Re-designing Medical Alerts for Nurses at NYC H+H
by Federica Bologna
The Challenge
NYC Health + Hospitals (NYC H+H) is the largest municipal network of healthcare facilities in the United States, comprising 11 hospitals, 30 health centers, and 5 long-term care centers. In 2019, NYC H+H transitioned to Epic as their provider for electronic health records (EHRs), and with this new system, came BestPractice Advisories (BPAs).
BPAs are alerts to improve clinical care that remind medical professionals of drug interactions, lab results, and referrals, and are typically embedded in electronic health records. For instance, if your provider writes you a prescription for a medication, a notification will pop-up to remind them of any interactions between that medication and your current treatment plan.
When carefully implemented, BPAs can improve clinical care and workflows. However, they can also be unintuitive to use or can appear too often and at inconvenient times, causing providers alert fatigue and desensitization.
In an effort to reclaim provider capacity, increase patient safety, and reduce alert fatigue, a team of health informatics experts was established at NYC H+H in 2020. This working group routinely assesses new and past BPAs, and modifies or removes any that are found to be unhelpful.
Over this summer, I supported these efforts as a Siegel Family Endowment PiTech PhD Impact Fellow. I studied the impact of the BPA working group on relieving nurses’ fatigue, and provided design recommendations based on nurses’ workflows.
Motivation
Focusing on BPAs that affect nurses is particularly important as there is a lack of studies dedicated to the evaluation and improvement of BPAs for nursing. For one, BPAs are rarely created specifically with nurses in mind, let alone created through rapid cycles of designing and testing with nurses’ feedback and involvement. Lastly, nursing BPAs are less frequently evaluated in terms of patient outcomes and usability.
Furthermore, at NYC H+H, nursing BPAs have always been the most popular type of BPA, comprising around 40% of all BPAs.
The Impact of the BPA Working Group
To begin my exploration of the BPA working group’s effectiveness at addressing nurses’ alert fatigue, I selected three of the most frequent nursing BPAs:
BASE IP MED DOSE GIVEN TOO CLOSE TO PREVIOUS DOSE MAR [30400001]: this alert informs healthcare professionals of when the current administration of a medication is too close to any previous administration.
BASE HHC C-SSRS SUICIDE RISK SCREENING NEEDS COMPLETION - NURSING [2301]: this alert is meant to evaluate the mental wellbeing of the patients.
HHC IP BH TREATMENT PLAN GOAL/RESTRAINT ORDER/DOCUMENTATION BPA BASE [2088]: this alert reminds nurses of whether they need to add a restraint order to the current treatment plan.
I then analyzed changes over time in: their alert rates (how often they pop-up) and their action rates (how often nurses carry out the action recommended by the alert). Thanks to the efforts of the BPA working group, all three alerts have seen decreased alert rates and two saw higher action rates between 2019 and 2023.
To gather a better sense of the impact of these changes on nurses’ fatigue, I interviewed six nurses across the Emergency Department (ED), Telemetry, Surgery, and Intensive Care Unit (ICU). All nurses understood the importance and utility of the medication alert (30400001) for preventing medical errors, however, they described it as one of the most incessant and inconveniently timed BPAs. Nurses in ED specifically felt that the BPA didn’t apply to their work, as they are required to use high risk medications due to the critical condition of their patients, and are already aware of the risks of doing so. Similarly, one nurse found that the mental health alert (2301) would fire even when the condition of the patient made it impossible to assess their mental health. However, in the case of alert 2088, two nurses agreed that it was helpful and didn’t appear too frequently. The interviews left me with the impression that, the integration of BPAs in nurses’ workflows could be further improved by following a few design recommendations.
Designing BPAs for Nurses’ Workflows
To provide helpful recommendations for the improvement of BPAs’ integration, I leveraged results from the six interviews mentioned above, as well as from an observation study I conducted in the ED over the course of a day. Here is an overview of both the results and my recommendations.
In terms of alert rate, my recommendation was to review the inclusion and exclusion criteria of frequent alerts (the rules determining when to fire or not fire an alert) routinely to prevent repeated firings when e.g. nurses complete the requested task, but do so outside the alert window, or have already provide a valid reason for not fulfilling the task.
With regards to the timing of the BPAs, it is fundamental to stop alerts from firing as soon as nurses open a patient’s chart to document their work after a round of check-ins, as this timing is unhelpful. In contrast, nurses appreciate alerts that appear when they are admitting new patients, as they streamline the admission process, or those at the end of their shift, as they help nurses take care of anything they might have forgotten.
Lastly, in terms of BPA appearance, nurses have suggested redesigning the alerts so they can be moved to the side or minimized in a corner of the screen. This way, if nurses are in the middle of a task, they can postpone addressing the alert rather than waiting for the alert to be fired again.
Impact and Path Forward
At the end of this summer, I provided the BPA working group with two deliverables:
Evidence of the BPA group’s effectiveness at relieving nurses’ alert fatigue;
Design recommendations to improve the integration of BPAs in nurses’ workflows;
I am confident that with this added knowledge, the BPA working group will continue its important efforts to improve nursing BPAs at NYC H+H. Indeed, the BPA working group has already passed a policy that imposes to routinely modify all alerts that are fired when opening a patient’s chart.