New push for standard approach to critical values

Kevin B. O’Reilly

September 2014—Newly reported survey data that show widely varying international practices on managing critical values may demonstrate the need for a new guideline—already in development—to help laboratories formulate evidence-based policies.

The new data from European labs were presented during a session at the American Association for Clinical Chemistry’s Annual Meeting and Clinical Expo in Chicago (“Critical Result Management Practices: Global Perspectives and Recommendations for Best Practices”). The session also provided a preview of a forthcoming draft guideline from the Clinical and Laboratory Standards Institute that represents the organization’s first formal attempt to advise laboratories around the world on critical values reporting.

The survey of 871 participating European labs was conducted by the Task and Finish Group on Critical Results of the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM), in partnership with the Australasian Association of Clinical Biochemists. While single-country surveys of critical result practices in Europe have been conducted previously, this work marks the first time that labs in many—30—different countries answered the same set of questions about how they operate in this sphere so essential to patient safety.

“It’s rather convincing that critical result management practices are really greatly varied. This means both within and between countries,” Eva Ajzner, MD, PhD, said at the session. Dr. Ajzner, from the Teaching Hospital of the University of Debrecen Medical and Health Science Center in Hungary, is chair of the EFLM group that conducted the survey.

Hungary represented about six percent of the responding laboratories, while French labs accounted for a plurality of respondents with 29 percent. Laboratories in the United Kingdom represented 13 percent of respondents, while another 11 percent of responses came from Italy. Between 20 and 73 laboratories in Serbia, Belgium, the Czech Republic, the Netherlands, Croatia, Slovenia, and Norway also participated in the survey, which was conducted online between September 2012 and March 2013. The survey drew responses from private and publicly owned labs serving different settings—hospitals, outpatient clinics, adults and pediatrics—in metropolitan and rural areas.

The data show a number of areas where European labs appear to be falling short of what they should be doing in managing critical results, Dr. Ajzner said. The major shortfall in European labs comes in the development of critical alert lists, thresholds, and auditing practices, Dr. Ajzner tells CAP TODAY. Such critical results management practices “should be performed in a shared policy between laboratory and clinical staff. Although this fundamental requirement is not fulfilled in many European laboratories, in some countries a majority of laboratories cooperate with clinicians in critical result management,” she says. In many European labs, the job of designing critical results lists “is mainly a task of lab professionals, and physicians are seldom involved,” Dr. Ajzner says. Less than half of European labs ask doctors about critical value thresholds, but among laboratories in Norway and the Netherlands, 72 percent and 88 percent, respectively, do involve clinicians in these decisions. That compares with 73 percent in the U.S. (Howanitz PJ, et al. Arch Pathol Lab Med. 2002; 126[6]:663–669).

“I’m not proud to show you these results,” she said at one point in her talk. “In 44 percent of European laboratories, there is absolutely no protocol to read back the results and for labs to record this.” Less than a quarter of labs said they do require read-back of results but that they do not record that process. “The proper actions are done in only about 33 percent of European laboratories and in about 60 percent of laboratories in the U.K., where read-back and documentation is most often applied,” Dr. Ajzner said. That compares with a 91 percent rate of read-back and documentation found in a survey of more than 700 American labs (Dighe AS, et al. Arch Pathol Lab Med. 2008; 132[10]:1666–1671).

European laboratories do have the upper hand in some areas of critical values management, Dr. Ajzner says. For example, the EFLM survey found that in nearly half of European labs, it is the medical doctor on duty who usually delivers the critical result, which Dr. Ajzner says provides an opportunity for immediate consultation at the time of notification.

The EFLM survey also provides key insight into the wide variation among labs in terms of which tests are selected for inclusion in critical value lists. For the adult lists, just 11 laboratory tests were shared by more than half of European labs and only three were listed by more than 90 percent. By comparison, 21 tests were shared by more than half of the 163 American labs that took part in a CAP Q-Probes study (Wagar EA, et al. Arch Pathol Lab Med. 2007;131[12]:1769–1775). Eight tests were listed by more than 90 percent of these U.S. laboratories.

“The variations seen in the European survey and other studies on which values and thresholds are chosen may be attributed, in part, to differences in the patient populations and clinical settings that laboratories serve, as well as differences in the test methodologies they employ,” Dr. Ajzner says.

She adds, however, that these factors do not tell the whole story.

“There is a lack of published, evidence-based clinical outcome data for all but a handful of laboratory medicine tests, which is probably the main contributor to the disparity in [critical value] list composition among laboratories,” Dr. Ajzner says.

The shortage of outcomes evidence to drive decisionmaking on critical values was noted by another speaker at the AACC session, Andrea R. Horvath, MD, PhD. She is clinical director of South Eastern Area Laboratory Services North at Prince of Wales Hospital in Sydney, Australia, and a member of the CLSI committee that is developing the draft guideline on critical values. Dr. Horvath also served as a consultant on the EFLM survey.

“We don’t have true outcome data to tell us at what critical thresholds urgent notification of results would save lives,” Dr. Horvath said at the session. “For that reason, it’s extremely difficult for us to decide. There are no universal standards.”

“Laboratories all around the world face difficulties when designing alert lists,” she added. “There’s hardly any evidence to help them.”

Andrew N. Young, MD, PhD, another speaker at the AACC session, echoed this point in an interview with CAP TODAY. Dr. Young is medical director of Quest Diagnostics’ Pittsburgh office and heads the CLSI’s document development committee on reporting critical results.

“That’s one of the most significant challenges, which is that there is not a lot of high-quality evidence that defines a level of risk for any given test result,” he says. “A lot of it is based on collective clinical experience and judgment.”

“You can imagine that we’re not going to have a controlled clinical trial where one group of patients will have their critical test results communicated and another won’t,” Dr. Young says. “That would never be permitted, for good reason, so it’s unlikely we’ll ever get that kind of evidence.”

Jonathan R. Genzen, MD, PhD, sees some promise for definitive answers in the transformation to electronic record keeping. Dr. Genzen, who spoke at another AACC session (“Critical Values: Improving the Design, Practice, and Communication of Critical Laboratory Results”), is medical director of the automated core laboratory at ARUP Laboratories and assistant professor of pathology at the University of Utah School of Medicine. He also is the lead author of a review article on critical value communications (Am J Clin Pathol. 2011;135:505–513).

“I do wonder if we will reach a point where clinical data warehouses—basically databases that contain not just laboratory data but also clinical outcomes data as part of the electronic record—could provide enough patients and enough lab values and enough clinical outcomes to look back retrospectively and say that overall mortality was higher for patients in this threshold or that threshold,” Dr. Genzen tells CAP TODAY. “That could provide the type of evidence we’re looking for, more of an evidence-based threshold for critical values and critical value limits.”

That kind of analysis is entering the medical literature. For example, researchers examined a cohort of nearly 40,000 patients admitted to Sarasota Memorial Hospital and looked to associate a risk of death one year after discharge with small intervals of values for five routinely performed laboratory tests: serum creatinine, blood urea nitrogen, serum sodium, serum potassium, and serum chloride (Solinger AB, et al. Clin Chem Lab Med. 2013;51[9]:1803–1813). The study led the authors to propose reducing the upper limits for some tests to avoid “high-normal” results where they found that the mortality risk rose two to three times above the average of the patient population studied. The hospital’s reference interval for potassium had been 3.5–5.1 mmol/L, but the research showed the upper limit should be cut to 4.3 mmol/L. Meanwhile, the reference interval for sodium was 136–145 mmol/L; the researchers proposed dropping the upper limit to 142 mmol/L.

This methodology “allows the potential for [decision limits] to be determined from unlimited data mining of any EMR or a variety of other sources without consideration of selecting and maintaining a healthy cohort,” the researchers wrote. “As the study can be done retrospectively at any hospital or laboratory with extant data, the statistical sample sizes are just about unlimited, meaning that 95 percent confidence intervals can be as small as desired.”

An earlier study that also mined electronic records for help in setting alert thresholds involved an examination of more than 100,000 sodium results over a six-month period. Pathologists from the State University of New York Health Science Center in Brooklyn looked beyond death risk to analyze how the lengths of stay and clinical actions changed for patients with hypernatremia and hyponatremia (Howanitz JH, et al. Am J Clin Pathol. 2007;127:56–59). Based on the study, researchers opted not to change their sodium thresholds of 120 mmol/L–155 mmol/L. They noted that increasing the lower sodium limit to 125 mmol/L would have generated more than 600 additional calls from the laboratory during the study’s six-month time frame.

Dr. Young says this kind of data mining could represent “a great approach” to using clinical outcomes to help set alert thresholds. “It does have some limitations,” he noted, insofar as this kind of retrospective review cannot conclusively demonstrate that the critical value was the cause of any particular patient’s death.

“But,” he says, “anything we can do to derive some objective assessment of risk that organizations can use, or that guideline development committees can use, to try to come to some level of agreement on starting points to help organizations say, ‘We’re going to reassess our approach to critical results and where do we begin?’ That’s a fantastic start.”

In light of the scarcity of gold-standard evidence to guide the setting of critical value limits, Dr. Horvath advised laboratory leaders to pursue a hierarchical approach to the process. Adapting a concept outlined in an earlier study (Sikaris K. Clin Biochem Rev. 2012;33[4]:141–148), she said lab professionals and their clinician collaborators ought to consider, in descending order of importance, these five levels of information: