November 2009
Feature Story

Karen Titus

Human dealings can be compactly, if crudely, summed up in two questions: Who are you? And what do you want?

It even works in health care, with a slight variation on the second question: What do you need?

Labs, and all health care practitioners, for that matter, know how to answer question No. 1. Spurred largely by the Institute of Medicine’s report “To Err is Human,” the past decade has seen no end of efforts to make sure the right patients get the right treatments. With patient safety and error prevention now front and center in medicine, fewer struggle today to figure out the urgent “who” questions: Who is this patient? Whose specimen is this? Who collected it?

Now labs are beginning to confront the second question and its offspring: What do you need? What test do you want us to do?

More specifically, they now have data, from a recent Q-Probes study, to help them look at a problem plaguing labs: test orders that need to be cleared up. “Labs know this is a problem. We all struggle with it,” says Teresa Darcy, MD, MMM, one of the study’s authors.

The Q-Probes undersells this all a bit, with its less-than-swashbuckling title: “Monitoring of Outpatient Orders Requiring Clarification.” Concedes Frederick A. Meier, MD, another of the study’s authors, “It’s one of those compromise titles.” In the lab, these orders go by other, less-PC names: bad orders, unclear orders, orders you can’t understand.

Behind the sleepy title are blunt data. “We’ve come up with some firm numbers about which there’s been a lot of speculation,” says Dr. Meier.

The first number to pay attention to is 1.8. Among outpatient test orders, a median of 1.8 percent require clarification. And until they’re cleared up, they bring the processing of that order to a halt.

Is 1.8 percent high or low?

Dr. Darcy frames it as a Six Sigma matter. A 1.8 percent error rate translates into 18,000 defects per million events. “That is a 3.6 Sigma process that is not good enough,” says Dr. Darcy, medical director of clinical laboratories, University of Wisconsin Hospital and Clinics, and associate professor of pathology and laboratory medicine, University of Wisconsin School of Medicine and Public Health.

On the other hand, 1.8 percent might sound low, given claims from other quarters that hint the numbers could have been higher. As the Q-Probes authors note in their report of the findings, a recent Commonwealth Fund survey reported five percent of American patients said they’d been given incorrect lab results, 11 percent said they’d experienced delays, and 14 percent said they’d experienced either lab or diagnostic errors. Although these numbers don’t lend themselves to an apples-to-apples comparison with the data collected in the Q-Probes study, they could make that 1.8 percent seem low.

Nonetheless, the consequences of 1.8 percent can be substantial. The median annual test volume at the labs in the study was 225,000. That means some 4,000 times a year, the median participating lab has to stop the line to make a call and try to figure out an unclear order. That’s several times a day. “Think about the patient who has to wait, and those in line behind them who are also kept waiting,” says Dr. Meier, a senior staff pathologist at Henry Ford Hospital, Detroit, and director, regional pathology services, Henry Ford Health System. The median clarification time was 10 minutes.

And, says Dr. Darcy, think of all the resources labs pour into correcting these problem orders. “Now we know how big the problem is. We have to stop wasting time and resources.”

That 1.8 percent can be broken down several ways. More than a third of those errors—38 percent—are cases involving orders that are unclear because of a missing test name or test code. That’s putting it dryly. “It’s just amazing to think of the number of times we can’t tell what test the physician intends for this patient,” says Dr. Darcy. Missing diagnostic (ICD) codes account for 28 percent of the order defects requiring clarification. Another 18 percent of the problems were due to “complete absence of a test order at the time that a patient presented for outpatient specimen collection,” says the report. That’s a polite way of saying the patient showed up at the lab, and no one knew why. “That’s huge,” says Dr. Darcy, not one to mince words. “Potentially it’s just a failure in communication between a doctor’s office and the lab, but it’s also possibly orders entered on the wrong patient.”

There are limits to this Q-Probes, naturally. This one looked at small to medium-sized hospitals, because those are the institutions that subscribed. Although the study offers insights into various sections of the lab, it didn’t focus on particular tests that may be, as Dr. Meier puts it, “mischievous.” And since the study took place over a relatively short period (the time it took for each participant to identify 30 outpatient test orders needing clarification, or until three weeks of test orders were reviewed), it could not have picked up seasonal variations in ordering problems, if any exist.

Neither Drs. Darcy nor Meier are ignoring the findings, however, and not simply because they wrote the report. Both say the findings are in line with some of the problems they have at their institutions, and both report some surprises as well. For Dr. Darcy, the study merely confirms what labs have known all along. “Every lab figures, ‘Here are my problem doctors, here are my problem clinics.’ And maybe we think someone else is handling it better. But we’re all in the same boat.”

Apart from the surprising numbers, the Q-Probes study also punctured a couple of myths that may have been woven into labs’ collective unconscious.

Send-out tests were not the chief problem. Although they were disproportionate among those orders needing clarification, a higher delay rate tended to occur with tests performed in-house, rather than those sent to a reference laboratory. No one expected this, anymore than a visitor expects to come across an Eames Lounge chair and ottoman combo at Mount Vernon. “Very surprising,” says Dr. Darcy.

Dr. Meier, who calls this finding “odd,” has a hypothesis: Send-out tests are more memorable. When called so a test order can be clarified, the person who placed the order is more likely to remember a send-out test order, unlike one of many CBCs or basic metabolic panels. Someone who’s ordered only one growth hormone test on the day in question is likely to remember what was ordered, and why.

Electronic order entry, the would-be savior, did not reduce the number of errors. While 74 percent of unclear orders stem from written requests, 19 percent arose from electronic orders—more than double the problems caused by oral orders (eight percent). (Nineteen percent was about the fraction of electronic orders evaluated. “So the error fraction was proportionate,” Dr. Meier says.) Dr. Darcy recalls her and her colleagues’ reactions to this particular finding: “That cannot be.”

The Q-Probes authors had another surprise when they learned who was charged with clearing up the confusing (or missing) orders. Few were medical technologists (14 percent), customer service reps (seven percent), or order entry supervisors (four percent). Instead, a common strategy at many institutions seems borrowed from chess —sacrifice a pawn. The staff members charged with clarifying problem orders tend to be those in the front line, including phlebotomists (a little less than 30 percent). Order entry personnel accounted for slightly more than 40 percent.

This move becomes even more interesting when paired with another piece of information from the Q-Probes report. For six percent of the orders that resisted clarification, the person handling the confusing order gave up trying to ascertain the correct order and just moved forward.

How? They decided to “guess at” the intended order, according to the Q-Probes report.

Dr. Meier can see that happening with a worker caught between two immediate tasks—a phlebotomist, say, who’s trying to clarify one order and move on to the next collection. “I imagine that many of these people are experienced and have learned what the likely guess is,” he says. Maybe they’re like parents of toddlers, who can hear full sentences in their children’s random pattern of syllables and grunts.

An engineered process such as Lean offers an alternative to such guesswork, with the offending order sent to a higher-up to solve. The phlebotomist, in the meantime, would move on to the next draw.

Dr. Meier prefers such an approach. Is it reasonable to expect detective work from somebody who has a room full of people waiting to have their blood drawn, he asks. At Henry Ford, “We try to bump it up to the level of phlebotomy coordinator.” That person is also drawing specimens, but is not necessarily on the front lines full time; rather, he or she deals with surges, rushes, and hard sticks, but otherwise should have more time to deal with problem orders.

One final piece of information: Only 43 percent of the order forms received by the Q-Probes labs contained test codes.

Dr. Meier reports that his lab includes codes on the forms it creates. But, he says with a sigh, other departments aren’t shy about creating their own order forms, with their own abbreviations, understood only by those in that particular department. It’s as if they’re speaking in dialect. “You would think they would ask what we use, or even look to see what the lab abbreviations are,” says Dr. Meier. There’s little sense of the institution as a whole; rather, actions often hinge on, “How will it work in my little world?” he says. It’s unsettling. The demand for diagnostic codes is now firmly rooted in federal statutes and regulations, yet their absence accounts for two-thirds of the problems encountered in the Q-Probes.

Rarely, but no less seriously, an unclear order will simply disappear, rather like the Plantagenet dynasty. Abandoned orders occur “only” one per every thousand orders, but there’s no way to put a good face on this. If a test isn’t done, it may interfere with patient care; at the same time, the physician, if not the patient, will likely form some less-than-flattering opinions of the lab.

If the problems have been identified, that should make them easier to solve. “You would think,” says Dr. Meier, who clearly believes otherwise.

Lab personnel can be forgiven if all this makes them want to hold their collective heads in their collective hands. It’s not as if these are new problems. Then again, years of unlearned behavior can take years to learn. “It comes up in every venue: We haven’t done a great job educating a generation of medical students and residents and attending physicians about lab tests,” Dr. Darcy says. “I wish we did it better.”

She recounts the history at her lab. “We started with accepting anything from anybody,” she says. Her health care system then moved toward a standard request form that could be customized for each specialty.

As the lab saw what tests physicians were ordering incorrectly, it amended the forms to make ordering those tests foolproof. The lab also included test codes on the paper request forms, which improved things further. At one point, the lab placed bar codes on its forms, though they took up too much space and reduced the number of tests that would fit on the form.

As you might guess, none of these efforts fully stemmed the tide of orders needing clarification. “Whenever our physicians were not finding the test on their request form, they handwrote it,” Dr. Darcy says. Sometimes such orders were illegible. Sometimes they weren’t written in understandable English. Sometimes the English was clear, but the words didn’t match a specific test.

Somehow, physicians always managed to find their way around the obstacle, otherwise known as the test order form. As one example, Dr. Darcy notes that her lab does not offer hepatic function panels. When physicians wrote in their requests for one, each physician would have a different panel in mind. “You’d have five physicians, and when you called for clarification, you’d get five different groups of tests.”

Now the lab and its health care system are sailing toward what should be smoother waters: the use of electronic orders. Yet, as the Q-Probes reveals, this might just be another rapids. “That’s one that’s keeping me up at night,” says Dr. Darcy. Electronic orders should make a difference, but they didn’t, at least in the settings evaluated in the study. “It’s so discouraging,” she says.

Adding electronic lab orders has been one of the most difficult tasks. When Dr. Darcy and her laboratory colleagues asked about training for physicians in computerized order entry, they were told physicians know how to order laboratory tests. “But they don’t,” says Dr. Darcy. While the names of the tests are standard and simple, at least from the lab’s view, “It’s not in physician minds.”

There’s something wrong, she says, with how the electronic orders are built. The problem could be big, or it could be small. It could be as big as to require a redesign of lab information systems, she says, or it may be as simple as physicians deciding it takes too long to place the order. Dr. Darcy’s lab has an order called “miscellaneous test,” meant to be used for rare tests—certain genetic tests, say, that might be ordered once or twice a year and therefore don’t require building a full-blown electronic order. But what sounds like a sensible solution is ripe for abuse. Sometimes busy physicians will order a miscellaneous test, accompanied by the comment, “I need the following 15 things,” which may not be correctly named. “Whenever they can free-text something, instead of picking it, it may require clarification,” Dr. Darcy says.

In the Q-Probes, the laboratory section with the most problems was clinical chemistry—61.2 percent. That’s one place laboratories can start rooting around to figure out the failures. “What is it about clinical chemistry tests? Is it hard for physicians to pick the right one, or is it something else?” Dr. Darcy asks. “We need to dig deeper.” Hematology/coagulation tests accounted for 32.4 percent of the orders needing clarification, and microbiology/serology weighed in with 13.8 percent. (These rates reflect the fractions of tests in these subgroups analyzed for the study, Dr. Meier says.)

Labs should also pay attention to that man behind the curtain, so to speak—who is filling out those forms? While test orders start with physicians, they might be handing them off to someone else, especially in the case of electronic orders. That someone else may not know the appropriate code, and they may not be able to easily hunt down someone who does. Even the ordering physician may not know the code.

Labs, for their part, at times act like tired salmon, and fail to head further upstream in search of the earlier problem that leads to an unclear order. “The amount of noise we tolerate is surprising,” says Dr. Meier. “There isn’t much effort to get to the physicians or practices that are sending you these orders.” There should be, says Dr. Darcy. “This is a significant error rate, and it just shouldn’t be acceptable in health care anymore.”

Rousing the ordering physician might solve the problem, or at least strike another blow for the cause. “My boss, Dr. Zarbo [Richard J. Zarbo, MD], has been sending things back, so people know they’ve created these problems and delays,” says Dr. Meier, though this new approach, which began in June, is too new to evaluate.

The Q-Probes contains its own advice for handling unclear orders, including:

• Don’t abandon a test order.
• Figure out the order before drawing a specimen. “There’s pressure in some practices, when you call up, they say, ‘Oh, just draw the patient, and we’ll call you back,’” says Dr. Meier. “It turns out that if you don’t clarify the order before drawing the patient, it actually takes longer to get the order clarified,” he says—often more than 30 minutes.

It’s curious that so many patients felt capable of reporting having any experience with their lab tests, let alone a bad one (as they did in the aforementioned Commonwealth Fund survey), given that the lab is so often considered to be an invisible partner in medicine.

Dr. Meier floats an explanation. Say a patient has an unexpected potassium result, one that’s elevated above the reference range, which most likely is a statistical aberration rather than a clinical threat. Yet anyone practicing medicine defensively, that is to say almost everyone, might feel compelled to urge their patient to have another specimen drawn and the test re-run.

At this point, “They do get a regression to the mean, and they do get one back that’s within the 95 percent range,” says Dr. Meier. “And the physician tells them, ‘Well, that first lab value was bad. You’re all right.’ It creates a sense that things are not to be trusted,” Dr. Meier says.

It also creates a delay. But any delay—including an unclear lab order—will be branded a laboratory problem, even if the problem originates in an order from a clinician that needs clarification.

It’s not a matter of merely bad PR for the lab. Dr. Darcy is firm: Unclear orders are bad for patient throughput and patient satisfaction, and they’re bad economics. “Whether we have health care reform this year or we have it next year,” she says, “there’s an urgency about making sure we’re using every dollar the best we can. Waste in the system has to go away.”

Even more to the point, unclear orders stop labs in their tracks. Says Dr. Darcy: “We can’t do anything in the lab if we don’t have a clear test order. We can’t do anything.”

Clearly, it’s time to do something. The Q-Probes might be the place to start.