A 54-year-old man was admitted to the hospital for preoperative evaluation and elective knee surgery. On the morning of surgery, the patient was awakened by the phlebotomist who drew his blood for basic laboratories and type and cross-matching.

To ensure proper patient identification, the hospital had implemented a policy requiring a registered nurse or physician to verify the identity of all patients screened for blood transfusion. In practice, after verification of identity, the nurse or physician was required to initial the patient label on the vial of blood.

As it was the change of nursing shift, the bedside nurse for the patient was not available and there were no physicians on the floor at the time. With another floor of patients still to see, the phlebotomist carried the labeled vial of blood out to the nurses’ station, and the label was signed by a random nurse. The sample was sent to the laboratory for analysis.

Later that morning, a laboratory technician noticed a large and surprising change (compared to the previous day’s sample) in the hemoglobin value for a different patient on the same floor. She chose to investigate the discrepancy. Upon review, she realized that the vials of blood for the 54-year-old man had been mislabeled with another patient’s label by the phlebotomist. The reason the hemoglobins were so discrepant for this other patient was that today’s value was that of the 54-year-old man, the wrong patient. On closer examination, it was determined that all the blood samples had been mislabeled, including the vial for type and cross-matching.

Despite the “near miss,” the patient suffered no harm, and another blood specimen was drawn prior to surgery.

by Michael Astion, MD, PhD

Background

Errors in laboratory services encompass a number of problems occurring outside or inside the laboratory.(1) Laboratory service errors are often classified as preanalytic, analytic, or postanalytic (see Table 1 for a standard taxonomy of errors [2]). Preanalytic errors occur before a specimen is analyzed in the laboratory. This case involved a mislabeling error, one of the most common preanalytic errors. Many spurious lab results, which physicians often attribute to errors inside the laboratory, are actually the result of preanalytic problems occurring before laboratory involvement. Analytic errors are problems during the manual or automated specimen analysis in the laboratory. Postanalytic errors occur after production of a laboratory result and include errors associated with misreporting results by oral, written, or electronic communication. In most studies, preanalytic errors are the most common (1–5), but clinically significant errors occur in all phases of testing, and hospital leaders should not assume that most injurious errors are preanalytic.

Errors in laboratory services are relatively uncommon, with published frequencies ranging from 0.1%–5% of tests ordered. Studies are difficult to compare because of large variations in error detection methods and error definitions. Published error rates are likely underestimates because of inadequate detection methods and reluctance to publish or otherwise share errors. For example, although peer-reviewed studies document overall error rates of approximately 0.1% (1) (with specimen mislabeling rates of approximately 0.05% [6]), pathologists involved in internal quality improvement typically encounter hospital mislabeling rates four to six times higher, closer to 0.2%–0.3%. Other preanalytic errors, such as contaminated blood draws from intravenous lines, are equally common; their rates are much higher than published rates.

Because of the gap between published and internal quality improvement data, benchmarks must be interpreted cautiously, and this is even true of the popular and highly regarded Q probes and Q tracks programs from the College of American Pathologists (www.cap.org). These voluntary programs involving laboratories across the United States and several foreign countries use consistent methods to provide benchmark error rates for many of the steps in the laboratory testing process. They also identify best practices associated with better performance.(5,7) For example, a Q probes study of critical value reporting in 623 institutions showed that 5% of critical values were not successfully communicated to hospital staff.(8) Other examples of previous Q probes studies include turnaround times for biochemical markers of myocardial injury and the rate of manual review of peripheral blood smears by laboratory personnel. Examples of ongoing Q tracks programs include monitoring errors in patient wristbands, blood culture contamination, and submission of unacceptable specimens.(7) Moreover, some successful Q track interventions are disseminated through peer-reviewed publication. For example, a study of wristband errors from 217 institutions showed that the wristband error rate in participating institutions decreased from 7% to 3% over the 2 years of study.(9) The most common error was patients without wristbands (72% of the errors), and the most important intervention to reduce errors was phlebotomist refusal to draw blood on patients without wristbands.

Few rigorous studies correlate errors in laboratory medicine with adverse patient outcomes. Based on limited data, it is likely that 5%–10% of laboratory service errors lead to harm (4,10,11), and that adverse events are more frequent in the acute care setting. If one conservatively estimates that 1% of laboratory results are erroneous, and 5% of the errors adversely affect patients, then a typical university hospital, performing 2 million tests per year, would have 1000 laboratory-related adverse events per year. Although most of these would involve minimal harm, a few cases annually are likely to be associated with serious injury. Some de-identified examples of laboratory errors associated with patient harm are described in Table 2.

Error Detection

Laboratory service errors are detected in many ways, including caregiver complaints, incident reports from inside or outside the laboratory, error checking protocols within the laboratory, and a variety of laboratory management reports.

Most mislabeling errors are detected by one of two methods (patient identifiers and delta checking, described below), and overall only 15% of mislabeled specimens result in the release of a laboratory result—in other words, 85% are detected before the specimen is run and released.(6) Most mislabeling errors are detected when the specimen and requisition first enter the laboratory, when technicians check that two patient identifiers and other information on the requisition match the two identifiers and other information on the blood tubes. Mismatch of patient identifiers or failure to label the tube leads to test cancellation and caregiver notification. This front-end error checking missed the mislabeling error in this case, probably because the requisition accompanying the mislabeled tubes was also mislabeled with the same identifiers.

The second common way mislabeling errors are detected is “delta checking” (12), and this was the case here. Delta checking is an automated comparison of the patient’s current and previous lab test value, and the comparison takes place inside the laboratory information system. If the difference between the values (“delta”) is greater than a defined limit, the result is flagged, and a procedure is implemented to determine if an error occurred. Delta checking procedures vary but usually involve repeating the test and investigating for misidentification. In this case, a different patient on the same floor received the hemoglobin value of the patient awaiting surgery, and the two values differed enough to trigger the error flag. The advantage of delta checking is that it does not rely on human vigilance. The shortcomings are that it requires previous values, and it is difficult to apply to text results or to numeric results where large changes are expected in common clinical conditions (for example, troponin in myocardial infarction). In addition, delta limits are imperfect since they involve the inherent tradeoff between false positive and false negative error detection.

Specific Errors in This Case

The particular error in this case is a mislabeling error occurring outside the laboratory because of two active errors. The primary error, committed by a phlebotomist, was mislabeling of patient A’s (pre-op for knee surgery) blood tube with labels from patient B. This error puts A and B at risk, since A fails to receive a result needed for care and B potentially receives test results that are not his.

We are not told if the phlebotomist's mislabeling error is due to lack of knowledge, or if it is noncognitive, that is, due to interruption in a process that is normally automatic. Most mislabeling errors such as this are noncognitive errors made by experienced phlebotomists. The usual contributing factors to noncognitive errors are high work volume, fatigue, and distractions, and these same factors provide the incentive structure that produces and sustains error-prone shortcuts.(13) In addition, although we are not told this, the specimen mislabeling error in this case may have been related to shortcuts (“workarounds”) by the phlebotomist, such as simultaneously carrying prelabeled tubes from multiple patients, or skipping the step where the patient actively identifies himself.

The second error in this case involves the floor nurse who initialed the blood tube without participating in patient identification. Simplistically, this error could be classified as a rule violation and failure of a double-check. However, a more useful way of understanding it is to assume that the error is related to a shortcut, and that the shortcut is conforming, not deviant. This means that the shortcut makes sense in light of the work environment (13) and would tend to be adopted by many of the nurses in this unit. Therefore, it would be ineffective to restrict interventions to this one nurse and more important to understand the work environment and the forces that may be leading to this and similar workarounds.

Interventions

Table 3 provides specific examples to illustrate “intermediate” and “strong” interventions to reduce errors in laboratory services.(14) “Weak” interventions, which are not shown, include training, memos, and warning labels. Intermediate and stronger interventions involve strategies such as standardization, automation, matching of workflow to staffing, and the elimination of error-prone steps.

Typical interventions to reduce mislabeling, from weakest to strongest, are:

• Weak: Retrain the phlebotomist and nurse regarding following the proper policy and procedure for blood collection.
• Weak, but slightly stronger: Retrain the phlebotomist and nurse and monitor their behavior.
• Intermediate: Retrain all phlebotomists and nurses and monitor their behavior.
• Stronger: Standardization of phlebotomy procedures around a new process (“single piece flow”) in which only one patient with one set of patient labels are handled at a time, and a second patient is not phlebotomized until the first patient’s specimens are submitted to the lab. This is combined with optimization of staffing, and coaching and monitoring phlebotomists and nurses regarding the new procedure.
• Strongest: All of the above, plus barcode-based, semi-automated patient identification and specimen collection incorporating automated error-proofing techniques to assure the correct patient, labels, specimen containers, and order of draw.(15)

The strong interventions are more expensive and difficult to implement, but they are likely to be the only truly robust solutions to the pervasive problem of mislabeling. Moreover, we would probably find that they have a reasonable financial return on investment if the true costs of mislabeling could be determined.

Take-Home Points

• Clinically significant laboratory errors can occur in any step in the laboratory testing process.
• Mislabeling errors are one of the most common preanalytic errors in laboratory services, and they are usually detected by front end error checking by the laboratory or by automated delta checking.
• Training is a weak intervention to reduce mislabeling errors. Health care systems would be better off focusing on standardizing less error-prone procedures and removing incentives in the work environment that encourage shortcutting.

Michael Astion, MD, PhD
Associate Professor and Director of Reference Laboratory Services
Editor-in-Chief, Laboratory Errors and Patient Safety Newsletter
University of Washington, Department of Laboratory Medicine

1. Bonini P, Plebani M, Ceriotti F, Rubboli F. Errors in laboratory medicine. Clin Chem. 2002;48:691–698. [go to PubMed]

2. Astion ML, Shojania KG, Hamill TR, Kim S, Ng VL. Classifying laboratory incident reports to identify problems that jeopardize patient safety. Am J Clin Pathol. 2003;120:18–26. [go to PubMed]

3. Lapworth R, Teal TK. Laboratory blunders revisited. Ann Clin Biochem. 1994;31:78–84. [go to PubMed]

4. Plebani M, Carraro P. Mistakes in a stat laboratory: types and frequency. Clin Chem. 1997;43:1348–1351. [go to PubMed]

5. Howanitz PJ. Errors in laboratory medicine: practical lessons to improve patient safety. Arch Pathol Lab Med. 2005;129:1252–1261. [go to PubMed]

6. Valenstein PN, Raab SS, Walsh MK. Identification errors involving clinical laboratories: a College of American Pathologists Q-Probes study of patient and specimen identification errors at 120 institutions. Arch Pathol Lab Med. 2006;130:1106–1113 [go to PubMed]

7. Zarbo RJ, Jones BA, Friedberg RC, et al. Q-tracks: a College of American Pathologists program of continuous laboratory monitoring and longitudinal tracking. Arch Pathol Lab Med. 2002;126:1036–1044. [go to PubMed]

8. Howanitz PJ, Steindel SJ, Heard NV. Laboratory critical values policies and procedures: a College of American Pathologists Q-Probes study in 623 institutions. Arch Pathol Lab Med. 2002;126:663–669. [go to PubMed]

9. Howanitz PJ, Renner SW, Walsh MK. Continuous wristband monitoring over 2 years decreases identification errors: a College of American Pathologists Q-Tracks study. Arch Pathol Lab Med. 2002;126:809–815. [go to PubMed]

10. Yuan S, Astion ML, Schapiro J, Limaye AP. Clinical impact associated with corrected results in clinical microbiology testing. J Clin Microbiol. 2005;43:2188–2193. [go to PubMed]

11. Astion ML, Krueger-Nielsen S, Davison B, Miller B. Errors and patient outcomes associated with problems in stat chemistry testing. Clin Chem. 2004;50(suppl 6):A114–A115.

12. Nosanchuk JS, Gottmann AW. CUMS and delta checks. A systematic approach to quality control. Am J Clin Pathol. 1974;62:707–712. [go to PubMed]

13. Dekker S. The Field Guide to Human Error Investigations. Aldershot, United Kingdom: Ashgate Publishing; 2002.

14. Astion ML. Putting power into patient safety interventions. Lab Errors Patient Saf. 2005;1:9–11.

15. Bologna LJ, Mutter M. Life after phlebotomy deployment: reducing major patient and specimen identification errors. J Healthc Inf Manag. 2002;16:65–70. [go to PubMed]

Table 1. A Representative Taxonomy of Laboratory Errors in Use by a Number of Laboratories with Minor Modifications (Adapted from Reference 2).

Table 2. Examples of Patient Harm Associated with Laboratory Errors

Table 3. A List of Intermediate and Strong Interventions with Specific Examples Related to Clinical Laboratory Services.(14)