On paper, doctors should know better — antibiotics treat only bacterial infections, and yet, physicians sometimes give them to patients who have viral infections. For patients, an unnecessary antibiotic can mean short-term side effects, like diarrhea, or more-persistent impacts, like microbiome disruption. But on a grand scale, the overuse and misuse of antibiotics pressure bacteria to gain resistance, the ability to thwart the drugs intended to kill them.

That can fuel the evolution of "superbugs" that evade most, if not all, antibiotics. In the worst-case scenario, this could contribute to tens of millions of extra deaths over the next 15 years, caused by illnesses that were once easily treated.

Given that antibiotic resistance is one of the world's leading public health threats, earlier this year, I went to Japan to investigate a program that has been remarkably effective at curbing the overuse and misuse of the drugs. I wanted to understand why doctors sometimes prescribe antibiotics when they're not needed and what approaches have been shown to improve their prescribing habits.

To answer those questions, I took a deep dive into the research on the topic and found the work of Julia Szymczak, a medical sociologist at the University of Utah School of Medicine, whose studies shed light on why doctors prescribe these medicines when they're not needed. I spoke with Szymczak about the complex social dynamics behind this behavior and whether there are reliable strategies for reining in antibiotic misuse.

Nicoletta Lanese: Could you explain the focus of your work?

Julia Szymczak: All of my work is really focused on two things. One, understanding why it is difficult for clinicians in real-world practice to use antibiotics the way that medical guidelines or evidence suggests they should be used. And then, more recently in my career, it's focused on developing interventions or strategies to help clinicians apply evidence that's informed by all that work.

I think about the decision-making about how an antibiotic is used as not simply a decision that is about pathophysiology or microbiology — it's about social dynamics. Clinicians are sensitive to a lot of other features in the care delivery environment beyond what they know to be true about antibiotics, what they know to be true or apparent about the potential infection that a patient has.

NL: What are some factors that shape that dynamic?

JS: Diagnostic uncertainty is a major challenge for clinicians. Differentiating viral versus bacterial is not [straightforward] — you don't have a slam-dunk perfect test. There are attempts to develop things to help, but the diagnostic uncertainty piece is really challenging.

Then there's the organizational characteristics around clinician decision-making, which is that everybody is incredibly time pressured, and so decision-making about antibiotics happens very quickly.

In the ambulatory or the outpatient setting, where the vast majority of human antibiotic use occurs, one of the more common themes that you will hear when you talk to clinicians is that patients often want antibiotics that are not needed. That relationship is more complicated than it appears on its face, but that is a major pressure point for clinicians.

NL: Are there other pressures that are unique to the outpatient setting, where most antibiotics are used?

JS: The major one is time pressure. I had a pediatrician who said they had — I can't remember the figure, but it was like 800 seconds for a sick visit. They broke it down into seconds. Their experience of time in the outpatient setting is so intense. Certainly clinicians in the inpatient setting [hospitals] feel time pressure, but the decision-making is distributed over an admission, which still might only be two days, but two days is different than literally five minutes.

The other thing is your interaction with that patient. That clinical encounter is very transactional, particularly in the United States, particularly for those clinicians who work in, for example, telemedicine, which is a whole other context but has similar features to urgent care or sick visits. This idea that "I'm trying to provide you with something of value" [is a big factor]. That could be a proper diagnosis. That could be the provision of a prescription. It could be reassurance that you're going to be fine. In some scenarios, people are looking for information that they can share with their employer.

Someone is coming to you to get something for a problem. Oftentimes, your assumption is that what they're coming to you for is an antibiotic. The encounter is already shaped by the patient's expectation — or your [the doctor's] expectation of the patient's expectation. There's literature that shows that, in many scenarios, clinicians might perceive that a patient wants an antibiotic when the patient actually doesn't.

Oftentimes, clinicians will say that [when] somebody has what is very likely a viral infection and they don't need antibiotics, the act of explaining why they don't need antibiotics is very difficult, particularly if they seem to want them or if they've had multiple similar episodes and they've always gotten antibiotics in the past. That discussion, the literal conversation, is difficult. It takes time. It's draining.

Then, you're in an environment where there are competing priorities around how that patient is going to evaluate your care. If a patient is unhappy because you didn't give them an antibiotic and you're concerned about the patient-satisfaction score, which is being watched by your leadership, but no one's monitoring your antibiotic use, that could tip you into the prescription of an antibiotic that isn't needed.

Then, of course, there's also the fear of missing something. On the off chance the patient has an infection and it helps them, that staves off a whole bunch of other imagined or real bad scenarios down the line.

NL: You said it's often difficult for doctors to explain their reasoning around antibiotics. Do you think that's because the technicalities of resistance are hard to explain, or something else?

JS: I don't think it's necessarily that they aren't confident in the medical explanation. A paper of mine called "I Never Get Better Without an Antibiotic" goes through all the reasons why the discussion is difficult.

Briefly: The biomedical stuff is often not the hard part. What's difficult is countering a patient who you think has already made up their mind about what they need and convincing them that they don't need it. It involves not just the provision of microbiological facts but having to explain why their past diagnoses might not have been accurate or their previous clinicians didn't make a good decision. Or people might talk about their social network: "Well, so and so got antibiotics for that." And it's like, I'm not their doctor. I didn't see them. I'm making a decision about you.

There are social reasons why that discussion is just difficult, and then you throw that into the time pressure and potentially add in even the glimmer of antagonism or conflict, and people just don't want to go there because they're exhausted.

I don't think it's about the education, about the likelihood of this being viral and "antibiotics don't work for viral infections." It's a lot more countering beliefs that aren't necessarily accurate [such as antibiotics always being needed for certain symptoms] and dealing with social awkwardness.

NL: I feel like that breaks with the common stereotype of doctors being very cold, calculating and logical.

JS: In my life of explaining to people, mostly clinical and epidemiologic audiences, there is a bit of a professional pride about evidence-based practice. Clinicians are educated deeply, and they're experts; they should be applying this evidence to every patient every time. But I always start [by saying], "You guys are human too, right?"

With antibiotics, emotions play a large role in how people are using these drugs. I've had many clinicians describe antibiotics as some of the best anti-anxiolytics — so like it's an anti-anxiety medicine for the clinician.

This idea of the cold, logical, rational actor, I mean, doesn't apply anywhere in medicine. But in particular, I think this is a great [example of a] scenario where that perfect model of decision-making just gets completely upended by contextual and structural factors, as well as social and emotional factors.

NL: Are there additional factors to consider in the context of pediatrics?

JS: A lot of my portfolio is in pediatrics, and in fact, that's where I started my work. I was a postdoctoral fellow at the Children's Hospital of Philadelphia, so I have spent a lot of time doing pediatric research.

As pediatricians say, "We have two patients: there's the child and the caregiver, the parent or the guardian." Maybe two. And so you're navigating the patient and their parents, and the interactions have a lot of complexity. There's often the challenge where the patient can't communicate what's wrong; it's difficult to convey symptoms. It adds a layer to the diagnostic uncertainty.

Then, of course, the fragility of children [is a factor], and the concern of the illness going off the rails. That feels more fearful than it does for a middle-aged adult.

But I would say one thing with pediatrics is that parents are more open to the idea of not wanting to give their kids medication that they don't need. The origins of that may come from different places than what an antibiotic steward would necessarily think of as the main reason why you want to avoid antibiotics, because it's often just about avoiding any medication. But I think that parents can be a partner in stewardship, engaging with clinicians around whether or not an antibiotic is necessary or potentially being open to this "watch and wait" — this idea of holding off to see if the body fights off the infection on its own.

When you look nationally [in the U.S.], pediatricians have done the best at improving their prescribing. Some of the biggest leaps and bounds in outpatient stewardship, it started in pediatrics. So pediatricians tend to be on the cutting edge, I would say.

NL: In pediatric outpatient settings, are there any strategies that work really well?

JS: One of the most common ones is the use of "audit with feedback," this idea of prescribing report cards where you give clinicians information at regular intervals about how well they use antibiotics and then compare it to their colleagues in their practice or in their entire health system. That's been demonstrated to work, but not in isolation.

[Editor's note: Szymczak's research suggests that certain social factors make this approach more likely to work. For instance, clinicians who respond best trust that the data they're being given is accurate, feel supported by their leadership, don't feel overly stressed or surveilled by the feedback, and are comfortable fielding patients' demands for antibiotics.]

Another piece that has been demonstrated to work, if clinicians use it, is that many electronic health records have pathways or order sets or guidelines embedded. So, if a clinician's like, "I'm going to diagnose [urinary tract infection] UTI in this patient," there's a UTI pathway that they can click on that will give them evidence-based laboratory testing and management strategies. It takes them fewer clicks to get the stuff that they need.

So, it's multifactorial, but [effective stewardship] usually involves some combination of data, education and making the right choice the easy choice.

NL: When it comes to interventions for outpatient settings, are there strategies that just don't seem to work?

JS: Education on its own, targeting clinicians or patients, is not sufficient to move the needle on prescribing.

NL: Could you elaborate on why educating patients isn't the best approach?

JS: Patient education is important but has not been demonstrated to move the needle very much. I think partially that is because the approach we have taken to education has been connected to abstract concepts like antimicrobial resistance, which is important at the population level, of course, but can be difficult to understand for the lay public and can be less motivating [for them to change their personal behaviors].

I do think we are seeing more interest from the lay public in things like the microbiome and gut health and the role of antibiotics in potentially disrupting those things. I think education to the public that directly connects to individual-level harms is more motivating than population-level harms.

NL: When it comes to interventions, do you think the surrounding cultural context affects which strategies work best? I'm thinking of the U.S. versus Japan, for instance.

JS: I do think that the surrounding cultural context will always play a role, to some degree, in how interventions to improve clinical decision-making will fare. I have also written a bit about that in another commentary that delves into the observation in the United States that we have considerable regional variation in antibiotic use that is not explained by clinical factors.

I am cautious, though, about how we think about the concept of "culture" in relation to clinical decision-making. When it comes to antibiotic prescribing, I do think there are universal factors that shape how people respond to efforts to intervene, including the management of diagnostic uncertainty; fears of missing something, leading to "just in case" prescribing; a desire to offer patients something of value; and the difficulty explaining why antibiotics are needed or not.

NL: Regarding Japan's incentive program, which pays pediatricians "tips" for improving their antibiotic use, do you think a similar approach would be motivating for U.S. pediatricians? Would it be feasible to implement that kind of strategy here?

JS: I think it could be difficult to implement here, but the details of how the program is operationalized would be very important.

We know, in general, that financial incentives have [a] mixed impact on physician decision-making. You especially have to be careful about unintended consequences. For antibiotics, it would be very important to have a clear definition of the outcome that will be incentivized and how it would be measured. With antibiotics, there can be gray areas and you don't want to incentivize undertreatment, especially if it is individual-level financial incentives.

A better approach may be in aggregate and [to] reward health systems or clinics for improved antibiotic use for conditions in which antibiotics are never needed, for example.

Editor's note: This interview has been condensed and edited for clarity.