Chapter 1 Variation Errors and Quality in The Clinical Laboratory 2013 Accurate Results in The Clinical Laboratory
Chapter 1 Variation Errors and Quality in The Clinical Laboratory 2013 Accurate Results in The Clinical Laboratory
1
Variation, Errors, and Quality in the
Clinical Laboratory
Jorge Sepulveda
Columbia University Medical Center, New York, New York
samples) [6]. Other frequent causes of pre-analytical noncognitive errors, commonly known as slips and
errors include the following: lapses, due to interruptions in a process that is routine
or relatively automatic. Whereas the first type can be
Patient identification error
prevented by increased training, competency evalua-
Tube-filling error, empty tubes, missing tubes, or
tion, and process aids such as checklists or cheat
wrong sample container
sheets summarizing important steps in a procedure,
Sample contamination or collected from infusion
noncognitive errors are best addressed by process
route
improvement and environment re-engineering to mini-
Inadequate sample temperature.
mize distractions and fatigue. Furthermore, it is useful
Table 1.1 provides a complete list of errors, includ- to classify adverse occurrences as activethat is, the
ing pre-analytical, analytical, and post-analytical immediate result of an action by the person perform-
errors, that may occur in clinical laboratories. ing a taskor as latent or system errors, which are sys-
Particular attention should be paid to patient identifi- tem deficiencies due to poor design or implementation
cation because errors in this critical step can have that enable or amplify active errors. In one study, only
severe consequences, including fatal outcomes, for approximately 11% of the errors were cognitive, all in
example, due to transfusion reactions. To minimize the pre-analytical phase, and approximately 33% of the
identification errors, health care systems are using errors were latent [5]. Therefore, the vast majority
point-of-care identification systems, which typically of errors are noncognitive slips and lapses perfor-
involve the following: med by the personnel directly involved in the
process. Importantly, 92% of the pre-analytical, 88% of
1. Handheld devices connected to the laboratory
analytical, and 14% of post-analytical errors were pre-
information systems (LIS) that can objectively
ventable. Undoubtedly, human factors, engineering,
identify the patient by scanning a patient-attached
and ergonomicsoptimization of systems and process
bar code, typically a wrist band.
redesigning to include increased automation and user-
2. Current laboratory orders can be retrieved from the
friendly, simple, and rule-based functions, alerts,
LIS.
barriers, and visual feedbackare more effective than
3. Ideally, collection information, such as correct tube
education and personnel-specific solutions to consis-
types, is displayed in the device.
tently increase laboratory quality and minimize errors.
4. Bar-coded labels are printed at the patients side,
Immediate reporting of errors to a database accessi-
minimizing the possibility of misplacing the labels
ble to all the personnel in the health care system,
on the wrong patient samples.
followed by automatic alerts to quality management
Analytical errors are mostly due to interference or personnel, is important for accurate tracking and timely
other unrecognized causes of inaccuracy, whereas correction of latent errors. In our experience, reporting
instrument random errors accounted for only 2% of all is improved by using an online form that includes
laboratory errors in one study [5]. According to that checkboxes for the most common types of errors
study, most common post-analytical errors were due to together with free-text for additional information
communication breakdown between the laboratory and (Figure 1.1). Reviewers can subsequently classify errors
the clinicians, whereas only 1% were due to miscommu- as cognitive/noncognitive, latent/active, and internal to
nication within the laboratory, and 1% of the results had laboratory/internal to institution/external to institution;
excessive turnaround time for reporting [5]. Post- determine and classify root causes as involving human
analytical errors due to incorrect transcription of labora- factors (e.g., communication and training or judgment),
tory data have been greatly reduced because of the software, or physical factors (environment, instrument,
availability of automated analyzers and bidirectional hardware, etc.); and perform outcome analysis.
interfaces with the LIS [5]. However, transcription errors Outcomes of errors can be classified as follows:
and calculation errors remain a major area of concern in
those testing areas without automated interfaces 1. Target of error (patient, staff, visitors, or
between the instrument and the LIS. Further develop- equipment).
ments to reduce reporting errors and minimize the test- 2. Actual outcome on a severity scale (from unnoticed
ing turnaround time include autovalidation of test to fatal) and worst outcome likelihood if error was
results falling within pre-established rule-based para- not intercepted, because many errors are corrected
meters and systems for automatic paging of critical before they cause injury. Errors with significant
results to providers. outcomes or likelihoods of adverse outcomes should
When classifying sources of error, it is important to be discussed by quality management staff to
distinguish between cognitive errors, or mistakes, which determine appropriate corrective actions and
are due to poor knowledge or judgment, and process improvement initiatives.
TEST ORDERING
Duplicate order Order misinterpreted (test ordered 6 intended test)
Ordering provider not identified Inappropriate/outmoded test ordered
Ordered test not performed (include add-ons) Order not pulled by specimen collector
SAMPLE COLLECTION
Unsuccessful phlebotomy Check-in not performed (in the LIS)
(Continued)
OTHER
Clearly, efforts to improve accuracy of laboratory continuously improve the quality of the products, a
results should encompass all of the steps of the testing concept known as continuous quality improvement. A
cycle, a concept expressed as total testing process or major component of a quality assurance program is
brain-to-brain testing loop [7]. Approaches to quality control (QC), which involves the use of periodic
achieve error minimization derived from industrial measurements of product quality, thresholds for
processes include total quality management (TQM); [8] acceptable performance, and rejection of products that
lean dynamics and Toyota production systems; [9] do not meet acceptability criteria. Most notably, QC is
root cause analysis (RCA); [10] health care failure applied to all clinical laboratory testing processes and
modes and effects analysis (HFMEA); [11,12] failure equipment, including testing reagents, analytical
review analysis and corrective action system (FRACAS) instruments, centrifuges, and refrigerators. Typically,
[13]; and Six Sigma [14,15], which aims at minimizing for each clinical test, external QC materials with
the variability of products such that the statistical fre- known performance, also known as controls, are run
quency of errors is below 3.4 per million. A detailed two or three times daily in parallel with patient speci-
description of these approaches is beyond the scope of mens. Controls usually have preassigned analyte con-
this book, but laboratorians and quality management centrations covering important medical decision levels,
specialists should be familiar with these principles for often at low, medium, and high concentrations. Good
efficient, high-quality laboratory operation [8]. laboratory QC practice involves establishment of a
laboratory- and instrument-specific mean and standard
QUALITY IMPROVEMENT IN THE deviation for each lot of each control and also a set of
CLINICAL LABORATORY rules intended to maximize error detection while mini-
mizing false rejections, such as Westgard rules [16].
Quality is defined as all the features of a product Another important component of quality assurance for
that meet the requirements of the customers and the clinical laboratories is participation in proficiency test-
health care system. Many approaches are used to ing (or external quality assessment programs such as
improve and ensure the quality of laboratory opera- proficiency surveys sent by the College of American
tions. The concept of TQM involves a philosophy of Pathologists), which involves the sharing of samples
excellence concerned with all aspects of laboratory with a large number of other laboratories and compari-
operations that impact on the quality of the results. son of the results from each laboratory with its peers,
Specifically, TQM approaches apply a system of statis- usually involving reporting of the mean and standard
tical process control tools to monitor quality and pro- deviation (SD) of all the laboratories running the same
ductivity (quality assurance) and encourage efforts to analyzer/reagent combination. Criteria for QC rules
FIGURE 1.1 Example of an error reporting form for the clinical laboratory.
not exceed the RCV has a greater than 95% probability deviations from expected results. In summary, the use of
that it is due to the combined analytical and intraindivi- TAAE and RCV brings objectivity to error evaluation,
dual biological variation; in other words, the difference QC and proficiency testing practices, and clinical decision
between the two creatinine results (measured without making based on changes in laboratory values.
repeats) should exceed 23.5% to be 95% confident
that the change is due to a pathological condition.
Conversely, for any change in laboratory values, the RCV CONCLUSIONS
formula can be used to calculate the probability that it is
due to analytical and biological variation [17,19,20]. See As in other areas of medicine, errors are unavoid-
Table 1.2 for examples of RCV at the 95% confidence able in the laboratory. A good understanding of the
limit, using published intraindividual variation and the sources of error together with a quantitative evaluation
authors laboratory imprecision. Ideally, future LIS of the clinical significance of the magnitude of the
should integrate available knowledge and patient- error, aided by the establishment of limits of accept-
specific information and automatically provide estimates ability based on statistical principles of analytical and
of expected variation based on the previous formulas to intraindividual biological variation, are critical to
facilitate interpretation of changes in laboratory values design a quality program to minimize the clinical
and guide laboratory staff regarding the meaning of impact of errors in the clinical laboratory.
References [11] Chiozza ML, Ponzetti C. FMEA: a model for reducing medical
errors. Clin Chim Acta 2009;404(1):758.
[1] Forsman RW. The value of the laboratory professional in the [12] Southard PB, Kumar S, Southard CA. A modified delphi meth-
continuum of care. Clin Leadersh Manag Rev 2002;16(6):3703. odology to conduct a failure modes effects analysis: a patient-
[2] Forsman RW. Why is the laboratory an afterthought for centric effort in a clinical medical laboratory. Qual Manag
managed care organizations? Clin Chem 1996;42(5):81316. Health Care 2011;20(2):13151.
[3] Hallworth MJ. The 70% claim: what is the evidence base? [13] Krouwer J. Using a learning curve approach to reduce labora-
Ann Clin Biochem 2011;48(Pt 6):4878. tory errors. Accreditation and Quality Assurance: Journal for
[4] Gaines AR, Pierce LR, Bernhardt PA. Fatal Iatrogenic Quality, Comparability and Reliability in Chemical
Hypoglycemia: Falsely Elevated Blood Glucose Readings with a Measurement 2002;7(11):4617.
Point-of-Care Meter Due to a Maltose-Containing Intravenous [14] Llopis MA, Trujillo G, Llovet MI, Tarres E, Ibarz M, Biosca C,
Immune Globulin Product. [cited; Available from: ,http://www. et al. Quality indicators and specifications for key analytical-
fda.gov/BiologicsBloodVaccines/SafetyAvailability/ucm155099. extranalytical processes in the clinical laboratory. Five years
htm.; 2009 [06.18.2009] experience using the six sigma concept. Clin Chem Lab Med
[5] Carraro P, Plebani M. Errors in a stat laboratory: types and fre- 2011;49(3):46370.
quencies 10 years later. Clin Chem 2007;53(7):133842. [15] Gras JM, Philippe M. Application of the six sigma concept in clini-
[6] Carraro P, Zago T, Plebani M. Exploring the initial steps of the cal laboratories: a review. Clin Chem Lab Med 2007;45(6):78996.
testing process: frequency and nature of pre-preanalytic errors. [16] Westgard JO, Darcy T. The truth about quality: medical useful-
Clin Chem 2012;58(3):63842. ness and analytical reliability of laboratory tests. Clin Chim
[7] Plebani M, Lippi G. Closing the brain-to-brain loop in labora- Acta 2004;346(1):311.
tory testing. Clin Chem Lab Med 2011;49(7):11313. [17] Fraser CG. Biological variation: from principles to practice.
[8] Valenstein P, editor. Quality management in clinical labora- Washington, DC: AACC Press; 2001.
tories. Northfield, IL: College of American Pathologists; [18] Westgard J. Desirable Specifications for Total Error,
2005. Imprecision, and Bias, derived from intra- and inter-individual
[9] Rutledge J, Xu M, Simpson J. Application of the Toyota produc- biologic variation. [cited 2012; Available from: ,http://www.
tion system improves core laboratory operations. Am J Clin westgard.com/biodatabase1.htm.; 2012.
Pathol 2010;133(1):2431. [19] Kroll MH. Multiple patient samples of an analyte improve
[10] Dunn EJ, Moga PJ. Patient misidentification in laboratory medi- detection of changes in clinical status. Arch Pathol Lab Med
cine: a qualitative analysis of 227 root cause analysis reports in 2010;134(1):819.
the Veterans Health Administration. Arch Pathol Lab Med [20] Fraser CG. Improved monitoring of differences in serial labora-
2010;134(2):24455. tory results. Clin Chem 2011;57(12):16357.