Surgical specimen identification

errors: A new measure of quality in

surgical care
Martin A. Makary, MD, MPH,a,d,e Jonathan Epstein, MD,b Peter J. Pronovost, MD, PhD,a,c,d,e
E. Anne Millman, MS,a,d,e Emily C. Hartmann, MS,a and Julie A. Freischlag, MD,a Baltimore, Md

Background. Communication errors are the primary factor contributing to all types of sentinel events
including those involving surgical patients. One type of communication error is mislabeled specimens.
The extent to which these errors occur is poorly quantified. We designed a study to measure the
incidence and type of specimen identification errors in the surgical patient population.
Methods. We performed a prospective cohort study that included all patients who underwent surgery in
an outpatient clinic or hospital operating room and for whom a pathology specimen was sent to the
laboratory. The study took place during a 6-month period (October 2004 to April 2005) at an urban,
academic medical center. The study’s main end-points were the incidence and type of specimen labeling
errors in the hospital operating room and the outpatient clinic. The specimen was the unit of analysis.
All specimens were screened for “identification errors,” which, for the purposes of this study, were
defined as any discrepancy between information on the specimen requisition form and the
accompanying labeled specimen received in the laboratory. Errors were stratified by the type of
identification error, source, location, and type of procedure.
Results. A total of 21,351 surgical specimens were included in the analysis. There were 91
(4.3/1000) surgical specimen identification errors (18, specimen not labeled; 16, empty container;
16, laterality incorrect; 14, incorrect tissue site; 11, incorrect patient; 9, no patient name; and 7, no
tissue site). Identification errors occurred in 0.512% of specimens originating from an outpatient clinic
(53/10,354 specimens) and 0.346% of specimens originating from an operating room (38/10,997
specimens). Procedures involving the breast were the most common type to involve an identification error
(breast ⫽ 11, skin ⫽ 10, colon ⫽ 8); in addition, 59.3% (54/91) of errors were associated with a biopsy
procedure. Follow-up was complete in all cases found to have an identification error.
Conclusions. Surgical specimen identification errors are common and pose important risks to all
patients. In our study, these events occurred in 4.3 per 1000 surgical specimens or an annualized rate
of occurrence of 182 mislabeled specimens per year. Given the frequency with which these errors occur
and their potential effect on patients, the rate of surgical specimen identification errors may be an
important measure of patient safety. Strategies to reduce the rate of these errors should be a research
priority. (Surgery 2007;141:450-5.)

From the Departments of Surgery,a Surgical Pathology,b and Anesthesiology,c Center for Surgical Outcomes
Research, John Hopkins University School of Medicine; Department of Health Policy and Management,d Johns
Hopkins Bloomberg School of Public Health; and the Johns Hopkins Quality and Safety Research Group,e Johns
Hopkins Medical Institutions, Baltimore, Md

Though communication failures are a major which to measure communication.3-5 Communica-

contributor to all types of sentinel events,1,2 we lack tion failures are particularly problematic among
scientifically sound and feasible methods with surgical team members and result in preventable
morbidity, mortality, and high costs of care. In
studies of surgical patients, communication failures
Accepted for publication September 9, 2006. have been identified as the root cause in 80% of
Reprint requests: Martin A. Makary, MD, MPH, Assistant Profes- sentinel events, 77% of wrong-site operations, and
sor of Surgery and Health Policy & Management, Johns Hopkins other medical errors in the operating room.6-8 One
University, 4940 Eastern Ave, Building A-5, Baltimore, MD type of communication failure that poses risks to
21224. E-mail: mmakary1@jhmi.edu. patients is the mislabeling of a surgical specimen
0039-6060/$ - see front matter prior to its arrival in a pathology laboratory. Spe-
© 2007 Mosby, Inc. All rights reserved. cifically, an error in verbal communication and
doi:10.1016/j.surg.2006.08.018 transcription during the hand-off increases the risk

450 SURGERY
Surgery Makary et al 451
Volume 141, Number 4

of mislabeling the specimen. Indeed, the relay of Source location was classified as clinic versus
vital surgical specimen information occurs within a operating room. Type of procedure was classified
local culture of teamwork and communication be- using the source organ of the specimen (eg, breast
tween a physician and nurse or technician.9 Al- procedure, skin procedure, prostate procedure, etc.).
though this context of information transfer is The results were descriptive, and we presented the
associated with rates of medical errors, the local data as proportions. Comparisons were performed
culture of teamwork is difficult to quantify.5 One using a chi-square test using SPSS version 12.0
consequence of poor teamwork and communica- (SPSS, Chicago, Ill). Institutional Review Board
tion that can harm patients and can be measured is (IRB) exemption approval was obtained from the
the occurrence of specimens being labeled incor- Johns Hopkins University for the publication of
rectly. these findings as quality measurement data.
Despite the known risk to patient safety, the rate
of mislabeled specimens remains poorly defined, RESULTS
and few hospitals routinely monitor these rates as a Of 21,351 specimens studied, 91 (4.3 per 1000
measure of patient safety. We hypothesize that mis- specimens) were associated with an identification
labeled surgical specimens occur commonly and error. These included the following types of iden-
that the rate of these occurrences can be used to tification errors: 18, specimen not labeled; 16,
measure patient safety in the surgical patient pop- empty container; 16, laterality incorrect; 14, incor-
ulation. The specific aim of this study was to define rect tissue site; 11, incorrect patient; 9, no patient
and measure the frequency of mislabeled speci- name; and 7, no tissue site identified (Fig 1).
mens in the setting of surgical procedures. Of 10,354 specimens originating from the oper-
ating room, 38 contained errors (3.7 per 1000 spec-
METHODS imens). Of the 10,997 specimens received from a
We used a prospective cohort design. All surgical clinic, 53 contained errors (4.8/1000) (P ⫽ .062).
pathology specimens were screened by the surgical Compared with specimens received from the oper-
pathology department over a 6-month period (Oc- ating room, specimens from a clinic had an in-
tober 2004 to April 2005) at an urban, academic creased proportion of identification errors
medical center. The primary dependent variable involving an empty container (1.00 vs 0.48/1000),
was the rate of mislabeled surgical specimens, and incorrect laterality (1.00 vs 0.48/1000), incorrect
the specimen was the unit of analysis. All specimens tissue site (0.82 vs 0.48/1000), and incorrect pa-
were screened for errors, which were defined as any tient (0.64 vs 0.38/1000); whereas specimens re-
discrepancy between information on the specimen ceived from an operating room were more likely to
requisition form and the accompanying labeled have a specimen not labeled (1.16 vs 0.55/1000),
specimen received in the laboratory. Errors were no patient name (0.58 vs 0.28/1000), and no tissue
stratified by the type of identification error, source site (0.97 vs 0.55/1000) (Table I).
location, and type of procedure. We classified type of Among the 91 mislabeled surgical specimens,
identification error by the type of discrepancy found the largest number came from surgical procedures
between the specimen and accompanying requisi- involving a biopsy (n ⫽ 54), followed by procedures
tion form. The following categories were identified: associated with an excision (n ⫽ 24), resection
“specimen not labeled,” which included specimens (n ⫽ 3), or other procedure (n ⫽ 10). Among the
without a label present on the container; “empty 91 mislabeled surgical specimens, breast tissue was
container,” which included containers with no tis- the most common (n ⫽ 11), followed by skin (n ⫽
sue specimen; “laterality incorrect,” which included 10), and colon (n ⫽ 9) (Table II).
specimens for which the left versus right designa-
tion was inconsistent with the label or requisition; CONCLUSION
“incorrect tissue site,” which included any organ- or In this study, we found that mislabeled surgical
body-site– designation discrepancy between the la- specimens are common, occurring at an overall
bel and requisition, excluding laterality discrepan- rate of 4.3 per 1000 specimens, with 4.8 per 1000
cies; “incorrect patient name,” which included any occurring in a clinic and 3.7 per 1000 in an oper-
wrong patient label on the specimen or requisition ating room; when annualized, the rate of occur-
form; “no patient name,” which included speci- rence at this single institution is 182 mislabeled
mens missing a patient name on the container specimens each year, all posing significant risks to
and/or requisition; and “no tissue site,” which in- patient safety. Given the high frequency with which
cluded any specimen for which the site designation these events occurred and the feasibility and valid-
of the specimen was absent. ity of measuring them, mislabeled surgical speci-
452 Makary et al Surgery
April 2007

Figure. Incidence of identification errors observed per 1000 specimens (n ⫽ 21,351).

Table I. Type of specimen identification error by source location (clinic vs operating room [OR])
Total Clinic OR
Proportion Proportion Proportion
per 1000 per 1000 per 1000
Events specimens Events specimens Events specimens
Error type (% of total) (n ⫽ 21,351) (% of total) (n ⫽ 10,997) (% of total) (n ⫽ 10,354)
Specimen not labeled 18 (19.8) 0.84 6 (11.3) 0.55 12 (31.6) 1.16
Empty container 16 (17.6) 0.75 11 (20.8) 1.00 5 (13.2) 0.48
Laterality incorrect 16 (17.6) 0.75 11 (20.8) 1.00 5 (13.2) 0.48
Incorrect tissue site 14 (15.4) 0.66 9 (17.0) 0.82 5 (13.2) 0.48
Incorrect patient 11 (12.1) 0.52 7 (13.2) 0.64 4 (10.5) 0.38
No patient name 9 (9.9) 0.42 3 (5.7) 0.27 6 (15.8) 0.58
No tissue site 7 (7.7) 0.33 6 (11.3) 0.55 1 (2.6) 0.97
Total 91 (100) 4.26 53 (100) 4.82 38 (100) 3.67

mens may serve as a useful measure for patient mislabeled surgical specimens. The nurse immedi-
safety in surgical patients. In fact, the data for this ately undertook an aggressive investigation to find
study were collected as part of routine quality au- the correct information needed to resolve the dis-
dits in the Division of Surgical Pathology. Further- crepancy. This individual searched medical records
more, not investigating these errors may represent a to identify the intended procedure and tissue
failure to recover from an error. sought for biopsy or resection. In addition, operat-
In the study presented, a full-time, dedicated ing room personnel were interviewed, and the sur-
quality improvement nurse was notified about all geon who performed the procedure was contacted
Surgery Makary et al 453
Volume 141, Number 4

Table II. Common procedure types associated used for this phase and the differences in local
with mislabeled specimens (n ⫽ 91) practices regarding specimen handling, labeling,
Procedure type Events (% total)
and transport.18 Although the benefit of second
opinions and tracking systems within pathology de-
Breast 11 (12.1) partments have been established, a standardized
Skin 10 (11.0) protocol for surgeon-to-nurse/technician commu-
Colon 9 (9.9)
nication of specimen identification information in
Prostate 6 (6.6)
labeling and handling the specimen, and an
Uterus 4 (4.4)
Bone marrow 4 (4.4) independent check of specimen information in the
Vertebrae 4 (4.4) same way that blood products are checked, are
Small bowel 4 (4.4) lacking.
Brain 3 (3.3) The new Joint Commission on Accreditation of
Cervix 3 (3.3) Healthcare Organization safety goal—requiring label-
Liver 2 (2.2) ing specimens with at least 2 patient identifiers and a
Lymph node 2 (2.2) preoperative verification step in the presence of the
Tonsil 2 (2.2) physician (www.jointcommission.org/PatientSafety/
Thyroid 2 (2.2) NationalPatientSafetyGoals/)—may help standardize
Larynx 2 (2.2)
this process. In addition, recent efforts to improve
Anus 2 (2.2)
communication in operating rooms, such as briefing
Other 23 (25.3)
and debriefing protocols,5,19,20 and systems to evalu-
ate adverse events, such as comprehensive safety pro-
grams,21 are aimed at improving the transfer of such
by pager to discuss the discrepancy. All incidents of critical information and the process used to evaluate
mislabeled surgical specimens were resolved with- mistakes. The range of procedure types and sources
out patient harm. However, the extent of patient (operating room and outpatient clinic) for misla-
harm prior to this program at our hospital, or any beled specimens included in our study indicates that
hospital, is unknown and has not been reported in the problem is prevalent across surgical care and a
the literature. Nevertheless, errors involving speci- function of systems, not particular individuals. To
men identification could have resulted in delay in address the issues of poor communication in the
care, the need for an additional biopsy or therapy, operating room, which can lead to mislabeled speci-
failure to use appropriate therapy, or therapy ad- mens and other potential errors, we have instituted a
ministered to the wrong body site, side, or patient. routine briefing19 and debriefing20 with every oper-
These system failures can have serious implications ating room procedure. The briefing and debriefing
to patients and providers and result in significant (in the form of checklists) is prompted by the circu-
harm to the patient, costs to the institution, and lator and reviewed by the surgeon, anesthesia pro-
distrust by a community. In one study of the clinical vider, and nurse. The debriefing includes the
impact of errors in diagnosing surgical specimens question “Has the surgical specimen been verified?”
at 4 hospitals, 39% to 45% of errors were associated as a double check for accuracy in the same way that
with patient harm.10 blood products are verified by 2 persons. We are also
Errors involving surgical specimens are recog- promoting briefings and debriefings for all clinic and
nized to occur during 1 of 3 discrete phases: the bedside procedures to identify and mitigate potential
preanalytical phase (transfer of information from issues that can result in patient harm. These efforts
physician to nurse during a procedure, and subse- are aimed at promoting a culture of safety, which we
quent specimen labeling, packaging, and trans- measure among all our staff annually using a psycho-
port), the analytical phase (handling a specimen metrically validated survey instrument.5
within a surgical pathology laboratory and interpre- Measuring safety rates in surgical care has
tation by a pathologist), and the postanalytical proved challenging. We generally lack standard-
phase (recording and relaying the interpretation ized definitions of harm (ie, what constitutes a
for the clinician). The latter two phases have been surgical complication); we are unsure of the appro-
well-studied and have error rates of 0.1% to 9%11-15 priate denominator (ie, the population at risk);
and 0.12% to 3.4%, respectively.16,17 In this study, and we often rely on self-reported events that are
we chose to focus on the preanalytical phase be- notoriously biased because we lack an independent
cause it is poorly understood and may represent a surveillance mechanism. Even if we could obtain a
significant and highly variable source of harm, rate of harm, variation in case mix, methods of data
given the paucity of checks and balances routinely collection, and random error would likely far ex-
454 Makary et al Surgery
April 2007

ceed variation in safety.22,23 Each type of quality cur.19,21,23,25 By applying these principles to mis-
measure has its weakness. For example, quality as labeled surgical specimens, we can standardize
measured by surgical site infection rates may be a requisition forms, the process for completing the
function of a hospital’s case mix of colorectal op- forms, and the training for staff who com-
erations (which more commonly result in wound plete the forms. We also can create a checklist that
infection secondary to intestinal spillage of bacte- a nurse or other member of the care team would
ria) versus nongastrointestinal laparoscopic opera- complete prior to sending a specimen to the labo-
tions (which involve incisions often too small to ratory. Finally, when a mislabeled surgical speci-
manifest a wound infection). Furthermore, the in- men is discovered, we can review how and why it
fection data is difficult and expensive to collect happened and whether we need to change the
because it requires short- and long-term follow-up system. In this way, mislabeled surgical specimen
and uniform expertise to define what constitutes an events can be used both as a surrogate marker of
infection. Other measures of quality, such as rates patient safety and as a quality improvement flag
of postoperative deep venous thrombosis and myo- that could signal the need to redesign the process
cardial infarctions, are also confounded by variabil- of specimen identification in surgical care. In con-
ity in detection and selection bias given the high clusion, mislabeled surgical specimens are com-
rate of asymptomatic occurrences. Thus, the real- mon, represent a significant threat to patient safety,
ized incidence of these events can simply be a and should be a focus of patient safety improve-
function of the tools and vigilance used to search ment efforts.
for them, not their true incidence.24 As a result, The authors would like to thank Christine Holzmuel-
hospitals that are good at collecting complication ler of the Johns Hopkins Quality and Safety Research
data are punished for having the worst results, and Group for her assistance in the preparation of this manu-
hospitals that do a poor job of capturing events are script, and Doris Deane, surgical pathology quality
rewarded for having the best outcomes. We found assurance coordinator, for her assistance with data man-
agement.
that tracking the number of specimens that are
mislabeled in the operating room or other pro-
REFERENCES
cedure room is easily performed using a simple
1. Joint Commission on Accreditation of Healthcare Organi-
screening process at the receiving desk in the Sur- zations. Sentinel events: evaluating cause and planning im-
gical Pathology Department. As such, this method provement. Oakbrook Terrace, IL: Joint Commission on
of monitoring quality is inexpensive, not subject to Accreditation of Healthcare Organizations; 1998.
interpretation, and has a high capture rate, which 2. Lingard L, Espin S, Whyte S, et al. Communication failures
makes it a scientifically sound, yet feasible, measure in the operating room: An observational classification of
recurrent types and effects. Qual Saf Health Care
of patient safety in surgical patients. 2004;13:330-4.
We recognize several limitations in our findings. 3. Firth-Cozens J. Why communication fails in the operating
First, we did not obtain detailed data regarding the room. Qual Saf Health Care 2004;13:327.
tissue or procedure type for all specimens. As a 4. Espin S, Levinson W, Regehr G, Baker GR, Lingard L. Error
result, we cannot provide rates of error stratified by or “act of god”? A study of patients’ and operating room
team members’ perceptions of error definition, reporting,
tissue or procedure type, although the hospital and disclosure. Surgery 2006;139:6-14.
studied is representative of large academic facilities 5. Makary MA, Sexton JB, Freischlag JA, et al. Patient safety in
with a broad spectrum of surgical services. Second, surgery. Ann Surg 2006;243:628-35.
we only captured errors that were identified. We 6. Edmonds CR, Liguori GA, Stanton MA. Two cases of a
did not evaluate whether there were additional wrong-site peripheral nerve block and a process to prevent
this complication. Reg Anesth Pain Med 2005;30:99-103.
errors, namely lost specimens that were not re- 7. Gawande AA, Studdert DM, Orav EJ, Brennan TA, Zinner
ceived by surgical pathology, or what percentage of MJ. Risk factors for retained instruments and sponges after
errors resulted in harm. Considering these events, surgery. N Engl J Med 2003;348:229-35.
our rate is likely a conservative estimate of the true 8. Gawande AA, Zinner MJ, Studdert DM, Brennan TA. Anal-
error rate and may understate the extent of the ysis of errors reported by surgeons at three teaching hospi-
tals. Surgery 2003;133:614-21.
problem. 9. Makary MA, Sexton JB, Freischlag JA, et al. Teamwork in the
Though we have not yet reduced the rates of operating room: teamwork in the eye of the beholder. J Am
these errors, safety science provides some guidance Coll Surg 2006;202:848-52.
on how we might accomplish this. In our safety 10. Raab SS, Grzybicki DM, Janosky JE, et al. Clinical impact and
efforts, we reduce errors by the following measures: frequency of anatomic pathology errors in cancer diag-
noses. Cancer 2005;104:2205-13.
(1) standardizing what we do and when we do it, 11. Safrin RE, Bark CJ. Surgical pathology sign-out. routine
(2) creating independent checks for key processes, review of every case by a second pathologist. Am J Surg
and (3) learning from mistakes when they oc- Pathol 1993;17:1190-2.
Surgery Makary et al 455
Volume 141, Number 4

12. Furness PN, Lauder I. A questionnaire-based survey of er- 19. Makary MA, Holzmueller CG, Thompson D, et al. Operat-
rors in diagnostic histopathology throughout the United ing room briefings. Jt Comm Jour Qual Saf 2006;32:351-5.
Kingdom. J Clin Pathol 1997;50:457-60. 20. Makary MA, Holzmueller CG, Sexton JB, et al. Operat-
13. Renshaw AA, Young ML, Jiroutek MR. How many cases need ing room debriefings. Jt Comm Jour Qual Saf 2006;
to be reviewed to compare performance in surgical pathol- 32:407-10.
ogy? Am J Clin Pathol 2003;119:388-91. 21. Pronovost PJ, Weast B, Rosenstein B, et al. Implementing
14. McBroom HM, Ramsay AD. The clinicopathological meet- and validating a comprehensive unit-based safety program.
ing. A means of auditing diagnostic performance. Am J Surg J Patient Saf 2005;1:33-40.
Pathol 1993;17:75-80. 22. Pronovost PJ, Nolan T, Zeger S, Miller M, Rubin H. How can
15. Raab SS, Nakhleh RE, Ruby SG. Patient safety in anatomic clinicians measure safety and quality in acute care? Lancet
pathology: Measuring discrepancy frequencies and causes. 2004;363:1061-7.
Arch Pathol Lab Med 2005;129:459-66. 23. Lilford R, Mohammed MA, Spiegelhalter D, Thomson R.
16. Nakhleh RE, Zarbo RJ. Amended reports in surgical pathol- Use and misuse of process and outcome data in managing
ogy and implications for diagnostic error detection and performance of acute medical care: avoiding institutional
avoidance: A college of American pathologists Q-probes stigma. Lancet 2004;363:1147-54.
study of 1,667,547 accessioned cases in 359 laboratories. 24. Jha AK, Kuperman GJ, Teich JM, et al. Identifying adverse
Arch Pathol Lab Med 1998;122:303-9. drug events: development of a computer-based monitor and
17. Raab SS, Grzybicki DM, Zarbo RJ, Meier FA, Geyer SJ, comparison with chart review and stimulated voluntary re-
Jensen C. Anatomic pathology databases and patient safety. port. J Am Med Inform Assoc 1998;5:305-14.
Arch Pathol Lab Med 2005;129:1246-51. 25. Pronovost PJ, Holzmueller CG, Martinez E, et al. A practical
18. Zarbo RJ, Meier FA, Raab SS. Error detection in anatomic tool to investigate defects in patient care. Jt Comm Jour
pathology. Arch Pathol Lab Med 2005;129:1237-45. Qual Saf 2006;32:102-8.