Electronic alert systems that warn providers that a patient has acute kidney injury (AKI) have no effect on AKI progression, the need for dialysis, or death, and there is some evidence the alerts could even lead to worse outcomes, a new multicenter, randomized trial indicates.
“We had good preliminary data to show that documentation of AKI was fairly low across a variety of health systems and we also saw that best practices in the face of AKI were not being completed — even simple diagnostic tests like urine analyses which are universally recommended were not getting done — so it was common sense to us that an alert might move these metrics and improve clinical outcomes,” lead author Perry Wilson, MD, told Medscape Medical News.
“But after our study, the safest conclusion is that we can’t provide any evidence that these alerts are beneficial,” added Wilson in a statement from his institution, Yale University School of Medicine, New Haven, Connecticut.
The research was published online January 16 in the British Medical Journal.
Alerts Seemed to Increase Death Rates in Nonteaching Hospitals
In total, 6030 patients were recruited from six hospitals — four teaching and two nonteaching hospitals — in the Yale New Haven Health System in Connecticut and Rhode Island between March 2018 and December 2019. The median age of patients was 71 years, almost half were women, and 16% were Black. Almost one third of patients were in the intensive care unit at time of randomization.
Providers received a pop-up alert indicating the presence of AKI or the alert was switched off and patients received usual care.
“The primary outcome was a composite of inpatient AKI progression (defined as an increase in AKI stage), receipt of dialysis, or death within 14 days of randomization,” the investigators explain.
Results showed that the primary outcome occurred in 21.3% of patients randomized to an AKI alert versus 20.9% of patients receiving usual care.
In a post-hoc analysis, Wilson and colleagues also found that the alerts appeared to be associated with harm at the two nonteaching hospitals included in the study.
In these two hospitals, the risk of AKI progression, need for dialysis, or death within 14 days of randomization was almost 50% higher in alert care patients than it was in the usual care group.
Specifically, 24.3% of patients in the alert care group met the primary outcome compared with 16.3% of the usual care group, at a relative risk of 1.49 (P = .006), the authors report.
This difference in patient outcomes appeared to be primarily driven by mortality, as death rates were significantly higher among patients receiving an alert, at 15.6%, compared with 8.6% in the usual care group (P = .003).
In contrast, no such effect was identified at the four teaching hospitals, the authors point out.
Patients randomized to AKI alerts were more likely to receive an order for intravenous fluids, to undergo urinalysis, to have their serum creatinine measured within 24 hours of randomization, and to have their AKI documented than those in the usual care group, but these interventions did not account for the difference in clinical outcomes, the investigators say.
No Single Explanation for Finding
Asked why the alerts used in the nonteaching hospitals seemed to be associated with harm, Wilson said they analyzed this specifically but couldn’t find any explanation.
“If we had seen the opposite result, if we’d seen that the alerts reduced death, we probably would not have found a specific thing that the alert triggered to save lives, we’d just say the care was better in general,” he explained.
“I think the opposite may hold true here,” he added.
“For different patients, different decisions were probably made that shouldn’t have been made. An alert does have an effect on a provider who thinks, ‘I’d better do something about this.’ And that is not always right — sometimes the best thing in medicine is not to do anything,” he stressed.
Alerts Should Give Specific Instructions
Wilson believes the best practice may be for AKI alerts to carry specific instructions for providers.
Therefore in a current, ongoing trial, Wilson and colleagues are again using alerts to identify patients with AKI, but in this study, the alert reminds the provider, for example, that not only does this patient have AKI, but that they are also on a nephrotoxic medication and the provider may want to think about stopping it.
“So instead of an alert simply saying, ‘Your patient has AKI,’ we need to give them [providers] more structure [and explain]…what those best practices are on an individual patient level,” he said.
In the meantime, Wilson and colleagues feel that any center currently using these alerts could be justified in discontinuing them, particularly in the absence of any compelling evidence that they benefit patients — “less might be more,” as he suggested.
Wilson is a Medscape contributor. He has reported reported no relevant financial relationships.
BMJ. Published online January 16, 2021. Abstract