OUP user menu

Potential medical adverse events associated with death: a forensic pathology perspective

Kentaro Sakai, Akihiro Takatsu, Akio Shigeta, Kenji Fukui, Kyoko Maebashi, Shuntaro Abe, Kimiharu Iwadate
DOI: http://dx.doi.org/10.1093/intqhc/mzp058 9-15 First published online: 3 December 2009


Objective To determine the incidence of potential medical adverse events among patients undergoing forensic autopsy, and to present the characteristics of potential medical adverse events.

Design Retrospective review of consecutive autopsy records.

Setting Department of Forensic Medicine, the Jikei University School of Medicine, Tokyo, Japan.

Participants A total of 3355 forensic autopsy cases between 1983 and 2006.

Main outcome measures Incidence of potential medical adverse events identified in decedents undergoing forensic autopsy, classified by actual occurrence as ‘confirmed’, ‘equivocal’ and ‘negative’ cases; proportion of potential diagnostic, performance and system errors among potential medical adverse events.

Results Of 291 autopsies (8.7%) with potential medical adverse events, 66 cases (22.7%) were confirmed, 42 cases (14.4%) were negative and 183 cases (62.9%) were equivocal. Confirmed cases consisted of potential diagnostic errors in 49 cases (74.2%) and performance errors in 17 cases (25.8%). Equivocal cases included 99 cases associated with potential diagnostic errors (54.1%) and 60 cases associated with potential system errors (32.8%). In 38 of the confirmed cases (57.5%), serious exacerbation of patient condition occurred outside the medical facility.

Conclusions Potential medical adverse events are not uncommon in decedents undergoing forensic autopsy. They are particularly associated with potential diagnostic errors. Forensic autopsy may provide information that could be used to improve care and reduce deaths due to potential medical adverse events.

  • adverse event
  • medical error
  • medical malpractice
  • malpractice claim
  • forensic autopsy


Difficulty in identifying medical adverse events and medical errors causes confusion for both patients and medical care providers. Some patients may be seriously injured due to medical negligence and should therefore be adequately compensated but are unaware of this fact. On the other hand, some medical professions could be beset with groundless claims filed by patients who allege that medical errors have harmed their health. Although only about 3800 malpractice claims are filed a year in the USA for an estimated 27 000 injuries caused by medical negligence [1], medical errors were not recognized in one third of malpractice claims. In 3% of registered claims, no adverse events from medical care were evident [2]. Although obstetrician-gynaecologists and surgeons are at high risk for malpractice claims [25], retrospective reviews have shown that, in 22% of claims against obstetrician-gynaecologists and two-thirds of claims against surgeons, adverse events were unpreventable [68]. In practice, indemnities are sometimes paid even for malpractice claims without error [2, 9].

In cases in which the medical outcome was fatal, the incentive for a patient's family to assert that the death was attributed to medical adverse events and that the adverse event was preventable may be particularly strong. As reported in previous studies, claims of medical error resulting in patient death comprise about 30% of all medical injury claims [2, 7, 911]. However, analyses of medical errors based on court claims or insurance claims are limited in that they do not include information from patients’ families who fail to file claims or are unaware of medical adverse events [12]. Furthermore, a number of previous studies on medical adverse events were based on medical record reviews [6, 7, 1316] and did not include most deaths occurring outside of medical facilities. Objective survey of all cases in which a medical adverse event may have led to death, regardless of recognition and claim by patients’ family and clinical evaluation, has not been straightforward.

The aim of this study was to determine incidence of potential medical adverse events among patients undergoing forensic autopsy and to examine the circumstances surrounding the potential medical adverse events using forensic autopsy cases performed in the Jikei University School of Medicine in Tokyo, Japan.

Materials and methods

This study is a retrospective analysis of 3355 forensic autopsies performed at the Department of Forensic Medicine of the Jikei University School of Medicine between January 1983 and December 2006. The department covers 2.6 million residents of the Tama district of Tokyo and performs approximately 140 forensic autopsies per year. In Japan, the Doctors’ Act obligates physicians to report unusual deaths to the police, and forensic autopsies are performed at the request of the investigation authority (except in five major cities—Yokohama, Nagoya, Osaka, Kobe and the central part of Tokyo—where medical examiner systems are involved).

Autopsy records and investigation reports were consecutively reviewed, and information was extracted from those cases associated with potential medical adverse events by two forensic pathologists who had been trained using simulated cases. Since access to original medical records was limited in most cases, investigation reports that included detailed medical information from forensic pathologists about the deaths contributed significant information to the review process.

Referring to the general definition for medical adverse events as injuries caused by medical management and for medical errors as preventable adverse events given medical knowledge available at the time of the event [14, 17, 18], cases with a ‘potential medical adverse event’ were defined as deaths that seemed to be attributed to any medical practice at the time of death. All the cases were analyzed regardless of the preventability of the event or validity of the judgement. There were the reasons for the inclusions of cases with uncertainty regarding preventability and validity; evaluation of preventability is not straightforward among different patients and medical providers and uncertainty regarding the occurrence of medical adverse event at the time of autopsy might include cases with an actual medical adverse event and also cases without a medical adverse event that were considered to include a medical adverse event based on a claim made by a patient's family.

Cases were evaluated for the validity of the actual occurrence based on autopsy findings and then classified into ‘confirmed’, ‘negative’ or ‘equivocal’. In confirmed cases, attribution of death due to medical adverse events was confirmed by autopsy findings (e.g. death caused by identified common carotid artery injury due to failure in access to external jugular vein). In negative cases, no medical adverse event was identified as a cause of death, based upon autopsy findings (e.g. sudden death during colectomy diagnosed as ischemic heart failure by post-mortem autopsy). In equivocal cases, presence of medical adverse event remained undetermined (e.g. death at home due to subdural haemorrhage without any specific findings on diagnostic imaging at consultation).

Moreover, cases were categorized according to types of potential medical error, assuming that the adverse events were preventable. To determine the types of errors, indices employed in the Harvard Medical Practice Study [14] were used, after modification to incorporate ‘diagnostic errors’, ‘performance errors’ and ‘system errors’. Diagnostic errors consisted of overlooking, delay or failure to evaluate severity (e.g. overlooking aortic dissection for a patient presenting chest pain, underestimating a patient with bronchopneumonia as common cold). Performance errors included inadequate drug treatments in addition to errors in technical procedures (e.g. intra-abdominal haemorrhage due to surgical operation trauma, incompatible blood transfusion). System errors included failures in management of patients’ health and prevention of adverse events or inadequate function of services (e.g. accidental injury of a disabled patient due to fall from a hospital bed without taking precautions). In cases with multiple errors, the most influential error with regard to the death was used for the classification.

Reviewers classified potential medical adverse events and potential medical errors independently. Cases with disagreement were re-classified after discussion. The site where a serious clinical exacerbation occurred and length of admission prior to death were also examined; clinical exacerbation or death did not always occur at medical facilities, even after performance errors. Clinical specialties involved were grouped into internal medicine, surgery, paediatrics, obstetrics and gynaecology and psychiatry based on the specialty that provided the majority of care before death. If cases were under the care of family practice or emergency departments, they were categorized according to the causes of death.

Frequencies and incidence of potential medical adverse events were determined for trends, characteristics of patients, types of potential medical errors, site of exacerbation or death, length of admission at death and the involved clinical specialties. Proportion of potential medical adverse events in forensic autopsies and age of patients were also compared using χ2 tests, mean and standard error. For inter-rater reliability, kappa statistics were calculated in classification of potential medical adverse events and potential medical errors. The study was reviewed and approved by the Jikei University School of Medicine Ethics Committee.


From 1983 to 2006, among 3355 forensic autopsies there were 291 cases (8.7%) with potential medical adverse events at time of death (Table 1). Although the proportion of potential medical adverse events for each study year varied, the mean proportion after the year 2000 was 12.6%, higher than the 6.2% recorded prior to 1990 and the 6.8% reported during the 1990s (P < 0.01). Among 291 cases with potential medical adverse events, medical adverse events were confirmed for 66 cases (22.7%), although negative cases and equivocal cases accounted for 42 cases (14.4%) and 183 cases (62.9%), respectively (Table 2). The kappa statistic was 0.59 and 0.68 for inter-rater reliability in the classification of potential medical adverse events and potential medical errors, respectively.

View this table:
Table 1

Number of forensic autopsies and potential medical adverse events

YearsForensic autopsies No.Potential medical adverse events No. (%)
1980s (1983–1989)75747 (6.2)
1990s (1990–1999)143397 (6.8)
2000s (2000–2006)1165147 (12.6)
Total (1983–2006)3355291 (8.7)
View this table:
Table 2

Number of patient characteristics in ‘confirmed', ‘negative' and ‘equivocal' cases of potential medical adverse events

Potential medical adverse events
No. (%)No. (%)No. (%)
Total66 (22.7)42 (14.4)183 (62.9)
 Male32 (48.5)26 (61.9)116 (63.4)
 Female34 (51.5)16 (38.1)67 (36.6)
 Mean age, year
 (±SE, Range)57.0 (±2.6, 1–98)49.4 (±3.3, 0–62)42.2 (±1.8, 0–92)
Types of potential medical errors
 Diagnostic49 (74.2)14 (33.3)99 (54.1)
 Perfomance17 (25.8)28 (66.7)24 (13.1)
 System0 (0)0 (0)60 (32.8)
Sites of exacerbation
 Within medical facilities
  Patient room16 (24.2)18 (42.9)72 (39.3)
  Consultation room3 (4.5)2 (4.8)9 (4.9)
  Laboratory3 (4.5)3 (7.1)7 (3.8)
  Operating room4 (6.1)5 (11.9)2 (1.1)
  Delivery room1 (1.5)2 (4.8)6 (3.3)
  Others1 (1.5)1 (2.4)18 (9.8)
 Outside medical facilities
  Home35 (53.0)9 (21.4)60 (32.8)
  Others3 (4.5)2 (4.8)9 (4.9)
Length of admission at death
 Within 24 h18 (27.3)13 (31.0)60 (32.8)
 Between 1 and 7 days5 (7.6)8 (19.0)25 (13.7)
 More than 1 week17 (25.8)13 (31.0)57 (31.1)
 None26 (39.4)8 (19.0)41 (22.4)
Clinical specialties
 Internal medicine44 (66.7)19 (45.2)55 (30.1)
 Surgery17 (25.8)13 (31.0)40 (21.9)
 Obstetrics and gynaecology2 (3.0)5 (11.9)14 (7.7)
 Paediatrics2 (3.0)0 (0)28 (15.3)
 Psychiatry1 (1.5)5 (11.9)46 (25.1)

Cases with confirmed medical adverse events

Of the 66 cases with confirmed medical adverse events, 34 cases (51.5%) were female (Table 2). The mean age was 57.0 years old (SE ± 2.6). Potential medical errors were classified as diagnostic errors in 49 cases (74.2%) and performance errors in 17 cases (25.8%). There were no cases categorized as system error. Patients developed fatal conditions at home in 35 cases (53.0%) and within medical facilities, primarily patient rooms, in 16 cases (24.2%). Deaths without hospitalization accounted for 26 cases (39.4%), and deaths within 24 h from admission accounted for 18 cases (27.3%), although length of hospitalization was more than 1 week at death in 17 cases (25.8%).

Internal medicine was the most frequently involved specialty with 44 cases (66.7%), followed by surgery with 17 cases (25.8%). Examples of death due to potential performance errors were accidental vascular injury in vascular access and surgery, failure in respiratory care and problems with perioperative care, invasive treatment and examination (Table 3). Causes of death from potential diagnostic errors in internal medicine patients included ischemic heart disease, rupture of aorta, ileus, gastrointestinal perforation and respiratory infection shortly after consultation for associated symptoms (Table 4). Among surgery cases, accidental internal organ injury, such as intestinal rupture, liver contusion, cervical spine injury and pelvic fracture, leading to death were undiagnosed at consultation; natural diseases such as rupture of aorta and gastrointestinal perforation were also represented. Ileus and respiratory infection were causes of death in paediatric patients, and gastrointestinal perforation was seen in a psychiatric patient; all cases received consultation for abdominal pain and flu-like symptoms shortly before death.

View this table:
Table 3

Examples of confirmed medical adverse events in potential performance errors

Vascular injury due to access to central vein, pulmonary artery and cerebrovascular
Vascular injury during operation in neurosurgery, gastrointestinal and gynaecological surgery
Failure of endotracheal intubation
Accidental removal of ventilator tube and vascular catheter
Total spinal block
Incompatible blood transfusion
Perforation of alimentary canal due to endoscopy
View this table:
Table 4

Causes of deaths from potential diagnostic errors in confirmed medical adverse events by clinical specialty

Internal medicine No.Surgery No.Paediatrics No.Psychiatry No.
Ischemic heart disease11
Rupture of aorta73
Gastrointestinal perforation311
Respiratory infection31

Cases negative for medical adverse events

Among the cases classified as negative for a medical adverse event, males comprised 26 cases (61.9%), and the mean age was 49.4 years old (SE: ± 3.3). Although these cases were considered to include medical adverse events at the time of death classified as diagnostic errors in 14 cases (33.3%) and performance errors in 28 cases (66.7%), autopsy disclosed the absence of actual medical adverse events. Serious exacerbations occurred in the hospital patient room in 18 cases (42.9%) and at home in 9 cases (21.4%). Cases admitted within 24 h prior to death accounted for 13 cases (31.0%) and cases admitted more than 1 week prior accounted for 13 cases (31.0%). Internal medicine accounted for 19 negative cases (45.2%), 13 cases (31.0%) were in surgery, five cases (11.9%) in obstetrics/gynaecology and five cases (11.9%) in psychiatry. There were no paediatric cases classified into this category.

Among classified potential diagnostic errors, there were eight cases (38.1%, Table 5) in internal medicine of deaths shortly after consultation, resulting in detection of an unrelated cause of death at autopsy (e.g. cerebral haemorrhage after abdominal pain and alcohol intoxication following consultation for nausea). In surgery, five cases (38.5%) were classified as potential diagnostic errors, as death attributed at autopsy to natural causes (e.g. ischemic heart disease, myocarditis, liver cirrhosis) occurred shortly after consultation for head and/or chest trauma.

View this table:
Table 5

Types of potential medical errors by clinical specialty

Internal medicineSurgeryObstetrics and gynaecologyPaediatricsPsychiatry
No. (%)No. (%)No. (%)No. (%)No. (%)
 Diagnostic37 (84.1)9 (52.9)0 (0)2 (100)1 (100)
 Performance7 (15.9)8 (42.1)2 (100)0 (0)0 (0)
 System0 (0)0 (0)0 (0)0 (0)0 (0)
 Diagnostic8 (42.1)5 (38.5)0 (0)0 (0)1 (20.0)
 Performance11 (57.9)8 (61.5)5 (100)0 (0)4 (80.0)
 System0 (0)0 (0)0 (0)0 (0)0 (0)
 Diagnostic39 (70.9)27 (67.5)4 (28.6)24 (85.7)5 (10.9)
 Performance5 (9.1)7 (17.5)10 (71.4)1 (3.6)1 (2.2)
 System11 (20.0)6 (15.0)0 (0)3 (10.7)40 (87.0)

Among potential performance errors, autopsies disclosed natural causes of death in 11 internal medicine cases (57.9%); these consisted of deaths during or within 24 h after drug administration, alimentary canal endoscopy, endotracheal intubation or vascular access. In surgery, deaths during or within 24 h after surgical operation or during hospitalization for surgical disease were considered to be potential performance errors, although all of them were eventually diagnosed as deaths due to natural causes such as pulmonary thromboembolism, ischemic heart disease and peritoneal carcinomatosis unrelated to medical adverse events. In obstetrics and gynaecology, potential errors were classified as performance errors in three maternal deaths and two neonatal deaths associated with perinatal care; the neonatal deaths were diagnosed as being due to amniotic fluid embolism, coiling of the cord or respiratory distress. In psychiatry, there were deaths after errors in drug dispensing, inadequate monitoring after administration of drugs or inappropriate treatment procedures; however, all cases were diagnosed as death due to other natural causes at autopsy.

Cases with equivocal medical adverse events

Among cases judged equivocal for the presence of an adverse event, males comprised 116 cases (63.4%). The mean age was 42.2 years old (SE ± 1.8). Potential medical errors were most frequently (99 cases) classified as diagnostic errors (54.1%), followed by system errors in 60 cases (32.8%) and performance errors in 24 cases (13.1%). Among the equivocal cases, patients developed fatal conditions within medical facilities in 114 cases (62.2%) and at home in 60 cases (32.8%). Death within 24 h from admission accounted for 60 cases (32.8%), and death without hospitalization accounted for 41 cases (22.4%). Internal medicine was involved in potential adverse events in 55 cases (30.1%), followed by psychiatry in 46 cases (25.1%).

In internal medicine, surgery and paediatrics, equivocal medical adverse events were frequently associated with diagnostic error, accounting for 70.9, 67.5 and 85.7%, respectively. Potential diagnostic errors included death due to natural diseases, such as ischemic heart disease, myocarditis, ileus, pulmonary thromboembolism, subarachnoid haemorrhage or respiratory infection, or to trauma, such as subdural hematoma, intestinal rupture or fat embolism, shortly after consultation for non-specific symptoms. Among potential performance errors, four maternal deaths and four neonatal deaths due to perinatal problems (e.g. uterine rupture, birth injury due to unidentified cause) were observed in obstetrics and gynaecology. There were seven deaths in total due to shock after administration of analgesia and/or anaesthesia in internal medicine and surgery. Among potential system errors, 40 cases in psychiatry included death due to latent diseases during hospitalization that could be associated with inadequate prevention or detection of fatal conditions such as pulmonary thromboembolism or ischemic heart disease. Deaths due to pulmonary thromboembolism during hospitalization were also observed in internal medicine and surgery. Furthermore, some deaths due to accidental injury in hospitalized patients were observed with potential but unconfirmed links to inadequate management (e.g. intracranial injury due to fall from examining table, drowning during assisted bathing). In addition, deaths associated with delay or inadequate transfer of patients from primary emergency facilities to acute or intensive care facilities were observed in three internal medicine cases and two surgery cases, although those cases noted natural disease or head injury as immediate cause of death.


This study found that potential medical adverse events accounted for 8.7% of patients undergoing forensic autopsy and that forensic autopsy can reveal the presence or absence of medical adverse events based on a perspective distinct from that of medical professionals, patients’ family or prosecutors. Assuming that potential medical adverse events were preventable, upon further categorization, potential diagnostic errors were more frequent than potential performance errors among confirmed cases of a medical adverse event. However, all of the cases with potential system errors were classified as equivocal with respect to the presence or absence of a medical adverse event.

Deaths with potential medical adverse events among forensic autopsies are not uncommon, although the incidence determined from these data is not directly comparable with incidence found in previous studies due to the difference in study population, which showed that incidence of death among all medical adverse events in hospitalized patients was 5–14% [1316]. A recent increase in the incidence of deaths with potential medical adverse events was found. This may be due to global awareness of patient safety raised by large-scale studies in the 1990s [13, 14, 16] and the major announcement from the United States Institute of Medicine in 1999 [17]. Additionally, in Japan, medical errors have frequently been reported in the media since 1999, when incorrect surgeries due to a switch between patients took place at a university hospital [19]. These circumstances may encourage more physicians than before to report deaths with medical adverse events to investigation authorities, in addition to natural increase of reported deaths with an increasing demand for medical care in aging population [20]. The higher frequency of equivocal cases compared with confirmed or negative cases indicates that the actual incidence of deaths with medical adverse events may be higher than this study has presented. Identifying medical adverse events only by forensic autopsy can also be inadequate for finding all cases.

This study's finding of the frequency of potential diagnostic errors is consistent with findings in previous studies [24, 9]. Currently in Japan, however, performance errors receive much of the attention in relation to deaths due to medical errors [21]. In previous studies regarding diagnostic accuracy with autopsy, frequency of major diagnostic errors in hospitalized patients ranged from 4.1 to 49.8% (23.5% of the median rate) and frequency of errors without which deaths may have been prevented range from 0 to 20.7% (9.0% of the median rate) [2233]; similar results have been reported in studies based on forensic autopsies [34, 35]. The incidence of diagnostic errors in this study's confirmed and equivocal medical adverse events was estimated to be 4.4%, consistent with these previous findings. The relatively low incidence in this study may be related to the sample we examined which included a number of deaths that had no association with medical management. In fact, if preventability is considered for identification of medical error, autopsy findings alone may not precisely determine whether failure in diagnosis at consultation is preventable because the risk for an exacerbation of an underlying condition varies by patient, and patients are not always treated in well-equipped hospitals [36]. However, the fact that a number of outpatients are allowed to return home with fatal condition such as ischemic heart disease or ileus after physicians assessed them non-urgently suggests that prevention of potential diagnostic errors may have a large impact on decreasing potential medical adverse events.

Potential performance errors are more likely to be identified in autopsy if the error was due to inadequate technical procedure, as it was in more than half of potential performance errors classified into confirmed or negative cases. This may be so because the adverse event is grossly or objectively observable at the time of autopsy. Potential system errors primarily consisted of failure to prevent fatal conditions in hospitalized patients with other minor diseases, failure to prevent accidental injuries within hospitals and inadequate transfer of patients. Although these factors for medical adverse events seem to be related to a patient death, only autopsy findings and death-scene investigation cannot demonstrate causal relationships between those factors and the death; therefore, potential system errors need to be further explored with more cases. Deaths associated with delay or inadequate arrangement of patient transfer from primary emergency facilities to acute or intensive care facilities suggest a scarcity and inefficiency of emergency medical resources in Japan [37].

One limitation of this study is a potential bias due to use of a forensic autopsy cohort; cases that did not undergo forensic autopsy at the discretion of the investigation authority were not included. Other limitations include problems with generalizability due to the different autopsy rates for unusual death among jurisdictional areas and the retrospective attribution of error based on limited available data. However, it was common for forensic autopsy to be performed when association between death and medical adverse events was suspected. Although investigation reports that lacked details of medical cases may have affected the classification of potential medical adverse events, they were supplemented with relevant information in medical records and interviews with attending physicians at the time of autopsy as needed.


Deaths associated with potential medical adverse events, particularly those associated with potential diagnostic errors, are not an uncommon finding in forensic autopsy, regardless of recognition by medical professionals, patients’ families or prosecutors. To reduce deaths from potential medical adverse events, further research and prevention strategies should focus on delay and underestimation in diagnosis, as well as on assuring the correct conduct of technical procedures. Additionally, proper management of patients with multiple medical problems should be encouraged to prevent death due to another avoidable cause.


This study was, in part, supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and the Japan Society for the Promotion of Science (Grant #: 18790425).


View Abstract