Our study suggests that the overall prescription rate of antibiotics among acute COVID-19 patients including outpatients and inpatients in Japan was around 13%. The rate was higher in winter and was associated with patient characteristics including older age, multi-morbidity, and severe disease requiring oxygen or invasive mechanical ventilation. Higher mortality and longer hospital stay in the prescribed inpatient group were likely to reflect that the prescribed group had severe disease. Although older age was a significant factor associated with prescription, older outpatients were more likely to receive antibiotics than older inpatients. Frequently used antibiotics included cephalosporins and penicillin with beta-lactamase inhibitors. To our knowledge, this may be the first study to determine prevalence and patient characteristics associated with antibiotic prescription in a large population with COVID-19.
In our study, the antibiotic prescription rate in COVID-19 patients was lower than the prescription rate in patients with upper respiratory symptoms in Japan prior to the COVID-19 pandemic (up to 20%)18, as well as in the US16. Although 38% of Japanese physicians in a national survey reported that they had clinical experience of prescribing antibiotics to patients with the common cold in outpatient clinics in the period before the pandemic, 96% of them recognized that they tried to follow the antibiotic stewardship program over the last year before the survey18. However, some patients with acute COVID-19 may have bacterial co-infection or secondary bacterial infection. A study noted that approximately eight percent of patients with acute COVID-19 who visited the emergency departments had presumed bacterial infection based on the finding of cultures growing clinically important bacteria at their ER visit9. In a meta-analysis19 based on data on acute COVID-19 up to April 2020, bacterial co-infection was identified in 3.5% of patients based on estimates on presentation and secondary bacterial infection in 14.3% of patients, but 73% of patients analyzed in this meta-analysis received antibiotics. Moreover, several studies which were published later, revealed that bacterial co-infection or secondary bacterial infection was noted in 1.2%20, 2.3%21 and 6.1%22 of all patients with acute COVID-19. When considering the low proportion of bacterial co-infection or secondary bacterial infection among patients with acute COVID-19, 13% of antibiotic prescriptions in patients with COVID-19 still may need improvement. Although a study outside Japan reported that physicians tend not to follow guideline statements against antibiotic prescription among patients with acute viral infections23, physicians in Japan need to address the prescription practice by safely reducing its use since patients with acute COVID-19 are likely to have a low probability of bacterial coinfection24.
The proportion among older outpatients with COVID-19 accounted for 32.0%, which was higher than that among older inpatients in our study. The US study also showed a proportion of around 30% among older outpatients, which is comparable to the result among older outpatients of our study12. Since admission is usually feasible for patients with severe diseases in Japan, and thus outpatients had less severe diseases than inpatients, it is unclear why the older outpatients in our study received more antibiotic prescriptions than the older inpatients. During the upsurge of COVID-19, a high proportion of patients with COVID-19 may have received antibiotics at outpatient settings even if they were indicated for admission, due to the high bed occupancy rate. Moreover, diagnostic uncertainty about COVID-19 might have been greater among outpatients than inpatients because of outpatients’ limited availability of testing to rule in COVID-19 as well as to rule out other possible causes of infectious disease.
In line with findings in a US study, the higher rate of antibiotic prescriptions in winter compared to summer was revealed by our monthly data12,25. Physicians examining patients with respiratory symptoms in winter might have a higher tendency to suspect bacterial pneumonia. Indeed, winter has a higher incidence of bacterial pneumonia and common respiratory viral infections25,26. Coinfection of bacteria and viruses, including SARS-CoV2, may occur, however, cases of COVID-19 with coinfection of bacterial pneumonia have been reported in mostly late-stage ventilator-associated pneumonia and rarely early-stage community-acquired bacterial pneumonia27,28. The prescription rate in the second winter (year of 2021) was lower than that of the first winter (year of 2020) in our study. This might have reflected the improvement of timely COVID-19 diagnosis as diagnostics gradually became available after the initial waves of the pandemic.
Bacterial coinfection such as pneumonia by Streptococcus pneumoniae or Staphylococcus aureus with COVID-19 was lower than those with influenza29. In a 10-year retrospective cohort study comparing bacterial co-infection in patients admitted with either COVID-19, influenza or RSV (respiratory syncytial virus) positive community-acquired pneumonia29, bacterial coinfection was identified in only four percent (46/1243) of COVID-19 patients who were observed between 2020 and 2021. In contrast, the coinfection was identified in 27% (209/775) of those with influenza between 2010 and 2021. In a comparison of outpatient antibiotic use between COVID-19 and influenza, a study in the U.S. showed that approximately one third of patients with PCR-confirmed influenza were prescribed antibiotics and that patients who received antibiotics were relatively older.
The findings of our study that older age, multi-morbidity, and severe COVID-19 requiring oxygen or invasive mechanical ventilation are associated with antibiotic prescription may reflect that the treatment threshold for antibiotic prescription is changed based on individual risk30. For example, physicians more frequently tried to manage patients with severe COVID-19 with possible treatable bacterial infection or those with high risk for severe disease (older patients or those with multi-morbidity). A large study on COVID-19 inpatients in Japan during the first to third waves showed that about 5% developed bacterial pneumonia and the risk was higher in older patients31. In the usual clinical course of COVID-19 patients, secondary bacterial infection may develop, if any, several days after onset and thus routine antibiotic prescription at the time of COVID-19 diagnosis is not recommended3,4. Nonetheless, there may be an opportunity to improve the antibiotic prescription practice by enhancing diagnostic excellence to aim for greater clarity of the threshold to commencing antibiotic treatment in individual patients.
The type of antibiotics prescribed in COVID-19 patients in our study differed from that of the US study on outpatients, with the highest prescribed being beta-lactams in the former and azithromycin (51%) in the latter12. The choice of azithromycin in US physicians might be related to the influence of several studies indicating the possible usefulness of azithromycin among acute COVID-19 patients32,33. In a survey study on physicians and their antibiotic prescription pattern for patients with suspected acute respiratory tract infection in Japan18, beta-lactams or beta-lactams/beta-lactamase inhibitors (52%) were prescribed more often than macrolides (35%) or fluoroquinolones (11%). Several antibiotics, such as cefazolin, ceftriaxone, and ampicillin/sulbactam, are available only for parenteral use. A potential explanation for the overuse of cefazolin or other antibiotics for community-acquired infections is the presence of a wide variety of bacterial co-infection in patients with COVID-19. A study suggested that bacterial co-infection in these patients may be of an etiology independent of COVID-19, commonly occurring in the elderly population. The presence of these infections possibly led to empirical overuse of these agents34, although the usefulness of antibiotics for the fatal cases remains unclear.
There are several limitations in our study. First, we could not adjust the confounding factors and draw a causal inference regarding antibiotic prescribing patterns due to the cross-sectional study design. Second, we could not exclude cases of concomitant bacterial infections of the lungs or other organ systems because of the lack of clinical reasons for the prescription in individual patients in the database. Moreover, we could not obtain clinical data of individual patients about the history and physical examination, culture results (sputum, blood, urine, and/or cerebrospinal fluid cultures), procalcitonin level, leukocyte count, and/or C-reactive protein level which physicians used to evaluate the probability of bacterial infection. Third, the participating hospitals that consented to data provision in the database were not randomly sampled. However, the database was large, and the distribution of demographics, diagnoses, and procedures was extremely similar to the national survey data, thus the practice patterns are likely to be similar across all Japanese acute care hospitals. Fourth, we could not collect antibiotic prescribing data in other types of healthcare institutions, including clinics, non-acute care hospitals, and geriatric long-term care facilities. Patients treated in these institutions may have different patterns in receiving prescriptions, which warrants further studies.
In conclusion, the antibiotic prescription rate among acute COVID-19 patients, including outpatients and inpatients, in Japan was relatively low, but room for improvement in antibiotic prescribing practice remains. The higher rate was recognized in winter and was associated with older age, multi-morbidity, and severe disease. In addition to the diagnostic uncertainty of COVID-19, possible bacterial coinfection and secondary bacterial infection may have caused physicians to prescribe antibiotics. Nevertheless, judicious use of antibiotics in patients with COVID-19 should be strongly encouraged even if the COVID-19 pandemic continues. Early detection of bacterial coinfection or secondary bacterial infection with wise use of bacterial cultures and inflammatory biomarkers should improve clinical management for COVID-19 patients.