AP Discrepancy Rates - College of American Pathologists
AP Discrepancy Rates - College of American Pathologists
AP Discrepancy Rates - College of American Pathologists
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
C<strong>AP</strong> Laboratory Improvement Programs<br />
Patient Safety in Anatomic Pathology<br />
Measuring <strong>Discrepancy</strong> Frequencies and Causes<br />
Stephen S. Raab, MD; Raouf E. Nakhleh, MD; Stephen G. Ruby, MD<br />
● Context.—Anatomic pathology discrepancy frequencies<br />
have not been rigorously studied.<br />
Objective.—To determine the frequency <strong>of</strong> anatomic pathology<br />
discrepancies and the causes <strong>of</strong> these discrepancies.<br />
Design.—Participants in the <strong>College</strong> <strong>of</strong> <strong>American</strong> <strong>Pathologists</strong><br />
Q-Probes program self-reported the number <strong>of</strong><br />
anatomic pathology discrepancies in their laboratories by<br />
prospectively performing secondary review (post–sign-out)<br />
<strong>of</strong> 100 surgical pathology or cytology specimens. Reasons<br />
for the secondary review included conferences, external<br />
review, internal quality assurance policy, and physician request.<br />
Participants.—Seventy-four laboratories self-reported<br />
data.<br />
Main Outcome Measures.—Frequency <strong>of</strong> anatomic pathology<br />
discrepancy; type <strong>of</strong> discrepancy (ie, change in<br />
margin status, change in diagnosis, change in patient in-<br />
The 1999 Institute <strong>of</strong> Medicine report increased the national<br />
awareness <strong>of</strong> medical errors and patient safety. 1<br />
Anatomic pathology errors are reported to occur in 1% to<br />
43% <strong>of</strong> all anatomic pathology specimens, 2–19 and this exceptionally<br />
wide range depends on the methods <strong>of</strong> detection<br />
and the definition <strong>of</strong> what counts as an error. On review<br />
<strong>of</strong> the literature, Raab2 estimated that the mean anatomic<br />
pathology error frequency ranged from 1% to 5%,<br />
although this frequency was largely based on studies using<br />
single-institution data. No large-scale, multi-institutional<br />
anatomic pathology error studies have been conducted,<br />
and information on the effect <strong>of</strong> anatomic pathology<br />
error on patient outcome is generally lacking.<br />
Error detection in anatomic pathology most <strong>of</strong>ten depends<br />
on some form <strong>of</strong> secondary case review. 2 Secondary<br />
case review has been built into some pathology quality<br />
assurance practices (eg, review <strong>of</strong> a set percentage <strong>of</strong> cases,<br />
intradepartmental ‘‘difficult case’’ conferences, cytolog-<br />
Accepted for publication December 2, 2004.<br />
From the Department <strong>of</strong> Pathology, University <strong>of</strong> Pittsburgh, Pittsburgh,<br />
Pa (Dr Raab); the Department <strong>of</strong> Pathology, St Luke’s Hospital/<br />
Mayo Clinic, Jacksonville, Fla (Dr Nakhleh); and the Department <strong>of</strong><br />
Pathology, Palos Community Hospital, Palos Heights, Ill (Dr Ruby).<br />
The authors have no relevant financial interest in the products or<br />
companies described in this article.<br />
Reprints: Stephen S. Raab, MD, Department <strong>of</strong> Pathology, University<br />
<strong>of</strong> Pittsburgh, UPMC Shadyside Hospital, 5150 Centre Ave, Pittsburgh,<br />
PA 15232 (e-mail: raabss@msx.upmc.edu).<br />
formation, or typographic error); effect <strong>of</strong> discrepancy on<br />
patient outcome (ie, no harm, near miss, or harm); and<br />
clarity <strong>of</strong> report.<br />
Results.—The mean and median laboratory discrepancy<br />
frequencies were 6.7% and 5.1%, respectively. Forty-eight<br />
percent <strong>of</strong> all discrepancies were due to a change within<br />
the same category <strong>of</strong> interpretation (eg, 1 tumor type was<br />
changed to another tumor type). Twenty-one percent <strong>of</strong> all<br />
discrepancies were due to a change across categories <strong>of</strong><br />
interpretation (eg, a benign diagnosis was changed to a<br />
malignant diagnosis). Although the majority <strong>of</strong> discrepancies<br />
had no effect on patient care, 5.3% had a moderate<br />
or marked effect on patient care.<br />
Conclusions.—This study establishes a mean multi-institutional<br />
discrepancy frequency (related to secondary review)<br />
<strong>of</strong> 6.7%.<br />
(Arch Pathol Lab Med. 2005;129:459–466)<br />
ic-histologic correlation, or review <strong>of</strong> all malignancies).<br />
Secondary case review also occurs in hospital patient-centered<br />
conferences (eg, tumor board); external consultation<br />
practices; or at the behest <strong>of</strong> clinicians, who may initiate<br />
communication when the pathology report does not correlate<br />
with the clinical findings. An error detected by 1 <strong>of</strong><br />
these processes may be referred to as a discrepancy or a<br />
difference in interpretation or reporting between 2 pathologists.<br />
Error detection frequencies based on the different methods<br />
<strong>of</strong> secondary review have been variably studied. For<br />
example, the <strong>College</strong> <strong>of</strong> <strong>American</strong> <strong>Pathologists</strong> (C<strong>AP</strong>) has<br />
intensively studied gynecologic cytologic-histologic correlation<br />
and reported that paired cervical-vaginal cytology-biopsy<br />
specimens had a sensitivity and specificity <strong>of</strong><br />
89.4% and 64.8%, respectively. 20–22 Based on nongynecologic<br />
cytologic-histologic correlation data, Clary et al 6 reported<br />
that 23% <strong>of</strong> interpretive discrepancies had a major<br />
impact on patient outcome. Other methods <strong>of</strong> secondary<br />
review have been studied in less detail. As a consequence,<br />
how these methods may be utilized to improve patient<br />
safety and be incorporated into laboratory continuous<br />
quality improvement programs is unknown.<br />
The C<strong>AP</strong>’s Q-Probes program has measured and defined<br />
a number <strong>of</strong> key quality indicators in anatomic and<br />
clinical pathology. 21,22 This Q-Probes study is the first multi-institutional<br />
study to measure and to document anatomic<br />
pathology discrepancy frequencies and the effect <strong>of</strong><br />
these discrepancies on patient outcome.<br />
Arch Pathol Lab Med—Vol 129, April 2005 Patient Safety in Anatomic Pathology—Raab et al 459
Table 1. Definition <strong>of</strong> <strong>Discrepancy</strong><br />
<strong>Discrepancy</strong>: A discrepancy has occurred if there is any difference between the original interpretation and the interpretation after the<br />
second review. Discrepancies were further classified by cause into one <strong>of</strong> the following categories:<br />
Change in margin status: The interpretation <strong>of</strong> the margin status was changed from benign to malignant or vice versa.<br />
Change in categoric interpretation: An interpretation was changed from one categoric diagnosis, such as benign, to another categoric<br />
diagnosis, such as malignant. For purposes <strong>of</strong> this study, interpretations were classified within categories that were graded by their<br />
probability <strong>of</strong> a malignant clinical outcome (eg, a benign diagnosis was assigned a 1; atypical, 2; suspicious, 3; and malignant, 4). We<br />
considered a difference <strong>of</strong> 2 or more steps between the original and the review interpretation as a discrepancy. For example, if the<br />
original diagnosis was benign, and the review diagnosis was malignant, the difference in steps between these 2 diagnoses was 4 1<br />
3, and this case was considered discrepant. If the step difference between the original and review interpretations was 1, we decided<br />
that a discrepancy had not occurred.<br />
Change within the same category <strong>of</strong> interpretation: An interpretation was changed from one benign interpretation to another benign<br />
interpretation or from one malignant interpretation to another malignant interpretation. A change from one tumor type to another fell<br />
within this category. For example, if the original interpretation was adenocarcinoma and the review diagnosis was epithelioid sarcoma,<br />
this case was placed within this category <strong>of</strong> discrepancy.<br />
Change in patient information: There was a change in the organ site, such as the left ovary to the right ovary.<br />
Typographic error<br />
Table 2. Definitions <strong>of</strong> Effects <strong>of</strong> Discrepancies<br />
Effect <strong>of</strong> discrepancy on patient management: Discrepancies were classified into the following categories based on patient outcome:<br />
Harm (significant event): A discrepancy that resulted in patient harm (eg, inappropriate treatment, loss <strong>of</strong> life or limb, psychological<br />
event). The effect <strong>of</strong> the significant event on patient outcome was assessed using a 3-point Likert scale (1 severe effect, 2 moderate<br />
effect, 3 mild effect). The pathologists performing the review judged the significance <strong>of</strong> the event.<br />
Near miss: A discrepancy that was detected before harm occurred, such as a discrepancy that was detected at a clinical pathologic<br />
conference before treatment was initiated.<br />
No harm: A discrepancy that did not result in patient harm, such as a typographic error that had no bearing on patient management.<br />
MATERIALS AND METHODS<br />
Laboratories enrolled in the C<strong>AP</strong>’s volunteer Q-Probes quality<br />
improvement program participated in this study in 2003. The Q-<br />
Probes program and the format for data collection and handling<br />
have been previously described in detail. 21<br />
Participants prospectively identified 100 consecutive surgical<br />
pathology or cytology specimens that were reviewed by a second<br />
pathologist after a first pathologist had signed out that case. In<br />
order to standardize the data-collection process across all participating<br />
laboratories, pertinent terms were defined (Tables 1 and<br />
2). For each case, the participating laboratory recorded the specimen<br />
type (surgical pathology or cytology), organ or anatomic<br />
site (chosen from a specified list), primary reason for secondary<br />
review (chosen from a specified list), clarity <strong>of</strong> the report, presence<br />
or absence <strong>of</strong> a discrepancy (Table 1), effect <strong>of</strong> discrepancy<br />
on patient outcome (Table 2), and modification <strong>of</strong> report (if performed).<br />
We subclassified discrepancies into several main groups <strong>of</strong><br />
causes that have been previously described in the pathology literature.<br />
2 The detection <strong>of</strong> particular discrepancy subtypes depends<br />
partly on the method <strong>of</strong> secondary review, described in<br />
more detail later. For example, change in margin status is more<br />
likely to be detected by a frozen-permanent section review, and<br />
change in categoric interpretation is more likely to be detected<br />
by cytologic-histologic correlation review. We arbitrarily chose a<br />
disagreement <strong>of</strong> 2 steps as constituting a categoric interpretation<br />
discrepancy, although even a 1-step discrepancy is an error. Raab 2<br />
reported that 2-step discrepancies have a much greater probability<br />
<strong>of</strong> clinical significance compared with 1-step discrepancies. In<br />
addition, because interobserver variability studies have shown<br />
that 1-step discrepancies are much more common, we did not<br />
want to collect data on a large subset <strong>of</strong> cases that had little effect<br />
on patient care. 2 For changes in categoric interpretation, the original<br />
and review diagnoses were recorded.<br />
The taxonomy <strong>of</strong> the effect <strong>of</strong> a pathology discrepancy on patient<br />
outcome was based on the medical patient safety literature,<br />
23–26 which uses a taxonomy related to the effect <strong>of</strong> a failure<br />
<strong>of</strong> a planned action to be completed or the use <strong>of</strong> a wrong plan<br />
to achieve an aim. 1 Diagnostic error does not fit neatly into the<br />
category <strong>of</strong> an ‘‘action’’ error, and the resulting outcome <strong>of</strong> the<br />
patient is <strong>of</strong>ten difficult to assess. In particular, distinguishing<br />
between a no-harm and a near-miss event may be problematic.<br />
We defined a near-miss event as an error that was detected before<br />
harm occurred; an example is when a diagnostic error was picked<br />
up at a conference (or another means <strong>of</strong> secondary review)<br />
before a particular treatment protocol was initiated. In this case,<br />
if the error had not been detected, we assume that some degree<br />
<strong>of</strong> harm would have occurred. We recognize that harm (eg, psychologic)<br />
may still have occurred in this case, but we classified<br />
this event as a near miss, because the ability <strong>of</strong> the review pathologist<br />
to identify this type <strong>of</strong> harm was limited in this study.<br />
We defined a no-harm event as occurring when a diagnostic error<br />
occurred that would not cause patient harm even if the error had<br />
been undetected. An example <strong>of</strong> a no-harm event was a typographic<br />
error, such as writing the word ‘‘brest’’ instead <strong>of</strong><br />
‘‘breast’’ in the diagnostic line.<br />
We defined a significant event as an error resulting in patient<br />
harm. 1 We provided the review pathologist a 3-point Likert scale<br />
to grade the severity <strong>of</strong> this harm, recognizing that this grading<br />
was subjective. Grzybicki et al 27 showed that even with more<br />
well-defined criteria for patient harm, there is little agreement<br />
among pathologists as to the effect <strong>of</strong> an error on patient outcome;<br />
thus, although subclassifying harm needs further study,<br />
we wanted to measure the general view <strong>of</strong> the participant institutions<br />
in assessing harm. We did not require the pathologist to<br />
perform medical record review but only to assess the error severity<br />
based on the information available at the time <strong>of</strong> detection.<br />
In general, the pathologist had some degree <strong>of</strong> knowledge <strong>of</strong> the<br />
clinical outcome. This information varied depending on the method<br />
<strong>of</strong> detection; for example, more information may be available<br />
if a clinician requests secondary review compared to if a cytologic-histologic<br />
correlation is performed.<br />
Autopsy cases were the only anatomic pathology case type excluded<br />
from this study. Multiple specimens with a secondary<br />
review from the same case or specimens from different cases<br />
associated with the same patient were included in the study. Secondary<br />
review is the main method used to detect anatomic pathology<br />
errors. The recorded reasons why cases were reviewed<br />
were as follows: intradepartmental conference, request by clinician,<br />
interdepartmental conference, specified institutional quality<br />
460 Arch Pathol Lab Med—Vol 129, April 2005 Patient Safety in Anatomic Pathology—Raab et al
assurance policy, and extradepartmental review. These methods<br />
encompass the majority <strong>of</strong> forms <strong>of</strong> secondary review, although<br />
participants also could choose the category <strong>of</strong> ‘‘other’’ if none <strong>of</strong><br />
these methods applied. We chose to examine multiple methods,<br />
because single institutions <strong>of</strong>ten do not review a sufficient number<br />
<strong>of</strong> cases using a single method for statistical significance testing<br />
in the time frame provided by this Q-Probes study. We also<br />
wanted to determine if discrepancies were detected at different<br />
frequencies and had different clinical import depending on the<br />
method <strong>of</strong> detection. Institutional quality assurance policies include<br />
cytologic-histologic correlation, random review, and frozenpermanent<br />
section correlation. We recognize that these methods<br />
<strong>of</strong> review detect only a fraction <strong>of</strong> discrepancies (because in most<br />
institutions, the majority <strong>of</strong> cases do not undergo secondary review)<br />
and exclude those errors detected prior to sign-out. However,<br />
some <strong>of</strong> these forms <strong>of</strong> secondary review are the most likely<br />
means to detect clinically significant error, the subcategory <strong>of</strong><br />
error that some authors believe is the most important to track<br />
and eradicate.<br />
The ‘‘clarity <strong>of</strong> the report’’ question referred to how understandable<br />
the original report was perceived to be from the standpoint<br />
<strong>of</strong> the review pathologist. The review pathologist determined<br />
the report clarity using a 4-point Likert scale (ie, clear,<br />
mildly unclear, moderately unclear, markedly unclear). A limitation<br />
in this assessment was that clinicians, who were the users<br />
<strong>of</strong> the diagnostic information, did not perform the analysis, and<br />
previous authors have indicated discrepancies between clinician<br />
and pathologist impression <strong>of</strong> report clarity. 28 We also recognize<br />
that the clarity assessment was subjective, because definitive criteria<br />
<strong>of</strong> clarity were not provided. The lack <strong>of</strong> clarity may be an<br />
example <strong>of</strong> error, particularly for reports assessed to be moderately<br />
or markedly unclear, because this lack <strong>of</strong> clarity could result<br />
in improper patient management.<br />
Laboratories could choose 3 options regarding computerized<br />
report modification after a discrepancy was detected: (1) the original<br />
report was deleted and replaced with a new report; (2) the<br />
original report was marked with a qualifier, and a new report<br />
was appended; or (3) the original report was not marked or modified,<br />
but a new report was appended.<br />
Institutional demographic and practice variable information<br />
was obtained by completion <strong>of</strong> a questionnaire. These data included<br />
annual specimen volume, number <strong>of</strong> institutional practicing<br />
pathologists, type and frequency <strong>of</strong> institutional review conferences<br />
(eg, breast, chest, or endocrine conference), and institutional<br />
quality assurance methods <strong>of</strong> secondary review. The discrepancy<br />
frequency (expressed as a percentage) was calculated<br />
for the aggregate data and for each institution. The discrepancy<br />
frequency was calculated as the number <strong>of</strong> discrepancies divided<br />
by the total number <strong>of</strong> cases reviewed, multiplied by 100. Discrepancies<br />
and report clarity assessments were subclassified by<br />
organ type (and further broken down by effect on patient outcome).<br />
The correlation <strong>of</strong> the main indicator variable (discrepancy<br />
frequency) was assessed for each <strong>of</strong> the predictor variables separately<br />
by using nonparametric Wilcoxon rank sum or Kruskal-<br />
Wallis tests. The 2 goodness-<strong>of</strong>-fit test was used to test for associations<br />
between the presence <strong>of</strong> a discrepancy and other caselevel<br />
variables. Statistically significant associations are defined at<br />
the P .05 level. The reason for case review was correlated with<br />
the effect <strong>of</strong> the discrepancy on patient outcome.<br />
Some participating institutions did not answer all <strong>of</strong> the questions<br />
on the demographics form or on the input forms. These<br />
institutions were excluded only from the tabulations and the<br />
analyses that required the missing data element.<br />
RESULTS<br />
A total <strong>of</strong> 74 institutions submitted data for this study.<br />
Most institutions (87.8%) were located in the United<br />
States, with the remainder located in Canada (n 7), Australia<br />
(1), and Saudi Arabia (1). Of the participating institutions,<br />
31% were teaching hospitals, and 23% had a pathology<br />
residency program. The Joint Commission on Ac-<br />
Table 3. Percentile Distribution <strong>of</strong> Anatomic<br />
Pathology <strong>Discrepancy</strong> Rate<br />
All Institutional Percentiles<br />
N 10th 25th<br />
50th<br />
(Median) 75th 90th<br />
<strong>Discrepancy</strong> rate, % 74 21.0 10.0 5.1 1.0 0.0<br />
creditation <strong>of</strong> Healthcare Organizations inspected the majority<br />
<strong>of</strong> the institutions (68%) participating in this study,<br />
and the C<strong>AP</strong> inspected 82% <strong>of</strong> laboratories contributing<br />
data. The participants’ institutional sizes were as follows:<br />
39.7% had fewer than 150 beds; 39.7%, 151 to 300 beds;<br />
9.5%, 301 to 450 beds; 3.2%, 451 to 600 beds; and 7.9%,<br />
more than 600 beds. Private nonpr<strong>of</strong>it institutions comprised<br />
54.9% <strong>of</strong> the institutions; 11.1% were private forpr<strong>of</strong>it<br />
institutions; 9.5% were state, county, or city hospitals;<br />
6.3% were university hospitals; 3.2% were federal<br />
governmental institutions; and 15.9% were other. Fiftythree<br />
percent <strong>of</strong> the institutions were within a city, 28.1%<br />
were suburban, and 15.6% were rural. The annual mean<br />
number (and SD) <strong>of</strong> surgical pathology, nongynecologic<br />
cytology, and gynecologic specimens per institution were<br />
16 241 (19 567), 1898 (2376), and 21 147 (44 903), respectively.<br />
The mean and median numbers <strong>of</strong> pathologists at<br />
each institution were 6 and 4, respectively.<br />
The 74 institutions performed secondary review <strong>of</strong> 6186<br />
anatomic pathology specimens (5268 surgical pathology<br />
specimens and 847 cytology specimens), and each institution<br />
collected data on a range <strong>of</strong> from 2 to 100 specimens<br />
(median, 99 specimens). In aggregate, 415 discrepancies<br />
were reported, and the overall mean anatomic pathology<br />
discrepancy frequency was 6.7%. The overall<br />
mean surgical pathology discrepancy frequency was 6.8%<br />
(356 discrepancies <strong>of</strong> 5255 reviewed cases), and the overall<br />
mean cytology discrepancy frequency was 6.5% (55 discrepancies<br />
<strong>of</strong> 844 reviewed cases); neither <strong>of</strong> the specimen<br />
types had a higher discrepancy frequency (P .78). The<br />
distribution <strong>of</strong> the anatomic pathology discrepancy frequencies<br />
is listed in Table 3. Higher percentile ranks indicate<br />
lower discrepancy frequencies.<br />
The number <strong>of</strong> reviewed specimens, overall discrepancy<br />
frequency, and the discrepancy classification by organ<br />
type are shown in Table 4. The most common organ types<br />
reviewed in this study were female genital tract and<br />
breast. None <strong>of</strong> the organ types had a higher frequency <strong>of</strong><br />
discrepancy than other organ types (P .15). For the organ<br />
types with higher volumes reviewed (200 cases), a<br />
change in categoric interpretation (ie, an error type more<br />
likely to be associated with harm) occurred more frequently<br />
in the female genital tract, male genital tract, and<br />
lymph node. A change in margin status occurred most<br />
frequently in breast specimens.<br />
In Table 5, the effect <strong>of</strong> the discrepancy on patient outcome<br />
and the report modification in response to a discrepancy<br />
are listed by organ type. Some form <strong>of</strong> harm was<br />
seen in the majority <strong>of</strong> organs, although specimen type<br />
(cytology or surgical pathology) or organ type did not<br />
correlate with effect on patient outcome (P .73 and P <br />
.83, respectively). Harm was observed in 20.8% (11 cases)<br />
<strong>of</strong> breast specimens and in 25.3% (21 cases) <strong>of</strong> female genital<br />
tract specimens in which a discrepancy was detected.<br />
Neither specimen type nor organ type correlated with the<br />
Arch Pathol Lab Med—Vol 129, April 2005 Patient Safety in Anatomic Pathology—Raab et al 461
Table 4. Number <strong>of</strong> Specimens and <strong>Discrepancy</strong> by Organ Type<br />
Change in Change in Change in<br />
Overall Change in Categoric Same Patient Typographic<br />
Specimens, No. <strong>Discrepancy</strong>, Margin, Interpretation, Category, Information, Error,<br />
Organ<br />
(% <strong>of</strong> Total) %<br />
%<br />
%<br />
%<br />
%<br />
%<br />
Genital, female<br />
982 (15.9) 7.1 2.3 28.7 36.8 18.4 13.8<br />
Breast<br />
796 (12.9) 8.3 10.9 12.5 42.2 10.9 23.4<br />
Lung<br />
463 (7.5) 5.0 0 13.6 50.0 4.6 31.8<br />
Genital, male<br />
355 (5.7) 7.1 0 41.7 54.1 0<br />
4.2<br />
S<strong>of</strong>t tissue<br />
345 (5.6) 7.5 0 12.5 66.7 8.3 12.5<br />
Lymph node<br />
288 (4.7) 5.9 5.9 23.5 52.9 0<br />
17.7<br />
Hepatobiliary<br />
240 (3.9) 6.7 0 13.3 53.3 13.3 20.0<br />
Urinary tract<br />
181 (2.9) 7.2 0 30.8 53.9 0<br />
15.4<br />
Pharynx<br />
141 (2.3) 5.0 0 28.6 42.9 0<br />
28.6<br />
Endocrine<br />
125 (2.0) 8.0 0 10.0 50.0 10.0 30.0<br />
Bone marrow<br />
107 (1.7) 6.6 0 14.3 85.7 0<br />
0<br />
Bone<br />
99 (1.6) 2.0 0<br />
0<br />
50.0 0<br />
50.0<br />
Neuropathology<br />
88 (1.4) 6.8 0 16.7 83.3 0<br />
0<br />
Kidney<br />
55 (0.9) 5.5 0 33.3 33.3 0<br />
33.3<br />
Pancreas<br />
31 (0.5) 6.5 0 50.0<br />
0<br />
0<br />
50.0<br />
Salivary gland<br />
29 (0.5) 3.5 0<br />
0 100.0 0<br />
0<br />
Spleen<br />
11 (0.2) 9.1 0<br />
0<br />
0<br />
0 100.0<br />
Gastrointestinal and other<br />
1841 (29.8) 5.6 5.0 19.0 48.0 8.0 20.0<br />
Total 6162 6.7 3.7 21.0 47.7 9.1 18.5<br />
Organ<br />
Genital, female<br />
Breast<br />
Lung<br />
Genital, male<br />
S<strong>of</strong>t tissue<br />
Lymph node<br />
Hepatobiliary<br />
Urinary tract<br />
Pharynx<br />
Endocrine<br />
Bone marrow<br />
Bone<br />
Neuropathology<br />
Kidney<br />
Pancreas<br />
Salivary gland<br />
Spleen<br />
Gastrointestinal and other<br />
Table 5. Effect <strong>of</strong> <strong>Discrepancy</strong> on Patient Management and Pathology Response by Organ Type<br />
Marked<br />
Harm,<br />
%<br />
2.4<br />
1.9<br />
5.3<br />
4.4<br />
8.0<br />
0<br />
0<br />
7.7<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
Moderate<br />
Harm,<br />
%<br />
6.0<br />
7.6<br />
0<br />
0<br />
0<br />
6.2<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
2.1<br />
Effect on Patient Outcome<br />
Mild<br />
Harm,<br />
%<br />
16.9<br />
11.3<br />
0<br />
8.7<br />
0<br />
0<br />
7.1<br />
23.1<br />
14.3<br />
0<br />
28.6<br />
0<br />
16.7<br />
0<br />
50.0<br />
0<br />
0<br />
12.6<br />
Near Miss,<br />
%<br />
No Harm,<br />
%<br />
Report Change<br />
Yes, % No, %<br />
462 Arch Pathol Lab Med—Vol 129, April 2005 Patient Safety in Anatomic Pathology—Raab et al<br />
4.8<br />
13.2<br />
21.1<br />
13.0<br />
12.0<br />
18.8<br />
0<br />
7.7<br />
0<br />
11.1<br />
0<br />
0<br />
0<br />
33.3<br />
0<br />
0<br />
0<br />
6.3<br />
69.9<br />
66.0<br />
73.7<br />
73.9<br />
80.0<br />
75.0<br />
92.9<br />
61.5<br />
85.7<br />
88.9<br />
71.4<br />
100.0<br />
83.3<br />
66.7<br />
50.0<br />
100.0<br />
100.0<br />
79.0<br />
58.3<br />
62.3<br />
33.3<br />
56.5<br />
58.3<br />
56.3<br />
42.9<br />
38.5<br />
14.3<br />
33.3<br />
42.9<br />
100.0<br />
50.0<br />
0<br />
100.0<br />
0<br />
100.0<br />
54.1<br />
Total 2.1 3.2 11.3 8.7 7.5 53.3 46.7<br />
report modification in response to a discrepancy (P .15<br />
and P .14, respectively).<br />
In Table 6, the clarity <strong>of</strong> the report is listed by organ<br />
type. Markedly and moderately unclear reports were seen<br />
infrequently and only in a few specimen types, such as<br />
female genital tract, breast, and lung. Because <strong>of</strong> the large<br />
number <strong>of</strong> cells with a value <strong>of</strong> 0, a 2 goodness-<strong>of</strong>-fit test<br />
was not performed.<br />
In Table 7, the discrepancy type, original and review<br />
interpretations for categoric discrepancies, and the effect<br />
<strong>of</strong> the discrepancy on patient outcome are shown. A<br />
change in the same category <strong>of</strong> diagnosis was the most<br />
common discrepancy detected. For changes in categoric<br />
interpretation, the review diagnosis tended to be shifted<br />
downward to a benign or upward to a malignant diagnosis,<br />
and there were fewer nondefinitive (atypical or sus-<br />
41.7<br />
37.7<br />
66.7<br />
43.5<br />
41.7<br />
43.7<br />
57.1<br />
61.5<br />
85.7<br />
66.7<br />
57.1<br />
0<br />
50.0<br />
100.0<br />
0<br />
100.0<br />
0<br />
45.9<br />
picious) diagnoses compared with the original diagnosis.<br />
When a discrepancy occurred, the most common classification,<br />
based on patient outcome, was a no-harm event.<br />
Anatomic pathology discrepancy specimen-centered<br />
variables, including specimen type and origin, discrepancy<br />
type, primary reason for review, the effect <strong>of</strong> discrepancy<br />
on patient outcome, and the response to a discrepancy<br />
in the form <strong>of</strong> a report change, were evaluated<br />
to identify any associations. The statistically significant associations<br />
are shown in Table 8. A request for review directed<br />
by a clinician was much more likely to be associated<br />
with a discrepancy than all other reasons for review. If a<br />
discrepancy occurred, a change in categoric interpretation<br />
was more likely to be seen in cytology specimens compared<br />
with surgical pathology specimens and related to<br />
extradepartmental review compared with all other rea-
Genital, female<br />
Breast<br />
Lung<br />
Genital, male<br />
S<strong>of</strong>t tissue<br />
Lymph node<br />
Hepatobiliary<br />
Urinary tract<br />
Pharynx<br />
Endocrine<br />
Bone marrow<br />
Bone<br />
Neuropathology<br />
Kidney<br />
Pancreas<br />
Organ Clear, %<br />
Salivary gland<br />
Spleen<br />
Gastrointestinal and other<br />
Table 6. Clarity <strong>of</strong> Report by Organ Type<br />
97.6<br />
96.5<br />
97.0<br />
98.6<br />
97.7<br />
97.6<br />
96.6<br />
98.3<br />
98.6<br />
99.2<br />
94.4<br />
99.0<br />
95.4<br />
96.2<br />
96.8<br />
100.0<br />
90.9<br />
97.4<br />
Mildly<br />
Unclear, %<br />
Moderately<br />
Unclear, %<br />
Markedly<br />
Unclear, %<br />
Total 97.4 2.3 0.3 0.07<br />
Table 7. <strong>Discrepancy</strong> Type, the Original and Review<br />
Diagnoses for Categoric Discrepancies, and the Effect<br />
<strong>of</strong> <strong>Discrepancy</strong> on Patient Outcome<br />
<strong>Discrepancy</strong> type<br />
Change within same category<br />
Change in categoric interpretation<br />
Typographic error<br />
Change in patient information<br />
Change in margin status<br />
Original interpretation for categoric<br />
discrepancies<br />
Benign<br />
Atypical<br />
Suspicious<br />
Malignant<br />
Reviewed interpretation for categoric<br />
discrepancies<br />
Benign<br />
Atypical<br />
Suspicious<br />
Malignant<br />
Effect on patient outcome<br />
Harm<br />
Marked<br />
Moderate<br />
Mild<br />
Near miss<br />
No harm<br />
Specimens, No. (%)<br />
415<br />
194<br />
85<br />
75<br />
37<br />
15<br />
25<br />
27<br />
16<br />
16<br />
28<br />
21<br />
7<br />
28<br />
63<br />
8<br />
12<br />
43<br />
33<br />
283<br />
(100)<br />
(47.8)<br />
(20.9)<br />
(18.5)<br />
(9.1)<br />
(3.7)<br />
(29.8)<br />
(32.1)<br />
(16.0)<br />
(16.0)<br />
(33.3)<br />
(25.9)<br />
(8.3)<br />
(33.3)<br />
(16.6)<br />
(2.1)<br />
(3.2)<br />
(11.3)<br />
(8.7)<br />
(74.7)<br />
sons for review. If a near-miss event occurred, the reason<br />
for case review was more likely to be extradepartmental<br />
review compared with all other reasons for review.<br />
Table 9 shows that the reason for case review correlated<br />
with patient outcome in discrepant cases (P .02). Harm<br />
occurred more frequently in discrepant cases that were<br />
reviewed at the request <strong>of</strong> a clinician (23.5%) and interdepartmental<br />
conference (25.0%). Clinician-directed review<br />
was the most common method that detected a discrepancy<br />
(23.0% <strong>of</strong> all cases reviewed). The majority <strong>of</strong><br />
discrepant cases detected at an intradepartmental conference<br />
were associated with no-harm events.<br />
Arch Pathol Lab Med—Vol 129, April 2005 Patient Safety in Anatomic Pathology—Raab et al 463<br />
1.9<br />
2.1<br />
2.8<br />
1.4<br />
2.1<br />
2.4<br />
3.4<br />
1.7<br />
1.4<br />
0.8<br />
3.8<br />
1.0<br />
4.6<br />
3.8<br />
3.2<br />
0<br />
9.1<br />
2.4<br />
0.3<br />
1.3<br />
0.2<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
1.9<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0.2<br />
0.2<br />
0.1<br />
0<br />
0<br />
Of the 74 institutions, the number that had a conference<br />
devoted to breast review was 33; chest, 21; endocrine, 8;<br />
gastrointestinal, 22; general surgical, 29; genitourinary<br />
tract, 18; gynecologic, 26; head and neck, 16; hematopathology,<br />
20; liver, 15; renal, 18; and tumor board, 60. Institutional<br />
quality assurance practices were measured as<br />
well: 52 institutions had an intradepartmental conference<br />
for difficult cases; 31 reviewed a percentage <strong>of</strong> cases after<br />
sign-out; 22 reviewed all malignancies before sign-out; 19<br />
reviewed a percentage <strong>of</strong> cases before sign-out; 6 reviewed<br />
all malignancies after sign-out; and 1 reviewed all cases<br />
before sign-out. Of the institutions that made changes to<br />
the reports after an error, 43 issued an amended report<br />
and did not retrieve the original report; 3 retrieved the<br />
original report, stamped it ‘‘in error,’’ and filed the report<br />
in the chart; 3 destroyed the original reports; and 3 handled<br />
the change using other methods.<br />
COMMENT<br />
This is the first study to determine a baseline anatomic<br />
pathology discrepancy frequency across multiple pathology<br />
laboratories. Based on secondary pathologist review,<br />
the mean anatomic pathology discrepancy frequency<br />
(based on cases reviewed through several different methods)<br />
was 6.7%, and the variability across laboratories was<br />
striking, with the 25th and 75th percentiles being 10.0%<br />
and 1.0%, respectively.<br />
<strong>Discrepancy</strong> represents one form <strong>of</strong> error, and based on<br />
literature review <strong>of</strong> a number <strong>of</strong> error-detection methods,<br />
Raab 2 estimated that the mean laboratory error frequency<br />
ranged from 1% to 5%. The discrepancy frequency established<br />
in this Q-Probes study is based on review <strong>of</strong> selected<br />
cases (a targeted group <strong>of</strong> cases studied), a bias that<br />
may overestimate overall laboratory error. Error frequencies<br />
partly depend on the method <strong>of</strong> case detection, and<br />
the more thoroughly one looks for error, the more frequently<br />
one will find it. 1,29 As expected, some <strong>of</strong> the secondary<br />
review methods used in this study detected more<br />
error than other methods; for example, clinician-directed<br />
review detected a discrepancy (23.0% <strong>of</strong> cases) more frequently<br />
than random review (4.3%). In general, methods<br />
0.3<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0
Primary reason for review (P .001)<br />
Request by clinician<br />
All other reasons<br />
<strong>Discrepancy</strong> type: change in categoric interpretation<br />
Specimen type (P .001)<br />
Cytology<br />
Surgical pathology<br />
Primary reason for secondary review (P .03)<br />
Extradepartmental review<br />
All other reasons for review<br />
<strong>Discrepancy</strong> type: change in same category <strong>of</strong> diagnosis<br />
Specimen type (P .001)<br />
Cytology<br />
Surgical pathology<br />
<strong>Discrepancy</strong> type: change in patient information<br />
Primary reason for secondary review (P .001)<br />
Request by clinician<br />
All other reasons for review<br />
Effect on patient outcome: near miss<br />
Primary reason for secondary review (P .001)<br />
Extradepartmental review<br />
All other reasons for review<br />
Effect on patient outcome: no harm<br />
Primary reason for secondary review (P .001)<br />
Intradepartmental review<br />
All other reasons for review<br />
Response to a discrepancy: report change<br />
Primary reason for secondary review (P .001)<br />
Interdepartmental conference<br />
Request by clinician<br />
All other reasons for review<br />
Table 8. Statistically Significant Associations<br />
No. <strong>of</strong> Specimens No. With <strong>Discrepancy</strong><br />
464 Arch Pathol Lab Med—Vol 129, April 2005 Patient Safety in Anatomic Pathology—Raab et al<br />
348<br />
5812<br />
54<br />
349<br />
89<br />
317<br />
349<br />
54<br />
79<br />
327<br />
80<br />
299<br />
44<br />
335<br />
43<br />
77<br />
250<br />
Table 9. Reason for Review Correlated With Effect on Patient Outcome<br />
Errors<br />
Detected, No.<br />
Effect on Patient Outcome<br />
(% <strong>of</strong> Cases Harm, Near Miss, No Harm,<br />
Reason for Review<br />
Reviewed) No. (%)<br />
No. (%)<br />
No. (%) Total<br />
Intradepartmental review<br />
47 (7.1)<br />
2 (5.3)<br />
2 (5.3)<br />
34 (89.5) 38<br />
Request by clinician<br />
80 (23.0) 12 (23.5) 3 (5.9)<br />
36 (70.6) 51<br />
Interdepartmental review<br />
48 (4.8)<br />
8 (25.0) 2 (6.3)<br />
22 (68.8) 32<br />
Selected by quality assurance review<br />
127 (4.3) 13 (14.6) 8 (9.0)<br />
68 (76.4) 89<br />
Extradepartmental review<br />
92 (8.6)<br />
8 (14.0) 12 (21.1) 37 (64.9) 57<br />
Total 43 27 197 267<br />
<strong>of</strong> case review that involve clinical input are a better<br />
means to detect error but are harder to perform.<br />
<strong>Pathologists</strong> have studied error frequency using intralaboratory<br />
quality assurance methods in some detail. The<br />
most commonly studied review method is correlation,<br />
such as frozen-permanent section correlation or cytologichistologic<br />
correlation, a form <strong>of</strong> review mandated by the<br />
Clinical Laboratory Improvement Amendments <strong>of</strong> 1988. 20<br />
Using this review process and the total number <strong>of</strong> cases<br />
as the denominator, Clary et al 6 reported that 2.26% and<br />
0.44% <strong>of</strong> nongynecologic cytology and histology cases<br />
were discrepant. This Q-Probes study showed that the error<br />
frequency based on extradepartmental conference review<br />
was 7.1%, whereas Raab et al 30 reported an error<br />
frequency <strong>of</strong> 8.9%, with severe significant events occurring<br />
in 7.0% <strong>of</strong> all errors. McBroom and Ramsay 11 reported that<br />
80<br />
335<br />
27<br />
56<br />
26<br />
59<br />
179<br />
15<br />
24<br />
13<br />
15<br />
18<br />
40<br />
243<br />
33<br />
53<br />
103<br />
9.0% <strong>of</strong> cases reviewed at a clinicopathologic conference<br />
had a change in diagnosis. This similarity in error frequency<br />
across studies and across many institutions may<br />
indicate a true benchmark.<br />
Benchmarking and reducing anatomic pathology error<br />
frequency clearly is just beginning, and prior to this study,<br />
the benchmarks were based on single-institution or anecdotal<br />
data. 2 The correlation <strong>of</strong> error frequencies with particular<br />
secondary review practices and existing laboratory<br />
error reduction programs is largely unknown. Lack <strong>of</strong><br />
subspecialty expertise may contribute to higher error frequencies,<br />
although error data detected using extradepartmental<br />
specialty academic institution review may be biased<br />
and may overreport error. 4,7,31 Underreporting <strong>of</strong> error<br />
because <strong>of</strong> biased review methods, lack <strong>of</strong> understandable<br />
error taxonomy, and individual fears invariably exists,
ut this has not been thoroughly studied in pathology laboratories.<br />
Some leading patient safety researchers, such as Resar<br />
et al, 32 have argued that error prevention programs should<br />
target errors that have an effect on patient outcome, rather<br />
than all errors. These Q-Probes study data show that the<br />
majority <strong>of</strong> anatomic pathology discrepancies do not result<br />
in harm, similar to the data reported in nonpathology<br />
fields. 1 Determining the effect <strong>of</strong> a pathology error on patient<br />
outcome is challenging because <strong>of</strong> contact and temporal<br />
barriers between pathology and clinical care. Thorough<br />
medical record review generally is not performed<br />
after a pathology error has occurred, and health care systems<br />
invariably lack personnel trained in the intricacies <strong>of</strong><br />
the triad <strong>of</strong> pathology reporting, patient safety, and assessing<br />
patient outcomes. Grzybicki et al 27 reported that<br />
even with data from medical record review, pathologists<br />
showed poor agreement in determining if harm actually<br />
occurred. We used pathologist self-assessment to determine<br />
error severity, and acknowledging the biases inherent<br />
in this process, we found that 16.6% <strong>of</strong> all errors resulted<br />
in some form <strong>of</strong> patient harm. This indicates that<br />
1.1% <strong>of</strong> all anatomic pathology cases that underwent secondary<br />
review were associated with a harmful significant<br />
event. Because secondary review processes tend to be varied<br />
across institutions, the probability is high that a relatively<br />
large percentage <strong>of</strong> pathology errors resulting in<br />
harm go undetected.<br />
Statistical significance testing did not show that any<br />
body site was most likely to be associated with an error.<br />
However, some organs (eg, female genital tract and breast)<br />
tended to have higher associations with a discrepancy resulting<br />
in clinical harm compared with other organs. Using<br />
medical chart review to determine clinical follow-up<br />
<strong>of</strong> errors detected through cytologic-histologic correlation,<br />
Clary et al 6 reported that the most common cytology specimens<br />
associated with harm were pulmonary and breast<br />
specimens. In a study examining errors associated with<br />
malpractice claims, Troxel and Sabella 15 also identified<br />
that diagnostic errors in breast cytology specimens were<br />
associated with harm; in addition, they identified other<br />
organ or specimen types, such as prostate needle biopsies,<br />
Papanicolaou tests, and melanocytic skin lesions, as being<br />
associated with errors resulting in harm. In this Q-Probes<br />
study, discrepancies detected in cytology specimens were<br />
more likely (compared with surgical pathology specimens)<br />
to have a 2-step change in diagnosis, and this<br />
change <strong>of</strong>ten meant that the original diagnosis was either<br />
a false-negative or false-positive diagnosis. These types <strong>of</strong><br />
errors are more likely to have a clinical effect than other<br />
types <strong>of</strong> errors. 2 Harm was more likely to be associated<br />
with cases reviewed at the behest <strong>of</strong> a clinician and cases<br />
sent for extradepartmental review.<br />
Although researchers have studied anatomic pathology<br />
diagnostic variability, the relationship <strong>of</strong> variability and<br />
error is not sufficiently addressed in the pathology literature.<br />
Strictly speaking, variability is a form <strong>of</strong> error, and<br />
secondary case review is simply a means to unearth differences<br />
in diagnosis or errors. In practice, some pathologists<br />
would like to maintain that ‘‘true’’ diagnostic errors<br />
are errors that most reasonable pathologists would agree<br />
are errors. However, establishing the ‘‘true’’ diagnosis is a<br />
complex and controversial task (eg, do we use a panel <strong>of</strong><br />
practicing pathologists or experts?) that has not been well<br />
studied in anatomic pathology. Our study did not address<br />
methods <strong>of</strong> establishing the accuracy <strong>of</strong> the original and<br />
review diagnoses. However, some methods <strong>of</strong> secondary<br />
review were performed with knowledge <strong>of</strong> clinical outcome,<br />
which, although biasing the review pathologist<br />
(compared with the original pathologist, who lacked this<br />
bias), provides a window on the actual effect <strong>of</strong> the diagnosis.<br />
This Q-Probes study recorded errors other than interpretive<br />
errors. Typographic errors and changes in patient<br />
information infrequently result in harm, but when harm<br />
occurs, it may be severe, with far-reaching consequences<br />
(eg, switching <strong>of</strong> patient specimens). Pathology laboratories<br />
place checks in the system in order to limit these error<br />
types, although these safeguards are generally not formally<br />
shared across laboratories. The frequency <strong>of</strong> marked<br />
harm as a result <strong>of</strong> these errors is known only through<br />
small studies or anecdotally, 9,33,34 resulting in a lack <strong>of</strong><br />
widespread system learning. Our study was not detailed<br />
enough to drill down into these error types.<br />
The lack <strong>of</strong> report clarity is an example <strong>of</strong> error that is<br />
difficult to quantify and falls within the realm <strong>of</strong> communication<br />
error. Poor communication is an important<br />
source <strong>of</strong> error in clinical medicine and may result in severe<br />
harm 35 ; Dovey et al 25 reported that 5.8% <strong>of</strong> family<br />
practice errors were a result <strong>of</strong> miscommunication. Report<br />
clarity is subjective, and Powsner et al 28 reported that surgeons<br />
misunderstood 30% <strong>of</strong> pathology reports. As expected,<br />
in this Q-Probes study, pathologists believed that<br />
the majority <strong>of</strong> reports were clear, and less than 1% were<br />
perceived as markedly unclear. This confirms the conclusion<br />
by Powsner et al 28 that a communication gap exists<br />
between pathologists and clinicians. The fact that clinicians<br />
request that a certain percentage <strong>of</strong> cases be reviewed<br />
underscores a potential communication problem<br />
but shows as well a functioning means <strong>of</strong> detecting error.<br />
Conclusions<br />
Using secondary pathologist review, the mean anatomic<br />
pathology diagnostic discrepancy frequency was 6.7%.<br />
More than 1% <strong>of</strong> all reviewed anatomic pathology cases<br />
may be associated with an error associated with patient<br />
harm.<br />
The <strong>College</strong> <strong>of</strong> <strong>American</strong> <strong>Pathologists</strong> provided financial support.<br />
The statistical reviewer was Molly Walsh, PhD, <strong>College</strong> <strong>of</strong><br />
<strong>American</strong> <strong>Pathologists</strong>.<br />
References<br />
1. Kohn LT, Corrigan JM, Donaldson MS, eds. To Err Is Human: Building a<br />
Safer Health System. Washington, DC: National Academy Press; 1999.<br />
2. Raab SS. Improving patient safety by examining pathology errors. Clin Lab<br />
Med. 2004;24:849–863.<br />
3. Adad SJ, Souza MA, Etchebehere RM, et al. Cyto-histological correlation <strong>of</strong><br />
219 patients submitted to surgical treatment due to diagnosis <strong>of</strong> cervical intraepithelial<br />
neoplasia. Sao Paulo Med J. 1999;117:81–84.<br />
4. Arbiser ZK, Folpe AL, Weiss SW. Consultative (expert) second opinions in<br />
s<strong>of</strong>t tissue pathology: analysis <strong>of</strong> problem-prone diagnostic situations. Am J Clin<br />
Pathol. 2001;116:473–476.<br />
5. Chan YM, Cheung AN, Cheng DK, et al. Pathology slide review in gynecologic<br />
oncology: routine or selective? Gynecol Oncol. 1999;75:267–271.<br />
6. Clary KM, Silverman JF, Liu Y, et al. Cytohistologic discrepancies: a means<br />
to improve pathology practice and patient outcomes. Am J Clin Pathol. 2002;<br />
117:567–573.<br />
7. Hahm GK, Niemann TH, Lucas JG, et al. The value <strong>of</strong> second opinion in<br />
gastrointestinal and liver pathology. Arch Pathol Lab Med. 2001;125:736–739.<br />
8. Furness PN, Lauder I. A questionnaire-based survey <strong>of</strong> errors in diagnostic<br />
histopathology throughout the United Kingdom. J Clin Pathol. 1997;50:457–460.<br />
9. Hocking GR, Niteckis N, Cairns BJ, Hayman JA. Departmental audit in surgical<br />
anatomical pathology. Pathology. 1997;29:418–421.<br />
10. Labbe S, Petitjean A. False negatives and quality assurance in cervicouterine<br />
cytology. Ann Pathol. 1999;19:457–462.<br />
Arch Pathol Lab Med—Vol 129, April 2005 Patient Safety in Anatomic Pathology—Raab et al 465
11. McBroom HM, Ramsay AD. The clinicopathological meeting: a means <strong>of</strong><br />
auditing performance. Am J Surg Pathol. 1993;17:75–80.<br />
12. Nakhleh RE, Zarbo RJ. Amended reports in surgical pathology and implications<br />
for diagnostic error detection and avoidance. Arch Pathol Lab Med. 1998;<br />
122:303–309.<br />
13. Ramsay AD, Gallagher PJ. Local audit <strong>of</strong> surgical pathology: 18 months<br />
experience <strong>of</strong> peer-review–based quality assessment in an English teaching hospital.<br />
Am J Surg Pathol. 1992;16:476–482.<br />
14. Safrin RE, Bark CJ. Surgical pathology signout: routine review <strong>of</strong> every case<br />
by a second pathologist. Am J Surg Pathol. 1993;17:1190–1192.<br />
15. Troxel DB, Sabella JD. Problem areas in pathology practice uncovered by<br />
a review <strong>of</strong> malpractice claims. Am J Surg Pathol. 1994;18:821–831.<br />
16. Whitehead ME, Fitzwater JE, Lindley SK, et al. Quality assurance <strong>of</strong> histopathologic<br />
diagnoses: a prospective audit <strong>of</strong> 3 thousand cases. Am J Clin Pathol.<br />
1984;81:487–491.<br />
17. Zuk JA, Kenyon WE, Myskow MW. Audit in histopathology: description <strong>of</strong><br />
an internal quality assessment scheme with analysis <strong>of</strong> preliminary results. J Clin<br />
Pathol. 1991;44:10–15.<br />
18. Lind AC, Bewtra C, Healy JC, et al. Prospective peer review in surgical<br />
pathology. Am J Clin Pathol. 1995;104:560–566.<br />
19. Zardawi IM, Bennett G, Jain S, et al. Internal quality assurance activities<br />
<strong>of</strong> a surgical pathology department in an Australian teaching hospital. J Clin<br />
Pathol. 1998;51:695–699.<br />
20. Jones BA, Novis DA. Cervical biopsy-cytology correlation: a <strong>College</strong> <strong>of</strong><br />
<strong>American</strong> <strong>Pathologists</strong> Q-Probes study <strong>of</strong> 22439 correlations in 348 laboratories.<br />
Arch Pathol Lab Med. 1996;120:523–531.<br />
21. Zarbo RJ. Monitoring anatomic pathology practice through quality assurance<br />
measures. Clin Lab Med. 1999;19:713–742.<br />
22. Raab SS, Jones BA. Q-TRACKS: Gynecologic Cytologic-Histologic Correlation:<br />
2003 Annual Summary. Northfield, Ill: <strong>College</strong> <strong>of</strong> <strong>American</strong> <strong>Pathologists</strong>;<br />
2003.<br />
23. Brixey J, Johnson TR, Zhang J. Evaluating a medical error taxonomy. Proc<br />
AMIA Symp. 2002:71–75.<br />
24. McNutt RA, Abrams RI. A model <strong>of</strong> medical error based on a model <strong>of</strong><br />
disease: interactions between adverse events, failures, and their errors. Qual Manag<br />
Health Care. 2002;10:23–28.<br />
25. Dovey SM, Meyers DS, Phillips RL Jr, et al. A preliminary taxonomy <strong>of</strong><br />
medical errors in family practice. Qual Saf Health Care. 2002;11:233–238.<br />
26. Tamuz M, Thomas EJ, Franchois KE. Defining and classifying medical error:<br />
lessons for patient safety reporting systems. Qual Saf Health Care. 2004;13:13–<br />
20.<br />
27. Grzybicki DM, Vrbin-Turcsanyi CM, Janosky J, Raab SS. Examining pathology<br />
errors to improve patient safety: pathologists don’t agree on the identification<br />
<strong>of</strong> errors due to pathologist misinterpretation. Paper presented at:<br />
AcademyHealth Annual Research Meeting; June 8, 2004; San Diego, Calif.<br />
28. Powsner SM, Costa J, Homer RJ. Clinicians are from Mars and pathologists<br />
are from Venus. Arch Pathol Lab Med. 2000;124:1040–1046.<br />
29. Nieva VF, Sorra J. Safety culture assessment: a tool for improving patient<br />
safety in healthcare organizations. Qual Saf Health Care. 2003;12S:ii17–23.<br />
30. Raab SS, Clary KM, Grzybicki DM. Improving pathology practice by review<br />
<strong>of</strong> cases presented at chest conference [abstract]. Mod Pathol. 2003;67:<br />
312A–313A.<br />
31. Tsung JS. Institutional pathology consultation. Am J Surg Pathol. 2004;28:<br />
399–402.<br />
32. Resar RK, Rozich JD, Classen D. Methodology and rationale for the measurement<br />
<strong>of</strong> harm with trigger tools. Qual Saf Health Care. 2003;12S:ii39–45.<br />
33. Cree IA, Guthrie W, Anderson JM, et al. Departmental audit in histopathology.<br />
Pathol Res Pract. 1993;189:453–457.<br />
34. Ramsay AD. Errors in histopathology reporting: detection and avoidance.<br />
Histopathology. 1999;34:481–490.<br />
35. Gawande AA, Zinner MJ, Studdert DM, Brennan TA. Analysis <strong>of</strong> errors<br />
reported by surgeons at 3 teaching hospitals. Surgery. 2003;133:614–621.<br />
466 Arch Pathol Lab Med—Vol 129, April 2005 Patient Safety in Anatomic Pathology—Raab et al