Since the novel coronavirus pandemic began, limited data from small studies in several countries have indicated high rates of antibiotic prescribing in COVID-19 patients and low rates of bacterial co-infections, raising concerns about unnecessary antibiotic use and the pandemic's potential impact on antimicrobial resistance (AMR).
A new study from researchers in Michigan highlights some of those concerns.
In the study, published in Clinical Infectious Diseases, researchers looked at data on more than 1,700 hospitalized COVID-19 patients treated at 38 Michigan hospitals from Mar 13 to Jun 18 and found that more than half received early antibiotic therapy, with antibiotic use as high as 84% in some hospitals. But only a small fraction of those patients had bacterial co-infections.
But in another study on COVID-19's impact on antibiotic use, researchers in Pittsburgh reported that monthly antibiotic use in the Veterans Affairs (VA) Pittsburgh Healthcare System was down significantly in March through June compared with previous years, as a result of COVID-19 restrictions that limited medical procedures at the hospitals.
Taken together, these findings suggest that COVID-19 has affected antibiotic use in different ways, creating a cloudy picture of how the pandemic will ultimately affect AMR.
Low rate of bacterial co-infection
Valerie Vaughn, MD, a hospitalist at the University of Michigan's academic medical center and the lead author on the Michigan study, said she wasn't surprised to find a high rate of antibiotic prescribing in COVID-19 patients. Michigan was one of the states hit hard early on in the pandemic, and she knew from her own experience that many patients were getting antibiotics.
There were several reasons for this. Patients were coming into hospitals very ill with pneumonia-like symptoms, COVID-19 test results were taking a long time to come back, and there was a real concern that many patients had secondary infections. But of the 1,705 patients in the study, only 3.5% were found to have a community-onset bacterial co-infection.
"I actually thought our co-infection rate would be much higher," Vaughn said. "A lot of us were really worried that patients coming into the hospital who were really ill with COVID were ill because they had a bacterial infection on top of their viral infection. But now, over the course of the last half a year, we've learned that that's not really the case."
Still, antibiotic use was high, with 56.6% of patients receiving early empiric antibiotic therapy. Across the 38 hospitals, the percentage of patients prescribed antibiotics ranged from 27% to 84%. And that didn't include prescribing of azithromycin, which some hospitals were using in combination with the antimalaria drug hydroxychloroquine as part of standard of care treatment. Patients were more likely to receive antibiotics if they were older and had more severe illness.
Vaughn and her colleagues found that antibiotic use in COVID-19 patients did decrease over the course of the study, as the turnaround time of the COVID-19 tests shortened. Early on, it took as long as 2 weeks for some test results to come back from the state health department. But by April, Vaughn said, most of the hospitals had tests that could produce results in less than a day, and some now have tests with a turnaround time of less than an hour.
"That's made a world of difference," she said. "When it takes days for a test to come back, or even 1 day…it's hard to hold off on antibiotics while waiting to see if a test comes back."
Vaughn said that the lack of treatments for COVID-19 early on played a role in heavy antibiotic prescribing, as did the crush of COVID-19 patients in some hospitals, which left clinicians with less time to make decisions about treatment.
"You want to do something, and I think that, early on, antibiotics were that something," Vaughn said.
The study showed that the percentage of patients treated with antibiotics fell from 66.7% in March to 46.9% in May, as the percentage of tests that came back in less than day increased from 54.2% to 89.2%. Vaughn and her colleagues also found that low procalcitonin—a protein in the blood associated with the body's response to a bacterial infection—had a high negative predictive value—98.3%, indicating that low procalcitonin levels are a strong indication that antibiotics aren't needed.
What Vaughn and her colleagues don't know, and would like to understand better, is why antibiotic use was so high at some of the hospitals studied. It could be, they suspect, that hospitals with better antibiotic stewardship programs and more resources were better equipped to figure out who needed antibiotics and who didn't.
"That's one thing that we think may be going on, [that] robust stewardship carries on to these other patients, whether it's because of the culture of the hospital or there are more resources," she said. "Maybe hospitals that don't have those resources outside of a pandemic aren't going to do well when their resources are stretched even thinner by the pandemic."
Vaughn thinks that quicker test turnaround time and more COVID-19 therapeutics will help keep antibiotic prescribing down if there's another big wave of infections in the coming months.
"I suspect we'll do better the next time around," she said.
Drop in monthly antibiotic use
At the Pittsburgh VA Healthcare System, the situation in the early months of the pandemic was not quite as chaotic. Unlike Michigan, Western Pennsylvania was not one of the areas hit hard by the early wave of COVID-19 cases. And that had an impact on antibiotic use at the acute care hospital.
In a study published last week in Antimicrobial Agents and Chemotherapy, researchers from the VA Pittsburgh Healthcare System, the University of Pittsburgh, and Johns Hopkins Center for Health Security reported that antibiotic days of therapy (DOT) for March through June 2020 fell by 151.5 per month compared with January 2018 through February 2020—a 6.5% monthly reduction. That was driven by a decrease of 285 bed days of care (BDOC) per month.
Essentially, antibiotic use fell because fewer patients were in the hospital during the early months of the pandemic. As in many parts of the country, shelter-in-place restrictions drastically reduced the number of people going to the hospital for non-COVID–related issues.
"The overall amount of patients admitted to the hospital was much less than seen in previous years," said Deanna Buehrle, PharmD, the study's lead author and an infectious diseases clinical pharmacist with the VA Pittsburgh Healthcare System. "There were less patients coming in for surgical procedures, and less people seeking healthcare and being admitted to the hospital."
And with fewer people getting surgery, there were likely fewer healthcare-associated infections that might require antibiotics.
But among the patients who were admitted to the hospital, antibiotic use actually went up, by an average of 8.1 DOT per 1,000 BDOC, compared with the previous 2 years. Of the 16 patients with COVID-19 treated at the hospital, 9 (56%) received antibiotics. And, as with Michigan, a major reason was the delay in getting test results back.
"A lot of [these patients] received coverage for community-acquired pneumonia until their tests came back," Buehrle said. "During the very early weeks, the COVID testing turnaround could take a couple of days."
In addition, Buehrle said, a lot of the patients were elderly, had comorbidities, and were coming from nursing facilities, which put them at increased risk of bacterial co-infections. Of the 16 COVID-19 patients, 6 had co-infections upon admission or acquired them during their hospital stay.
Even though the number of COVID-19 patients seen at the hospital in the early months of the pandemic was small, Buehrle said the data have enabled her and her colleagues to identify the patients who don't need antibiotics, those in whom antibiotics can quickly be stopped, and those who have or may be at risk for co-infections and who may need to antibiotics that target specific pathogens. That will be helpful if there is a surge of cases in the coming months.
"It's important to have a stewardship plan going forward," she said.
Impact on AMR remains unclear
So what do these studies tell us about how the COVID-19 pandemic is going to affect antibiotic resistance rates?
That remains unclear. As Buehrle and two of her co-authors on this study—Neil Clancy, MD, and M. Hong Nguyen, MD, of the University of Pittsburgh—wrote in a commentary in JAC-Antimicrobial Resistance in July, no conclusions about the relationships between COVID-19 and antibiotic resistance can be drawn from the reports that have been published to date.
But they argue, ultimately, that excessive antibiotic use during the pandemic will likely increase AMR in some regions of the world, particularly in hard-hit areas that already have a high prevalence of drug-resistant pathogens, such as China; Lombardy, Italy; and New York.
"Hospitals and regions with high AMR prevalence that have moderate-to-high numbers of COVID-19 patients should be prepared for potential upswings in nosocomial infections by these pathogens," they wrote.
In addition, they said, there could be significant selection pressure for resistant pathogens in hospital units that were converted to COVID-19 care, even if the rest of the hospital saw an overall reduction in antibiotic use. And broad-spectrum antibiotic use in older, comorbid COVID-19 patients who are at risk for bacterial co-infections caused by resistant pathogens could exacerbate the problem.
Furthermore, as healthcare restrictions are lifted, overall antibiotic use is likely to return to prior levels.
A counterpoint commentary in the same issue of JAC-Antimicrobial Resistance, however, argues that the pandemic should lead to a drop in resistance rates in many countries because of improved infection prevention and control practices and restrictions on travel.
But another factor that could affect antibiotic resistance, Vaughn suggested, is outpatient prescribing during the pandemic, something that researchers are starting to look at. The question is, with a lot of routine healthcare visits now being done virtually and clinicians unable to do a complete medical examination of patients, are clinicians more or less likely to prescribe antibiotics?
Both Vaughn and Buehrle agree that the picture will likely remain unclear for several years.
"Understanding how this affects patient outcomes and resistance overall will definitely take time," Vaughn said.
"I'm not sure we have a great grasp on it yet," Buehrle added.