Summary
Washington University in St. Louis implemented a clinical decision support intervention to reduce unnecessary magnesium testing. The intervention uses a modified soft stop, alerting providers after two normal magnesium levels, prompting them to consider discontinuing the daily test. This approach encourages providers to engage in reflective thinking, potentially reducing over-ordering and improving patient care.
Sherrie Rice
February 2026—Clinical decision support, when done right, can lower the volume of over- or misordered daily laboratory tests.
The question is not only how to cut the volume of such tests but which ones.
In an ADLM session last July, Grace Mahowald, MD, PhD, of Massachusetts General Hospital, explained why her team at MGH zeroed in on the CBC with differential (see December issue, https://bit.ly/CT_1225_CBC). Ronald Jackups, MD, PhD, of Washington University in St. Louis, told the story of his lab’s magnesium testing intervention, which went live in March 2025.
“What we were looking for is that quick win to get buy-in for future efforts—that high-volume test but somewhat narrower use” than the CBC with differential. “For us, that turned out to be magnesium,” said Dr. Jackups, professor in the Department of Pathology and Immunology, Washington University in St. Louis School of Medicine, and associate chief medical information officer, BJC HealthCare.
The test is generally used to drive magnesium supplementation decisions in the ICU and in cardiac services, and the threshold is 2 mg/dL, with supplementation indicated if it’s under two. Dr. Jackups has read the studies that support that and “is not entirely convinced,” he said, but it is the desired threshold, noting it’s in the middle of the WashU reference range (1.4–2.5). ICU and cardiac service providers requested that the reference range be moved from 1.4 to 2 mg/dL so they will be alerted when a supplement is needed, but the laboratory declined.
One reason: A histogram of inpatient magnesium results “more or less recapitulates our reference range,” Dr. Jackups explained. Another: Half the patients are at less than 2 mg/dL. “Is it right to be supplementing half of all patients? It turns out it’s not,” he said. “They’re not supplementing all patients who are less than two, even though you would think that,” although “the chance for supplementation goes up once the patient is below two, so they seem to be using it as a threshold. But,” he added, “there are a lot of patients getting a lot of tests and it doesn’t appear they’re acting on it”—making it a “good target,” with data that could be used to convince stakeholders of the need for intervention.
The data pulled from Epic (by WashU colleague Mark Zaydman, MD, PhD) revealed that half of all magnesium orders were placed with an “in AM daily” frequency, and the top three order sets driving it were oncology and bone marrow transplant admissions and general adult admissions. “All are commonly used order sets, so it was quite a bit of a problem,” Dr. Jackups noted. It was easy to see why: “There’s a default, so all they have to do is go to that order set and press ‘OK’ and it will automatically give them the daily order.”
The clinical decision support intervention options consist of hard or soft stops, order questions, and noninterruptive alerts.
“Hard stops should be extremely rare and the last thing you do, in my opinion,” Dr. Jackups said. “You start with the less aggressive interventions, and if those fail, you can come back and say, ‘We tried it. It failed. Can we do something tougher?’” That gets more buy-in for the hard stops, he added. When to use one? “For things that are really important, to us or to leadership.” An example: drug orders. “If you order a dose that is going to kill any patient, that’s a hard stop.” For lab orders, a hard stop is used only for Clostridioides difficile testing, stemming from concern about asymptomatic carriage of C. difficile leading to reimbursement penalties that could add up to millions of dollars. “That’s an easy case to make for a hard stop,” he said.