Using swarm intelligence to boost the root cause analysis process and enhance patient safety
In an effort to strengthen patient safety, leadership at the University of Kentucky HealthCare (UKHC) decided to replace its traditional approach to root cause analysis (RCA) with a process based on swarm intelligence, a concept borrowed from other industries. Under this process, when a problem or error is identified, staff quickly hold a swarm — a meeting in which all those involved in the incident or problem quickly evaluate why the issue occurred and identify potential solutions for implementation.
- A pillar of the swarm concept is a mandate that there be no punishments or finger-pointing during the swarms. The idea is to encourage staff to be forthcoming to achieve effective solutions.
- Typically, swarms last for one hour and result in action plans designed to correct problems or deficiencies within a specific period of time.
- The ED was one of the first areas where UKHC applied swarms. For example, hospital administrators note that the approach has been used to address issues involving patient flow, triage protocols, assessments, overcrowding, and boarding.
- After seven years, incident reporting at UKHC has increased by 52%, and the health system has experienced a 37% decrease in the observed-to-expected mortality ratio.
The idea of delving deeply into an error so that you can figure out what went wrong makes perfect sense to most people. But in the healthcare environment, such a process, often referred to as root cause analysis (RCA), can easily bog down in political, legal, and bureaucratic concerns. For instance, when frontline providers are not forthcoming about problems or issues because they fear punitive repercussions, it is much harder to see the truth, let alone devise solutions that will actually improve patient safety. Similarly, if an RCA process is designed to meet a requirement rather than to fix problems, it is easy to see how the results may be off the mark or ignored.
With these issues in mind, in 2009 the University of Kentucky HealthCare (UKHC) decided to replace its traditional RCA process with a new approach that relies on swarm intelligence, a concept borrowed from other industries, to quickly assess problems in a blame-free environment and identify fixes that potentially stand a better chance of correcting or improving flawed processes.
Hospital administrators acknowledge that it took time for the process to work, but after seven years, the results are impressive. For example, investigators at UKHC note that incident reporting has increased by 52% and the health system has experienced a 37% decrease in the observed-to-expected mortality ratio.1 Furthermore, it’s an approach that UKHC administrators believe could well deliver dividends at other health systems.
Don’t play blame game
Paula Holbrook, RN, BHS, JD, CPHRM, associate general counsel and associate director of risk management at UKHC, recalls how “difficult and unwieldy” the RCA process was at UKHC.
“It was basically reserved mostly for sentinel events. It consisted of one person in quality interviewing people and making notes, going through this huge, onerous … unintelligible framework that just didn’t work,” she says. “It was a boondoggle.”
Holbrook credits Paul DePriest, MD, who is now the executive vice president and chief operating officer at Baptist Memorial Health Care in Memphis, TN, but at the time was the chief medical officer at UKHC, for recognizing the need for reform and spearheading the effort to rid the RCA process of blame and finger-pointing. However, Holbrook acknowledges that it was an uphill battle.
“One of the early challenges was trying to explain what [swarming] was,” she recalls. “Trying to introduce a new concept into an old culture is difficult.”
During swarming, when something goes wrong, the people involved come together as quickly as possible after the problem becomes apparent to figure out a solution, explains Mark Williams, MD, FACP, MHM, director of the Center for Health Services Research at UKHC.
“It leverages acute memory and knowledge about what is going on versus looking at the problem much later when memories fade and they begin to reinterpret their memories based on other activities that are going on or comments by other people,” he says.
However, everyone recognized from the start that the approach would not be successful if the people called on to discuss and solve problems feared they were being pulled into a swarm to be punished, Williams adds. “We heard very clearly from the staff [involved when the swarm approach was first being implemented] that Paul DePriest would step in if there was any attempt by others to point fingers or assign blame,” he says.
As time went on, other leaders fulfilled this role. Trained swarm facilitators also made sure to protect frontline staff members from intimidation or punishment. “That happened so repeatedly that people began to become increasingly comfortable, and they recognized the effectiveness of these swarms,” Williams notes.
While reformers started small, they focused on important issues, Holbrook recalls.
“The first reforms had to do with unrecognized clinical decline and chaotic codes, particularly in the ED,” she explains. “We had to proceed slowly, carefully, and reliably in order to establish some credibility to the process.”
In 2009, the first year of swarming, the health system only completed 22 swarms, but the number of swarms more than tripled in the second year, and by the third year nearly 170 swarms were completed, Holbrook says. She adds that in 2014, the health system completed more than 300 swarms, and she is still tallying the number of 2015 swarms.
In the ED, for example, the approach has been applied to patient flow, triage protocols, assessments, overcrowding, and boarding, Holbrook says.
“This swarm process applies to the very basics of healthcare, and what we have found in the course of swarming … is that often the basics are lacking,” she says.
For instance, assessment, escalation, communication, and handoffs are critical steps, Holbrook notes.
“The nuts and bolts of care, such as taking vital signs and communicating those or obtaining an accurate weight [on a patient],” are very important, she explains.
“We overcame the challenges by explaining [the concept] and reinforcing it at every swarm. We provided feedback to people, and people learned it well,” Holbrook says. “You have the right people in the right place, you hear the stories, and you hear the perspectives of people who were involved: The people in authority, accountability, and responsibility for that unit, service, or department.”
The insight and understanding gained from listening to different viewpoints on a problem or error reduces finger-pointing and blame, Holbrook adds.
“That shared perspective, even if it is different, is helpful,” she says. “As we did more of these, we found a way to develop a more standard process in terms of documentation and communication of the findings … and doing this in a blame-free environment really helped. People are more honest.”
Holbrook observes that one standardized form that is particularly helpful is a sign-in sheet that reminds all swarm participants that their discussions are privileged under federal law.
“As an attorney, I want to make sure that oral and written communications that are part of the swarm process are protected and privileged so that people have a safe place to be open and transparent,” she says. “We very zealously guard our discussions.”
The commitment from top leadership was key to the successful implementation of the swarm process, adds Jing Li, MD, MS, administrative director and an assistant professor at the Center for Health Services Research at UKHC.
“We needed the leadership to set the tone, and then for everyone else to get the same message,” she says. “Another [key] strategy at that time was that there was a designated group to start the process.”
Add structure to the approach
Over the last seven years, UKHC administrators and staff have continued to refine the swarming process. When problems arise, there are two types of swarms to correct problems or improve processes.
“There might be a local unit-level swarm when an issue doesn’t have system-level implications. In that instance, people will deal with the issue right there by pulling their group together, led by a unit manager or local area staff,” Williams explains. “Then there are system-level swarms … which invariably involve a senior leader, such as the chief nursing officer, chief operating officer, or chief medical officer.”
In addition, like many health systems, UKHC has an incidence reporting system. Staff review reported issues on a weekly basis according to established criteria to determine which issues or problems should be swarmed at the system level.
Once the decision is made to hold a swarm, key staff members who were involved in the incident or issue are contacted, and someone from the patient safety department works to coordinate attendance.
“The [swarms] are routinely scheduled at three particular times, either at noon, 7 a.m., or at 4 p.m. or 5 p.m. so that other issues don’t interfere,” Williams says. “Also, since this is a priority for the organization, managers know that if one of their frontline staff [members] is invited to attend a swarm, it is their responsibility to support their staff being there.”
In fact, typically, when frontline staff members are invited to a swarm, their managers also attend the meeting to support their personnel, even if they were not involved in the episode under discussion, Williams adds.
“[Similarly], when residents are invited to a swarm, the residency program director for that particular specialty is invited to the swarm to support the resident,” he says.
Swarms typically take place in a neutral location where participants have access to the health system’s electronic medical record and a white board to document findings or suggestions. The meetings are designed to last for 1 hour, and involve a review of all the particulars that led to the swarm, an investigation of any underlying systems factors, and the identification of specific areas in need of improvement, which are formulated into an action plan. In addition, specific individuals are tasked with carrying out each of the identified action steps, most commonly within 60 days.
“After this one-hour session, there is still a follow-up mechanism in place … and there will be one person as the key point of contact for the action items,” Li notes. Further, the health system is now developing a database of all swarms that have taken place so that administrators can quickly identify any swarms that are recurring, indicating that the issue has not been solved.
After several successful years of swarming, UKHC is developing a standardized, instructional approach so that other organizations can take advantage of lessons learned. In addition, Holbrook has already begun to visit other hospitals to help train their personnel to implement the process and offer advice to organizations that are interested in traveling a similar path.
Holbrook stresses that senior leadership must lead and promote swarming initiatives.
“Make sure that you have credible people [taking charge] of the effort on an ongoing basis,” she says. “Don’t relegate the responsibility to a manager. Make sure that the initiative has support from the top down and that you have engagement from leadership.”
- Li J, et al. SWARMing to improve patient care: A novel approach to root cause analysis. Jt Comm J Qual Patient Saf 2015;41:494-501.
- Paula Holbrook, RN, BHS, JD, CPHRM, Associate General Counsel, Associate Director of Risk Management, University of Kentucky HealthCare, Lexington, KY. Email: [email protected]
- Jing Li, MD, MS, Administrative Director and Assistant Professor, Center for Health Services Research, University of Kentucky HealthCare, Lexington, KY. Email: [email protected]
- Mark Williams, MD, FACP, MHM, Director, Center for Health Services Research, University of Kentucky HealthCare, Lexington, KY. Email: [email protected]
The key to the swarming approach is a blame-free environment and rock-solid support from hospital leadership.
Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.