College of American Pathologists
Printable Version

  ER, ICU testing: limits, logistics, and laurels


CAP Today




August 2011
Feature Story

Anne Paxton

If there was ever any doubt, expanding test menus and surging test orders from critical care clinicians are proving that the clinical laboratory is the linchpin of the hospital critical care unit as well as a lifesaver for the more than 5 million patients admitted to ICUs each year.

Demand for critical care has never been higher, says Charles Cairns, MD, professor and chair of the Department of Emergency Medicine at the University of North Carolina, Chapel Hill. Throughout the country, “there’s a wide variation in who gets intensive care, and part of that is due to the availability of resources because critical care is expensive to build, maintain, and staff. Even though we’ve gotten better at standard hospital care—so not as many patients run into conditions where they have to go to intensive care—overall utilization of the ICU continues to increase,” says Dr. Cairns.

While the laboratory responds to the resulting increased testing demand, critical care in turn, like a crucible, tests the mettle of the laboratory. With its continual, high-stakes urgency, critical care presents unique challenges to labs’ ability to solve diagnostic mysteries, alert clinicians to unexpected risks, flexibly manage heavy workloads, and get accurate results back fast enough to count.

The exploding use of lactate and blood gas testing; ways to get toxicology results fast enough to help a critical care diagnosis, control over preanalytical errors, and faster turnaround time; and growth in point-of-care testing are a few of the aspects that critical care testing experts cite when asked to describe the changing role of their institutions’ laboratories in their ICUs.

An ever-lengthening test menu for critical care at Johns Hopkins Healthcare, for example, is placing extra pressure on the clinical laboratory, says William Clarke, PhD, associate professor of pathology, director of clinical toxicology, and director of point-of-care testing.

“When critical care testing started, there was just blood gas testing and CO-oximetry. What’s happened over the past few years is the addition of all sorts of tests, including electrolytes, glucose, ionized calcium, and many other analytes.” Until now, most of the Johns Hopkins critical care laboratories have been satellite labs staffed by medical technologists. “It’s now shifting to point of care, where it’s not just going to be done by medical technologists or respiratory therapists but increasingly by nurses.”

The hospital’s three satellite labs currently perform critical care testing supplemented by critical care analyzers in the emergency department and in the neonatal and pediatric ICUs. “They’re spread out, but we have pockets of testing.” However, next April, when all the labs move to a new clinical building, “all of these areas, particularly the NICU and PICU, will be much further spread out, over an area equivalent to two or three city blocks. So we’ll have to go to a mixed model, where we’ll still have a critical care lab, but will also have to have decentralized testing with POC devices or mobile platforms.”

“Do we need to have every test on our very large menu available on every analyzer? That is the question we’re addressing right now.” The mobile platform—for example, a Radiometer America ABL 80—has a “stripped down” menu, “so creatinine is not available, and lactate is not available yet. We’ll have to be able to match up the analyzers for how they’re going to be used in each clinical setting.” Right now his laboratory is having discussions with the medical directors of all the ICUs to work out the best mix.

Nowadays, the typical critical care analyzer, such as the Siemens 1200 RapidLab series, the Instrumentation Laboratory Gem 3000 or 4000, or Nova Biomedical’s Stat Profile Critical Care Xpress, or CCX, or the Radiometer 800 series, are benchtops, not handheld devices, he notes. “As these analyzers become more sophisticated, they weigh anywhere from 40 to 60 pounds depending on the vendor. They’re not large compared to a chemistry analyzer in the central lab, but they’re not easily mobile.” The true point-of-care devices, like the iStat cartridge-based handheld blood analyzers, are much lighter, “but those tend to cost much more per sample.” More intermediate devices like the Radiometer ABL 80 or IL Gem 4000 are relatively compact. They can sit on a benchtop but also be put on a cart or an IV pole, Dr. Clarke says. “So they’d have the potential to move from room to room.”

Next year’s move will also bring a sharply higher number of device operators. “One significant concern for us is the appropriate mixing of the specimens. Particularly for hemoglobin analysis, the specimens need to be mixed in syringes, and that can be a difficult process. We train our MTs to be very thorough, and when we have respiratory therapists doing this, that is again a very small population of users.” It’s relatively easy to make sure they’re trained correctly and to oversee them, he says.

But with the expansion to nursing, he expects there will be less and less compliance with the appropriate technique. “The big analyte I’m concerned about with the expansion of these menus is potassium, because none of the analyzers have any way to detect hemolysis. In the central lab, we have mechanisms in place on the analyzers to automatically detect hemolysis, but these mechanisms do not exist in any of the critical analyzers, whether bench, mobile, or handheld.”

In the satellite labs, now staffed by medical technologists, “we do a manual check of hemolysis. But at the point of care, it’s unlikely people will do this because they’ll be engaged in patient care. There are no automated safety checks. So you can have a falsely elevated potassium that would appear to be at critical level, or you could have a low potassium but hemolysis would make it appear normal, and you’d take no action when you should take an action. This is what makes this decentralized testing less safe.” These preanalytical considerations are not going to be at the forefront for nurses, he says. “They will be engaged in patient care rather than thinking, ‘Did I take all the precautions with the specimen?’”

As the hospital opens up the menu outside of the critical care labs, the respiratory therapists will do the test if a blood gas is needed, “but they won’t be able to run every sodium, potassium, or whole blood creatinine that is ordered,” points out Dr. Clarke, who is expecting to go from 10 to 20 POC users to 200. “That’s a much different picture. The onus is on the lab to do increased education and oversight of the nursing staff. And that’s going to become a significant challenge to us moving forward.”

Even clinicians can lack appreciation of preanalytical problems. “With ED specimens, 10 percent to 15 percent of the specimens come in hemolyzed, and that’s the example we use in discussions. When you explain to clinicians, ‘Well, this is how hemolysis will affect your result, and the test becomes useless for the most part,’ there are some clinicians who would disagree and say, ‘Even if there’s some hemolysis, I’m better off with a ball park potassium.’ And those are discussions we can have. But I think overall, when we discussed with clinicians here what the challenges are of measuring potassium on a whole blood analyzer in the ED, they quickly lost interest in doing that testing.”

The laboratory/ICU setup at Duke University Health System is somewhat unusual, says John Toffaletti, PhD, director of blood gas services and clinical pediatric labs, and professor of pathology. “Every institution evolves in different ways, and while many larger hospitals have a separate area within the main laboratory for testing pediatric samples, very few have a clinical pediatric lab, but we have a very busy one on the fifth floor near the pediatric ICUs.” On the third floor, the blood gas lab is located near the operating room and the other ICUs.

A pneumatic tube system serves all the ICUs. “We do well over 100,000 blood gases a year at Duke, and that would be really hard to do point of care. But transporting samples is always a problem.” One reason: The tube system can affect PO2 levels. “If you think about what the pneumatic tube system does, it accelerates and decelerates specimens fairly quickly, so it’s basically the same as violently shaking up the tube.”

“So if you leave an air bubble in a blood sample, that equilibrates the blood with that air bubble, so you want to be really careful about removing all the air bubbles—tapping the tube and expelling bubbles before putting the cap back on—before sending a sample. Unfortunately, caregivers have other things to worry about and may not notice if an air bubble remains in the syringe.”

Another concern is the trend to collect more venous samples for blood gas testing instead of arterial samples. “It’s for convenience, because venous is much easier to obtain and much less painful. If they want a quick lactate, arterial or venous are pretty much the same; they’re not going to miss the diagnosis by getting venous. However, the PO2 result is pretty much worthless.”

Turnaround time is always a big issue, Dr. Toffaletti says. “But one of the things that’s happened recently is our pneumatic tube system becoming overburdened.” Few hospital pneumatic tube systems work like the ones in a bank, he points out. “Instead, they go from point A to a central holding area. Then they stay for a few seconds while a directional device rotates and blows them back out again to where they’re supposed to go. That way you don’t need 50 different tubes coming out of each place.”

But if the system is very busy, it will not accept the sample until the path is cleared. “Because more specimens are being sent on our tube system, sometimes tubes end up sent to the wrong place. I think that’s been happening more frequently lately, because when we get complaints about turnaround time, it’s often related to the pneumatic tubes.” While the tubes are the responsibility of the hospital’s engineering department, Dr. Toffaletti says, “we want to know about that, because we’re responsible for the turnaround time.”

For the last two years, the laboratory at San Francisco General Hospital has offered a novel program to address a common shortcoming of emergency and critical care: the inability to get toxicology results quickly enough to make meaningful treatment decisions in many cases of apparent drug toxicity or overdose.

While immunoassay testing of ur­ine is widely used for testing ED and ICU patients, it only partly meets the needs of clinicians caring for toxicology patients, says Alan Wu, PhD, chief of clinical chemistry and toxicology at the hospital and professor of laboratory medicine at the University of California. “Tests using immunoassays from the central laboratory or point-of-care devices do not have adequate performance or cover the menu of drugs needed to meet all clinical needs.”

The problem is that immunoassays can be used to detect each of the families of amphetamines, barbiturates, benzodiazepines, opiates, and tricyclic antidepressants—but the assays have significant limitations. The opiate assay, for instance, is sensitive to codeine, morphine, hydromorphone, and hydrocodone, but has little reactivity to oxycodone or oxymorphone, Dr. Wu points out, while the tricyclic antidepressants assay produces false-positive results to structurally similar drugs like cyclobenzaprine, diphenhydramine, and carbamazepine.

Gas chromatography (GC) coupled with mass spectrometry (MS) can be conducted, but is not amenable to stat analysis, so it is useful mainly for confirmatory testing, not for making urgent clinical decisions. For that reason, Dr. Wu decided to develop an expanded clinical toxicology service using advanced technologies that produce rapid test results. Through a collaboration with a regional poison control center and use of research funds, San Francisco General has become a crucial resource for area clinicians and, Dr. Wu hopes, a model for other regions’ clinical toxicology programs.

Intoxications, of course, are only one issue for an ICU. “There are so many other things: stroke, cardiac arrest, motor vehicle accidents, meningitis, sepsis.” For detecting the presence of certain illicit drugs like cocaine and opiates, immunoassays can do a decent job fairly quickly, but for other causes of toxicity, testing is often much more complex.

Under Dr. Wu’s model, the assumption is that the ICU doctor ­doesn’t have the toxicology expertise, and calls one of the 35 poison centers across the U.S. to get expert advice from a trained medical toxicologist on the best way to proceed given the clinical history available. In the San Francisco area, his laboratory then operates on referral from the poison center. “We’re sort of ‘their lab.’”

A 2010 case illustrates the potential impact of his proposed model. “We had a large rock concert where a number of kids got really sick from Ecstasy. They were critically ill and several died.” On referral from the regional poison center, Dr. Wu’s laboratory was sent blood and urine samples while the patients were in several hospitals’ ICUs, along with confiscated drug evidence that could indicate what the toxic agent was. Within about eight hours, “we were able to determine it was not a contaminant causing this but an overdose; the kids were taking a concentrated version of a street drug at three times the usual dosage.”

If, while the results are awaited, patients are given supportive care—that is, making them breathe, taking care of bleeding, fixing their blood gas status, and giving them fluids, “measures that are not specific to any particular problem,” Dr. Wu says—an eight-hour turnaround time on a toxicology test might be quick enough to get them a specific treatment that works. In this case, “we were able to provide the survivors, in real time, information on what we were dealing with.” But the more typical 24- to 48-hour wait for toxicology results is likely to be too long for some patients.

Once his laboratory has a sample, it can get preliminary results in a couple of hours. “Every week we have cases where a patient has some kind of toxic syndrome and an unknown drug of abuse is suspected, and we do testing to rule in or rule out what it is,” Dr. Wu says. For example, a child recently was admitted with significant bleeding disorders and the suspected agent was rat poison. “Within a few hours we determined it was not rat poison and we went on to look for other things; it turned out to be a genetic disease.”

Changing the toxicology testing specimen from urine to serum is also part of his model, he says. “Everybody does urine testing, and urine is okay if you want to cover the last 72 hours, but it doesn’t necessarily tell you what happened in the last seven hours. And in an ICU patient, you want what’s causing their present symptoms, not what they were on three days ago, which may or may not be contributory.”

There is an information gap on the part of most critical care physicians about what a toxicology laboratory can do with a true unknown, he notes. The clinicians are frequently “toxicology-naive.” In 2009, “why did it take six weeks for the Los Angeles medical examiner to find out that Michael Jackson died of Propofol? It’s because nobody told him what to look for, and they had to be absolutely sure of the results.” Unfortunately the message in that case, as in others, was “here’s a sample and it’s your job to tell me what it is,” Dr. Wu says. “But it doesn‘t work that way. It becomes not a laboratory test but a science project, because you have to look under every stone, and figure out every possible reason for why he died, clinical versus toxicological, and that becomes a thesis project.”

Screens of general unknowns can be done, he says, but they take weeks or months, as in the Jackson case. Such screens are appropriate for forensic toxicology but not clinical toxicology, which is tasked with finding answers quickly or not providing them at all. “Toxicologists’ attitude is frequently ‘we provide the right answer and we can go to court.’ And in the forensic environment that’s absolutely true. But in the clinical world you have to make a choice: Am I going to provide some information that might even be wrong but is more likely right, or provide no information for three or four days?”

Targeted, on-site liquid chromatography (LC)-MS analysis for clinical purposes under Dr. Wu’s proposed model could be performed in eight to 16 hours. But because LC-MS technologies are considerably more expensive than GC-MS—on the order of $300,000 versus $50,000—the National Academy of Clinical Biochemistry proposed in 2003 that regional laboratories be set up widely, working through referrals from regional poison centers, to provide rapid clinical toxicology services. “You’re not going to get more than one or two hospitals in a major area that can afford this equipment for clinical toxicology purposes only. So a regional lab is important because you want to distribute costs.”

A high percentage of the 120 million patients who come through U.S. emergency rooms each year are there for episodic, unexpected care, says Dr. Cairns of UNC, “and in that group those with life-threatening illness or injury are the ones we need to identify quickly.” In the ER, “we’re faced with increasing numbers of patients, and these patients are undifferentiated; they don’t have specific disease states, and in order to better characterize them, we rely increasingly on lab tests.”

That’s one reason for the extraordinary rise in the use of lactate and blood gas testing in EDs and critical care settings over the past five to 10 years, he says. “It really stems from the basic concept of emergency medicine that time makes a difference and early intervention saves lives. In some conditions, we now know that time makes a difference in terms of recognition and intervention on the order of minutes to hours, and the one area where this is clearly changing lactate utilization is in sepsis.”

Lactate testing, which has been growing exponentially, has been shown to predict who’s going to have complications of illness or injury, especially in sepsis, so higher lactate levels have been associated with increased risk for complications including death, and therefore predict the need for earlier, more intensive therapy and hospitalization, Dr. Cairns explains. “It’s evolved so far that some hospitals literally do point-of-care lactate testing at the triage desk at the ED.”

Making clinical decisions based on the lactate level within six hours of an ER visit has been shown to lower death rates, he notes—a significant accomplishment through use of a test that is rapid, easily obtainable from blood, and relatively inexpensive.

The first article showing that such early intervention makes a difference was published in the New England Journal of Medicine in 2001 (Early goal-directed therapy in the treatment of severe sepsis and septic shock, 345:1368–1377). “It was this idea of early goal-directed therapy for sepsis—making sure that the patient has their physiology supported, in other words, their cardiovascular system, the oxygenation of blood, enough blood, and how it gets delivered to tissues—that was the first really big step. And then subsequent studies not only replicated those findings but showed that a surrogate for a lot of complicated invasive markers is lactate.”

In addition, Dr. Cairns says, studies have shown lactate to be a valuable surrogate for therapies, “so we’re not just taking a lactate to predict who’s really going to be at risk but also to show the benefit of therapy.”

Most cases of fatal sepsis or complicated sepsis present in the ER, he says. “It’s clear that patients have local infections that somehow go systemic. Some of them then develop septic shock, some develop cardiovascular compromise, or their organs become dysfunctional, while some, on the other hand, will have infections and inflammation and not have any particular complication.” So understanding the impact of infection as part of that host response is an increasingly important area for emergency medicine and critical care clinicians to identify.

Some medical centers, in fact, do lactate point-of-care testing at the first patient encounter, literally at the triage desk. “At our hospital that’s not how we do it, but we certainly set up lactates in anyone in whom we suspect an infection or who has significant inflammation.” Lactate, he points out, has been shown to be more sensitive in detecting severe sepsis than clinical evidence of shock with low blood pressure, for example. “You can see the real value of lactate. It‘s a tool that’s literally instrumental in saving lives by making the recognition of the severity of septic state earlier on and allowing intervention and the ability to reduce patients’ mortality.”

Continued lactate testing varies depending on the patient’s condition and the protocols used at the institution. “At some centers the test is used as often as every two to four hours, at others as infrequently as once a day. From our standpoint in the ED, it’s that initial lactate that’s really important, but in borderline cases a repeat test in two hours can help further guide the choice of therapy.”

Integrated point-of-care blood gas machines are also used at UNC in the ED on the ground floor and in some of the different ICUs throughout the hospital to assess patients with suspected infection and evidence of inflammation, Dr. Cairns says. “Our lab has worked with the respiratory therapy department so that dedicated, trained respiratory therapists run that machine. We have the capability of doing essentially fingerstick testing in the triage area with these machines, which have automated cleaning functions, automated quality functions, and of course automated reporting to the EMR.“

Primarily it is nurses who are doing collection in the ED, especially in time-sensitive cases. “At the average critical care unit, there’s always an ongoing need for personnel training for understanding machine maintenance and to work on the quality of the testing program to make sure it remains valuable to clinicians.”

But more and more critical care testing is being performed at the point of care. “There has been consistent pressure and a pretty broad movement to try to get these time-sensitive tests that have shown to be so valuable in clinical management—especially those that identify life-threatening conditions—closer to the area of care so they can rapidly be incorporated into clinical decisionmaking,” Dr. Cairns says. “That’s been a trend that’s really accelerated in the last few years.”

While it’s not pertinent to lactate and blood gases, a chronic issue in emergency and critical care is the multiple testing platforms with varying kinds of standards. “Certainly for tests we use all the time like troponin and D-dimer, I can tell you every time they change platforms there is confusion. A number that was important before is unimportant now, and vice versa. So patients are misclassified because people don’t realize the standards have changed.”

The Centers for Medicare and Medicaid Services’ Hospital Compare program has helped the trend toward more POC testing in emergency and critical care, starting in academic centers and extending to community centers, Dr. Cairns says. “For the last few years, there is a movement on public reporting for time-sensitive diseases by the federal government, including CMS. Hospitals literally have to report what their compliance is with getting timely antibiotics to patients with pneumonia or acting on laboratory results for patients with heart attacks.” Every hospital in the country has to be on this system now, which offers reporting of patients’ experience, process-of-care measures and outcomes of care, and allows direct comparison on those measures among three hospitals at a time.

“The government has clearly become more interested in time-sensitive emergency and critical care,” Dr. Cairns points out. “The National Institutes of Health has started sponsoring studies to look at what the best approaches are for these patients. Frankly, a lot of these efforts were organized by medical care societies and physicians just to put evidence together into helpful pathways, but they’ve shown repeatedly that death rates for sepsis have gone down, and that’s gotten the attention of the federal government.”

Patients who are undifferentiated don’t often present with a classic syndrome, he notes. “It’s one of the key challenges we face. It can be very difficult to identify their disease state to treat them effectively, and we’re going to need more and more tools—lactate is just one—to better characterize patients earlier and more actively.”

So-called cognitive traps that prevent clinicians from considering sepsis as a diagnosis are very real, Dr. Cairns adds. “We’ve been looking at the presentation of sepsis patients in a study last year. And we were able to identify a group who didn’t look very sick based on their vital signs, especially heart rates and blood pressure. But within 24 to 48 hours they markedly deteriorated while in the hospital, and 17 percent of them died. This illustrates if you’re just going to look at blood pressure from someone in severe septic shock, you’re going to miss a serious number of patients.”

A lot of progress has been made in handling sepsis, Dr. Cairns believes. “It’s still deadly. It’s increasing in incidence, and unfortunately the interventional clinical trials have not been as successful as we would have hoped.” But laboratory testing is one of the ways clinicians have been able to improve their treatment of sepsis.

What are the major issues for laboratory testing in critical care? “I think the really big areas for need are, No. 1, to get earlier accurate identification of the infectious agent. No. 2, to really understand host responses—which pathways of inflammation are at work, how much organ dysfunction is occurring, and who are the patients really at risk. And No. 3, to be able to then better tailor therapy, not just to individual patients but to a population of patients. Clinical phenotyping of patients will help match them with therapies and resources, and decide which patients should be in the ICU versus the hospital floor versus home, and which patients would benefit from antiviral therapy or antiinflammatory therapy in addition to antibiotics.”

In Dr. Cairns’ view, “the clinical laboratory is the future of emergency and critical care.”

“This earlier characterization of undifferentiated patients is going to be important to identify those with life-threatening illnesses in whom we can intervene, as well as those who don’t need so many resources and can be safely sent home to be cared for. The movement in our field is earlier, more accurate identification, and the laboratory is our partner in being able to do that.”

Anne Paxton is a writer in Seattle.