January 2009
Feature Story

Anne Ford

Richard C. Friedberg, MD, PhD, and Christopher M. Lehman, MD, helped author a recent Q-Probes study on specimen mislabeling in blood banks for more or less one reason: “If the label’s wrong from the get-go, everything else we do about accuracy and reliability and precision goes out the window,” Dr. Friedberg says. “Your car may run perfectly, but if you’re on the wrong highway, you’re going to end up in the wrong spot.”

The study, “Blood Bank Safety Practices,” aimed to determine how frequently specimens submitted for ABO typing at health care institutions bear an incorrect or incomplete label. “We saw this Q-Probe as an opportunity to assist participants in meeting Joint Commission and AABB requirements, and to create a benchmark against which participants and other transfusing institutions could assess their practice,” says Dr. Lehman, assistant director of hospital laboratories at the University of Utah Health Sciences Center and associate professor of pathology, University of Utah School of Medicine. In providing their reactions to the study’s findings, he and Dr. Friedberg, chair of the Department of Pathology at Baystate Health and medical director of Baystate Reference Labs, Springfield, Mass., also make recommendations for health care institutions to consider when reviewing their own blood bank safety procedures.

Going into the study, Dr. Lehman says, he had hoped to discover that the International Society for Blood Transfusion’s Biomedical Excellence for Safer Transfusion 2003 blood grouping sample collection study—as well as the Joint Commission’s general emphasis on patient identification—had led to a lower rate of mislabeled samples and a decreased rate of wrong blood in tube, or WBIT. Instead, he and Dr. Friedberg write, “the rates of mislabeled samples and WBIT were found to be alarmingly high.” Among 112,112 specimens from 122 participating institutions, the mislabeled specimen rate was 1.12 percent, as compared with another recent Q-Probes finding of a .09 percent mislabel rate for hematology, chemistry, and coagulation specimens. Meanwhile, the estimated aggregate rate of WBIT (uncorrected for silent WBIT cases) was one in 2,500 samples, a rate Dr. Lehman calls “consistent with older studies.”

How can an institution lower its rate of mislabeled specimens and WBIT? First, the authors say, it’s crucial to ID patients with not one, but two unique identifiers before collecting specimens. “For example, my institution uses patient name and a unique medical record number,” Dr. Lehman says. “For outpatients, we require name and date of birth if a crossmatch is not ordered.” Of course, that’s not unusual; 100 percent of participants in the Q-Probes study use first and last patient names, and nearly as many (97.5 percent) use a medical record number or other unique number. He notes, too, that in the International Society for Blood Transfusion’s 2003 study, “participants from countries that used unique identification numbers for all citizens had rates of WBIT too low to accurately estimate.”

As this Q-Probes study revealed, however, a few institutions persist in using much less reliable identifiers. Slightly more than 17 percent of respondents use patient ward or other patient location as a means of identification. “There have been fatalities when the ID has been a bed number or the room number,” Dr. Friedberg warns, speaking not of the study but in general. And, lest anyone be tempted to use patients’ driver’s license numbers rather than generate a unique medical record number, he points out that “where I sit right now in the western part of Massachusetts, I’m probably three miles from Connecticut and 40 miles from Vermont and 60 miles from New York. So the next five people I grab, I’m likely to get an out-of-state driver’s license and therefore more likely to get a duplicate driver’s license number.”

Obviously, the more commonly used identifiers aren’t fail-safe, either. “Is birth date a unique identifier? Well, it narrows it down, but it’s certainly not unique. Is name? Depending on where you live, there could be a thousand Maria Rodriguezes in the phone book, or a thousand John Millers,” Dr. Friedberg says. Even with medical record numbers, “mixups on handwritten things can certainly occur: ‘Is this a 3? Is it an 8?’” he adds. “Absent a bar code on the [patient’s] forehead, the rest are all compromises. You’re stuck with name, birth date, medical record number.”

Then there’s the question of whether and under what circumstances an institution will allow deviations from ABO typing specimen acceptance criteria and permit specimen re-labeling. Most study participants (94.2 percent) reject specimens if the first name, last name, or unique number on the label is incorrect. However, respondents are more divided on whether to reject specimens on the basis of incorrect date of birth (as 49 percent do), phlebotomist ID (56.8 percent), or phlebotomy date (49.2 percent).

Rather than call for across-the-board acceptance/rejection criteria, the study urges readers to perform a risk-benefit analysis of allowing each type of deviation. What it comes down to, Dr. Friedberg says, is: “How precious is the specimen?”

“You want to be compulsive, but at the same time put your compassion hat on,” he says. “If it’s a neonate, and the baby weighs all of a pound and a half, and there are maybe 75 mLs of blood in the whole baby, do you want to go get a few more? There has to be some room for clinical judgment in these things. Sometimes you’re going to have to make a call based on your best judgment. Just be sure that you’re willing to stand up in front of your peers, let alone a court, and explain why you made the call.”

As for label correction, Dr. Lehman considers it a risky practice, particularly when patient identifiers are the elements being corrected. “Since the potential consequences of WBIT are so severe, I think it is safer to have a uniform process of discarding samples that do not meet institutional labeling requirements,” he says.

Should an institution go so far as to require a new sample for ABO typing when a patient’s name is changed or updated during admission? Survey participants are almost evenly split on this, with 50.8 percent responding “yes” and 49.2 percent responding “no.” Dr. Lehman argues for the former.

“From a practical point of view, we believe that this is good practice, since at my institution, most of the patients for whom clinicians request a name change have crossmatched blood set up in the blood bank,” he says. “In order for the crossmatch tag on the units to match the patient’s new wristband, a crossmatch with a sample properly labeled with the changed or updated name must be performed so that the unit and wristband can be matched at the bedside prior to transfusion.”

Institutions that do allow name changes without requiring new specimens should carefully evaluate their wristband placement processes, he adds. “Placement of wristbands is a critical control point in the patient identification process. The Patient Identification Accuracy Q-Tracks monitor has demonstrated a median wristband error rate—that is, a missing, incomplete, or inaccurate wristband of 0.5 percent.”

However, the Q-Probes study also yielded the counterintuitive finding that, in the authors’ words, “a lower rate of ABO mislabeled specimens tended to occur in institutions that do not require submission of a new specimen for ABO typing when a patient’s name is changed or updated during an admission.” Did the study’s authors find this surprising? “Yes, because our experience has been that this can lead to labeling problems,” Dr. Lehman says. “However, I believe the response to this question is probably acting as a surrogate for other unmeasured factors that contribute to an apparent decreased mislabel rate. For example, institutions that do not require a new sample may have strict policies for production and placement of wristbands on patients as a part of a vigilant patient identification process, or they may just be more efficient at identifying sample mislabels.”

Requiring a new sample for ABO typing in the case of a changed or updated patient name is not the only issue on which study participants are divided. There’s also the question of whet­her to require two ABOs on patients with no historical ABO type before issuing non-group O RBCs in nonemergency situations. The almost 40 percent of respondents that do not do so may wish to consider it, Dr. Lehman says, given a UCLA study (Am J Clin Pathol. 2006;126:422–426) suggesting that requiring ABO typing on two independently drawn samples can be an effective method of catching WBIT collection errors.

The key phrase, however, is “two independently drawn samples.” Of the respondents that do require two ABOs on patients with no historical type, 68 percent of them don’t require that the typing be done on two different specimens. And that’s a bad idea, Dr. Friedberg says.

“If I’m taking the same tube and typing it twice, all I’m checking is the technical accuracy of the typing. Typically that’s not where the problem is. If I draw the specimen on the person in the bed next to you instead of you, it doesn’t matter how many times I type it. If you’re getting two types in order to make sure it’s the right specimen from the right patient, you need to get two types on two specimens.”

In another study finding, nearly all (93.5 percent) of respondents allow nonlaboratory personnel to collect and label blood bank specimens, albeit with specific specimen-labeling training. Why? “To control expenses,” Dr. Lehman says. He’s not a fan of the trend. “The CAP Q-Tracks that monitors specimen acceptability has consistently demonstrated that samples collected by nonlaboratory personnel are rejected—for reasons including mislabeling at a rate three to four times that of trained phlebotomists. Based on these data, I think it is reasonable to expect that sample collection by nonlaboratory personnel increases the risk of WBIT collections.”

After all, as Dr. Friedberg puts it, “People who do lots of things often do lots of things well, but people who do only one or two things often do them extremely well.” Still, he cautions, using only laboratory personnel to collect specimens for ABO typing is “a tough business case to make,” since “you have to be big enough to keep them busy the whole time.” And, he points out, it’s better to have well-trained nonlaboratorians doing the collecting rather than insufficiently trained laboratorians: “Poorly trained people under your control are certainly no better.”

Nearly 16 percent of respondents reported that in the last 12 months, their laboratories had identified at least one instance of a patient being identified with an incorrect medical record number as a result of the patient intentionally using another person’s identifying information, presumably to gain access to health insurance. “I think we can expect the rate to increase during the current recession,” Dr. Lehman warns. One way to prevent this practice from leading to ABO-incompatible transfusions, Dr. Friedberg says, is to insist on two separate specimens from each admission rather than trusting that the historical blood type is accurate. It may also help to require photo identification when registering a patient; just over half (56 percent) of responding institutions already do so.

Regarding the practice of separate blood-bank-specific armbands or patient identifiers, whether for inpatients or outpatients, study participants are again nearly evenly split. Drs. Lehman and Friedberg have no specific recommendations, agreeing that institutions should evaluate their own processes and decide what’s appropriate in their own cases. On one hand, as Dr. Friedberg says, “If there’s a unique blood-bank-specific armband, it probably raises the level of the concern” about avoiding armband errors in general. On the other hand, a specific blood bank armband isn’t used at the University of Utah Health Sciences Center because, Dr. Lehman explains, “We believe that patient identification is a significant concern for all hospital services, and that proper patient identification needs to be stressed globally rather than just in the transfusion medicine service.” So, your call.

One practice that should probably not be your call, however: allowing staff to produce and save labels for specimen labeling at a future collection. Seventy percent of study respondents don’t have a policy prohibiting this. “That’s actually scary,” Dr. Friedberg says bluntly. “That’s probably one of the most dangerous practices out there. I’ve walked into hospital rooms in a number of different states where on the window or on the door or on the bed there’s a strip of labels. When that patient changes beds, that strip is still there.”

“This practice tends to be more prevalent in intensive care units, where laboratory tests are ordered frequently,” Dr. Lehman notes. Even if an institution has a policy against generating labels ahead of time, it’s important to make sure it’s strictly enforced. At Utah, he says, “we conduct periodic audits of patient care areas to verify compliance with the preprinted label prohibition.”

In fact, he adds, “We find that periodic audits of blood sample collection processes [in general] are essential for keeping everyone’s eye on the ball. Laboratories should also monitor mislabel rates and conduct root cause analyses of WBIT cases to identify opportunities to improve institutional processes.” Because, as he says, “a no-tolerance policy for WBIT should be the norm.”