The impact of autopsy participation on clinical residency

Abstract Background Autopsy has had an essential role in ensuring the quality of education and medical care. However, its role in clinical residency has not been clarified. This study assessed actual autopsy circumstances during clinical residency and evaluated the association between autopsy and clinical knowledge. Methods We conducted a cross‐sectional study involving postgraduate second year residents in Japan who took the General Medicine In‐Training Examination in 2019. We modeled the General Medicine In‐Training Examination scores of the residents to examine their association with autopsy experiences and the number of autopsy experiences to assess its predictors. Results Of 2715 postgraduate second year residents, 353 (13.8%) had no autopsy participation, and 1015 (39.7%) had only one experience. Although autopsy participation was not related to the mean General Medicine In‐Training Examination score, the residents' clinicopathological conference participation, self‐study for more than 60 min per day, and wish to be pathologists were significantly associated with autopsy experiences. They experienced more autopsies when they belonged to small‐sized hospitals in rural areas performing many autopsies. Conclusion We reported the current status of autopsy in clinical residency and showed that more than half of the residents experienced no or only one autopsy. General Medicine In‐Training Examination scores were not correlated with the number of autopsy experiences.


| INTRODUC TI ON
Historically, autopsy has made an invaluable contribution to medicine, from understanding novel illnesses to answering patient management issues and maintaining quality of medicine. The value of autopsy in detecting diagnostic errors has been demonstrated. [1][2][3][4][5] In a retrospective review of 2 year records from an educational hospital, 34% of autopsy cases had an unexpected pathological diagnosis leading to death, and 93% of the physicians who attended the autopsies rated them as being a valuable educational experience. 6 However, as a global trend, the autopsy rate has been continuously decreasing. 7 In Japan, it was 5.2% in 2012, and an autopsy was performed in approximately half of all in-hospital deaths in the 1960s. 8 In Japan, residents were required to participate in autopsies and clinicopathological conferences (CPCs) since the new postgraduate medical education (PGME) program was introduced in 2004. 9 Residents must attend the autopsy explanation to the bereaved families, the autopsy itself, and CPCs to understand the pathophysiology in detail during their 2 years of clinical residency training. The purpose of PGME is training physicians with a holistic approach and acquiring primary care skills. This training, from the autopsy explanation to CPCs, is considered a part of this holistic approach. In a survey for each teaching hospital conducted jointly by the Japanese Society of Pathology and the Japanese Society of Internal Medicine, residents learned autopsy explanation, including its permission, by observing their supervising physicians. 10 In addition, the survey has found that autopsy education has become an on-the-job training.
How many autopsies residents experience and how effective autopsies are in the PGME program are unclear.
During these 16 years, we have verified the new PGME program using the General Medicine In-Training Examination (GM-ITE), which was developed using a similar methodology for developing the Internal Medicine ITE (IM-ITE) in the United States. 11 The purpose of the GM-ITE (the same as that of the IM-ITE) is providing residents and program directors with an objective, reliable, and valid assessment of each resident's clinical knowledge in a multiple-choice examination, and the mean scores of each program were compared with those of their peers. 12,13 We have been examining the characteristics of residents regarding their clinical knowledge using a questionnaire survey conducted at the same time as the GM-ITE. In our previous studies, we have reported the characteristics of residents with more excellent clinical knowledge, such as those with appropriate emergency department and inpatient caseloads and those working in provincial community hospitals with many beds. [14][15][16][17] We hypothesized that autopsy would be one of the factors associated with clinical knowledge because autopsy has become a compulsory part of PGME. Therefore, we used the GM-ITE and a concurrent questionnaire survey to clarify the actual circumstances of autopsy among residents and evaluated the association between their autopsy experiences and clinical knowledge.

| Study design and study population
We conducted a cross-sectional study involving postgraduate second year (PGY-2) residents of 441 teaching hospitals in Japan who took the GM-ITE in 2019. The participants were trainees of 2 year postgraduate rotation training programs, including internal and emergency medicine, required for all residents regardless of their specialty before entering the specialty training programs (PGY-3 or later). All participants provided informed consent, which was obtained under the opt-out agreement. The residency program directors were required to assemble residents in a room at each hospital at a scheduled time and administer the GM-ITE to their residents. Then, each program director collected the completed examination answer sheets and sent them back to us in an envelope we provided. Since the academic calendar in Japan starts on April 1 and ends on March 31 of the following year, the GM-ITE was conducted in February or March 2019. Immediately after the test, we provided a self-reported questionnaire sheet regarding the residents' autopsy experiences (i.e., the number of autopsies experienced, autopsy participants when they were in charge, and CPC participation). In addition, the sheet included the question whether residents wished to be pathologists.
We collected the number of autopsy cases and deaths at each hospital from the Annual of the Pathological Autopsy Cases in Japan by the Japanese Society of Pathology. This database has registered all autopsies performed in Japan since 1960. We used the data from 2017 (11,089 cases registered from 808 hospitals). In addition, we obtained additional data, including whether it was a university hospital, whether it was located in an urban area, how many beds it had, and whether it had a general medicine department, from the website of each hospital.
This study was approved by the Institutional Review Board of Mito Kyodo General Hospital, Mito City, Ibaraki, Japan.

| General Medicine In-Training Examination
In Japan, since the new PGME program was introduced in 2004, the knowledge. The maximum and minimum scores for the examination were 60 and 0, respectively, with higher scores indicating a better performance of the general medicine knowledge base. Before conducting the examination, a question review was organized, and the content validity was confirmed by the peer review of each committee member.

| Statistical analyses
We summarized the residents' (resident-level variables) and workplace (hospital-level variables) characteristics using the self-reported number of autopsy participation during their clinical residency programs. The purposes of our analysis were as follows: (a) to assess the association between GM-ITE scores and residents' autopsy ex- However, note that because the highest category of the number of autopsy experiences was censored seven times (i.e., outcome values range from 0 to 7), estimates may include the bias from the "ceiling effect." Hence, we conducted sensitivity analyses. First, we excluded residents who participated in more than seven autopsies (n = 36). Second, the number of autopsy participation was censored using the discrete-time hazard model (with the robust variance estimator clustering hospitals). The model was interpretable as a continuation-ratio logit model for ordinal categories (i.e., the numbers of autopsy experiences) and provided estimates of common odds ratios for the probability of taking several outcomes (autopsy participation) among the residents with more than or equal to that number. Tables S1 and S2 show the results of the sensitivity analyses.
All analyses were conducted through SAS version 9.4 (SAS, Inc.).

| RE SULTS
A summary of the baseline characteristics is shown in Table 1 The mixed-effect model results showed that an increasing number of autopsy experiences were not associated with higher GM-ITE scores ( Table 2). Significant variables associated with a more excellent GM-ITE score were internal medicine rotation for 11 - The results of the log-linear and linear models showed that the more residents participated in CPCs, the more they experienced autopsies (

| DISCUSS ION
To our knowledge, this is the first study that investigated the actual autopsy conditions during a 2 year clinical residency in Japan.
More than half of the study participants experienced no or only one autopsy per year. 18 No correlation was observed between the number of autopsy experiences and GM-ITE score. Based on these two facts, autopsy experience does not significantly affect residents' clinical competence. With the current low autopsy participation, autopsy is not beneficial for improving residents' clinical competence.
Hospital characteristics affecting whether residents' participation in many autopsies were those located in rural areas, with a small number of beds, and performing a large number of autopsies. The former two characteristics, located in rural areas with a small number of beds, are also related to high GM-ITE scores. 14,17 Although we could not verify the relationship between autopsies and the clinical knowledge of residents, autopsy experience might be a factor ensuring the quality of clinical residency at rural hospitals.
Resident characteristics associated with participation in more autopsies were longer study time, more CPC experiences, and pathologist aspirants. In addition, longer study time is also related to      d "Autopsy participants in which residents were in charge" variables were removed due to a strong correlation with "Autopsy experiences" variables.
e "CPC participation" variables were categorized in the same way as "Autopsy participations" variables.

TA B L E 2 (Continued)
TA B L E 3 Generalized estimating equations for linear and log-linear models for mean number of autopsy participations among all residents (the number of autopsy experiences was censored at 7) a  19 Moreover, this attitude leads to the ideal physicians setting out in the Japanese PGME. 9 The reported causes for the decrease in autopsy rate worldwide are cause of death determined before death, no financial support to the pathologist for autopsy procedures, fear of medical malpractice, and changes in public awareness toward autopsies. 2,3,7,20,21 According to the questionnaire survey, 80% of attending physicians in Japanese teaching hospitals were educating about autopsies.
Some requested the redefinition of an autopsy. 10 We found that several highly motivated residents, such as those with more CPC participation, experienced more autopsies. The improvement of a CPC integrating clinical medicine and pathology does increase the number of autopsy participation. 10 Moreover, an autopsy may support clinical competence if residents can actively participate in their patients' autopsies.
This study has several limitations. First, this study involved a small sample size. Only 2254 of the 8489 PGY-2 residents in Japan participated in this study. 22 Their program directors were responsible for the decision to participate in the GM-ITE. PGY-2 residents taking the GM-ITE, who participated in this survey, were analyzed. There might have been a sampling bias in which the highly motivated teaching hospitals might have involved more participants in the GM-ITE.
However, the number of residents who took the GM-ITE increased, and only 341 residents (11%) who took the GM-ITE were denied participation in this survey. Second, resident characteristics, such as emergency room duty per month, number of inpatients handled, and number of autopsies experienced, could be influenced by recall or cognitive bias, since it was a questionnaire study. We used the term "on average" in the questionnaire for these items over the 1 year before the test date or the 2 year clinical residency. Third, the number of autopsies performed in each hospital used in this study was from 2017. Although we could not use the 2018 data, we thought that the number of autopsies at each hospital rarely changed.
In conclusion, we reported the current status of autopsy in clinical residency in Japan based on this questionnaire study involving PGY-2 residents. We found that more than half of the residents experienced no or only one autopsy. GM-ITE scores were not correlated with the number of autopsy experiences. Further study is required for improving autopsy training during clinical residency.

ACK N OWLED G EM ENTS
We thank Norio Takayashiki for contributing research idea and constructive pathological advice. We also thank Non-Profit Organization iHOPE International, Kyoto, Japan, for contributing study design.
Finally, we special thank the Japanese Society of Pathology for providing us Annual of the Pathological Autopsy Cases of 2017.

YN and YT participated on the General Medicine In-training
Examination project committee of the Japan Organization of