Special Article: Autopsy
Austin J Surg. 2023; 10(4): 1312.
Insights of Validating Death Certificates through Clinical Autopsies
Schubert D1,3; Christ H2; Fries JWU3
1Medical Training Center, Department of Medicine, University of Bielefeld, Bielefeld, Germany
2Institute for Medical Statistic, Informatic and Epidemiology (IMSIE), University of Koeln, Koeln, Germany 3 Institute of Pathology, University Hospital of Koeln, Koeln, Germany
*Corresponding author: Jochen W.U. Fries Professor of Pathology, Institute of Pathology, University Hospital of Koeln, Koeln, Germany. Tel: +49223858840; Fax: +492214786360 Email: Jochen.fries@uni-koeln.de
Received: September 11, 2023 Accepted: October 11, 2023 Published: October 18, 2023
Abstract
Background: With decreasing autopsy numbers, the cause of death listed in the death certificate is only based on an external inspection, which is increasingly criticized as insufficient. This investigation analyses 447 death certificates from the archives of the University Hospital of Koeln, Germany and their corresponding autopsy reports from the Institute of Pathology.
Methods: For this analysis, quality assurance protocols, death certificates, clinical/autopsy notes, and final autopsy reports were used. Cases were categorized: 1. in four validation classes (Goldman criteria, identifying unknown diseases with therapeutic relevance as cause of death); 2) in four nosological causal chains (WHO ICD10; type: linear/ divergent/ convergent/ complex).
Results: Differences in diagnoses between death certificate and autopsy report occurred in 32% (143/447 cases). In 7%, only the autopsy identified the cause of death (Goldman, type 2a). Nosological causal chains were established in 21% (linear/divergent) vs 28% (convergent/complex). Myocardial infarct, septicemia and cardiac insufficiency caused death in more than two thirds of cases. Diabetes and obesity did not play a major role as cause of death.
Conclusion: Autopsies are highly advisable if death occurs within 48 hrs. of admission and as sudden death in the hospital setting. Regular interdisciplinary autopsy conferences are important for quality control, assessing cases of the convergent/ complex type. The position of an autopsy commissioner as mediator between relatives, clinicians and pathologists seems recommendable in a hospital setting. While an electronic patient file is still controversial, medical data collection as source of information in emergencies by the patient’s medical practitioner seems advantageous.
Keywords: Death certificate; Autopsy report; Goldman criteria; Nosological causal chains; Autopsy commissioner; Electronic patient file.
Introduction
In Germany, about one million people die per year. Board certified medical doctors are oblieged by law to prepare death certificates. Basic diseases, their consequences and a resulting death occuring from a natural event have to be certificed. The massive development in medical knowledge and procedures over the last 30 years and procedures as well as the change in population structure (age increase, ethnicity) has increased the complexity of daily medical practice. In addition, lack of knowledge of the previous medical history of the patient and his preexisting diseases and comorbidites may lead to wrong conclusions. Thus not unexpectedly, during the last decade, the quality of the death certificate and its relation to an autopsy (so called internal body examination) has been increasingly criticized.
To recognize potential problems, the Health Ministry of North Rhine-Westfalia, Germany has suggested to study autopsy-based death certificates versus their respective autopsy reports. Previous analyses focused on particular patient collectives like newborns [1] liver transplantation [2,3], posttraumatic events [19] or intensive care patients [5-7]. An analysis in a community-hospital [8] revealed that a complete agreement between death certificate and autopsy report was only found in 42%. However, a study in a university clinic setting as hospital with all currently available surgical and medical devices regarding the agreement or disagreement between death certificate and autopsy report and its reasons based on nosological causal chains has not been done. We performed this study to analyse the reliability on the diagnoses in a death certificate, the need for autopsies based on the recognition of clinical relations, show problematic clinial scenarios and propose ways to improve potential pitfalls.
Material and Methods
Death certificates were validated using clinical autopsies by a team (a board certified anesthesiologist/intensive care specialist (DS) and a board certified pathologist being 20 years head of the autopsy service in the Pathology department of a university hospital). Reports were used from 1420 autopsies performed from 2005 to 2014 at the Institute of Pathology, University hospital of Koeln, Koeln, Germany, one of the largest university hospitals in Germany. Only 447 could be analyzed further due to incomplete/missing clinical information and/or death certificates. Potential diagnostic insights were standardized using case-related quality assurance protocols built on death certificates; applications for autopsy; notes from case discussion between clinicians and pathologists; the autopsy protocol itself and the final autopsy report. This analysis was done in two parts.
Part 1:
Cases were categorized into one of the 4 valuation classes [9] with respect of their role in the cause of death and their diagnoses comparable with those during the life of the patient.
• Class I error: Misdiagnosis that may have affected survival and probably would have required alteration tretatment.
• Class II error: Misdiagnosis that did not affect survival and would not have required alteration of treatment
• Class III and class IV errors: Missed minor diagnoses unrelated to the disease course
Part 2:
Advances in medical knowledge and technology, and the degree of multimorbidity makes the analysis of the cause of death considerably harder than before. As recommended by the WHO, we used the new ICD10 (International Statistical Classification of Diseases and Related Health Problems), where death should be considered the result of a multiple cause chain instead of a single event. In ICD 10, four nosological chains causing death should be used [10,11].
1. Linear type: the classical, monocausal chain (i.e. aspiration – pneumonia – death)
2. Divergent type: a chain whose basic disease is caused/characterized by two different disease processes (i.e. diabetes: macro- / microangiopathy; nerve paralysis and “diabetic foot” etc.).
3. Convergent type: a causal chain with a common final denominator (e.g. hypertension/diabetes/fatty acid disturbance: all causing microangiopathy leading potentially to cardiac infarction).
4. Complex type: several parallel existing diseases with an independent and equally strong contribution leading to death (e.g. inflammatory bowel disease and prostatic abscess both leading to septicemia).
5. All cases were subjected to nosological causal chain evaluation.
The following statistical analyses were performed:
i. Quality assurance protocols were established quantifying the diagnostic gain using autopsies by differentiating the relationship between basic/essential versus secondary illnesses
ii. Frequency distribution of valuation classes (according to Goldman; methods, part 1)
iii. Frequency of nosological causal chains
Results
Patient Data
We analyzed death certificates and autopsy reports from 447 patients, 192 men (43%) and 255 women (57%) (age distribution see Table 1). Patients came from either the university clinic (353; “inpatients”) or as outpatients/emergency cases (94) (see Table 2). The majority in both groups constituted patients from cardiology (inpatients: 44.4 vs emergency patients: 50%) and the GI department (inpatients: 22.3%). General surgery patients were dominant as emergency patients (41.5% vs. 12.1% as inpatients); patients from oncology were the next largest group for inpatients (14.7%). A more detailed differentiation was not possible in the emergency cases.
Age
Number of patients
in %
males
in %
females
in %
M : F
18 – 35
21
4.7
8
3.1
12
6.3
1.8 : 2.7
36 - 50
75
16.8
44
17.2
29
15.0
9.8 : 6.5
51 - 70
212
47.4
133
52.2
80
41.7
29.8 : 17.9
71 - 99
139
31.1
70
27.5
71
37.0
31.3 : 31.7
255
100.0
192
100.0
447 = 100.0
Gender
M
192
43.0
F
255
57.0
447
100.0
Table 1: Age and gender distribution.
Primary patient from
university clinic (“inpatients”)Patient transfered as
emergency case (“outpatients”)Medical discpline/Institute
total #
in %
total #
in %
Kardiology
157
44,4
47
50.0
Gastroenterology
79
22.3
Oncology
53
14.7
Gerneral surgery
43
12.1
39
41.5
Vascular surgery
4
4.2
Orthopedics
3
0.8
Other surgical disciplines
6
1.7
Neurology
12
3.4
4
4.2
Total number of all cases
353
100
94
100
Table 2: Patient recruitment.
Emergency Diagnoses
Patients were delivered as emergency cases from their homes/outside hospitals with a variety of serious different symptom / suspected diseases, of which circulatory arrest and myocardial infarct were the most frequent ones (representing more than 20% of all cases) followed by dyspnea, thorax pain and arrhythmias (Table 3).
Emergency diagnoses
sum #
in percent
circulation arrest of unknown cause
30
22.2
myocardial infarct
28
20.7
dyspnea
17
12.6
thorax pain
16
11.9
arrhythmia
12
8.9
neurologic deficits/syncope
12
8.9
hypotension
8
5.9
others
12
8.9
total
100
Table 3: Emergency diagnoses.
The Compilation of Diseases from Inpatients
Cardiac diseases such as myocardial infarct (41.8%) and cardiac insufficiency/cardiomyopathy (21.2) represent about two thirds of all cases, followed by liver (39.3%), renal diseases (22.6) and gastrointestinal bleedings (4.4%) (Table 4). There were no significant gender difference between respective major diseases.
Diseases of various organ systems
Sum #
in percent
myocardial infarct
148
22.4
liver failure/- diseases
139
21.1
renal failure/ - diseases
80
12.2
cardiac insufficiency/ cardiomyopathy
75
11.4
gastrointestinal bleeding
51
7.8
pulmonary edema/ lung bleeding
42
6.6
diverticulosis/ diverticulitis
38
5.7
pancreatitis
37
5.5
metastases
27
4.1
Others
21
3.1
all organ systems
658
100
Table 4: Diseases of various organs.
Major Basic Diseases
Major basic diseases from autopsy reports were arterial hypertension, pulmonary emphysema well as other cardiac diseases (Table 5) Oncologic diseases/immunosuppression with 8.7 percent were less prominent.
Diseases confirmed by autopsy (multireferencing)
sum #
in percent
arterial hypertension
216
14.9
pulmonary emphysema
190
13.1
arteriosclerosis
130
8.9
Coronary heart disease
117
8.0
Cor pulmonale
104
7.2
cardiac insufficiency/cardiomyopathy
85
5.8
myo-/endo-/pericarditis
75
5.2
pneumonia
74
5.1
other inflammatory diseases
162
11.1
leukema/immunosuppression
126
8.7
others
175
12.0
total # of diseases
1454
100.0
Table 5: Diseases confirmed by autopsy (multireferencing).
There were no significant gender difference between respective major disease; arterial hypertension and pulmonary emphysema being slightly more frequent in female ´than in male patients (Table 6).
major basic diseases
male patients
female patients
sum #
in percent
sum #
in percent
arterial hypertension
138
21.7
113
23.5
pulmonary emphysema
131
20.5
110
22.9
arteriosclerosis
108
17.0
71
14.8
coronary heart disease
84
13.2
60
12.5
cardiac insufficiency
110
17.3
81
16.9
Oncologic disease
66
10.3
45
9.4
Total # of diseases
637
100
480
100
Table 6: Major basic diseases in males/females (multireferencing).
Nosological Causal Chains
Cases categorized in nosological causal chains show an equal distribution for the linear/ divergent type in 21% versus 28.5% for the convergent/ complex type (Table 7).
Nosological causal chain
# of patients
In %
Type 1: linear
95
21.2
Type 2: divergent
97
21.7
Type 3: convergent
128
28.6
Type 4: complex
127
28.5
Total #
447
100
Table 7: Nosological causal chains.
Cause of Death
Causes of death were evaluated as described in autopsy reports. Among the five most frequent causes we found cardiac/circulatory failure (29.8%), septicemia (21.6%), myocardial infarct (18.8%), respiratory failure (5.9%), and hemorrhagic shock/aortic dissection (4.2%). Others were less frequent (Table 8).
Cause of death according to autopsy reports (inpatients)
In percent
cardiac/circulatory failure
29.8
septicemia
21.6
myocardial infarct
18.8
respiratory failure
5.9
hemorrhagic shock/aortic dissection
4.2
left heart failure
4.0
right heart failure
2.6
lung arterial embolism
3.6
gastrointestinal bleeding
2.8
myo-/ endo-/ pericarditis
1.8
pneumonia
1.5
others
3.8
Total number of cases: 354
100.0
Table 8: Causes of death according to autopsy reports.
According to age of death (Table 9), we observed in the young adult (age 18 – 35), death due to bleeding was the most prominent followed by cardiac failure. With increasing age, myocardial infarction became the most dominant cause of death (34% aged group 51 – 70; 32.7% in age group 71 – 99). The second most drominent cause was septicemia, even outnumbering myocardial infarction particularly in the 36 – 50year old patients.
18 - 35
in %
36 - 50
in %
51 - 70
in %
71 - 99
In %
myocardial infarct
3
14.3
19
22.3
60
34.1
54
32.7
septicemia
3
14.3
36
42.4
39
22.2
38
23.0
cardiac failure
5
23.8
17
20.0
28
15.9
26
15.8
hemorrhagic shock/hypotension
6
28.6
8
9.4
22
12.5
21
12.7
respiratory failure
4
19
5
5.9
27
15.3
26
15.8
total # of cases: 447
21
100.0
85
100
176
100
165
100
Table 9: Causes of death (primary inpatients and outpatients/emergencies from outside).
Significant differences in the frequency of basic diseases (arterial hypertension, emphysema, arteriosclerosis, cardiac insufficiency, coronary heart disease, oncologic diseases, obesity, diabetes) as potential forerunners of the 5 most prominent causes of death were identified (Table 8) at different age groups (Figure 1). For inpatients, in 70% a heart/circulation failure, septicemia or a myocardial infarct was the cause of death. For myocardial infarct as cause of death, arterial hypertension, arteriosclerosis, cardia insufficiency and the degree of pulmonary emphysema showed an increasing importance over age (Figure 1a). Similarly, the same factors affected the death of cardiac failure (Figure 1b). In respiratory failure, the amount of pulmonary emphysema was most prevalent with a stepwise increase from 36 to 99 years of age. In the youngest age group respiratory failure was predominantly caused by pneumonia (Figure 1c). Death due to septicemia was influenced by arterial hypertension and arteriosclerosis (Figure 1d); the pulmonary function also playing a major role. Oncologic diseases were dominating in the age group 36 – 50 years as forerunners of septicemia. In later age groups, oncologic diseases play only a minor role, while arteriosclerosis is rising. All other disease entities were of minor importance (less than 10%). In case of bleeding (Figure 1e) the arterial pressure and the degree of arteriosclerosis were most important (>20%). Oncologic diseases played a prominent role in pulmonary failure and in septicemia in the younger age groups. In contrast, obesity and diabetes were only of minor importance. In summary, with increasing age, myocardial infarct, respiratory failure and septicemia are rising and comprise about two thirds of all deaths.
Figure 1:
Discrepancy between Causes of Death in Death Certificates vs. Autopsy Reports
The analysis between the cause of death on a death certificate versus the respective information on the autopsy report showed an agreement in 68% overall (304/447 cases). In 32%, however, the diagnoses between the two documents did not match (143/447) (Table 10) Myocardial infarct (36/143; 25.1%) and septicemia (28/143; 19.6%) were the most commonly misdiagnosed, occurring concurrently in 3.5% (5/143). At third place, hemorrhagic/hypovolemic shock due to bleeding (19/143; 13.3%) was misinterpreted. Pulmonary embolism was unrecognized in 6.3% (9/143). In 10/143 misdiagnosed cases (7%), the autopsy was essential in detecting the cause of death; the death certificates listed here as cause unknown/unclear/none.
Case #
cause of death by autopsy
cause of death by death certificate
1
aortic dissection
cardiac/circulatory failure, septicemia
arrythmia
none
brain stem compression
respiratory insuffciency
cardiac insufficiency
myocarditis, pulmonary embolism, myocardial infarct
5
cardiac insufficiency
septicemia
cardiac metatsases, failure
pneumonia, lung metatsases
cardiac tamponade
pulmonary embolism
cardiac/circulatory arrest/ arrhythmia
septicemia
cardiac/circulatory arrest / arrhythmia
unclear
10
cardiac/circulatory arrest / arrhythmia
unclear
cardiac/circulatory arrest/ arrhythmia
myocardial infarct
cardiac/circulatory arrest/ arrhythmia
respiratory insufficiency
cardiac/circulatory arrest / arrhythmia
thyreotxoic crisis
cardiac/circulatory arrest / arrhythmia
aspiration, septicemia
15
cardiac/circulatory arrest / arrhythmia
myocardial infarct
cardiac/circulatory arrest/ arrhythmia, intracerebral bleeding
septicemia, intestinal ischemia
circulatory failure
septicemia
endocarditis
pulmonary edema
endocarditis
right heart failure
20
endocarditis
electromechanical decoupling, cardiac failure
endocarditis, moycardial infart
unknown
fungal septicemia
meningitis
gastrointestinal bleeding, hypovolemic shock
liver failure, multiple organ failure
gastrointestinal bleeding, hypovolemic shock
septicemia, circulatory failure
25
gastrointestinal bleeding
multiple organ failure
hemorhagic shock, respiratory insufficiency
cardiac/circulatory arrest/ arrhythmia, septicemia
hemorrhagic shock
pneumonia, right heart failure
hemorrhagic shock
liver failure
hemorrhagic shock
electromechanical decoupling
30
hemorrhagic shock
aspiration
hemorrhagic shock
septicemia, pneumonia
hemorrhagic shock
respiratory failure
Hemorrhagic shock
septicemia
hemorrhagic shock
myocardial infarct
35
hemorrhagic shock, aortic rupture
pulmonary embolism, cardiac failure
hemorrhagic shock, myocardial infarct
electromechanical decoupling
hemorrhagic shock, myocardial infarct
electromechanical decoupling
hemorrhagic shock, myocardial infarct
septicemia
hemorrhagic shock, myocardial infarct
liver failure
40
hemorrhagic shock, myocardial infarct
septicemia
hemorrhagic shock, myocardial infract
cardiogenic shock
hypovolemic shock
arrythmia, upper gastrointestinal bleeding
hypovolemic shock/ myocardial infarct
septicemia
hypovolemic sock
septicemia
45
left heart failure
intestinal ischemia, liver failure
left heart failure
unknown
left heart failure
septicemia, multiple organ failure
left heart failure
anphylaxia, unclear
left heart insufficiency
septicemia, multiple organ failure
50
liver failure
aspiration
liver failure, septicemia
brain edema, bleeding
metabolic /toxic circulatory failure
paralytic ileus
metabolic/ toxic circulatory failure
unclear
metabolic/ toxic circulatroy failure
lymphoma
55
myocardial infacrt
cardiac/circulatory arrest caused by bleeding
myocardial infacrt
right heart failure
myocardial infarct
cardiopulmonal decompensation
myocardial infarct
septicemia
myocardial infarct
pulmonary embolism
60
myocardial infarct
intestinal ischemia
myocardial infarct
unknown, multiple organ failure
myocardial infarct
septicemia, multiple organ failure
myocardial infarct
right heart failure
myocardial infarct
pulmonary embolism, aortic rupture
65
myocardial infarct
endocarditis
myocardial infarct
right heart failure
myocardial infarct
cardiac failure, electromechanical decoupling
myocardial infarct
septicemia, acute abdomen
myocardial infarct
septicemia, cardiac shock
70
myocardial infarct
septicemia, pneumonia
myocardial infarct
left heart failure
myocardial infarct
pulmonary embolism, cardiac/circulatory arrest / arrhythmia
myocardial infarct
pulmonary embolism, electromechanical decoupling
myocardial infarct
hypotonia, metabolic acidosis
75
myocardial infarct
septicemia
myocardial infarct
cardiac failure , cardiac insufficiency
myocardial infarct
electromechanical decoupling
myocardial infarct
septicemia
myocardial infarct
hemorrhagic shock
80
myocardial infarct
pneumonia
myocardial infarct
hemorrhagic shock
myocardial infarct
myocarditis
myocardial infarct
cardiac/circulatory failure
myocardial infarct
cardiac/circulatory arrest / arrhythmia
85
myocardial infarct
septicemia, rhabdomyolysis
myocardial infarct
cardiac/circulatory arrest /arrhythmia
myocardial infarct
septicemia
myocardial infarct, aortic dissection, hemorrhagic shock
septicemia
myocardial infarct, left heart failure
septicemia
90
myocardial infarct, myocardial bleeding
none
myocarditis
unclear
pericardial tamponade
pericardial effusion
peritonitis, septicemia
liver failure, postop cardiac shock
pneumonia
electromechanical decoupling
95
pneumonia
pulmonary bleeding, hypoxia
pneumonia
electromechanical decoupling
pneumonia
intracerebral bleeding
pulmonary embolism
cardiac shock
pulmonary embolism
myocardial infarct
100
pulmonary embolism
electromechanical decoupling
pulmonary embolism
endocarditis
pulmonary embolism
unclear
pulmonary embolism
intracerebral bleeding
pulmonary embolism
intestinal ischemia, acute abdomen
105
pulmonary embolism
cardiac failure
pulmonary embolism
herniation
respiratory insufficiency
pneumonia
respiratory insufficiency
none
respiratory insufficiency, stroke
pulmonary embolism
110
respiratory insuficieny, right heart failure
brain edema, encephalitis
right heart failure
myocardial infarct
septicemia
intestinal ischemia, disseminated intravascular coagulation
septicemia
pneumonia
septicemia
aspiration
115
septicemia
aspiration, ileus
septicemia
multiple organ failure
septicemia
right heart insufficiency, liver failure
septicemia
respiratory insufficiency
septicemia
left heart insufficiency
120
septicemia
respiratory insufficiency
septicemia
pneumonia, aspiration
septicemia
intestinal perforation
septicemia
bilateral pulmonary embolism
septicemia
herniation, brain edema
125
septicemia
electromechanical decouling, respiratory insufficiency
septicemia
sudden cardiac death, arrythmia
septicemia
intestinal ischemia
septicemia
respiratory insufficinecy
septicemia
respiratory insufficiency, acute respiratory distress syndrome
130
septicemia
liver failure
septicemia
gastrointestinal bleeding
septicemia
cardiac failure
septicemia
hemorrhagic shock, cardiac failure
septicemia
amyloidosis
135
septicemia, myocardial infarct
respiratory insufficiency, aspiration
septicemia, myocardial infarct
pulmonary embolism
septicemia, myocardial infarct
septicemia
septicemia, myocardial infarct
septicemia
septicmeia, myocardial infarct
cardiac/circulatory arrest /arrhythmia
140
sudden cardiac death
septicemia, intestinal ischemia
thrombophlebitis
paralytic ileus
toxic cardiac/circulatory failure
septicemia
toxic cardiac/circulatory failure, respiratory failure
cardiac failure
Table 10: Discrepancy of cause of death between autopsy report and death certificate.
Valuation by Goldman Criteria
Our data show a class I error in 32% (Table 10). These cases show the same frequency distribution in type 1 and 2 nosological causal chains. The frequency of the convergent type is with 34.3% slightly higher; the type 4 is unexpectedly slightly less frequent (23.9%) compared to the overall number of cases (28.5%) (Table 11 vs.7).
Discussion
The analysis of 447 death certificates plus autopsy reports showed that cases belonged in 21% to type 1 and 2 nosological causal chains and in about 28% to type 3 and 4. We found an overall discrepancy in the cause of death between death certificate and autopsy report in 32%. We experienced difficulties to posthumously differentiate in major errors (clinically missed diagnoses involving a primary cause of death) versus class I errors (likely to have affected patient outcome), because several cases being part of either category. Thus, we combined these cases as class I errors (misdiagnosis that may have affected survival and probably would have required alteration of treatment) [12]. Wiitekind and Gradistanac (2018) have reported frequencies of such errors ranging from 2.4 to 19%. Shojania et al. (2003) analysed 53 autopsy-series and calculated 23.5% (range, 4.1%-49.8%) for major errors and 9.0% (range, 0%-20.7%) for class I errors, being in the same total range of 32% of our study.
Decreasing Autopsy Numbers and Its Possible Causes
The concern raised by the North Rhine Westfalia Ministry of Health regarding the major discrepancy between cause of death in death certificates and in autopsy reports seems to be justified not only for regional hospital [8] But also at the university level (discrepancy 32%). This problem is most likely caused by the dwindling numbers of autopsy performed at the university hospital, where a decline of autopsy numbers of 58% is reported from 1993 to 2014 [14]. The problem is, however, widespread: already in 2008, Tavora et al. reported from the United States a major discrepancy involving the cause of death in 17.2% (50 cases) analyzing autopsy records from three institutional settings. Various other publications over the past decades have pointed to decreasing autopsy numbers and the subsequent lack of quality control of clinical measure [13,16,17,19]. This decrease has been made responsible for the substantial number of incorrect diagnoses on death certificates [18]. Winter et al (2012) reported up to 40.500 cases of misdiagnoses per year particularly in the Intensive Care Units (ICUs). Even in the era of “high-tech” medicine, autopsies have detected previously unknown causes of death [19]. In the literature, this discrepancy occurred in 8.4 to 24.4% [13], while from some university hospitals still relatively high autopsy numbers are reported (median 6.1%; mean 2.4%; Nemetz et al., 2006). This is not the case in the vast majority of non-academic hospitals where only a few or no autopsies at all are performed. This decrease is regarded as a negative effect for medical education and quality of care in a yet unprecedented scale [21]. This may already be reflected in the study of Betz et al (2008) from the Beth Israel Hospital in Boston. They conducted an anonymous Web-based test, questioning MDs regarding prior training and experience in completion of death certificates. In contrast to the clinicians’ high level of confidence in their ability to complete a death certificate accurately, only 52% filled out the death certificate with the correct primary cause of death and 10% listed the three major contributing causes of death. However, the use of autopsies for educational purposes is evaluated with regional differences: residents in pathology and internal medicine working at the Massachusetts General Hospital being associated with the Harvard Medical School as teaching hospital positively reflected the value of autopsies for educational purposes, to answer open clinical questions and as source for scientific research [21].
What are the Measures Being Recommended/Taken to prevent this Trend?
In 2017, the German Society of Pathology [14] renewed its guidelines recommending scenarios in which an autopsy should be considered. The major guidelines concerning the recommendation for autopsies in cases of death of adults with regard to our study in a university hospital are the following four:
i. Unexpected death in a hospital setting/ death from non-plausible causes
ii. Death occurring within 48hrs after hospitalization
iii. Post-/perioperative death following invasive procedures (ev. after release by prosecuting attorney’s office)
iv. death in the context of a suggested occupational disease
In the following paragraph, we will briefly discuss relevant cases to highlight these four recommendation and name potential problems.
Unexpected Death in a Hospital Setting/ Death from Non-Plausible Causes
Case 1: One of the most problematic causes of death is a pulmonary embolism [12]. We could observe this scenario in a variety of oncology patients with different tumors such as vein thrombosis following bladder carcinoma; prostate cancer with massive thrombi formation in the periprostatic veins, pancreatic and prostate cancer causing ascending leg vein thrombosis, all of which were followed by a fatal pulmonary embolism. In almost all cases the cause of death (before autopsy) given on the death certificate was myocardial infarct.
Case 2: A patient was diagnosed with a Coronavirus infection during his stay in the hospital but seemed otherwise unaffected in the previous afternoon. In the evening hours, he got up, collapsed and fell, breaking his cheek bone. He was found the next morning, successfully operated but experienced a progressive septicemia leading to his death about 24 hrs. after the incident. Since these events occurred in a hospital setting, he was autopsied by a pathologist, revealing as signs of death small peripheral lung emboli, bleeding and septicemia fitting the diagnosis of a Corona virus induced septicemia.
Death Occurring within 48hrs after Hospitalization
In case of a sudden death within 48 hours of delivery into a hospital an autopsy is highly recommended, particularly if this happens on a weekend. In most cases, any laboratory investigation will not yet be completed; radiologic investigations are not yet performed. The cause of death is solely depending on the limited information, which accompanied the patient (e.g. reports, patent file). Since death occurs so suddenly, relatives are often concerned that not everything possible was done for the patient, which admittedly may be true regarding the insufficient time for any investigation. Thus, an autopsy may help the relatives to understand the course of the disease and the final demise at this time.
As exemplary case we present a 30 year old male who has been imprisoned in Eriwan, at the time part of the Russian federation, where he got infected with tuberculosis. He was treated with a tuberculostatic monotherapy only. After his release, he traveled to Rome, from there to Paris and finally by train to Cologne, Germany, where he collapsed at the train station. Hospitalized as emergency in the university hospital, a renal insufficiency was noted. The patient died less than 48 hrs. after admission. At autopsy, a massive tuberculoid necrosis of his left psoas muscle extending to the inguinal ligament was found. No other site or lymph nodes were positive for Tbc.
Both kidneys showed histologically and immune histologically an AA amyloidosis as cause of his renal insufficiency. The autopsy result was cause of great concern to the respective Italian/French and German authorities; however, no infectious spread was ever reported.
Post-/Perioperative Death Following Invasive Procedures (Ev. After Release by Prosecuting Attorney’s Office)
Patent’s death following an intervention is often expected to be a case for forensic pathology; occasionally, the prosecuting attorney’s office will release the body for autopsy at the Pathology department. However, in our experience, most often the underlying cause is not negligence but rather an unknown or prior undetectable anatomic malformation. In one of those cases, the catheterization of the right coronary artery lead to a dissection with a major intramural myocardial bleed which got infected leading to a combination of infarct plus myocarditis affecting the atrium and the major parts of the right ventricle. This condition ultimately caused myocardial death. The dissection was caused by an extremely narrow opening into the right coronary artery from the aorta.
In TAVI procedures, the success can be hampered by the high degree of valve calcification of the aortic cusps as well as complications following an unrecognized endocarditis [23].
Death in the Context of a Suggested Occupational Disease
Occupational diseases such as pneumoconiosis can be autopsied by request from the relatives, since compensations are in order if the pulmonary disease played a substantial role in causing death. However, there is often a difference in the clinical appearance (lung function) and the detectable degree of structural damage. Occasionally, the development of tuberculosis, or even a pleural mesothelioma are suspected, which are considered a major cause of deterioration of the occupational disease. However, an autopsy is mandatory since such a course cannot be considered on clinical/radiologic means only. Mesotheliomas are potentially difficult to diagnose due to the mantel-like fibrosis encasing the lung lobe. However, more often than not the patient has another underlying disease: malignant neoplasms of the lung as well as other organs (skin, pancreas, prostate or kidney) can mimic pleural mesothelioma and if so, are defined as ’pseudo-mesothelioma’. Saleh et al. (2021) described in a study of 8 autopsy cases for occupational disease 3 cases of malignancy due to pulmonary adenocarcinoma, and one of squamous cell carcinoma originating from the diseased lung. All of which are not recognized as pathophysiologic disease related to a compensated occupational hazard. A bizarre case even revealed a papillary renal cell carcinoma metastasizing into the pleural cavity.
In spite of these recommendations, there seems to be a lack of willingness to ask for and perform an autopsy. This has several reasons.
Autopsy Consent Depends on Patient’s Relatives and Attitude of the Treating Physicians
Autopsy consent is multifactorial [20] and depends on patient’s relatives as well as the treating physicians in the case. The communication between the relatives and the former treating physicians is of the essence [18,25]. A strong recommendation for an autopsy will clearly increase the likelihood of a consent from the relatives [18]. Further reasons for relatives to refuse an autopsy request as listed by medical staff from pathology and internal medicine varied considerably (discomfort; cause of death “is already known”; confusion about the actual procedure; autopsy considered disrespectful behavior; disturbance of the peace of the dead; religious/morale reasons) [21]. In contrast, Tsitsikas et al (2011) reported that the public was in favor of an autopsy as long as results were presented honestly and in detail as well as open information about the autopsy procedure itself (e.g. organ analysis; tissue preservation and examination).
Objections to an autopsy from the medical staff are primarily economical in nature [20] as well as the erroneous belief that the progressing medical diagnostics and effective therapy would reduce the value of an autopsy to such a minimum of new information, that it is not worth while risking an autopsy injury with contraction of an infectious disease (e.g. particularly hepatitis B and C, tuberculosis; HIV). In addition, the work load is intensive, and an autopsy binds experienced staff, which represents an unattractive monetary situation for a hospital [20,26]. On the other hand, in a Norwegian study of 250 general practitioners and clinical doctors, 42% voiced a positive opinion about the value of an autopsy in spite of medical advancements in diagnostic procedures and therapy [27].
Full Body Autopsy Alternatives only in Exceptional Cases
Alternative procedures have been assessed as potential replacement for a full body autopsy such as minimal invasive investigations (needle biopsies; endoscopy) as well as different radiologic techniques. While these procedure may appeal to relatives being less “aggressive”, they are much less accurate than the autopsy in gaining information and sampling of tissues samples, while they need specially trained personnel and equipment leading to increasing costs [18].
However, the option of an autopsy limited to a specific organ/organs may suffice. A sudden and unexpected death in a 5 year old girl occurred, who had experienced previous signs of upper respiratory infection. A myocarditis was suspected. Since the parents were strongly opposing a full body autopsy, heart muscle tissue was taken (from an autopsy limited to the heart), in which molecular analysis identified a Parvo virus myocarditis.
In another case of an elderly woman suffering from liver failure, a full body autopsy was refused by the relatives. A liver-only autopsy was agreed which showed massive liver metastases. To reveal the source of this previously unknown tumor, immune histologic analysis was done, which showed markers in accordance with a papillary serous carcinoma of the ovary.
The Problem of Time for the Post-Mortem Examination
Delaying an autopsy for more than 24 hrs. after death may well lead to difficulties in the evaluation of the diagnoses listed on the death certificate, particularly if premortal diseases are suspected leading to an accelerated decay of body tissues (such as septicemia, chemotherapy, hematologic diseases; post bone marrow transplant; as well as no suitable cooling chamber for preservation of the body).
A critical time factor arises in an autopsy which has been primarily directed to forensics, of which organ parts are subsequently sent to pathology for detailed evaluation. If there is insufficient material being preserved by fixation or there is a prolonged time period between time of death and this subsequent analysis (which often happens due to the administrative /judicial course of events) materials may be no longer preserved well enough for molecular analysis, as exemplified in the following case.
After a tonsillectomy, a 30 year old female bled so profusely, that in spite of surgical intervention by ligation of neck arteries, the patient died. The surgeon being accused of negligence, called for a forensic autopsy. Since specialized histologic analyses were needed; pathology was involved, showing a unique malformation of superior neck arterioles (segmental mediolysis). The value of this autopsy lies in the exculpation of the operating surgeon from the initial accusation. However, the attempt of a molecular analysis of this conatal condition, was thwarted due to the degradation of the RNA, which would have been necessary for further analysis [28].
A Good Relationship to Forensic Pathology is Mandatory for Achieving Best Results in Autopsy Cases
A well working interaction between pathology and forensic medicine is very beneficial and important. Histological/immune histological/ molecular aspects can be of an asset for pathology to aide the forensic analysis. On the other hand, there are cases in which the cause of death is appears highly likely resulting from wrongdoing so that the autopsy should be stopped and continued by the forensic pathologist. As exemplary case we observed a 28 year old male with a history of recurrent pneumonias and pleural effusions. The patient returned to an outside hospital with a massive pleural effusion which needed to be drained. During the night the puncture site bled internally clinically unnoticed, and he died the same night. A regular pathology autopsy was performed unbeknown of the exact medical history (as death following recurrent pneumonia) (lack of clinical data from an outside hospital). Opening the pleura on the former puncture site, a massive hemothorax was discovered. At that point, the autopsy was transferred to the forensic department for specific analysis. The result showed several adhesions, which were undetectable by X-ray following the massive effusion. Fatefully, the treating physician happened to puncture through one of them leading to the deadly bleeding.
Improving Communications is Crucial for Autopsy Permission: the Autopsy Commissioner
To overcome relatives’ arguments of an autopsy rejection it seems evident from the above discussion that an improved communication is imperative. This involves training physicians to favor an autopsy, sufficient time without time pressure and an environment (special room, furniture setting, flowers, pictures, etc.) which is conducive for such a difficult topic to discuss. In our opinion for the university /larger hopsitals, an autopsy commissioner, being a highly qualified (board certified) and experienced (older) clinician could be an ideal mediator to assess and discuss the case with the treating clinicians involved as well as acting as an understanding partner for the relatives. He should also be well aware of the applying legal situation, and be able to competently interact with the funeral director/ staff.
Interdisciplinary Autopsy Conferences Based on Quality Assurance Protocols and Autopsy Reports
The understanding for the need of autopsies can be greatly improved when regular interdisciplinary autopsy conferences are part of the weekly clinical curriculum and will be of great help in teaching students. Hereby, quality assurance protocols and autopsy reports as well as a review of the discussed case via photo documentation (clinical data, macro-photos, histology) are essential tools. As it appears from our evaluation, cases of the convergent/complex type according to ICD10 present with the highest degree of difficulty in the external examination. This will very much depend on the timeliness of information of the existing pre-morbidities and their up-to-date status/therapy. As example, we refer to a female patient with a breast carcinoma 10 year prior, who dies of tumor cachexia; however, the oncologic disease was not known at the time of death to the clinician filling out the death certificate.
Emergently Available Medical Records May Save Lives
Finally, we are convinced that individual patient’s medical records should be readily available in an emergency. For interaction in case of emergencies where a fast and protected exchange of medical history information is needed, a so-called telematic infrastructure is developed (in Europe). It allows data from the electronic patient file to be sent via App to end users (e.g. smart phones, tablets) or PCs. Since Jan 1, 2021, all health insurance companies in Germany have to supply their customers with such an App without extra costs. However, the use of this device is met with some skepticism regarding abuse; although patients have the control, which data they allow to be used by whom. Medical staff can only access this data, if the patient agrees. While electronic patient data files have been introduced in several countries in Northern Europe successfully (Estland 2001; Great Britain 2007; Sweden 2010), in others like Germany (2021) or Greece (2023) this has met much more resistance. Until this is an obligation for all patients, medical data collection by the patient’s medical practitioner seems advantageous as an interim solution.
Conclusion
The present study analyzed the frequency of agreement/ of discrepancy between the cause of death listed in death certificates versus autopsy reports in a university hospital setting from 2005 – 2014. The patients (447) were 353 inpatients and 94 patients being sent from an outside hospital/as emergency (outpatients). While the number of deceased patients in the same time frame was about 15.000 only 1420 autopsies were done (about 9.4%) , of which only a limited number (447) could be used for this analysis because of missing clinical data/death certificates.
Overall, we found an agreement in 68%/ discrepancy in 32%, of which in 7% of cases no cause of death was given. Looking at nosological causal chains, an equal distribution was found for the linear/ divergent type in 21% versus 28.5% for the convergent/ complex type (Table 7). Furthermore, regarding the 143 cases with discrepant diagnoses, the frequency distribution between the different types was almost identical (Table 11). Thus, the frequency relation between the four nosological causal chains is maintained.
Nosological causal chain
# of patients
In %
Type 1: linear
28/ 95
19.5/21.2
Type 2: divergent
31/ 97
21.6/21.7
Type 3: convergent
49/128
34.3/28.6
Type 4: complex
35/127
23.9/28.5
Total #
143/447
32.0/100.0
Table 11: Nosological causal chains in discrepant diagnoses between death certifictaes and autopsy reports vs total number of cases in each category.
The valuation by classes using the Goldman criteria shows a combined major/type I error in 32%. This discrepancy is mainly due to the unknown diseases which accelerated a deteriorating health, and - because they were undiagnosed - had been left untreated.
While about 50% of outpatients were surgical cases, inpatients were cases from the internal medicine departments and only 12% came from general/vascular surgery. Myocardial infarct and cardiac insufficiency were the most frequent diseases followed by liver – and renal failure, gastrointestinal bleedings, and pulmonary insufficiency (Table 4).
Expectedly, the cause of death was mainly related to the 5 major categories (cardiac/circulatory failure; septicemia; myocardial infarct; respiratory failure and shock through internal bleeding (Table 8).
Looking at diseases most likely accompanying/accelerating causes of death main we could identifiy: arterial hypertension, pulmonary emphysema, arteriosclerosis, cardiac insufficiency, and coronary heart disease.
Oncologic diseases played only a relatively minor role except as forerunner of septicemia and respiratory failure. Unexpectedly, obesity and diabetes seem to play a rather small role (about 10% in all age groups): this may be due to the overall limited number of patients due to the high density of universities and major clinics in North Rhine Westfalia region. Overall, hypertension is the leading cause of death amounting from 15% at ages below 35 years to 35% in the patients over 70 years of age.
Author Statements
Ethics
Since human autopsy protocols and death certificates were analyzed, procedures were followed as outlined in accordance with ethical standards, formulated in the Helsinki Declaration 1975 (and revised in 1983), with pre- approval by the Ethic’s Committee of the Medical Faculty of the University of Koeln (Koeln, Germany; file reference 20-1632).
Acknowledgements
The authors thank Prof. Dr. med. R. Buettner for the permission to use the autopsy reports, and Dr. med. Anne Bunte / Prof. Dr. med. Gerhard A. Wiesmüller; Ministry of Health, Koeln, for the access to the death certificates.
We further gratefully acknowledge the many years of trustful and collegial collaboration with the autopsy auxiliary staff in the Institute of Pathology, University of Koeln.
References
- Kabra NS, Udani RH. Correlation between clinical diagnoses at the time of death and autopsy findings in clinically sick neonates at a regional neonatal intensive care unit in India. J Trop Pediatr. 2001; 47: 295-300.
- Ong AW, Cohn SM, Cohn KA, Jaramillo DH, Parbhu R, McKenney MG, et al. Unexpected findings in trauma patients dying in the intensive care unit: results of 193 consecutive autopsies. J Am Coll Surg. 2002; 194: 401-6.
- Fröhlich S, Ryan O, Murphy N, McCauley N, Crotty T, Ryan D. Discrepancies between clinical and autopsy diagnosis in liver transplant recipients–a case series. Acta Gastroenterol Belg. 2013; 76: 429-32.
- Kuijpers CC, Fronczek J, van de Goot FR, Niessen HW, van Diest PJ, Jiwa M. The value of autopsies in the era of high-tech medicine: discrepant findings persist. J Clin Pathol. 2014; 67: 512-9.
- Tai DY, El-Bilbeisi H, Tewari S, Mascha EJ, Wiedemann HP, Arroliga AC. A study of consecutive autopsies in a medical ICU: a comparison of clinical cause of death and autopsy diagnoses. Chest. 2001; 119: 530-6.
- Twigg SJ, McCrirrick A, Sanderson PM. A comparison of postmortem findings with post hoc estimated clinical diagnoses of patients who die in a United Kingdom intensive care unit. Intensive Care Med^1. 2001; 27: 706-10.
- Pastores SM, Dulu A, Voigt L, Raoof N, Alicea M, Halpern NA. Premortem clinical diagnoses and postmortem autopsy findings: discrepancies in critically ill cancer patients. Crit Care. 2007; 11: R48.
- Friemann J. Klinische Obduktionen. Pathol Eur. 2010; 31: 256-67.
- 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: 1000-5.
- von Kress HF. Klinische Aspekte des Sterbens. [Clinical aspects of dying]. Naturwissenschaften. 1970; 57: 1-5.
- Leiss J. Die Todesursache unter individual-pathologischen Gesichtspunkten. DMW. 1982; 107: 1069-72.
- Wittekind C, Gradistanac T. Post-mortem examination as a quality improvement instrument. Dtsch Ärztebl Int. 2018; 115: 653-8.
- Shojania KG, Burton EC, McDonald KM, Goldman L. Changes in rates of autopsy-detected diagnostic errors over time: a systematic review. JAMA. 2003; 289: 2849-56.
- Wittekind C, Friemann J, Gradistanac T, Habeck J-O, Schalke W, Schulz-Schaeffer WJ. Leitline zur Durchführung von Obduktionen in der Pathologie. [Guidelines for autopsies in pathology] (in German). Publication from the Bundesverband Deutscher pathologen e. V. Deutsche Gesellschaft für Pathologie e.V. 2017.
- Tavora F, Crowder CD, Sun CC, Burke AP. Discrepancies between clinical and autopsy diagnoses: a comparison of university, community, and private autopsy practices. Am J Clin Pathol. 2008; 129: 102-9.
- Burton JL, Underwood J. Clinical, educational, and epide miological value of autopsy. Lancet. 2007; 369: 1471-80.
- Winters B, Custer J, Galvagno SM Jr, Colantuoni E, Kapoor SG, Lee H, et al. Diagnostic errors in the intensive care unit: a sys tematic review of autopsy studies. BMJ Qual Saf. 2012; 21: 894-902.
- Shojania KG, Burton EC. The persistent value of the autopsy. Am Fam Phys. 2004; 69: 2540-2.
- Kuijpers CC, Fronczek J, van de Goot FR, Niessen HW, van Diest PJ, Jiwa M. The value of autopsies in the era of high-tech medicine: discrepant findings persist. J Clin Pathol. 2014; 67: 512-9.
- Nemetz PN, Tanglos E, Sands LP, Fisher WP Jr, Newman WP 3rd, Burton EC. Attitudes toward the autopsy–an 8-state survey. MedGenMed. 2006; 8: 80.
- Hull MJ, Nazarian RM, Wheeler AE, Black-Schaffer WS, Mark EJ. Resident physician opinions on autopsy importance and procurement. Hum Pathol. 2007; 38: 342-50.
- Betz ME, Kelly SP, Fisher J. Death certificate inaccuracy and underreporting of injury in elderly people. J Am Geriatr Soc. 2008; 56: 2267-72.
- Loeser H, Wittersheim M, Puetz K, Friemann J, Buettner R, Fries JW. Potential complications of transcatheter aortic valve implantation (TAVI)-an autopsy perspective. Cardiovasc Pathol. 2013; 22: 319-23.
- Saleh MM, Büttner R, Fries JWU. Differentiation between pleura-mesothelioma versus pseudo-mesothelioma demonstrated in eight autopsy cases. Clin Oncol Res. 2021; 4: 2-7.
- Tsitsikas DA, Brothwell M, Chin Aleong JA, Lister AT. The attitudes of relatives to autopsy: a misconception. J Clin Pathol. 2011; 64: 412-4.
- Shojania KG, Burton EC. The vanishing nonforensic autopsy. N Engl J Med. 2008; 358: 873-5.
- Midelfart J, Aase S. The value of autopsy from a clinical point of view. A survey of 250 general practitioners and hospital clinicians in the county of Sor-Trondelag, Norway. APMIS. 1998; 106: 693-8.
- Feld K, Mellin W, Melzer B, Rothschild MA, Fries JWU. Fatal hemorrhage after tonsillectomy. A case of segmental mediolysis in arterioles. [Rechtsmedizin 2021 https://doi.org/10.1007/s00194-021-00502-8].
Download PDF