JOURNAL OF BLOOD SERVICES MANAGEMENT

Root cause analysis of transfusion error: identifying causes

to implement changes _2943 2772..2777

Priti Elhence, S. Veena, Raj Kumar Sharma, and R. K. Chaudhary

INTRODUCTION
BACKGROUND: As part of ongoing efforts to improve

T
transfusion safety, an error reporting system was imple- he ultimate aim of the blood transfusion service
mented in our hospital-based transfusion medicine unit is to provide safe and quality blood transfusion to
at a tertiary care medical institute. This system is based patients. However, “transfusion safety” extends
on Medical Event Reporting System–Transfusion Medi- much beyond the scope of “safe blood unit.” It
cine (MERS-TM) and collects data on all near miss, no depends upon the coordinated linkage of many processes
harm, and misadventures related to the transfusion from donor selection to transfusion.1 Data from reporting
process. Root cause analyses of one such innocuous systems around the world demonstrate that non-
appearing error demonstrate how weaknesses in the infectious hazards are the leading cause of serious morbid-
system can be identified to make necessary changes to ity and/or mortality resulting from transfusion.2,3 Adverse
achieve transfusion safety. events are however infrequent and it is difficult to gather
STUDY DESIGN AND METHODS: The reported error enough comprehensive data to generate statistically sig-
was investigated, classified, coded, and analyzed using nificant information on patterns, trends, and root causes of
MERS-TM prototype, modified and adopted for our such events. On the other hand “near miss events” and “no
institute. harm events” are much more frequent and are more forth-
RESULTS: The consequent error was a “mistransfu- coming in revealing weaknesses in processes and systems,
sion” but a “no-harm event” as the transfused unit was giving chance for improving patient safety.4
of the same blood group as the patient. It was a high We have recently started an error reporting and man-
event severity level error (level 1). Multiple errors pre- agement system with the aim of identifying deficiencies
ceded the final error at various functional locations in and lapses for taking corrective actions. Event reporting,
the transfusion process. Human, organizational, and investigation, classification and coding, risk assessment
patient-related factors were identified as root causes index calculation, and root cause analysis is done using
and corrective actions were initiated to prevent future Medical Event Reporting System for Transfusion Medicine
occurrences. (MERS TM) prototype.5 Prior to this, only adverse transfu-
CONCLUSION: This case illustrates the usefulness of sion events were reported and resolved on an as and where
having an error reporting system in hospitals to high- basis. Now, any unusual event/deviation or adverse event
light human and system failures associated with trans- is reported to the reporting officer from the transfusion
fusion that may otherwise go unnoticed. Areas can be service, who completes the event discovery form to capture
identified where resources need to be targeted to all aspects of the event; the event is then reported to the
improve patient safety. Quality Assurance Systems Operator (QA Sys Op), who
starts detailed investigation to trace and code all the ante-
cedent events to find where and what type of errors

From the Department of Transfusion Medicine, Sanjay Gandhi

Postgraduate Institute of Medical Sciences, Lucknow, India.
Address reprint requests to: Priti Elhence, Associate Prof,
Department of Transfusion Medicine, Sanjay Gandhi Postgradu-
ate Institute of Medical Sciences, Lucknow-226014, India;
e-mail: pelhence@sgpgi.ac.in.
Received for publication April 5, 2010; revision received
May 22, 2010, and accepted July 23, 2010.
doi: 10.1111/j.1537-2995.2010.02943.x
TRANSFUSION 2010;50:2772-2777.

2772 JBSM Volume 50, December 2010

 TRANSFUSION SAFETY IN THE HOSPITAL

have occurred. A causal tree is then constructed to vial. There was no discrepancy in patient identification
find the root causes of the error so that they can be eli- details on the sample label and blood requisition; there-
minated/minimized to improve the system and prevent fore, blood requisition was accepted. Blood grouping by
recurrences. tube technique using monoclonal blood grouping antis-
An interesting and unusual event was reported during era (Diamed AG, Switzerland) was found to be A Rh(D)
this period. Root cause analysis of this event revealed more positive. Two units of RBCs were crossmatched by
than what was visible on the surface. This event illustrates polyspecific Anti Human Globulin gel cards (Diamed AG).
the weak links in the transfusion chain and how and where Results were entered on HIS and a “compatibility report”
errors occur at bedside. It also shows how important it is to was generated. A “compatibility label” with patient and
have an error reporting system and root cause analysis to unit identification details was stuck on both the units, and
make meaningful recommendations to reduce the trans- the units were prepared for issue. Nothing unassuming so
fusion risk. Unless the transfusion services take the initia- far, and the system seems to be perfect in giving safe blood
tive to work on such events, the cases that would come to transfusion to the patient.
notice would only be “misadventures” or “adverse events” However, when ward attendants came to collect
where the patient has already been harmed. blood for the patient the next day, it came to their notice
that blood units had been already collected by another
ward attendant on the previous day itself. The units were
THE BACKGROUND
neither found in the ward refrigerator nor transfused to
Ours is a hospital-based transfusion service and an aca- the patient. In the first instance, it appeared to be a simple
demic department of transfusion medicine supporting case of issue to another patient with the same name, but it
blood component needs of an 870-bed super specialty was found that there was only one requisition for a patient
teaching institute and hospital in North India. The trans- with this name, and mix-up was thus ruled out. Routine
fusion service issues around 20,000 units of red blood cells approach would have been to presume that those two
(RBCs) annually. units were misplaced, raise another requisition for two
The hospital has a computerized hospital informa- units, and STOP investigating further. However, as the
tion system (HIS), which has a special module for blood reporting officer was informed and a detailed investiga-
transfusion service, with “live” interface with the rest of tion was initiated, the units were found in ward Y after a
HIS. On registration with the hospital, each patient is continued search. One unit had already been transfused
assigned a unique identification number known as central uneventfully to patient Jennifer (Patient II) with the same
registration number (CR No.) for identification. CR No. is a first name, and another was still stored in the ward refrig-
10-digit numerical number—the first four digits represent erator. The compatibility label and compatibility form
year of registration and the next six are sequentially gen- indicated that these units were indeed meant for Patient I,
erated identification numbers. and transfusion to Patient II was a mis-transfusion. This
Blood component requisitions are generated in wards was a “no harm event,” as the patient’s blood group was A
on HIS using the patient’s CR No. as an identification Rh(D) positive and the blood units were also A Rh(D) posi-
access for a given patient by the concerned resident doctor. tive. If the search was unyielding, this mistransfusion
The rest of the patient’s details and historical blood group would have gone undetected. This represents the major
records are automatically updated from the system. The chunk of errors represented by a submerged portion of
type of component, number of units, and time of require- iceberg, showing that errors are occurring at much higher
ment are indicated by the requesting doctor. A request is rates than they are detected.
thus generated with unique sequential requisition number The remaining unit along with the compatibility form
and all other details. The printout of this requisition, signed was then taken to ward X for transfusion to Patient I
by the resident doctor, is sent along with the labeled blood (Jennifer Smith) for whom it was originally meant. Her
sample of the patient to the blood service. The procedures relatives, however, informed that she was group AB Rh(D)
are according to Drugs and Cosmetic Act of India and positive (historical blood group), whereas the blood
national guidelines.6,7 Standard operating procedures issued for this patient indicated that she was A Rh(D)
(SOPs) are also available to all the users. positive. Hence, the transfusion was stalled and the blood
bank staff was informed of this discrepancy (near miss
event). The availability of the historical blood group has
Case report been shown to prevent many such errors.
A blood request for two units of RBCs was received for a ABO and RhD grouping was repeated on a pretrans-
patient named Jennifer Smith (Patient I) admitted in Ward fusion sample (compatibility samples are required to be
X. This was a printed computerized physician order entry preserved for at least 7 days) and was found to be A Rh(D)
(CPOE) form with the prescribing doctor’s signature. The positive again. A repeat blood sample was then ordered
request was accompanied by a blood sample in a labeled from Patient I (Jennifer Smith) and was confirmed to be

Volume 50, December 2010 JBSM 2773

ELHENCE ET AL.

AB Rh(D) positive, as told by the patient’s relatives, prepared the blood sample label with the information on
meaning that the previous sample indicated a “wrong the requisition form of Patient I’s identification details
blood in the tube” (WBIT). Two units of AB Rh(D) positive (i.e., Jennifer Smith) (second breach) and collected a
blood were crossmatched for Patient I and transfused blood sample from Patient II (i.e., Jennifer) (third
uneventfully. However, according to computer records, breach). Thus, the pretransfusion sample of request for
the blood group of Patient I was saved as A Rh(D) positive. Patient I (Jennifer Smith) indeed belonged to Patient II
This is unmodifiable data and modifications can be done (Jennifer). Positive patient identification was omitted at
only by HIS in charge. It is important to correct such record all stages as the patient was unconscious and wrist bands
as this serves as future reference. were not available (SOPs for raising blood requisition,
The problems yet unsolved were whose blood (A sample labeling, and sample collection were not fol-
Rh(D) positive) was there in the pretransfusion sample lowed). Thus, the blood in the sample tube belonged to
and how Patient II (Jennifer), who was admitted in the correct patient but the requisition and sample label
another ward, got involved in the first transfusion, The belonged to another patient with the same name but dif-
historical blood group of Patient II was A Rh(D) positive, ferent CR No. The ward attendant had collected blood
but she was admitted in a physically and administratively from the blood center by patient name and, hence, the
separate ward and, hence, the sample collection from this error was also not detected at that stage (omission of
patient was a remote possibility. The case was discussed SOPs) (fourth breach). At the time of transfusion, again,
with the concerned consultant to identify the source of the SOPs for confirmation of identification on compat-
error and problem areas in the transfusion chain. The resi- ibility label and form with the patient identity were not
dent doctor who had sent the request and the nursing staff followed, hence, the error was not detected (fifth breach).
responsible for the collection of the sample was called The patient was lucky enough to survive after receiving
for first-hand information. It is important to enquire from one unit of blood not intended for her and despite so
the people involved in the incident. Initial response was of many errors, as the blood in the pretransfusion sample
denial and blaming the blood bank for the mixup of sent indeed belonged to her. All these errors would not
samples, but when the importance of working up such have come to be noticed if the blood for Patient I
errors for identifying weak links in the
transfusion process with the aim of pre-
venting such errors in the future was
emphasized, active cooperation was
extended. It is important to have a pro-
active nonpunitive approach to error
management. The event was recognized
as an unusual event of high severity level
and root cause analysis was undertaken.

EVENT
CHRONOLOGY—RESULTS
OF INVESTIGATION
On close scrutiny of the medical
records of both patients, it was found
that Patient II (Jennifer) needed trans-
fusion during dialysis and she was
physically temporarily transferred to a
dialysis ward at another location. The
patient was unconscious and medical
records and CR No. were unavailable at
the time of raising requisition (there
was no wrist band). The resident doctor
searched the hospital information
system for CR No. of an admitted
patient with the name “Jennifer,” and a
blood request was raised. This record
actually belonged to Patient I, i.e., Jen-
nifer Smith (first breach). The nurse Fig. 1. Root cause analysis of transfusion error.

2774 JBSM Volume 50, December 2010

 TRANSFUSION SAFETY IN THE HOSPITAL

(Jennifer Smith) was not requested for or if the Patient I Based on the findings of the case, the hospital admin-
was of the same blood group as Patient II. In fact, it was istration was apprised of the need for continuing medical
found that genuine requisitions and sample for Patient I education of existing paramedical staff, newly joining
(Jennifer Smith) raised from ward X were not accepted at paramedical and resident doctors about the bedside pro-
the transfusion service, as HIS showed that blood was cedures to be followed and need of wrist bands with
already crossmatched for the patient. This event also patient details for patient identification at least in uncon-
delayed the transfusion to Patient I. scious patients to start with. SOPs for patient identifica-
Using the MERS TM system of classification of errors, tion, sample collection, and transfusion were prepared as
the present error was classified as a no harm event with charts which are now displayed in wards for ready refer-
high event severity level (ESL): level 1 with a risk assess- ence (Figs. 2 and 3). Use of technology to reduce the
ment Index 0.9. The events are coded to indicate where in chances of human errors is also being considered. This
the process the error occurred (indicated by alphabets), case illustrates and highlights the importance of role of
and what actually happened (indicated
by numbers). Descriptive event codes
by location of events were Patient
Request (PR), Order Entry (OR), Sample
Collection (SC), Unit Issue (UI), and
Unit Transfusion (UT). Thus, error
occurred at multiple levels. Consequent
discovery code was 2UT 001, that is, no
recovery no harm event and administra-
tion of correct product to wrong patient,
i.e., mistransfusion. Significant ante-
cedent codes were PR 001—Order for
wrong patient, SC 001—Sample labeled
incorrectly, SC 003—Sample collected
from wrong patient, UI 009—Not check-
ing or incorrect checking of patient
information at unit issue, UI 011—Unit
delivered to incorrect location, UT
005—Administrative review of patient/
unit information not done at bedside
before transfusion and UT 007—
Transfusion delayed.
Figure 1 shows the root cause
analysis of the event and the causal
tree. The discovered event is in the top
box and all the antecedent events are
traced back to their origins. In this
case, the root causes were identified as
human rule-based verification (HRV)
errors, human rule-based intervention
(HRI), organizational management
(OM), organizational policy (OP), and
patient-related factors (PR). The causal
tree construction also identified the
corrective actions that will be helpful
in the prevention of such errors.
Another lesson learned from this exer-
cise was that blood units involved in
any event should be brought back to
the blood bank and repeat samples for
blood grouping and crossmatching
should be collected from all involved
patients before issuing blood. Fig. 2. Poster chart for instructions for sending blood requests and samples.

Volume 50, December 2010 JBSM 2775

ELHENCE ET AL.

other nations. The United Kingdom

Serious Hazards of Transfusion program
collects data on the full range of adverse
transfusion events and there too, mis-
transfusion accounts for the largest pro-
portion of all adverse events.3 The
frequency and types of errors observed
suggest that these are inevitable, in that
mistakes are inherent to human nature,
unless significant changes, including the
use of computerized instruments, are
made to procedures.8-10
In India, we do not have such a
reporting system as yet, though it is felt
that there is a need. Mistransfusion
events are surprisingly easy to miss, as
illustrated by this case, and might
involve more than one error and be con-
fusing to sort out. However, once prop-
erly investigated, the full picture
emerges.
The problem, however, is the way
the errors have been managed in the
past. The strategies of “blame and
shame” and “blame and tame” result in
hiding the errors. A “just culture” that
balances the accountability with the
need for learning environment is what is
required to improve the system. This
system acknowledges and does not take
disciplinary action for the unintended
nature of human errors and takes cor-
rective actions for latent errors.11 One
needs to establish error reporting
systems for uniform classification and
logical root cause analysis, creating a
“safety culture,” inter institutional
sharing of strategies to reduce the fre-
quency of errors, regular error-related
feedback to frontline staff, suggesting
required changes to hospital policies
and procedures, and creation of bench-
Fig. 3. Poster chart for instructions for collection of the components from blood
mark rates of error as a quality assess-
center and transfusion.
ment tool. It also provides a learning
opportunity and gives us a chance to see
transfusion services in error management programs to how things are actually working in our organization, in
ensure transfusion safety for the patients. contrast to how we think they are working.

COMMENTS AUTHOR IDENTIFICATION

Among the noninfectious hazards, mistransfusion repre- Dr Priti Elhence, MD, principal author in collaboration with Dr S.
sents the most frequently observed serious hazard Veena, Prof R.K. Chaudhary, and Prof R.K. Sharma.
and is a leading cause of death from transfusion reported to Priti Elhence, MD is an associate Professor in the Depart-
the U.S. Food and Drug Administration since such ment of Transfusion Medicine at Sanjay Gandhi Postgraduate
reporting began.2 Similar results are also reported from Institute of Medical Sciences (SGPGIMS), Lucknow, India.

2776 JBSM Volume 50, December 2010

 TRANSFUSION SAFETY IN THE HOSPITAL

S. Veena, MBBS, is a postgraduate student pursuing an MD events in the Republic of Ireland. Vox Sanguinis 2007;92:
in “Blood Transfusion and Immunohematology” at the Depart- 233-41.
ment of Transfusion Medicine at SGPGIMS, Lucknow, India. 5. Kaplan HS, Callum JL, Rabin Fastman B, Merkley LL. The
R.K. Chaudhary, MD, DNB, is Professor and Head, Depart- medical event reporting system for transfusion: will it help
ment of Transfusion Medicine at SGPGIMS, Lucknow, India. get the right blood to the right patient? Transfus Med Rev
R.K. Sharma, MD, MNAMS, FAMS, FASN, and FISN is Pro- 2002;16:86-102.
fessor and Head, Department of Nephrology and Director, 6. Malik V. Drugs and cosmetic act1940. Lucknow: Eastern
SGPGIMS, Lucknow, India. Book Company; 2003. 16th ed. p. 279-303.
7. National Blood Policy. National AIDS Control Organiza-
tion, Ministry of Health and Family Welfare, Government of
CONFLICT OF INTEREST
India. 2007.
None. 8. Miyata S, Kawai T, Yamamoto S, Takada M, Iwatani Y,
Uchida O, Imanaka H, Sase K, Yagihara T, Kuro M. Network
computer assisted transfusion-management system for
REFERENCES accurate blood component-recipient identification at the
1. Dzik WH. Emily Cooley lecture 2002: transfusion safety in bedside. Transfusion 2004;44:364-72.
the hospital. Transfusion 2003;43:1190-9. 9. Dzik WH. New technology for enhanced transfusion safety.
2. Sazama K. Reports of 355 transfusion-associated deaths: Br J Haematol 2007;136:181-90.
1976 through 1985. Transfusion 1990;30:583-90. 10. Bennardello F, Fidone C, Cabibbo S, Calabrese S, Garozzo
3. Love EM, Soldan K, The Serious Hazards of Transfusion G, Cassarino G, Antolino A, Tavolino G, Zisa N, Falla C,
Steering Group. SHOT annual report 2000-2001. Available Drago G, Di Stefano G, Bonomo P. Use of identification
from: URL: http://www.blood.co.uk/hospitals/library/ system based on biometric data for patients requiring
shot/body01.htm transfusions guarantees transfusion safety and traceability.
4. Lundy D, Laspina S, Kaplan H, Fastman BR, Lawlor E. Blood Transf 2009;7:193-203.
Seven hundred and fifty nine chances to learn: a 3-year 11. Reason J. Human error: models and management. BMJ
pilot project to analyse transfusion related near-miss 2000;320:768-70.

Volume 50, December 2010 JBSM 2777