NEW YORK – Overtesting is a significant issue for hospitals and clinical labs, leading to wasted resources and in some cases an increased risk of patient harm.
One strategy institutions have adopted to curb this practice is the use of "once-per-visit" alerts — notices that are automatically triggered when a doctor reorders a test for a patient within a time period where it is unlikely to be clinically useful.
A recent Clinical Chemistry study by physicians at St. Louis's Barnes-Jewish Hospital, however, found that compliance with these alerts is relatively poor, with staff overriding them roughly half the time.
This suggests substantial room for improvement as hospital systems work to address overtesting while not unduly inconveniencing doctors and other staff, said Ronald Jackups, senior author on the study and an associate professor of pathology and immunology at Washington University School of Medicine as well as co-medical director of the Barnes-Jewish hematology lab.
While Jackups said that the exact amount of lab overtesting is hard to quantify, the Clinical Chemistry study cites estimates suggesting as many as two-thirds of ordered tests "do not contribute to patient care."
This, he noted, contributes to a waste of lab resources and can negatively impact patient care by increasing the chance of false positive findings or even depleting their blood to the point where transfusions can be required.
Jackups said that most hospital electronic ordering systems have duplicate test alerts, but these are typically meant to flag doctors ordering two of the same test by mistake, "for instance, on the same day or within hours of each other."
Once-per-visit alerts like those employed at Barnes-Jewish are meant to alert doctors that a test they are reordering is unlikely to change patient care if performed again within a certain time window. The authors provided the example of glycohemoglobin levels, which they noted "are governed by red blood cell turnover," and are unlikely to detect changes that would inform patient care if done within three months of the previous test.
Such once-per-visit alerts are still relatively rare, Jackups said, and largely confined to large academic medical centers, but the study findings suggest that even in this setting they are not as effective as they might be.
In their study, the researchers analyzed compliance with alerts set for 18 lab tests between September 2012 to October 2016, finding that the overall compliance rate was 54.5 percent.
Looking at the data broken down by personnel, attending physicians had the highest average compliance rate, at 62.8 percent, while nurses had the lowest, at 22.9 percent. While rates for non-attending physicians and nurse practitioners were not given in the paper, Jackups said they were just below the rates for attending physicians.
"What you see is that people [physicians and nurse practitioners] who are authorized to order tests on their own had a much higher compliance than the one group [nurses] who didn't have the authority to order tests on their own," he said, adding that this suggests a dynamic where nurses who encounter an alert when placing an order for a physician or nurse practitioner may be reluctant to take the time to follow up with them about how to proceed and are likely to just override the alert.
"I think the biggest lesson we learned from this study is that the person who directly interacts with the alert may not be the person who is actually responsible for the test request," Jackups said. "I think what we need to do is revise the alerts so that when a nurse gets the alert there is some outlet — for example, to have them identify the physician who is requesting the test so that we can go back to that physician and ask them why they think this [test] was important."
Another observation was that compliance peaked around day four of a patient's stay in the hospital. Jackups noted that while it was not surprising to see compliance decline as a patient's stay extended beyond this mark, it was striking that compliance was also lower in the first days of a patient's stay.
He suggested this could be due to the high volume of tests ordered initially as doctors try to get a handle on a new patient's condition.
"On that first day of admission, there are lots of tests being ordered," Jackups said. "And if [doctors] don't see the results of a test, they may be inclined to reorder it just to be on the safe side."
He noted that the Barnes-Jewish ordering system allows doctors to check where in the process a given test is but said that accessing this information typically requires clicking through to a different screen.
"Doing that once doesn't seem like a huge burden, but if you are trying to keep track of a dozen tests on a dozen patients, it becomes much more of a burden," he said. "Busy clinicians can be tempted to take the easier route of just ordering the test again."
One approach that would likely boost compliance significantly would be to implement what Jackups called a "hard stop," in which doctors would have to make a call to the hospital laboratory to override a retesting alert. This was explored several years ago by researchers at Rijnstate Hospital in the Netherlands, who found that their alerts were overridden only 18 times per year during the three years of the study.
However, Jackups said that implementing this sort a hard stop would require extremely solid evidence that reordering of specific tests was unnecessary and could cause patient harm to secure physician buy-in.
He noted that during the course of the study, Barnes-Jewish did decide to change its alert for Clostridium difficile testing to a hard stop based on the fact that overordering of that test was clearly linked to overtreatment of patients.
"The issue with that was not just the waste from overdoing the test, it's that there are a lot of people who may have that bacteria but not actually be" harmed by it, he said. "But if you are identifying those patients, then you have to treat them or you have to keep them in the hospital longer than they should be. And that is both a risk of harm to the patient as well as a huge waste of resources."
Even implementing "soft-stop" alerts like those evaluated in the Clinical Chemistry study requires careful attention to winning over physicians.
"You have to put in the alerts very carefully and only include ones that are really well defended by the [scientific] literature," Jackups said, noting that Barnes-Jewish originally placed alerts on a set of five tests and then built to its current list of 18 as it established the usefulness of the system.
He said he and his colleagues are planning to add to the list.
"There are a lot of tests that we know really don't need to be duplicated," he said. "It's definitely our plan to expand it."