Review Article

The detection and prevention of errors in laboratory medicine

Mario Plebani
Department of Laboratory Medicine, University Hospital of Padova, Padova, Italy
Corresponding author: Mario Plebani, Department of Laboratory Medicine, University Hospital of Padova, Via Giustiniani, 2, 35128
Padova, Italy. Email: mario.plebani@unipd.it
This article was prepared at the invitation of the Clinical Sciences Reviews Committee of the Association of Clinical Biochemistry.

Abstract
The last few decades have seen a significant decrease in the rates of analytical errors in clinical laboratories. Evidence
demonstrates that pre- and post-analytical steps of the total testing process (TTP) are more error-prone than the analytical
phase. Most errors are identified in pre-pre-analytic and post-post-analytic steps outside of the laboratory. In a patient-
centred approach to the delivery of health-care services, there is the need to investigate, in the TTP, any possible defect that
may have a negative impact on the patient. In the interests of patients, any direct or indirect negative consequence related to a
laboratory test must be considered, irrespective of which step is involved and whether the error depends on a laboratory
professional (e.g. calibration/testing error) or non-laboratory operator (e.g. inappropriate test request, error in patient
identification and/or blood collection). Patient misidentification and problems communicating results, which affect the
delivery of diagnostic services, are recognized as the main goals for quality improvement. International initiatives aim at
improving these aspects. Grading laboratory errors on the basis of their seriousness should help identify priorities for quality
improvement and encourage a focus on corrective/preventive actions. It is important to consider not only the actual patient
harm sustained but also the potential worst-case outcome if such an error were to reoccur. The most important lessons we
have learned are that system theory also applies to laboratory testing and that errors and injuries can be prevented by
redesigning systems that render it difficult for all health-care professionals to make mistakes.

Ann Clin Biochem 2010; 47: 101– 110. DOI: 10.1258/acb.2009.009222

made once consensus has been achieved on a comprehen-

Introduction sive definition of errors in the laboratory testing.
During the past decade, after the publication of the Institute Errors in laboratory medicine are intrinsically obscure as
of Medicine (IOM) report, To Err Is Human, patient safety is they are difficult to identify and, when found, are less
finally the object of medical and public attention.1 The easily understood than other types of medical error.
awareness and understanding of medical errors have Compared with adverse events related to surgery or other
expanded rapidly, with an energetic patient safety move- treatment errors that are often glaring and obvious, labora-
ment promoting safer health care through ‘systems’ sol- tory errors tend to be more insidious and difficult to pin-
utions, thanks to a major message from the IOM report: point in time and place. The difficulties depend largely on
the cause of medical errors and preventable deaths was there being several steps involved. Firstly, there is a time
not careless or incompetent people but bad systems.2 lapse between laboratory testing, physicians’ action and
Compared with other types of medical error, however, patient outcomes. Potential failures in the process steps
errors in laboratory medicine received little attention and nearest to patient intervention are, in fact, more likely to
the reasons for this neglect are complex, as shown in result in patient injury or harm. Failures that occur earlier
Table 1. on in the process are more likely to result in process disrup-
Most of the many different terms used in the literature to tion but ‘active and passive’ defensive barriers – which rely
define errors in laboratory medicine (e.g. mistakes, blun- on technology, people, procedures and administrative con-
ders, defects, outliers, unacceptable results and quality trols – may mitigate their potential harm or may prevent
failure) have negative connotations of blame, individual the recognition of their effect on the final adverse event.
failure and culpability and, even worse, pertain to studies Secondly, the testing process is complex, consists of numer-
focusing on a limited number of total testing process ous steps and stretches across multiple providers. Moreover,
(TTP) steps. A key step toward initiatives aiming to reduce only the analytical phase falls under laboratory control,
errors and improve patient safety in the discipline will be while the pre- and post-analytic phases pertain to different

Annals of Clinical Biochemistry 2010; 47: 101– 110

102 Annals of Clinical Biochemistry Volume 47 March 2010
................................................................................................................................................

Table 1 Errors in laboratory medicine: reasons for a neglect analytical errors (e.g. outliers, unacceptable results) or, as
1. Heterogeneous and ambiguous definition of what a laboratory error in the case of the split-specimen design, are insensitive to
really is; many steps in the testing process, particularly those at the
2. Difficulties in discovering and identifying all types of errors and need beginning and/or at the end of the cycle. One recent and
for well-designed study protocols aiming to evaluate all steps in the
interesting proposal made is to use a neutral term such as
total testing process (TTP);
3. Complexity of TTP and need for cooperation and integration between ‘quality failure’, which mitigates the negative connotations
different health-care providers; associated with previously reported terms, and the related
4. Poor perception by physicians and other stakeholders of the fear of culpability and blame. According to the author,
harmfulness of errors in laboratory medicine; this term means ‘any failure to meet the required output
5. Laboratory professionals reluctant to report and disclose data on
types of errors and their frequency;
quality necessary for optimum patient care anywhere in
6. Increasing use of complementary/alternative testing options (e.g. the pathway from test selection to the return of an appro-
point-of-care, near-patient and self-monitoring). priately interpreted report to requesting clinician’.4 This
definition has a clear focus on patient care and patient out-
comes rather than on processes and procedures. However,
stakeholders other than the laboratory such as the clinician, the term ‘error’ is used in the medical literature, and
the nurse, the patient and others involved in patient identi- should therefore be employed also for errors in laboratory
fication, data entry, specimen collection and transport. In medicine, particularly as they are part of the broader issue
the postanalytic phase there is the possibility of inappropri- of diagnostic errors.7 The Technical Specification released
ate response to the receipt, interpretation and utilization of by the International Organization for Standardization
laboratory information.3 Carefully designed studies, a (ISO/TS 22367) defines laboratory error as
multidisciplinary approach and teamwork are therefore
required for a thorough investigation of TTP. Thirdly, failure of planned action to be completed as intended, or
physicians responsible for making clinical decisions use a wrong plan to achieve an aim, occurring at any part
seldom perceive laboratory errors as a harmful source of of the laboratory cycle, from ordering examinations to
patient adverse events, nor do they understand that most reporting results and appropriately interpreting and
laboratory defects may arise from the pre- and post-analytic reacting to them.8
steps. Fourthly, laboratory professionals are reluctant to
divulge data on the frequency and types of errors observed This comprehensive definition has several advantages and,
in their own setting for fear of a sense of blame, individual in particular, encourages a patient-centred evaluation of
failure and culpability associated with these events.4 This, in errors in laboratory testing. It has been stressed that ‘the
turn, makes it difficult to evaluate the entire testing process promotion of patient-centred care should be translated
and set quality specifications for each step in order to ident- into the need to investigate any possible defect that occurs
ify weakness in policies and procedures to provide opportu- in the TTP and that may eventually have a negative
nities for quality improvement through the formulation and impact on the patient. From the patient viewpoint, any
prioritization of corrective actions. Finally, laboratory direct or indirect negative consequence related to a labora-
testing is no longer performed only in the clinical laboratory tory test must be considered, irrespective of whether the
setting: point-of-care testing, the fastest growing segment of source lies in the pre-, intra-, or post-analytic phase; it is,
current clinical laboratory testing market, near-patient moreover, irrelevant whether an error has been caused by
testing and self-monitoring are widely used alternative or a laboratory professional (e.g. calibration or testing error)
complementary testing options. or by a non-laboratory operator (e.g. patient/specimen mis-
There is therefore an urgent need to evaluate errors in lab- identification, inappropriate test request or interpretation).5
oratory medicine within the reliable framework of the TTP. Therefore, TTP is the unique framework for considering
From the patient’s viewpoint, the integrity of the entire and identifying laboratory errors, both for testing in ‘tra-
process is important and there is a need to prevent any ditional’ clinical laboratories and with point-of-care testing
error in the pre-, intra- or post-analytic phase. From this per- (POCT) or alternative testing types (devices for near-patient
spective, any possible defect in TTP should be investigated testing and self-monitoring).
in order to prevent and obviate any negative impact
on patient care, irrespective of the step involved, and of Types of errors in laboratory medicine
whether the error has been caused by a laboratory
and their prevalence
professional (e.g. calibration or testing error) or by a
non-laboratory operator (e.g. inappropriate test request, Laboratory medicine, as a specialty that had prioritized
error in patient identification, blood collection and/or quality control, has always been at the forefront of error
result interpretation).5,6 reduction. In terms of quality control and error rates, labora-
tory medicine has a far better record than most other fields
in health care. Some studies indicate that, in the analytic
phase, the average error rate is as low as 0.002%; this is func-
Definition of error tioning at the five sigma level. As a comparison, the rates of
The different terms used as synonyms in the literature on infections and medication errors are closer to three sigma,
laboratory errors are the fruit of different study designs that is defect rates .3000-fold those in the clinical labora-
that have almost exclusively allowed the evaluation of tory (Figure 1).2 (The concept of ‘sigma’, according to
 Plebani. Detection and prevention of errors in laboratory medicine 103
................................................................................................................................................

Post-MI
β-blockers Overall health care in the USA (Rand)
Hospital acquired infections
1,000,000 Airline baggage handling

100,000
Detection and
treatment of Adverse
10,000 depression drug Laboratory medicine
Defects
per events
1000
million

100

10 US industry
best-in-class
1
1 2 3 4 5 6
(69%) (31%) (7%) (0.6%) (0.002%) (0.00003%)
σ Level (% defects)

Figure 1 Defect rates in laboratory medicine as compared with other sectors (from reference,2 modified)

stringent statistical quality control criteria, means that a tests ( part per million, ppm) to 447 ppm.17 – 20 This dramatic
process is considered to be in control if the variation – and impressive reduction in errors, by about 300-fold,
expressed as standard deviation [sigma, s] – is less than derives from automation, improved laboratory technology,
1/6 of the difference between the process mean and the assay standardization, well-defined rules for internal
control limit.9 A process reaching Six Sigma [6s] perform- quality control, effective quality assurance schemes and
ance, i.e. a process variation of 6s, will create very few better trained staff. However, recently collected data
defects: 3.4 defects per million opportunities [DPMO]).10 demonstrate that analytical quality is still a major issue.
However, the data are less impressive when one considers Westgard has demonstrated that estimates on a s scale for
the entire testing process of selecting, requesting, identifying common clinical chemistry and coagulation tests are not
patients, collecting and transporting specimens, analysing, satisfactory, ranging from 3 to 4s, at best. Unsatisfactory
reporting, interpreting and utilizing laboratory results. As analytical performance has been described not only in the
stated by Mark Graber11 ‘errors related to laboratory field of clinical chemistry, but also in haematology, coagu-
testing are too common and constitute a significant fraction lation and molecular biology tests.21 In particular, a rela-
of diagnostic errors in medicine today’. Laboratory testing tively high frequency of analytical error has been
in modern clinical medicine is assuming an increasingly documented for immunoassays with associated adverse
important position in the diagnostic process, and in moni- clinical outcomes. In some cases, analytical interference
toring the effects of therapy. Therefore, even a low incidence in immunoassays has resulted in grossly erroneous
of laboratory testing errors among the billions of tests per- results.22,23 Recently collected data on the interference of
formed every day worldwide might have important health para-proteins in many laboratory measurements, including
and patient safety implications. Data collected on laboratory glucose, bilirubin, C-reactive protein, creatinine and
error rates will depend critically on the study design and in albumin, demonstrate that the frequency of this type of
particular the TTP steps investigated. It is therefore easy to error is variable and probably underreported.24 In addition,
understand why the error rates in the literature may vary it has been demonstrated that haemolysis still causes facti-
from one identified error every 33 – 50 events to 1000 tiously high biochemical parameter levels, thus stressing
events, or from 214 to 8300 laboratory results.12 – 16 the need for more appropriate guidelines for the identifi-
cation and appropriate treatment of unsuitable specimens.25
As already stressed, there is no discrepancy between the
Analytical errors impressive reduction in analytical errors achieved over the
Early studies in the field of error in laboratory medicine last few decades and current evidence that analytical
were devoted to identifying analytical errors, the analytical quality is not satisfactory when evaluated on the sigma
phase being the ‘core’ of laboratory work and analytical scale, the Six Sigma being one of the best available
processes under the control of the laboratory staff. An analy- approaches for providing objective estimates and metrics
sis of the data collected and reported in the literature, start- in several industries.11,26 Therefore, despite the impressive
ing with the paper published by Belk and Sunderman in improvement achieved in analytical quality, a body of evi-
1947 through the results collected by the College of dence demonstrates that further improvements are needed.
American Pathologists in the 90s and, finally, the data pub- This should be achieved by setting down and using
lished by Witte and co-workers in 1997 shows that error evidence-based analytical quality specifications in every-
rates have decreased from 162,116 per million laboratory day practice; if this were done, rules for internal quality
104 Annals of Clinical Biochemistry Volume 47 March 2010
................................................................................................................................................

control and external quality assessment procedures would and utilization of laboratory results.31,32 However, on
be more appropriate. Moreover, there is an urgent need exploring the beginning and the end of the loop, it
for standardization programmes aiming at improving emerges that currently these steps, performed neither in
metrological traceability and correcting biases and systema- the clinical laboratory nor, at least in part, under the
tic errors. Finally, more stringent metrics, such as the Six control of laboratory personnel, are more error-prone than
Sigma, should be introduced into clinical laboratories to others.33,34 Recent data on errors in the pre-pre analytical
improve upon current analytical quality. phase – initial procedures performed neither in the clinical
laboratory nor, at least in part, under the control of labora-
tory personnel – underline that failures to order appropriate
Pre- and post-analytical phases diagnostic tests, including laboratory tests, accounted for
While the frequency of laboratory errors varies greatly, 55% of observed incidents of missed and delayed diagnoses
depending on the study design and TTP steps investigated, in the ambulatory setting and 58% of errors in the
a series of papers published between 1989 and 2007 drew Emergency department.35 – 38
the attention of laboratory professionals to the pre- and In the final steps of the loop, the incorrect interpretation of
post-analytical phases, which currently appear to be more diagnostic or laboratory tests was found to be responsible
vulnerable to errors than the analytical phase. Our group for a high percentage of errors in the ambulatory setting
published two papers, in 1997 and 2007, using one study as well as in Emergency departments (Table 3). A very
design that allowed us to investigate most TTP steps in recent paper underlined that failure to inform outpatients
the same clinical context; it also used the same menu of of clinically significant abnormal test results or to document
tests (stat laboratory).27,28 The results (Table 2) demonstrate that the relevant information has been given, appear to be
a significant, although not dramatic, decrease in the error relatively common, occurring in one of every 14 tests.
rates in 2007 but a similar distribution of errors. The pre- Examples include patients not being informed of results of
analytic phase had the highest error rate, the most frequent total cholesterol as high as 8.2 mmol/L (318 mg/dL), a haem-
problems arising from mistakes in tube filling, inappropriate atocrit level as low as 28.6% and a potassium level as low as
containers, and requesting procedures. Identification errors 2.6 mmol/L. The overall rate of failure to inform the patient
were noted for three patients and 14 related tests (875 or record/document communication of information was
ppm) in the 2007 study. There was a significant reduction 7.1%, ranging between practices from 0% to 26%.40 This
in the proportion of specimens collected inappropriately failure to inform patients of clinically important results
from the infusion route in the later study. The main hinders a move away from the traditional paternalistic role
reasons for errors in the postanalytic phase were an exces- of physicians towards a new model incorporating ‘shared
sive turnaround time in the later study, errors in keyboard decision-making’ in which the physician attempts to
entry and missed correction of erroneous findings in the provide the patient and family with the full range of infor-
earlier study. Thanks to improved information procedures, mation, including laboratory results, about the clinical con-
a reduction has been achieved in errors in test transcription dition.41 Further evidence of errors in reacting to laboratory
and ward identification. However, in both the pre- and post- information is given in a study on the prescription
analytic phases, new types of error have emerged, particu- of potassium despite the presence of hyperkalemia.42
larly those attributable to the staff’s application of novel Another study found that more than 2% (2.6% in 2000,
information system procedures. Further studies confirm 2.1% in 2007) of patients with thyrotropin (TSH) results
that the pre- and post-analytical phases are much more higher than 20 mU/mL had no follow-up.43 Finally, an
error-prone than the analytic phase.29,30 interesting study of hospital inpatients showed almost half
of 1095 discharged patients had laboratory and radiology
test results pending and that 9% of these results were poten-
Pre-pre and post-post-analytical steps tially actionable.44 Overall, the above data demonstrate that
While the concept of brain-to-brain loop was developed the initial and final steps of the TTP process, in particular
by Lundberg in 1981, laboratory professionals were not test requesting and reaction to laboratory results, not only
concerned enough about the initial and final TTP steps, are more error-prone than all the other steps, but are more
namely the appropriateness of test requesting, patient and important causes of potential adverse outcomes for patients.
specimen identification and, respectively, the physician’s Currently available data on the relative frequency of errors
reaction to the laboratory report, and the interpretation in the TTP are summarized in Table 4.

Table 2 Frequency and types of errors according to the phase of the

TTP (data from reference28)
Table 3 Post-post analytical errors: frequency of incorrect
Absolute Relative frequency interpretation of diagnostic tests in different clinical settings
frequency ( ppm) (%)
Primary Internal Emergency
1996 2006 1996 2006 Setting care medicine department
Total errors 4667 3092 Incorrect interpretation of 37 38 37
Preanalytic 3186 1913 68.2 61.9 diagnostic tests:
Analytic 617 646 13.3 15.0 estimate (%)
Postanalytic 864 715 18.5 23.1 References 35 39 38
 Plebani. Detection and prevention of errors in laboratory medicine 105
................................................................................................................................................

Table 4 Types and relative frequency of errors in the different serious challenges, particularly in relation to operator com-
phases of the TTP (data from reference7) petence and non-adherence to procedures. A potentially
Relative more dangerous possibility is that the rapid availability of
Phase of the TTP Type of error frequency (%) results and immediate therapeutic intervention might
Pre-pre-analytical Inappropriate test request amplify the clinical impact of errors and translate into
Order entry adverse events for patients.48,49 Recently, we have demon-
Patient/specimen strated a significant number of errors in data transcription
misidentification
Sample collected from
and incomplete data reported using portable glucose
infusion route meters in the hospital setting, thus stressing the vulner-
Sample collection 46– 68.2% ability of post-analytical phase when using POCT.50 While
(haemolysis, clotting, these errors do not arise in the laboratory, they pertain to
insufficient volume, etc.)
the utilization of results by clinicians and care-givers and,
Inappropriate container
Handling, storage and as part of the overall testing and diagnostic process,
transportation should be taken into consideration and managed according
Pre-analytical Sorting and routing to a patient-centred perspective. TTP thus provides the
Pour-off unique framework for analyzing and reducing errors and
Aliquoting, pipetting and 3.0– 5.3%
the risk of errors, not only in ‘centralized’ laboratory
labelling
Centrifugation (time testing but also in POCT and all other alternative site
and/or speed) testing options.
Analytical Equipment malfunction
Sample mix-ups 7.0– 13%
Interference (endogenous or
exogenous)
Impact of errors in laboratory medicine
Undetected failure in It seems likely that only a small proportion of laboratory
quality control
errors results in actual patient harm and adverse events
Post-analytic Erroneous validation of
analytical data thanks to the several barriers and defensive layers present
Failure in reporting/ between the release of laboratory information, the decision-
addressing the report making process and, ultimately, the action on the patient.
Excessive turn-around-time 12.5– 20% The data reported in the literature on the impact of labora-
Improper data entry and
tory errors on patient care are summarized in Table 5.
manual transcription error
Failure/delay in reporting The risk of adverse events and inappropriate care due to
critical values laboratory errors ranges from 2.7% to 12%, while in a larger
Post-post-analytic Delayed/missed reaction percentage of cases (24.4% to 30%), the laboratory error
to laboratory reporting translates into a patient care problem. In the studies pub-
Incorrect interpretation
Inappropriate/inadequate 25– 45.5%
lished by our group, errors resulted in inappropriate admis-
follow-up plan sion to critical care units, inappropriate transfusions,
Failure to order appropriate modifications in heparin and digoxin therapies.27,28 The
consultation impact of laboratory error on the patient’s journey as
regards further inappropriate investigations (both labora-
tory and imaging examinations) and more invasive testing
Errors in POCT and alternative site testing and additional consultations is much higher and although
not necessarily harmful, creates discomfort and incurs
In the literature, data on errors in POCT are scarce, the main higher costs for both patients and the health-care system.
focus being on analytic errors. The claimed advantage of From a risk management viewpoint, the great majority of
POCT, in addition to its reduced turnaround-time is that it laboratory errors with little direct impact on patient care
calls for fewer steps in producing laboratory results. In provide important learning opportunities. In fact, any
addition, ‘errors originating during transport are substan- error, regardless of its apparent triviality, might indicate
tially reduced and post-analytical errors are practically weaknesses in policies and procedures that may not lead
totally eliminated, since results are presented directly to to adverse events in their particular context, but might
the care-giver’.45 This, in turn, should reduce associated cause the patient harm in slightly different circumstances.
errors. However, despite the illusion of simplicity, POCT Therefore, a suitable system for grading laboratory errors
devices are affected by several environmental and operator-
related factors. Managing the pre-, intra- and post-analytic
Table 5 Impact of errors in laboratory medicine on patient outcomes
processes is a major challenge in POCT, just as it is in centra-
lized laboratories. Recently we analysed errors and patient Effect on Adverse events
Authors Number of patient care or risk of adverse
safety problems related to POCT adopting a modified
(Reference) errors (%) events (%)
Kost error classification framework that takes into account
all steps of the testing process, thus demonstrating that Ross JW13 336 30 7
Nutting PA51 180 27 12
POCT reduces errors and risks of error only in a few steps Plebani M27 189 26 6.4
of the entire testing process.46,47 Furthermore, from a risk Carraro P28 160 24.4 2.7
management perspective, POCT has given rise to new and
106 Annals of Clinical Biochemistry Volume 47 March 2010
................................................................................................................................................

according to their seriousness should help identify priorities patients. Errors with negative consequences are considered
for quality improvement and focus corrective/preventive due to negligence or even recklessness and therefore call
actions; the grading system would be designed to consider for sanctions. From this perspective, the connection
not only the real patient harm sustained but also the poten- between proximal actions and bad outcomes is far easier
tial worst-case outcome if such an error were to recur.52 to prove than that between organizational issues and man-
According to the ISO Technical Specification ‘Medical agement decisions. The convenience for lawyers in chasing
laboratories-reduction of error through risk management individual errors rather than collective ones is further
and continual improvement’, any clinical laboratory must reinforced by the willingness of professionals, including
implement processes for: (a) identifying high-risk processes physicians, to accept responsibility for their actions. This
where the potential error could lead to a safety risk for model has several drawbacks, including the fact that often
patients; (b) identifying actual incidents associated with the best people make errors. Even worse, it prevents any
deviations from standard requirements; (c) estimating and initiative designed to disclose medical errors. Hospitals
evaluating the associated risks to patient safety; (d) control- fear public disclosure of reports, which damage reputation,
ling these risks and (e) monitoring the effectiveness of the and cause loss of business, and litigation, while experts
control undertaken.8 In addition to the ISO/TS 22367, a agree that a voluntary system for the reporting of medical
recent proposal, with several merits, suggests that it is poss- errors and adverse events has great potential for improving
ible to assign both an actual (A) and a potential (P) score to safety.55,56 The legal approach, furthermore, encourages
describe the seriousness of an individual laboratory error by defensive medicine which can, in the laboratory setting,
grading it according to a 5 point severity scoring system translate into excessive and inappropriate testing, thus
based on patient outcome.4 The lower score identifies ‘no leading to excessive costs and related inefficiencies.
change in patient management; no adverse clinical
outcome’ as a result of the individual error; and the
higher score, ‘significant adverse clinical outcome’. In a System approach
recent study using this score system over a 30-month The basic premise in the system approach is that humans are
period and considering 714,988 requests for laboratory fallible and errors are to be expected, even in the best poss-
tests received and 658 errors, 75% of errors were given an ible organizations. Errors, seen as consequences rather than
‘A’ score of 1 (no adverse event) while 67.9% were allocated causes, originate in systemic factors, including recurrent
on a ‘P’ score of 5 ( potential significant adverse clinical error traps in the workplace and the organizational pro-
outcome). Once again, these data demonstrate that labora- cesses that give rise to them. Countermeasures are based
tory errors may play a significant role in affecting the on the assumption that although the human condition
overall quality of patient care, including its safety. cannot be changed, the conditions under which humans
work can be improved upon. In particular, defence, barriers
and safeguards occupy a central role in this approach.
Paradigms of error and strategies to improve
High-technology systems, including clinical laboratories,
patient safety
have many defensive layers but sometimes they are
The human error problem, particularly error in medicine, riddled with holes like slices of Swiss cheese and the holes
can be viewed in three ways: the person, the legal and the in the numerous layers may momentarily line up to
system model.53 permit a trajectory of accident opportunity, bringing
hazards into damaging contact with victims.57 The practice
Person approach of laboratory medicine is highly complex. Of the factors
The longstanding and widespread tradition of the person linked to the complexity of TTP, perhaps the most signifi-
approach focuses on the unsafe act, errors and procedural cant are the several steps and the different professionals
violations of individuals at the sharp end: nurses, phys- involved in those steps, which are only partially under
icians, surgeons, anaesthetists, pharmacists and, in rare the control of the laboratory professionals. Figure 2
cases, laboratorians. It views unsafe acts as arising primarily shows the Swiss cheese model adapted to the specific
from aberrant mental processes such as forgetfulness, setting of laboratory medicine, focusing on the most impor-
inattention, poor motivation, carelessness, negligence and tant gaps, and defence layers, the most effective of which
recklessness. Countermeasures, directed mainly at reducing are the identification and documentation of all processes
unwanted variability in human behaviour include disciplin- and procedures, automation and simplification, adequate
ary measures, threat of litigation, retraining, blaming and personnel training, supervision and quality indicators.
shaming. The person approach, the dominant tradition in According to this model, the ability to detect the incipient
medicine, has serious shortcomings: it precludes a detailed indicators and the collective will to implement corrective
analysis of mishaps, incidents, near-misses and isolates measures are essential prerequisites of an effective risk
unsafe acts from their system context, thus precluding an management programme. Process control and proactive
effective risk management policy.54 hazards analysis tools such as FMEA (failure mode and
effect analysis), HACCP (hazards analysis and critical
Legal approach control points) and HAZOP (hazards and operability
According to this model, responsible professionals should studies) have already demonstrated their effectiveness
not make errors as this is part of the duty of care. Such in identifying weaknesses in laboratory processes and
errors are rare but sufficient to cause adverse events to minimizing the risk of errors.58 – 60
 Plebani. Detection and prevention of errors in laboratory medicine 107
................................................................................................................................................

Defences Well-designed procedures The gaps giving guidance on interpretation as well as the procedural
and processes information required to carry out that investigation on
Simplification and automation
the individual patient.70 Similar tools should be used to
Training Complex TTP
Several steps and
improve the appropriateness of test requesting. The accurate
Supervision analysis of all testing processes and documented procedures
different professionals
Effective through proactive tools such as FMEA and HAZOP have
Troubles in the boundaries
lab/clinical already proven effective in reducing the risk probability
interface Shortage of staff index and, therefore, in improving patient safety in labora-
tory testing.58 – 60 These proactive tools are increasingly,
Increasing complexity in test and more readily, accepted by laboratorians and clinicians
Patients’ ordering/result interpretantion because they exploit professional competences through a
adverse events
positive approach to problems by focusing on the examin-
Figure 2 The Swiss cheese model applied to laboratory medicine: the gaps
ation of the entire testing process, thus anticipating major
and defences. According to the model described by Reason57, the presence adverse events and pre-emptively implementing changes
of holes in any one defensive layer does not normally cause adverse events. to prevent them.
Usually this happens only when the holes in many layers line up to permit a
trajectory of accident opportunity
International initiatives to reduce errors
Processes to reduce errors in laboratory medicine in laboratory medicine
In the last few years, in addition to efforts aiming to reduce Recently, the World Alliance for Patient Safety ( promoted
analytic errors and improve analytic quality, important by the World Health Assembly in 2004 to improve patient
achievements have been made in addressing errors in lab- safety as a global initiative) included the communication
oratory medicine. Thanks to the introduction of pre-analytic of critical test results among 23 potential patient safety sol-
workstations, a significant reduction has been achieved in ution topics, thus acknowledging the importance of avoid-
pre-analytic errors in the automation of procedures such ing errors in laboratory testing.71 The second goal of the
as specimen preparation, centrifugation, aliquoting, pipet- Joint Commission 2008 National Patient Safety Goals for
ting and sorting.61,62 The increasing interest shown in devel- Laboratories is to ‘improve the effectiveness of communi-
oping guidelines and standard operating procedures for cation among care-givers’, the first goal being to improve
patient identification, blood collection, sample handling ‘the accuracy of patient identification’, thus underlining
and specimen acceptance or rejection will surely translate the initial and final steps of the testing process.72 The
into higher quality standards.63 – 65 Likewise, significant Working Group on ‘Laboratory errors and patient Safety’
improvements have been made to the postanalytic phase (WG-LEPS) of the International Federation of Clinical
in data transcription as a result of interfacing analysers Chemistry and Laboratory Medicine (IFCC) has undertaken
and laboratory information systems. Importantly, data tran- a project named ‘Model of quality indicators’ based on the
scription is a source of serious errors, particularly if many identification of valuable and consensually accepted
numbers and results have to be entered into the laboratory quality indicators in all steps of the testing process.
computer manually. Further important achievements in the Briefly, 25 quality indicators were selected after discussing
post-analytic phase concern policies and procedures used and analysing the proposal made by 26 clinical laboratories
for reporting critical values as well as initiatives to better enrolled in the Working Group: 16 for the pre-analytic, three
understand and improve upon the efficiency of test report for the analytic and six for the post-analytic phase.
delivery to requesting physicians.66,67 Automatic computer- Currently, participant laboratories may introduce the data
ized communication systems have recently been developed collected in their own institution on each and all quality
to improve the timeliness of notification and avoid potential indicators in a specifically developed website (www.3.
errors for which accreditation programmes require read- centroricercabiomedica.it). A preliminary analysis of col-
back of the result. After being validated by laboratory lected data has been made and reported, but more data
physicians on call, critical values are automatically are needed to allow robust statistical analysis.73 Further
communicated (in real time) to the clinicians, short steps of the project are: (a) to define preliminary quality
message services or alert messages appearing on desktop specifications for each quality indicator; (b) to assess the
computers.67,68 These IT systems, which improve the likeli- data with respect to the preliminary quality specifications;
hood of reaching the physician on call, are easily adapted (c) to re-evaluate the quality specifications and (d) to
to reach patients on their mobile phone or desktop compu- implement an external quality assurance programme by
ter, thereby representing an effective means of reducing or which participating laboratories may evaluate their
indeed eliminating the failure to communicate abnormal performances on the basis of a comparison between the
outpatient test results to users. Further developments results obtained and the desirable quality specifications
concern the introduction of more effective automated identified for each indicator. The overall purpose of the pro-
procedures for data validation and reporting as well as the gramme, therefore, is to encourage each clinical laboratory
implementation of systems which allows an effective knowl- to assess and monitor its own performance not only in its
edge management to support data interpretation and clini- analytic aspects but also in the pre- and post-analytic
cal decision-making at the point of care.69 At the simplest phases. In addition, it should be possible to identify and
this can include a direct link into the laboratory handbook monitor error rates in TTP and to improve upon the
108 Annals of Clinical Biochemistry Volume 47 March 2010
................................................................................................................................................

process on the basis of objective and desirable quality Table 6 Priority areas of patient safety improvement in
specifications defined by the scientific community. laboratory medicine
A further proposal is to undertake a project aimed at iden- 1. Accuracy of patient/specimen identification;
tifying and promoting the adoption of a sentinel event 2. Effectiveness of laboratory data communication;
3. Communication of critical test results;
policy in laboratory medicine. A sentinel event is an unex-
4. Sample acceptability and rejection criteria;
pected occurrence involving death or serious physical or 5. Appropriateness of test request;
psychological injury, or the risk thereof, thus signalling 6. Avoiding manual transcription of data.
the need for immediate investigation and response. The
terms ‘sentinel event’ and ‘medical error’ are not synon-
ymous: not all sentinel events occur because of an error medicine, process analysis, the recording/documentation
and nor do all errors result in sentinel events. Since the of all procedures and processes according to quality stan-
assessment of clinical outcomes in relation to laboratory dards, particularly the ISO 15189: 2007 which has been
diagnostics is notoriously challenging, a further problem is specifically developed for medical laboratories, are key
the identification of those laboratory events arising across tools for changing and improving upon everyday clinical
the totality of the testing process that are most closely associ- practice.76 The accurate analysis and control of all pro-
ated with patient harm. This highlights the compelling need cedures and processes included in the testing process,
for the development and use of reliable and universally particularly if effective tools such as FMEA and HAZOP
agreed performance indicators that would reflect the ‘best techniques are adopted, may significantly reduce weak-
practice’ throughout TTP as well as the identification of ‘lab- nesses and vulnerable steps, thus maximising patient
oratory sentinel events’ that would be conducive to gaining safety. In laboratory medicine we have learnt that TTP is
further knowledge on incidents, and hold providers accoun- the unique framework for identifying and reducing error,
table for patient safety. Available data indicate that potential including initial steps such as patient identification and
‘sentinel events’ might include inappropriate test requests appropriateness in test requesting, and final steps, such as
for critical diseases (e.g. myocardial infarction, pulmonary communication and interpretation of test results.
embolism), patient misidentification, the use of inappropri- The second lesson is that teamwork is the essence of safety,
ate assays, severe analytical errors, critical tests performed particularly if we wish to improve the appropriateness of test
on unsuitable samples (e.g. haemolysed, clotted), the requesting and the reaction to test results. The availability of
release of laboratory results in spite of poor quality expert support systems, which provide information on diag-
control results, the failure to alert critical values and the nostic efficiency and interpretation criteria at the point of
wrong report destination.74 Finally, the lesson we have care, may play a role, but multidisciplinary co-operation
learnt from the worst laboratory error in Italy (a report tran- and collaboration is mandatory for assuring a patient-centred
scription error resulting in HIV transmission to three trans- approach to error reduction. International projects aiming to
planted patients) is the need to avoid manual transcription develop quality indicators for all steps in the testing process,
of data.75 This tragic event caused by a human error, once and to establish related quality specifications, may enable
again, demonstrated weaknesses in the system and holes clinical laboratories to compare, monitor and improve their
in the defensive layers. In particular, it was underlined performances in the every-day practice, not only in the
that this incident occurred because of ‘inadequate awareness analytic phase. Finally, the goals selected by international
of the specificity of the donation process and the organizations, such as the World Alliance for Patient Safety
non-evaluation of the consequences for the already trans- and the Joint Commission, should lead to the prioritization
planted patients or on waiting lists and for the personnel of improvement programmes addressing well-recognized
involved’ and related procedures.75 Taking into consider- critical issues, such as patient identification and communi-
ation current international initiatives and recommendations, cation of laboratory results.
some priority areas of improvement in patient safety and in
error reduction in laboratory medicine might be summar-
DECLARATIONS
ized, as shown in Table 6.
Conflicts of interest: None.
Funding: No funding applicable to this review.
Ethical approval: Ethical approval not required with regard
Conclusions
to the content of this review.
In the last two decades significant advances have been Guarantor: MP.
achieved in the comprehension and reduction of errors in Contributorship: The author is the sole contributor to the
medicine. Finally, those involved are aware that, rather article.
than being caused by ‘bad’ people, errors are indicative of Acknowledgements: This article was prepared at the invita-
weaknesses of the system, a system that includes almost tion of the Clinical Sciences Reviews Committee of the
all the processes and methods we use to organize and Association of Clinical Biochemistry.
carry out virtually everything we do in medicine, including
laboratory medicine. The first lesson we have learnt is,
REFERENCES
therefore, that system theory works and that errors and inju-
ries can be prevented by redesigning systems so as to make 1 Kohn LT, Corrigan JM, Donaldson MS. To Err is Human: Building A Safer
it difficult for care-givers to make mistakes. In laboratory Health System. Washington, D.C.: National Academies Press, 1999
 Plebani. Detection and prevention of errors in laboratory medicine 109
................................................................................................................................................

2 Leape LL. Errors in medicine. Clin Chim Acta 2009;404:2– 5 34 Stroobants AK, Goldschmidt HM, Plebani M. Error budget calculations
3 Boone DJ. Is it safe to have a laboratory test. Accred Qual Assur in laboratory medicine: linking the concepts of biological variation and
2004;10:5 –9 allowable medical errors. Clin Chim Acta 2003;333:169 –76
4 O’Kane M. The reporting, classification and grading of quality failures in 35 Hickner J, Graham DG, Elder NC, et al. Testing process errors and their
the medical laboratories. Clin Chim Acta 2009;404:28 –31 harms and consequences reported from family medicine practices: a
5 Plebani M. Errors in clinical laboratories or errors in laboratory study of the American Academy of Family Physicians National Research
medicine? Clin Chem Lab Med 2006;44:750 – 9 Network. Qual Saf Health Care 2008;17:194 – 200
6 Plebani M. Errors in laboratory medicine and patient safety: the road 36 Gandhi TK, Kachalia A, Thomas EJ, et al. Missed and delayed diagnoses
ahead. Clin Chem Lab Med 2007;45:700 –7 in the ambulatory setting: a study of closed malpractice claims. Ann
7 Plebani M. Exploring the iceberg of errors in laboratory medicine. Clin Intern Med 2006;145:488 –96
Chim Acta 2009;404:16 – 23 37 Wahls TL, Cram PM. The frequency of missed test results and associated
8 International Organisation for Standardisation/Technical Specification. treatment delays in a highly computerized health system. BMC Fam Pract
Medical laboratories – reduction of error through risk management and 2007;8:32 –42
continual improvement. ISO/TS 22367: 2008 38 Kachalia A, Gandhi TK, Pupolo AL, et al. Missed and delayed diagnoses
9 Hinckley CM. Defining the best quality control systems by design and in the Emergency department: a study of closed malpractice claims from
inspection. Clin Chem 1997;43:873 –9 4 liability insurers. Ann Emerg Med 2007;49:196 –205
10 Gras JM, Philippe M. Application of the Six Sigma concept in clinical 39 Graber ML, Franklin N, Gordon R. Diagnostic error in internal medicine.
laboratories: a review. Clin Chem Lab Med 2007;45:789 –96 Arch Intern Med 2005;165:1493 – 9
11 Graber ML. The physician and the laboratory. Partners in reducing 40 Casalino LP, Dunham D, Chin MH, et al. Frequency of failure to inform
diagnostic error related to laboratory testing. Am J Clin Pathol patient of clinically significant outpatient test results. Arch Int Med
2005;124(Suppl. 1):S1 –S4 2009;169:1123 – 9
12 McSwiney RR, Woodrow DA. Types of error within a clinical laboratory. 41 Hartzband P, Groopman J. Keeping the patient in the equation –
J Med Lab Technol 1969;26:340 –6 humanism and health care reform. N Engl J Med 2009;361:554– 5
13 Ross JW, Boone DJ. Institute on Critical Issues in Health Laboratory Practice. 42 Schiff GD, Aggarwal HC, Kumar S, McNutt RA. Prescribing potassium
Wilmington, DE: DuPont Press, 1989:173 despite. hyperkalemia: medication errors uncovered by linking
14 Khoury M, Burnett L, McKay MA. Error rate in Australian chemical laboratory and pharmacy information systems. Am J Med 2001;109:494 – 7
pathology laboratories. Med J Aust 1996;165:128 –30 43 Schiff GD, Kim S, Krosnjar N, et al. Missed hypothyroidism diagnosis
15 Lapworth R, Teal TK. Laboratory blunders revisited. Ann Clin Biochem uncovered by linking laboratory and pharmacy data. Arch Intern Med
1994;31:78 –84 2005;165(5):574 –7
16 Bonini P, Plebani M, Ceriotti F, Rubboli F. Errors in laboratory medicine. 44 Roy CL, Poon EG, Karson AS, et al. Patient safety concerns arising from
Clin Chem 2002;48:691 –8 test results that return after hospital discharge. Ann Intern Med
17 Belk WP, Sunderman FW. A survey of the accuracy of chemical analyses 2005;143:121 – 8
in clinical laboratories. Am J Clin Pathol 1947;17:853 –61 45 Drenk NE. Point of care testing in critical care medicine: the clinician’s
18 Steindel SJ, Howanitz PJ, Renner SW. Reasons for proficiency testing view. Clin Chim Acta 2001;307:3 –7
failures in clinical chemistry and blood gas analysis: a College of 46 Kost GJ. Preventing problems, medical errors, and biohazards in
American Pathologists Q-Probes study in 665 laboratories. Arch Pathol point-of-care testing. Point of Care 2003;2:78 – 88
Lab Med 1996;120:1094 –101 47 Plebani M. Does POCT reduce the risk of error in laboratory testing? Clin
19 Howanitz PJ. Errors in laboratory medicine: practical lessons to improve Chim Acta 2009;404:59 – 64
patient safety. Arch Pathol Lab Med 2005;129:1252 – 61 48 Jones BA, Meier FA. Patient safety in point-of-care testing. Clin Lab Med
20 Witte VanNess SA, Angstadt DS, Pennell BJ. Errors, mistakes, 2004;24:997 –1022
blunders, outliers, or unacceptable results: how many? Clin Chem 1997; 49 Meier FA, Jones BA. Point-of-care testing error. Arch Pathol Lab Med
43:1352 – 6 2005;129:1262 – 7
21 Westgard JO, Westgard SA. The quality of laboratory testing today. Am J 50 Carraro P, Plebani M. Post-analytical errors with portable glucose meters
Clin Pathol 2006;125:343 –54 in the hospital setting. Clin Chim Acta 2009;404:65– 7
22 Tate J, Ward G. Interferences in immunoassay. Clin Biochem Rev 51 Nutting PA, Main DS, Fisher PM, et al. Problems in laboratory testing in
2004;25:105 –20 primary care. JAMA 1996;275:635 – 9
23 Ismail AA. Interference from endogenous antibodies in automated 52 O’Kane MJ, Lynch PL, McGowan N. The development of a system for
immunoassays: what laboratorians need to know. J Clin Pathol the reporting, classification and grading of quality failures in the clinical
2009;62:673 –8 biochemistry laboratory. Ann Clin Biochem 2008;45:129 – 34
24 Dalal BI, Bridgen ML. Factidious biochemical measurements resulting 53 Reason JT. Human Error. New York, NY: Cambridge University Press,
from hematologic conditions. Am J Clin Pathol 2009;131: 195– 204 1990
25 Lippi G, Blanckaert N, Bonini P, Green S, Kitchen S, Palicka V, et al. 54 Leape LL, Fromsom JA. Problem doctors: is there a system level
Haemolysis: an overview of the leading cause of unsuitable specimens in solution? Ann Intern Med 2006;144:107 – 15
clinical laboratories. Clin Chem Lab Med 2008;46:764 – 72 55 Gostin L. A public health approach to reducing error: medical
26 Stankovic AK, Romeo P. The role of in vitro diagnostic companies in malpractice as a barrier. JAMA 2000;283:1742 – 3
reducing laboratory error. Clin Chem Lab Med 2007;45:781 –8 56 Leape LL. Reporting of adverse events. N Engl J Med 2002;14:1633 –8
27 Plebani M, Carraro P. Mistakes in a stat laboratory: types and frequency. 57 Reason J. Human error: models and management. BMJ 2000;320:768 – 70
Clin Chem 1997;43:1348 –51 58 Chiozza ML, Ponzetti C. FMEA: a model for reducing medical errors.
28 Carraro P, Plebani M. Errors in a stat laboratory: types and frequencies 10 Clin Chim Acta 2009;404:75– 8
years later. Clin Chem 2007;53:1338 –42 59 Plebani M, Ceriotti F, Messeri G, Ottomano C, Pansini N, Bonini P.
29 Astion ML, Shojana KG, Hamil TR, Kim S, Ng VL. Classifying laboratory Laboratory network of excellence: enhancing patient safety and service
incidents reports to identify problems that jeopardize patient safety. Am J effectiveness. Clin Chem Lab Med 2006;44:150 –60
Clin Pathol 2003;120:18– 26 60 Signori C, Ceriotti F, Sanna A, et al. Process and risk analysis to reduce
30 Kalra J. Medical errors: impact on clinical laboratories and other critical errors in clinical laboratories. Clin Chem Lab Med 2007;45:742 –8
areas. Clin Biochem 2004;37:1052 – 62 61 Holman JW, Mifflin TE, Felder RA, Demers LM. Evaluation of an
31 Lundberg GD. Acting on significant laboratory results. JAMA automatic pre-analytical robotic workstation at two academic health
1981;245:1762 – 3 centers. Clin Chem 2002;48:540 –8
32 Lundberg GD. The need for an outcome research agenda for clinical 62 Da Rin G. Pre-analytical workstations: a tool for reducing laboratory
laboratory testing. JAMA 1998;280:565 –6 errors. Clin Chim Acta 2009;404:68 –74
33 Laposata M, Dighe A. ‘Pre–pre’ and ‘post – post’ analytic error: high 63 Lippi G, Guidi GC, Mattiuzzi C, Plebani M. Preanalytical variability: the
incidence patient safety hazards involving the clinical laboratory. Clin dark side of the moon in laboratory testing. Clin Chem Lab Med
Chem Lab Med 2007;45:712 –9 2006;44:358 –65
110 Annals of Clinical Biochemistry Volume 47 March 2010
................................................................................................................................................

64 Lippi G, Guidi GC. Risk management in the preanalytical phase of 72 Joint Commission on Accreditation of Healthcare Organizations.
laboratory testing. Clin Chem Lab Med 2007;45:720 – 7 International patient safety goals. 2007. See http//www.
65 Lippi G, Banfi G, Buttarello M, et al. Recommendations for detection and jointcommission.org/PatientSafety/NationalPatientSafetyGoals (last
management of unsuitable samples in clinical laboratories. Clin Chem Lab checked 8 January 2009)
Med 2007;45:728 – 36 73 Sciacovelli L, Plebani M. The IFCC working group on laboratory errors
66 Piva E, Plebani M. Interpretative reports and critical values. Clin Chim and patient safety. Clin Chim Acta 2009;404:79 – 85
Acta 2009;404:52 –8 74 Lippi G, Mattiuzzi C, Plebani M. Event reporting in laboratory
67 Piva E, Sciacovelli L, Zaninotto M, Laposata M, Plebani M. Evaluation of medicine. Is there something we are missing? MLO Med Lab Obs
effectiveness of a computerized notification system for reporting critical 2009;41:23
values. Am J Clin Pathol 2009;131:432 –41 75 Italian National Transplant Centre. Conclusions of the first meeting of
68 Plebani M. Interpretative commenting: a tool for improving the the Joint Commission on the case of HIV transmission from an organ
laboratory-clinical interface. Clin Chim Acta 2009;404:46 – 51 donor in Italy. See http://sysnews.multimodo.com/compoweb/590/
69 Oosterhuis WP, Ulenkate HJ, Goldschmidt HM. Evaluation of File/Conclusions%20of%20the%20first%20meeting.pdf (last checked 13
LabRespond, a new automated validation system for clinical laboratory August 2009)
test results. Clin Chem 2000;46:1811 –7 76 ISO 15189:2007. Medical laboratories – particular requirements for
70 Kay J. Technology to improve quality and accountability. Clin Chem quality and competence.
Lab Med 2006;44:719 – 23
71 World Health Organization: Field review of patient safety solutions;
2008. See http//www.who.int/patientsafety/solutions/patientsafety/
2008_field_review/en/ (last checked 18 January 2009) (Accepted 7 October 2009)