January 2009
Feature Story

Anne Paxton

As increasingly complex laboratory tests start being performed rapidly at the bedside and elsewhere, the steady migration of testing from central laboratory to the point of care seems inexorable. But laboratory managers and directors in many care settings can confirm that, when point-of-care testing is adopted too hastily, it can be a case of the technology tail wagging the diagnostic testing dog.

At Baystate Health in Springfield, Mass., for example, “We do a complex menu of point-of-care testing, and we’re dealing with dozens of sites, hundreds of devices, and thousands of operators,” says James H. Nichols, PhD, DABCC, FACB, director of clinical chemistry. Sometimes, “physicians say ‘I need POC testing,’ and think it’s going to solve all the problems of the world—when in fact it can just throw another wrench of technology into an already overburdened system.”

The faster test, he stressed, isn’t necessarily the better test. “I’m not necessarily pro-point of care. I’m for using the right technology for the right patient at the right point of time.”

Speaking at CAP ’08 in a panel on “Point-of-Care Testing: A Tale of Three Institutions,” Dr. Nichols was one of three point-of-care testing experts who described the benefits of their expanding programs and some of the pitfalls that have cropped up along the way—or may lie ahead.

Point-of-care testing helps, Dr. Nichols said, by eliminating transportation of the specimen to the lab, receipt in the lab, specimen clotting, centrifugation, and aliquotting. “But you still have analysis and result reporting and acknowledgment of the result by the physician‚ who is not always at the bedside at the time the test is run. Improved patient outcomes can only really result when POC testing is linked to treatment paths,” he said.

With POC testing devices, the potential for quality problems is high. Complaints about self-managed blood glucose devices represent the largest number of complaints filed with the FDA for any medical device. They have been a factor in more than 3,200 incidents, including 15 in which deaths were attributed to the devices—“usually because a patient at home gets a result, gives himself insulin, and goes into hypoglycemic shock and no one is around to notice,” Dr. Nichols said.

Other risks abound: A recent study showed that poorly managed urinometers and blood gas analyzers can act as an infectious reservoir for resistant microbes. Another reported on nine patients in Southern California who, as molecular genotyping showed, contracted hepatitis B infections because of blood glucose monitoring. “The nursing staff were doing rounds, and although they changed needles between patients, the needle holder was not changed,” Dr. Nichols explains.

There are many sources of error that POC testing has in common with main lab testing, of course. “You can select the wrong test for that patient, you can have patient and specimen misidentification, and so on. I see this all the time where an operator walks up, scans a patient wristband, and the number doesn’t exist in our system because the patient was transferred in from another hospital without cutting off the wristband,” Dr. Nichols said. Other known sources of error: inappropriate collection, wrong techniques, tube types, and specimen clotting before analysis.

“POC testing is a complex system. In the main lab, you have limited instrumentation and a limited number of highly skilled, highly trained individuals operating those instruments, and their sole responsibility is turning out quality test results. With POC testing, we’re dealing with staff who have no laboratory background or minimal laboratory training.”

A chronic problem is that most POC testing is delegated to the staff with the lowest level of education on the unit. “It’s not being done by your master’s-level RN nursing staff; it’s done by nurses’ aides who have less experience and less exposure to preanalytical variables.”

How do you manage POC testing quality in this environment? It can be a problem, Dr. Nichols said. For example, Baystate introduced i-Stat blood analysis modules with data management in the late 1990s, but removed them in 2001. “The problem was they did not have positive patient identification. An ICU patient could be transferred while the patient ID remained in the monitor for the next two or three patients; a patient would be discharged and still showing up in the results. That’s a huge problem.” Bar coding and better data management have eliminated most, but not all, such patient identification errors, he said.

“We use automation wherever we can,” to remind staff to fulfill the processes and help them do so. Automatic operator identification or QC lockouts are basic checks and balances that help ensure test quality, and operators are able simply to dock POC testing devices, with results reporting, billing, and medical record updates all taking place automatically. However, the policies and procedures, device validation, maintenance records, and other parts of the POC testing “information infrastructure” are still manual at this point. “The challenges of managing POC testing are really all this paperwork,” he said.

Dr. Nichols refers to two philosophies of POC testing management. “You can insert yourself as the laboratory into a process, and that’s effective in certain situations. But in the long run, I’ve learned from my experience that inevitably it will fail. The best way to really manage a POC process is to integrate that process into team patient care. The nursing staff and clinicians are involved in patient care, and point of care should be seen as an integral part of their day-to-day patient care responsibilities.”

This kind of partnership can be a gamble because the laboratory is the one with the CLIA license. “But I can’t be out there 24 hours a day holding every operator’s hand to make sure they’re doing the testing appropriately. I have to give the responsibility, tell them what needs to be done, and then trust they’re going to do the right thing. When they don’t, I have to help them improve.” He views his role as being a resource rather than serving as a “dictator,” and this approach has made him more successful in the long run.

POC testing is an entirely different technology, and clinicians need to take that into account in interpreting results. “Is that fast result really making a better result for patients?” In Baystate’s oncology center two blocks away, patients undergoing chemotherapy needed more rapid creatinine results than the core lab could report. “We evaluated different devices that had creatinine available, but the problem was no two creatinines match each other. One device overestimates the level, so if you don’t adjust the reference range or clinical cutoffs, you’re going to be decreasing the amount of drug given these patients, and they won’t be getting as effective therapy,” he said.

As a result, Baystate POC testing results are not freely interchangeable with other methods, and electronic reporting must keep results separate. POC test results require selection of site location, which prevents intermixing of lab and POC results and misinterpretation in the patient electronic medical record.

Even with glucose meters, unreliable results can be a problem. “You can get really erroneously low results if a patient is severely dehydrated, hypertensive, in shock, or in a hyperglycemic/hyperosmolar state. That’s why every patient walking into your ED should not get a glucose until that patient has been assessed by a clinician to make sure they don’t have one of these conditions.”

Baystate clinicians didn’t believe this at first. “They wanted to have a glucose done on every patient who walked in the door to streamline care.” To make the case against this practice, the laboratory did glucose tests on 60 patients in diabetic ketoacidosis and was able to demonstrate that the POC glucose results were indeed incorrectly biased downward in these patients compared with central laboratory instrumentation.

In New York, glucose testing in the ED is routine even though it should not be done, said Peter J. Howanitz, MD, vice chair and director of laboratories at SUNY Downstate Medical Center in Brooklyn and a speaker on the CAP ’08 POC panel. “Our emergency department tells us it’s done in every ED in New York City. When the patients hit the door, they end up getting a glucose as a triage patient.”

But some strange consequences have come to light. “At that point, we don’t have a medical record for the patient. So what does the operator do? They make up one. It might be ‘1234567’ or ‘111111.’ But when we do have a patient who has that number, this patient must have over 400 glucose results in his medical record, some in the ED, some from other times when he was supposedly hospitalized.” Other problems are created by the large percentage of results where the patient does not have a physician.

Despite its best efforts, the staff so far has gotten only about 18 percent of ED glucose values into the patients’ medical records. “So whatever can go wrong certainly will go wrong,” Dr. Howanitz said. “And with POC systems, no matter how good they are, people will do their best to try to get around the system.” The hospital is making a concerted effort to improve recording of glucose results and expects the next glucose meter vendor will help. “Three of the manufacturers now have systems available that look after having the correct medical record number and actually lock out if they don’t. Some are so specific they will only take a medical record number of patients who are at that unit.”

“But this is a problem that commonly occurs, and if you are looking at glucose at your institution, it’s one of the things you should look at.”

As at Baystate, the physicians at SUNY Downstate tend to believe POC testing can solve just about every problem, he said. “A little over a year ago, our OR physicians came to me and wanted i-Stats for 60 anesthesiology residents and 18 nurses. So that’s 78 new operators. As in most discussions, they had no evidence presented; it was just what they wanted.” The cardiothoracic physicians on staff now want the ability to do PT/INR on the i-Stat, he added. “Even though it’s closer than the men’s room to our clinical lab, they felt they needed this. So these are the kinds of issues we get on a very continuing basis.”

Since the late 1980s, the Mayo Clinic in Rochester, Minn., had had its 24/7 phlebotomy team performing glucose monitoring in the inpatient setting, said Paula J. Santrach, MD, co-director of point-of-care testing at Mayo and a panel speaker. “But a few years ago, we found we were unable to meet the need for coordination of meals, testing, and insulin administration, so we shifted it all from phlebotomy to the nursing service for the non-ICU testing. We did it through the connectivity and data management that Dr. Nichols described [in the panel]. It’s the only way we could ever have accomplished moving from 300 operators to 3,800 operators.”

A specialized phlebotomy team continues to perform testing in the ICU setting, which has high volumes because of the hourly testing needed for intensive insulin protocols. “We used Lean principles to develop a coordinated collection schedule in each ICU,” Dr. Santrach said.

“That’s made life easier, but we still have issues with patient ID compliance because of unavailable patient wristbands, and we have sources of potential contamination. Our infection control specialists determined that all meters have to be cleaned between patients, but using bleach causes significant damages to the devices. We’re now trying to move to an alternative cleanser, but it’s also really hard on the devices. We get problems with bubbling of the plastic, which leads to cracking.”

In their practice, specimens for blood gas analysis are collected by the stat laboratory technicians who also record the ventilator settings at that point in time, she said. There­fore, “Mayo clinicians are actually used to getting results together with the ventilator settings in the same place in the medical record. If all blood gas testing was done at the point of care, the ventilator settings would only be available at the bedside. So we’re working to see if there’s a way for them to download and view the ventilator settings from the device at the same time they view the test results, to see them together. That’s another challenge for us.”

Comparability of results is a particular issue with POC INR (international normalized ratio of prothrombin time) monitoring for patients on anticoagulant therapy, Dr. Santrach explained. “You want comparability with the current laboratory method, but all the technologies are different.” In side-by-side comparisons of most of the devices on the market, “you can see there are different levels of comparability between point of care and the laboratory. But even between labs, the INR often doesn’t match. We have sometimes unrealistic expectations about how well INR eliminates the differences.”

Some INR results should be checked using the lab method, she noted, pointing to the increasing divergence among methods when results show higher INR values. “The devices are designed to be used with patients with stable anticoagulation; they’re more problematic if the patient is unstable. Our own policy is to check POC results higher than 6 with another specimen sent for lab testing, but other labs might start checking at results higher than 4.” Laboratories also have to be careful when the patient is concurrently being administered low-molecular-weight heparin. “Some devices are susceptible to that, so it’s a good question to ask your vendor.”

Recent analyses showed that anticoagulation clinics and patient self-testing do make a difference by increasing the patients’ time in therapeutic range and reducing the incidence of recurrent thromboembolism and major hemorrhage, depending on the frequency and intensity of INR testing. “So this is why POC INR is very evidence-based,” and the reason for its widespread adoption in outpatient settings, she said.

Aside from the lack of data management for POC INR, which Dr. Santrach said is changing, the challenge is understanding the appropriate patient populations for this kind of testing. Is POC INR appropriate as a pre-procedural screening tool? Who is responsible for following up unexplained results? “Especially for us as a referral practice, our INR may be different from the INR you get at home,” she said.

Baystate uses i-Stats to perform point-of-care Hgb/Hct testing in the OR and critical care units, Dr. Nichols said, while the ER uses pneumatic tubes to send samples to the main laboratory for blood gas, ionized calcium, and coagulation testing.

In 2006, the medical center standardized ACT testing to the ITC Hemochron Elite, which minimized training for the float staff in cardiology. That has allowed use of single targets for low-dose and high-dose heparin rather than device-specific targets. Again, “the biggest problem we’ve had is data management,” Dr. Nichols said. However, operator QC lockouts have improved compliance.

Baystate resolved another problem it was having in the Coumadin clinics with moderate-complexity coagulation testing by standardizing all outpatient PT/INR testing to waived prothrombin time. Each clinic now self-manages its own testing and data, which has also brought improved compliance over moderate-complexity PT/INR where the laboratory had to be the medical director on the CLIA license.

Thromboelastography (TEG), or viscoelastic tests of how fast a clot forms, how strong it is, and how fast it dissolves, has become of increasing interest at the point of care as hospitals face blood management initiatives and the need for better decisions about transfusion during surgery, Dr. Santrach said. “People who use it at the point of care predominantly look at the pattern it produces in order to monitor the effects of heparin administration, platelet dysfunction, or fibrinolysis. It’s actually the only really good way we have to pick up fibrinolysis in the operating room.”

Again, research has shown differences between conventional and POC coagulation testing, she said. “You won’t necessarily get the same answer.” A 2006 study of TEG testing in liver transplantation found only moderate agreement in transfusion decisions; it concluded that transfusion practice is likely to differ according to the method of coagulation monitoring used.

However, a TEG-based transfusion algorithm has been particularly helpful for cardiac bypass surgery. “When you look at patient outcomes, we’ve found that use of the algorithm can really reduce the number of plate­lets transfused, although there isn’t a significant impact on blood loss. So it’s been very helpful from my perspective.”

TEG testing devices are not portable but can be used at the point of care. “Quality control is an issue because it’s nearly impossible to simulate patient samples,” Dr. Santrach pointed out. “You can do it, but I’m not sure how accurate it is. High testing volume also becomes a problem because most devices can do one or two patients at a time.” Finally, there is a question as to who performs the interpretation, the pathologist or the anesthesiologist. “That varies according to institution. In our facility, we give results to the anesthesiologist, who is usually the ordering physician. In other institutions it’s the pathologist who interprets.”

SUNY Downstate took over the anticoagulation unit about eight months ago when the cardiologists on staff decided they no longer wanted it. As typically happens, “they dumped it on the lab because it was a money loser,” Dr. Howanitz said. In the process, SUNY has experienced some of the same problems as Mayo. “As Dr. Santrach pointed out, all of these coagulation analyzers give entirely different results from what we have in the clinical lab, and the lab is unable to standardize our instruments against each other. The therapeutic ranges are different, and one has to work on each of these instruments individually.”

With TEG testing, interpretation of results is especially important because it has to be done in real time. “One has to look at the shape of the curve over 30 minutes, then interpret it. If the clinician is in the OR, TEG is usually done for patients who have ongoing surgery, so they need to get a result back to them immediately.”

The answer at SUNY Downstate was e-mail. “We would do the test in our OR lab and would then e-mail the pattern to one of our pathologists who would interpret it. It’s considered a formal consultation.” However, the hospital ultimately gave it up because a new cardiac surgeon came in who ­doesn’t require TEG measurement for his patients.

Another problem arose because of the instruments’ sensitivity to vibration, Dr. Howanitz said. “We put the instrument in the OR lab, which is a small room off the OR, and when we were spinning blood in the centrifuge, it created enough vibration to make problems with operation of the instrument. That was another reason our new surgeon decided against it; he had had the same experience at another medical center.”

There is an explosion of HIV testing through rapid testing, the panelists reported. At this point, Mayo uses HIV POC testing only in specific situations with employee or patient exposure, Dr. Santrach said. “But there are large numbers of POC HIV tests being done, and it’s probably going to become even more widespread,” said Dr. Howanitz. While SUNY Downstate is located in Brooklyn, there are plans to make the Bronx the first New York City borough where every person between 18 and 64 knows his or her HIV result. “There was great interest in spreading this throughout the entire city of New York, although I don’t know if the city has the money to do this.”

However, the rapid test of oral fluids that is used at the point of care produces false positives, and any positive result must be followed by confirmatory testing by Western blot. “We were one of the sites that chose not to use the oral fluids,” Dr. Howanitz said. Like most states, New York requires that hospitals do HIV prescreening of women who are delivering. “But there are Centers for Disease Control recommendations that everyone be screened when they see their physician. If we did that, we figured we would need 65,000 HIV tests done in our lab. POC devices would not be the way to do the tests because you would need an army of people to do all the testing.”

At Baystate, HIV testing is also done as a rapid test, not a POC test per se, just for the subgroup of women who show up in the delivery room. “It’s a very small number—people who may be traveling or we don’t have medical records on them, or for some reason they don’t have prenatal care,” Dr. Nichols said. He is not in favor of making the rapid HIV test universal. “If you asked every patient every time they see a physician if they want a free HIV test, the health system would go bankrupt because the rapid tests are $20 to $30 apiece and we see hundreds of thousands of patient visits a year.”

Helicobacter pylori testing at the point of care is another test that has become important. At SUNY Downstate, H. pylori testing produces a color change on a card. “But if you ask physicians to read these results and put the results into the computer, it’s not going to happen,” Dr. Howanitz said. “They all want POC testing, but if they’re the ones doing the procedure, and they wait 30 minutes to see a color change, they are not going to enter it as a lab result.”

His laboratory worked out a way around this recording lapse. “We have the physician look at the result, then send it to the lab where our people would officially read it. The result is stable over almost 24 hours.” The physicians like this service. “We’ve actually gone to private suites where we’ve done this and they used it just to get the result into their medical records. So it’s kind of a marketing issue as well; it’s a service we can offer to physician practices,” he said.

At Mayo, the endoscopy group expressed interest in using the CLO test for H. pylori, but it was never implemented, Dr. Santrach said. Biopsy specimens have a typical 24-hour turnaround time, and breath, stool, and serologic testing are readily available from the lab.

Perhaps a yellow flag of caution should be attached to POC testing in general, the panelists suggested, because it tends to be overused. “From clinicians’ standpoint, it’s a convenient means of getting faster laboratory tests. But they don’t quite think about the differences in actual techniques as well as the cost differences compared to the main lab,” Dr. Nichols said.

He suggests figuring out what’s wrong with the system first if you’re thinking about POC testing. “Facilitate communication, map­ping and diagramming the flow of patients, and find where POC is going to play a role. Start there, rather than use POC as a convenience or as a tool, because if you don’t understand the complexities and the holdups of your current system, just getting a patient a fast test result isn’t going to move patients faster through the system.”

Of course, if the cost of POC testing can be translated into moving the patient through the health care system more quickly, then it’s all to the good, Dr. Nichols said. “But if you’re doing it just because you want a faster test, then there’s no reason to get a point-of-care test. You have to integrate it into the whole health care picture.”