Study: Outpatient antibiotics widely prescribed without an infection diagnosis

Several studies in recent years have documented a significant amount of inappropriate antibiotic prescribing in outpatient settings for respiratory infections that are generally caused by viruses and don't require antibiotics. In emergency rooms, doctor's offices, and urgent care clinics, anywhere from a third to nearly half of the antibiotics prescribed for coughs, sore throats, and other respiratory ailments have been found to be unnecessary.

Jeffrey Linder, MD, MPH, a general internist at Northwestern University Feinberg School of Medicine, says that those studies, which have focused on the type of ailments that drive people to visit a provider, have revealed only one aspect of the problem.

"The way we've been looking at outpatient antibiotic prescribing, only focusing on in-person visits and only focusing on a restricted number of diagnoses, has been missing a lot of antibiotic prescribing," he told reporters today at IDWeek 2018.

Linder was presenting the results of a new study, conducted with researchers from the University of Michigan Medical School and Harvard Medical School, that found that 46% of antibiotics prescribed at 514 outpatient clinics were prescribed without an infection-related diagnosis. Even more troubling, 20% of all antibiotic prescriptions were given to patients without an in-patient visit.

"You should almost never get an antibiotic without being seen for conditions like a cold or the flu, symptoms like a cough or a sore throat," he said. "There are things that we as doctors need to do, particularly for coughs and sore throats, to determine whether an antibiotic is actually indicated or not."

Diagnostic codes and infection-related diagnoses

Using the electronic health record system of an integrated health delivery system, the researchers evaluated 509,534 antibiotic prescriptions made to 279,169 unique patients by 2,413 clinicians. They then sorted those prescriptions into three baskets, based on the same-day diagnostic code associated with the prescriptions. The idea was to see whether there was a diagnostic code that could explain why an antibiotic was prescribed.

Prescriptions were considered infection-related if they were associated with a diagnostic code that may signify an infection, while prescriptions linked to diagnostic codes that do not signify an infection were considered non-infection-related. The third basket was prescriptions associated with no diagnosis.

The researchers found that 54% of the prescriptions were for infection-related diagnoses. Of the 46% without an infection-related diagnosis, 29% were non-infection-related and 17% were associated with no diagnosis. Linder explained that some of the antibiotics prescribed for non-infection-related diagnoses were for viral rather than bacterial infections. But in others, the diagnostic codes were "completely irrelevant" to antibiotic prescribing, indicating conditions like hypertension or signifying an annual wellness visit.

Recognizing that not all prescriptions are written on the day the diagnosis is recorded in the electronic health record, Linder and his colleagues expanded their analysis to include diagnoses made 30 days before and 30 days after a prescription was written. Even then, they still found that 35% of prescriptions had no infection-related diagnostic code.

Linder suggested that part of the problem could be attributed to sloppy diagnostic coding that doesn't accurately reflect what happened during a patient's visit with a provider. But he doesn't think that diminishes the findings. "Even if it's bad record-keeping, it's a significant problem," he said.

Linder also acknowledged that some of the 54% of prescriptions associated with infection-related diagnoses were for viral conditions like sinusitis and probably didn't require an antibiotic. "We were very forgiving to doctors in terms of what we considered appropriate," he said.

The most common antibiotic classes prescribed were penicillins (30%), macrolides (23%), cephalosporins (14%), fluoroquinolones (11%), tetracyclines (10%), and sulfonamides (6%).

Prescriptions without in-patient visits

Of the 20% of antibiotic prescriptions outside an in-person visit, half (10%) were prescribed over the phone, 4% were simply entered into the electronic health record, 4% were refills, and 1% were made through an online portal.

Linder said there are cases where getting an antibiotic without seeing a physician is appropriate, citing women with recurrent urinary tract infections and teens taking antibiotics for acne as examples. "There is an example where people might call in, not be seen in person, and an antibiotic prescription might be reasonable," he said.

But he was particularly concerned that 4% of antibiotic prescriptions that didn't involve a visit were refills. While getting refills over the phone for chronic medications, like blood pressure or cholesterol medicine, is common, Linder said the system shouldn't enable refills for antibiotics. "It's a little concerning that we see a refill for an antibiotic prescription," he said.

For those who may think getting an antibiotic without seeing a doctor isn't a big issue, Linder warned that indiscriminate antibiotic use can promote antibiotic resistance. He also stressed the potential adverse reactions to antibiotics, from allergic reactions to diarrhea to Clostridium difficile infection.

Linder said the study indicates the problems with outpatient antibiotic prescribing, which accounts for 80% of all antibiotic prescribing in healthcare, run much deeper than previously understood, and illustrate how much antibiotics have been taken for granted. He called for providers to take a more "righteous" attitude about antibiotics.

"We've been pretty cavalier about the use of antibiotics, not appreciating the harms as much as possible and not educating patients as much as we should," Linder said. "I'm not anti-antibiotic, but I would like us to use them responsibly so we can use them effectively when indicated."

Linder said the next steps will be to go through the data more carefully to understand what was happening in the three categories of prescribing and to look at provider notes in the electronic health record. And they're setting up a prospective study in which they'll call a subset of doctors within a week of an unexplained antibiotic prescription to get a better understanding of the decision.

The study was funded by the Agency for Healthcare Research and Quality.

See also:

IDWeek abstract #1632

Newsletter Sign-up

Get CIDRAP news and other free newsletters.

Sign up now»


Unrestricted financial support provided by

Bentson Foundation 3MAccelerate DiagnosticsGilead 
Grant support for ASP provided by


  Become an underwriter»