Original papers

Pre-analytical errors management in the clinical laboratory: a five-year study

Angeles Giménez-Marín1*, Francisco Rivas-Ruiz2, Maria del Mar Pérez-Hidalgo3, Pedro Molina-Mendoza1
1ClinicalLaboratory, Hospital de Antequera, Antequera, Málaga, Spain
2Hospital Research Unit, Hospital Costa del Sol, Marbella, Málaga, Spain
3Library and Knowledge Management Consultant, Hospital de Antequera, Antequera, Málaga, Spain

*Corresponding author: angeles.gimenez.sspa@juntadeandalucia.es

Abstract
Introduction: This study describes quality indicators for the pre-analytical process, grouping errors according to patient risk as critical or major, and
assesses their evaluation over a five-year period.
Materials and methods: A descriptive study was made of the temporal evolution of quality indicators, with a study population of 751,441 anal-
ytical requests made during the period 2007-2011. The Runs Test for randomness was calculated to assess changes in the trend of the series, and the
degree of control over the process was estimated by the Six Sigma scale.
Results: The overall rate of critical pre-analytical errors was 0.047%, with a Six Sigma value of 4.9. The total rate of sampling errors in the study pe-
riod was 13.54% (P = 0.003). The highest rates were found for the indicators “haemolysed sample” (8.76%), “urine sample not submitted” (1.66%)
and “clotted sample” (1.41%), with Six Sigma values of 3.7, 3.7 and 2.9, respectively.
Conclusions: The magnitude of pre-analytical errors was accurately valued. While processes that triggered critical errors are well controlled, the
results obtained for those regarding specimen collection are borderline unacceptable; this is particularly so for the indicator “haemolysed sample”.
Key words: total quality management; quality indicators, health care; laboratories; patient safety

Received: November 18, 2013 Accepted: April 27, 2014

Introduction
Strategies in the field of risk management and pa- tribution to patient care (2). Moreover, the Europe-
tient safety are aimed at the prevention, detection an Committee of Experts on Management of Safe-
and mitigation of adverse events through the anal- ty and Quality in Health Care has proposed that
ysis of errors. In the clinical laboratory all errors indicators should be useful, identify the critical
must be measured and controlled, from the obvi- steps in each process, reflect their potential and
ous ones to those that do not originate in the lab- make it possible to continuously assess safety in
oratory (1), by means of indicators providing an healthcare procedures, in order to accredit sus-
objective assessment of the problem and, when tained improvement and to determine when defi-
appropriate, by carrying out comparisons between ciencies occur (3).
different laboratories and different periods of time. Since the 1960s, with the introduction of analytical
According to point 4.12.4 of ISO 15189, “Medical quality control, the approach taken to ensuring
laboratories – particular requirements for quality quality has evolved, focusing in turn on concepts
and competence”, laboratory managers should such as quality guarantees, the management of
implement quality indicators for the systematic quality, the use of quality goals, the specification
monitoring and evaluation of the laboratory’s con- of operational processes and of the resources

Biochemia Medica 2014;24(2):248–57 http://dx.doi.org/10.11613/BM.2014.027

248 ©Copyright by Croatian Society of Medical Biochemistry and Laboratory Medicine. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License
(http://creativecommons.org/licenses/by-nc-nd/3.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

needed to meet the objectives of the system (4) The Clinical Laboratory of Antequera Hospital is a
and total quality management, to the more recent Clinical Management Unit, certified in accordance
focus on clinical safety and risk management (5-8). with ISO 9001:2008, and by the Andalusian Agency
From this standpoint, Technical Specification ISO/ for Healthcare Quality, Advanced level. It provides
TS 22367, “Medical laboratories. Reduction of error the specialised services of clinical biochemistry (4
through risk management and continual improve- physicians), microbiology (1 physician) and haema-
ment” was developed to indicate how risk man- tology (3 physicians), assisted by 21 technical and
agement should be implemented in the structure, 3 administrative staff, using a laboratory informa-
organisation, operation and quality management tion system (LIS) with no electronic request facility
system of the clinical laboratory, with special em- (Open for Labs ®). It is organised into working units
phasis on the pre and post-analytical phases (9). according to the technology involved, with a cen-
Of the three key processes in the clinical laborato- tral laboratory that carries out both scheduled and
ry, the analytical phase is the most highly stand- urgent tasks. This laboratory is equipped with 2
ardised, with well-defined indicators and interna- haematology cell counters, 2 automated haemos-
tionally-accepted specifications for a large number tasis analysers, 2 biochemistry analysers and 2 an-
of biological magnitudes (10-12). All studies agree alysers for immunochemical techniques, with no
that it is in the extra-analytical processes where pre-analytical automation. In addition, the bio-
the greatest number of errors occur, especially in chemical area has an automated urinalysis system,
the pre-analytical stage. These processes, moreo- two blood gas analysers and an osmometer. A spe-
ver, are the most critical (13-15) and the hardest to cial technical unit performs the analyses for stud-
manage, due to the decentralisation of extractions, ies of allergies and autoimmune diseases, using an
involving the participation of various professionals automated system for autoimmunity and allergy,
(physicians, specialists of laboratory medicine, and seminal fluids.
nurses, laboratory technologists and technicians, The laboratory attends a population of 115,155 in-
phlebotomists etc.), organisations and healthcare habitants (2010 census), divided among four Basic
centres. Health Areas with 19 Clinics and Neighbourhood
In this study we describe quality indicators of the Clinical Sampling Centres. In 2011, it received
pre-analytical process, grouping errors according 155,999 requests for analysis, giving rise to
to patient risk as critical or major, and assess their 1,628,852 individual results. All clinical samples,
evaluation over a five-year period. both in hospitals and in primary healthcare, are
taken by external nursing staff, unrelated to the
Material and methods laboratory, and all specimens are centrifuged at
the primary healthcare centre before their transfer
Design to the laboratory. The centrifugation process is su-
A descriptive study was carried out of the tempo- pervised by the nursing department at the centre.
ral evolution of the pre-analytical quality indica- The indicators were designed in 2006 following a
tors proposed for monitoring clinical risk, at the descriptive Modal Failure and Effects Analysis
Clinical Laboratory of Antequera Hospital. (MFEA), carried out as an integral part of the pre-
analytical phase, a process in which the Neigh-
Data source bourhood Clinical Sampling Centres were includ-
The study population was constituted of all the ed. This analysis made it possible to define the crit-
analytical requests received by the laboratory (no ical steps in clinical sampling processes that can
exclusion criteria were applied) from primary lead to mistakes being made, thus enabling cir-
healthcare providers (approximately 41%), special- cuits to be redesigned and procedures restruc-
ists (33%), hospital inpatient departments (6%) and tured. In this analysis, mistakes are classified in ac-
accident and emergency departments (20%), dur- cordance with the corresponding Risk Priority
ing the period 2007-2011, inclusive. Number (RPN), which enables us to prioritise those

http://dx.doi.org/10.11613/BM.2014.027 Biochemia Medica 2014;24(2):248–57

249
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

to be monitored. This number is obtained by mul- tory (a request in which the patient is incorrect-
tiplying the scores given to evaluate the frequency ly identified and/or the sample presents such a
of occurrence of the error and its severity (on a misidentification) or within it, due to an identifi-
scale of 1 to 10, from least to most frequent/se- cation error in producing the aliquot.
vere) and the probability of error detection (on an • Total rate of critical errors.
inverse scale, from 10: low probability to 1: high • Haemolysed sample: for both serum and plas-
probability). ma samples, any degree of haemolysis from
Three main categories of mistake were identified: ‘slightly haemolysed’ to ‘highly haemolysed’, is
“critical” mistakes (RPN > 500), arising mainly from logged as ‘haemolysed sample’, by direct ob-
the type of analytical request and the means used servation and checking against a colorimetric
to communicate and record it, with severe conse- table published by the Spanish Society for Clin-
quences for the patient if not rapidly detected and ical Chemistry, which expresses the equivalent
corrected (16); “major” ones, arising from inade- of a haemoglobin concentration ranging from
quate application of the sampling procedure (RPN: 0 (absence of haemolysis) to 10 g/L.
112-315); and “minor” ones, considered as such due • Clotted sample: in blood with ethylenediami-
to the low likelihood of their occurrence, high netetraacetic acid (EDTA) or with sodium citrate,
probability of detection or absence of severity by direct observation in the presence of coagu-
(RPN < 100). These latter mistakes were only taken lation.
into account for the purposes of reviewing proce- • Inadequate sample container (refers to any type
dures and technical instructions (17). This analyti- of sample).
cal tool has its limitations, especially in the health- • Insufficient sample (refers to any type of sam-
care context, because it is designed for industrial ple).
use before a product is marketed; in any case, any • Blood sample not submitted.
critical error should be seen as such regardless of • Urine sample not submitted.
the NPR value determined. The project was ap-
• Total rate of major errors.
proved by the laboratory’s senior management.
To ensure ongoing improvement, it is of funda- Data collection
mental importance to provide training in safety To compile data on critical errors, a database was
procedures. For this reason, accredited courses designed to record the errors occurring in the lab-
were given to laboratory staff in 2008, 2009 and oratory, when analytical requests are manually en-
2010. In the latter year, the course was oriented to- tered into the LIS. Such errors are expected to
ward transfusion safety, while practical sessions cease when an electronic request system is imple-
for nursing staff in specimen-taking procedures mented. These errors are currently detected either
were organised in 2009. It is essential for staff to when the healthcare clinic requests an analytical
actively learn from mistakes, and so the indicator report that cannot be found, or when primary
results are reported to the services involved every healthcare centres check the analytical reports re-
three months. An annual retrospective analysis is ceived against the lists of patients who provided
made of the results, and this is published in the clinical samples. A second data recording system
corresponding management report. used, for any type of critical error detected in the
laboratory (in the present context, pre-analytical
Outcome variables: indicators and other than those mentioned above), is initiat-
• Critical error in patient identification, when ad- ed by means of a notice of “non-conformity” (18).
ministrative staff manually enters the analytical Such errors may be detected by the analyst on
request into the LIS. checking the report, or also by the physician, on
• Critical errors that could compromise patient observing that the details do not correspond to
safety, recorded by a notice of non-conformity. the clinical data, and are communicated to the lab-
Such errors might originate outside the labora- oratory. In either case, a procedure has been im-

Biochemia Medica 2014;24(2):248–57 http://dx.doi.org/10.11613/BM.2014.027

250
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

plemented to declare and record such events, via defects per million, an error rate of 6.68% corre-
the quality manager or the laboratory manage- sponds to a Sigma value 3, equivalent to border-
ment system. line unacceptable, while an error rate of 0.62% cor-
All errors related to the absence of sample quality responds to a Sigma value 4, which reflects a good
are entered into the LIS directly by the laboratory level of control (19,20).
staff, on receipt of the sample, and are appropri-
ately described by reference to a thesaurus previ- Results
ously created for this purpose. During the study
period, the data-compilation methodology re- A total of 751,441 analytical requests were made to
mained unchanged. the Laboratory Service between 2007 and 2011.
For the study period, comprehensive information
Statistical analysis was obtained on all the proposed quality indica-
Quarterly rates were calculated for the proposed tors, except for one critical error which was record-
indicators, with their respective mean values, com- ed as a non-conformity notice, and not quantified
pared to the total volume of activity, in terms of in 2007. In 2007, there were 141,561 requests, and
analytical requests. The formulas used for the cal- the volume increased by over 10% each year after
culations are summarised in Table 1. The Runs Test this. The rates obtained for each indicator, the to-
was applied to assess changes in the trend of the tal sampling errors for the study period and the P-
series, assuming statistical significance at P < 0.050. value for the Runs Test are summarised in Table 2.
To determine the annual volume of requests and Table 3 shows the Six Sigma values and the aver-
statistically significant indicators according to the age for each indicator.
Runs Test, the base 100 index was calculated with
Pre-analytical critical error indicator
respect to the first year (or quarter) valued, and
these indicators were represented by frequency The total rate of pre-analytical critical errors was
polygons. To estimate the degree of control over 0.047%, obtained from the average non-conformi-
the processes, the results were transformed ac- ty critical error of 0.022% (since 2008), and the val-
cording to the Six Sigma scale, using the Westgard ue of 0.028% of errors in data entered into the LIS.
calculator (19), thus obtaining the frequency with The Runs Test reflected no changes in the trend
which an error is likely to occur. According to the for either type of error. According to the Westgard
table of equivalence between the Sigma level and calculator this error rate corresponds to a Sigma
Six value of 4.9.

Table 1. Formulas for calculation of pre-analytical indicators.

Indicator Formula
Haemolysed sample Number of haemolysed samples x 100 / Total number of requests
Clotted sample Number of clotted samples x 100 / Total number of requests
Inadequate container Number of inadequate containers for each sample type x 100 / Total number of requests
Insufficient sample Number of insufficient samples for each sample type x 100 / Total number of requests
Blood sample not submitted Number of samples with no blood sent x 100 / Total number of requests
Urine sample not submitted Number of samples with no urine sent x 100 / Total number of requests
Total sample incidents Total number of sample quality incidents x 100 / Total number of requests
Critical error “Non conformity” Number of non-conformities x 100 / Total number of requests
Critical error: entry of requests into LIS Number of LIS critical errors x 100 / Total number of requests
Total critical errors Total number of non-conformities + Critical LIS errors x 100 / Total number of requests

http://dx.doi.org/10.11613/BM.2014.027 Biochemia Medica 2014;24(2):248–57

251
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

Table 2. Quarterly rates of pre-analytical indicators.

No urine sent, %

LIS critical error,

Non-conformity
Total incidents,
Total Requests,

critical error, %
No blood sent,
Haemolysis, %

container, %

Total critical
Inadequate

Insufficient
sample, %

errors, N
Clotted,
Quarter
Year-

%
N

2007-1 36,420 9.26 1.56 0.11 0.43 1.24 2.14 14.74 0.055
2007-2 36,587 7.66 1.28 0.14 0.28 1.22 1.73 12.30 0.025
2007-3 32,812 8.31 1.80 0.11 0.25 1.44 1.64 13.55 0.037
2007-4 35,742 9.12 1.79 0.14 0.40 1.58 1.51 14.54 0.053
2008-1 37,340 11.90 1.93 0.11 0.52 1.42 1.63 17.50 0.024 0.035 0.059
2008-2 39,747 10.29 1.34 0.10 0.34 1.32 1.56 14.96 0.033 0.045 0.078
2008-3 34,716 10.38 1.59 0.06 0.41 1.20 1.79 15.42 0.023 0.009 0.032
2008-4 37,708 10.78 1.61 0.05 0.33 1.33 1.88 15.98 0.019 0.037 0.056
2009-1 40,554 13.44 1.62 0.10 0.40 1.45 1.73 18.75 0.015 0.015 0.030
2009-2 40,071 10.34 1.35 0.11 0.39 1.39 1.75 15.33 0.017 0.042 0.060
2009-3 34,906 7.68 1.25 0.05 0.49 1.41 1.64 12.53 0.017 0.017 0.034
2009-4 37,871 7.39 1.19 0.04 0.41 1.32 1.61 11.97 0.024 0.050 0.074
2010-1 39,496 7.64 1.22 0.04 0.39 1.11 1.71 12.11 0.013 0.020 0.033
2010-2 39,750 5.85 1.15 0.06 0.27 1.03 1.40 9.76 0.030 0.023 0.053
2010-3 34,700 5.65 1.46 0.07 0.33 1.19 1.53 10.22 0.026 0.012 0.037
2010-4 37,111 7.60 1.24 0.05 0.26 1.28 1.46 11.90 0.019 0.013 0.032
2011-1 40,803 8.19 1.17 0.05 0.28 1.06 1.70 12.45 0.039 0.020 0.059
2011-2 41,617 8.24 1.01 0.06 0.19 1.18 1.73 12.41 0.024 0.036 0.060
2011-3 35,773 7.68 1.40 0.05 0.26 1.43 1.49 12.30 0.011 0.011 0.022
2011-4 37,717 7.28 1.42 0.03 0.35 1.48 1.52 12.09 0.016 0.011 0.027
TOTAL/
751,441 8.76 1.41 0.08 0.35 1.30 1.66 13.54 0.022 0.028 0.047
Mean
P* 0.039 0.491 0.314 0.491 0.041 0.265 0.003 0.578 0.619 0.196
Pre-analytical indicators are presented as a frequency of total number of requests; *P value of Runs Test.

Of the 134 non-conformity errors reported, 50.74% glucosaline solution from the sampling apparatus).
of the samples did not correspond to the patient The remaining three cases originated in the labo-
described in the analytical request, while 44.03% ratory: twice, samples from different paediatric pa-
of the samples corresponded to an error made by tients were combined in order to obtain a greater
the physician in formulating the analytical request. sample volume, and once, an error arose in the
All of these errors remained undetected by the manual preparation of the sample aliquot.
nursing staff, who failed to confirm the identity of
the patient before performing the extraction. In Indicators related to specimen-collection
one case, a sample required for judicial proceed- errors
ings was invalidated by incorrect collection, and The total error rate for the study period was
on three occasions results were found to be con- 13.54%, with a P-value for the Runs Test of 0.003.
taminated (one with EDTA and the other two with The lowest rate per 100 requests was for the indi-

Biochemia Medica 2014;24(2):248–57 http://dx.doi.org/10.11613/BM.2014.027

252
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

Table 3. Average Six Sigma values for indicator

Year 2007 2008 2009 2010 2011 Total

Indicator N 6 Sigma N 6 Sigma N 6 Sigma N 6 Sigma N 6 Sigma N 6 Sigma
Haemolysed 12,161 2.90 16,202 2.80 15,075 2.80 10127 3.00 12,262 3.00 65,827 2.90
Clotted 2264 3.70 2410 3.70 2090 3.80 1907 3.80 1937 3.80 10,608 3.70
Inadequate container 176 4.60 121 4.70 118 4.70 79 4.80 73 4.90 567 4.70
Insufficient sample 483 4.30 596 4.20 643 4.20 471 4.30 422 4.30 2615 4.20
No blood sample sent 1940 3.80 1972 3.80 2138 3.70 1734 3.80 1993 3.80 9777 3.80
No urine sample sent 2487 3.70 2561 3.70 2588 3.70 2307 3.70 2518 3.70 12,461 3.70
Total incidents 19,511 2.60 23,862 2.50 22,652 2.60 16,625 2.80 19,205 2.70 101,855 2.70
Non-Cfty critical error 37   28   33   36      
LIS critical error 60   48   48   26   31   213  
Total critical errors   85 4.80 76 4.80 59 4.90 67 4.90 287 4.90
Total requests 141,561 100* 149,511 105.62* 153,402 108.36* 151,057 106.71* 155,910 110.14* 751,441  
*Requests (Base 100 = 2007)

cator “inadequate container” (0.08%) followed by 20

18
“insufficient sample” (0.35%), while the highest
16
rates corresponded to the indicators “haemolysed 14
sample” (8.76%), “urine sample not submitted” 12
(1.66%) and “clotted sample” (1.41%). 10
8
Statistically significant differences were identified 6
4
by the Runs Test, in addition to the total rate of in-
2
cidents, for the indicators “haemolysed sample” 0
and “blood sample not submitted”, both at P =
2007-1
2007-2
2007-3
2007-4
2008-1
2008-2
2008-3
2008-4
2009-1
2009-2
2009-3
2009-4
2010-1
2010-2
2010-3
2010-4
2011-1
2011-2
2011-3
2011-4
0.039, and the total rate of errors decreased appre-
ciably from the second quarter of 2009 (Figure 1). Figure 1. Quarterly total rate of major errors (rate per 100 re-
Translated into Sigma metrics, the error rate was quests).
2.7%, fundamentally regarding the indicator
“haemolysed sample”, with a Sigma Six value of
2.9. For the remaining processes, the Sigma values
ranged from 3.7 (“clotted sample”) to 4.7 (‘inade-
quate container”). to another patient or to erroneous results being
reported, thereby provoking a misdiagnosis or in-
correct treatment decisions being established, as a
Discussion result of the procedure and the means by which
analytical requests are made and entered into the
A precise calculation was made of the magnitude
system. A second group of errors, arising from mal-
of the pre-analytical errors associated with our ac-
practice in the specimen collection procedure, and
tivity and area of organisation, after implementing
considered “major” insofar as they affect analytical
a FMEA, during the period 2007-11. This analysis
quality, cause delays, require additional or repeat-
considered, first, critical errors, i.e., those which
ed analyses and increase costs.
could lead to the analytical report being assigned

http://dx.doi.org/10.11613/BM.2014.027 Biochemia Medica 2014;24(2):248–57

253
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

The issue of pre-analytical quality indicators is not these corresponded to the terms “error in patient
a novel one, but this study proposes an innovative identification in inputting data to the LIS”, “patient
approach to developing such indicators and defin- name incorrect on the request” or “no specimen/
ing their scope, in accordance with the recommen- analytical request traceability” (13,14,25-29), al-
dations of the European Committee of Experts on though these studies are methodologically differ-
Management of Safety and Quality in Health Care ent. It is unrealistic to believe that all critical errors
(3), the requirements of ISO 15189 (2), and the are recorded, but all healthcare centres check their
methodology proposed in the Technical Specifica- lists against analytical requests and we analysed
tion ISO 22367:2008. (9) To the best of our knowl- all analytical incidents in search of possible errors.
edge, no previous study has been made to moni- Furthermore, the act of recording them, as pro-
tor pre-analytical errors after the implementation posed in Technical Specification ISO/TS 22367,
of a FMEA. Perhaps the main contribution of the through the creation of a “non-conformity” notice
present study is its scope: the fact that it has incor- (9), means they can be investigated on an individu-
porated all the processes involved, in the hospital al basis, to analyse the cause and the potential
and in primary health care, taking into account the harm to the patient, and to take preventive and
multidisciplinary professionals involved in the corrective measures.
process of obtaining analytical samples, and ad- It is essential to learn from mistakes, and for this
dressing all areas relevant to the clinical laboratory. reason a quarterly report of the results of the indi-
Moreover, the study design is highly effective. We cators is communicated to senior management, to
propose a prior descriptive analysis that achieves the medical officers, to the hospital’s quality con-
high sensitivity and controls the efficiency of the trollers, to the heads of unit (in hospitals and pri-
process, in contrast to a record of predefined or mary care) and to nursing managers. These activi-
fortuitous errors that presents low sensitivity; this ties, together with the laboratory’s policy of always
latter approach only ensures good results with re- rejecting doubtful samples, and in such cases de-
spect to individual activities, and provides little or manding a new sample request and specimen col-
no overall control of the process (21). lection, has led to the establishment of a culture
Indeed, measuring and managing pre-analytical oriented toward safety and the recognition of er-
critical errors continues to be the major challenge rors committed. It is beyond the scope of this study
facing clinical laboratories (7,8,22,23), and the main to investigate possible harm caused to the pa-
difficulty in this respect lies in achieving an effec- tient.
tive, systematic design to ensure that safe proce- After application of FMEA to obtain suitable indi-
dures and processes are adopted, accompanied by cators, the Sigma methodology was used as a sta-
all necessary corrective procedures (3,24). Many tistical tool to quantify the results, expressed as
factors affect patient safety and many variables af- defects per million opportunities (30,31), to evalu-
fect each of the sub-processes, services and health- ate the effectiveness of the training and informa-
care professionals involved in the pre-analytical tion actions aimed at improving the system and to
process. This complexity makes it difficult to man- compare them with the standards proposed by
age pre-analytical critical error, and is one reason Llopis et al. (27) and with the preliminary data on
why the scientific literature in this respect provides quality indicators obtained by the IFCC Working
very limited data, being focused more on estimat- Group Project “Laboratory Errors and Patient Safe-
ing the frequency and specifications of errors re- ty” (32).
lated to sample quality than on examining those
which are potentially critical. In general, insuffi- For critical errors, we obtained a Sigma value of
cient attention is paid to such errors, as regards 4.9, which indicates that the processes involved
their impact on health care (13-15,23). Analysis of are well controlled. However, as suggested by
our results revealed relatively few “critical errors” Llopis et al. (27), given the potential danger to the
compared with previously published data, and patient of such incidents, their specification should

Biochemia Medica 2014;24(2):248–57 http://dx.doi.org/10.11613/BM.2014.027

254
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

be 0%, equivalent to a Sigma value of 6. In this re- for the bias arising from the subjective determina-
spect, there were differences in the type of indica- tion performed, based on visual inspection, and
tors used. In contrast to those proposed in the not on the determination of an index of haemoly-
above-mentioned studies, which focus more on sis using the automated clinical chemistry system.
quality or compliance with minimum requirements This is a limitation of our study. However, the latest
for analytical requests, in the present study, which recommendations for improving the assessment
is aimed at measuring the risk of critical error in of this indicator specify how haemolysis should be
our pre-analytical process, the indicators used are identified and quantified and what threshold
the result of an initial determination of what steps should be used for rejecting the sample, express-
in the processes are to be measured, how and ing the indicator in terms of the total number of
when. These questions are then prioritised accord- samples (32). In our study, this indicator produces
ing to the level of risk. overvalued results, fundamentally because it
In monitoring specimen collection, it is logical to counts all stages of haemolysis, and in conse-
use the percentage of total error for which a bad quence 25% of the comments made lead to the
result is obtained, mainly caused by the indicator test being invalidated. The main cause of the high
“haemolysed sample”. This means that the process rate of in vitro haemolysis in our context is multi-
is borderline unacceptable and requires a thor- factorial: a) in general, serum rather than plasma is
ough review. employed; b) the vacuum system is not widely
used, the specimen tubes being filled directly us-
In this study, the criteria for deciding which indica- ing a syringe (33); c) staff are insufficiently skilled in
tors should be monitored are determined by the the sampling process.
MFEA, specifically, by the NPR values obtained, re-
gardless of any preanalytical incident that may be Our review of the literature revealed different ways
recorded in the analytical report. Comparison with by which this type of indicator could be formulat-
the results of Llopis et al. (27) shows that it is in ed, but these proposals are not comparable with
haemolysis where we are still far from reaching the our results, because in some cases they refer to the
specifications stipulated. In (27), the largest group total number of tests performed (13,15,23), and this
of laboratories used the same method of visual in- type of indicator will always achieve better results
spection, and it is not stated what minimum de- when larger magnitudes are involved and when
gree of haemolysis is recorded. A limiting factor of analytical correctness is less well managed. In oth-
our study, as regards comparison with previous re- er cases, which are more recent and more accu-
search, is the fact that our results are expressed as rate, they are not comparable because they refer
a percentage of analytical requests, whether se- to the total number of samples obtained (28,29,32).
rum or coagulation tube in the absence of a sys- All the studies reviewed agree that the problems
tem enabling us to count the number of tubes, associated with obtaining specimens are the lead-
and that only a single analytical request form was ing cause of pre-analytical errors, with haemolysis
available for all areas of knowledge. presenting greatest variability and comprising the
most frequent cause of sample rejection
Moreover, in the case of haemolysis, due to the ab- (13,23,27,34,35). Plebani et al. highlighted the ur-
sence of recommendations as to the minimum de- gent need to harmonise quality indicators, in view
gree of haemolysis that should be recorded, our of how they can be affected by different interpre-
study used a single heading “haemolysed sample”. tations and how they are measured by each labo-
Nevertheless, the analytical report specified any ratory. These indicators are subject to three main
degree of haemolysis detected, regardless of requirements: they must be patient-centred, con-
whether it interfered to a greater or lesser extent sistent with the requirements of the International
with the analysis of the analyte and whether or not Standard for medical laboratories accreditation,
the test was invalidated in strongly haemolysed and address all stages of the total testing process.
samples. This may compensate, to some degree, (36).

http://dx.doi.org/10.11613/BM.2014.027 Biochemia Medica 2014;24(2):248–57

255
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

In future studies we will consolidate this approach national Standard for medical laboratories accredi-
and analyse the changes presented by the indica- tation, incorporating all the sub-processes and
tors after the implementation of definitive forms personnel involved in the pre-analytical process.
of improvement and after investigating the possi- The processes that trigger “critical” errors, while
ble harm to the patient and the inefficiency costs not achieving the zero-error level, are well control-
generated by such errors. led, and errors in the sample-taking process are of
only marginal importance, mainly concerning the
indicator “haemolysed sample”.
Conclusions
This paper presents a rigorous, patient-centred Potential conflict of interest
analysis of the magnitude of pre-analytical errors, None declared.
in accordance with the requirements of the Inter-

References
  1. De la Fuente Capdevila B, García Vitoria M, Ricós Aguilá C, 11. Buño A, Calafell R, Morancho J, Ramón F, Ricos C, Salas A.
Ventura Solà MC, Gimferrer Campmajó M, Berlanga Caro [Consensus on minimum specifications for analytic quali-
R, et al. [Laboratory medicine and process quality manage- ty]. Lab Cli 2008;1:35-9. (In Spanish)
ment]. Quim Clin 2003;22:44-7. (In Spanish) 12. Calafell R, Gutierrez G, Jou JM, Morancho J, Ramón F, Ricós
2. ISO 15189:2007. Medical laboratories. Particular require- C, et al. [Consensus on the minimum analytical quality spe-
ments for quality and competence. cifications for haematology and special biochemistry para-
  3. The European Health Committee. Committee of experts meters]. Lab Cli 2010;3:87-93. (In Spanish)
on management of safety and quality in healthcare. [Pre- 13. Plebani M, Carraro P. Mistakes in a stat laboratory: types
vention of adverse effects in healthcare: a system approa- and frequency. Clin Chem 1997;43:1348-51.
ch]. Council of Europe CDSP (2004) 41. Strasbourg, 2004. (In 14. Wiwanitkit V. Types and frequency of preanalytical mista-
Spanish) kes in the first Thai ISO 9002:1994 certified clinical labo-
  4. AENOR. [Quality control and quality assurance UNE ratory, a 6 month monitoring. BMC Clin Pathol 2001;1:5.
8402:1995]. Madrid: AENOR, 1995. (In Spanish) http://dx.doi.org/10.1186/1472-6890-1-5.
  5. Westgard JO. Historical perspective on laboratory QC: Whe- 15. Carraro P, Plebani M. Errors in a Stat Laboratory: Types and
re we’ve been and where we’re going. Available at: http:// Frequencies 10 Years Later. Clin Chem 2007;53:1338-42.
www.westgard.com/history-and-future-of-qc/pdf.htm. http://dx.doi.org/10.1373/clinchem.2007.088344.
Accessed July 25, 2012. 16. Lippi G, Blanckaert N, Bonini P, Sol G, Kitchen S, Palic-
  6. CLSI. Laboratory Quality Control Based on Risk Manage- ka V, et al. Causes, consequences, detection, and preven-
ment; Approved Guideline. CLSI document. EP23-A. Wayne, tion of identification errors in laboratory diagnostics. Clin
PA: Clinical and Laboratory Standards Institute, 2011. Chem Lab Med 2009;47:143-53. http://dx.doi.org/10.1515/
  7. Nichols JH. Laboratory quality control based on risk ma- CCLM.2009.045.
nagement. Ann Saudi Med 2011;31:223-8. http://dx.doi. 17. Giménez A, Molina P, Ruiz JJ, Acosta F, López M, Jiménez
org/10.4103/0256-4947.81526. M, et al. [Application of failure mode and effects analysis
  8. Malone B. A new approach to quality control-How can of the pre-analytical phase in a clinical laboratory]. Lab Cli
risk management help labs? Clinical Laboratory News 2010;3:161-70. (In Spanish)
2011;37:11. Available at: http://www.aacc.org/publicati- 18. ISO 9000:2005. Quality management systems. Fundamen-
ons/cln/2011/november/Pages/ANewApproachtoQuali- tals and vocabulary.
tyControl.aspx. Accessed October 30, 2012. 19. Westgard JO. Six Sigma Calculators. Available at: http://
  9. Technical Specification ISO/TS 22367:2008. Medical Labo- www.westgard.com/six-sigma-calculators-2.htm. Acce-
ratories-Reduction of error through risk management and ssed July 25, 2012.
continual improvement. 20. Westgard JO. Six Sigma Quality Design and Control: Desira-
10. Kenny D, Fraser CG, Petersen PH, Kallner A. Strategies to ble precision and requisite QC for laboratory measurement
set global analytical quality specifications in laboratory processes. 2nd ed. Madison, WI: Westgard QC Inc; 2006, 338
medicine – consensus agreement. Scand J Clin Lab Invest pp.
1999;59:585.

Biochemia Medica 2014;24(2):248–57 http://dx.doi.org/10.11613/BM.2014.027

256
Giménez-Marín A. et al. Pre-analytical errors management in the clinical laboratory

21. Govindarajan, R. [Healthcare disorder can be cured: from 29. Beletic A, Sumarac Z, Milinkovic N, Ignjatovic S, Majkic-Sin-
patient safety to the sustainability of the healthcare system gh N. Experience in introduction of quality indicators in
through process-focused management]. 1st Ed. Barcelona: pre-analytical phase. 2nd EFLM-BD European Conference
Marge Médica Books; 2009. (In Spanish). on Preanalytical Phase. Biochem Med 2013;23:A46.
22. Boone DJ. How can we make laboratory testing safer? Clin 30. Gras JM, Philipe M. Application of the Six Sigma con-
Chem Lab Med 2007;45:708-11. http://dx.doi.org/10.1515/ cept in clinical laboratories: a review. Clin Chem Lab Med
CCLM.2007.169. 2007;45:786-96. http://dx.doi.org/10.1515/CCLM.2007.135.
23. Plebani M, Ceriotti F, Messeri G, Ottomano C, Pansini N, Bo- 31. Iren Emekli D, Aslan D, Zorbozan N. Six Sigma in the prea-
nini P. Laboratory network of excellence: Enhancing pa- nalytical phase for continuous quality improvement in cli-
tient safety and service effectiveness. Clin Chem Lab Med nical laboratories. 2nd EFLM-BD European Conference on
2006;44:150-60. http://dx.doi.org/10.1515/CCLM.2006.028. Preanalytical Phase. Biochem Med 2013;23:A11.
24. Shahangian S, Snyder SR. Laboratory medicine quali- 32. Sciacovelli L, O’Kane M, Skaik YA, Caciagli P, Pellegrini C, Rin
ty indicators: A review of the literature. Am J Clin Pathol GD, et al. Quality indicators in laboratory medicine: from
2009;131:418-31. http://dx.doi.org/10.1309/AJCPJF8JI4ZL- theory to practice. Preliminary data from the IFCC Working
DQUE. Group Project “Laboratory Errors and Patient Safety”. Clin
25. Wagar EA, Tamashiro L, Yasin B, Hilborne L, Bruckner DA. Chem Lab Med 2011;49:835-44. http://dx.doi.org/10.1515/
Patient safety in the clinical laboratory. A longitudinal CCLM.2011.128.
analysis of specimen identification errors. Arch Pathol Lab 33. Carraro P, Servidio G, Plebani M. Hemolyzed Specimens: A
Med 2006;130:1662-8. Reason for Rejection or a Clinical Challenge? [Letter] Clin
26. Ricós C, García Victoria M, De la Fuente B. Quality indicators Chem 2000;46:306-7.
and specifications for the extra-analytical phases in clinical 34. Lippi G, Bassi A, Brocco G, Montagnana M, Salvagno
laboratory management. Clin Chem Lab Med 2004;42:578- GL, Guidi GC. Preanalytic error tracking in a labora-
82. http://dx.doi.org/10.1515/CCLM.2004.100. tory medicine department: results of a 1-year experien-
27. Llopis MA, Trujillo G, Llovet MI, Tarrés E, Ibarz M, Biosca C ce. Clin Chem 2006;52:1442-3. http://dx.doi.org/10.1373/
et al. Quality indicators and specifications for key anal- clinchem.2006.069534.
ytical-extranalytical processes in the clinical laboratory. 35. Lippi G, Guidi GC. Risk management in the preanal-
Five years’ experience using the Six Sigma concept. Clin ytical phase of laboratory testing. Clin Chem Lab Med
Chem Lab Med 2011;49:463-70. http://dx.doi.org/10.1515/ 2007;45:720-7. http://dx.doi.org/10.1515/CCLM.2007.167.
CCLM.2011.067. 36. Plebani M, Chioza ML, Sciacovelli L. Towards harmonizati-
28. Carraro P, Zago T, Plebani M. Exploring the initial steps of on of quality indicators in laboratory medicine. Clin Chem
the testing process: Frequency and nature of pre-prea- Lab Med 2013;51:187-95. http://dx.doi.org/10.1515/cclm-
nalytic errors. Clin Chem 2012;58:638-42. http://dx.doi. 2012-0582.
org/10.1373/clinchem.2011.175711.

http://dx.doi.org/10.11613/BM.2014.027 Biochemia Medica 2014;24(2):248–57

257