Applying human factors engineering to the chaotic and intense environment of the emergency department (ED) shows promise in reigning in routine overprescribing of antibiotics, said Michael Pulia, MD, MS, an ED physician and researcher. Pulia is the director of the emergency medicine antibiotic stewardship research program at the University of Wisconsin in Madison. “Human factors engineering is not something we really talk about in medicine very often, and it was definitely not incorporated in my training to start thinking about these things,” he said at the IDWeek 2019 conference, held Oct. 2-6, 2019, in Washington, DC.

Broadly classified as ergonomics in much of the world, human factors engineering ultimately may lead to changes in practices and behaviors entrenched in healthcare that endanger patients with infections and other harms.

“Why do we have such a hard time moving the needle on antibiotic stewardship [in the ED], and more globally, why do we have such a problem with patient safety in healthcare as compared to other complex industries like the airlines?” Pulia said. “The one word that I would boil it down to is ‘design.’ We don’t think about things in healthcare from a design point of view and building better systems.”

Instead, healthcare is prone to so-called “fundamental attribution errors,” which is targeting individuals or a group of providers “who are doing the bad behavior — vs. just understanding human nature,” he said. “We [need to] design better systems around this to accommodate those behaviors. That is the different lens that systems engineers use.”

In general, human engineering design puts the person in the middle of a paradigm that extends out to all of their interactions and decision points in a complex system. To put it simply, it is “the study of factors that make work easy or hard,” Pulia said. A good overview of the approach as it applies to antibiotic stewardship in ambulatory care is a 2018 review article1 by a different research team, he recommended.

Downstream Effects

A human factors assessment must consider the effect on patients and healthcare workers. A design change for patient well-being that increases clinician burnout is self-defeating.

“If we don’t design good interventions, we have adverse effects in other areas,” Pulia said. “Even if we change the behavior we were looking for, there can be downstream impacts that are not good.”

An emergency physician, Pulia offered humorous advice to anyone contemplating taking on antibiotic stewardship in the ED: “First of all, good luck to you. I think this is where the term ‘herding cats’ came from.”

Indeed, by the measures of human subject behavior, the ED is one of the most rapid-decision, high-consequence systems imaginable.

“It is high-pressure, high-density decision making,” he said. “It has the most decision-making density than any other field of medicine, and some would argue in any occupation known to mankind. There are more decisions per minute in this setting.”

The general perception is that emergency clinicians use antibiotics liberally because they are not necessarily involved in patient follow-up and they want to “hedge their bets,” Pulia said. That contention is vastly oversimplified, as human behavior research shows.

“ED overprescribing of antibiotics is actually quite complicated,” he said. “There are a lot of different factors and it is probably even specific to local institutions in some regard.”

Barriers to system change include staff burnout, turnover, lack of equipment and IT support, and weak leadership structure, he says, describing the ED as the “juxtaposition of chaos with the need for judicious and expeditious thinking.” Thus, rigorous design principles, with some notable exceptions, really have not been applied to ED settings, Pulia said.

Blank Slate

If you seek to change behavior in the ED — antibiotic stewardship or otherwise — it is best to begin with a “blank state,” he said, underscoring the point by showing a blank slide.

“Forget what you know or think that you know about even your own discipline or area,” he says. “You have your own clinical biases [formed by] a lifetime of practice, education, your mentors and teachers.”

Pulia said he was fortunate to have non-clinical human factors engineers working with him on the stewardship research.

“They ask the basic questions that we sort of gloss over, like, ‘Why are [practitioners] doing it that way?’” he says. “They see things that we don’t see.”

Try to partner with engineers if possible, but if not, begin with an open mind when you look at processes and patterns in the ED, he said.

“We had human factors engineers come to our emergency department and shadow our clinicians,” he said. “They documented every part of the care process like they were exploring an alien world for the first time. They are incredibly meticulous in figuring out the care process and all the things involved.”

Another aspect of the research is to have confidential conversations with emergency providers. “I am usually not involved with that because as an emergency physician, I am biased,” Pulia said. “Research staff and other non-clinicians interview these people because they are not introducing bias.”

Clinicians talk to them anonymously about work problems, then they try to address common themes through design change. In research, Pulia studies these changes and then continues to get feedback for any necessary redesigns in a rigorous, iterative process.

“It has to be really informed by the frontline providers — the people using it — and by systematic observation,” he said.

Too often ED guidelines and recommendations fail to consider issues such as diagnostic uncertainty, concern over access to care and follow-up, and patient expectations. Resource and environmental factors include available tests and diagnostics, crowding, and time pressures.

“They are actually quite concerned about patients when they leave, [concerned that] they are going to fall through the cracks,” he said. “There are quite a bit of drivers of antibiotic prescribing.”

The key is to focus change on the most “salient” barriers, which may require a high level of specificity to identify.

“What exactly are you trying to target — is this a pneumonia thing, a UTI [urinary tract infection], or skin and soft tissue?” he said. “What is the professional culture — the people that are working there?

Studies show the “uptake of an intervention is drastically different at various sites. That implies that the level of local barriers to implementation were not addressed,” he added.

Over-Testing Equals Over-Treatment

In a project to reduce over-testing and unnecessary treatment, Pulia and colleagues specifically looked at UTI issues in the ED.

Research has shown that, in general, older adults with nonspecific symptoms (i.e., confused mental state) frequently get a urinary analysis (UA), are diagnosed with asymptomatic bacteriuria, and then are put on unnecessary antibiotics.

Indeed, there is a “perverse incentive” in EDs to get some kind of diagnosis, and in an elderly patient that often ends being an asymptomatic bacteriuria that gets treated with antibiotics only needed for a UTI.

“And it is often an education piece, too, that they don’t know that an [asymptomatic bacteriuria] should not be treated,” Pulia said. “But there is this perverse incentive — if you label something a UTI, the tracks are just greased for you. [Start some antibiotics] and we’re good to go — no pushback. That’s really what is going on.”

Educating emergency clinicians about this problem will help, but that is not going to be as effective as “designing something that truly accounts for that behavior.”

An unidentified community hospital taking part in the research developed a simple but ingenious solution, Pulia said.

“They developed a protocol where an adult with these vague symptoms and a positive UA can get admitted without antibiotics — no questions asked,” he said. “They could watch as long as they were clinically stable, and they wouldn’t feel that pressure to give them antibiotics.”

Though the protocol used at a small hospital — linked to one long-term care facility — may not be applicable to larger healthcare institutions, it clarifies that patients can be admitted without administration of antibiotics for 24-48 hours “while the hospitalist evaluates for more likely causes of altered mental status in this population.”

If patient is still in an altered state after 48 hours — with no other explanation of the symptoms, the policy calls for beginning antibiotic treatment for UTI.

“There is a vast amount of over-testing or urine analysis in older adults,” Pulia said, adding that he noticed at his own facility that virtually every UA from the emergency department included an order for a reflex culture.

“On every single patient — we found a tremendous amount of over-culturing going on, and basically eliminated that order from the ED,” he said. “We saw a 40% decline in cultures out of the ED.”

A Problem Arises

These findings are not yet published, but Pulia revealed a problem in removing the routine order for culture: Some cultures that really should be done fell by the wayside.

“So, I’m getting pressure from one side, knowing this is vastly overused, and the pressure from the [infectious disease physicians] saying we need cultures,” he said.

The reflex culture order was revised and restored after getting buy-in from urologists, infectious disease clinicians, and others, he said. “There were some process instructions but basically we said, ‘Do not send this test for anything other than suspecting a UTI, and you have an appropriate indication.’”

Under the new policy, the following were not considered indications for a UA with reflex culture:

• Abnormal urine quality, such as color and order, without another indication;

• Abdominal pain without suspicion for UTI;

• Routine component of “pan-culture”;

• Screening for UA for patients with nonspecific complaints, such as fatigue, weakness, or falls.

The idea is to eliminate unnecessary testing while not missing cultures that need to be done. The work is still in process, but a 20% reduction in treating asymptomatic bacteriuria has been achieved thus far, Pulia said.

When trying to make design changes, you have to make sure they are integrated into the system — specific actions, not just a link to hospital policy, he said. The information must be accurate, and it must get to the right people in the right format, for example, text, computer, or electronic medical records.

In addition, the design change must kick-in at the precise point of care, not too early or too late. Expect challenges. “In systems engineering, sometimes building a better mousetrap makes a smarter mouse,” he said. “Human beings are quite interesting in their ability to overwhelm really well-designed systems. You don’t want to get to a point where you are doing a lot of hard stops and you are making work hard for everybody else because of an outlier.”

If the behavior change needs to be made with an outlier, it is probably better to try to do that face-to-face rather than redesigning a system that is working for everyone else, he said.


  1. Keller SC, Tamma PD, Cosgrove SE, et al. Ambulatory antibiotic stewardship through a human factors engineering approach: A systematic review. J Am Board Fam Med 2018;31:417-430.