- Email: [email protected]
The value of the hospital autopsy. A study of causes and modes of death estimated before and after autopsy
Forensic Science International, 21 (1983) Elsevier Scientific Publishers Ireland Ltd.
23-32
23
THE VALUE OF THE HOSPITAL AUTOPSY. A STUDY OF CAUSES AND MODES OF DEATH ESTIMATED BEFORE AND AFTER AUTOPSY
S. ASNAES, V. FREDERIKSEN
and C. FENGER
University Institute of Forensic Medicine, Copenhagen, and Holstebro Medical and Pathological Departments, Holstebro (Denmark)
Central Hospital,
(Received March 28,1982) (Revision received June 17, 1982) (Accepted July 9, 1982)
Summary Among 312 consecutive deaths in a Danish Central Hospital autopsy was performed in the pathology department on 266 cases, i.e. 85%. Retrospectively, the underlying causes of death were estimated from the clinical information alone by an experienced clinician and subsequently compared with the autopsy report. The definite cause of death was determined jointly by the clinician and the pathologist. The clinician’s diagnosis was thereby confirmed as incorrect in 18% of the cases if small differences in site and type of malignant tumours were not considered. This is less than in many other investigations, but it is stressed that this could partly be because formal errors in completing the death certificate were avoided, The main causes of death were ischaemic heart disease and neoplasia. Clinical diagnosis of malignant diseases was never found to be erroneous. There was a slight tendency to clinically overestimate ischaemic heart disease, but in general the different errors outweighed each other, so that the total number of different causes of death before and after autopsy was nearly the same. The original death certificate was investigated in 12 accidental cases. Hereby it was found that the mode of death was originally stated erroneously as natural in 7 cases, i.e. 4.5%. It is concluded that hospital autopsy is still needed for the control and correction of causes of death, and it is stressed that clinicians as well as pathologists should be more aware of cases with a trauma in the history to avoid errors in the mode of death. Such errors can imply legal as well as insurance problems. Key words: Hospital autopsy; Errors in causes of death
Introduction Periodically it medical research that the cause of cases if only the
is debated how significant the hospital autopsy is for and treatment [l-3]. Many studies [4-111 have shown death will be stated erroneously in about one third of all clinical information is considered. Errors will occur more
0379-0738/83/0000-0000/$03.00
0 Elsevier Scientific Publishers Ireland Ltd. Printed and Published in Ireland
24
frequently among elderly people and many contributory causes of death will often not be discovered until autopsy is performed [ 121. With a high autopsy rate in a well-defined population the severeness, extension and frequency of many diseases can be determined in detail, and in the epidemiological research the mortality statistics are still a useful aid in elucidating the need of medical assistance, making prognoses and judging the effect of medical assistance [ 131. In 1952 Munck compared the presumed cause of death before autopsy and the result of the autopsy and found 20% of errors. Since that time the diagnostic methods within clinical medicine have changed and been modernized. This could perhaps give a higher degree of security in establishing the correct cause of death. The present retrospective study was therefore performed to elucidate what significance the hospital autopsy has for the correct statement of the cause of death to-day. Materials and methods Materials
266 autopsies (114 women and 152 men) were performed in 1974 in a 530-bed Central Hospital, situated in a county with 250,000 inhabitants. The total number of deaths in the hospital in the same period was 312, i.e. the autopsy rate was 85%. Methods
First the cause of death was estimated from clinical information alone (V.F.). The logical sequence between the different conditions leading to death was secured according to the WHO regulations for underlying causes of death [ 141. The aim was to avoid formal errors, which even if they reflect the uncertainty of our mortality statistics, would not give any information about the value of the autopsy. Secondly the clinically estimated cause of death was compared with the autopsy findings. In each case the clinician and the pathologist jointly estimated whether the autopsy findings could change the clinical cause of death. TABLE
I
AGE IN DECADES
Males Females Total
O-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-
Total
11 12 23
1
1 1 2
12 9 21
17 10 27
48 23 71
45 39 84
14 14 28
3 6 9
152 114 266
1
25
In all cases where deaths were considered to be non-natural a copy of the original death certificate was asked for. Results Eightyone percent of the women and 84% of the men were more than 50 years of age (Table 1). Totally we found 48 differing causes of death (18%), if 12 cases, where a malignant disease was stated wrongly concerning origin and histological type, are not included. In two cases the mode of death was estimated erroneously, i.e. 0.8%. The different causes and modes of death in certain main groups are shown in Table 2. In 136 cases opening of the skull was performed, i.e. 51%. In 146 cases microscopy was performed after the autopsy, i.e. 55%. We found no characteristic relation between the number of differing causes of death and age groups. Besides a slightly pronounced tendency to overdiagnose ischaemic heart disease (IHD) as cause of death among men as compared to women, no characteristic difference in the two sexes was found.
Accidents In 14 cases an accident was suggested as the cause of death. This was correct in 12 cases (5%). Among women the most common accident was stumbling at home and in men it was traffic accidents. One case where the clinician wrongly estimated that death was due to an accident was that of a 74-year-old woman, who was taken to hospital after a trauma against the head. The blood pressure was 230/130 and an atypical plantar reflex was found. An X-ray contrast investigation was performed with suspicion of subdural haematoma, and death occurred. At autopsy a swelling of the brain and scattered haemorrhages in the pons were found. There was no sign of trauma against the skull and the findings were considered to be due to hypertension. By investigating the original death certificates it was found that all traffic accidents were correctly reported to the police, external medico-legal examinations were performed and all cases were correctly registered as accidents. Five cases, where death was caused by stumbling in the home were all reported to the police, but medico-legal examination was not asked for. Despite this, the mode of death was stated as natural in the hospital and later registered as such in the General Register. Two cases of accidental deaths were not reported to the police and stated as natural. One case was that of a 7-year-old boy who had a trauma against the head. Shortly afterwards he developed a headache and a slight fever. He was taken to the local hospital where it was assumed he had got a cold and was sent home. Since then he suffered from an occasional headache, and when he again became febrile he was taken to the hospital where he died
1
412
TOTAL
37
21
2
Others (disagreement)
1
1
12
6
Accidents
Others (agreement)
Pneumonia
Subarachnoid haemorrhage
1
34
410
Cerebral haemorrhage
Stillborn
Pulmonary embolism
Malignant disease
Ischaemic heart disease
63
1
1
60
1
14
3
2
6
2
1
20
20
22
2
18
1
1
4
4
11
2
1
44
44
12
12
CAUSES AND MANNERS OF DEATHS IN 266 HOSPITAL DEATHS ESTIMATED BEFORE AND AFTER AUTOPSY
TABLE 2
18
8
1
1
3
2
1
2
266
12
14
44
13
6
25
20
10
60
17
45
K
27
from a meningococcical meningitis. At autopsy a small fracture in the lamina cribosa was found, and this was assumed to be the entry of the infection. Natural death
There was a total of 254 cases of natural deaths, i.e. 95%. From Table 2 it appears that the cause of death in 62 cases (45 + 17) was reported as IHD, i.e. 23%. This was correct in 46 cases, i.e. 74%. Out of 45 estimated cases of fresh myocardial infarction 34 were correct. The differing causes of death are shown in Table 2. In one case the cause of death was stated as unknown after autopsy. It was a 35-year-old man, who according to his wife had previously developed symptoms of tachycardia, but who was otherwise healthy. He was suddenly taken unconscious and died shortly after arrival in hospital. At autopsy only a slight coronary sclerosis was demonstrated. Among 17 presumed deaths from pure coronary sclerosis 12 were correct. The differing causes of death were fresh myocardial infarction, bronchogenic carcinoma, pulmonary embolism and thrombosis of a mesenterial vein. Sixty cases were correctly stated as a malignant disease (Table 2). The specific type of the tumours was stated erroneously in 12 cases, i.e. 20% (Tables 3, 4 and 5). In six out of 10 cases it was correctly presumed that the cause of death was pulmonary embolism. The differing cases were cerebral haemorrhage, aortic stenosis, myocardial infarction and peritonitis. Cerebral haemorrhage was presumed in 25 cases. Eighteen were correct. The correctly stated cases all concerned patients who had been in the ward for several days. The cliniCal TABLE
3
FORTY-EIGHT
CORRECTLY
STATED
DEATHS
MALIGNANT
DISEASES
Men
Women Cancer coli Adenocarcinoma ovarii Bronchogenic carcinoma Cancer mammae Pancreatic cancer Gastric cancer Renal cancer Leiomyosarcoma Cholangiocarcinoma Carcinoma planocellulare vaginae Reticulosarcoma Chorioepithelioma
(5P (3) (2) (2) (2) (1) (1) (1) (1) (1) (1) (1)
TOTAL aThe figures in parentheses
FROM
(21) indicate
the number
Bronchogenic carcinoma Gastric cancer Pancreatic cancer Malignant ependymoma Epidermoid cancer (on neck) Epidermoid cancer (on tongue) Carcinoma urinaric bladder Colonic carcinoma Hepatoma Cholangiocarcinoma Renal cancer Melanocarcinoma
(14) (2) (2) (1) (1) (1) (1) (1) (1) (1) (1) (1)
TOTAL
(27)
of cases.
28 TABLE
4
DIAGNOSIS IN SIX AUTOPSY (MEN)
DEATHS
FROM
MALIGNANT
DISEASES
BEFORE
Clinical diagnosis
Au topsy
Total
c. occultus
Carcinoid tumor of the rectum Bronchogenic carcinoma Pancreatic cancer Seminoma
1 1 1 1
Bronchogenic carcinoma Cholangiocarcinoma
1 1
Prostatic
cancer
(2)
TOTAL
AND AFTER
6
picture was in all cases rather.characteristic with a hemiparesis, unconsciousness and a high blood pressure. Different causes of death were subarachnoid bleeding, thrombosis of a mesenterial vein and bronchopneumonia. Subarachnoid haemorrhage was stated as cause of death in six cases. One of the differing causes of death was a 64-year-old woman with a headache during many years who died shortly after 2 h of intensive pain. At autopsy a cerebral haemorrhage was found. Bronchopneumonia was overdiagnosed. Among 13 presumed cases only 8 were correct. In the other cases a bronchogenic carcinoma, ulcerative endocarditis with cardiac insufficiency and pulmonary embolism were found. In 44 cases the cause of death was correctly stated (Table 6). In 12 cases other differing causes of death were found. Four cases occurred in women. Two cases were presumed to be bronchitis, but the autopsy revealed TABLE
5
DIAGNOSIS IN SIX AUTOPSY (WOMEN)
DEATHS
FROM
MALIGNANT
DISEASES
BEFORE
Clinical diagnosis
Au topsy
Total
c. occultus
Cholangiocarcinoma Carcinoma of the cervix Alveolar cell carcinoma Adenocarcinoma ovarii
1 1 1 1
Colonic
Adenocarcinoma
1
cancer
Adenocarcinoma TOTAL
ovarii
Pancreatic
cancer
ovarii
1 6
AND AFTER
29
TABLE
6
FORTY-FOUR
CASES OF NOT DIFFERING
CAUSES OF DEATH
Women
Men
Diabetes Hypertensio arterialis Aortic stenosis Diverticulosis coli Choledocholithiasis Arteriosclerosis universalis Bronchitis chronica Peritonitis Cholecystitis Paraoesophageal hernia Polycytaemia Colitis ulcerosa Chronic pyelonephritis Perforated ulcus ventriculi Hyperparathyoidism Essential pulmonal hypertension Sepsis
Chronic bronchitis Pancreatitis acuta Diabetes Adipositas Hyperlipoproteinaemia Rheumatic endocarditis Acute pyelonephritis Amyotrophic lateralsclerosis Thrombosis of mesenterial vein Arterial hypertension Duodenal ulcer with penetration Ischaemic colititis Uraemia after renal infarction v01v01us
TOTAL
TOTAL
(341
a The figures in parentheses indicate the number of cases.
a myocardial infarction and uremia. In one case the clinical findings were similar to hyperparathyroidism, but at autopsy a cirrhotic liver with bleeding oesophageal varices was found. One case of presumed chronic renal insufficiency was mitral stenosis and heart insufficiency. The causes of death among 8 men are shown in Table 7. In one case the cause of death was stated as septicaemia. It was a 74-yearold man who was taken to hospital after suffering at home for a long time TABLE
7
CAUSES OF DEATH BEFORE
AND AFTER
AUTOPSY
Clinical
Au topsy
Chronic bronchitis Chronic bronchitis Chronic bronchitis Esophageal varices Polycytaemia’ Unknown Septicaemia Abdominal tumor
Coronary sclerosis Aortic rupture Brochogenic carcinoma Bleeding ulcer Pancreatitis Meningococcal sepsis Pancreatic cancer Coronary sclerosis
TOTAL
IN EIGHT MEN Number
30
from pneumonia. At first he got a renal insufficiency. 10 days later amputation of the right leg was performed because of gangrene and after a few days death occurred. At autopsy a pancreatic carcinoma with metastases was found as an underlying cause of death. Totally, there were 37 fresh myocardial infarctions. In 14 of these, thrombosis was found, i.e. 35%. Three cases of malignant diseases were discovered, i.e. 1%.
Discussion Erroneous statements for the underlying causes of death before autopsy were 18%. This is rather low as compared to many other studies. Britton for example found a total of 30% of erroneous clinical diagnoses in hospital data from the Serafimer Hospital in Stockholm [6], and Asnaes found that the cause of death was erroneously estimated in about 50% before autopsy in non-medico-legal material from Copenhagen [15] . One reason for the discrepancy could be that we paid a lot of attention in discussing the correct and logic sequence of the different causes of death, whereby formal errors were avoided. Such formal errors can, as demonstrated by Asnaes et al., [16] and Skullerud and Skullzrud [17], give up to 20% wrongly estimated causes of death. If only errors, where the revised diagnosis belongs to different main groups of diseases within the ICD classification of the ‘list of three digit categories’, are counted, a total of 27 erroneously stated underlying causes of death exists, i.e. 10%. If the cases of differing modes of death are eliminated too, the cause of death is found differing in 9%. This comes very close to Britton’s investigation, where she found different causes of death in main groups in 7% [6]. The approach of this study concerned the estimation of the causes of death. However, it was also found that the mode of death was differing in 0.8% of the cases. This is only a theoretical problem in this retrospective study. It is more important that by investigating the original death certificate we found that the mode of death was originally stated wrongly as natural death in seven accidents, i.e. 5%. This is in accordance with the investigations of Otterland and Pihl [ 181, who found accidents to be under-recorded in the official statistics. Otterland and Pihl mention cases of fractures which, clinically, were classified as senile osteoporosis, without any mentioning of a previous trauma. While errors in the cause of death can influence mortality statistics, epidemiological investigations, etc., it is more serious if the mode of death is stated wrongly. Such errors can imply legal as well as insurance problems. Ischaemic heart disease was clinically overdiagnosed, but not so pronounced as in the medico-legal investigations, where ischaemic heart disease is overdiagnosed in about 30% of all cases [ 191.
31
In the present study fresh myocardial infarction was overdiagnosed to the disregard of chronic ischaemic heart oisease. From a clinical point of view overdiagnosis of myocardial infarction must be to the benefit of the patients (coronary care units, etc.). Among infarctions, coronary thrombosis was found in 35% of the cases. This is not-far from Asnaes’ and Qstergaard’s study [ 201, where eight thromboses among 16 recent infarctions were found in patients who died after a very short treatment in hospital. In contrast to this Asnaes [ 151 in a material of sudden unexpected deaths found 11 cases of coronary thrombosis, but only eight myocardial infarctions. The difference could be due to the fact that in the study of Asnaes and pstergaard the formazan test was used. Another explanation could be that the hospital treatment, even short, could give the myocardium the possibility of developing an infarction. Pulmonary embolism was slightly underestimated, pointing out that this is still a difficult clinical diagnosis. Cerebral haemorrhage and pneumonia were slightly overestimated. This is in contrast to similar medico-legal investigations [ 191. It must be considered, however, that the brain was only investigated in about 50% of all cases. Clinical diagnosis of neoplastic disorders were never found to be erroneous, but in 20% of the cases the histological type and the origin of the different tumours were first found at autopsy. Three cases of malignant disease were discovered, i.e. 1%. This is less than that found in similar medico legal investigations [ 191, where up to 4.5% malignant diseases were discovered. But it is not surprising that malignant disease is discovered more seldom among hospital deaths. Many of the diseases, where the clinical diagnosis was wrong, are wellknown as offering particular diagnostic difficulty (endocarditis, pancreatic cancer, ruptured aortic aneurysm). The aim of the present study was to investigate whether autopsies are important in determining theunderlying cause of death. We found the answer obviously positive, but in this study, as in others, it is characteristic that clinical overdiagnosis and underdiagnosis, to some degree, outweigh one another. In the present study the total number of different causes of death was nearly the same before and after autopsy. It is unsatisfactory, however, to know that a wrong cause of death would be corrected in another fault if autopsy is not performed. Besides a certain number of contributory causes of death would be discovered, whereby some clinical abnormalities could be explained. It is therefore concluded that hospital autopsies are still needed with a view to control and correction of causes of death. The autopsies mean a continuous control of diagnostic errors, but it is also important, when completing the death certificate, that the clinician as well as the pathologist are aware of whether the death is natural or due to an accident. A closer cooperation between the pathologist and the clinician in com-
32
pleting the death certificate help.
as well as more central control could possibly
References 1 2 3 4 5 6 7 8 9 10
11 12 13 14 15
16 17 18 19 20
J.T. Freeman, The Autopsy.Pa. Med., 71 (1968-1971). What about the autopsyEditorial, 193 (1965) 172. A. Juul and J. Andersen, Rutineautopsier. Ugeskr. Laeg., 139 (1977) 323. S. Asnaes and F. Paaske, The significance of medicolegal autopsy in determining mode and cause of death. Forensic Sci. Znt., 14 (1979) 23. W. Borris, Die Bedeutung der Vervaltungssektion fiir die Morbiditats- und Mortalitiitsstatistik. 2. Arztl. Fortbild., 46 (1952) 471. M. Britton, Diagnostic errors discovered at autopsy. Acta Med. &and., 196 (1974) 203. R.H. Gruver and E.D. Freis, A study of diagnostic errors. Ann. Intern. Med., 47 (1957) 108. M. Heasman and L. Lipworth, Accuracy of Certification of Cause of Death. Her Majesty’s Stationery Office, London, 1966. W. Munck, Autopsy finding and clinical diagnosis. A comparative study of 1000 Cases. Acta Med. Stand. Suppl. 226 (1952) 775. N. Schulz and Schaarschmidt, Zur Relation zwischen klinischen und pathologischanatomischen Diagnosen in der Todesursachenstatistik. Zentralbl. Gesamte. Hyg., 16 (1970) 928. I. Wallgren, Obduktionsfyndet och de kliniska diagnoserna. Nord. Med., 26 (1945) 1311. M. Britton, Clinical diagnostics: Experience from 383 autopsied cases. Acta Med. Stand., 196 (1974) 211. K. Juel, Autopsihyppighedens betydning for d&ls%rsagsstatistikken. Ugeskr. Laeg., 143 (1981) 2669. WHO, International classification of diseases, 1965 Revision WHO, Geneva, 1967. S. Asnaes, Uncertainty of determining mode and cause of death without autopsy: An autopsy study of medically unattended non-medicolegal deaths. Forensic Sci. Znt., 15 (1980) 191. S. Asnaes, C.E. Mabeck and B. Wichmann, The significance of autopsy for statistice concerning causes of death. An investigation of death certificates. Ugeskr. Laeg., 142 (1980) 261. K. Skullerud and J. Skullerud, Mortality Statistics. Tds. Skr. Nor. Laegeforen., 98 (1978) 1748. A. Otterland and E. Pihl, Klinisk, patologisk-anatomisk og officiel d&is%rsagsstatistik med udgangspunkt i 327 cbduktioner. Lakartidningen, 61 (1964) 68. S. Asnaes and F. Paaske, Uncertainty of determining cause of death in a medicolegal material without autopsy. An autopsy study. Forensic Sci. Int., 15 (1980) 103. S. Asnaes and K. Qtergaard, Reliability of death certificates. An autopsy investigation. Ugeskr. Laeg., 142 (1980) 265.
23-32
23
THE VALUE OF THE HOSPITAL AUTOPSY. A STUDY OF CAUSES AND MODES OF DEATH ESTIMATED BEFORE AND AFTER AUTOPSY
S. ASNAES, V. FREDERIKSEN
and C. FENGER
University Institute of Forensic Medicine, Copenhagen, and Holstebro Medical and Pathological Departments, Holstebro (Denmark)
Central Hospital,
(Received March 28,1982) (Revision received June 17, 1982) (Accepted July 9, 1982)
Summary Among 312 consecutive deaths in a Danish Central Hospital autopsy was performed in the pathology department on 266 cases, i.e. 85%. Retrospectively, the underlying causes of death were estimated from the clinical information alone by an experienced clinician and subsequently compared with the autopsy report. The definite cause of death was determined jointly by the clinician and the pathologist. The clinician’s diagnosis was thereby confirmed as incorrect in 18% of the cases if small differences in site and type of malignant tumours were not considered. This is less than in many other investigations, but it is stressed that this could partly be because formal errors in completing the death certificate were avoided, The main causes of death were ischaemic heart disease and neoplasia. Clinical diagnosis of malignant diseases was never found to be erroneous. There was a slight tendency to clinically overestimate ischaemic heart disease, but in general the different errors outweighed each other, so that the total number of different causes of death before and after autopsy was nearly the same. The original death certificate was investigated in 12 accidental cases. Hereby it was found that the mode of death was originally stated erroneously as natural in 7 cases, i.e. 4.5%. It is concluded that hospital autopsy is still needed for the control and correction of causes of death, and it is stressed that clinicians as well as pathologists should be more aware of cases with a trauma in the history to avoid errors in the mode of death. Such errors can imply legal as well as insurance problems. Key words: Hospital autopsy; Errors in causes of death
Introduction Periodically it medical research that the cause of cases if only the
is debated how significant the hospital autopsy is for and treatment [l-3]. Many studies [4-111 have shown death will be stated erroneously in about one third of all clinical information is considered. Errors will occur more
0379-0738/83/0000-0000/$03.00
0 Elsevier Scientific Publishers Ireland Ltd. Printed and Published in Ireland
24
frequently among elderly people and many contributory causes of death will often not be discovered until autopsy is performed [ 121. With a high autopsy rate in a well-defined population the severeness, extension and frequency of many diseases can be determined in detail, and in the epidemiological research the mortality statistics are still a useful aid in elucidating the need of medical assistance, making prognoses and judging the effect of medical assistance [ 131. In 1952 Munck compared the presumed cause of death before autopsy and the result of the autopsy and found 20% of errors. Since that time the diagnostic methods within clinical medicine have changed and been modernized. This could perhaps give a higher degree of security in establishing the correct cause of death. The present retrospective study was therefore performed to elucidate what significance the hospital autopsy has for the correct statement of the cause of death to-day. Materials and methods Materials
266 autopsies (114 women and 152 men) were performed in 1974 in a 530-bed Central Hospital, situated in a county with 250,000 inhabitants. The total number of deaths in the hospital in the same period was 312, i.e. the autopsy rate was 85%. Methods
First the cause of death was estimated from clinical information alone (V.F.). The logical sequence between the different conditions leading to death was secured according to the WHO regulations for underlying causes of death [ 141. The aim was to avoid formal errors, which even if they reflect the uncertainty of our mortality statistics, would not give any information about the value of the autopsy. Secondly the clinically estimated cause of death was compared with the autopsy findings. In each case the clinician and the pathologist jointly estimated whether the autopsy findings could change the clinical cause of death. TABLE
I
AGE IN DECADES
Males Females Total
O-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-
Total
11 12 23
1
1 1 2
12 9 21
17 10 27
48 23 71
45 39 84
14 14 28
3 6 9
152 114 266
1
25
In all cases where deaths were considered to be non-natural a copy of the original death certificate was asked for. Results Eightyone percent of the women and 84% of the men were more than 50 years of age (Table 1). Totally we found 48 differing causes of death (18%), if 12 cases, where a malignant disease was stated wrongly concerning origin and histological type, are not included. In two cases the mode of death was estimated erroneously, i.e. 0.8%. The different causes and modes of death in certain main groups are shown in Table 2. In 136 cases opening of the skull was performed, i.e. 51%. In 146 cases microscopy was performed after the autopsy, i.e. 55%. We found no characteristic relation between the number of differing causes of death and age groups. Besides a slightly pronounced tendency to overdiagnose ischaemic heart disease (IHD) as cause of death among men as compared to women, no characteristic difference in the two sexes was found.
Accidents In 14 cases an accident was suggested as the cause of death. This was correct in 12 cases (5%). Among women the most common accident was stumbling at home and in men it was traffic accidents. One case where the clinician wrongly estimated that death was due to an accident was that of a 74-year-old woman, who was taken to hospital after a trauma against the head. The blood pressure was 230/130 and an atypical plantar reflex was found. An X-ray contrast investigation was performed with suspicion of subdural haematoma, and death occurred. At autopsy a swelling of the brain and scattered haemorrhages in the pons were found. There was no sign of trauma against the skull and the findings were considered to be due to hypertension. By investigating the original death certificates it was found that all traffic accidents were correctly reported to the police, external medico-legal examinations were performed and all cases were correctly registered as accidents. Five cases, where death was caused by stumbling in the home were all reported to the police, but medico-legal examination was not asked for. Despite this, the mode of death was stated as natural in the hospital and later registered as such in the General Register. Two cases of accidental deaths were not reported to the police and stated as natural. One case was that of a 7-year-old boy who had a trauma against the head. Shortly afterwards he developed a headache and a slight fever. He was taken to the local hospital where it was assumed he had got a cold and was sent home. Since then he suffered from an occasional headache, and when he again became febrile he was taken to the hospital where he died
1
412
TOTAL
37
21
2
Others (disagreement)
1
1
12
6
Accidents
Others (agreement)
Pneumonia
Subarachnoid haemorrhage
1
34
410
Cerebral haemorrhage
Stillborn
Pulmonary embolism
Malignant disease
Ischaemic heart disease
63
1
1
60
1
14
3
2
6
2
1
20
20
22
2
18
1
1
4
4
11
2
1
44
44
12
12
CAUSES AND MANNERS OF DEATHS IN 266 HOSPITAL DEATHS ESTIMATED BEFORE AND AFTER AUTOPSY
TABLE 2
18
8
1
1
3
2
1
2
266
12
14
44
13
6
25
20
10
60
17
45
K
27
from a meningococcical meningitis. At autopsy a small fracture in the lamina cribosa was found, and this was assumed to be the entry of the infection. Natural death
There was a total of 254 cases of natural deaths, i.e. 95%. From Table 2 it appears that the cause of death in 62 cases (45 + 17) was reported as IHD, i.e. 23%. This was correct in 46 cases, i.e. 74%. Out of 45 estimated cases of fresh myocardial infarction 34 were correct. The differing causes of death are shown in Table 2. In one case the cause of death was stated as unknown after autopsy. It was a 35-year-old man, who according to his wife had previously developed symptoms of tachycardia, but who was otherwise healthy. He was suddenly taken unconscious and died shortly after arrival in hospital. At autopsy only a slight coronary sclerosis was demonstrated. Among 17 presumed deaths from pure coronary sclerosis 12 were correct. The differing causes of death were fresh myocardial infarction, bronchogenic carcinoma, pulmonary embolism and thrombosis of a mesenterial vein. Sixty cases were correctly stated as a malignant disease (Table 2). The specific type of the tumours was stated erroneously in 12 cases, i.e. 20% (Tables 3, 4 and 5). In six out of 10 cases it was correctly presumed that the cause of death was pulmonary embolism. The differing cases were cerebral haemorrhage, aortic stenosis, myocardial infarction and peritonitis. Cerebral haemorrhage was presumed in 25 cases. Eighteen were correct. The correctly stated cases all concerned patients who had been in the ward for several days. The cliniCal TABLE
3
FORTY-EIGHT
CORRECTLY
STATED
DEATHS
MALIGNANT
DISEASES
Men
Women Cancer coli Adenocarcinoma ovarii Bronchogenic carcinoma Cancer mammae Pancreatic cancer Gastric cancer Renal cancer Leiomyosarcoma Cholangiocarcinoma Carcinoma planocellulare vaginae Reticulosarcoma Chorioepithelioma
(5P (3) (2) (2) (2) (1) (1) (1) (1) (1) (1) (1)
TOTAL aThe figures in parentheses
FROM
(21) indicate
the number
Bronchogenic carcinoma Gastric cancer Pancreatic cancer Malignant ependymoma Epidermoid cancer (on neck) Epidermoid cancer (on tongue) Carcinoma urinaric bladder Colonic carcinoma Hepatoma Cholangiocarcinoma Renal cancer Melanocarcinoma
(14) (2) (2) (1) (1) (1) (1) (1) (1) (1) (1) (1)
TOTAL
(27)
of cases.
28 TABLE
4
DIAGNOSIS IN SIX AUTOPSY (MEN)
DEATHS
FROM
MALIGNANT
DISEASES
BEFORE
Clinical diagnosis
Au topsy
Total
c. occultus
Carcinoid tumor of the rectum Bronchogenic carcinoma Pancreatic cancer Seminoma
1 1 1 1
Bronchogenic carcinoma Cholangiocarcinoma
1 1
Prostatic
cancer
(2)
TOTAL
AND AFTER
6
picture was in all cases rather.characteristic with a hemiparesis, unconsciousness and a high blood pressure. Different causes of death were subarachnoid bleeding, thrombosis of a mesenterial vein and bronchopneumonia. Subarachnoid haemorrhage was stated as cause of death in six cases. One of the differing causes of death was a 64-year-old woman with a headache during many years who died shortly after 2 h of intensive pain. At autopsy a cerebral haemorrhage was found. Bronchopneumonia was overdiagnosed. Among 13 presumed cases only 8 were correct. In the other cases a bronchogenic carcinoma, ulcerative endocarditis with cardiac insufficiency and pulmonary embolism were found. In 44 cases the cause of death was correctly stated (Table 6). In 12 cases other differing causes of death were found. Four cases occurred in women. Two cases were presumed to be bronchitis, but the autopsy revealed TABLE
5
DIAGNOSIS IN SIX AUTOPSY (WOMEN)
DEATHS
FROM
MALIGNANT
DISEASES
BEFORE
Clinical diagnosis
Au topsy
Total
c. occultus
Cholangiocarcinoma Carcinoma of the cervix Alveolar cell carcinoma Adenocarcinoma ovarii
1 1 1 1
Colonic
Adenocarcinoma
1
cancer
Adenocarcinoma TOTAL
ovarii
Pancreatic
cancer
ovarii
1 6
AND AFTER
29
TABLE
6
FORTY-FOUR
CASES OF NOT DIFFERING
CAUSES OF DEATH
Women
Men
Diabetes Hypertensio arterialis Aortic stenosis Diverticulosis coli Choledocholithiasis Arteriosclerosis universalis Bronchitis chronica Peritonitis Cholecystitis Paraoesophageal hernia Polycytaemia Colitis ulcerosa Chronic pyelonephritis Perforated ulcus ventriculi Hyperparathyoidism Essential pulmonal hypertension Sepsis
Chronic bronchitis Pancreatitis acuta Diabetes Adipositas Hyperlipoproteinaemia Rheumatic endocarditis Acute pyelonephritis Amyotrophic lateralsclerosis Thrombosis of mesenterial vein Arterial hypertension Duodenal ulcer with penetration Ischaemic colititis Uraemia after renal infarction v01v01us
TOTAL
TOTAL
(341
a The figures in parentheses indicate the number of cases.
a myocardial infarction and uremia. In one case the clinical findings were similar to hyperparathyroidism, but at autopsy a cirrhotic liver with bleeding oesophageal varices was found. One case of presumed chronic renal insufficiency was mitral stenosis and heart insufficiency. The causes of death among 8 men are shown in Table 7. In one case the cause of death was stated as septicaemia. It was a 74-yearold man who was taken to hospital after suffering at home for a long time TABLE
7
CAUSES OF DEATH BEFORE
AND AFTER
AUTOPSY
Clinical
Au topsy
Chronic bronchitis Chronic bronchitis Chronic bronchitis Esophageal varices Polycytaemia’ Unknown Septicaemia Abdominal tumor
Coronary sclerosis Aortic rupture Brochogenic carcinoma Bleeding ulcer Pancreatitis Meningococcal sepsis Pancreatic cancer Coronary sclerosis
TOTAL
IN EIGHT MEN Number
30
from pneumonia. At first he got a renal insufficiency. 10 days later amputation of the right leg was performed because of gangrene and after a few days death occurred. At autopsy a pancreatic carcinoma with metastases was found as an underlying cause of death. Totally, there were 37 fresh myocardial infarctions. In 14 of these, thrombosis was found, i.e. 35%. Three cases of malignant diseases were discovered, i.e. 1%.
Discussion Erroneous statements for the underlying causes of death before autopsy were 18%. This is rather low as compared to many other studies. Britton for example found a total of 30% of erroneous clinical diagnoses in hospital data from the Serafimer Hospital in Stockholm [6], and Asnaes found that the cause of death was erroneously estimated in about 50% before autopsy in non-medico-legal material from Copenhagen [15] . One reason for the discrepancy could be that we paid a lot of attention in discussing the correct and logic sequence of the different causes of death, whereby formal errors were avoided. Such formal errors can, as demonstrated by Asnaes et al., [16] and Skullerud and Skullzrud [17], give up to 20% wrongly estimated causes of death. If only errors, where the revised diagnosis belongs to different main groups of diseases within the ICD classification of the ‘list of three digit categories’, are counted, a total of 27 erroneously stated underlying causes of death exists, i.e. 10%. If the cases of differing modes of death are eliminated too, the cause of death is found differing in 9%. This comes very close to Britton’s investigation, where she found different causes of death in main groups in 7% [6]. The approach of this study concerned the estimation of the causes of death. However, it was also found that the mode of death was differing in 0.8% of the cases. This is only a theoretical problem in this retrospective study. It is more important that by investigating the original death certificate we found that the mode of death was originally stated wrongly as natural death in seven accidents, i.e. 5%. This is in accordance with the investigations of Otterland and Pihl [ 181, who found accidents to be under-recorded in the official statistics. Otterland and Pihl mention cases of fractures which, clinically, were classified as senile osteoporosis, without any mentioning of a previous trauma. While errors in the cause of death can influence mortality statistics, epidemiological investigations, etc., it is more serious if the mode of death is stated wrongly. Such errors can imply legal as well as insurance problems. Ischaemic heart disease was clinically overdiagnosed, but not so pronounced as in the medico-legal investigations, where ischaemic heart disease is overdiagnosed in about 30% of all cases [ 191.
31
In the present study fresh myocardial infarction was overdiagnosed to the disregard of chronic ischaemic heart oisease. From a clinical point of view overdiagnosis of myocardial infarction must be to the benefit of the patients (coronary care units, etc.). Among infarctions, coronary thrombosis was found in 35% of the cases. This is not-far from Asnaes’ and Qstergaard’s study [ 201, where eight thromboses among 16 recent infarctions were found in patients who died after a very short treatment in hospital. In contrast to this Asnaes [ 151 in a material of sudden unexpected deaths found 11 cases of coronary thrombosis, but only eight myocardial infarctions. The difference could be due to the fact that in the study of Asnaes and pstergaard the formazan test was used. Another explanation could be that the hospital treatment, even short, could give the myocardium the possibility of developing an infarction. Pulmonary embolism was slightly underestimated, pointing out that this is still a difficult clinical diagnosis. Cerebral haemorrhage and pneumonia were slightly overestimated. This is in contrast to similar medico-legal investigations [ 191. It must be considered, however, that the brain was only investigated in about 50% of all cases. Clinical diagnosis of neoplastic disorders were never found to be erroneous, but in 20% of the cases the histological type and the origin of the different tumours were first found at autopsy. Three cases of malignant disease were discovered, i.e. 1%. This is less than that found in similar medico legal investigations [ 191, where up to 4.5% malignant diseases were discovered. But it is not surprising that malignant disease is discovered more seldom among hospital deaths. Many of the diseases, where the clinical diagnosis was wrong, are wellknown as offering particular diagnostic difficulty (endocarditis, pancreatic cancer, ruptured aortic aneurysm). The aim of the present study was to investigate whether autopsies are important in determining theunderlying cause of death. We found the answer obviously positive, but in this study, as in others, it is characteristic that clinical overdiagnosis and underdiagnosis, to some degree, outweigh one another. In the present study the total number of different causes of death was nearly the same before and after autopsy. It is unsatisfactory, however, to know that a wrong cause of death would be corrected in another fault if autopsy is not performed. Besides a certain number of contributory causes of death would be discovered, whereby some clinical abnormalities could be explained. It is therefore concluded that hospital autopsies are still needed with a view to control and correction of causes of death. The autopsies mean a continuous control of diagnostic errors, but it is also important, when completing the death certificate, that the clinician as well as the pathologist are aware of whether the death is natural or due to an accident. A closer cooperation between the pathologist and the clinician in com-
32
pleting the death certificate help.
as well as more central control could possibly
References 1 2 3 4 5 6 7 8 9 10
11 12 13 14 15
16 17 18 19 20
J.T. Freeman, The Autopsy.Pa. Med., 71 (1968-1971). What about the autopsyEditorial, 193 (1965) 172. A. Juul and J. Andersen, Rutineautopsier. Ugeskr. Laeg., 139 (1977) 323. S. Asnaes and F. Paaske, The significance of medicolegal autopsy in determining mode and cause of death. Forensic Sci. Znt., 14 (1979) 23. W. Borris, Die Bedeutung der Vervaltungssektion fiir die Morbiditats- und Mortalitiitsstatistik. 2. Arztl. Fortbild., 46 (1952) 471. M. Britton, Diagnostic errors discovered at autopsy. Acta Med. &and., 196 (1974) 203. R.H. Gruver and E.D. Freis, A study of diagnostic errors. Ann. Intern. Med., 47 (1957) 108. M. Heasman and L. Lipworth, Accuracy of Certification of Cause of Death. Her Majesty’s Stationery Office, London, 1966. W. Munck, Autopsy finding and clinical diagnosis. A comparative study of 1000 Cases. Acta Med. Stand. Suppl. 226 (1952) 775. N. Schulz and Schaarschmidt, Zur Relation zwischen klinischen und pathologischanatomischen Diagnosen in der Todesursachenstatistik. Zentralbl. Gesamte. Hyg., 16 (1970) 928. I. Wallgren, Obduktionsfyndet och de kliniska diagnoserna. Nord. Med., 26 (1945) 1311. M. Britton, Clinical diagnostics: Experience from 383 autopsied cases. Acta Med. Stand., 196 (1974) 211. K. Juel, Autopsihyppighedens betydning for d&ls%rsagsstatistikken. Ugeskr. Laeg., 143 (1981) 2669. WHO, International classification of diseases, 1965 Revision WHO, Geneva, 1967. S. Asnaes, Uncertainty of determining mode and cause of death without autopsy: An autopsy study of medically unattended non-medicolegal deaths. Forensic Sci. Znt., 15 (1980) 191. S. Asnaes, C.E. Mabeck and B. Wichmann, The significance of autopsy for statistice concerning causes of death. An investigation of death certificates. Ugeskr. Laeg., 142 (1980) 261. K. Skullerud and J. Skullerud, Mortality Statistics. Tds. Skr. Nor. Laegeforen., 98 (1978) 1748. A. Otterland and E. Pihl, Klinisk, patologisk-anatomisk og officiel d&is%rsagsstatistik med udgangspunkt i 327 cbduktioner. Lakartidningen, 61 (1964) 68. S. Asnaes and F. Paaske, Uncertainty of determining cause of death in a medicolegal material without autopsy. An autopsy study. Forensic Sci. Int., 15 (1980) 103. S. Asnaes and K. Qtergaard, Reliability of death certificates. An autopsy investigation. Ugeskr. Laeg., 142 (1980) 265.