College of American Pathologists
Printable Version

  Unmasking specimen ID errors, every step of the way


CAP Today




June 2010
Feature Story

Anne Paxton

A recent Q-Probes study focusing on mislabeling of surgical pathology specimens, blocks, and slides reinforces increasingly familiar messages about quality control: Formal checks and continuous individual case processing keep error rates down, and beware the handoffs.

The study, “Mislabeling Rate of Specimens, Blocks, and Slides in Surgical Pathology,” looked at hundreds of thousands of cases at 136 institutions to determine where errors occur in the procurement, accessioning, and processing of surgical pathology specimens. In reviewing 427,255 cases, the study found mislabeled rates of 1.1 per 1,000 cases, 1.0 per 1,000 specimens, 1.7 per 1,000 blocks, and 1.1 per 1,000 slides.

In almost all cases—97 percent—the mislabeling errors were corrected. Twenty-four percent of incidents were detected before the slides were released to the pathologist, and 21 percent were detected by the pathologist before the case was released. Two practice variables found to be associated with lower mislabeled slide rates were use of continuous individual case accessioning and a formal check at accessioning to detect and track errors.

“Surgical pathology is a very labor-intensive process that involves many, many steps,” says study co-author Raouf E. Nakhleh, MD, professor of pathology at the Mayo Clinic in Jacksonville, Fla., and chair of the CAP Quality Practices Committee. “We know from day-to-day work that errors occur. So we wanted to have a really good over­view of where the problems are in the process.”

“The idea was to document error-prone steps, and at the same time we asked labs about their processes, how they were doing things, to try to figure out if some methods were better than others.” Proper specimen labeling, the study authors write in their analysis, is an important patient safety initiative of the CAP and the Joint Commission.

Dr. Nakhleh, who is an editor and author of the CAP manual, Quality Management in Anatomic Pathology: Promoting Patient Safety Through Systems Improvement and Error Reduction, believes others have done similar studies before, within their own labs. “But I don’t think this has ever been done in multiple hospitals,” he says.

Study co-author Leonas G. Bekeris, MD, chair of the pathology department at Phoenixville (Pa.) Hospital, became interested in mislabeling for practical reasons. “I’m a practicing pathologist, and this is an issue we monitor day in and day out in the hospital,” says Dr. Bekeris, who is clinical professor at the University of Pennsylvania School of Medicine and a former member of the Quality Practices Committee. “It’s a major safety issue, and everybody is interested in making sure labels are done properly because of the potential consequences of mislabeling specimens.”

The study was conducted prospectively, with participants reviewing surgical pathology cases for an eight-week period or until 30 error incidents related to mislabeled cases, specimens, blocks, or slides were identified. Participants recorded the point at which the error occurred, what was mislabeled (case, specimen, blocks, or slides), the number of cases, specimens, blocks, and slides affected, the point at which the error was detected, and the outcome of the error.

Not included in the error tracking were surgical pathology cases processed before the study period but subsequently identified as mislabeled during the study period, prepared slides received from other institutions, and blocks or slides cut in other laboratories.

More than half of the 136 participating institutions were nonprofit hospitals, about 16 percent were private independent laboratories, and the remainder consisted of government or university hospitals, group practices, and others. Nearly 90 percent had undergone a CAP inspection within the past two years.

The study’s authors were surprised to find that out of the 427,255 cases reviewed, only 1,811 had some type of mislabeling. “We thought the number would be higher,” Dr. Bekeris says. “But you have to realize, when you look at the number of cases, you’re looking at a spectrum of labs, many of which are extremely well run and have practically no mislabeling.” In fact, laboratories at the 90th percentile had error rates of zero across the board.

Because of the nature of surgical pathology, there are multiple possibilities for error, he says. “You have a situation in which there are multiple handoffs of specimens from one person to another—from the surgeon to the nurse, from the nurse probably to an aide who might bring the case to pathology, then to someone who cuts and labels, then to the pathologists.”

A single specimen can generate 10 or 20 blocks, and then double that number of slides, Dr. Nakhleh notes. “So there’s a progression. A single case could have multiple specimens. Some labs have 50,000 specimens, which can generate 200,000 slides, so you’ve got enormous numbers. Even at an error rate of one in 1,000, the potential for patient harm is real. Our study certainly demonstrated that potential.”

Mislabeling errors occurred at a nearly equal rate among cases, specimens, blocks, and slides. “We tried to document the rate at which errors occur, and where in the process they occur. We found that errors occur pretty equally throughout the process,” Dr. Nakhleh says. “About one-third of errors occur when the specimen is transferred from outside the lab and is accessioned in the lab. The second third [31.9 percent] occur when the case is dissected and the tissue is transferred into blocks. Roughly the last third [30.4 percent] occur as the tissue is transferred from the block to the slide. At every point where the tissue is transferred or a handoff occurs, the potential for a mistake is present and it manifests itself equally.”

The type of mislabeling varies. A mislabeled case is defined as one in which the wrong accession number is applied to the entire case. A mislabeled specimen would be wrong labeling due to mixup of specimens within a case, such as right versus left specimens in a bilateral biopsy from the same patient.

A mislabeled block refers to a histologic block labeled with the wrong patient or case identification (surgical pathology number and at least one other unique identifier), or wrong sequence number or letter, and could occur when the wrong label is applied to the block or when tissue is placed in the wrong block. A mislabeled slide might have the wrong specimen or patient identification, sequence number, or letter.

Certain practices seem to affect the incidence of error. Lower mislabeled specimen rates tended to occur in laboratories that routinely include in the gross description a statement indicating the specimen is labeled with the patient’s name and is identified properly. This seemed to be associated with a median specimen mislabeling rate (.6 per 1,000) that was less than half that of labs that said they do not routinely include such a statement (1.3 per 1,000).

Even more significant was the drop in error rate when continuous individual case accessioning rather than batch processing was used. Laboratories that said they typically processed surgical pathology cases continuously had a median slide mislabeling rate of .5 per 1,000, compared with those that typically did batch accessioning (2.1 errors per 1,000). Similarly, laboratories with a formal (documented) check for labeling errors at accessioning had a median mislabeling rate of .6 per 1,000 slides, while those without such a formal check had a median rate of 2.3 errors per 1,000 slides.

Seventy-five percent of laboratories in the study have a system in place to detect and track accessioning and labeling errors. Fifty-three percent said blocks are visually checked with the slides for accuracy of cut section before the slides are distributed, and 72 percent of labs check blocks against the gross description before slide distribution.

When do the errors come to light? The Q-Probes study showed a clear pattern in which errors are most often picked up in the one or two steps immediately after the step where the error is made. For example, about 60 percent of errors made at accessioning are detected almost immediately or during gross dissection. The majority of block labeling errors are caught while grossing, at embedding, while sectioning, or before the slides are released to the pathologist. A minority of errors do seem to “cascade” through the entire system, the authors write, but they concluded that quality checks for errors should focus on the most immediate prior step. Mislabeling errors are usually corrected without consequence, but 3.2 percent in this study required a corrected report and 1.3 percent affected patient care. In this study, the impact of the affected patient care was not assessed.

The study demonstrated that having multiple checks in the laboratory does help prevent or circumvent errors, Dr. Nakhleh says. “In the study, errors are discovered by pathologists 21 percent of the time, and usually it’s because things don’t look right.” For example, “Let’s say I have a specimen and the protocol that comes with it says skin, but when I look at the slide it doesn’t look like skin; it looks like stomach. So I’m thinking something’s wrong here.

“When two different organs are mixed up, the error is usually caught; the danger is when there is a mixup of the same organ—prostate, for example—from two different patients. That type of mixup is very hard to identify.”

Some errors can happen when the surgeon takes a specimen and hands it to a nurse, Dr. Bekeris says. “It’s supposed to be labeled at that moment with the patient ID and what type of specimen it is, right at the site of surgery, but although relatively rare, errors can happen there. The wrong specimen could be attached to the wrong requisition. Or the person in charge of accessioning may use a manual system, with a logbook in which he or she writes the number and the information about the specimen. It’s not unusual in labs that use a manual login system for someone to miss the numerical sequence; for example, they go from 1999 to 1901.”

Or, when the blocks are prelabeled and the case is being cut, the labeled blocks are there and the tissues can be put into the wrong cassette, Dr. Bekeris explains. “It could happen that you have two cases sitting on the grossing station next to each other and you could grab the wrong cassette and put the tissues in it. Even with automatic labeling of cassettes one could theoretically grab the wrong one.” Even when everything is labeled properly, errors might occur: “If you have two or three consecutive cases that are similar, they may be sitting in a tray next to each other. You may grab the wrong slide and read it thinking it’s patient B, when it’s actually patient C. To try to prevent this, in my laboratory we label consecutive specimens with alternating colors using diluted India ink.”

Increasing the possibility of error is that pathologists and the histology personnel are subject to distraction, Dr. Bekeris notes. “For some reason you’d never think of calling the OR and interrupting a surgeon, but when a pathologist is working, it’s not unusual to have 10 or 15 interruptions a day.”

Batch processing is often the norm as well, he says—processing multiple cases together and increasing the chance of mixing them up. The Q-Probes showed that 57 percent of laboratories did continuous individual case accessioning, but 36 percent did multiple small batches, 4.7 percent did multiple large batches, and 2.3 percent did one batch.

At the block labeling stage, 57 percent of labs say they use an automated block labeler but often to process multiple cases; less than half did labeling for one case at a time. During slide labeling, the majority of labs put handwritten accession numbers on the slides, often doing a batch of cases at a time.

“The ideal situation is when a case is being processed, during dissection the bar code of that case is scanned in at that moment and generates blocks only for that specimen,” Dr. Nakhleh says. “Then when the case is ready to be cut, the bar code on the block is scanned and the slides are automatically generated just for that case. But those types of systems don’t really exist except at one or two institutions—where the in-house IT department has developed software to take control of specimens and cut many steps out of the process.” In this study, only one laboratory used a device to attach slides and blocks before cutting and slide mounting.

Often, logistical issues affect these choices, Dr. Bekeris points out. “Every lab develops QC processes, and whether you do things in batches or you don’t do them in batches depends on the volume of work the lab has.” As a former director of a commercial lab, he found that often so many specimens would arrive at the same time, “there was no way to avoid a big batch.”

Participants were asked whether their laboratories had undergone a systematic improvement or Lean redesign, and 58.5 percent said no; 23.8 percent said they had undergone such a redesign in the past one to two years. “Recently people have been touting Lean production as a possible solution in the lab to mislabeling errors. So one of the things we asked about is whether labs have done Lean,” Dr. Nakhleh says. “But I don’t think the study really demonstrated if Lean was helpful or not.”

The Q-Probes did not make any determination about the effectiveness of bar coding, says study co-author Michael O. Idowu, MD, MPH, assistant profes­sor of pathol­ogy at Virginia Commonwealth University and a member of the Quality Practices Committee. The study found that bar codes are used to scan in cases at accessioning, at the gross room table, and at micro­scopy in just over half of labs, but are not used for block and slide labeling in most labs. “You would expect that if you automated something, you could reach some kind of improvement. And it probably does remove some human error, but somehow we couldn’t demonstrate that in this study. That’s not to say it is inefficient or ineffective—just that the difference is not apparent in this study,” Dr. Idowu says.

The technology for bar coding blocks and slides does exist, but only a small percentage of places actually use bar codes, says Dr. Nakh­leh. Bar coding alone doesn’t solve the problem, in his view. “The vendors have the pieces—the bar code generators, the block labelers, the bar code scanners—but they don’t sell you the system that actually manages the whole process. What we need is the software that ties everything together, that tracks specimens through the system and prevents you from making errors. But there is no commercial product that will do this yet.”

An interesting, perhaps counterintuitive finding of the study was that labs that have a committee dedicated to patient safety and quality assurance have higher error rates than labs that don’t have one. “That may be due to the fact that they have committees because they have a problem,” Dr. Bekeris points out. Or, Dr. Nakhleh suggests, it may be an oddity of the statistics. “Many times with quality assurance people focus on a problem and start looking at it harder, and with more effort, people find more problems.”

Dr. Idowu, who is director of the breast pathology service and of quality assurance in anatomic pathology at Virginia Commonwealth University Medical Center, says he and his colleagues recently changed their policy, by mandating that the personnel/resident performing the gross examination actually pick up the cassette and repeat/ dictate the surgical pathology number, the patient name, and the block designation directly from the cassette. “This way, if the number doesn’t match the specimen, you know at that point and you can do something about it. If this step is omitted, the transcriptionists will place a hold on the dictation and inform personnel charged with ensuring that this procedure/policy is followed. Since we’ve implemented that, I actually think we’ve improved a lot.”

Dr. Bekeris’ preference when he examines grossly is to dictate all the accession numbers and full patient names to avoid the possibility of a patient ID mixup. “I will also dictate the site of origin of the case. And once or twice a year, a secretary will come and say, ‘Here the requisition says left and you said right.’ You realize what you did is look at the wrong specimen and you dictated it out of sequence.” The Q-Probes found that labs that do name the patient do better than ones that do not, he says. “That also helps in making sure you have the right case in front of you.”

A mislabeled slide or specimen should, of course, never reach the pathologist without the error having been caught and corrected. “You hope it’s caught before the pathologist sees it,” Dr. Bekeris says. “But this Q-Probes study showed there are quite a few that do get to the pathologist. I see this from time to time where a case comes to me, where somebody has attached the wrong label to the wrong slide. You have to be aware that this could happen.”

This Q-Probes demonstrates there are no special problems causing errors, Dr. Nakhleh says. Rather, problems exist throughout the system. “You have to put in checks after virtually every step, and that will add to your work, but if you see it through to completion and you can catch problems early, you will reap the benefits; your workload will eventually go down.”

Dr. Bekeris says his first reaction when things go wrong is: Who messed up? “But you have to step back and consider that if somebody, an educated person who has been doing everything right for years, makes a mistake, there may be something wrong in the system that might be a factor. You have to look at all the different possibilities.”

In his view, the most important lesson to be drawn from the study is one that labs are already well aware of but may need to hear again. “They have to have a quality control system in place that allows the laboratory to track these errors, actively monitor them, and take corrective action wherever necessary. If they are constantly having problems in the lab, there may be a system problem that needs to be addressed.”

Anne Paxton is a writer in Seattle.