Attitudes toward clinical autopsy in unexpected patient deaths in Japan: a nation-wide survey of the general public and physicians
- 1Department of Social Welfare, University of Kochi, Kochi, Japan
- 2Graduate School of Health Management, Keio University, Fujisawa, Kanagawa, Japan
- 3Department of Internal Medicine, University of California, San Diego, California, USA
- 4Department of Forensic Pathology and Sciences, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan
- Correspondence to Dr Shoichi Maeda, Keio University, Graduate School of Health Management, 4411 Endo, Fujisawa, Kanagawa 252-8530, Japan;
- Received 15 June 2012
- Revised 23 August 2012
- Accepted 29 August 2012
- Published Online First 27 September 2012
Context Autopsy is a useful tool for understanding the cause and manner of unexpected patient death. However, the attitudes of the general public and physicians in Japan about clinical autopsy are limited.
Objective To describe the beliefs of the general public about whether autopsy should be performed and ascertain if they would actually request one given specific clinical situations where patient death occurred with the additional variable of medical error. To compare these attitudes with previously obtained attitudes of physicians practising at Japanese teaching hospitals.
Design, setting and participants We conducted a cross-sectional study of the general public. We sent standardised questionnaires in 2010 to a randomly selected non-physician adult population using a survey company for participant selection. Respondents gave their opinions about the necessity of autopsy and how they might act given various clinical scenarios of patient death. We compared these results with those of a previous survey of Japanese physicians conducted in 2009.
Results Of the 2300 eligible general adult population, 1575 (68.5%) responded. The majority of the general public indicated they believed an autopsy was necessary. However, in cases of unclear medical error or unclear cause and effect relationship of medical care and patient death, the general public were much less likely to indicate they would actually request an autopsy than were physicians (p<0.0001). Currently in Japan the debate about the role autopsy should play in the case of error related to death is underway. The results from this study will be important in informing related decisions.
When patients die unexpectedly while receiving medical care, medical error may or may not be present. When medical error has occurred, the direct relationship of the medical error to that death may or may not be present.1 ,2 That is to say, the cause of death may be entirely unrelated to medical error even though one has occurred. When it is unclear if there has been a medical error or if the medical error was related to death, autopsy becomes an important tool for investigating the cause and manner of death.3–8
When unexpected patient death occurs, there is no uniform system in Japan to handle autopsy.9 ,10 We will provide some background. After the second world war, some localities established systems akin to a medical examiner's office. However, for various reasons related to the local governments of the time, many of these programmes were abandoned, and in many areas, no obligatory system to handle unexpected death exists to this day. Compare this with Western countries: while the names and qualifications of various systems and positions within those systems may differ, for example, ‘coroner’ versus ‘medical examiner’, a system is usually required to be in place to deal with autopsy in all regions in the event of an unexpected death, and often these systems are well established. In Japan, however, the options are often limited to calling the police or initiating an internal hospital investigation, depending on the situation. More often than not, it is the physicians who recommend autopsy. It is up to the physicians to take the initiative in obtaining families’ consent for autopsy in cases where it is unclear whether there has been a medical error or whether there was a cause and effect relationship between medical care and patient death. Physicians, for reasons of self-incrimination or otherwise, may be reluctant to recommend autopsy even when it might be appropriate after unexpected patient death.1 Of course, the families can also request an autopsy even when the physicians do not offer,11 ,12, so understanding how willing families would be to request an autopsy becomes valuable information.
Although research has described autopsy in Japan, its focus has been on more routine autopsy, such as for better understanding disease progression or confirmation of presumed diagnosis for the purpose of furthering medical understanding, not on autopsy in the context of possible medical error and iatrogenic death. Furthermore, prior research has focused on the attitudes of forensic and clinical pathologists, not the physicians and the general public who are more directly responsible for initiating autopsies in general. We previously performed a cross-sectional survey of all residency programme directors (physicians) in Japan in 2009 and found that the majority reported they would recommend an autopsy when there may have been medical error, although over 10% of physicians said that they would not recommend an autopsy.1 In this study, we aim to (1) investigate the attitudes of the general public toward autopsy in specific scenarios and (2) to compare these attitudes with those of physicians.
Subjects and methods
Two thousand three hundred people were randomly selected to participate in the study from a nationwide database compiled by a survey company (Japanese Management Association Research Institute, Inc.) without regard to sex, social status, age, or residence on 4 February 2010. Healthcare workers such as physicians and nurses were excluded. The questionnaire was sent by fax machine to potential participants who had agreed to participate in survey research. Of note, different from other countries, in Japan fax machines are still in widespread use with one in nearly every home. Participants did not receive compensation directly but could earn points by survey completion that were redeemable for prizes through the survey company.
We have described the physician participants in detail elsewhere.1 Briefly, in January 2009 we sent a questionnaire to the attending physicians in charge of graduate medical education at every hospital in Japan that participates in the standardised national general residency programme (all teaching hospitals with residents), including 1004 non-university and 109 (1113 total) university-affiliated hospitals, as listed on the government's residency programme information repository.13
A structured, anonymous, self-administered questionnaire presenting various scenarios of unexpected patient death consisted of four sections: (1) the significance of autopsy, (2) the causative factors for low autopsy rates, (3) opinions on necessity for autopsy and attitudes towards requesting/recommending autopsy in specific clinical scenarios and (4) respondent demographics. The questionnaire for the physicians was more lengthy and included additional questions about hospital policies on autopsy, and reporting of patient deaths to the police.
We defined specific clinical scenarios where we believe autopsy would be required to determine whether medical error occurred and whether medical error contributed to the death of the patient. Each scenario described the presence, absence, or uncertainty of medical error and the presence, absence, or uncertainty of the causal relationship of care rendered to patient death. This information was presented in a graph format for clarity and was worded simply and clearly (see figure 1).
The questionnaire administered to the general public was similar to that previously administered to physicians; we first asked if respondents believed autopsy was necessary in each scenario. To see if respondents would act upon their assessments of necessity, members of the general public were asked if they would request autopsy to their treating physicians (regardless of their belief if it was necessary or not).
Multiple scenarios were presented with the commonality being a patient who died during the course of receiving medical care. If a medical error had occurred, the medical error was described. It is important to note that not all medical errors led to patient death; for example, a wrong drug given to a patient who died from severe injuries related to an automobile accident died regardless of the medical error. To make this clear, the relationship between the medical error and patient death was explicitly stated. We also presented scenarios where medical error occurred but was not so clearly related to patient death. We also presented scenarios where medical treatment led directly to patient death but no error occurred (eg, routine surgery resulting in death as a complication).
For ease of discussion, the various scenario conditions will be hereafter denoted in shorthand with ‘error’ and/or ‘relationship’ followed by a symbol (‘+’, ‘−’, or‘?’). ‘Error+’ denotes medical error was definitely present and ‘error−’ denotes medical error was definitely absent. ‘Relationship+’ means medical care (erroneous or not) definitely led to patient death and ‘relationship−’ denotes that medical error was definitely not related to patient death. Finally, ‘error?’ denotes that it is unclear if an error occurred and ‘relationship?’ means it is unclear if medical error was related to patient death.
Statistical analysis was completed using JMP8.0 software. χ2 Analysis was used to compare the differences between the respondent groups who did or did not think clinical autopsy was necessary. χ2 Analysis was also used to compare the differences between respondent groups who would or would not request/recommend autopsy. Significance was set at an α <0.05.
Response rates and patient demographics
Of the 2300 people selected from the general public and the 1113 physicians approached, completed surveys were returned from 1575 (68.5%) and 466 (41.9%), respectively (table 1). Unfortunately, education levels were not available from the survey company and we did not ask them on the questionnaire. We inquired about non-delivery rates and the survey company was also unable to provide us with data; however, they estimate, in general, a roughly 5% non-delivery rate across all surveys they send.
Should an autopsy be performed?
Table 2 shows whether or not participants believed autopsy was necessary (should be performed) in a given scenario. Statistical differences in the attitudes of the general public versus physicians are highlighted below.
Scenario 1: Patient death as a result of medical care
Except for the case of definite error leading to patient death (relationship+/error+), the general public was significantly less likely than physicians to believe an autopsy was necessary (p<0.0001). They were also significantly less likely than physicians to report they would request an autopsy (p<0.0001).
Scenario 2: Medical error
In all scenarios (table 2, lower half), the general public was significantly less likely than physicians to believe an autopsy was necessary (p<0.0001).
Acting on beliefs
As presented in the section ‘physicians’ above, we asked participants, ‘Do you think an autopsy should be performed?’ Based on their responses, we grouped them into participants who believed an autopsy would be necessary (see section ‘I believe an autopsy is necessary, but would I ask for one?’) and those who felt an autopsy was unnecessary (see section ‘I believe an autopsy is unnecessary, but would I stand by that belief by not asking for one?’). Section ‘acting on beliefs’ conveys the follow-up question of if the participants would act on their assessments and go on to ask for an autopsy (or in the case of physicians, recommend an autopsy).
I believe an autopsy is necessary, but would I ask for one?
Scenario 1: Patient death as a result of medical care
Significantly more physicians than the general public (8.8% vs 12.1%, p=0.0487) indicated that they would not recommend an autopsy even when they believed an autopsy was necessary in the presence of error leading to patient death (relationship+/error+). Meanwhile, in the case where error was unclear (relationship+/error?), significantly more of the general public indicated they would not request an autopsy even when they believed an autopsy was necessary (15.4% vs 9.6%, p=0.0025). The largest difference was seen in the case where error was absent (relationship+/error−). The general public were much more likely than the physicians to indicate they would not request an autopsy when error was absent even though they thought autopsy was actually necessary (41.1% vs 9.6%, p<0.0001) (table 3).
Scenario 2: Medical error
There was no difference in participants who would not request/recommend an autopsy even if they believed an autopsy was necessary (9.7% vs 9.6%, p=0.9769) in the case of relationship (error+/relationship+). Additionally, in the case where relationship was unclear (error+/relationship?), the general public and physicians did not differ (13.5% vs 9.9%, p=0.0646) in their opinions. The largest difference, however, was seen when no relationship was present (error+/relationship−). The general public were much more likely to indicate they would NOT request an autopsy when relationship was absent, even when they thought autopsy was actually necessary (24.6% vs 11.0%, p<0.0001) (table 3).
I believe an autopsy is unnecessary, but would I stand by that belief by not asking for one?
Scenario 1: Patient death as a result of medical care
Significantly more physicians indicated that they would recommend an autopsy even if they believed an autopsy was unnecessary (6.0% vs 16.7%, p=0.0236) in the presence of error (relationship+/error+). Additionally, in the case where the presence of error was unclear (relationship+/error?), significantly more physicians indicated that even if they believed an autopsy was unnecessary, they would still recommend an autopsy (14.1% vs 33.3%, p=0.0119). The largest difference; however, was seen when error was absent (relationship+/error−). The general public were much less likely to request autopsy when error was absent, and felt autopsy was unnecessary (1.4% vs 23.1%, p<0.0001) (table 4).
Scenario 2: Medical error
There was no difference in participants who would request/recommend an autopsy even if they believed an autopsy was unnecessary (4.1% vs 10.2%, p=0.0592) in the case of relationship (error+/relationship+). In the case where relationship was unclear (error+/relationship?), more physicians said they would recommend autopsy even if they felt it was unnecessary compared with the general public (13.0% vs 3.0%, p=0.0004). The largest difference, however, was seen when no relationship was present (error+/relationship−). The general public were much less likely than the physicians to request an autopsy when relationship was absent, when they thought autopsy was unnecessary (1.2% vs 11.3%, p<0.0001) (table 4).
Analysis by demographics
Analysis of the data regarding the general public (presented in table 1) by gender and age showed no statistically significant differences. Physicians were also analysed by gender and age and no group differences emerged.
Relatively high participation rates were obtained in the study. Reasons for this are unclear, but the results are in line with typical response rates from survey companies in Japan targeting the general public. For example, a study conducted by the Cabinet Office of the Japanese Government in 2010 sent to 3000 people using a similar data bank and survey company had a participation rate of 64%.14
This study aimed (1) to reveal the attitude of the general public on desire for autopsy in unexpected patient death, and (2) to compare these attitudes with those of physicians using the data obtained in our previous research.1 In the sections that follow, we will attempt to analyse the responses of the general public and compare and contrast them with those of physicians. However, to do so requires some background understanding of the beliefs held in Japan around death and autopsy in general.
In Japan, the body of the deceased is considered by many to be sacred. Indeed, this was the topic of a recent hit movie called ‘Departures’ (‘Okuribito’ in Japanese, which refers to the person who handles the ‘send off’ of the dead), which received an Academy Award for best foreign language film in 2009. The movie reveals some of the cultural nuances of how the Japanese view the dead and viewers get a sense of the sacred nature of the deceased. This sense of sacredness has been offered as a partial explanation for the low rate of transplantation surgery in Japan.15 Any discussion of autopsy must take into account that, especially in the older generation, these attitudes, while not exactly religious are rooted in Shintoism, also have influence over autopsy, and that autopsy rates are low in Japan. Indeed, because autopsy rates are so low in Japan, postmortem medical imaging using CT is often used as an alternative.16
Scenario 1: Patient death as a result of medical care: should autopsy be performed?
Error caused patient death (relationship+, error+)
When a clear causal relationship between medical error and death exists, autopsy is not strictly necessary in theory. However, approximately 90% of the general public believed that autopsy was necessary in this situation.
There are several possible reasons why the general public may believe that autopsy is necessary. The first is that some people desire to know not only the presence of medical error or causal relationship of medical care to patient death, but also the specific manner of death. All the research on bereaved families and autopsy indicates that the majority of families desire to know the exact, detailed cause of death.17–20
The second reason is that there are some who may desire to preserve concrete evidence of medical error in order to make a claim in the future.21 ,22 In fact, there have been cases in which the medical side has later denied medical error within the court even though they had initially acknowledged it.19 Also, some of the general public may not understand the significance of medical error unequivocally related to patient death, and think that autopsy is necessary whenever there is any medical error.20–24
The approximately 10% who responded that autopsy is unnecessary may be those who realised that autopsy is not technically necessary in this situation or simply people who do not desire autopsy in general.
Error may have caused the death (relationship+, error?)
When the relationship between error and death could not be readily established, 81% of the general republic responded that autopsy was necessary. While most thought it was necessary, a greater proportion of the general public than physicians (20% vs 5%) thought it was unnecessary, and fewer thought it was necessary than when medical error was clearly present. This is the opposite of what we had expected—again perhaps pointing to patients believing when medical error is clearly present that autopsy can be used as a tool for documentation or otherwise. The general public may have different underlying goals related to autopsy.4 ,20–22 ,25 Perhaps education of the general public in the significance of autopsy should be instituted as others have suggested.20–24
Medical care caused death, but without error (relationship+, error−)
Patients often die in the course of routine medical care as a complication of care, most commonly in the case of surgery, both elective and emergency. In this case, the majority of the general public did not think autopsy was necessary. However, it was interesting to find that more of the general public than physicians thought that autopsy was not necessary by a wide margin (81.7% vs 23.4%, p<0.0001).
Medical professionals are likely see autopsy as a tool for learning and improvement of the medical system, whereas the general public may not have heard of this concept. Additionally, if autopsy is primarily viewed by patients as a tool to investigate error, and error is clearly absent, they may believe that autopsy is unnecessary. Finally, if error is absent, the general public may believe that the body of the deceased should not be subject to further insult.4 ,20–22 ,25
Scenario 2: Medical error occurred. Should autopsy be performed?
Error led to patient death (error+, relationship+)
This scenario considers the presence of both medical error and a causal relationship of medical care to patient death (relationship+/error+), as in scenario 1 (error+/relationship+), essentially serving as an internal control.
As expected from the results of scenario 1, a majority of the general public also considered autopsy necessary in scenario 2, but the number differed significantly: 89% vs 74%.
Physicians, on the other hand, responded similarly to each of these questions. It is not easy to make sense of this discrepancy. From these results, there are several possibilities for the 15.6% difference. First, it is possible that there are people in the general public who do not understand the relationship between medical error or causal relationship of medical care to patient death and the need for clinical autopsy.17–21 Perhaps it was our choice of wording that made the difference. In scenario 1, we used the Japanese word ‘kashitsu’ in the main question stem, which is translated as ‘medical error’ and focuses on the error. In scenario 2 we used the word ‘jikou’ in the main question stem (although we used ‘kashitsu’ in the supporting text) which also translates into ‘medical error’ and is used interchangeably but has more a focus on the results of the error in relationship to the patient. The use of slightly different language might have affected our results. If so, using the word ‘kashitsu’ may actually increase the likelihood that families will request autopsy and may be a preferable term if autopsy is desirable. Further research could confirm this finding.
Error may have led to patient death (error+, relationship?)
When medical error is present but may or may not have caused patient death (error+/relationship?), the ratio of the general public who indicated that autopsy is necessary was lower than that of scenario 2 with medical error and causal relationship. As previously mentioned, we believe autopsy is most indicated in cases when the causal relationship is unclear. The significance of autopsy in such cases should be better publicised to persuade families to request autopsy even when it is not proposed by the physicians.
That more physicians would recommend autopsy than the general public in all permutations of scenario 2 (error+) is perhaps not surprising since physicians have probably been exposed to the idea that autopsy is valuable. What is interesting, however, is the relative constancy of the percentage of physicians who would recommend autopsy—even when the connection between medical error and death is firmly established without one. Our assumption was that autopsy is only necessary when the relationship of error to death is ambiguous and therefore more physicians would have chosen autopsy in this situation. Perhaps clinicians reason differently, that the purpose of autopsy is to confirm definitively what is already known—that is, that the medical error caused the death, in both situations where medical error led to the patient death directly and situations where this relationship is ambiguous. Alternatively, physicians may desire autopsy in all cases of patient death.
Error was present but did not lead to patient death (error+, relationship−)
In this case, in our opinion, autopsy is not necessary. It is interesting to note that like scenario 1, more of the general public than physicians thought that autopsy was unnecessary (82.7% vs 35.4%, p<0.0001).
‘Should’ and ‘would’: the translation of belief of autopsy necessity into an autopsy request
I believe autopsy should be performed, would I request an autopsy?
A majority of the general public who consider autopsy necessary responded that they would request autopsy. Mirroring their beliefs, more people would actually request an autopsy in cases of medical error being clearly present than in cases where medical error is unclear. There are significant differences in the attitudes of the general public and physicians in scenario 1 in all cases, regardless of the presence of medical error. As in scenario 2, all the cases without causal relationship displayed a significant difference in responses. In both scenarios, the ratio of physicians proposing autopsy to families was greater than the general public requesting autopsy to physicians. There are several possible reasons for this.
First, throughout their medical studies, physicians are repeatedly taught to propose autopsy to families, even when the cause of patient death is illness.5 Secondly, this study targets teaching hospitals, where autopsy is likely to be especially emphasised. In order to be certified as a teaching hospital in Japan, hospitals are required ‘to carry out more than 20 autopsies a year, and autopsy performed on >30% of the patients’.26 Physicians working at teaching hospitals are requested by administrators to propose autopsy to families when the patient dies in order to fulfil this requirement. Probably these physicians, when presented with the opportunity for autopsy, will take it, especially if they deem it necessary. Finally, for the reasons aforementioned, families, while knowing that autopsy may be necessary, may rather elect for doing nothing to preserve the body of the deceased.
On the other hand, the relatively high request rate for autopsy in the general public may be taken as surprising given the overall low rate of autopsy in Japan.
One way to interpret these results is through the lens of physician trust (or lack thereof). Physician trust in Japan is perhaps greater than that in many Western countries but has been eroded in recent years because of a series of well-publicised medical errors beginning in the 1990s.27 It may be that, in the specific scenario related to medical error, patients are less trusting of physicians and want autopsy as mentioned above to confirm that what the physicians/hospitals have determined about patient death is correct.
I believe autopsy should not be performed, but would I still request an autopsy?
Those of the general public who do not think autopsy necessary actually tended to stick to their guns and the majority would not seek an autopsy. Physicians, on the other hand, were much more likely to recommend an autopsy even when they thought it unnecessary. Again, this may reflect that the physicians feel compelled to request autopsy even when they feel it unnecessary.
The future of autopsy: project for investigation of death associated with medical practice
In Japan, ‘The project for the investigation of death associated with medical practice’ is currently underway.9 ,10 ,28 This project is a third-party system that has been piloted in certain areas since September 2005. A third party looks into deaths to find errors, and works specifically to prevent similar errors from occurring.
Under the Model Project, family consent is required in cases where autopsy is deemed necessary. In order for the model to be effective, an understanding of patient knowledge of autopsy is necessary, and in some cases, it may be essential to raise the families’ awareness of such issues. This study, highlighted the lack of understanding of the necessity of autopsy by the general public was revealed by their responses given to the scenario 1 and 2. The results of this research will help determine how the model should be developed, particularly in the educational field.
Limitations of this study
This study has a number of limitations. First, the survey nature of the study has many inherent limitations. Second, we have few data about non-responders, which may bias our results. Third, lack of data about education level is unfortunate. While the education levels probably reflect the general position in Japan where >90% of citizens graduate from high school and many go on to advanced education, this is conjecture and a major limitation. Finally, we did not specifically ask about the reasons why participants would or would not pursue autopsy. This additional information would provide insight into how to achieve better autopsy rates and suggest how practitioners might best broach the subject of autopsy to grieving families. The generalisability of these findings beyond members of the general public who participate in survey research and teaching hospitals is also not proved. These could all be topics of future research.
This cross-sectional descriptive study on attitudes to autopsy in various clinical scenarios involving medical error shows that the general public are most likely to believe autopsy should be performed and to request autopsy in situations where medical error is present, regardless of whether or not the presence of medical error is clear or unclear. The general public are less likely to believe autopsy is necessary or pursue autopsy than physicians in the scenarios presented. These results provide a starting point for research into attitudes about autopsy in the case of medical error and further research is needed.
The authors thank Professor Ehara (Graduate School of Health Care Sciences, Jikei Institute) for valuable discussions. We also thank Igaku-Shoin (the rights holder) for permission to use figure 1.
Contributors EK, SM made an equal contribution, participated in the statistical analysis and contributed to the manuscript. All authors conceived and designed the study and read and approved the final manuscript.
Competing interests None.
Ethics approval Kyushu University institutional review board.
Provenance and peer review Not commissioned; externally peer reviewed.