Risk factors for positive follow-up blood cultures in Gram-negative bacteremia among immunocompromised patients with neutropenia

Abstract

Introduction:

Gram-negative bacillary bloodstream infection (GN-BSI) is a frequent clinical challenge among immunocompromised hosts and is associated with a high mortality. The utility of follow-up blood cultures (FUBCs) for GN-BSI in this population, particularly in the setting of neutropenia, is poorly defined.

Methods:

We conducted a single-center, retrospective cohort study between the period of July 2018 and April 2022 to investigate the utility of FUBCs and delineate risk factors for positive cultures among neutropenic patients with monomicrobial GN-BSI. Univariate logistic regression was performed to assess risk factors associated with positive FUBCs.

Results:

Of 206 patients, 98% had FUBCs performed and 9% were positive. Risk factors for positive FUBCs included multidrug-resistant GN infection (OR 3.26; 95% CI 1.22–8.72) and vascular catheter source (OR 4.82; CI 1.76–13.17). Among patients lacking these risk factors, the prevalence of positive FUBCs was low (2.8%) and the negative predictive value was 92%. Those with positive and negative FUBCs had similar rates of all-cause mortality (16.7% vs 16.6%; p = 0.942) and microbiologic relapse (11.1% vs 6.0%; p = 0.401) within 90-days of treatment completion. However, positive FUBCs were associated with prolonged hospitalization and longer duration of antimicrobial therapy.

Conclusion:

Positive FUBCs were infrequent in neutropenic patients with GN-BSI, and their occurrence did not significantly impact mortality or microbiologic relapse. Risk factors for positive FUBCs included MDR-GN infection and vascular catheter source. Prospective studies will be necessary to elucidate the benefits and risks of FUBCs when managing GN-BSI in patients with underlying immune compromise.

Graphical Abstract

In neutropenic patients with monomicrobial Gram-negative bacteremia, 9% of follow-up blood cultures (FUBCs) were positive but did not impact 90-day mortality or microbiologic relapse. Risk factors for persistent bacteremia included vascular catheter source and multi-drug resistant Gram-negative infection.

INTRODUCTION

Gram-negative bacillary bloodstream infection (GN-BSI) is associated with high mortality (13–20%) [1, 2] and represents a common clinical challenge among immunocompromised patients [3]. Knowledge regarding optimal management of GN-BSI including choice of antimicrobials, duration of therapy, and need for follow-up blood cultures (FUBCs) to document clearance of infection is rapidly evolving. Obtaining FUBC is the current standard of practice in Staphylococcus aureus bacteremia as it has important prognostic implications [4]. However, utility of FUBC in GN-BSI remains controversial. While some studies demonstrate increased mortality associated with positive FUBC and improved outcomes with early intervention [5–7], others report minimal utility given low rates of positive cultures and minimal impact on clinical management [8–11].

A systematic review and meta-analysis of 15 non-randomized cohort studies supported obtaining FUBC in patients with GN-BSI because both obtaining FUBC and identifying those with positive FUBCs resulted in reduced mortality [12]. FUBCs likely provide opportunity for early detection of potential metastatic foci of infection and assess adequacy of source control, thereby providing opportunities for optimization of clinical and antimicrobial management. Unfortunately, patients with underlying immunocompromising conditions, including those with hematologic malignancy or hematopoietic stem cell transplant (HSCT), are underrepresented in these studies [12]. Limited reports evaluating role for FUBC in immunocompromised hosts have demonstrated low rates of positive FUBC (2.6–6%) with no impact on survival [8, 9]. Additionally, FUBC can lead to false positive cultures with isolation of bacterial contaminants, prolong hospital stay, and may lead to unnecessary antimicrobial exposure [9]. We therefore sought to investigate the utility of FUBC and identify risk factors for positive FUBC among patients with GN-BSI in the setting of neutropenia.

METHODS

We conducted a single-center, retrospective cohort study of adult patients with neutropenia (polymorphonuclear neutrophil (PMN) count ≤1.0×10⁹/L) at the time of or within 24 hours of a monomicrobial GN-BSI managed at our tertiary care center between the period of July 1, 2018 to April 30, 2022. The study was deemed exempt by our Institutional Review Board (IRB# 21–010370).

Definitions.

Detailed study methodology, including inclusion/exclusion criteria and definitions were previously described [13]. Relevant study definitions regarding sources of BSI (including catheter-related bloodstream infection (CRBSI)), follow-up blood cultures, positive FUBCs, persistent bacteremia, and time to clearance of BSI are further highlighted in Appendix Table 1. Source of bacteremia was defined following clinical and radiographic evaluation by an Infectious Diseases-trained physician or the primary care team. Owing to clinical challenges among patients with neutropenia, stringent definitions were used and multiple or unknown sources were recorded if a clear source was not discernible.

Data collection and outcomes.

Individual patient data including demographics, clinical comorbidities [14], microbiology of GNB and antimicrobial susceptibility, source of bacteremia [15], and management were obtained from electronic medical records, collected, and stored in a secure REDCap electronic data capture tool hosted at the Mayo Clinic [16, 17]. Among patients who had FUBCs completed, the frequency and timing of FUBC, source control measures and timing of effective source control, effectiveness of empiric and definitive antimicrobial therapy, and antibiotic resistance was evaluated. Primary outcomes included all-cause mortality and microbiologic relapse within 90-days following therapy completion. Secondary outcomes included incidence of Clostridioides difficile infection (CDI) [18] or rates of MDR-GN infection [19] within 90-days following therapy completion.

Statistical analysis.

Baseline data are expressed as median and interquartile range (IQR) for continuous variables and counts and percentages for categorical variables. Kruskal-Wallis and Pearson χ² (or Fisher exact) tests for continuous and categorical variables, respectively. Statistical significance was indicated by a 2-tailed P < 0.05. Risk factors for positive FUBC were evaluated using univariate logistic regression. All analyses were conducted using R, version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

During the study period, we identified 206 patients with concurrent monomicrobial GN-BSI and neutropenia, of whom 201 (98%) had FUBCs completed at least 24 hours after initial identification to document clearance of infection. Majority of patients (95%) had the first set of follow-up cultures obtained within 72 hours of index blood cultures (Table 1). 111 patients (55%) had clearance of blood cultures on day one following initial bacteremia diagnosis. 18 patients (9%) had positive FUBCs with median time to blood culture clearance of 3 days (interquartile range [IQR] 2.25 – 4.75) in comparison to 1 day (IQR 1–2) among those with negative FUBCs.

Table 1:

Baseline demographics, microbiologic profile, and outcomes of GN-BSI among neutropenic patients with positive and negative FUBCs ^A,B.

Variable	Negative FUBC (n=183)	Positive FUBC (n=18)	p value
Age (years)	64 (57–70)	63 (51–71)	0.5921
Gender: Female	56 (30.6%)	4 (22.2%)	0.4592
Charlson Comorbidity Index (CCI)	6 (4–7)	5 (4–6.75)	0.3791
Pitt Bacteremia Score	0 (0–1)	0 (0–1)	0.4571
Hospital admission	169 (92.3%)	17 (94.4%)	0.7472
Hospital length of stay (days)	8 (4–23)	16 (8–37)	0.027 1
Immunocompromising Condition	183 (100%)	17 (95%)	0.6893
Type of immunocompromising condition
Hematopoietic cell transplant	87 (47.5%)	11 (64.7%)
Hematologic malignancy	64 (35.0%)	6 (35.3%)
Solid organ malignancy	19 (10.4%)	0 (0.0%)
Solid organ transplant	9 (4.9%)	0 (0.0%)
HIV/AIDS	1 (0.5%)	0 (0.0%)
Primary immunodeficiency	1 (0.5%)	0 (0.0%)
Other	2 (1.1%)	0 (0.0%)
Duration of neutropenia (days)	10.5 (7–20)	10 (7–13.5)	0.3702
Profound neutropenia	153 (83.6%)	13 (72.2%)	0.2242
Absolute neutrophil count	0 (0–0)	0 (0–203)	0.5872
Microbiologic cause of BSI			0.0922
Escherichia coli	94 (51.4%)	6 (33.3%)
Klebsiella species	20 (10.9%)	3 (16.7%)
Pseudomonas aeruginosa	41 (22.4%)	4 (22.2%)
Enterobacter species	9 (4.9%)	1 (5.6%)
Citrobacter species	0 (0.0%)	1 (5.6%)
Stenotrophomonas maltophilia	2 (1.1%)	1 (5.6%)
Other c	8 (4.4%)	2 (11.1%)
MDR-GNB isolated	43 (23.5%)	9 (50.0%)	0.014 2
Source of BSI D
Vascular catheter	45 (24.6%)	11 (61.1%)	< 0.001 2
Neutropenic enterocolitis/gut translocation	73 (39.9%)	4 (22.2%)	0.2822
Intra-abdominal	30 (16.4%)	0 (0.0%)	0.0632
Pulmonary E	11 (6.0%)	2 (11.1%)	0.4012
Urinary	15 (8.2%)	2 (11.1%)	0.6722
Skin and soft tissue	4 (2.2%)	1 (5.6%)	0.3812
Unknown F	12 (6.6%)	0 (0.0%)	0.2632
Source control achieved
Yes	51 (78.5%)	11 (91.7%)	0.1072
No	14 (21.5%)	1 (8.3%)
Not applicable	118	6
Time to appropriate antibiotic initiation (days)	0 (0–1)	1 (0–1.75)	0.028
Time to blood culture clearance (days)	1 (1–2)	3 (2.25–4.75)	< 0.001 1
Timing of initial FUBC
Day 1	111	8
Day 2	58	10
Day 3	11	0
Day 4	3	0
Outcomes at 90-days following therapy completion
Mortality	31 (16.9%)	3 (16.7%)	0.9422
Microbiologic relapse	11 (6.0%)	2 (11.1%)	0.4012
CDI incidence	8 (4.4%)	1 (5.6%)	0.5962
MDR-GN emergence	5 (2.7%)	3 (16.7%)	0.1182

^A.Abbreviations and definition: MDR-GN – multi-drug resistant gram negative (defined as Enterobacteriaceae, Pseudomonas, or Acinetobacter species resistant to at least 1 drug in 3 drug categories); CCI – Charlson comorbidity index; BSI – bloodstream infection; CDI – Clostridioides difficile infection;

^B.Values represent median (quartile 1, quartile 3) for continuous variables and frequency (percentage) for categorical variables. P values calculated by

¹Wilcoxon rank sum tests,

²Pearson χ².

^C.Other species included Serratia, Citrobacter, Bacteroides, Leptotrichia, Capnocytophaga, Acinetobacter, and Elizabethkingia species

^D.The source of infection columns do not summate to total number of patients as a subset of patients may have had multiple potential sources of infection.

^E.Pulmonary source included hospital and ventilator acquired pneumonia, pulmonary abscess, and empyema.

^F.Unknown source was noted if no work-up to identify source was pursued or no clear source was identified despite appropriate workup.

We further evaluated clinical and microbiologic differences among patients with positive and negative FUBCs to delineate risk factors for positive FUBCs (Table 1). Baseline characteristics including patient comorbidities, immunocompromising condition, severity of infection, and need for hospital admission were similar between patients with negative or positive FUBC. The primary sources of BSI included neutropenic enterocolitis/gut translocation (38%) and vascular catheter (28%). When feasible, effective source control was achieved at similar rates among those with positive (91.7%) and negative (78.5%) FUBCs (Table 1).

The most common organisms causing GN-BSI included Escherichia coli (50%), Pseudomonas aeruginosa (22%), and Klebsiella spp. (11%); no clear association between microbiology and risk of positive FUBCs was identified. Interestingly, however, rates of MDR-GN infection were significantly higher among those with positive FUBC (50% versus 23.5%). Patients with MDR-GN received appropriate empiric therapy significantly less frequently than those without (44.2% vs 91.9%, p <0.001). Additionally, the time to initiation of appropriate antimicrobial therapy was delayed in those with positive FUBCs (Table 1). Lastly, patients with positive FUBC received longer duration of definitive therapy with median 15.5 days (IQR 15–17) compared to those with negative FUBC treated for a median of 14 days (IQR 10–15) (Table 1).

Risk factors for positive FUBC.

Univariate logistic regression demonstrated MDR-GN infection (OR 3.26; 95% Confidence Interval [CI] 1.22–8.72; p=0.019) and vascular catheter source (OR 4.82; 95% CI 1.76–13.17; p=0.002) as important risk factors for positive FUBCs (Appendix Table 2). In the absence of these factors, the prevalence of positive FUBCs was low (2.8%; 3 of 106 patients) and carried a high negative predictive value (92%) within this cohort.

To further delineate characteristics of patients with positive FUBCs (n=18), we conducted a narrative review of their clinical course and management (Table 2). Vascular catheter was the most common source of infection (11 of 18 patients), with a high rate of MDR-GN infection (50%) observed. Among those with positive FUBC in setting of CRBSI, source control was attained in 10 of 11 patients. However, delay in line removal (defined as line removal ≥ 72 hours after index bacteremia) was observed in 7 of 11 patients (Table 2), with line removal prompted by identification of persistent bacteremia.

Table 2:

Narrative review of clinical course and management of patients with positive FUBCs ^A.

ID	Age	CCI	Immunocompromising conditions	GN pathogen	Time to clear	Source	Source Ctrl.	MDR-GN	Empiric Tx	Definitive Tx	Tx. Days	90-day mortality	90-day relapse
1	62	4	N/A	Pseudomonas aeruginosa	2	CRBSI	Yes @ Day 1	No	Piperacillin-tazobactam Effective: Yes	Piperacillin-tazobactam	18	No	No
2	72	6	Mantle cell lymphoma Active chemotherapy: Yes	Klebsiella variicola	2	Presumed gut translocation	N/A	No	Cefepime Effective: Yes	Ceftriaxone Gentamicin-line lock	15	No	No
3	78	7	DLBCL Active chemotherapy: No	Pseudomonas aeruginosa	2	Pneumonia	N/A	No	Cefepime Effective: Yes	Cefepime	16	No	No
4	72	6	Follicular lymphoma Active chemotherapy: Yes	Escherichia coli	2	Pyelonephritis	N/A	Yes	Cefepime Effective: No	Ertapenem	15	No	No
5	65	9	Multiple myeloma Active chemotherapy: Yes	Klebsiella pneumoniae	3	CRBSI	Yes @ Day 2	Yes	Cefepime Effective: No	Ceftazidime/ avibactam	17	Yes	No
6	25	3	Allogenic SCT due to AML	Pseudomonas aeruginosa	3	Fournier’s gangrene	Yes	No	Piperacillin-tazobactam Effective: Yes	Meropenem	15	Yes	No
7	59	4	Autologous SCT due to Multiple myeloma	Pseudomonas aeruginosa	3	CRBSI	Yes @ Day 3	No	Cefepime Effective: Yes	Cefepime	15	No	No
8	75	6	Autologous SCT due to Multiple myeloma	Elizabethkingia meningoseptica group	4	CRBSI	Yes @ Day 4	Yes	Piperacillin-tazobactam Effective: Yes	Levofloxacin	10	No	No
9	52	5	Acute lymphocytic leukemia Active chemotherapy: Yes	Escherichia coli	3	Neutropenic enterocolitis	N/A	Yes	Cefepime Effective: No	Meropenem	17	Yes	No
10	62	5	Autologous SCT due to Multiple myeloma	Escherichia coli	3	CRBSI	Yes @ Day 6	Yes	Cefepime Effective: No	Ertapenem	13	No	No
11	71	8	Allogenic SCT due to B-ALL	Citrobacter freundii	3	CRBSI	No*	Yes	Cefepime Effective: Yes	Ertapenem	16	No	No
12	71	7	Autologous SCT due to Multiple myeloma	Escherichia coli	4	Presumed gut translocation Pneumonia	N/A	Yes	Meropenem Effective: Yes	Meropenem	17	No	No
13	64	5	Autologous SCT due to DLBCL	Stenotrophomonas maltophilia	4	CRBSI	Yes @ Day 5	No	Cefepime Effective: No	Levofloxacin	7	No	No
14	43	3	Allogenic SCT due to AML	Enterobacter cloacae complex	5	CRBSI	Yes @ Day 2	No	Cefepime Effective: Yes	Ertapenem	20	No	Yes
15	51	4	Autologous SCT due to DLBCL	Escherichia coli	5	CRBSI	Yes @ Day 5	No	Cefepime Effective: Yes	Cefepime	17	No	No
16	28	3	Allogeneic SCT due to aplastic anemia	Achromobacter xylosoxidans	5	CRBSI	Yes @ Day 2	Yes	Meropenem Effective: Yes	Piperacillin-tazobactam	15	No	No
17	25	3	T-cell lymphoma Active chemotherapy: Yes	Klebsiella pneumoniae	6	Urinary	N/A	Yes	Ceftazidime/ avibactam Effective: No	Colistin, Tigecycline, Cefiderocol	15	No	Yes
18	65	7	Allogeneic SCT due to MDS	Escherichia coli	6	CRBSI Presumed gut translocation	Yes @ Day 6	No	Cefepime Effective: Yes	Ertapenem	21	No	No

^A.Abbreviations and definitions: N/A – not applicable; CCI – Charlson comorbidity index; GN -Gram-negative; MDR – multidrug resistant; Tx – treatment; CRBSI – catheter related bloodstream infection; DLBCL – diffuse large B cell lymphoma; SCT – stem cell transplantation; AML – acute myelogenous leukemia; ALL – acute lymphocytic leukemia;

^*No – Lack of source control was due to ongoing need for line due to poor vascular access and salvage was attempted using gentamicin lock therapy

Outcomes.

The primary outcomes of all-cause mortality or microbiologic relapse within 90-days of treatment completion were not significantly different between those with positive and negative FUBCs. Similarly, rates of CDI and emergence of resistance was similar between the two cohorts. Patients with positive FUBC had significantly longer median hospital length of stay than those with negative cultures (16 days versus 8 days) (Table 1).

DISCUSSION

Both obtaining FUBCs and early identification of persistent bacteremia through positive cultures have been demonstrated to be a prognostic indicator for improved clinical outcomes in patients with GN-BSI [5, 7, 12]. However, utility of FUBC is challenging to study among immunocompromised hosts as most clinicians err on the side of obtaining FUBCs due to complex patient comorbidities and high mortality with GN-BSI. In keeping with this, follow-up cultures were obtained in 98% of patients within our study and ranged from 67–84% in prior studies evaluating patients with underlying malignancy or HSCT [8, 9]. Fortunately, the rate of positive FUBC was low (9%) and comparable to prior reports (2.6–6%).

From a microbiologic standpoint, no clear association between type of GN organism and positive FUBC was identified in this study despite anecdotal concern for persistent bacteremia with non-fermenting GNB like Pseudomonas or Stenotrophomonas spp [20]. Prior reports have similarly demonstrated that <10% of patients with Pseudomonas BSI have positive FUBCs and source control is an important determinant of persistent BSI [21–23]. Rather than the organism, antibiotic resistance appears to be an important predictor of positive FUBC in our analysis, with MDR-GN infection associated with a high risk of positive FUBCs [24]. This is particularly critical among immunocompromised hosts as both incidence of MDR-GN infection and associated mortality rate is higher than the general population [13, 25, 26].

Among neutropenic patients with GN-BSI, the most common sources of GN-BSI included intra-abdominal (enterocolitis or gut translocation) and CRBSI. Vascular catheter source was identified to be an independent risk factor for positive FUBCs, as previously described [24]. Interestingly, among patients with positive FUBCs in the setting of CRBSI, source control was achieved in 91%. However, a delay in time to line removal was noted in the majority of patients. This delay is not entirely surprising as persistent bacteremia is frequently in itself an indication for line removal. However, this highlights the potential importance of advocating for line removal, particularly in the setting of MDR-GN BSI. While appropriate source control is critical to management of GN-BSI, timing of source control is an underappreciated factor and requires consideration in future studies evaluating utility and risk factors for persistent BSI.

Fortunately, positive FUBCs did not significantly impact post-treatment all-cause mortality or microbiologic relapse. However, as described in several studies [9, 11, 27], FUBCs are not without consequence and were noted to result in prolonged hospital length of stay and extend the total duration of antibiotic exposure. This is particularly relevant as cumulative antibiotic exposure significantly increases the risk of CDI [28] and development of antimicrobial resistance [29].

Limitations.

There are several limitations to note when interpreting the results of this study. First, the retrospective nature of this observational study limits the ability to capture potential confounding factors that may impact the risk for persistent GN-BSI. Additionally, we excluded patients with GN-BSI who did not complete therapy due to death or transition to hospice. Consequently, the exclusion of patients with early mortality may falsely underestimate the impact of persistent bacteremia or positive FUBCs on all-cause mortality. Thus, these results may only apply to those who complete planned treatment. We also noted challenges to defining source of bacteremia among patients with neutropenia particularly in the setting of impaired mucosal barriers and presence of vascular catheters; a subset of patients had multiple or unknown sources of infection. Additionally, timing of FUBCs was not standardized, but the majority of patients had initial cultures within 3 days of initial diagnosis. While this is a limitation, it is likely to overestimate the median time to clearance among those with negative FUBC, without impacting those with positive FUBCs. We also emphasize the descriptive nature of Table 1 wih p values provided solely to highlight differences between those with positive and negative FUBCs, with additional analysis limited by low event rate. Consistent with this, we conducted a univariate logistic regression as additional adjustment was limited by low incidence of positive FUBCs.

CONCLUSION

Within this study, we observed a low rate of persistent GN-BSI in the setting of neutropenia. In the setting of appropriate source control and antimicrobial therapy, we noted that positive FUBCs did not significantly impact mortality or microbiologic relapse. Risk factors for positive FUBCs included MDR-GN infection and vascular catheter source, with the latter being driven by a delay in time to source control. Prospective studies investigating role for FUBCs in patients with immunocompromised status will be necessary to elucidate the risks and benefits of this approach on management of GN-BSI.

Appendix Table 1:

Definitions of clinical variables ^A

Source of bacteremia	Source of bacteremia was defined by clinical and radiographic evaluation by an Infectious Diseases-trained physician or the primary care team, with adherence to the Centers for Disease Control and Prevention criteria [30].
Intra-abdominal source	Due to the nature of mucosal disruption in neutropenic hosts, intra-abdominal source included the following: • Mucositis/neutropenic enterocolitis (based on clinical and radiographic evidence on abdominal imaging) • Presumed gut translocation (based on clinical presentation, but negative abdominal imaging or lack of further imaging) • Intra-abdominal as per standard CDC criteria, not meeting above definitions.
Catheter-related bloodstream infection (CRBSI)	CRBSI was defined as per IDSA guidelines [31] with inclusion of both catheter-related and catheter colonization as vascular catheter source. • Catheter-related infection was defined by isolation of causative pathogen from both intravascular device and peripheral blood culture with differential time to positivity, with clinical manifestation of infection, in the absence of alternate source for BSI. • Catheter colonization was defined as growth on organism in catheter cultures only, without positive peripheral cultures.
Follow-up blood culture (FUBC)	FUBC was defined as any blood culture [at least one aerobic and one anaerobic bottle] obtained at least 24 hours after and within 7 days of initial positive blood culture during the same hospitalization [6, 9]. Date of initial blood cultures was defined as day 0. The date of FUBCs were collected in order to establish the time from index blood cultures to positive or negative FUBCs.
Positive FUBC	Defined as a FUBC noted to be positive with the same GNB as the index bacteremia at least 24 hours after index blood cultures.
Persistent bacteremia	Defined as identification of positive FUBCs with same GNB > 72 hours after index blood cultures, irrespective of the number of sets of FUBCs obtained prior to 72 hours (i.e. if a patient obtained first set of blood cultures between day 4–7, but noted to be positive, this was considered persistent bacteremia)
Time to clearance	Defined as time between index positive culture (day 0) to first negative follow-up blood culture (quantified in days).
Source control	Defined by removal of any infected devices or hardware, surgical or non-surgical drainage of abscess or fluid collections, as well as resolution of any obstructions including pancreatic, biliary, or urinary within 72 hours of bacteremia.
Appropriateness of antimicrobial therapy	Antimicrobial therapy was deemed appropriate if the GN-organism was susceptible to antibiotic based on in vitro AST. Time to initiation of appropriate antibiotic thearpy was identified and recorded within this study.
Multi-drug resistant organism	MDR-GN organism was defined as Enterobacteriaceae, Pseudomonas species, or Acinetobacter species resistant to at least 1 drug in 3 of the following drug categories: extended-spectrum cephalosporins, fluoroquinolones, aminoglycosides, carbapenems, piperacillin/tazobactam, and ampicillin-sulbactam (specifically for Acinetobacter species) [32].

Abbreviations: GN-BSI: Gram-negative bloodstream infection; FUBC: follow-up blood culture; CDC: Centers for Disease Control and Prevention; AST: antimicrobial susceptibility testing; MDR: multi-drug resistant; IDSA – Infectious Diseases Society of America

Appendix Table 2:

Unadjusted univariate logistic regression model for risk of positive FUBC in neutropenic patients with GN-BSI

Variable	OR (95% confidence interval)	p value
Age (years)	0.98 (0.95–1.02)	0.327
Charlson Comorbidity Index (CCI)	0.87 (0.65–1.16)	0.351
Prophylactic antibiotic therapy	0.87 (0.35–3.44)	0.903
MDR-GN infection	3.26 (1.22–8.72)	0.019
CRBSI source of infection	4.82 (1.76–13.17)	0.002
Appropriate empiric therapy used	0.47 (0.17–1.35)	0.161