How to overcome operational inefficiencies in the ED using AI and machine learning

Ayla Ellison -

Hospital operational inefficiencies can be difficult to overcome and can have a negative effect on employee satisfaction and patient experience. However, with the right data and tools, hospitals can take a proactive approach to defeating these issues.

 

To overcome operational inefficiencies, such as long wait times in the emergency department, innovative hospital leaders are looking to real-time platforms that reveal trends and factors that contribute to such issues. While these solutions can provide healthcare organizations with retrospective insight for understanding past trends, hospitals and health systems are using them to act in real-time to improve efficiency, meaningfully impact the patient and clinician experience, and sustain results.  

How Mercy Fort Smith transformed its ED
Before being named COO of Mercy Hospital Springfield (Mo.) last year, Brent Hubbard served as COO of Mercy Fort Smith (Ark.), a role he began in 2013. "When I joined Fort Smith, we [encountered] several challenges," said Mr. Hubbard during a March 22 webinar sponsored by Qventus and hosted by Becker's Hospital Review. He said patient satisfaction was suffering, the hospital's left-without-being-seen rate was too high, and its discharge length of stay and admitted length of stay exceeded national averages.  

To address these challenges, hospital leadership knew they had to start by fixing the root of the hospital's operational inefficiency issues — the emergency department. Mr. Hubbard said many of the hospital's challenges, including those in the ED, were attributable to only having access to retrospective data analysis. The hospital had data and analytics, but insights derived from such systems were not available in real time. "Every day we were being reactive rather than proactive," he said. "We were always looking in the rear-view mirror."

The hospital decided to partner with Qventus, which offers a software platform that uses artificial intelligence and machine learning to help hospital teams make better operational decisions in real-time. The Qventus platform enables hospital leaders to monitor data in real time, predict potential issues, identify course corrections and communicate them to frontline teams in the form of "nudges." The platform uses the hospital’s preferred form of communication, including text messages, to share those course corrections with decision-makers.

Within a few months of using the real-time solution, Mercy Fort Smith significantly improved patient satisfaction and achieved a 30 percent reduction in its left-without-being-seen rate, a 14 percent reduction in length of stay, a 20 percent reduction in door-to-doc time and a 40 percent reduction in unnecessary testing. The hospital produced these results while experiencing an 18 percent increase in patient volume.

Mr. Hubbard said the hospital has sustained the results since implementing the Qventus system and has rose in patient satisfaction from 29th of 33 hospitals in the Mercy system to number three. "I couldn't be prouder of what my team accomplished," he said.

How Mercy Ardmore improved ED throughput
Mercy Ardmore (Okla.) is a 190-bed hospital in rural Oklahoma that is also part of the St. Louis-based Mercy system. The hospital had high performance with many ED-specific metrics, but the hospital's nursing staff was frustrated with the transition process from ED to inpatient floor.

To assess the level of the hospital's ED throughput problem, Jennifer Bramlett, RN, director of nursing and emergency services at Mercy Hospital Ardmore, looked to CMS Core Measures disposition to admit data. The data revealed that the top 10 percent of hospitals move patients from ED to bed in 42 minutes. Mercy Hospital Ardmore's disposition to admit time was nearly 89 minutes, on average.

To overcome its inefficiencies, Mercy Hospital Ardmore partnered with Qventus. The hospital began using Qventus Mission Control to manage the ED in real time. "I can see when metrics are deviating from our goals," said Ms. Bramlett. "It highlights areas where we need to respond immediately and provides us insight into areas that are likely to be problems so we can get ahead of the issues." Ms. Bramlett said she can also monitor how the team is responding to system recommendations and how well they're collaborating to solve the problems.

The hospital set a goal of reducing disposition to admit time to 42 minutes and worked with Qventus to develop a play focused on reaching this target. When a patient waits more than 42 minutes, each team member receives a nudge via their phone, encouraging them to take action. Ms. Bramlett said the system provides her with real-time data she can use to understand what team or person is having issues and what day and time those issues typically occur. "I'm also able to see where we are being successful so we can replicate best practices," said Ms. Bramlett.

Ms. Bramlett said she was blown away with the results the hospital achieved. After using the system for five months, the hospital reduced its disposition to admit time by 10 percent. The hospital also experienced other ED improvements, including a 19 percent reduction in arrival-to-room time, a 55 percent reduction in the left-without-being-seen rate and a 17 percent reduction in door-to-doc time.

"I am passionate about improving our ED," said Ms. Bramlett. "I love that everyone is coming to the table to own their piece of the problem and solution so it raises ownership and accountability."

Operational improvements created better patient and staff experience
By applying AI and machine learning to the terabytes of data that Mercy hospitals were already generating, the care teams at Mercy Fort Smith and Mercy Ardmore where able to make better operational decisions in real-time. The result was improvement across all measurable performance metrics and improved cross functional communication.

Copyright © 2024 Becker's Healthcare. All Rights Reserved. Privacy Policy. Cookie Policy. Linking and Reprinting Policy.