Context.—Autopsy rates have been declining throughout the world, although preservation of the autopsy is considered a fundamental principle of medical care. In France, the 1994 bioethics law requires physicians to inform relatives before performing an autopsy.
Objective.—To analyze the following factors that potentially influence hospital autopsy rates: legal constraints, autopsy reporting times, opinions of physicians requesting autopsies and pathologists regarding the usefulness of autopsy in patient care, and use of autopsy material in research publications.
Design.—Record of the annual numbers of deaths and autopsies during a 10-year period (1988–1997). Record of the delays for transmission of final autopsy report to the requesting physician. Questionnaire analyzing the possible factors influencing autopsy rate. Categorization of articles published by pathologists according to the use of autopsy material.
Setting.—A 1000-bed, university teaching hospital in the Paris, France, area.
Participants.—Questionnaire addressed to physicians, head nurses, and mortuary staff.
Results.—A total of 1454 autopsies were reviewed. The autopsy rate declined from 15.4% in 1988 to 3.7% in 1997. This decline was marked after 1994 and tended to be slower for neurologic indications than for other indications. The final report had not been communicated within 180 days in 620 (42.6%) of 1454 autopsies. Fifty-five of 105 respondents considered that the bioethics law was one cause of the recent decrease of autopsy rate. Considering the contribution of autopsy to medical research, 94 (81%) of 116 articles dealing with central nervous system but only 28 (6%) of 464 articles dealing with other organs used autopsy-derived material.
Conclusions.—The 1994 bioethics law seems to contribute to the decline of autopsy. Inadequate delays for communicating autopsy results are frequent. Except for neuropathologists, autopsy is a minor source of research material.
Autopsy rates have been declining throughout the world during the last 30 years.1–3 Numerous reasons have been suggested. Some physicians hesitate to ask for the relatives' consent or fear that the autopsy would reveal misdiagnosis.4,5 Most pathologists became reluctant to perform autopsies for fear of an infectious risk and because they feel that many autopsies are unuseful, considering modern diagnostic technologies.4,6,7 Inadequate delays in communication of autopsy results by pathologists may take part in the decline of autopsy rates.8,9 Whether long autopsy reporting times are the rule or the exception has been seldom studied.2,10 In France, one additional factor might be the recent change in the law governing autopsy permits.11 Before 1994, autopsy could be performed without informing the family, except if it had been formally refused by the patient before death or by relatives after death. Since the 1994 bioethics law,12 physicians must inform relatives that “postmortem samples are taken,” which implies that they have to inform the relatives of their intention to perform an autopsy and ask if the patient had refused autopsy. Practically speaking, the relatives' consent is now needed to perform an autopsy.13
Preservation of autopsy is often claimed a fundamental principle of medical care and quality control in medicine.14,15 The still persisting high discordance rate between clinical and autopsy diagnoses is also given as confirming the value of postmortem examination by its defenders.16–19 These defenders also claim that autopsy is a valuable tool for detection and evaluation of emerging diseases.15,20 Medical research could be an additional incentive to perform autopsies,1,8,21 but only limited quantitative information is available on the contribution of autopsy findings to medical research. One study indicated that the proportion of autopsy-based research papers increased between 1966 and 1993.1 In brain pathology, where biopsy is a minor source of pathological material, a higher contribution of autopsy findings to medical research could be expected but has not been determined.1
Studies conducted about autopsy usually address either the requesting physicians' or the pathologists' point of view.2,10,22 In our study, we asked both requesting physicians and pathologists from one hospital the same questions about their view of autopsy. Our objective was to analyze some factors that potentially influence the decline of hospital autopsy rates: recent legal constraints, autopsy reporting times, opinions of requesting physicians and pathologists regarding the usefulness of autopsy in patient care, and use of autopsy material in research publications.
Analysis of Autopsy Rates
We recorded the annual numbers of deaths and autopsies during a 10-year period (1988–1997) at Henri-Mondor Hospital, a 1000-bed university teaching hospital in the Paris, France, area.
Analysis of Autopsy Request and Communication of Autopsy Reports
For each autopsy performed during the period studied, we noted the grade of the physician writing the autopsy request (head of department or senior medical doctor, registrar, or resident) and the existence of a specific question in the request. We recorded the delay for transmission of final autopsy report. In all cases, tissue samples were taken for histologic examination, but histologic examination was not systematically performed. Final autopsy report turnaround time was evaluated in 4 situations: (1) autopsies including a macroscopic but no histologic examination of central nervous system (CNS), (2) autopsies including both macroscopic and histologic examination of CNS, (3) autopsies without CNS examination but with histologic examination of some samples from other organs, and (4) autopsy without CNS examination and without any histologic examination.
We also noted the autopsies for which no final report had been transmitted to the requesting physician on December 31, 1998. In evaluating the median delay for communication of autopsy reports, we did not consider autopsies for which no final report had been transmitted. To consider all autopsies, we categorized the final autopsy turnaround times as follows. We categorized the turnaround times for each autopsy, including examination of organs other than CNS, as less than 30 days, 30 to 59 days, 60 to 89 days, or 90 days or more. Because of the mandatory need for brain fixation during 3 to 4 weeks,23 we categorized the turnaround times for each CNS examination as less than 60 days, 60 to 89 days, 90 to 179 days, or 180 days or more. Autopsies for which no final report had been transmitted were categorized with those reported after 90 days or more (organs other than CNS) or 180 days or more (CNS). We evaluated the rates of autopsies for which the final report (CNS or other organs) had not been communicated within 180 days.
To analyze the possible factors that influence autopsy rate, we addressed a questionnaire to the physicians and head nurses from hospitalization departments, pathologists, and mortuary staff. The questionnaire was divided into 5 headings: the 1994 bioethics law, autopsy request, the interest of autopsy, the performance of autopsy, and causes of autopsy refusal by relatives (Table 1). The questionnaire ended with a suggestion for adding open comments.
Contribution of Autopsy to Medical Research
We collected all articles published by pathologists from our hospital from 1988 through 1997. We used the MEDLINE database and considered only research articles. We considered 3 categories of articles according to the organ involved, namely articles focusing on CNS, other organs, or both. In a second classification, articles were categorized according to the use of tissue specimens obtained at autopsy as part of the material used for the study. We categorized the articles using the abstracts, and, when necessary, we asked the authors to classify their work correctly.
Differences between groups were assessed using the χ2 and Fisher exact test. Correlations were evaluated using Spearman correlation coefficient. All tests were 2-sided and conducted at the .05 significance level.
To characterize the evolution with time of the autopsy rate, 3 models were considered: (1) constant linear decrease from 1988 to 1997; (2) linear decrease from 1988 to 1993 and from 1993 to 1997, with distinct slopes; and (3) constant rate from 1988 to 1993, followed by linear decrease from 1993 to 1997. To assess whether the rates of autopsies concerning CNS and those concerning other organs decreased with different slopes during 1993 through 1997, 2 models were fitted to the data. The first model assumed parallel decline, whereas the second model allowed for different slopes. Models were compared using the likelihood ratio test, assuming gaussian errors with zero mean and constant variance.
Autopsy Rate During the Surveyed Period
A total of 1454 autopsies have been performed between 1988 and 1997 at Henri-Mondor Hospital. The annual number of deaths was stable (range, 1208–1308; median, 1270). The autopsy rate, defined as the ratio of the number of autopsies to the total number of deaths, declined from 15.4% in 1988 to 3.7% in 1997) (Figure 1). The number of autopsies declined from 198 in 1988 to 45 in 1997. For the evolution in time of the autopsy rate, the best model was model 3, which assumes that the rate of autopsy was constant from 1988 to 1993 and then decreased linearly from 1993 to 1997 (model 3 versus model 1, P < .005; model 3 versus model 2, P > .90). The likelihood ratio test statistic for the hypothesis of a common slope for the decline of CNS autopsy versus autopsies concerning other organs yielded a P value of .053 (Figure 2).
Analysis of Autopsy Request and Communication of Autopsy Reports
The median rate of requests containing a specific question was 72.7% (range during 10 years, 50.7–78.5). It did not vary significantly over time (r = −0.41, P = .24). Data for median autopsy reporting times are reported in Table 2. No autopsy report had been transmitted for 130 (8.9%) of 1454 autopsies. The annual percentage of autopsies without reports transmitted (median, 8%; range, 4%–31%) did not correlate with the number of autopsies performed. Data for categorized final autopsy reporting times are reported in Table 3.
For CNS autopsy reports, the median delay for communication increased and the percentage communicated within 180 days decreased over time both for macroscopic (r = 0.65, P = .05 and r = −0.78, P = .008) and histologic (r = 0.73, P = .02 and r = −0.75, P = .01) examinations. For reports concerning organs other than CNS, the median delay for communication and the percentage communicated within 90 days did not correlate with calendar years both for macroscopic (r = 0.47, P = .17 and r = −0.20, P = .58) and histologic (r = 0.05, P = .89 and r = −0.17, P = .67) examinations. The global rate of autopsies for which the final report had not been communicated within 180 days was 620 (42.6%) of 1454. Annual rates of such cases (range, 57 [30.3%] of 188 and 113 [55.4%] of 204) did not correlate with calendar years (r = 0.36, P = .31).
A total of 117 members of the hospital staff were asked to complete a questionnaire between March and April 1998. Twelve persons (10.3%) did not answer; 105 completed questionnaires were collected from 14 departments. The respondents were as follows: 12 heads of department, 29 senior medical doctors, 30 registrars, 21 residents, 11 head nurses, and 2 mortuary staff members. In some cases, all questions were not answered, for example, when the respondent had no opinion or when the question did not make sense to him or her (eg, pathologists for a question about the relatives' information).
The first section dealt with questions about the 1994 bioethics law. Only 17 respondents (16%) considered that their knowledge of the 1994 bioethics law was precise, 62 (59%) considered that it was not, and 26 (25%) did not know anything about this law (Table 4). Table 5 shows the responses to other questions from this heading. Most respondents regarded the bioethics law as a cause of the recent decrease of autopsy rate.
The second section concerned questions about autopsy request. Responses regarding what prevents physicians from requesting autopsies are shown in Table 5. Responses about the difficulty of informing relatives varied according to the grade of the respondents (Table 6). Most clinicians considered that autopsy was always justified and that autopsy request was sufficiently precise. But, for both questions, pathologists' opinions differed significantly (Table 7). Only 10% of all respondents considered that the possibility for autopsy to allow demonstration of medical malpractice hindered physicians from requesting autopsies.
The third section involved questions about the interest of autopsy. Almost all respondents claimed that autopsy was still useful (Table 5). Opinions about autopsy reports are shown in Table 5. The opinions about the delay for verbal communication of results varied according to the grade of the respondents (Table 8). Most clinicians complained about the delay for obtaining written autopsy reports.
The fourth section contained questions about the performance of autopsy (Table 5). Most pathologists considered that the risk of blood exposure hindered the performance of autopsies and complained about the conditions in which the autopsies are performed (Table 7). Most pathologists but only a few other respondents were satisfied with the current decrease of autopsy rates (Table 7).
The fifth section dealt with causes of autopsy refusal by relatives. For the respondents, the main reasons that could lead relatives to refuse an autopsy were the fear of a disrespectful handling of the corpse, religious objections, and moral suffering in the context of death (Table 9).
Of the 105 respondents, 63 (60.0%) added open comments to the questionnaire. Points most frequently raised included the lack of communication about autopsy between requesting physicians and pathologists (11 [17.5%] of 63), the insufficient motivation of pathologists (8 [12.3%] of 63), the insufficient motivation of requesting physicians attested by their physical absence during autopsy (6 [9.5%] of 63), the inadequate delay for communicating autopsy results (6 [9.5%] of 63), and the need for a definite policy regarding the indications of autopsy (6 [9.5%] of 63).
Contribution of Autopsy to Medical Research
Of the 641 articles collected, we considered only the 568 research articles. We identified 116 CNS-based articles and 464 articles based on other organs. Twelve articles dealing with both CNS and other organs were counted in both pools of articles. The overall proportion of articles using autopsy results or material obtained at autopsy was 111 (20%) of 568. Ninety-four (81%) of 116 articles dealing with CNS used autopsy-derived material. Twenty-eight (6%) of 464 articles dealing with other organs used autopsy-derived material. The difference between the 2 proportions was significant (P < .001, Fisher exact test). The annual proportions of articles using tissue specimens obtained at autopsy did not correlate with calendar years (Figure 3).
Autopsy rates at Henri-Mondor Hospital declined during the last 10 years, which is in accordance with what has been documented in the United States (a decrease from 41% in 1964 to 10.3% in 19922,8) and in England and Wales (a decrease from 8.9% in 1966 to 1.7% in 19911). However, an additional factor in France is the introduction in 1994 of a new regulation that makes it necessary for physicians to inform the families that postmortem samples are taken.12,13
Figure 1 shows the sharp decrease between 1993 and 1994, followed by a further decrease, since final modalities of the law were mostly promulgated in 1997. Half of the respondents to the questionnaire recognized that this new regulation was a cause for the decline in autopsy rate, although only one third claimed that their behavior was modified. It is not clear from this study if physicians maintained the same level of requests and if they had more refusals from the families or reduced by themselves the number of autopsy requests. Interestingly, the residents, who have to inform the family, feel this is much more difficult to do than heads of departments, who probably rarely do it themselves (Table 6). Residents also feel that their knowledge of the bioethics law is only vague (10 of 21) or nonexistent (11 of 21) (Table 4). The main reason for family refusal of autopsy, as expressed by the respondents, was the disrespectful handling of the corpses, a finding similar to that of direct evaluation of public or medical students' discomfort at the thought of an autopsy.24,25
The rapid and obvious decline of autopsies performed at Henri-Mondor Hospital contradicts the conventional affirmation by nearly all respondents (97%) that autopsies are still “useful.” However, if the respondents really believe this, an intriguing question is why, faced with a new legal obligation, clinicians and pathologists were not able to maintain the level of autopsies before 1994. From the answers to the questionnaire, it is clear that the reasons are not the same for these 2 populations of physicians. For clinicians, a main reason for the obvious disinterest that was expressed in the questionnaire was the long responding time, since shortening those delays has been reported to contribute to higher autopsy rates.4,8,9,26,27 Most requesting physicians were not satisfied with the delay for obtaining written autopsy reports. Of particular note is the significant percentage of autopsies (8.9%) for which no report had been transmitted to requesting physicians. Hospital accreditation standards for laboratories in the United States require the final autopsy report to be on the medical record within 60 days.2 In about half the cases, final autopsy reports were transmitted after more than 60 days. The observed autopsy reporting times are higher than those observed in the United States and Great Britain.2,10 Faced with difficulties in obtaining the consent from the families, clinicians may be discouraged in requiring autopsies because they are not expecting results before months. An additional symptom of disinterest of clinicians for autopsies is their absence during the procedure, as mentioned by several pathologists in their free comments. Requesting physicians also indicated that they were not “satisfied with the content of autopsy report,” which indicates that they are not expecting too much from postmortem examination. This is at odds with most articles on autopsy, which invariably document findings that were not known by the attending physician before death.15–19 However, in the era of evidence-based medicine, it has not been demonstrated that informing clinicians about such findings did modify their clinical practice. Fear of documenting errors or mismanagement at autopsy does not seem to be determinant, since malpractice suits in France are still uncommon. Finally, high autopsy rates are not required in the accreditation standards that are presently implemented in France.
As to the pathologists, the questionnaire identified several key factors. The motivation of pathologists for autopsies can be questioned from the excessive delay in reporting results. Moreover, the decreasing number of autopsies during the period we studied was not associated with decreasing reporting times, unlike what has been documented in the United States,2 which suggests that those delays are not directly related to the workload induced by autopsies. A first factor of reluctance to perform autopsies, which our questionnaire documented, is the fear of contracting an infectious disease, as admitted by 10 of 14 pathologists. A second factor is the low contribution of autopsies to the scientific production of the department. Our MEDLINE survey showed that only 6% of articles published during the 1988–1997 period by the department of pathology (excluding neuropathology ) were based on material obtained at autopsy (Figure 3). Such a situation explains, at least in part, the lack of interest for autopsies by these academic pathologists. Pathologists also indicated that they were “not satisfied of the conditions in which they had to perform autopsies”; only 5 of 14 pathologists found that the clinicians' request was adequately formulated. Moreover, only 1 (7%) of 14 pathologists but 63 (69%) of 91 other respondents found that autopsy was “always” justified. In addition, pathologists commonly note that requesting physicians are not present during the procedure and that, unlike what happens with biopsy or cytology samples, they usually do not ask for autopsy results. Pathologists also stress that the sharp decline in autopsy rate will induce loss of expertise. Finally, the clear message from the Henri-Mondor pathologists (10 of 12 respondents) was that they were “satisfied” with the current decline of hospital autopsy rate in the conditions in which they are currently performed.
The particular situation of neuropathology underlines the coexistence of multiple internal and external factors that might cause the decline of autopsy. With the exception of brain tumors that were removed or had undergone biopsies by neurosurgeons, medical research in brain pathology needs autopsy-derived material.28,29 Indeed, we found that most research articles published by neuropathologists used autopsy material and that decline of autopsy rate was slower when autopsy was performed for neurologic indications than for other indications.
This study has some limitations. First, it relies on a questionnaire, which reports on opinions and may differ from real behaviors. Second, this study was performed in a single hospital, and this may possibly not accurately reflect the situation in France. However, available data show that the decline of autopsy rate documented in Henri-Mondor Hospital is similar to that observed in other French academic institutions (Assistance publique—hôpitaux de Paris, unpublished data, 1998). In addition, such an extensive survey can be performed only in 1 or a limited number of institutions.14,16
In conclusion, this study confirms the decline of autopsy rate in our institution. Faced with recent legal constraints, physicians do not seem to have increased their efforts to explain to the families the reasons for performing autopsies, apparently because they feel discouraged by long reporting times and low informative value of reports. Pathologists, except neuropathologists, do not complain about decrease of autopsy rate. They feel that autopsies are a low priority and are not stimulated by requesting physicians, who are usually not present during procedure and rarely express interest in final results. It is clear that, in our institution, except for neuropathologists, autopsy material is no longer the basis of major scientific production. Paradoxically, a renewal of autopsy in future years might result from mentality changes due to the recent law. Motivation of both requesting physicians and pathologists could improve, because indications will be more debated between clinicians and pathologists and reports will be transmitted more quickly, since family is now informed of the procedure and frequently asks for results. The autopsy procedure will no longer be hidden from relatives and will possibly have the same value as pathological examination of samples obtained from living patients, for which communication between requesting physicians and pathologists is of much higher quality.
We are indebted to all participating members of Henri-Mondor Hospital staff.