Aecuracy of clinical Diagnosis in a Teaching Hospital :a Study of 1OO Consecutive Medical Autopsies

Rupali R Bavikar+, Arvind G Valand**

Abstract\Ilith the current emphasis on cost effectiveness and quality assurance in healthcare, autopsy results can help evaluate the antemortem diagnosis and medicaltreatments or lntenrentions that the patlent receives, and whether or not thoseintenrentions were appropriate. The aims of this study were to audit autopsies ofdeaths within 48 hours and deaths after 48 hours of hospital admission by comparingantemortem and postmortem dlagnosis. The study sample comprised rooconsecutive autopsies. ln l7o/o autopsies, major discrepancies were noted betweenantemortem and postmortem diagnosis. Patients who died within 48 hours ofadmission had a statistically significant higher chance of having a major discrepancybetween antemortem and postmortem diagnosis. Minor discrepancies were seen in18 cases. ln LLo/o cases, gross autopsy Iindings did not match histopathologiealdiagnosis.

Introduction

I utopsy has been called "the ultimateI lmedical consultation". Autopsies areabsolutely necessary for the progress ofmodern clinical medicine and for solvingmany problems : diagnostic, control oftreatment quality, detection of infections,hereditary and systemic diseases,establishing causes and mechanisms ofsudden death, qualification of uncleardiseases and s5mdromes, classificationand nomenclature of nosological entities.The role of autopsy in teaching andtraining, postgraduate education cannotbe emphasized enough.'

In this study, we examine thediscrepancies between altemortem andpostmortem diagnosis, discuss thecorrelation between gross andhistopathological findings and attempt toevaluate the factors responsible for the*Lecturer in Pathologr, **Prof. of Pathologl, GrantMedical College and Sir J. J. Hospital, Mumbai.

30

declining rates of autopsy.

Material and MethodsThis is a prospective study and

analysis of 10O consecutive medicalautopsies of patients over a period of 12months. During this period, there were atotal of 1558 deaths. A total of ll2 medicalautopsies were performed of which thefirst hundred were included in this study.Of the LOO deaths studied, 26 occurredwithin 48 hours of admission to thehospital while the rest (n=74) occurredafter 48 hours. Clinical impression wasdocumented in 86 cases. Autopsies wereperformed by residents and iecturersunder supervision ofprofessor in charge ofautopsy. Information regarding age, sex,duration of hospitat stay, clinicalpresentation, antemortem investigationresults, clinical diagnosis/differentialdiagnosis, postmortem gross findings,provisional cause of death and final causeof death after histopathological tissueexamination was recorded in a proforma

Bonfiay Hospital Journal,Yol. 53 , No. t ,201I

and analyzed. Major and minordiscrepancies between antemortem andpostmortem diagnosis were analyzed. Amajor discrepancy was defined asdiscrepant primary diagnosis with anadverse impact on survival. A minordiscrepancy was defined as a discrepantsecondar5r diagnosis not directly related tothe cause of death. This also includesoccult or accidental findings and findingsrelated to senile changes or supportingmain cause of death.

Obserations and resultsThe 100 autopsies were classified into

two groups :

Group I : 83 autopsies where theclinical impression matched the finalcallse of death on histopathologicalexaminatisn. Minor discrepancies werenoted in 18 ofthese cases.

Group II : 17 autopsies where theclinical impression did not match the finalcause of death and mqior discrepancieswere present between antemortem andpostmortem.

Each group was again analyzed fordeaths within 48 hours of admission anddeaths after48 hours ofadmission.

In group I death occurred in 12 cases(14%) within 48 hours of admission whilein group II death occurred in 10 cases(59%) within 48 hours of admission.Patients who died .udthin 48 hours ofadmission had statistically significanthigher chance of having a majordiscrepancy between antemortem andpostmortem diagnosis (p <0.O0005, Oddsratio O. 1.2 with confidence interval O.O3 too.421.

Gross autopsy findings did not matchfinal histopathological diagnosis in l4%ocases. 9 of these cases were from group I{11%)while 5 belonged to group II (29%1.

Bombay Hospital Journal, Vol. 53 , No. I , 201 I

The final cause of death was given aftercomplete histopathologic examination ofvarious tissues. Table 1 gives a list of majordiscrepancies between antemortem andfinal (postmortem) diagnosis. The largestTable 1 : Major discrepancies between antemortem

and postmortem diamosis.postmortem

No. of Cases Flnal cause of dcath

7

2

2

2

2

1

1

Pulmonary or disseminated tuberculosis

Cerebral malryia Malignancy

Malignmcy

Acute myocardial inftrction

Submassive hepatic necrosis

Pseudomembrmous enterocolitis

Acute haemorhagic pmcreatitis

single category of unsuspected diagnoseswas undiagnosed infections liketuberculosis and malaria followed byacute myocardial infarction andmalignancy.

DiscussionAs early as 1912, Cabot recognized the

value of autopsy in detecting commonerrors in clinical diagnosis. Several morerecent studies have confirmed its merit inmodern medical practice.2'3'4 All haveshown that despite many advances indiagnostic techniques a significantnumber of diagnoses are either notconfirmed at or not suspected untilautopsy. The overall figure of 35% fordiscrepant diagnosis (major 17 and minor18) is comparable to the figures quoted inthe literature which range from 7o/o lo39.7%. Problems the clinicians have notrecognized, false positive and negative stilloccur frequently.

Infectious diseases like tuberculosisand malaria continue to be the commonestdiseases that remain undiagnosed andwere found to be the most important majordiscrepancy between antemortem andpostmortem diagnosis. This is clearly in

31

contrast to studies reported from thewestern world where pulmonarlr embolismand malignancies remain the major causeof discrepancies in antemortem andpostmortem diagnosis.t

Hospital stay of less than 48 hoursbefore death was more common in group II(i.e patients who were found to have majordiscrepancies in antemortem andpostmortem diagnosis), suggesting thatthe clinicians are now re\ring more ondiagnostic tests to arrive at a diagnosis. Ashorter hospital stay might not enable thetreating clinicians to utilize all diagnosticmodalities available (especially in publichealth care setup).

Possible factors contributing todiagnostic errors are lack of adequatehistory, er-rors of omission (failure toobtain admission screening test, failure tocarry out required, indicated procedures),errors ofjudgement (failure to account forsigns/symptoms noted, failure to accountfor abnormal investigation report,prejudiced viewpoint.), misleadinginvestigation reports and failure to reviewa case as a whole or to repeat physicalexamination..

It may not have been possible to avoidthe diagnostic errors in the cases withshort hospital stay, but in those with along hospital stay, treatment and outcomewould have been different if they couldhave been diagnosed antemortem. As longas diagnosis and therapy continues toevolve discrepancies w'ill continue to existat a substantial level. Thus it is beyonddoubt that autopsies still continue torevea-l a considerable number of errors inclinical diagnosis in patients dying in thehospitals. There is no convincing sign thatthere is decrease in the rate of errorsalthough the diseases and types ofmistakes have changed.

32

The controlling function of autopsyacts at two levels viz. al at the individuallevel, the autopsy helps the clinician tounderstand the signs and symptoms in theactual case. b) at the collective level,analysis of a large number of autopsiescan expose errors in the clinical diagnosticprocess and initiate correction of theseerrors.

Gross autopsy findings canoccasionally prove inadequate andhistopathological examination of involvedtissues is recommended. Sections fromorgans that appear grossly normal andwould not contribute to the diagnosisusually do not need to be examinedhistopathologically.

Lack of documentation of clinicalimpression in 14 cases probably reflectsthe lack of concern or fear of litigation orboth. Clinicians should be sensitized thatthe information that they provide is vital tothe pattrologist.

Medical autopsies are progressivelydecliningworldwide.u'' Some of the reasonsare discussed below :

a) Clinicians lack motivation and feelautopsies are unnecessar5r. Theciinicians are more confident oftheir diagnosis of the underlyingcause of death and with theincreased confidence come a lackof interest in autopsy.

Difficulty in obtaining the consentfor autopsy from patient'srelatives. This often stems fromlack of good communicationbetween the treating clinicians andpatient's relatives.

In some cases it is the fear in thetreating doctors that the autopsyfindings may not be consistentwith the presumed clinicaldiagnosis. This, the clinicians may

Bonfiay Hospital .Iournal,lbl. 53 , No. I , 201 I

b)

c)

I

fear, might expose their mistakesin diagnosis and or treatment.

d) There is an increasing dependenceand over reliance of the clinicianson sophisticated moderndiagnostic modalities. This hasresulted in making the autopsysuperfluous. Several studies haveshown that despite the improvedquality of diagnostic technologrthe frequency of incorrectdiagnosis has not decreasedappreciably.'

e) A long turnaround time for finalhistopathologic autopsy reportsreduces the clinician's interest inthe process. Hence autopsy reportsshould be given in a time boundmanner i.e. provisional reportsshould be issued the same day andfinal reports withing a month ofautopsy.

0 All autopsy examinations shouldbe carried under the supervision ofa senior pathologist; this will helpin accurate interpretation anddiagnosis of gross examinationfindings.

g) The entire process should involveall concerned parties closely i.e.the clinicians involved in patientscare, patient's relatives and thepathologists.

h) With the easy availability of digitalphotography, a compact disccontaining gross and microscopicpictures can be provided to theclinicians. There should bebimonthly meetings of cliniciansand pathologists to discuss thecases.

If no autopsies were performed, whatwould be missed and what would begained? We would gain a little space in the

Bonfiay Hospital Jountal, Vol. 53 , No. I , 201 I

hospital mortuarSz / postmortem room. Wewould gain variable amounts of file andstorage space and histotechnician time.The incidence of occupational hazard forpathologists and mortuar5r workers mightcome down. We would save a few autopsycalls, but have fewer opportunittes t0 knourwhat is going on in the hospita-l and in thecommunit5r.

To conclude, autopsy still remains the"gold standard" of final diagnosis.lnfectious diseases continue to be underdiagnosed and are the most commoncause of discrepancy between antemortemand postmortem diagnosis. The decliningrate of medical autopsies is a cause ofconcern and needs urgent attentionespecially by clinician's and pathologistsin teaching institutes.References1. Nemetz PN, Ludwig J, Kurland LT. Assessing the

autopsy. AmJ ofPathol 1987 ;128:362-379.2. Tse GMK, Lee JCK. A 12-month review of

autopsies performed at a universit5r-affiliated' teaching hospital in Hong Kotg. HKMJ

2000;6: 1 9O- 1 9:1.

3. Roosen J, Franke, Wilmer A, Knockaert DC,Bobbaers H. Comparison of premortem clinicaldiagnosis in critically ill patients andsubsequent autopsy findings. Mago Clin Proc2000; 75:557-558.

4. Jashnaai KD, Deshpande JR, Amonkar GP.Medical autopsy : whose gain is it? An audit.Indian J Patlrcl Micrcbiol 2OO 6',49 : L 88- 1 92.

5. Karwinski B, Hartveit F. Death certification :

increased clinical conlidence in diag-nosis andlack of interest in conlirmation by necropsy isnotjustified. J Clin Pathol 1989 ; 42: 13-77 .

6. Veress B, Alafuzoff I. A retrospective analysis ofclinical diagnosis and autopsy findings in 3, O42cases during tu'o different time periods. HumPathol. 7994;25: 14O- 145.

7. McKelvie PA, Rode J. Autopsy rates andclinicopathological conference audit in anAustraiian metropolitan hospital - cause forconcern? Med J Aust 1992;756:456-462.

8. Lee Goldman, Diagnostic advances v the value ofthe autopsy 1912-1980 Arch Pathol Lob MedI 984;108:50 1-505.

33