008) 178–182
cardiomyopathy (ARVD), two patients with dilated cardiomyopathy, and three patients with hypertrophic cardiomyopathy (HCM). Three of the cardiomyopathy sudden deaths occurred during exercise. These included HCM (Fig. 1) and ARVD. The aneurysms included four dissections, two abdominal aortic aneurysm ruptures, one thoracic aneurysm rupture that had been followed medically, and one sinus of Valsalva aneurysm rupture. One patient with aortic stenosis had a prior diagnosis and was found dead the day before the scheduled aortic valve replacement. The other patient had no prior history of heart disease and was found to have subaortic stenosis with left ventricular hypertrophy at autopsy (Fig. 2). Of the seven patients with diagnosed pulmonary embolism, five had underlying risk factors, two of which (pancreatic adenocarcinoma) were first diagnosed at autopsy. Several of these patients had a history of coronary risk factors. In 52% of death certificates, there was discrepancy with the autopsy cause of death. There was no significant difference in the discrepancy rate between community hospital and private autopsies (Table 4). The rate was lowest in patients with known symptomatic heart disease (33%) and highest in patients without medical history (60%), although the difference was not statistically significant. Of the seven patients with known coronary artery disease, two died of other causes (28%). The majority of incorrect diagnoses were coronary disease on the death certificate when autopsy showed another cause of death (22/28 or 78%). The sensitivity of coronary artery diagnosis was high, at 85%, but the specificity and the predictive value were low, with only 50% of cases of coronary disease death certificate causes of death accurate as reflected by autopsy findings. The majority of noncoronary deaths were attributed to coronary disease on the initial death certificates. These included seven cases of
Fig. 2. Subvalvular aortic stenosis. A discrete membrane is present in the outflow tract of the left ventricle, with dysplastic changes of the valve leaflets.
181
Table 4 Rate of discordance between death certificates versus autopsy diagnosis, by level of prior medical history Autopsies with incorrect cause of death, as indicated on death certificate
No prior medical history Prior medical history, excluding symptomatic heart disease History of symptomatic heart disease Total
Community (n=13)
Private (n=41)
Total (n=54)
2/3 3/6
7/12 11/18
9/15 (60%) 14/24 (58%)
1/4
4/11
5/15 (33%)
6/13 (46%)
22/41 (54%)
28/54 (52%)
cardiomyopathy, seven cases of aneurysm, six cases of pulmonary embolism, and two cases of valve disease. 4. Discussion The current study demonstrates that a large proportion of death certificates mistakenly assign coronary artery disease as the cause of out-of-hospital sudden deaths. This finding is easily explained, as coronary artery disease is the most common cause of sudden unexpected death and, therefore, is the most likely correct diagnosis without the benefit of autopsy. Furthermore, many patients who die suddenly of noncoronary causes have risk factors for coronary disease, as these are very prevalent and occasionally increase the risk for noncoronary disease such as hypertensive cardiomyopathy. Although coronary artery disease is the most common cause of sudden death in adults, it comprised fewer than 50% of cases in the current study. Because of the overdiagnosis of coronary disease, the predictive value of atherosclerotic heart disease on the death certificate was barely 50%. Relatively uncommon causes of sudden death, such as cardiomyopathy, ruptured aneurysms, and valve disease, were rarely correctly identified on the death certificate, unless there was a specific prior history and diagnosis. This issue is of particular value in cases where the diagnoses may carry a genetic risk for the living relatives, such as in cases of ARVD and HCM. In these instances, the importance of cardiovascular assessment of out-of-hospital sudden deaths is reinforced in the current study. The role of autopsy selection bias in the current study is undoubtedly significant. Out-of-hospital sudden death with any doubt about manner of death would be excluded from the current sample, as these cases would be referred to the medical examiner for autopsy. Family members requesting an autopsy and were willing to pay for the service, as in the case of the private autopsies, may be motivated by a doubt about the death certificate diagnosis, increasing the risk for death certificate inaccuracy. However, several of the deaths in the current study had well-documented underlying heart disease that is known to predispose to sudden death. Therefore, the selection bias is not exclusively toward deaths in which the diagnosis is likely suspect but may also include
182
F. Tavora et al. / Cardiovascular Pathology 17 (2008) 178–182
instances in which the family wished to confirm a diagnosis for litigious or family risk issues. Previous studies addressing autopsy discrepancies have largely focused on inpatients. Discrepancies have been classified as major (Class I), which would have changed management and potentially prolonged survival, and Class II, which were major discrepancies but without impact on treatment [9]. For out-of-hospital deaths that are sudden, these criteria have limited meaning, as the patient has often already expired before a diagnosis is rendered. There have been at least two studies comparing initial death certificate diagnoses and autopsy diagnosis in sudden nonhospital-based deaths. One study, comparing antemortem diagnoses in the emergency department with postmortem findings, demonstrated similar results as the current study. O'Connor et al. [10] discovered complete agreement between clinical impression and postmortem result in 51% of deaths. In another study by Ermenc [6], there was a 43% discrepancy rate in forensic cases, including complete and partial disagreements. In epidemiologic studies, in which accuracy of death certificates in out-of-hospital deaths is assessed by retrospective review that does not always include autopsy information, there is a wide variation in results. Unlike the current study, Folsom et al. reported a high sensitivity and specificity of coronary heart disease diagnosis on out-ofhospital sudden deaths in the Minneapolis–St. Paul region. This study sampled random death certificates but did not always have autopsy data available; the current study involved a highly selected population but showed that, even with a history of coronary artery disease, autopsy findings often disclose a different cause of death [8]. In contrast, Goraya et al. demonstrate that the predictive value of coronary heart disease diagnosis, as listed on the death certificate, was only 52% in cases of symptom duration less than 1 h. Therefore, definition of sudden death is important; in the current study, we used 1 h, as virtually all of the out-ofhospital sudden deaths were instantaneous [5]. The prevalence of disease also appears important. The current study demonstrates a low rate of sensitivity for diagnoses other than coronary artery disease; Tokashiki et al. [7] reported
that, in Okinawa, the highest rate of accuracy was stroke, which has a high prevalence in Japan. In conclusion, in a select group of autopsies performed in sudden unexpected deaths outside the hospital, we have shown that the accuracy rate of the death certificate cause of death is only about 50%, with coronary artery disease as the virtual default diagnosis in cases without specific prior history of other cardiovascular diseases. These findings are not inconsistent with prior reports and underscore the need for caution when interpreting death certificate data when estimating rates of sudden death due to coronary artery disease.
References [1] Iribarren C, Crow RS, Hannan PJ, Jacobs DR, Luepker RV. Validation of death certificate diagnosis of out-of-hospital sudden cardiac death. Am J Cardiol 1998;82(1):50–3. [2] O'Sullivan JP. The coroner's necropsy in sudden death: an under-used source of epidemiological information. J Clin Pathol 1996;49(9): 737–40. [3] Maudsley G, Williams EM. “Inaccuracy” in death certification—where are we now? J Public Health Med 1996;18(1):59–66. [4] Reid DW, Hendrick VJ, Aitken TC, Berrill WT, Stenton SC, Hendrick DJ. Age-dependent inaccuracy of asthma death certification in Northern England, 1991–1992. Eur Respir J 1998;12(5):1079–83. [5] Goraya TY, Jacobsen SJ, Belau PG, Weston SA, Kottke TE, Roger VL. Validation of death certificate diagnosis of out-of-hospital coronary heart disease deaths in Olmsted County, Minnesota. Mayo Clin Proc 2000;75(7):681–7. [6] Ermenc B. Discrepancies between clinical and post-mortem diagnoses of causes of death. Med Sci Law 1999;39(4):287–92. [7] Tokashiki T, Muratani A, Kimura Y, Muratani H, Fukiyama K. Sudden death in the general population in Okinawa: incidence and causes of death. Jpn Circ J 1999;63(1):37–42. [8] Folsom AR, Gomez-Marin O, Gillum RF, Kottke TE, Lohman W, Jacobs DR. Out-of-hospital coronary death in an urban population— validation of death certificate diagnosis. The Minnesota Heart Survey. Am J Epidemiol 1987;125(6):1012–8. [9] Goldman L, Sayson R, Robbins S, Cohn LH, Bettmann M, Weisberg M. The value of the autopsy in three medical eras. N Engl J Med 1983;308(17):1000–5. [10] O'Connor AE, Parry JT, Richardson DB, Jain S, Herdson PB. A comparison of the antemortem clinical diagnosis and autopsy findings for patients who die in the emergency department. Acad Emerg Med 2002;9(9):957–9.
Original Article
Discrepancies in initial death certificate diagnoses in sudden unexpected out-of-hospital deaths: the role of cardiovascular autopsy Fabio Tavora a,b,⁎, Clinton Crowder a , Robert Kutys c , Allen Burke a,c a
Department of Pathology, University of Maryland, 22 South Greene Street, Room NBW43, Baltimore, MD 21201, USA b Department of Pathology, The Johns Hopkins University, School of Medicine, Baltimore, MD 21231, USA c CVPath Institute, Inc., 19 Firstfield Road, Gaithersburg, MD 20878, USA Received 31 May 2007; received in revised form 10 July 2007; accepted 24 July 2007
Abstract Background: The accuracy of death certificates issued for out-of-hospital sudden deaths has been questioned. Methods: We retrospectively studied a series of consecutive autopsies performed at two community hospitals. Results: Fifty-four autopsies in which the deaths were sudden and that occurred outside the hospital were retained for study. The indication for autopsy was largely driven by the wishes of family or physician who was uncertain about the diagnosis. The overall discrepancy rate was 52%. The death certificate diagnosis, rendered before autopsy, was coronary artery disease in 44/54 autopsies (81%). At autopsy, coronary artery disease was the cause of death in 26 cases (48%), cardiomyopathy in 10 (18%), ruptured aneurysm in 8 (15%), pulmonary embolism in 7 (13%), and valve disease in 3 (6%). The diagnosis of coronary artery disease on the death certificate was accurate only 50% of the time. The discrepancy rate was lowest in patients with a history of cardiovascular disease (33%) and was 60% in patients with no prior medical history. The accuracy rate of death certificates was under one third in cases of cardiomyopathy, valve disease, ruptured aneurysm, pulmonary embolism, and valve disease. Conclusions: We conclude that in a highly selected group of sudden deaths, in which there was often a question about cause of death, the rate of initial death certificate accuracy is only one half. Furthermore, coronary artery disease as the cause of death is less than 50%, far less than initial death certificate diagnoses would indicate. © 2008 Elsevier Inc. All rights reserved. Keywords: Autopsy discrepancy; Coronary disease; Ruptured aneurysm; Quality control
1. Introduction Studies assessing the validity of death certificates issued on out-of-hospital sudden deaths have shown a high rate of discrepancy between clinical and autopsy findings [1,2]. There are various methods of assessing death certificate accuracy, but none is straightforward or easily verifiable [3]. Methods of assessing death certificate accuracy include retrospective review of medical records and death certificates by experts [4]; retrospective review of random fraction of population-based deaths, including death certificates and coroners' files [1,5]; comparison of causes of death before and after autopsies were performed after filling out the initial
⁎ Corresponding author. Tel.: +1 410 3285555; fax: +1 410 3255508. E-mail address: [email protected] (F. Tavora). 1054-8807/08/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.carpath.2007.07.010
death certificates [6]; and retrospective review of all available information in population-based studies of sudden unexpected death [7]. Some studies [1], but not others [8], demonstrate a high rate of discrepancy when assessing validity of death certificate diagnoses, specifically in cases of sudden death. This rate of discrepancy is especially high if the death occurs within 1 h of onset of symptoms before medical diagnostic procedures can take place [5]. In the state of Maryland, out-of-hospital sudden deaths are handled by emergency services, and death certificates are signed by the deceased's physician, based on prior knowledge of the patient's medical condition. If there is no available physician, the death is referred to the regional medical examiner, an employee of the statewide medical examiner's system. If there are no suspicious circumstances, the body is released to the funeral home, and a death certificate is signed by the medical examiner based on
F. Tavora et al. / Cardiovascular Pathology 17 (2008) 178–182 Table 1 Case characteristics, sudden death patients
Autopsy setting Community hospital Private Gender Male Female Medical history No prior medical history Prior medical history, excluding symptomatic heart disease a History of symptomatic heart disease b Total
n
Age (in years), mean±S.D.
13 41
65±15 65±11
27 27
61±9 69±13 ⁎
15 24
66±12 65±12
15 54
67±13 66±12
a
There were 10 patients with multiple coronary risk factors, including 3 with diabetes; 7 with neoplasia; 3 with prior negative cardiac evaluation, including negative stress echocardiogram; 2 with pulmonary disease (asthma, pulmonary nodule); 1 with prior genitourinary surgery; and 1 with neurologic disorder. b There were 7 patients with known coronary artery disease, 2 with prior cerebrovascular accident, 1 with known amyloid, 1 with known aortic stenosis, 1 with a history of pulmonary embolism and giant cell arteritis, 1 with known abdominal and thoracic aortic aneurysm, 1 with known dilated cardiomyopathy, and 1 with a history of aortic valve replacement. ⁎ P=.01 versus men.
interviews with the family and investigation of the scene. The death certificate allows for an amended diagnosis, made after autopsy, if performed. In out-of-hospital deaths that are turned down by the medical examiner or signed by a physician who had contact with the patient, the family may request an autopsy based on a desire for more specific cause of death, concerns about hereditary disease, litigious concerns, or documentation of death certificate accuracy for insurance purposes. In many cases, the autopsies are referred to private (paid) services because community hospitals are generally unwilling to provide autopsy services unless the deceased was a recent inpatient. The purpose of this study was to report a series of out-ofhospital sudden deaths, in which autopsy was performed based on family request and in which a cardiovascular event was either diagnosed or considered. The autopsy diagnosis was compared to the death certificate diagnosis, and concordance was assessed based on final autopsy diagnosis, degree of clinical evaluation, and age and gender of the patient. 2. Materials and methods Out-of-hospital deaths were referred to two settings for autopsies, occurring between 2001 and 2006: a community hospital and a private autopsy service, which performed all autopsies in a second community hospital, prior to embalming. The cases referred to the community hospital were performed because of the deceased's physician's attending privileges at the institution; in these cases, the deceased's
179
physician signed the death certificate in every instance. The private autopsy service performed autopsies at the family's request, and the death certificate was signed either by a family physician or by the medical examiner, after nonnatural causes of death were excluded by way of investigation of the death scene and interview with family members. Unlike community hospitals, private autopsies required payment by the deceased's family. Cases were included for study if the death occurred outside the hospital or if death was confirmed in the emergency room of a hospital, before the patient was admitted. The time frame encompassed the period 2001– 2006. The cases comprised a subset of autopsies that included inpatient deaths at both institutions. Sudden death was defined as death within 1 h of onset of symptoms. Patients with prior heart disease were included if they were asymptomatic prior to the sudden death. The same prosector with subspecialty expertise in cardiovascular pathology performed all autopsies. In all cases, a minimum of 20 histologic sections was performed, including sections of all viscera. Examination of the brain was performed if there were any neurologic symptoms or if examination of the cardiovascular system did not demonstrate a clear-cut cardiovascular cause. History was available in the community hospital based on medical records of prior admissions and telephonic conversations with the family physician. History was available in the private autopsies based on telephone conversations with the next of kin, who requested and paid for the autopsies. The historical information was gathered at the time of autopsy. Correlation between autopsy findings, death certificate findings, and clinical information was done retrospectively at the time of study, based on archived materials. Discrepancy was defined as the absence of the anatomic cause of death anywhere in the death certificate or if the primary cause of death as recorded on the death certificate was not found at autopsy. The autopsy causes of death were classified into five groups: coronary artery disease, cardiomyopathy, pulmonary embolism, valve disease, and ruptured aneurysm.
3. Results There were 54 autopsies fulfilling the inclusion criteria, including 13 performed at the community hospital and 41 referred for private autopsy. These formed a subset of 158 community adult autopsies and 82 private autopsies, representing 8% and 50%, respectively, of the entire series of autopsies. There were 27 men and 27 women. The mean age of the women was significantly higher than that of the men (Table 1). Fifteen patients had no prior medical history, and 15 patients had a history of symptomatic cardiovascular disease. The remaining 24 patients had prior medical care but no history of symptomatic cardiovascular disease.
180
F. Tavora et al. / Cardiovascular Pathology 17 (2008) 178–182
Table 2 Case distribution by cause of death, age, and gender Cause of death, autopsy
n (%)
Men:Women
Age (in years), mean±S.D.
Coronary artery disease Cardiomyopathy Aneurysm Pulmonary embolism Valve disease Total
26 (48) 10 (18) 8 (15) 7 (13) 3 (6) 54
12:14 6:4 5:3 2:5 2:1 27:27
67±11 56±9 ⁎ 72±9 67±17 58±12 66±12
⁎ P=.005 versus coronary artery disease; P=.003 versus aneurysm.
The most common autopsy cause of death was coronary artery disease in 48%, followed in decreasing incidence by cardiomyopathy, aneurysm, pulmonary embolism, and valve disease (Table 2). There were no significant gender differences, although cardiomyopathy patients were the youngest, significantly younger than coronary and aneurysm patients (Table 2). The majority of patients suffered from presumed instantaneous sudden death. Four of the 54 patients complained of chest pain, shortness of breath, or abdominal discomfort moments before death; 22 patients were found unresponsive but were in their normal state of health while last seen, and 28 were witnessed arrests without apparent symptoms. The majority of coronary deaths had no evidence of prior infarction, and 15 of 26 deaths demonstrated epicardial thrombosis (Table 3). Of 26 hearts in the coronary artery disease groups, 11 demonstrated moderate to severe cardiomegaly, with heart weights N500 g (Table 3). Of the
Table 3 Cardiac findings and gender distribution, coronary artery disease deaths Finding
n
Heart weight Male:Female N500 g
No prior infarct 18 5:13 Acute plaque rupture 7 3:4 Acute plaque rupture with acute infarct 2 1:1 Acute plaque rupture with ruptured 2 0:2 acute infarct Severe coronary atherosclerosis, 4 1:3 no thrombus Severe coronary atherosclerosis, 1 0:1 status post-CABG Nodular calcification with rupture 1 0:1 and thrombus Acute infarct, without thrombus 1 0:1 Prior infarct (healed myocardial infarct) 8 7:1 Plaque rupture, without acute infarct 2 2:0 Plaque rupture with acute infarct 1 1:0 Severe coronary atherosclerosis, 2 2:0 status post-CABG Severe coronary atherosclerosis, 1 1:0 status post-stenting Severe coronary atherosclerosis, 2 1:1 no thrombus CABG, coronary artery bypass graft surgery.
7/18 4/7 0/2 0/2 1/4
Fig. 1. Hypertrophic cardiomyopathy. (A) Gross photograph demonstrates septal hypertrophy and left ventricular outflow tract plaque (arrow). (B) Microscopic section demonstrates myofiber disarray.
0/1 1/1 1/1 4/8 1/2 1/1 1/2 1/1 1/2
15 patients in the entire study with a history of heart disease, only 7 were in the coronary artery disease autopsy group, including 3 with prior coronary artery bypass graft surgery and 1 with prior coronary stenting. Three patients had recent negative cardiac evaluations, including stress testing; interestingly, the initial death certificates indicated atherosclerotic heart disease as the cause of death. Of the noncoronary sudden cardiac deaths, there were 10 cardiomyopathies, 8 ruptured aneurysms, and 3 valve diseases as the cause of sudden cardiac death. The cardiomyopathy group included one patient with amyloidosis, two patients with arrhythmogenic right ventricular
F. Tavora et al. / Cardiovascular Pathology 17 (2008) 178–182
cardiomyopathy (ARVD), two patients with dilated cardiomyopathy, and three patients with hypertrophic cardiomyopathy (HCM). Three of the cardiomyopathy sudden deaths occurred during exercise. These included HCM (Fig. 1) and ARVD. The aneurysms included four dissections, two abdominal aortic aneurysm ruptures, one thoracic aneurysm rupture that had been followed medically, and one sinus of Valsalva aneurysm rupture. One patient with aortic stenosis had a prior diagnosis and was found dead the day before the scheduled aortic valve replacement. The other patient had no prior history of heart disease and was found to have subaortic stenosis with left ventricular hypertrophy at autopsy (Fig. 2). Of the seven patients with diagnosed pulmonary embolism, five had underlying risk factors, two of which (pancreatic adenocarcinoma) were first diagnosed at autopsy. Several of these patients had a history of coronary risk factors. In 52% of death certificates, there was discrepancy with the autopsy cause of death. There was no significant difference in the discrepancy rate between community hospital and private autopsies (Table 4). The rate was lowest in patients with known symptomatic heart disease (33%) and highest in patients without medical history (60%), although the difference was not statistically significant. Of the seven patients with known coronary artery disease, two died of other causes (28%). The majority of incorrect diagnoses were coronary disease on the death certificate when autopsy showed another cause of death (22/28 or 78%). The sensitivity of coronary artery diagnosis was high, at 85%, but the specificity and the predictive value were low, with only 50% of cases of coronary disease death certificate causes of death accurate as reflected by autopsy findings. The majority of noncoronary deaths were attributed to coronary disease on the initial death certificates. These included seven cases of
Fig. 2. Subvalvular aortic stenosis. A discrete membrane is present in the outflow tract of the left ventricle, with dysplastic changes of the valve leaflets.
181
Table 4 Rate of discordance between death certificates versus autopsy diagnosis, by level of prior medical history Autopsies with incorrect cause of death, as indicated on death certificate
No prior medical history Prior medical history, excluding symptomatic heart disease History of symptomatic heart disease Total
Community (n=13)
Private (n=41)
Total (n=54)
2/3 3/6
7/12 11/18
9/15 (60%) 14/24 (58%)
1/4
4/11
5/15 (33%)
6/13 (46%)
22/41 (54%)
28/54 (52%)
cardiomyopathy, seven cases of aneurysm, six cases of pulmonary embolism, and two cases of valve disease. 4. Discussion The current study demonstrates that a large proportion of death certificates mistakenly assign coronary artery disease as the cause of out-of-hospital sudden deaths. This finding is easily explained, as coronary artery disease is the most common cause of sudden unexpected death and, therefore, is the most likely correct diagnosis without the benefit of autopsy. Furthermore, many patients who die suddenly of noncoronary causes have risk factors for coronary disease, as these are very prevalent and occasionally increase the risk for noncoronary disease such as hypertensive cardiomyopathy. Although coronary artery disease is the most common cause of sudden death in adults, it comprised fewer than 50% of cases in the current study. Because of the overdiagnosis of coronary disease, the predictive value of atherosclerotic heart disease on the death certificate was barely 50%. Relatively uncommon causes of sudden death, such as cardiomyopathy, ruptured aneurysms, and valve disease, were rarely correctly identified on the death certificate, unless there was a specific prior history and diagnosis. This issue is of particular value in cases where the diagnoses may carry a genetic risk for the living relatives, such as in cases of ARVD and HCM. In these instances, the importance of cardiovascular assessment of out-of-hospital sudden deaths is reinforced in the current study. The role of autopsy selection bias in the current study is undoubtedly significant. Out-of-hospital sudden death with any doubt about manner of death would be excluded from the current sample, as these cases would be referred to the medical examiner for autopsy. Family members requesting an autopsy and were willing to pay for the service, as in the case of the private autopsies, may be motivated by a doubt about the death certificate diagnosis, increasing the risk for death certificate inaccuracy. However, several of the deaths in the current study had well-documented underlying heart disease that is known to predispose to sudden death. Therefore, the selection bias is not exclusively toward deaths in which the diagnosis is likely suspect but may also include
182
F. Tavora et al. / Cardiovascular Pathology 17 (2008) 178–182
instances in which the family wished to confirm a diagnosis for litigious or family risk issues. Previous studies addressing autopsy discrepancies have largely focused on inpatients. Discrepancies have been classified as major (Class I), which would have changed management and potentially prolonged survival, and Class II, which were major discrepancies but without impact on treatment [9]. For out-of-hospital deaths that are sudden, these criteria have limited meaning, as the patient has often already expired before a diagnosis is rendered. There have been at least two studies comparing initial death certificate diagnoses and autopsy diagnosis in sudden nonhospital-based deaths. One study, comparing antemortem diagnoses in the emergency department with postmortem findings, demonstrated similar results as the current study. O'Connor et al. [10] discovered complete agreement between clinical impression and postmortem result in 51% of deaths. In another study by Ermenc [6], there was a 43% discrepancy rate in forensic cases, including complete and partial disagreements. In epidemiologic studies, in which accuracy of death certificates in out-of-hospital deaths is assessed by retrospective review that does not always include autopsy information, there is a wide variation in results. Unlike the current study, Folsom et al. reported a high sensitivity and specificity of coronary heart disease diagnosis on out-ofhospital sudden deaths in the Minneapolis–St. Paul region. This study sampled random death certificates but did not always have autopsy data available; the current study involved a highly selected population but showed that, even with a history of coronary artery disease, autopsy findings often disclose a different cause of death [8]. In contrast, Goraya et al. demonstrate that the predictive value of coronary heart disease diagnosis, as listed on the death certificate, was only 52% in cases of symptom duration less than 1 h. Therefore, definition of sudden death is important; in the current study, we used 1 h, as virtually all of the out-ofhospital sudden deaths were instantaneous [5]. The prevalence of disease also appears important. The current study demonstrates a low rate of sensitivity for diagnoses other than coronary artery disease; Tokashiki et al. [7] reported
that, in Okinawa, the highest rate of accuracy was stroke, which has a high prevalence in Japan. In conclusion, in a select group of autopsies performed in sudden unexpected deaths outside the hospital, we have shown that the accuracy rate of the death certificate cause of death is only about 50%, with coronary artery disease as the virtual default diagnosis in cases without specific prior history of other cardiovascular diseases. These findings are not inconsistent with prior reports and underscore the need for caution when interpreting death certificate data when estimating rates of sudden death due to coronary artery disease.
References [1] Iribarren C, Crow RS, Hannan PJ, Jacobs DR, Luepker RV. Validation of death certificate diagnosis of out-of-hospital sudden cardiac death. Am J Cardiol 1998;82(1):50–3. [2] O'Sullivan JP. The coroner's necropsy in sudden death: an under-used source of epidemiological information. J Clin Pathol 1996;49(9): 737–40. [3] Maudsley G, Williams EM. “Inaccuracy” in death certification—where are we now? J Public Health Med 1996;18(1):59–66. [4] Reid DW, Hendrick VJ, Aitken TC, Berrill WT, Stenton SC, Hendrick DJ. Age-dependent inaccuracy of asthma death certification in Northern England, 1991–1992. Eur Respir J 1998;12(5):1079–83. [5] Goraya TY, Jacobsen SJ, Belau PG, Weston SA, Kottke TE, Roger VL. Validation of death certificate diagnosis of out-of-hospital coronary heart disease deaths in Olmsted County, Minnesota. Mayo Clin Proc 2000;75(7):681–7. [6] Ermenc B. Discrepancies between clinical and post-mortem diagnoses of causes of death. Med Sci Law 1999;39(4):287–92. [7] Tokashiki T, Muratani A, Kimura Y, Muratani H, Fukiyama K. Sudden death in the general population in Okinawa: incidence and causes of death. Jpn Circ J 1999;63(1):37–42. [8] Folsom AR, Gomez-Marin O, Gillum RF, Kottke TE, Lohman W, Jacobs DR. Out-of-hospital coronary death in an urban population— validation of death certificate diagnosis. The Minnesota Heart Survey. Am J Epidemiol 1987;125(6):1012–8. [9] Goldman L, Sayson R, Robbins S, Cohn LH, Bettmann M, Weisberg M. The value of the autopsy in three medical eras. N Engl J Med 1983;308(17):1000–5. [10] O'Connor AE, Parry JT, Richardson DB, Jain S, Herdson PB. A comparison of the antemortem clinical diagnosis and autopsy findings for patients who die in the emergency department. Acad Emerg Med 2002;9(9):957–9.