By Renee DiIulio

Patient A is being monitored for glucose tolerance. Her latest report indicates she has a fasting blood glucose of 75 mg/dL, but because of a collection error, this is not correct. Her blood glucose level is really 95 mg/dL. However, because normal glucose tolerance falls within 70 mg/dL to 95 mg/dL, the error is unproblematic and goes unnoticed.

Patient B, who is lying in the intensive care unit on a ventilator, is not so lucky. A mixup has led to his results also being wrong, an error that leads to his cardiac arrest. The patient does not die but does endure a longer and more expensive stay, involving more treatment, more lab tests, and more discomfort.

What went wrong in each case? A clear audit trail will provide an answer, but only if the error is noticed and examined. In the first case, this is unlikely to occur. In the latter case, the error may be found, but the information may not be used. Error reporting is discouraged in today’s medical culture, and too little data has slowed the development of evidence-based medicine practices.

But this is changing. Government organizations, such as the Centers for Disease Control and Prevention (CDC of Atlanta)’s Institute for Quality in Laboratory Medicine (IQLM), and industry associations, such as the College of American Pathologists (CAP of Northfield, Ill), are investigating qualifiers and practices that can help reduce errors.

The Layout
Before errors can be reduced, they must first be identified and tracked. Errors can occur at any point in the test cycle—there is plenty of room for error from the moment the test is ordered until the moment the physician interprets the results and acts on them. Between these two points lies patient preparation and identification; specimen collection; transport and delivery; accessioning; processing; testing; reporting, including format and clarity; and report delivery.

Fred Meier, MD

Unfortunately, the lab does not have control over all of these steps. The beginning and end stages typically happen outside the lab’s purview. The physician or nurse completes the order, and according to one source, rumors about physician handwriting have merit. Nurses or phlebotomists, not lab personnel, frequently collect the sample, and collection can sometimes affect its stability. For example, improper care can cause red blood cells to hemolyze. On the other end, physicians interpret results within their knowledge base and could benefit from the insight of someone who more fully understands what the data mean. Currently, however, labs have no role in these portions of the cycle.

Stephen Raab, MD

Even so, labs are responsible for the results produced. Information from the lab affects at least 70% of medical decisions. Errors anywhere in the process tend to reflect badly on the lab. By simply tracking errors, labs can not only identify where most of the problems occur and develop processes to reduce their frequency, but they can also immediately improve their performance.

“The Q-Probes reports have shown that if people track measures over time, things get better. This was seen in reports on patient and/or specimen misidentification, blood culture contamination, and the reporting of critical values,” says Fred Meier, MD, section head for system laboratories, Henry Ford Health System (Detroit).

Problem Areas
Error tracking can also determine where in the process mistakes are more likely to occur. Data collected thus far indicates that a greater number of errors occur in the pre- and postanalytic stages.

In 2002, an Italian group reviewed the literature published during the previous 8 years and found that 68%–87% of errors were reported in these two stages.¹ Information shared at the IQLM conference in April also supports these conclusions, according to Julie Taylor, PhD, senior service fellow, CDC. And Meier states, “Some have suggested that the pre- and postanalytic phases are, at their best, at two sigma while the analytic stage is at four sigma.” Six sigma processes aim to keep the defects per million (DPM) low, close to fewer than 1 in 1 million; a 5.8 sigma would have an error rate of 0.001%.

Not all errors will negatively impact patient care, but they do have the potential to kill patients as well as lead to increased testing, misdiagnosis, unnecessary surgery, and medicine errors. How often this really happens, though, is not well-understood.

Faux Pas Frequency
The Institute of Medicine (IOM of Washington, DC) determined in 1999 that, each year, between 44,000 and 98,000 patients in the United States die due to medical error. However, this number does not specify where the errors have occurred.

“The exact number of laboratory errors is not known because we do not have a standard system for reporting errors that covers all aspects of testing for all types of laboratory services in all of the various practice settings. One of the goals of the IQLM is to work toward developing systems that help us understand and track the number and types of errors that occur,” says Taylor.

In the meantime, individual efforts have produced some data. A 1997 study of clinical chemistry results by Witte et al found 447 errors per 1 million samples, with 196 in control samples and 251 in patients’; 41 errors per million were judged “likely to alter patient care.”²

CAP’s Q-Probes and Q-Tracking have provided even more specific information, with each study focusing on one particular aspect, such as blood-culture contamination. At a median institution in 2003, 2.7% of all such samples were contaminated, states Meier. “This rate fell to 1.7% if just the best 10% of performers were analyzed and rose to 4% in the worst 10% of performers,” he says.

An even larger variation was found in the 2003 study on the reporting of critical values. “About 1.3% of all things defined as critical values were abandoned because the lab could not identify who to tell,” says Meier. When broken down, in the best 10% of performers, 0.14% of these results were abandoned. In the worst 10% of performers, 10% of the critical values were abandoned. “If critical values are truly critical, this can negatively impact patient care,” says Meier.

However, the impact thus far has not been measured well. “There are not clear population-based measures of effects. There have been clear bad effects in individual cases, but no one has figured out a way to measure them,” says Meier.

Design Flaws
Meier suggests that once errors are tracked, the real question will become: How tolerant is society of medical errors? “Would society be intolerant of 12 deaths yearly due to a mismatched identification? Eliminating those 12 deaths will take a tremendous amount of effort. Society must ask itself how much time and effort does it, as a whole, want to spend on decreasing medical errors. We may need to decide by how much we want to reduce the rate,” he suggests.

The major obstacle is people. “We’re all human, so a 100% error-free lab may not be possible,” says Margaret Peck, director of the laboratory accreditation program at JCAHO (Joint Commission on Accreditation of Healthcare Organizations in Oakbrook Terrace, Ill).

It is an even greater challenge in labs operating below full staffing levels. “Some labs just don’t have enough folks,” says Elissa Passiment, EdM, CLS (NCA), executive vice president, The American Society for Clinical Laboratory Science (ASCLS of Bethesda, MD). Understaffed labs may resort to hiring individuals with biology or chemistry degrees, who can perform in the areas in which they are trained, but can’t work beyond these responsibilities because they lack a basic understanding of laboratory science. “Without the appropriate level of staffing or competency, labs will never be error free,” says Passiment.

Modern Accents
Automation can help in some regard. “We’re not likely to recruit the numbers needed, and automation can help to free up staff and allow them to learn about quality control,” says Passiment.

It also can improve accuracy. “The fewer times the reagent has to be made, the better. The less handling of a specimen, the better,” says Passiment.

“Many people believe automation can reduce errors. For example, over 80 percent of respondents to the IQLM-sponsored, Clinical Laboratory Management Association-conducted survey want their future patient identification systems to contain automation, such as handheld devices that read bar-coded identification bands that may be used in the administration of blood,” says the CDC’s Taylor.

Automation reduces the risk of error that is hazardous to both the patient and the technologist. By handling manual tasks, such as decapping and aliquoting, automated systems reduce the possibility of repetitive-motion disorders and biohazardous exposure for staff, while reducing the chance of error.

“There is always a chance with manual labor that a person can mix up samples and improperly label an aliquot. An automated system will safeguard against this,” says Stephen Wasserman, vice president, Olympus Diagnostics Systems Group (Melville, NY). Olympus’ latest equipment takes a snapshot of the primary bar code label and applies it to the aliquot or daughter. “So the system now has a positive identification from the primary tube to the daughter tube, reducing mislabeling error,” says Gernaey.

Robin Stombler, president, Auburn Health Strategies, LLC (Arlington, VA), which is working with the CDC to advance the concept of an institute for quality in laboratory management, agrees that automation plays a role in patient safety but feels it does not solve all problems. “For example, automated systems must be able to ‘speak’ with one another to ensure information is properly transmitted and obtained. Also, information fed to an automated system must be properly entered to assure safety,” says Wasserman.

Color Me Perfect
If automation alone cannot eliminate errors, what else can be done? “We can overcome many errors by encouraging cooperation among all laboratory stakeholders, exhibiting leadership, and developing an openness to appropriate change,” says Stombler.

At the recent IQLM conference, several best practice presentations focused on improving communication between the laboratory professional and the clinician. JCAHO also believes that improved communications can reduce errors.

“Miscommunications are one of the biggest contributors to errors. JCAHO standards attempt to address this. For instance, accredited labs must verify all phone orders and test results by reading back the order or results to assure they were properly communicated and understood,” says Peck.

JCAHO has other guidelines, including those directed toward accurate patient and specimen identification, but Peck feels that what is first needed is a change in the environment that encourages the profession to share information about problems and ways to improve them.

A culture shift may allow larger changes, such as greater involvement by the lab in the pre- and postanalytic phases. “The lab needs to spend more time with caregivers, collecting samples or interpreting results, to make sure that it’s done properly. There are manuals, but they are long and dry, and few people read them cover to cover. Having a laboratorian within reach means questions about collection or result interpretation can be answered immediately. Laboratories need to be better resources,” says Passiment. ASCLS expects to produce a paper this summer on what such a practitioner should look like.

Jennifer McGeary, director of standards and quality, Clinical and Laboratory Standards Institute (CLSI of Wayne, Penn), feels that a quality systems approach to health care that provides guidance for every step will also help reduce errors. The problem, many agree, is that there is a lack of data to help develop these guidelines.

As data is collected, best practices are developed. “Proficiency testing has been used for many years to improve the quality of results. As another example, it is recommended, based on evidence, that using two patient identifiers when collecting laboratory samples is preferred to improve the accuracy of patient identification,” says Taylor.

This data has also been used to develop the JCAHO guidelines. “Our national patient-safety goals are developed with an evidence-based medicine perspective,” says Peck.

High-End Quality
Currently, the medical community is trying to determine what quality measures should be used to evaluate the quality of a lab’s results. “At last count, there were in the neighborhood of 240-plus benchmarks that can be used for quality,” says McGeary.

“What the medical community is looking for is not just errors, but how to develop measures of quality for the lab. You want to pick monitors that provide an opportunity to improve,” says Meier, who was part of an IQLM workgroup focused on this topic. “We proposed 12, including patient identifiers; blood culture contamination; adequacy of patient identification; critical value reporting; turnaround time; order accuracy; clinician satisfaction; and lab results that are important to population health, such as lipid screening and levels of hemoglobin A1c,” he says.

Meier suggests that monitors will become more important if the pay-for-performance movement becomes standard. Pay for performance is a government-initiated policy that reimburses physicians and facilities based on performance rather than treatment. This will likely mean increased standards. Currently, the government provides a minimum level for quality control. Many organizations set standards that are higher.

“Government guidelines are designed to create baseline standards but not best standards. It sets a minimum but not the good or high bar,” says Stephen Raab, MD, chief of pathology, University of Pittsburgh Medical Center (UPMC) Shadyside (Pittsburgh, Penn). “And even though quality is regulated, we still don’t understand how those regulations affect patient care,” he adds.

Even so, quality control has been used for years to manage the analytical phase of testing, notes Taylor. All labs should have a very comprehensive quality-control plan designed to check the reliability of the test systems and operator performance. “It’s not as simple as running quality controls every 2 hours or 8 hours, or once per day. We need to make the quality control match the usefulness of the test and the decision points of the physicians,” says Passiment.

A number of guidelines are out there, including those from Clinical Laboratory Improvement Amendment, JCAHO, and CLSI. “There are some national benchmarks on quality control, and textbooks and courses focus on this topic. The lack of uniformity is a problem, but labs can use the material that addresses the areas of quality control they are interested in,” says Raab.

The Designers
“Many, many stakeholders have a responsibility in developing best practices for the delivery of laboratory services, including physicians, patients, nurses, payors, manufacturers, information technology experts, and, of course, laboratory professionals,” says Stombler.

The government plays a role, too. “I would love to say governmental organizations don’t have much of a role, but if someone isn’t monitoring it, it doesn’t get done. When everyone is being crunched from all sides to cut costs and produce more, corners that can be cut are cut, and the only way they are not is that someone requires it,” notes Passiment.

Labs should take the foremost role, however. “Labs are complex and have gotten more so. A lot of understanding about improving safety is on labs because they have the expertise, but they don’t necessarily agree on how to look at safety,” says Raab.

With more effort, though, this is changing. And as labs improve their performance, they become a more valued part of the health care team. “The lab is not an isolated entity, though it may seem to be. Other factors influence the department — for instance, preanalytics tie in nursing, physicians, and the patients themselves. Quality systems across the entire path of work flow will see better communication across the institutions,” says CLSI’s McGeary. And with that will come less room for error.

Renee DiIulio is a contributing writer for Clinical Lab Products.

References
1. Bonini P, Plebani M, Ceriotti F, Rubboli F. Errors in laboratory medicine. Clin Chem. 2002;48(5):691–698.
2. Witte DL, VanNess SA, Angstadt DS, Pennel BJ. Errors, mistakes, blunders, outliers, or unacceptable results: how many? Clin Chem. 1997;43(8 pt 1):1352–1356.