A rapid review of the overuse of antibiotics during the COVID- 19 pandemic: lessons learned and recommendations for the future [version 1; peer review: awaiting peer review]

The coronavirus disease 2019 (COVID-19) pandemic has had severe implications on healthcare systems and the patients affected by this infectious disease. To improve outcomes for patients with COVID-19 and limit future antimicrobial resistance, there is continued urgency to improve our understanding of the rates and causative agents of secondary bacterial infections in patients with COVID‐19, and recognise whether antibiotics are being overused in patients prior to and following COVID-19 diagnosis. This article presents the results of a rapid review comparing reported rates of secondary bacterial infections with rates of antibiotic use in patients with COVID-19 predominantly in a hospital setting, within the context of treatment guidelines and recommendations. The review revealed rates of antibiotic use in patients with COVID-19 of 37–100%, far outweighing rates of secondary bacterial infections which were typically below 20%. There was a lack of consistent reporting of causative microorganisms of secondary infections, and the distinction between bacteriallyand virally-induced sepsis was rarely made. Early in the pandemic, healthcare agencies published treatment guidelines recognising the importance of antimicrobial stewardship. However, many are yet to provide updated guidance detailing the most appropriate antibiotics to treat patients with concurrent COVID19 and secondary bacterial infections in a way which limits the emergence of drug-resistant infections and does not negatively impact patient outcomes. Without significant improvements to the testing and reporting of causative organisms and corresponding updates to antimicrobial treatment guidelines, there is a risk of worsened clinical outcomes and increased burden on healthcare systems from antimicrobial resistance during the remainder of the COVID-19 pandemic and beyond. Open Peer Review Reviewer Status AWAITING PEER REVIEW Any reports and responses or comments on the article can be found at the end of the article. AMRC Open Research Page 1 of 15 AMRC Open Research 2021, 3:17 Last updated: 06 JUL 2021


Introduction
During the ongoing coronavirus disease 2019 (COVID-19) pandemic, caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), there have so far been more than 150 million confirmed cases and over 3 million deaths worldwide; at the time of writing, cases and deaths continue to increase rapidly 1 . Clinical outcomes from previous viral pandemics, such as the 1918-1919 influenza, 2003 SARS and the 2009 influenza A[H1N1]pdm09 pandemics, have shown that viral respiratory infections can often lead to secondary bacterial infections that significantly affect morbidity and mortality [2][3][4][5][6] . With evidence from previous pandemics demonstrating a link between viral infections and secondary bacterial infections, there was concern at the beginning of the COVID-19 pandemic over our the lack of understanding of whether patients infected with SARS-CoV-2 could also be at increased risk of developing life-threatening secondary bacterial infections 7 .
For the purposes of antibiotic stewardship, there is a need to better understand the true rates of secondary bacterial infections and bacterial sepsis in patients with COVID-19 to guide appropriate antibiotic treatment. The overuse of antibiotics in previous viral pandemics has led to increases in antibiotic resistance 5 ; with drug-resistant bacterial infections currently causing around 700,000 deaths globally per year, and predictions suggesting that this figure will increase to around 10 million deaths per year by 2050 8 , antibiotic stewardship during the remainder of the COVID-19 pandemic (and in future pandemics) is of paramount importance.
Here, we report the findings of a rapid literature review investigating the prevalence of secondary bacterial infections and rates of antibiotic use in patients with COVID-19. In addition, we detail how guidance regarding antibiotic treatment for patients with COVID-19 has changed over the course of the pandemic. In light of our findings, we recommend changes to health policy and treatment guidelines to improve the clinical outcomes of patients with severe COVID-19 and to reduce the emergence of antimicrobial resistance.

Methods
A rapid literature review was conducted to determine the relationship between reported rates of secondary bacterial infections, antibiotic usage and sepsis in patients with COVID-19. The review was initially conducted in August 2020, and updated in February 2021 due to the volume of COVID-19 research published in recent months.

Information sources and search strategy
During the initial review and the update, Embase and MEDLINE were searched simultaneously via OvidSP using a mixture of text words and subject headings relating to COVID-19/SARS-CoV-2 and sepsis or secondary bacterial infections (initial search conducted on 12 August 2020). In the review update (conducted on 26 February 2021) the search terms were narrowed to exclude sepsis owing to the high volume of published literature between the first and second searches. Full search terms can be found in Table 1. All study and literature types were considered, including online pre-prints. Included studies could be from any geographic region, provided the publication was available open access in English and reported data in humans.

Selection process
Two authors (WC, MB) screened the abstracts to determine their relevance. The first 50 abstracts were screened by both authors to align on acceptance and rejection criteria, before the remainder of the abstracts were each screened by a single author. The same two authors then screened full texts to prioritise the sources, before key features of each included study were extracted.
Supplementary hand searches and searches of the reference lists of included articles were used to identify further relevant articles. Hand searches of key grey literature were used to identify resources providing clinical guidance for antibiotic treatment in patients with COVID-19. The grey literature was initially searched in August 2020, and a second search was conducted in January 2021 to identify notable updates to these resources.

Data collection
Data extraction was performed by a single author for each individual study (WC, MB). The extracted information was independently verified by a second individual to check for accuracy of data extraction (WC, MB). The data extracted included the following, where available: characteristics of the included studies, such as study design, geographic region and population size; reported rates of sepsis/septic shock, bacterial infection and/or antimicrobial treatment; identified microorganisms; details of antimicrobials prescribed; and/or clinical guidance related to antibiotic treatment in patients with COVID-19. Extracted data were collated by hand into Table 2 and Table 3.

Results
A total of 488 abstracts were identified in the initial database search. Of these, 46 were categorised into three groups for full text screening: literature reporting concomitant COVID-19 and secondary bacterial infections; COVID-19 and sepsis; and COVID-19, secondary bacterial infections and sepsis. Records in the 'COVID-19 and secondary bacterial infections' and 'COVID-19, secondary bacterial infections and sepsis' were prioritised for the review. Six texts were not open access and were excluded. Overall, thirteen key references were included from the initial review. Five additional records were included from hand searches of grey literature sources, and ten references were included from reference list searches ( Figure 1).
In the review update, an additional 215 abstracts were identified after de-duplication of any articles identified during the initial review. Four texts were not open access and were excluded. 80 remaining texts were assessed for full text screening, of which 19 were included in the final review. Of the five grey literature sources identified in the initial round of review, updates to two documents were identified and included. Two additional references were included from reference list searches ( Figure 2).
Incidence of sepsis in patients with COVID-19 varied across the literature in the initial review, ranging from 13.5% in an ICU setting 20 to 100% in patients who did not survive 35 . There was similar variation in the reported rates of septic shock 35,36 ( Table 2). Sepsis was widely recognised as a predictor of patient mortality 20,35 and bacterial sepsis in particular was correlated with far worse ICU patient outcomes than viral sepsis 37 . However, the literature neither commonly differentiated between bacterially-and virally-induced sepsis, nor reported causative microorganisms of sepsis in patients with COVID-19 20,21,36 .
Although sepsis was not defined in the search criteria of the review update, overall rates of bloodstream infections in patients with COVID-19 were often reported and were generally lower than overall rates of secondary bacterial infections (1.0% 34 , 1.1% 32 , 6.5% 33 and 7.9% 14 ). By contrast, one paper reported that 31.2% of hospitalised patients in three medical centres in the United States had a positive secondary bloodstream infection; however, the paper did not clarify whether the rate had been adjusted for contaminants, and the study only included patients with 'severe' COVID-19 30 .
Despite typically low rates of secondary bacterial infections among patients with COVID-19, the majority of hospitalised patients with COVID-19 were reported to have received broad-spectrum antibiotics, at rates far surpassing those of secondary bacterial infections 7 . Alarmingly, one paper reporting antibiotic use in patients with COVID-19 in New York stated that 70% of patients who received antibiotics had been prescribed >3 different classes of antibiotic 9 . One study reported that even after a negative diagnostic test for a secondary bloodstream infection, 70.5% of patients were given antimicrobials 30 . Another review found that empiric antibiotic use was reported in 71-100% of hospitalised patients with COVID-19 36 . These rates were further corroborated in the literature, with rates of 43.7% 11 , 64% 32 , 71% 9 , 71.8% 7 , 72% 38 , 86% 39 , 88.3% 15 , 97.1% 12 and 100% 11 described, often despite a lack of effective diagnostic testing to confirm the presence of bacterial infections, and with limited evidence suggesting improved outcomes for patients administered empiric antimicrobial therapy 28 (Table 2 and Table 3).
Notably lower rates of empiric antibiotic prescribing of 37% were reported in one study on a Midwestern US Emergency Department, possibly attributable to the use of rapid procalcitonin testing to guide antibiotic use, even before a positive COVID-19 test 38 .

Guidance on the use of antimicrobials in patients with COVID-19
In order to address the discrepancy in the rates of secondary bacterial infections and empiric antibiotic use, clear treatment guidance and health policy is required. Despite rapid research into, and constant reassessment of, the clinical guidelines surrounding the use of antiviral drugs and immune-based therapies to treat patients with COVID-19, the guidance for antibiotic treatment has remained largely unchanged since the beginning of the pandemic.
Much of the clinical guidance continues to recommend against the widespread use of antibiotics in the treatment of COVID-19, unless a patient's symptoms are characteristic of a bacterial infection 13,[40][41][42][43] . Healthcare organisations recognise the importance of antimicrobial stewardship, and therefore have recommended that any antimicrobial treatment is reviewed regularly in light of the patient's clinical symptoms and microbiological test results, with the aim to switch to a narrow-spectrum antibiotic or initiate de-escalation protocols when appropriate 13,40-43 .
The National Institute of Health and Care Excellence (NICE) guidance provides a protocol that encourages judicious antibiotic use during the COVID-19 pandemic 40,41 . However, the guidance also recommends that for patients with suspected bacterial pneumonia and/or septic shock, clinical judgement is used to prescribe empiric, broad-spectrum antibiotics prior to microbiological test results 40,41 . Guidance from the World Health Organization (WHO) also suggests immediate treatment with antimicrobials within one hour of recognition of septic shock 13 . Considering that it can be difficult to differentiate between pneumonia and sepsis caused by bacteria or COVID-19 based on the clinical presentation alone 7,40-43 , however, it seems likely that these current guidelines could lead to the overuse of indiscriminate antimicrobials. A call in the first edition of the NICE guidance (published 1 May 2020) for research into the utility of procalcitonin testing alongside clinical judgement to guide antibiotic treatment decisions, remains unaddressed in the latest version (updated 9 October 2020), and evidence for the use of procalcitonin in the literature remains mixed 11,15,38 . Without the development of novel, rapid, and differential diagnostic tools such as these, healthcare professionals (HCPs) will continue to lack the information to determine the best treatment strategy for patients.
The WHO and NICE both provide some guidance for antibiotic choice in a hospital setting in the form of the Access, Watch, Reserve (AWaRe) classification tool and prescribing tables for patients with community-or hospital-acquired pneumonia, respectively 13,40,41 . Despite this, there is no specific guidance describing which antibiotics are the most effective and/or safe to treat patients who have both COVID-19 and a secondary bacterial infection, beyond whether the patient has 'non-severe' or 'severe' pneumonia and is at a low or high 'risk of resistance' 40,41 . In addition, a clinical guideline from the Dutch Working Party on Antibiotic Policy stated that there have yet to be any studies assessing the efficacy and safety of specific antibiotic regimens in patients who have COVID-19 with confirmed or suspected bacterial pneumonia which could help guide antibiotic choice 17 . This may be partly due to a lack of clarity on the bacterial infections most commonly associated with secondary infection in patients with COVID-19 as outlined above.
Although in our review the reporting of bacteria responsible for secondary infections was limited, given the high incidence of resistance to broad-spectrum antibiotics in those that were mentioned, it is concerning that the clinical guidelines recommend the empiric use of broad-spectrum antibiotics in the first instance, prior to microbiological testing results.

Discussion
The data summarised from this rapid review showed rates of antibiotic use typically far exceeded the rates of secondary bacterial infections in patients with COVID-19. The literature published more recently tended to include more detailed information on causative agents of secondary bacterial infections, antimicrobials prescribed and diagnostic tests than that published towards the beginning of the COVID-19 pandemic.
Antibiotic overuse in the hospital setting during COVID-19 may be driven by a combination of factors, including the high prevalence and severity of secondary bacterial infections in previous influenza pandemics (such as H1N1) 2-6 , guidelines recommending empiric antibiotics for the treatment of patients presenting with severe pneumonia 40-43 , and the time required to identify safe and effective antiviral and supportive therapies 44 . The overuse of antibiotics may also reflect the difficulty to clinically distinguish between bacterial-and COVID-induced pneumonia 7,40,41 and limited access to rapid diagnostic tests 7 . Furthermore, some procedures known to produce highly specific and accurate results, such as bronchoalveolar lavage, are aerosol-producing and have been avoided during the COVID-19 pandemic to prevent the spread of infection 10 . However, without accurate testing, our knowledge of the interaction between severe COVID-19, bacterial infections and sepsis will remain limited 42,43 , and lives may continue to be lost as a result of indiscriminate antibiotic treatment.
Although antibiotic treatment may be clinically beneficial for some patients with concurrent COVID-19 and secondary bacterial infections, their overuse in the COVID-19 pandemic will likely facilitate a surge in drug-resistant infections 45 , as was observed after the 2003 SARS epidemic 5 . Some therapeutic combinations may also have additional unintended consequences, such as the induction of cardiac side effects 44,[46][47][48] in patients treated simultaneously with azithromycin and hydroxychloroquine 49 . Other antimicrobials, particularly those taken in the absence of a bacterial infection, may induce an inflammatory response and a cytokine storm leading to septic shock 35 . Ceftazidime, listed in the NICE guidelines as an option for treating severe pneumonia 40,41 , has been suggested as one such driver of inflammatory responses 35 . The inappropriate use of antibiotics may also increase the risk of secondary nosocomial infections, such as Clostridioides difficile 7,13 .
Other articles have also highlighted the current gaps in the reporting of secondary infections and the implications of antibiotic overuse during COVID-19 and previous viral pandemics, such as one authored by Manohar et al. 3 . However, whilst the authors of this article suggest that the solution lies in developing new antibiotics and alternative antibacterial therapies, we propose that this would have little effect during the remainder of the COVID-19 pandemic, owing to the timelines associated with developing new treatments. Instead, we suggest the following changes to enhance reporting of secondary infections and refine treatment guidelines to reduce the use of broad-spectrum antibiotics and accurately advise on narrow-spectrum antibiotics that are likely to be effective.
Firstly, there is a need for rapid, accurate and accessible diagnostic tests to identify the causative microorganisms of secondary infections. Enhanced reporting of this diagnostic information will lead to a better understanding of the prevalence of secondary bacterial infections, common causative bacterial agents and antimicrobial sensitivity, and ultimately guide HCPs in more effective treatment decisions to improve patient outcomes and limit broad-spectrum antibiotic use 50,51 .
Secondly, bacterially-and virally-induced sepsis need to be accurately reported as distinct diagnoses. This is vital to inform HCPs on the likely rates of sepsis in patients with COVID-19, and to facilitate the use of appropriate treatments that improve patient outcomes whilst limiting the emergence of antibiotic resistance.
Thirdly, the accurate reporting of the cause of death on death certificates, as either COVID-19, a secondary bacterial infection, or bacterially-induced sepsis, is needed to fully understand the increased risks associated with secondary infections to establish the urgency of treatment with antimicrobials in lieu of microbiological test results.
Finally, there is a need for clinical guidance to recommend appropriate, narrow-spectrum antibiotics to treat suspected bacterial infections in patients who also have COVID-19. This guidance should be constantly reviewed and updated as soon as new evidence relating to the most commonly-occurring secondary bacterial infections emerges. This review had several limitations. Firstly, the review and update conducted were rapid and pragmatic, and did not utilise exhaustive search terms or comprehensively explore potential grey literature sources. The search terms used for the review were narrowed further for the update, although this was primarily due to the high volume of published literature between the two review periods. Secondly, the overwhelming majority of literature we identified focused on patients in a hospital setting, so our findings may not reflect the rates of secondary bacterial infections and antimicrobial use in an outpatient setting.
However, this review explored a range of literature sources published throughout the course of the COVID-19 pandemic so far, and the literature identified reported data from a diverse range of countries and patient subgroups.

Conclusions
This paper reports the results of a rapid review on the reported rates of secondary bacterial infections and antibiotic use during the COVID-19 pandemic. Worryingly, the reported rates of antibiotic use greatly outweighed the reported incidence of secondary bacterial infections, and clinical guidelines regarding antibiotic treatment in patients with COVID-19 do not appear to have been updated over the course of the pandemic to recommend specific antimicrobials for the treatment of patients with COVID-19 and secondary bacterial infections.
If the gaps in understanding, testing and reporting of causative microorganisms of secondary bacterial infections highlighted here remain, and treatment guidelines and policies are not amended to be more specific, the overuse of empiric, broad-spectrum antibiotics is likely to continue. This will likely lead to patients receiving inappropriate antibiotic treatment for both the remainder of the COVID-19 pandemic and in future pandemics, and will ultimately result in further emergence of antimicrobial resistance.

Data availability
Underlying data All data underlying the results are available as part of the article and no additional source data are required.

Reporting guidelines
Open Science Framework: PRISMA checklist for 'A rapid review of the overuse of antibiotics during the COVID-19 pandemic: lessons learned and recommendations for the future', https://doi.org/10.17605/OSF.IO/5NBPE 52 .
Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).