College of American Pathologists
Printable Version

  Critical values—looking closely at calls, cutoffs


CAP Today




November 2008
Feature Story

Karen Lusky

Managing critical lab values comes with more than its fair share of challenges. For the laboratories at Emory Healthcare, that meant making 1,500 calls a week. For Dennis Dietzen, PhD, medical director of the core lab at St. Louis Children’s Hospital, and others in labs nearly everywhere, it means managing the preferences of individual physicians, which are known to trigger conversations like this one:

Infectious disease physician: “Hello. ... What is your critical ALT [alanine transaminase] value?”
Lab: “600 U/L, doctor.”
ID physician: “Well, that is waaaay too high. I changed a patient’s dose of azithromycin two weeks ago and he’s fine, but his ALT went from 125 to 350. Why wasn’t I called?”
Lab: “What would you have done following such a call, doctor?”
ID physician: “Well, probably nothing immediately, but it would have been nice to know.”

And had the lab changed the critical value for ALT to 300 U/L, it might have heard the following from a gastroenterologist: “Hello. ... Look, we have gotten 25 phone calls this morning about elevated transaminases in our patients. This is a GI ward for cryin’ out loud. ... Please, nothing less than 600. Give me a break.”

Dr. Dietzen shared those conversations (the first real, the second imagined) with a standing-room-only crowd at an AACC annual meeting session last summer, at which he and two Emory laboratory representatives shared their experiences in changing their critical value cutoffs and reporting systems last year.

As for those 1,500 calls a week that technologists in Emory’s two core labs were making: The lion’s share involved critical value cutoffs that were conservative compared with those reported in the literature and those of a sampling of Emory’s peers at other academic centers, says Corinne Fantz, PhD, co-director of the core laboratories at Emory.

Emory’s core labs were actually flagging about 3,000 critical values in their systems each week, Dr. Fantz said. But Emory’s medical technologists call clinicians about an initial critical value and then not a subsequent one unless there is an intervening normal result (except for potassium or glucose, which the laboratory calls every time).

As the first step in a quality-improvement project aimed at safely reducing the call volume, Dr. Fantz and her team found that 93 percent of the calls involved just eight tests: potassium, glucose, white blood cells, hemoglobin, hematocrit, platelets, calcium, and troponin.

The other seven percent included an assortment of critical values for magnesium, phosphorus, and therapeutic drugs. (For this study, blood bank and microbiology calls were excluded.) At the time of the project, Emory’s two core labs served three main hospitals: Emory University Hospital, Emory University Crawford Long Hospital, and Wesley Woods Geriatric Hospital, as well as the Emory Clinic. Emory has since added an orthopedic hospital, which has not driven up the number of critical value test results that much, Dr. Fantz says.

As the second step in the quality-improvement project, Dr. Fantz and her team reviewed the literature on critical values for the eight identified tests. They focused on a CAP Q-Probes study of critical values in hospitals (Howanitz PJ, et al. Arch Pathol Lab Med. 2002; 126 : 663–669) and a survey of hospitals published in the Journal of the American Medical Asso­ciation (Kost GJ. 1990; 263: 704–70 7). In addition, Dr. Fantz spoke with laboratory directors at six academic centers to find out what their critical values were for the eight tests.

Next, she and her team took the new proposed cutoffs for the eight tests to the medical practice committee so they could explain the rationale for changing them.

The committee agreed to change the hemoglobin and hematocrit critical values, which made a huge dent in the number of calls. Previously, the lab had been calling clinicians if a patient’s hemoglobin was less than 8.1 g/dL, Dr. Fantz says. By comparison, the Q-Probes and JAMA-reported studies were 6.6 and 7, respectively, and the academic centers, 6.3, on average. Lowering Emory’s critical value to 6.6 turned 202 calls per week into 33.

Reducing the critical value cutoff for hematocrit from 25 percent to 20 percent eliminated more than 320 calls per week. “The Kost and How­an­itz surveys were 18 and 20 [percent], respectively, and our peers, 20 percent,” Dr. Fantz says.

The medical practice committee decided to leave the hypokalemia critical value at 3 mEq/L rather than adopt the lab’s proposed cutoff of less than 2.7. The Kost and Howanitz surveys were less than 2.8, and the six academic institutions, less than 2.7. “We have a large outpatient population at Emory, and the medical practice committee was concerned we could easily miss people with potassium values of 2.7 or 2.8 if we set a critical cutoff too low,” Dr. Fantz explains.

Other changes to the tests and their effect on call volume were as follows:

  • Calcium. The lab reduced the critical value from 7 mg/dL to less than 6.6 for hypocalcemia, trimming the number of calls from 64 to 22 per week. “We did not adjust the upper critical limit beyond 13,” Dr. Fantz says.

  • Platelet count. Emory’s previous critical value, which was less than 50,000, is now set at less than 25,000. The Kost survey identified less than 37,000 as the mean critical value, and the academic center peers were less than 26,000. But Emory “may have more acute clientele than some centers used in those surveys, and we have a hematology service on site that’s more comfortable with a lower cutoff,” Dr. Fantz says. Adjusting the platelet count cutoff reduced call volume from 134 to 42 calls per week.

  • White blood cell count. The lab moved the lower critical value cutoff from 2,500 to less than 2,000. “We did not adjust the upper limit, which was set at greater than 50,000,” Dr. Fantz says. Emory’s peers had an upper limit of greater than 44,000, the JAMA-reported survey, 37,000, and Q-Probes, 30,000. “Critical white count calls went from 117 to 65 calls after adjusting the lower limit,” she says.

  • Glucose. Adjusting glucose critical values on the lower and upper ends from 50 mg/dL and 400 mg/dL, respectively, to 45 and 450, reduced calls only slightly, from 78 to 51, which is probably not even statistically significant, says Dr. Fantz. Due to the hospitals’ tight glycemic control protocols, the majority of glucose testing is done on the nursing units—about 66,000 tests a month.
  • Settling on a troponin critical value grew a bit complicated. For one, the literature on troponin didn’t offer a clear consensus for defining a cutoff, Dr. Fantz says. The Q-Probes and JAMA studies, perhaps because of when they were done, did not address critical values for troponin or any other cardiac marker. And only three of the six academic centers had critical troponin values.

    Yet troponin is a sensitive marker for myocyte damage, and anything above the 99th percentile is considered elevated, though “whether or not ‘elevated’ should be defined as critical and called to the licensed caregiver will depend on whom you ask,” Dr. Fantz says.

    There was another glitch: In working with the ED on turnaround times for troponin, the lab discovered that using a lower cutoff for a troponin critical value caused a delay in reporting the test result because of the documentation requirements for critical values.

    Autovalidated results can be reported in three minutes, on average, moving from the instrument to the middleware to the medical record, Dr. Fantz says. By contrast, it was taking on average 30 minutes (median time, 11 minutes) for the critical values to go into the middleware and await technologist intervention, and then for the phone call to be made, received, and read back/verified, and for the results to be released.

    Thus, to improve the turnaround time for troponin, the lab collaborated with the emergency department and agreed to increase the critical value cutoff to 0.5 ng/mL, which is 10-fold higher than what the lab was using previously. As a result, the lab’s calls for critical troponin values dropped from 167 to 43 per week. The emergency department was comfortable with this change because it has a tracking board to inform ED staff when the results are ready and would rather have them released than delayed for documentation reasons.

    Raising the troponin critical value cutoff helped the lab report the low-positive troponins faster, Dr. Fantz says. But the lab had to consider whether it’s worthwhile to have such a high critical cutoff to allow for autovalidation of the lower results. For that reason, and because of demands from the ED for ever-faster troponin turnaround times, the lab is now considering point-of-care testing for troponin in that setting.

    The lab completed its six-month changeover in critical value cutoffs in June 2007. Now, Dr. Fantz says, Emory’s critical value thresholds overall probably fall somewhere in the 50th percentile range for academic medical centers.

    Overall, the quality-improvement project decreased the technologists’ call volume for critical values by almost two FTEs a year, freeing up those staff to spend their time verifying results instead. And “the clinicians are happy about the changes as they are now doing so much less documentation on their end,” Dr. Fantz says.

    That’s not to say the lab doesn’t occasionally receive complaints from physicians about not getting called about certain results. Dr. Fantz recently received such a call from a physician who saw a patient’s potassium of 6 and wanted to know why no one called her about the critical value. “But our cutoff is greater than 6.0 or 6.1,” Dr. Fantz says. “It’s unfortunate but, operationally, you have to make the cutoff somewhere.”

    “And keep in mind,” she adds, “that the clinicians are getting the results—for example, that 6.1 potassium—but they aren’t getting called to alert them to it. Some don’t want to be called that often when they are seeing 30 to 40 patients a day.”

    Emory’s overhaul of its critical value reporting system called for changes to the lab information system and to the protocol for closing the loop in communicating results.

    At the outset of the project, the technologists in the lab were not only calling all of the critical value results, but also continuing to try to reach someone until they succeeded. So Emory had a computer programmer write a program to identify any critical value for which the technologist had not completed the communication loop.

    Now when technologists make the first call and do not reach anyone, they enter a comment to that effect in the computer, identifying whom they tried to call and at what time, says Alexis Carter, MD, director of pathology informatics at Emory University Hospital. “Entering the attempted call comment triggers the program to flag it as incomplete,” she says. The computer prints out a list of incomplete calls every seven minutes for customer service representatives who continue to call until they hand off the critical value. “If there are no patients on the list, the report still prints and states there are no patients with incomplete calls,” Dr. Carter says.

    “It never rolls back to the tech,” she adds. If customer service representatives have trouble communicating the result, they call the resident on call who also attempts to reach the patient’s licensed care provider. “If that fails, then the resident communicates [the critical value] to the attending,” and will call an outpatient directly to let him or her know about the critical value and the need to go to the emergency department for evaluation, Dr. Carter says.

    Emory has two additional studies in the works of critical lab value reporting. In one, charts are reviewed for the patient care impact of not calling a subsequent critical value (except for potassium and glucose) unless there is an intervening normal result.

    Dr. Fantz notes that one survey in the literature suggested that only 11 percent of hospitals had such a policy. Many of the subsequent critical values involve patients being closely monitored in the ICUs whose condition may not change over a short period, she says. And calling the caregivers repeatedly may, in fact, only make it more difficult for them to recognize patients with truly critical conditions.

    Nathan Spell, MD, chief quality officer at Emory University Hospital, who asked the lab to do the study of the impact of the policy on patient outcomes, agrees that in some cases, a clinician will find it acceptable that the second critical value was not called.

    An example is “a young patient with leukemia who can tolerate a very low hemoglobin because it’s been low for weeks.” But suppose someone who had surgery yesterday has that same hemoglobin value, Dr. Spell says. “That’s critical because it may indicate internal bleeding. And if it drops further, it tells you the person isn’t responding to whatever you did after being called [about] the first abnormal value.” He thus thinks it’s probably best “to err on the side of calling all critical values,” but is “willing to be proven wrong.”

    Emory is also designing and studying a computer decision support system to catch patients at discharge who have critical values or ones that are trending abnormally. In such cases, an alert will pop up on the computer monitor during the discharge process asking the physicians if they still wish to proceed in releasing the patient, says George Mathew, MD, instructor of medicine in the section of hospital medicine at Emory, who is helping to develop and test the system.

    Dr. Mathew and his team are looking at 12 commonly used lab tests, including creatinine and white blood cell trends. They have created computer rules to identify, for example, patients who left the hospital with a hemoglobin of less than 7, platelets less than 20,000, and INRs more than 4.

    They considered including calcium but did not do so because low calcium is often actually normal when corrected for the albumin level, Dr. Mathew says. “And if you don’t have an albumin level, you can’t determine that.” They also decided against including troponin because for now they want the program to be simple and user-friendly for clinicians.

    The computerized program will first be applied to the inpatient setting and later extended to outpatients. “The goal,” Dr. Mathew says, “is to save one person here, save one there—and over time, you save a lot of people.”

    For laboratorians trained in adult health care settings, managing pediatric critical lab values isn’t a totally different ballgame, but it is one with some nuances and different rules that can catch newcomers off guard.

    As a case in point, St. Louis Children’s Dr. Dietzen explained in his AACC presentation how he had been trained in an adult institution that hyperkalemia will cause abnormal EKGs (for example, peaked T waves, smaller Q waves, and prolonged QRS). So when he started working in pediatrics and made his first call to the NICU about a high potassium, he asked the doctor: “What does the baby’s EKG look like?”

    This question is usually met with indifference. It turns out, Dr. Dietzen said, that children’s hearts are a “little bulletproof” to moderate potassium elevations early in life.

    Dr. Dietzen shared what he views as a handful of tests unique to pediatrics that represent “must know now” results for clinicians, as well as potentially confounding pre- and postanalytical considerations that labs should consider for some of the tests. And potassium is on the list on both counts.

    Hyperkalemia is common in neonates, Dr. Dietzen said. Assessment is confounded by hemolysis, an occurrence particularly common when using capillary specimens. “The problem is further complicated in our setting because we encourage the use of whole blood for basic electrolyte profiles to save time and blood volume,” he said. “Often, there is not enough residual specimen to examine for signs of hemolysis.”

    Clinicians’ responses to calls from the lab about hyperkalemia vary. “Some say, ‘I know the potassium is high. It’s always high. Don’t call me again.’” Others appreciate the calls, he said. Either way, the lab calls clinicians about a critical potassium result and lets them know that a plasma specimen is required to accurately determine potassium concentration.

    Sometimes a specimen that’s clearly hemolyzed has a normal or marginally elevated potassium concentration, raising questions about whether the hemolysis is masking a patient’s hypokalemia.

    “If we can establish a reason for in vivo hemolysis in such a case, we don’t get so upset about it,” Dr. Dietzen says. In these circumstances, he will usually call the emergency department, for example, to get an idea of whether the hemolysis occurred inside the patient rather than inside the tube. “Conditions that would cause in vivo hemolysis include anything cytotoxic—chemotherapy and immune hemolytic anemias, as examples—but in those situations the hyperkalemia is usually rather mild because the kidneys adapt and get rid of the excess.”

    Other “must know now” test results include:

  • Bilirubin. The “unconjugated beast” is the bad guy in little kids because it gets in the brain and causes irreversible brain damage, Dr. Dietzen said. The American Academy of Pediatrics guidelines for dealing with unconjugated hyperbilirubinemia dictate intervention in the first hours of life. Fast communication in these cases is essential, he noted.
    “Conjugated bilirubin is also scary but not for the same reason—it indicates that the baby has biliary pathology,” which includes a long differential diagnosis. Biliary atresia is the one that scares the gastroenterologists the most, though it’s not common, Dr. Dietzen said. If the condition is detected soon enough, surgical intervention can delay the eventual need for liver transplantation.
    Conjugated bilirubin is a test where you have a little bit of time, however, though confirming a diagnosis requires additional workup, he added.
  • Acetaminophen and aspirin levels. Because the lab is “blinded to the ingestion time, we consider any value above the detectable limit of our assay to be a critical value,” Dr. Dietzen says. “We can do that because we don’t have hundreds [whereas] if an adult lab did that, they would be on the phone forever.”
  • Ammonia. Elevated ammonia can cause irreversible neurological damage in children and is one of those results where the clock is ticking. “The main physiologic causes of elevated ammonia levels are inborn errors in the urea cycle,” Dr. Dietzen says. “There are two ways to reduce ammonia: pharmaceutical intervention that traps ammonia, resulting in excretion. The other is dialysis.”
    Ammonia testing also involves preanalytical issues that can drive up the value. Transport times to the lab—even as little as 30 minutes—drive up ammonia mostly due to deamination of glutamine, cautions Dr. Dietzen.
  • Total CO2. While a low level indicates potential acidosis, it more often reflects a preanalytical issue. “Often the low CO2 is due to the low blood volumes collected from babies,” explains Dr. Dietzen, noting that the partial pressure of CO2 in blood is higher than that in the atmosphere. Thus, when you draw a very small sample from a baby, a large amount of surface area is exposed, and the CO2 can diffuse out of the specimen in short order. Thus, “by the time the lab result is reported, it looks like a low total CO2, which also happens in metabolic acidosis and gets people thinking about mitochondrial metabolism errors and other inborn errors of metabolism that result in accumulation of organic acids.”
  • Because of this preanalytical problem, some pediatric labs do not report total CO2 on these types of specimens, Dr. Dietzen says. Instead, the lab at St. Louis Children’s Hospital is trying to educate house staff physicians and interns to avoid overreacting to low total CO2. “Once you get a blood gas, the ‘acidosis’ often goes away,” he says.

    Dr. Dietzen also reviewed a couple of tests that fall into what he calls the “must know” rather than the “must know now” or critical category for pediatrics. One is sweat chloride for cystic fibrosis. “The pulmonologists want to get ahold of kids [who test positive] as quickly as they can,” but a child is not going to need a lung transplant sooner if the diagnosis isn’t confirmed for a couple of days, he said.

    Lead is another example. “You want to get chelation going quickly, but it’s not a situation where intervention will proceed in minutes or hours.”

    Mayo Clinic in Rochester, Minn., is looking at a special system for reporting test results that do not make the critical value cut but could cause significant problems if they escape the attention of physicians.

    The lab there is working with the cardiology and radiology departments to develop a computerized “second tier” of notification that would electronically flag “semi-urgent results” from all three departments in the electronic medical record. That way, the ordering pro­vider would be alerted to the condition, says Paula Santrach, MD, Mayo Clinic’s co-director of point-of-care testing and chair of the Clinical Practice Quality Oversight Subcommittee. “We are also trying to implant [the results] in the electronic medical record so they can be used for clinical decision support,” such as preventing medication ordering errors, she adds.

    The classic example in cardiology, she says, is a prolonged QT interval on an EKG, which in some patients can indicate a higher risk of sudden cardiac death. Certain antibiotics must be avoided in those patients, Dr. Santrach says.

    The lab is in the midst of identifying lab results for the second-tier notification system. INR is the first one it came up with. “We are looking at INRs greater than 5 in the inpatient setting,” says Dr. Santrach. The laboratory has a critical INR value of 6 and calls clinicians about that. She’s unaware of any protocols where a clinician would want a patient’s INR to be at 5.

    Developing the second-tier notification system helps keep critical lab results critical. In reporting critical values, Dr. Santrach says, the key is to ask whether the result is important enough that you need to interrupt the physician’s clinical workflow and divert the person’s attention to this patient. “For critical results, the answer to that is yes. For semi-urgent results, no.” But the latter are important, and “we are going to send [clinicians] an inbox message to make sure they don’t miss them.”

    Karen Lusky is a writer in Brentwood, Tenn.