The coronavirus disease 2019 (COVID-19) pandemic has been a colossal challenge unparalleled in modern times. However, experts have emphasized that continued focus on antimicrobial stewardship (AMS) is warranted. Reference Rawson, Moore and Castro-Sanchez1,Reference Spernovsilis and Kofteridis2
The role of antibiotics in the management of COVID-19 remains controversial. One meta-analysis suggests a bacterial coinfection rate of ∼8% in COVID-19 patients. Reference Rawson, Moore and Zhu3 However, up to 84% of COVID-19 patients receive antibiotics. Reference Vaughn, Gandhi and Petty4 Although patients may require antibiotics for later secondary infections, bacterial coinfection at the point of viral diagnosis may be able to be ruled out by utilizing biomarkers and culture results. Reference Rawson, Moore and Castro-Sanchez1,Reference Heesom, Rehnberg and Nasim-Mohi5
Methods
This retrospective cohort study of patients hospitalized with COVID-19 who received respiratory antibiotics during admission was conducted between June 1, 2020, and July 31, 2020. The study was conducted at Norton Healthcare, a community healthcare system in Louisville, Kentucky, with 1,500+ licensed beds and 4 adult hospitals. The study was approved by the University of Louisville Institutional Review Board. Inclusion and exclusion criteria can be found in Figure 1.
Screening and inclusion/exclusion.
Date of COVID-19 diagnosis was defined as the date the positive test result populated in the electronic medical record. If the patient was diagnosed with COVID-19 prior to admission, the date reported in the history and physical was used. If neither was available, the admission date was used. The target window was defined as 48 hours before and after the date of COVID-19 diagnosis. Antibiotic length of therapy included the active course at the time of COVID-19 diagnosis. Antibiotics started or restarted later in the admission were not counted toward overall length of antibiotic therapy. Data from AMS interventions to discontinue antibiotics during the target window were collected. Additionally, culture data for urinary, respiratory, blood and cerebrospinal fluid (CSF) samples were collected if samples were obtained within 48 hours of COVID-19 testing. Blood cultures with coagulase-negative Staphylococcus spp in a single set were considered contaminants and were excluded. The first procalcitonin value of admission was collected if available. Lastly, the quick sequential organ-failure assessment (qSOFA) score was collected from the day of COVID-19 diagnosis.
AMS recommendations to discontinue all antibiotics in COVID-19 patients without evidence of bacterial coinfection were made by trained clinical pharmacists. At our institution, dedicated AMS pharmacists independently review patients on antimicrobials and to provide recommendations to the primary treatment team. During their review, AMS pharmacists assess symptomology, imaging reports, microbiologic culture data, vital signs, and relevant laboratory results including biomarkers such as procalcitonin. Recommendations are documented as part of the electronic medical record, including provider response as “accepted” or “rejected.”
The primary outcome was the percentage of patients for whom antibiotics were discontinued within the target window, referred to as early antibiotic discontinuation. If antibiotics were not stopped within this time frame, patients were assigned to the late antibiotic discontinuation group. Secondary outcomes included percentage of patients with a recommendation from the AMS program within the target window, percentage of AMS recommendations accepted, length of antibiotic therapy, length of stay, and inpatient mortality.
Data were summarized using descriptive statistics. We used the Wilcoxon rank-sum test to compare quantitative characteristics and the Fisher exact test to compare categorical variables. All analyses were conducted using R version 4.0.0 statistical software (R Foundation for Statistical Computing, Vienna, Austria).
Results
Overall, 226 patients were screened for inclusion. Among them, 53 patients were excluded, yielding a total sample of 173 patients (Fig. 1). In total, 91 patients (52.6%) met criteria for the early antibiotic discontinuation group and 82 patients (47.4%) were in the late discontinuation group. Groups were similar regarding most characteristics (Table 1). Patients in the late antibiotic discontinuation group were more likely have a fever (P = .007) and a first procalcitonin value >0.5 ng/mL (P < .001). In the early discontinuation group, median recorded antimicrobial length of therapy was 3 days versus 7 days in the late discontinuation group (P < .001).
Baseline Characteristics and Severity of Illness
Note. CSF, cerebrospinal fluid; FiO2, fraction inspired oxygen; ICU, intensive care unit; IQR, interquartile range; O2, oxygen; PaO2, partial pressure oxygen; qSOFA, quick sequential organ-failure assessment; PCT, procalcitonin; NA, not applicable. Growth from a positive respiratory culture: Serratia marcescens. Growth from positive blood cultures: Escherichia coli (n=2), Klebsiella pneumoniae (n=1), and Staphylococcus capitis (n=1).
Antimicrobial stewardship
Overall, 62 patients (35.8%) had a documented AMS recommendation with an acceptance rate of 74.2% (n = 46). Among patients for whom antibiotics were discontinued early, 38 (41.8%) had an AMS intervention, with an 86.8% acceptance rate. In the group for whom antibiotics were discontinued later, 24 (29.3%) had an AMS intervention, with a 54.2% acceptance rate. There was no significant difference in the prevalence of recommendations to stop antibiotics between groups (P = .112), but was a significant difference in acceptance rates (P = .003).
Mortality and length of stay
The overall in-hospital mortality rate was 17.3%. In the early discontinuation group, 13 (14.3%) died, and in the late discontinuation group, 17 (20.7%) died (P = .316). The median lengths of stay were similar between the early and late discontinuation groups at 7 and 9 days, respectively (P = .063). For patients admitted to the intensive care unit (ICU), the median ICU lengths of stay were similar between these groups: 7 days versus 7 days (P = .402).
Discussion
We found no difference in mortality or length of stay between the early and late antibiotic discontinuation groups, suggesting that early discontinuation of antibiotics for respiratory tract infections in COVID-19 may be safe. Vaughn et al Reference Vaughn, Gandhi and Petty4 reported that many patients with COVID-19 received antibiotics during their hospitalizations, with rates ranging from 27% to 84% across 38 hospitals. The high prevalence of antibiotic prescribing demonstrates the need for antimicrobial stewardship in COVID-19 patients.
Interestingly, admission to ICU at time to diagnosis, oxygen requirement, vasopressor requirement, and qSOFA score were not major drivers in the decision to discontinue antibiotics. Fever and procalcitonin value >0.5 ng/mL were more common in the late antibiotic discontinuation group, which may have influenced both appropriateness to make recommendations and provider willingness to accept. A single-center retrospective cohort study found that utilizing procalcitonin guidance shortened antibiotic length of therapy in critically ill COVID-19 from 7 days to 5 days (P < .001). Reference Heesom, Rehnberg and Nasim-Mohi5 Our findings also suggest that procalcitonin may be a useful tool to AMS programs.
A survey of physicians found that the majority of respondents believed COVID-19 patients needed a mean 7.12 days of antibiotics, suggesting an assumption of bacterial coinfection, despite evidence that this is not common. Reference Rawson, Moore and Zhu3,Reference Beović, Doušak and Ferreira-Coimbra6 AMS efforts could be a key influence in challenging this notion. In our cohort, 35.8% of all COVID-19 patients had an AMS recommendation. This finding highlights the need for all prescribers to critically evaluate whether antibiotics are necessary in COVID-19 patients.
The strengths of this study include its description of AMS efforts and outcomes in COVID-19 patients at the point of diagnosis. Within this cohort, there appeared to be no worse outcomes in patients whose antibiotics were discontinued early versus not; however, this study was not powered to detect a mortality difference. Nevertheless, these findings may add to prescriber confidence when discontinuing antibiotics in confirmed COVID-19 patients. Our findings also demonstrate that AMS recommendations may lead to shorter lengths of therapy in COVID-19 patients.
Future studies should describe AMS efforts stratified by severity of illness; this factor did not appear to be predictive of antibiotic decision making in our study. The utility of procalcitonin and other biomarkers to guide AMS efforts in this patient population could warrant larger-scale investigations. The study period occurred prior to spread of SARS-CoV-2 variants and it remains relatively unknown whether infection due to these mutated strains may change risk of bacterial coinfection or provider risk assessment. 7 AMS recommendations could be a key factor in promoting appropriate antibiotic use in COVID-19 patients.
Acknowledgments
We thank Doug Lorenz, Richard Head, and Stephen Wyatt for their aid in data acquisition and analysis.
Financial support
No financial support was provided relevant to this article.
Conflicts of interest
All authors report no conflicts of interest relevant to this article.
