Bloodstream Infections in Hospitalized Patients with COVID-19: A Systematic Review and Meta-Analysis

Background: Little is known about the occurrence of bloodstream infections in hospitalized patients with COVID-19 and the related clinical consequences. The aim of this systematic review and meta-analysis was to estimate the pooled occurrence of BSIs among hospitalized patients with COVID-19 and mortality of this patient population. Methods: A systematic search was performed on PubMed, EMBASE, and Web of Science from inception to 19 April 2021. The primary outcome was the occurrence of BSIs among hospitalized patients with COVID-19. The secondary outcome was mortality at the longest available follow-up. Results: Forty-six studies met the inclusion criteria, with a total of 42,694 patients evaluated. The estimated occurrence of BSIs was 7.3% (95% CI 4.7–1.1%) among hospitalized patients with COVID-19, with a mortality rate of 41% (95% CI 30%–52.8%). The subgroup analysis conducted on patients admitted to ICU provided an estimated occurrence of 29.6% (95% CI 21.7%–38.8%). A higher occurrence of BSI was observed in patients with COVID-19, in comparison with patients without COVID-19 (OR 2.77; 95% CI 1.53–5.02; p < 0.001). Conclusions: Our analysis estimated the occurrence of BSIs among hospitalized patients with COVID-19 at around 7%. A four-times higher occurrence was estimated among patients admitted to ICU.


Introduction
Bloodstream infections (BSIs) are frequently causes of infection, sepsis, or septic shock in hospitalized [1] and critically ill patients [2]. Whether community or hospital acquired, BSIs may complicate patients' hospital stay and have been associated with negative outcomes [1]. The COVID-19 pandemic has caused the hospitalization of a substantial number of patients with acute respiratory failure. Previous reports have described the occurrence of BSIs in cohorts of patients with other viral pneumonia (e.g., influenza) [3,4]. The clinical features of SARS-CoV-2 infection, the use of immunomodulatory drugs, and the high rate of admission to ICU may pose patients with COVID-19 at a high risk of developing superinfections [5], such as ventilator-associated pneumonia and BSIs [6]. To date, fragmented data are available on the occurrence of BSIs in populations of hospitalized patients with COVID-19, and little is known about how BSIs may influence the outcome of these patients.
The aim of this systematic review and meta-analysis is to estimate the pooled occurrence of BSIs among hospitalized patients with COVID-19 and mortality of this patient population.

Materials and Methods
The protocol of this systematic review and meta-analysis was prospectively registered in the Open Science Framework (https://osf.io/ys8kd). For the purpose of this systematic review and meta-analysis, a systematic search was performed on PubMed, EMBASE, and Web of Science from inception to 19 April 2021. The search strategy included keywords as exact phrases and subject headings, according to database syntaxes, and is available as Supplementary Material. The reference list of relevant articles was also screened (i.e., the snowballing method). Full literature search records were then screened independently and by four authors in blinded pairs (MI, BS, CF, FRC) to identify all relevant records from titles and abstracts. Studies selected as relevant were then evaluated from the full text and included if two reviewers independently agreed on their eligibility. Nonrandomized studies, both prospective and retrospective, were considered eligible when specifically addressing the occurrence of bloodstream infections (BSIs) in adult and hospitalized patients with COVID-19, defined as the proportion of patients who developed at least one BSI during the study period, which divided the cohort of the study. The occurrence rate of BSI was the primary outcome of the study. The secondary outcome was mortality of patients with BSI and COVID-19 at the longest available follow-up. Case-control studies with a predefined or fixed proportion of enrollment between infected and non-infected were excluded, due to their design, as they were not able to provide a reliable estimate of the occurrence of BSI. Studies presenting data on only one family or species of microorganism were excluded, as they were not able to provide a reliable estimate of the overall occurrence of BSI. Studies including less than 10 patients (i.e., case series), case reports, abstracts, not peer-reviewed articles, and articles not in the English language were excluded. The Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) [7] checklist is provided in Supplementary Material. Data collection was performed in duplicate and using an electronic standardized data extraction form. Discrepancies at any stage are resolved by discussion. When the disagreement was due to a difference in interpretation, arbitration was conducted by another author (GC or CM). The corresponding authors of the included studies were contacted to obtain additional information regarding eligibility or data presentation, if needed, by two authors (MI and AC).

Qualitative Analysis
Qualitative analysis of included studies was performed using the Methodological Index for Non-Randomized studies (MINORS) [8], due to its ability to be used for the assessment of single arm studies, independently and in duplicate by two authors (GC, MI). Disagreements over the assessment were resolved by a third author (AC). The items were scored as 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate). The global ideal score was 16 for non-comparative studies and 24 for comparative studies.

Statistical Analysis
Meta-analysis was performed in the case of two or more included studies reporting data on the outcomes of interest. The summary estimates were derived from logit transformation of individual study proportions of the outcomes and presented along with the corresponding 95% confidence interval (CI), calculated using random-effect meta-analysis. The I-squared (I 2 ) statistical model was used to describe the percentage of variation across the included studies due to heterogeneity. Prespecified subgroup analyses were performed on the basis of the setting (e.g., intensive care unit, non-intensive care wards) and the number of centers per study (e.g., multicenter, single center). We also performed a sensitivity analysis, including studies comparing the outcomes of patients with COVID-19 to those of patients without COVID-19. A post-hoc subgroup analysis was performed on studies including only hospital-acquired BSI. All the analyses were performed by MI with inputs from AC, using Open Meta-Analyst 8 [9].

Characteristics of Included Studies and Patients
A total of 1172 records were retrieved. The full search output is available as Supplementary Material. After the screening of the records, a total of 46 studies were included in this systematic review and meta-analysis [6,. The included studies provided data on a total of 42,694 patients with COVID-19. The inclusion/exclusion process is presented as a PRISMA flow diagram, as shown in Figure 1.  19 to those of patients without COVID-19. A post-hoc subgroup analysis was performed on studies including only hospital-acquired BSI. All the analyses were performed by MI with inputs from AC, using Open Meta-Analyst 8 [9].

Characteristics of Included Studies and Patients
A total of 1172 records were retrieved. The full search output is available as Supplementary Material. After the screening of the records, a total of 46 studies were included in this systematic review and meta-analysis [6,. The included studies provided data on a total of 42,694 patients with COVID-19. The inclusion/exclusion process is presented as a PRISMA flow diagram, as shown in Figure 1. The characteristics of the included studies are presented in Table 1. Sixteen studies were multicentric and 30 were conducted in a single center. Twenty-nine studies were conducted in EU, 10 in USA, four in China, and three in other countries. The included patients had an age ranging from 32 to 70 years, with a percentage of male gender in the cohorts ranging from 48% to 94%. All the studies were single-arm studies, except 11 also presenting data on patients without COVID-19 as a comparator. Details on isolates and source of the infection are provided in Table S2, Supplementary Material. Data on causative microorganisms showed that the infections were mainly bacterial and rarely fungal BSIs, with Gram positive species prevalent in many studies. The qualitative assessment per study, according to MINORS, is provided in Tables S3 and S4, Supplementary Material. Only two [52,55] of the studies reported protocol registration and none reported information on sample calculation. Among the 11 studies with a comparison group, all ex- The characteristics of the included studies are presented in Table 1. Sixteen studies were multicentric and 30 were conducted in a single center. Twenty-nine studies were conducted in EU, 10 in USA, four in China, and three in other countries. The included patients had an age ranging from 32 to 70 years, with a percentage of male gender in the cohorts ranging from 48% to 94%. All the studies were single-arm studies, except 11 also presenting data on patients without COVID-19 as a comparator. Details on isolates and source of the infection are provided in Table S2, Supplementary Material. Data on causative microorganisms showed that the infections were mainly bacterial and rarely fungal BSIs, with Gram positive species prevalent in many studies. The qualitative assessment per study, according to MINORS, is provided in Tables S3 and S4, Supplementary Material. Only two [52,55] of the studies reported protocol registration and none reported information on sample calculation. Among the 11 studies with a comparison group, all except two had historical comparison groups [27,35]. Furthermore, 8 out of 11 studies reported only unadjusted statistical analysis [10,12,18,20,22,27,28,44]. These were the most frequently downgraded domains.
We also conducted a sensitivity analysis considering the occurrence of BSI patients without COVID-19, in comparison with those with COVID-19. Eleven studies provided data on a comparison group, but one study [18] did not provide the denominator to calculate the occurrence. Thus, in this analysis, 10 studies [6,10,12,16,20,22,27,28,35,44] providing unadjusted data on occurrence of BSI in the comparison group were included. There was found a significantly higher risk of BSI in patients with COVID-19 compared to patients without COVID-19 (OR 2.18; 95% CI 1.35-3.51; p < 0.001; I 2 = 79%; Figure 4).

Discussion
To the best of our knowledge, this is the most updated systematic review and metaanalysis estimating the occurrence of BSIs among hospitalized patients with COVID-19. Our main finding was that around 7% of hospitalized patients with COVID-19 may experience a BSI. This data was comparable with that measured among hospitalized patients prior to the pandemic [56]. Interestingly, around one out of three patients with COVID-19, once admitted to ICU, may have a clinical course complicated by a BSI. This finding is also in line with the previously estimated prevalence of BSIs in adult patients with sepsis or septic shock [1,57] and almost double than the prevalence of BSIs in cohorts of adult patients admitted to ICU [58].

Discussion
To the best of our knowledge, this is the most updated systematic review and metaanalysis estimating the occurrence of BSIs among hospitalized patients with COVID-19. Our main finding was that around 7% of hospitalized patients with COVID-19 may experience a BSI. This data was comparable with that measured among hospitalized patients prior to the pandemic [56]. Interestingly, around one out of three patients with COVID-19, once admitted to ICU, may have a clinical course complicated by a BSI. This finding is also in line with the previously estimated prevalence of BSIs in adult patients with sepsis or septic shock [1,57] and almost double than the prevalence of BSIs in cohorts of adult patients admitted to ICU [58].

Discussion
To the best of our knowledge, this is the most updated systematic review and metaanalysis estimating the occurrence of BSIs among hospitalized patients with COVID-19. Our main finding was that around 7% of hospitalized patients with COVID-19 may experience a BSI. This data was comparable with that measured among hospitalized patients prior to the pandemic [56]. Interestingly, around one out of three patients with COVID-19, once admitted to ICU, may have a clinical course complicated by a BSI. This finding is also in line with the previously estimated prevalence of BSIs in adult patients with sepsis or septic shock [1,57] and almost double than the prevalence of BSIs in cohorts of adult patients admitted to ICU [58].
A higher occurrence of BSI in patients with COVID-19 was also observed, in comparison with patients without COVID-19 (OR 2.77; 95% CI 1.53-5.02; p < 0.001). Although caution is needed considering this result, coming from an unadjusted analysis, the development of a BSI in patients with COVID-19 may share some risk factors with other superinfections, frequently observed in patients with COVID-19 [5,59]. Hospital and ICU overcrowding, a difficult application of infection prevention strategies while using PPE in such settings, patients' immunological impairment, and the frequent need for central lines can be counted as risk factors specifically belonging to patients with COVID-19.
During the early phases of the pandemic, many centers supported the protocolized administration of antibiotic therapy at the time of diagnosis or hospital admission due to COVID-19. Considering our data, in addition to what was already issued on the topic [60], that approach may not be probably justified by the real occurrence of superinfections, and a tailored approach is needed. A clinical suspect should still probably remain the condictio sine qua non to consider the beginning of an antibiotic therapy in patients with COVID-19, taking into account that those admitted to ICU may deserve specific attention.
Our study has strengths, such as the comprehensive search, the methodology, and the reporting according to PRISMA 2020 [7], the large number of studies and patients evaluated, and the geographical variety of the settings. Moreover, incompletely reported data were retrieved from the corresponding authors when feasible. Our analysis also has limitations. The nature of this study was descriptive, and no analysis was performed to look for associations with potential risk factors. Our analysis also had a high statistical heterogeneity. The high statistical heterogeneity across the studies may be attributed to the different criteria used to include patients, based on the origin of the infection (i.e., community-acquired, hospital-acquired, ICU-acquired), the variable sampling rate, or the different settings (i.e., ICU or non-ICU wards). Other factors, such as patients' characteristics or clinical severity may also explain heterogeneity. However, we tried to reduce the impact of heterogeneity with several pre-planned sensitivity and subgroup analyses. Data on causative microorganisms were also heterogeneously reported across the studies; thus, we decided to report them as per the authors' descriptions, and we did not perform any quantitative analysis on causative microorganisms.

Conclusions
A total of 7% of hospitalized patients with COVID-19 may experience BSI with a mortality rate of around 40%. Almost 30% of patients admitted to the ICU may develop BSI during their ICU stay. Hospitalized patients with COVID-19 may have three-times higher odds of developing BSI than patients without COVID-19.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/microorganisms9102016/s1, PDF S1: Search strategy; PDF S2: Figure S1: Forest plot with the results of the subgroup analysis conducted on the occurrence of bloodstream infections in patients with COVID-19 in multicenter studies; Figure S2: Forest plot with the results of the subgroup analysis conducted on the occurrence of bloodstream infections in patients with COVID-19 in single center studies; Figure S3: Forest plot with the result of the post-hoc subgroup analysis, considering only data from studies specifically addressing hospital-acquired bloodstream infections; Table S1: PRISMA Main Checklist and PRISMA Abstract Checklist; Table S2: Characteristics of bloodstream infections and isolates, Table S3: Quality assessment of studies according to MINORS score for non-comparative studies; Table S4: Quality assessment of studies according to MINORS additional score for comparative studies; Excel S1: Full search output.
Author Contributions: A.C. and M.I. conceived and designed the work. M.I., B.S., C.F., F.R.C., G.C., C.M., and A.C. collected the data. M.I. and A.C. performed the analysis. G.M. and A.G. gave substantial contributions to data acquisition and interpretation for the work; M.I. and A.C. drafted the work. B.S., C.F., F.R.C., G.C., C.M., G.M., and A.G. revised the work critically for important intellectual content. All the authors gave the final approval of the version to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement:
The data presented in this study are available in Supplementary Materials. The unpublished data retrieved by private correspondence with the authors and used in this study analysis are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.