Fixing the flaws in ID,
  labeling tactics

June 2001
Cover Story

Karen Titus

There is a time and a place for stomping on the bright and tender heart of creativity. It would be at the patient’s bedside.

While innovation is to be applauded in many arenas—inventing gadgets, problem solving, on stage—it can unleash chaos when it asserts itself in the spheres of patient identification and specimen management. Just one creatively misapplied wristband or label can mean disaster. Indeed, health care has received some of its biggest black eyes in recent years thanks to egregious breakdowns in patient ID and specimen oversight.

And if the anecdotes are bad, the statistics are nothing to cheer about either. For that matter, they may never be, given the zero-tolerance mindset that permeates the issue. George Hardy, MT(ASCP), MPA, has plenty of company when he says, "One error is one too many."

So what’s a laboratory to do?

The better starting point might be, Do labs need to do anything?

When Hardy and his colleagues at South Georgia Medical Center, Valdosta, began intensive tracking of preanalytical errors in their laboratory two years ago, the numbers, at first glance, looked good. "The error rate was very low," says Hardy, administrative director of the laboratory. In 1999 the institution had 250,000 phlebotomy-related opportunities to create a preanalytical error; 89 errors did occur, and 37 of those were considered to be critical, i.e. either a patient was misidentified or a specimen was mislabeled.

It might be tempting to look at such low numbers with relief rather than as a call to arms. "If you’re out there doing a quarter of a million venipunctures a year, and you only have 37 of these errors, they don’t come up on your radar screen very often. So you manage the 37 a year," Hardy says. And why not? "We don’t like to make errors, but I don’t think we’ve [the medical center] ever done any patient harm because we misidentified the patient or mislabeled a tube," he says.

Even if an extensive patient ID/specimen management system could get that critical error rate down to zero, the cost of such a system would stop many labs in their tracks. For the most part, Hardy says, "We have other traps in the system to catch bad lab results. The computers will delta check." Little wonder labs might question the value of spending hundreds of thousands of dollars or more to manage a few dozen errors.

Institutions may also question whether their own error rates are high enough to warrant additional action, though Hardy and others suggest such laboratories may be a bit deluded when it comes to gauging their own numbers.

"It just may be they’re naive to the number of errors they actually have," Hardy says. "I think everybody has these errors. They have more than they’re even willing to acknowledge. If they did an honest study, they’d be shocked at how many patient identification and specimen labeling and drug administration errors they truly have."

Lawrence J. Bologna, MBA, MS, MT(ASCP), director of laboratory services at The Valley Hospital, Ridgewood, NJ, agrees. His institution, like Hardy’s, conducted a painstaking analysis of its error rate, a process that drove home at least one important point: If you don’t seek, you will not find.

"If you go to a number of institutions and ask, ’How many ID errors do you have?,’ if they’re not tracking it closely, you’ll find their numbers are very low," Bologna says. "Does that mean they’re successful in not having errors?" Maybe not. "You don’t really know how many errors are occurring, because they’re not following it," Bologna reasons.

If a recent CAP Q-Tracks study on wristband ID errors is any indication, the numbers may be uncomfortably high. Two years of data are now available from the ongoing study. In 2000, the overall number of patient ID errors in 142 institutions was 22,267, a number the report emphasized with a rare exclamation point. "A very large percentage of laboratories started this study with over 10 percent of their patients not having wristbands," says the study’s author, Peter J. Howanitz, MD. "There was even one around 30 percent. That’s a very, very significant error rate."

That’s the bad news. The good news lies on the flip side: With such lofty numbers, the room for improvement is equally significant.

"We’re starting at a place where improvement needs to occur. And it can occur," says Dr. Howanitz. "Based on our data, it looks like the error rate can be driven under one percent in a large number of institutions. Some labs are now around 0.2 percent." Total wristband error rates were lower in institutions that use wristband errors as a permanent QA monitor, he notes, lending credibility to the idea that the problem needs to be tracked before it can be solved.

Still not convinced these errors are worth hunting down? It might not matter. Eventually, labs could have little say in whether they decide to tackle the problem more aggressively. Says Bologna: "I think we’re all going to be mandated at some point to reduce our errors. This issue is not going to go away. It’s going to become part of how we all do business in health care over the next five to 10 years."

There is a high road and a low road to patient identification and specimen management. To achieve either, labs must take both.

Think of the high road as the path of technology: bar-coded wristbands, radio frequency identification (RFID) tags, portable bedside label printers, wireless Palm-like devices, and large-scale automated systems.

None of these are new. Think back to the days of the first George Bush-occupied White House—remember when the erstwhile President, during a rare visit to a grocery store, was wowed by a scanner that read a bar code torn into seven pieces? RFID technology has been lurking around roughly since the mid-’90s, about the time people began tapping regularly on their Palm Pilots. Even comprehensive patient ID/specimen management systems have had a market presence for several years.

It seems reasonable to ask why these approaches haven’t seeped into health care yet. Cost is the obvious answer—and one with plenty of merit—but more subtle explanations are at play as well.

Sam Riccitelli, vice president and general manager of BD.id, a product and service area of Becton, Dickinson and Co. focused on reducing medical errors, admits he and his colleagues spend plenty of time pondering that very question. The answer, at least in part, is "a bit of a chicken-and-egg mentality," as Riccitelli puts it.

"For years in hospitals, nothing was bar coded because there was nothing that could read them to any useful purpose. So what do we do first—develop hardware systems that read bar codes, or do we bar code things and then wait for the hardware to come along?" he asks.

Hospitals have also "been burned by value propositions before," Riccitelli adds. "So there’s been some real skepticism about making such a big investment in systems that hadn’t proved their worth."

Steve McDermott, director of marketing for Bio-Logics Products, a Salt Lake City-based maker of patient identification systems, suggests the slow acceptance of patient ID/specimen management technology is relative. "One of the most unfair comparisons that has been made is saying, Well, we’ve been using bar codes in grocery stores for decades, so why not hospitals?" he says. "What they fail to acknowledge is the hospital environment is a lot tougher than a grocery store, and a lot tougher than most people realize." What may also be overlooked is though bar-code scanning technology has been around since 1974, it took another 15 years for the grocery industry to begin fully exploiting its vast data-tracking capabilities.

In health care, reading bar codes is a two-handed operation that requires good line of sight, often an easier-said-than-done proposition on a typical hospital floor. "Most patients’ arms are hooked up to an IV, or are under a blanket," says Alex Gelbman, formerly of eID Solutions and now president and CEO of another ID technology firm, Visible Tech-knowledgy, Hoboken, NJ. "So the nurse has to disturb the patient, move the arm, orient the wristband toward the bar-code reader. And most nurses do not see the benefit compared to that inconvenience to them and to the patient."

Bologna takes a less dim view of bar codes, noting that the advent of vertical bar codes on wristbands has eliminated the readability problem that plagued earlier incarnations. The bigger issue, he says, is more nebulous. "The staff that is going to use the system has to be comfortable with it, and it has to intermingle with the way they’ve been doing things previously," he says. "Let’s face it, people are very uncomfortable with change."

Big change—as in cha-ching!—makes people nervous, too. "Bar codes are easier to read, and the equipment is getting better and better and cheaper and cheaper—but it will only become truly desirable when health care can afford it," says Ann S. Merkley, MT(ASCP), CPHQ, quality management coordinator at Intermountain Health Care, Salt Lake City.

And if you think costs are of concern only to hospitals, think again. "The hardware cost to us is a big challenge," Bio-Logics’ McDermott admits. "In fact, we’re building our own hardware, just to keep costs down."

But cost isn’t always the ogre everyone makes it out to be, either. At Intermountain Health Care, which soon will pilot test an upgraded Bio-Logics bar-code patient armband system at one of its 22 hospitals in Utah and Idaho, "I don’t get any sense that my corporation is flinching at the cost of the Palm hardware. At this point, we’re just waiting to see if it works," says Sarah J. Ilstrup, MD, clinical pathologist and director of the transfusion medical service at Intermountain’s LDS Hospital, Salt Lake City.

Bologna takes that pragmatic approach one step further. His hospital served as a beta site for BD’s Dx system for specimen management and is assessing the company’s Rx system for medication management in a pilot study. He compares error-reduction systems to safety needles, whose higher costs are more than offset by increased safety to patients and health care workers.

"It’s great to be able to tell the board of trustees that you’re going to be able to save x number of dollars, but even if you can’t, if you can produce quality care and reduce errors and have patients feel that you’re doing a better job, then it’s worth every penny you spend," he says. "So even if you just break even, it’s a great deal."

Using those criteria, the Dx system is turning out to be a bargain at The Valley Hospital. It is in use on 10 hospital units and will roll out in an additional 20 or so units after a system upgrade—aimed at creating a more intuitive and multi-application platform—takes place. The new version, for which beta testing was wrapped up in March, features a user interface based on the Palm operating system.

The Dx system uses bar-coded wristbands and portable label printers at the patient’s bedside as well as a portable data terminal. Caregivers log onto the system by scanning their user ID with the data terminal; they then re-dock the terminal in a cradle, which triggers the BD server to download the patient collection orders from the LIS. After selecting a patient for collection, the caregiver scans the patient’s wristband for a positive identification. The terminal then displays the collection order, which includes patient demographic information, tests to be drawn, tube types, volumes, and order of draw as well as any special instructions entered into the LIS. After the specimen is collected, the caregiver prints a bar-code label—with the LIS accession number—and applies it to the appropriate tube.

Bologna reports that the average error rate on the floors using the system dropped from 2.46 percent to 1.29 percent; in addition, collection time fell 13 percent and receipt time 55 percent.

Just as beneficial, he says, has been the information generated by the system. Prior to implementation, a team of hospital staff and BD.id industrial engineers mapped out every step related to specimen management, from the clinician’s ordering a test to the analytical phase. That revealed where errors could occur, and, later, showed Bologna and his colleagues where they did occur.

"One of the things that happens with this is that your error numbers skyrocket once you focus on it," he says ruefully. "And certainly it’s not a banner day when the hospital board gets to see that happening." But that’s the only way a hospital will begin to tackle the real problem, he continues. "Information is your best weapon."

The system also lets the laboratory track who draws each specimen and time of draw. "The exact collection time," he says enthusiastically. "Not a defaulted time, not a probable time, not a possible time."

Such information is golden, says South Georgia’s Hardy, another satisfied user of the Dx system. "We can track where the problems are occurring, and we can make sure we’re doing therapeutic drugs correctly." His institution went live with the BD.id technology in June 2000; in the areas where it’s being used, he reports, "we have eliminated patient identification and specimen-labeling errors. To zero. It’s very impressive."

The beauty of such a highly automated system is manifold, Hardy says. By printing just-in-time labels at the bedside, caregivers no longer run the risk of mislabeling specimens, a common occurrence with preprinted labels and a paper collection list. The system can lock out users who are not validated and track those who are overriding built-in phlebotomy policies. And, when the upgrade is completed this summer, collection orders and detailed work instructions will be available on each unit.

That latter benefit will be useful when the South Georgia laboratory decentralizes its phlebotomy process; in fact, Hardy says, it makes such decentralization possible. "I don’t think we could do it any other way. The people doing the collection will not have to manage the phlebotomy process, other than to recognize the cradle is blinking green, which means a specimen needs to be collected. [For stat collections, the device blinks red.] The nurse only has to pick it up, go to the bedside, do the collection, bring the device back and cradle it, and tube the blood to the lab. They don’t have to manage a collection list or anything."

The most successful technologies in both patient ID and specimen management, most agree, will be those that are highly people-proof, so
to speak.

To sidestep the problem of armbands being cut off, for example, Bio-Logics has developed a small plastic tag—embossed with patient identification and information—that can be clipped to an armband and slides in and out of a pouch on the band. "If the armband is cut off, you can clip the tag and re-attach it to another armband, without having to go back out to the nurses’ station and print out a whole new armband," McDermott explains. Data from the tags can also be duplicated onto specimen labels, via bedside printers.

Visible Tech-knowledgy is banking on the promise of RFID technology. RFID tags, manufactured by semiconductor companies, contain electronically recorded information that can be read by a transceiver, which can either be placed in a handheld device or physically constructed into a less portable object, such as a tabletop or even a doorway. RFID does not require a line of sight to be used. Even if a tube is in a receiving bag, for example, a transceiver can read that tagged tube. Or tubes, for that matter—RFID can read entire groups of tagged objects, unlike traditional bar-code scanners, Gelbman explains. That makes it ideal for accessioning specimens-an error-prone, no-value-added process when done manually. Stored RFID data can also be updated easily with handheld devices, which could earn RFID tags a berth on patient wristbands.

The disadvantage of RFID is its invisibility. "It’s like a CD," Gelbman explains. "You know the information is there, but you can’t see it with your eye." Visible Tech-knowledgy hopes to solve this problem by putting a readable face on the RFID tag.

Cerner Corp. is marshalling its energies to develop a single handheld device that could march across nearly every patient safety front line: checking armbands for medication administration, monitoring patient vital signs, performing real-time tracking information during blood transfusions, and handling collection lists and positive patient ID. It’s not enough to address phlebotomy concerns alone, says Julie Hull, MT(ASCP), a Cerner product manager. "How do you ensure the physician has the most up-to-date information, and then how do we create real-time awareness of specimens that have been collected, orders that have been placed?" The company is working with a Brighton, Mass.-based technology partner, PatientKeeper (formerly Virtmed) to bring Cerner’s HNA Millennium architecture to a Palm-like device, and aiming for alpha installation by July 1.

An automated accessioning system developed by Specialty Laboratories, called TARO (an acronym for total accessioning reorganization), permits the Santa Monica, Calif.-based reference lab to easily direct its flow of specimens to 250-plus workstations and across a 3,500-test menu. As each client shipment arrives, it’s opened and sent through a data entry group, where each specimen is given a unique Specialty bar-code label. Every subsequently created daughter tube also receives a unique bar code. After the specimens have been routed through the appropriate workstations for analysis, they’re scanned again for archival purposes.

Prior to TARO’s installation late last year, the incumbent manual scanning and tracking system created its share of tracking errors and misdirected specimens, "including incoming work that was bypassing the lab and ending up in the archive refrigerators and freezers," says Tom England, PhD, Specialty’s vice president of clinical lab operations. "By going through our TARO system, we immediately identified those manual mis-sorts and redirected the specimens to the appropriate workstations."

The system has also reduced Specialty’s turnaround times, reports Bob Powell, assistant vice president of logistics. For the lab’s top 20 tests, improvement ranged from 25 to 60 percent. And it has allowed Specialty to rigidly structure testing schedules.

Explains Dr. England: "The culture in the lab has changed significantly in that the technologists and technicians now focus on testing the work in hand, i.e. the work that’s delivered to them via the TARO system. In the past, like with all other laboratories, when technologists received their work, they were always trying to reconcile it with their work list and spending one or two hours scurrying about the lab to find that one sample that got mis-sorted.

"Now they’re only responsible for testing what is delivered," he continues. The responsibility for finding samples that didn’t make it to the correct department falls to the specimen management team, which operates the TARO system. The net result—and very likely the reason for the improved turnaround times—has been to expedite the testing of the work in hand, versus delaying a hundred samples to look for one.

"As part of the culture change, we now think of the specimens as being owned by the specimen management team and on loan to the lab for analysis," says Powell, who oversees the team. "Whereas previously... well, let’s just say that technicians love to hold on to specimens. And while we haven’t fully changed that culture, we’re working on it. The technicians now are much more comfortable with knowing that if they return a specimen to us and then need it after the fact, they can find it. Before, if they let it go, they felt they might never find it again. But I think we’re getting over
that hump."

Apparently so. "I have a lot more empty refrigerators and freezers now than I did six months ago," Dr. England reports.

Despite the sophistication of these technologies, no one is suggesting labs have finally drifted into some lotusland of perfect patient ID and specimen management. Far from it. Specialty’s TARO system can’t handle frozen specimens, nor can it accommodate non-NCCLS-standard transport tubes. BD.id is still several years away from solving the complex bedside piece of the transfusion management puzzle, and it’s not set up to handle hospital outreach business without the use of bar-coded wristbands. Interfaces between the spiffiest specimen management systems and a lab’s preanalytical systems are tenuous at best. And not even the best and the brightest of armband manufacturers have figured out a reliable way to affix identification to tiny neonates.

Even if these problems have solutions—and they probably do—no expert will ever fully figure out how to work around workarounds—the bane of every laboratory’s existence. Or, as Valley Hospital’s Bologna says, "If your people don’t buy into the technology, no matter how good it is, then you might as well take all your handhelds and dump them out the window."

Think of this—people management—as the low road, or rather, low-technology road to successful patient identification and specimen management. Unlike technology breakthroughs, this is a relatively low-cost solution. It works—but not without constant effort. And it can cause even the most articulate laboratory personnel to sound like a bunch of consultants in very short order. (The words "process" and "communicate" pop up a lot, as in, "You need to communicate the process to everyone or else the system won’t work.")

At Intermountain, where audits of wristband compliance are performed often, Dr. Ilstrup has seen it all: Wristbands stuffed in nurses’ pockets. Wristbands clinging to the end of the portable label maker and never quite making it onto the patient. "I’ve even seen people ID the end of the bed," she says. "So even if you have the best armband in the world, they’ll still manage to cut it off the patient, tape it to the bedside table, and make labels from that bedside table."

That’s hardly comforting, but Intermountain is not alone. The CAP’s Q-Tracks study reveals missing and wrong wristbands are among the most common problems plaguing institutions, Dr. Howanitz reports. Old-fashioned legwork, such as routine audits on the floors to make sure patients have armbands, or periodic search-and-seize missions to scoop up preprinted bands, can alleviate such problems, the study notes.

One Intermountain solution has been to lay down hard-and-fast rules governing who can affix patient armbands. "Our phlebotomists cannot put a regular ID band on," Dr. Ilstrup says. "They can put an emergency green armband on"—a prenumbered band that’s also used in the emergency room and which is used to initiate stat lab tests in acute situations—"but that’s it." Only nursing and admitting personnel are authorized to place an armband on the patient.

Though nursing grumbled at first, Dr. Ilstrup reports, the laboratory stuck to its guns. "You can’t have phlebotomists sticking bands on patients," she says. "The phlebotomists didn’t take the patient’s history, and they don’t know the patient. We don’t want the phlebotomists—many of whom are inexperienced—in a vulnerable situation, where they don’t have the information to make the right decision."

Likewise, phlebotomists at Intermountain are not allowed to draw a patient who does not have a wristband—another sticking point with nonlab personnel at first, but one that has eventually subsided. "People have gotten pretty savvy about armbands," Dr. Ilstrup says. "Plus they know that getting angry isn’t going to change what the lab is going to do. If the phlebotomist runs into a problem, they simply page the pathologist on call—and it’s rarely an issue anymore." The ability to default to a prenumbered green armband in emergency situations has also helped, she notes. "We only have to do that once or twice a year, but it shows the lab is responding to the acute patient need; at the same time, we’re not putting the patient at undue risk by a misidentification."

Policies don’t solve everything, of course. When SwedishAmerican Health System, Rockford, Ill., decentralized phlebotomy eight years ago, it noted a bump up in its error rate. Hoping to ratchet it down, the total quality management team crafted a policy, and it did get results, recalls Patty Needham, MT(ASCP), HEM, safety officer at SwedishAmerican. Errors in the pediatrics unit bottomed out, for example, once staff stuck by the rule to take labels into the room with them, Needham says.

But the policy was only a start. "We found we had to drill down to unit-specific issues," Needham says. On the nursery floor, she recalls, "we found that without having our babies preadmitted, we didn’t have that critical medical record number at the time of birth, when we’re collecting the cord blood sample. So we started preadmitting babies."

Without technology’s help, SwedishAmerican has cut its error rate by 50 percent, Needham says. The hospital will begin implementing the BD.id Dx system in August. Based on a three-year analysis of error tracking and where those errors fall on the process flowchart the team developed, Needham expects to see about 87 percent of the specimen collection and labeling errors eliminated. "We’d like to get our rate as close to zero as possible," she says.

They may get close, but that’s it. As Intermountain’s Dr. Ilstrup says, "The technology should let you get to zero. People, though, are incredibly creative."

Karen Titus is CAP TODAY contributing editor and co-managing editor.