Behavioral Nudges in EHR Workflows May Reduce Healthcare Waste

Behavioral nudges within EHR workflows can help reduce erroneous decision-making and mitigate healthcare waste, according to a study published in JAMIA.

Researchers studied three clinical workflows at an academic medical center where the existing choice architecture, or the EHR design, was potentially nudging providers toward erroneous decisions, waste, and misuse.

The researchers changed the architecture to nudge providers toward better practice and found success to varying degrees.

“By changing the direction of these nudges—in one case, via making the less appropriate order more difficult to find and use; in the second case, by making the more frequently desired imaging easier to find; and in the final case, by presenting an easy to find alternative—we attempted to nudge providers toward reduced waste and misuse,” the study authors wrote.

The first nudge involved ordering free phenytoin levels, an expensive send-out test that leads to delays in patient care in the inpatient setting. The researchers explained that a total phenytoin level is appropriate in most circumstances and is available quickly.

However, the EHR alphabetically presented “free phenytoin” before “total phenytoin” to providers searching for “phenytoin level,” which could have influenced them to choose the more expensive and time-consuming test.

Researchers removed the ability to order free phenytoin independently. In its place, they created an order panel presented to a provider searching for “phenytoin” or “free phenytoin.”

“In this order panel, an explanation is given of the rare circumstances in which a free phenytoin level is appropriate and includes both total phenytoin, which is ‘pre-checked’ and free phenytoin, which is defaulted to ‘unchecked,’ nudging the provider toward the order that is almost always correct,” the authors wrote.

Another challenge that wasted clinician time and delayed patient care was ordering a CT abdomen when the provider wanted a CT abdomen/pelvis study.

Researchers hypothesized this was because the EHR presented CT abdomen alphabetically before CT abdomen/pelvis.

They changed the computerized provider order entry for “CT abdomen/pelvis” to “CT abdomen /pelvis.” The extra space in front of the slash displayed “CT abdomen /pelvis” at the top of the imaging list, nudging the provider to select the first order in the list.

The last workflow the researchers examined was related to the prescription of benzodiazepines for procedural anxiety. Originally, the default quantity for such orders in the EHR was the same for patients taking these medications on a routine basis for chronic disease. Hence, the system nudged providers to prescribe far more benzodiazepine pills than necessary.

The researchers created a new order called “Lorazepam (Ativan) tablet 0.5 mg for imaging/procedure” that appears in the ambulatory orders preference list for all ambulatory providers when they search for lorazepam (or ativan).

“This order defaults to a quantity of 2 tablets with zero refills, as well as a default PRN comment ‘for anxiety (prior to imaging study or procedure),’ nudging providers to prescribe the more appropriate quantity,” the authors said.

“The final change was successful in terms of nudging providers ordering sedation for Magnetic Resonance Imaging (MRI) or procedures to order two or fewer pills of low-dose lorazepam; however, this intervention improved from a very low baseline, and the ‘correct’ amount of nudging is not clear,” they added.

The researchers emphasized that healthcare organizations must balance the benefit of these interventions against the costs of their implementation.

“In the case of phenytoin, our nudge consumed an estimated six hours of institutional time to implement, inclusive of the time of all those involved to discuss (via email) the change with stakeholders and our EHR analysts, the time for the change to be built and test the change, and the time to enable it in the EHR. In the case of the CT orders, the total time involved was estimated at 3 hours; in the case of the lorazepam orders, 16 hours.”