RT Journal
A1 Atkins CD
T1 AUtopsy rates and diagnosis
JF JAMA
JO JAMA
YR 1999
FD June 16
VO 281
IS 23
SP 2181
OP 2181
DO 10-1001/pubs.JAMA-ISSN-0098-7484-281-23-jac90005
UL http://dx.doi.org/10-1001/pubs.JAMA-ISSN-0098-7484-281-23-jac90005
AB 
To the Editor: Dr Burton and
colleagues1 present important 
new information on the
correlation between clinical and autopsy diagnosis of malignancies.
However, their analysis suffers due to the use of the wrong
denominators, which leads to misleading conclusions. Although it is
true that only 125 of the 225 patients with malignancies found at
autopsy were correctly diagnosed, the appropriate denominator to
measure discordance is 1105 cases since we must consider those patients
correctly determined not to have malignancies. Additionally, of the 111
malignancies in the 100 patients whom the authors considered
underdiagnosed, 34 were correctly diagnosed as having malignancies on
clinical grounds and 24 had incidental malignancies. Only 8
malignancies were given an incorrect diagnosis. Thus, only about 45
(4%) of 1105 cases had clinically unsuspected malignancies that were
not merely incidental (the authors analyzed their underdiagnosed cases
by malignancy, not by patients, so the number of cases may be somewhat
greater or less than 45). Apparently all 880 patients without
malignancies were correctly identified as such clinically (the
authors do not specifically state this). Only 8
(6%) of 133 pathologically diagnosed malignancies were assigned an
incorrect histology or primary site. These data suggest to me that we
have excellent modern diagnostic tools which have led to highly
accurate clinical diagnoses, at least in New Orleans, La.