Changing epidemiology, microbiology and mortality of bloodstream infections in patients with haematological malignancies before and during SARS-CoV-2 pandemic: a retrospective cohort study


Bacterial bloodstream infection (BSI) is one of the most common complications in patients with haematological malignancies (HMs), which is associated with high morbidity and mortality.1–3 The incidence of BSI in these patients can exceed 50%, with an overall mortality >6%.4 5 Due to the underlying immunosuppressive, myelosuppressive cytotoxic chemotherapy, prolonged and severe neutropenia, severe mucosal damage and the use of broad-spectrum antibiotics, the mortality of BSI in haematological malignant patients is dramatically higher than in other patients.6–8 It is crucial to understand the microbiological spectrum and antimicrobial susceptibility of pathogens across different time periods and geographical regions to facilitate timely prevention and effective empirical antimicrobial therapy.

SARS-CoV-2 outbreak occurred at the end of 20199 and until 8 January 2023; China lifted Class A infectious disease prevention and control measures for SARS-CoV-2. Although the impact of SARS-CoV-2 infection on the outcomes of cancer patients remains controversial, most studies showed that SARS-CoV-2 exhibited an unfavourable outcome.10 11 During the pandemic of SARS-CoV-2, more stringent hygiene measures are conducted throughout the whole patient stay by hospital staff, patients and visitors.12 13 Thus, the epidemiology and microbiology may differ significantly before and during SARS-CoV-2 pandemic.14 The aim of this study was to explore the difference in epidemiology, microbiology and mortality of BSI in patients with haematological malignancies before and during the SARS-CoV-2 pandemic.

Patients and methods

Setting and study design

This retrospective cohort study was performed at the haematology department of Nanfang Hospital, which is the largest haematological centre in southern China, consisting of two ordinary wards and a haematopoietic stem cell transplantation ward, with a total of 176 beds. The study was evaluated and deemed exempt from a formal review by the Ethics Committee of Nanfang Hospital due to no personally identifiable information collected. The demographic, clinical and microbial data were prospectively collected in the hospital information system and microbiology department records. Patients were followed up 30 days after BSI by the chief doctor and infection control staff.

Patients aged ≥18 years with haematological malignancy such as acute myeloid leukaemia (AML), acute lymphoblastic leukaemia (ALL), lymphoma, multiple myeloma, chronic myeloid leukaemia, myeloproliferative neoplasm, chronic lymphocytic leukaemia and myelodysplastic syndrome (MDS) were included. Patients who were admitted to the hospital for other haematological diseases such as idiopathic thrombocytopenia, haemophilia, haemolytic anaemia, aplastic anaemia or coagulation disorders were excluded. The patients did not receive any prophylactic antibiotic treatment during the study period.


BSI was defined as the isolation of a bacterial or fungal organism from at least one blood culture.15 Multidrug resistance (MDR) is defined as the strain non-susceptible to at least one agent in ≥3 classes of antibiotics, including carbapenems, combinations of beta-lactams plus beta-lactamase inhibitors, cephalosporins, aminoglycosides and fluoroquinolones. The onset of BSI was defined as the date of collection of the first positive blood culture sample. For coagulase-negative staphylococci and common skin contaminants, at least two sets of positive blood cultures were required.15 16 Neutropenia was defined as an absolute neutrophil count <0.5×109/L.17 Appropriate initial antibacterial therapy was defined as receiving one or more antimicrobial agents with in vitro activity within 48 hours of the onset of BSI.15 The time from January 2018 to December 2019 and January 2020 to December 2021 were, respectively, defined as ‘before’ and ‘during’ the SARS-CoV-2 pandemic.

Patient and public involvement

Patients or the public were not involved in the design, conduct or reporting of this study.

Statistical analyses

Distributions of clinical characteristics between the two groups were performed using Mann-Whitney tests for continuous variables and χ2 tests for categorical variables. Variables with p<0.10 in the univariate analysis were subsequently included in the multivariate models. The multivariate analysis was performed using a Cox proportional hazard model. All p values were two-sided, and the significance was defined as p<0.05. Data were analysed by the R V.4.2.2.


Patient population

During the 4-year study period, a total of 599 episodes of BSI occurring in 22 717 inpatients were included in this cohort (online supplemental figure 1). 487 (81.3%) Gram-negative, 81 (13.5%) Gram-positive and 31 (5.2%) fungal organisms were isolated from blood cultures. The median age of patients was 40 years (range, 14–76 years) and 56.1% were men. AML (45.4%) was the most common haematological malignancy, followed by ALL (29.2%), MDS (11.02%), lymphoma (5.84%), plasma cell disorder (3.34%) and others (5.18%). Neutropenia was detected in 89.3% of patients with BSI. There were no significant differences in clinical features including age, gender, haematological malignancy types, disease status, neutropenia, duration of neutropenia, comorbidities and concurrent infection in BSI patients before and during the SARS-CoV-2 pandemic (p>0.05). Baseline characteristics of these BSI patients before and during the pandemic were displayed in table 1.

Supplemental material

Table 1

Demographic and clinical characteristics of BSI patients before and during the pandemic

Incidence and trend of BSI among the patients with HMs

Among them, 341 episodes of BSI occurred before the SARS-CoV-2 pandemic, and the frequency of total, Gram-negative and Gram-positive BSI was 2.90%, 2.49% and 0.27%, respectively. 258 episodes occurred during the pandemic, and the frequency was 2.35%, 1.77% and 0.44%, respectively, for total, Gram-negative and Gram-positive BSI (p=0.011, p<0.001 and p=0.027, respectively). BSI due to yeasts was relatively uncommon and remained stable during the study period. The trends in the aetiology of BSI among patients were shown in figure 1.

Figure 1
Figure 1

Trends in microbiology of bloodstream infections among inpatients with haematological malignancies in our cohort before (January 2018 to December 2019) and during (January 2020 to December 2021) the SARS-CoV-2 pandemic.

Susceptibility profiles of BSI isolates

293 (85.9%) Gram-negative and 32 (9.4%) Gram-positive organisms were isolated from blood cultures before the SARS-CoV-2 pandemic as well as 194 (75.2%) Gram-negative and 49 (19.0%) Gram-positive pathogenic organisms during the pandemic. Klebsiella pneumoniae was the most common Gram-negative organism, followed by Escherichia coli (25.9%), Pseudomonas aeruginosa (11.6%), Enterobacter cloacae (6.8%) and Acinetobacter baumannii (6.5%) before the pandemic and P. aeruginosa (20.6%), E. coli (20.1%), Stenotrophomonas maltophilia (5.2%) and E. cloacae (4.1%) during the pandemic (figure 2A). The frequency of MDR Gram-negative bacteria BSI was significantly increased during the pandemic (48.5% vs 57.7%, p=0.045, figure 2B), but carbapenem-resistant enterobacteriaceae (CRE) BSI was decreased (10.0% vs 2.7%, p<0.001, figure 2C). The susceptibility patterns of the common isolates were detailed in online supplemental table 1.

Supplemental material

Figure 2
Figure 2

The distribution and the incidence of different bloodstream infection isolates. The distribution of common Gram-negative isolates (A), the incidence of multidrug-resistant Gram-negative bloodstream infections (B) and carbapenem-resistant enterobacteriaceae bloodstream infections (C) and the distribution of Gram-positive bacteria (D) before and after the SARS-CoV-2.

Staphylococcus epidermidis (18.8%) was the most frequently isolated pathogen among Gram-positive bacteria before the pandemic, followed by Staphylococcus hominis (15.6%), Staphylococcus aureus (12.5%), Enterococcus faecium (9.4%) and Streptococcus pyogenes (9.4%). Different from the profiles of BSI isolates before the pandemic, the frequency of S. aureus BSI was markedly increased (26.5% vs 12.5%), followed by S. epidermidis (20.4%), E. faecium (12.2%), Enterococcus faecalis (8.2%) and Streptococcus pneumoniae (6.1%) during the pandemic (figure 2D). There were no significant differences before and during the pandemic on MDR Gram-positive bacteria BSI (34.4% vs 32.7%, p=0.872). Only one vancomycin-resistant enterococci BSI was found during the study period.

Outcomes and prognostic factors for mortality

A total of 109 episodes (18.2%) lead to 30-day mortality after the onset of BSI. Among them, 92 patients died from BSI, 11 from cerebral haemorrhage and 6 from disease progression. Of all patients with BSI, 72 patients (21.1%) died before the pandemic and 37 patients (14.3%) during the pandemic (p=0.043). We analysed factors associated with 30-day mortality concerning hosts, pathogens and treatments. In the univariate analysis, specific BSI pathogen, age, disease status, pulmonary infection, duration of neutropenia and shock were risk factors associated with 30-day mortality (online supplemental table 2). Various pathogens caused different mortality, especially CRE and Candida tropicalis which contributed more to 30-day mortality than other pathogens (figure 3). Multivariate analysis revealed that disease status, pulmonary infection and shock were independent predictors of 30-day mortality (table 2).

Supplemental material

Figure 3
Figure 3

The 30-day mortality for the major bacterial organisms due to bloodstream infections. CRE, carbapenem-resistant enterobacteriaceae; G-, Gram negative; G+, Gram positive; MDR G-, multidrug-resistant Gram negative; MDR G+, multidrug-resistant Gram positive.

Table 2

Multivariate analysis of prognostic factors of 30-day mortality in patients with BSI


To the best of our knowledge, our real-life, observational cohort study is among the first reports to compare the epidemiological trend and causative pathogens of BSI in patients with HMs before and during the SARS-CoV-2 pandemic. We found that the SARS-CoV-2 pandemic led to a significant shift in the incidence, microbiology and antibiotic resistance profiles of BSI.

During the study period, although Gram-negative organisms kept the predominant causes of BSI as reported,1 18 19 the incidence of Gram-negative BSI decreased during the pandemic, which may directly decrease the incidence of total organisms BSI. The top pathogens of BSI including K. pneumoniae, E. coli, E. cloacae, A. baumannii and S. maltophilia remained stable except P. aeruginosa in our study. P. aeruginosa BSI dramatically increased during the pandemic in our and others’ study.20–22 Potential reasons for the lower Gram-negative BSI incidence during the pandemic are unclear. A similar explanation is that rigorous and stringent hygiene measurements might have resulted in lower rates of Gram-negative BSI. Healthcare staff and patients were consistently wearing face masks, and contact restrictions during the pandemic reduced the number of visits to the patients by relatives and other non-healthcare professionals.23 The widespread use of broad-spectrum antibiotics during the pandemic might also cause a decline incidence of Gram-negative BSI but raise the MDR rates.24–26 For example, quinoline prophylaxis was found to be effective in reducing the frequency of BSI in HAMs, but also associated with a higher rate of bacterial resistance.27 This assumption is consistent with the fact that the MDR Gram-negative bacteria BSI markedly increased during the pandemic, but CRE BSI decreased in our study. The latter may be caused by antibiotic stewardship interventions in our department. Although stewardship strategies differ substantially in different institutions, it is effective in reducing drug resistance and improving outcomes.28 29 Since January 2019, the selection of empirical antibiotics changed every 3 months according to susceptibility profiles of Gram-negative organisms in our department to avoid the induction of CRE.

The incidence of Gram-positive BSI increased during the pandemic along with the change of main isolates. The top five Gram-positive pathogenic organisms were S. epidermidis, S. hominis, S. aureus, E. faecium and S. pyogenes before the pandemic, but the incidence of S. aureus and S. epidermidis BSI markedly increased followed by E. faecium, E. faecalis and S. pneumoniae during the pandemic. Similar results were reported by different studies in patients with non-haematological diseases.14 21 22 30 Those changes seen during the pandemic may reflect increased rates of contamination.30 During the pandemic, many of the practices that mitigate infection risk were adversely affected. Venepuncture and nursing processes might be more complex and difficult during the pandemic due to all staff performing procedures in unfamiliar personal protective equipment (PPE). Additionally, more than 75% of patients were performed central line for chemotherapy in our cohort, but routine central line-associated BSI prevention practices were altered during the pandemic. Due to contact restrictions and panic during the pandemic, patients could not get to the hospital in time for central line care.

Regarding survival, our multivariate analysis showed that malignancy type and state, pulmonary infection, previous antibiotic therapy and shock were independent predictors of 30-day mortality. No large differences were found before and during the pandemic. However, the mortality due to BSI was higher before the pandemic. The possible reason may be the difference in profiles of BSI isolates. Our data on pathogenic organisms and survival showed that Gram-negative organisms, especially CRE, caused more mortality than Gram-positive organisms BSI.

Our study has several limitations. The retrospective nature of this study may lead to data collection biases. All included patients in our cohort did not suffer from COVID-19 infection because patients with COVID-19 were admitted to the designated infectious disease department and hospital at that time. We were also unable to evaluate the unprecedented stress on hospitals, including staffing limitations, shortages in PPE and lack of ICU beds, which might affect patient outcomes. Notwithstanding these limitations, our study represents a large evaluation of BSI in patients with HMs before and during the SARS-CoV-2 pandemic in local China.

In conclusion, our data showed that the incidence of total and Gram-negative organisms BSI decreased, but Gram-positive BSI incidence increased in patients with HMs during the pandemic along with the changes of main isolates and susceptibility profiles. Although the 30-day mortality due to BSI was lower during the pandemic, the changes in the incidence, microbiology and antibiotic resistance profiles may give us insights into the impact of SARS-CoV-2 on BSI and provide an important reference for new infection prevention strategy in any future pandemics.

This post was originally published on