Consecutive autopsies on an internal medicine service.

Objectives: Autopsy rates continue to decline in the United States despite the demonstrated value of this procedure in many different settings. We sought to review clinical pathologic discordance information generated by autopsies on an internal medicine service in the urban United States and to determine whether resident services appear to influence autopsy rates.

Methods: We reviewed consecutive deaths and autopsies on an inpatient internal medicine service during a 30-month period at a 400-bed community hospital in Baltimore, MD.

Results: There were 622 deaths and 65 autopsies (10.3%). Resident teaching status correlated with a higher rate of autopsies performed (P = 0.048). Clinical pathologic discordance was common, with a major discordance rate of 39%. Major discordance was indicated by only one of nine autopsies performed on patients with human immunodeficiency virus.

Conclusions: The autopsy was a valuable educational and quality improvement tool on the urban internal medicine service. Residency influences may be a major factor in continuing this exercise. In our study, although the numbers were small, patients with human immunodeficiency virus had a very low discordance rate.

Key Words: autopsy rate, clinical pathologic discordance, internal medicine residency programs

**********

Autopsies should be essential components of performance-improvement activities in hospitals. (1,2) They are effective teaching instruments, and are crucially important in a practice-based learning environment. Studies have consistently demonstrated the importance of autopsies in determining correct pathologic diagnoses and in identifying clinical pathologic discordance. (3-9)

Despite the evident value of autopsies, there has been a persistent decline in autopsy rates, particularly in the United States. (10) The current US autopsy rate may be as little as 5% of all hospital deaths. (11) A pragmatic adjustment to the declining autopsy rate has been made by the Accreditation Council for Graduate Medical Education. It has removed specific minimum autopsy rates from the special requirements. (12) No longer is "a less than 15% autopsy rate" one of the most frequent educational citations in internal medicine training programs. The current requirement has eliminated a number; replaced by "autopsies performed whenever possible".

The objective of our study was to review autopsies in an urban hospital internal medicine service, and to include all medical inpatients in order to determine whether certain variables may influence the clinical pathologic discordance rate. Most recent autopsy reviews have been published outside the United States (1,3,5,8) or have focused on very specialized services such as restrictive critical care units and transplant centers. (4,6,7)

Materials and Methods

This study was conducted at Union Memorial Hospital, a 400-bed acute care institution in central Baltimore, Maryland. The hospital has a very diverse patient population. We studied records of all deaths on the internal medicine service between January 1, 1999, and June 30, 2001. Deaths were identified from the hospital registry. All patients assigned to the internal medicine service at the time of their death were included, including intensive care, telemetry unit, cancer unit, geriatric service, and general medicine wards. The only medical patients excluded were those who died in the emergency room shortly after admission and had not yet reached the inpatient unit.

A chart and case review team consisted of residents and faculty from the department of medicine. Either a resident or a medical student performed data extraction. Information was collected on patient demographics, whether they were admitted to the resident teaching service, the performance of cardiopulmonary resuscitation (CPR), and clinical findings at the time of death. Those admitted to the resident service were designated "teaching patients," as opposed to "nonteaching patients," who were those cared for exclusively by their private physicians with or without consultants and the assistance of board certified "house physicians." Clinical information collected included clinical diagnosis at the time of death as ascertained by a consensus of our review team, including senior faculty members. The review team was blinded to the autopsy results.

A review of the autopsies by a senior member of the pathology faculty followed. Subsequently, a consensus was achieved regarding concordance and discordance. Discordant autopsies were categorized as "major class I," "major class II," or "minor." (4) Major class I discordance included a missed diagnosis that would have had a direct impact on therapy or prognosis. Class II included major unexpected findings that would have been unlikely to have affected therapy or prognosis. Minor discordances were incidental findings of interest, but without management significance. The Fisher exact test was used to test for statistical differences between study groups.

Results

During the 30-month study period, 18,806 patients were discharged from the Union Memorial medical service; of these, 622 had died, a 3.3% death rate. Records were available for review on 592 (95.2%) of the deaths. Thirty records could not be located at the time of the study. None of these thirty cases had autopsies. Of the reviewable cases, 50.7% were males and 49.3% were females, 51.6% were black, 48.8% were white, and 439 (74.1%) of the 592 were teaching. Mean age was 70 with a range of 17 to 98. CPR was performed on 131 patients (22.1%). The rate of CPR performed on teaching cases was 25.3% compared with 13.1% for nonteaching (P = 0.002). There were 64 total autopsies, 10.3% of all deaths. The autopsy rate was 12.3% on the teaching service compared with 6.5% on the nonteaching service (P = 0.048). Forty-one patients (6.9%) were positive for human immunodeficiency virus (HIV). HIV-positive cases represented 38% of deaths among those aged 50 and under during the study period. There were five patients (7.8%) who had major pulmonary emboli at autopsy, three of which were unexpected. Table 1 shows major class I, major class II, and minor discordance rates. The rate of major discordances was 39.1%. There were insufficient numbers to statistically correlate sex, age, race, teaching status, and other variables with discordance rates. However, it is noteworthy that only one of nine autopsies on HIV-positive patients had a significant major finding, and that finding was a class II. The clinical details of the major discordance cases are listed in Table 2.

Discussion

Over the last 15 years, autopsy studies have reported discordance rates between 29 and 49%. (1,4,6-9) Regardless of ever-improving diagnostic technologies, our discordance rate of 39% was consistent with these prior studies and, indeed, suggests that the autopsy continues to be a very important educational tool on general internal medicine services in the United States. Despite this, there are consistently declining autopsy numbers overall. In addition, there do not appear to be reliable ways to target certain types of cases for autopsy. (1,13) Thus, maintaining a reasonable frequency of autopsies is important.

In our study, teaching patients were twice as likely to receive autopsies as nonteaching patients. We believe this is the first time this correlation has been reported. Autopsies have long been considered an essential component of internal medicine training. A more aggressive pursuit of knowledge that can affect the quality of care is one of the standards that separates the teaching from the nonteaching medical environment. It is likely that autopsy rates would decline even further if not for the pressure exerted by teaching programs.

Our unsuspected pulmonary embolus rate of 7.8% is similar to the total reported in a recent review. (14) One interesting finding was the low discordance rate among HIV-related cases, although the numbers were too small to prove statistical significance. There has been controversy involving the perceived risk of infection from the autopsy procedure in certain infectious viral illnesses. (15) It is our impression that many hospital departments, particularly in nonteaching services, do not encourage autopsies on HIV-positive patients for this reason. A low discordance rate in our study suggests further investigation would be helpful in assessing the risk versus the benefits of autopsies on HIV-positive patients. Attempts to obtain important pathologic information using the apparently lower-risk postmortem biopsy technique (16) may be a reasonable alternative.

Table 1. Discordant cases

                    No.      %

Major Class I (a)   10   15.6%
Major Class II (b)  15   23.5%
Total Major         25   39.1%
Minor (c)           25   39.1%

(a) Missed diagnosis which would have had a major impact on therapy or
prognosis.
(b) Missed diagnoses which were unlikely to have a major impact.
(c) Missed diagnoses that were incidental.

Table 2. Details of major discordance findings

Class I                          Class II

Pulmonary embolus                Squamous cell of lung with
Osteomyelitis, aortic              metastases
  aneurysm                       Pneumonia, Cerebrovascular accident
Myocardial infarction            Cerebrovascular accident
Bladder cancer metastases        Pulmonary embolis, pneumonia
Lung cancer with metastases      Mitral stenosis, aortic stenosis
Hemorrhagic colitis peritonitis  Lung cancer metastases
Hemorrhagic colitis              Colon cancer metastases
Bronchopneumonia                 Hemorrhagic colitis
Vasculitis                       Endocarditis, emboli to brain
                                 Myocardial infarction
                                 Pericarditis
                                 BOOP
                                 Adult Respiratory Distress
                                   Syndrome
                                 Adult Respiratory Distress
                                   Syndrome
                                 Gallbladder adenocarcinoma

Acknowledgments

The authors would like to acknowledge Patricia Reese for her assistance with manuscript preparation, medical students Azitah Zadek and Teddy Greaves for data collection and analyses, and Mohamed Badawi for statistical assistance.

Accepted October 29, 2003.

Copyright [c] 2004 by The Southern Medical Association

0038-4348/04/9704-0335

Please see Ronald C. Hamdy's editorial on page 325 of this issue.

References

1. Boers M, Nieuwenhuyzen Kruseman AC, Eulderink F, et al. Value of autopsy in internal medicine: a 1-year prospective study of hospital deaths. Eur J Clin Invest 1988;18:314-320.

2. Pelletier LL Jr, Klutzow F, Lancaster H. The autopsy: its role in the evaluation of patient care. J Gen Intern Med 1989;4:300-303.

3. Grundmann E, Menke GG. Autopsy diagnosis versus clinical diagnosis, particularly in malignant disease: comparison of two periods: 1961-70 and 1978-87. IARC Sci Publ 1991;112:81-90.

4. Tai DY, El-Bilbeisei H, Tewari S, et al. A study of consecutive autopsies in a medical ICU: a comparison of clinical cause of death and autopsy diagnosis. Chest 2001;119:530-536.

5. Mollo F, Bertoldo E, Grandi G, et al. Reliability of death certifications for different types of cancer: an autopsy survey. Pathol Res Pract 1986;181:442-447.

6. Blosser SA, Zimmerman HE, Stauffer JL. Do autopsies of critically ill patients reveal important findings that were clinically undetected? Crit Care Med 1998;26:1332-1336.

7. Mort TC, Yeston NS. The relationship of pre mortem diagnoses and post mortem findings in a surgical intensive care unit. Crit Care Med 1999;27:299-303.

8. Stevanovic G, Tucakovic G, Dotlic R, et al. Correlation of clinical diagnoses with autopsy findings: a retrospective study of 2,145 consecutive autopsies. Hum Pathol 1986;17:1225-1230.

9. Bayer-Garner IB, Fink LM, Lamps LW. Pathologists in a teaching institution assess the value of the autopsy. Arch Pathol Lab Med 2002;126:442-447.

10. Roberts WC. The autopsy: its decline and a suggestion for its revival. N Engl J Med 1978;299:332-338.

11. Burton EC. The autopsy: a professional responsibility in assuring quality of care. Am J Med Qual 2002;17:56-60.

12. American Medical Association. Graduate Medical Education Directory. Chicago, American Medical Association, 2002, p 89.

13. Friederici HH, Sebastian M. Autopsies in a modern teaching hospital: a review of 2,537 cases. Arch Pathol Lab Med 1984;108:518-521.

14. Pineda LA, Hathwar VS, Grant BJ. Clinical suspicion of fatal pulmonary embolism. Chest 2001;120:791-795.

15. Wilkes MS, Fortin AH, Jacobs TA. Physicians' attitudes toward the autopsy of patients with AIDS. N Y State J Med 1991;91:386-389.

16. Guerra I, Ortiz E, Portu J, et al. Value of limited necropsy in HIV-positive patients. Pathol Res Pract 2001;197:165-168.

RELATED ARTICLE: Key Points

* Autopsy rates continue to decline, and are being deemphasized by the internal medicine accreditation bodies.

* Clinical pathologic discordance rates remain high despite ever advancing diagnostic techniques.

Robert P. Ferguson, MD, Linda Burkhardt, MD, George Hennawi, MD, and Loveen Puthumana, MD

From the Departments of Medicine and Pathology, Union Memorial Hospital, Baltimore, MD.

Reprint requests to Robert P. Ferguson, MD, Department of Medicine, Union Memorial Hospital, 201 E. University Parkway, Suite 405, Baltimore, MD 21218. Email: Robert. [e-mail]