Early Onset Bacteremia in Kidney Transplant Recipients in the Current Era

ABSTRACT

Background

Bloodstream infection (BSI) after kidney transplantation increases the risk of graft loss and mortality. We aimed to identify contemporary risk factors and outcomes of early BSI (eBSI) at our transplant center.

Methods

We conducted a retrospective chart review of adult kidney transplant recipients (KTRs) from 2019 to 2023. Those with the eBSI (within 90 days of KTR) were matched by age and gender to controls. Risk factors were identified using multivariable logistic regression. All KTRs surviving to 90 days, were included in graft and patient survival analysis.

Results

Among 1408 KTRs, 75 (5.3%) developed eBSI, at a median of 21 days (IQR: 12–42.5). On multivariable analysis, each doubling of hospital duration was independently associated with a 49% increase in the risk of eBSI (OR: 1.49, 95% CI: 1.19–1.91; p < 0.001), and each doubling of postsurgical drain retention duration was independently associated with a 68% increase in the risk of eBSI (OR: 1.68, 95% CI: 1.17–2.48; p 0.006). eBSI was associated with 1‐year mortality (aHR: 3.32, 95% CI: 1.51–7.28; p 0.003) and 3‐year graft failure (aHR: 2.78, 95% CI: 1.31–5.93; p 0.008).

Conclusion

eBSI was associated with graft failure and mortality. The donor and recipient factors identified highlight the need for further studies to better characterize their impact on eBSI.

Abbreviations

BSI

blood stream infection

DBD

donation after brain death

DCD

donation after circulatory death

DGF

delayed graft function

eBSI

early blood stream infection

KDPI

kidney donor profile index

KTR

kidney transplant recipients

OR

operating room

RRT

renal replacement therapy

SHK

simultaneous heart kidney

SLK

simultaneous liver‐kidney

1. Introduction

Bacterial blood stream infections (BSIs) are reported to occur in 8.3% of kidney transplant recipients (KTRs) in the first year of transplant [1]. Gram negative bacteria, including Pseudomonas and enteric organisms, are most frequently reported [1, 2, 3]. The rising rate of antibiotic resistance poses a challenge to empiric management and successful treatment, with many studies demonstrating adverse outcomes for KTRs infected with drug‐resistant strains [2, 4, 5, 6].

BSI have been associated with elevated death censored graft failure and all‐cause mortality in KTRs [7, 8]. Risk factors identified for BSI include older age > 60 years, longer duration of renal replacement therapy (RRT), recipient diabetes, delayed graft function (DGF), and elevated donor age [2, 3, 8, 9]. Such studies report risk factors and outcomes for BSI that occur for months to years following kidney transplant [8, 10, 11]. However, most BSI occur early posttransplant and may highlight important donor and surgical risk factors that affect later transplant outcomes [2].

We sought to describe early BSI (eBSI) within 90 days following KT and to evaluate the risk factors and outcomes in recent years at our center.

2. Material and Methods

2.1. Patient Population

We conducted a retrospective cohort study of adults (age > 18) receiving a kidney transplant at the Cleveland Clinic Main Campus between January 1, 2019, and December 31, 2023. Patients with the first bacterial BSI within 90 days of transplant (eBSI) were identified through microbiologic data records and chart review. This study was reviewed and approved by the Institutional Review Board (IRB) IRB# 24–1067. Clinical trial registration was not applicable as this study is a retrospective chart review and does not involve an intervention.

Bacteremia was defined as either a single positive blood culture with a gram‐negative microorganism, Staphylococcus aureus, S. lugdunensis or Enterococcus species, or two positive blood cultures with the same gram‐positive microorganism, other than the mentioned, in the setting of symptomatic infection. Polymicrobial bacteremia was defined as presence of more than one pathogenic microorganism in blood cultures drawn within 24 h (excluding common skin contaminants). Extended spectrum beta lactamase (ESBL) producing bacteria was identified through molecular testing. Carbapenemase producing bacteria was defined by detection of one out of five carbapenemase genes on the tested sample when testing was triggered by the resistance profile.

Per kidney transplant protocols, all patients received induction with anti‐thymocyte globulin (ATG) or basiliximab. Maintenance immunosuppression typically included tacrolimus, mycophenolate mofetil, and prednisone. Belatacept was used as a calcineurin inhibitor sparing agent in selected patients. Perioperative antibiotics directed against gram positive skin flora include cefazolin, or in cases of allergic reaction to cephalosporins, vancomycin. Infectious disease physicians are consulted for all donor urine culture growth (excluding coagulase negative staphylococci) to determine if recipient treatment is indicated. Urine cultures with growth other than coagulase negative staphylococci are typically treated with 5–7 days of directed antimicrobials. During the transplant surgery, a double J ureteral stent was inserted and maintained for 6 weeks. All had a subcutaneous Jackson–Pratt (JP) drain placed, excluding those who underwent robotic transplantation, with removal decided by the transplant surgeon based on drain output (when drain output is consistently < 30 mL/day). Finally, a Foley catheter was maintained for at least 3 days, but longer based on intraoperative findings of bladder muscular health. At our institution, urine and blood cultures are obtained at the clinician discretion of infection is suspected. Routine cultures are not obtained per protocol.

2.2. Study Design

We described the timing and characteristics of the first eBSI, including organism identified, and resistance detected by genetic and in vitro antimicrobial susceptibility methods. The source of infection was determined by the transplant infectious disease physician consulting on the case, with validation by the study investigators from data acquired from the electronic medical record. In instances where growth of the same microorganism was found in the urine and fluid from perinephric fluid aspiration, the source was determined to be perinephric fluid related.

Recipient and donor characteristics were reported for those with and without eBSI. Recipient pretransplant characteristics included age at transplant, gender, race, comorbidities (diabetes, autoimmune disease needing immunosuppressants, HIV), genitourinary disease (obstructive uropathy, presence of recurrent stones, benign prostatic hyperplasia, and prior surgical changes to the genitourinary tract anatomy), time and modality of dialysis prior to transplant, and underlying kidney disease. Transplant characteristics included duration of index admission, return to the operating room (OR) prior to eBSI or until 90 days for those without eBSI, number of OR take back, cold ischemic time, duration of urinary catheter, ureteral stent, and surgical drains. Posttransplant elements included diagnosis of transplant rejection (biopsy proven or elevated donor‐specific antibodies) and DGF (need for RRT within 7 days from transplantation), duration and route of dialysis if DGF, presence of central venous catheters and exposure to critical care units 14 days prior to eBSI or up to 90 days for those without BSI.

Donor characteristics included donor type (living, donation after circulatory death [DCD], donation after brain death [DBD]), age, kidney donor profile index (KDPI), duration of terminal hospitalization, urine and blood culture results, hepatitis C virus status, SARS‐COV2 screening result, presence of donor acute kidney injury (defined as creatinine above 1.1 mg/dL), donor needing RRT, donor needing extracorporeal membrane oxygenation (ECMO), and “open organ offer” as communicated by the organ procurement organization. Donor elements were collected through DonorNet review.

Risk factors for eBSI were sought using a case‐control study. Cases were those who had eBSI within 90 days of transplant. Controls were selected from those who survived 90 days after transplant without a BSI, by propensity score matching on age and gender. Two best‐matched controls were selected for each case.

All KTRs at the Cleveland Clinic in the study period, who survived at least 90 days following transplant, were also evaluated for survival and graft function. Graft failure was defined as the need for maintenance dialysis or listing for re‐transplantation. Estimated glomerular filtration rate (eGFR) was calculated using the CKD‐EPI 2021 creatinine equation based on serum creatinine, age and sex as parameters and was collected from lab results at the 1‐year mark after transplant, or the closest lab result within 30 days before or after the 1‐year mark.

2.3. Statistical Analysis

Risk factors for eBSI were identified using multivariable logistic regression. The initial multivariable model included all variables with univariable associations with eBSI with p‐values < 0.20. Variable selection was then done with stepwise backward elimination, forcing hospital length of stay into the model, until only variables that remained significant at a level of 0.05 remained.

Multivariable Cox proportional regression models were used to compare the occurrence of death within 1 year and graft failure within 3 years between those with and without eBSI. Graft failure up to 3 years was assessed because of plateauing of graft failure only after 3 years on cumulative incidence curves. Variables with univariable associations with the outcome with p‐values < 0.2, were included in the initial multivariable models. Variable selection was then done using stepwise backward elimination until only variables that remained significant at a level of < 0.05 remained in the final models. DGF was forced into the graft failure model.

Estimated GFR at 1 year was compared for those with and without eBSI using multivariable linear regression. Variable selection was done in a similar manner to that for the Cox proportional hazards regression model for graft failure.

Statistical analyses were done using R version 4.4.1 [12].

3. Results

Amongst 1408, adult first time KTRs, 75 (5.3%) had eBSI. The characteristics of the recipients with eBSI and their age and gender matched controls are shown in Table 1. The median time to first eBSI was 21 days (IQR: 12–42.5; Figure S1).

TABLE 1.

Baseline characteristics of those with eBSI and their matched controls.

Variable	Cases (n = 75)	Controls (n = 150)	p‐value
Pretransplant elements
Age (mean [SD])	59 (9.6)	59 (9.6)	1
Gender
Race			0.62
Black	27 (36.0)	46 (30.7)
White	43 (57.3)	96 (64)
Other	5 (6.6)	8 (5.3)
Diabetes	37 (49.3)	68 (45.6)	0.7
Lower GU disease	25 (33.8)	36 (24.5)	0.19
Anuria prior	24 (32.0)	35 (23.6)	0.24
Duration of dialysis prior to transplant (days) (mean [SD])	1148.2 (1215.7)	8291.7 (72 986.0)	0.45
Peritoneal dialysis	11 (18.3)	31 (30.4)	0.13
Transplant elements
Transplant year			0.14
2019	5 (6.7)	26 (17.0)
2020	12 (20.0)	21 (14.0)
2021	12 (16.0)	22 (15.0)
2022	19 (25.0)	45 (30.0)
2023	24 (32.0)	36 (24.0)
Delayed graft function	26 (34.7)	25 (16.7)	0.004
Cold ischemic time (mean [SD])	23.34 (9.8)	18.6 (10.6)	0.001
Multi‐organ transplant	15 (20.0)	10 (6.7)	0.006
SLK	11	7
SHK	4	1
Double kidney		2
ATG induction	59 (78.7)	111 (74.0)	0.55
Posttransplant elements
Duration of index admission (mean [SD])	23.7 (48.2)	4.91 (6.1)	< 0.001
Foley duration (mean [SD])	13.96 (10.47)	9.48 (4.9)	< 0.001
Drain duration (mean [SD])	21.71 (13.4)	15.5 (10.3)	< 0.001
Stent duration (mean [SD])	32.4 (17.4)	30.9 (30.5)	0.71
Rejection prior to eBSI	4 (5.3)	17 (11.3)	0.22
Donor related elements
Donor type			0.05
DBD	38 (50.7)	64 (42.7)
DCD	32 (42.7)	58 (38.7)
Living	5 (6.7)	28 (18.7)
Donor admission duration (mean)	8.7 (10.0)	6 (5.1)	0.015
KDPI	0.42 (0.2)	0.45 (0.2)	0.42

Note: Data presented as n (%).

Abbreviations: ATG, antithymocyte globulin; DBD, donation after brain death; DCD, donation after circulatory death; eBSI, early blood stream infection; GU, genitourinary; KDPI, kidney donor profile index; SD, standard deviation; SHK, simultaneous heart kidney; SLK, simultaneous liver‐kidney.

3.1. Microorganisms, Antimicrobial Resistance, and Sources

Of the 75 episodes of first eBSI, four were polymicrobial, yielding a total of 79 bacterial isolates (Figure 1). Gram negative bacilli (GNB) accounted for 80% of episodes, with Klebsiella spp. (n = 20, 25%), Escherichia coli (n = 17, 22%), and Pseudomonas aeruginosa (n = 17, 22%) predominating. Antimicrobial resistance occurred in 32/79 (40%) of eBSI, with ESBL producing strains of Enterobacteriaceae occurring in 18/45 (40%). One E. coli strain was carbapenemase (KPC) producing. Pseudomonas isolates were resistant to piperacillin‐tazobactam 18%, Cefepime 18%, meropenem 20%, and ciprofloxacin 47%.

FIGURE 1.

Microorganisms isolated with their respective resistance profile.

For gram positive organisms, Enterococcus faecium and E. faecalis accounted for 10 BSI (13%) followed by S. aureus, and coagulase negative Staphylococcus. All E. faecium isolates were vancomycin resistant, with one also being daptomycin resistant. Recurrence of bacteremia with the same microorganism within 30 days of the first positive blood culture occurred in 13 (17%) patients.

The most common source for BSI was urinary (n = 43, 57%), followed by transplant perinephric fluid collection (n = 10, 13%) and pneumonia (n = 5 7%). The source was not identified in 16% of the episodes. None of the patients suffered a surgical site infection. All pneumonia patients were on mechanical ventilation.

3.2. Risk Factors for eBSI

Risk factors identified on univariable and multivariable analysis on the age and gender‐matched dataset are shown in Table 2.

TABLE 2.

Risk factors for eBSI on univariable and multivariable analyses.

	Univariable analysis				Multivariable analysis
Variable	OR	95% CI	p‐value	OR	95% CI	p‐value
Peritoneal dialysis	0.51	0.23–1.1	0.09	—	—	—
Transplant year (ref: 2019)
2020	3.71	1.22–12.97	0.03	—	—	—
2021	2.84	0.9–10.08	0.09	—	—	—
2022	2.2	0.78–7.24	0.16	—	—	—
2023	4.25	1.24–11.37	0.03	—	—	—
Cold ischemic time	1.046	1.017–1.08	0.002	—	—	—
Multi‐organ	3.50	1.5–8.47	0.004	—	—	—
Index admission duration a	1.99	1.47–2.79	< 0.001	1.78	1.29–2.54	< 0.001
Foley catheter duration a	3.73	2–7.37	< 0.001	—	—	—
Surgical drain duration a	2.59	1.55–4.5	< 0.001	2.12	1.26–3.71	0.01
Delayed graft function	2.65	1.4–4.8	0.003	—	—	—
Ureteral stent duration	1.28	0.78–2.61	0.43
CNI based maintenance b	5.29	1.11–37.5	0.05	—	—	—
CVL after transplant	0.94	0.53–1.67	0.85
Critical care exposure	1.85	0.97–3.49	0.06
Return to OR	1.77	0.95–3.28	0.07
Number of OR return	1.2	1.06–1.36	0.005
Donor type (ref: DBD)
DCD	0.92	0.51–1.68	0.80	—	—	—
Living	0.3	0.01–0.79	0.02	—	—	—
HCV donor	0.52	0.22–1.13	0.11	—	—	—
Donor hospitalization duration a	1.14	0.91–1.48	0.29
Positive donor urine culture	1.29	0.54–2.96	0.56	—	—	—

Abbreviations: CNI: calcineurin inhibitor; DBD: donation by brain death, DCD: donation after circulatory death.

^a Duration was log‐transformed prior to analysis.

^b Compared to Belatacept and everolimus‐based regimens.

On multivariable analysis, each 1‐unit increase in ln(recipient index hospitalization) was associated with a 78% increase in the odds of eBSI (OR: 1.78, 95% CI: 1.29–2.54; p < 0.001), and each 1‐unit increase in ln(postsurgical drain reception) was associated with a 112% increase in the odds of eBSI (OR: 2.12, 95% CI: 1.26–3.71, p 0.006). These correspond to each doubling of hospital duration being independently associated with a 49% increase in the risk of eBSI (OR: 1.49, 95% CI: 1.19–1.91, p < 0.001), and each doubling of postsurgical drain retention duration being independently associated with a 68% increase in the risk of eBSI (OR: 1.68, 95% CI: 1.17–2.48, p 0.006). However, duration of urinary catheterization, time on dialysis prior to transplantation, year of transplantation, DGF, cold ischemic time, and diabetes were not associated with increased odds for eBSI. Multi‐organ transplantation, exposure to critical care, take back to OR, number to OR take back, and presence of central venous catheters were not statistically significant on multivariable analysis. The type of organ donation, donor length of hospitalization, HCV positive donation, and positive donor urinary culture growth were not significant and not associated with eBSI.

3.3. Allograft and Patient Survival

Out of 1408 adult first time kidney transplants, 25 patients died within 90 days of transplant (4 with eBSI that died from infectious complications within 30 days of transplant, and 21 without eBSI). The remaining 1383 were evaluated for survival and graft function. The median time to follow up was 1070 days (IQR: 685–1531).

The cumulative incidence of patient death over 1 year for those with and without eBSI are compared in Figure 2a. On multivariable Cox proportional hazards regression eBSI was associated with increased risk for patient death within the first year, adjusted HR 3.32 (95% CI: 1.51–7.28, p 0.003) (Table S1). Of those with eBSI, 14% were known to have died within 1 year after transplant compared to 9% with no eBSI.

FIGURE 2.

(a) Cumulative incidence of death over 1 year for patients with and without early bloodstream infection. (b) Cumulative incidence of graft failure over 3 years for patients with and without early bloodstream infection.

The cumulative incidence of graft failure over 3 years for those with and without eBSI are compared in Figure 2b. On multivariable Cox proportional hazards regression eBSI was associated with increased risk of graft failure within 3 years, adjusted HR 2.78  (95% CI: 1.31–5.93, p 0.008) (Table S2). If the outcome was only evaluated up to 1 year, the adjusted HR was 3.00 (95% CI: 0.98–9.15, p 0.054).

After excluding those who died or were censored within 1 year (55 patients), and those without 1‐year eGFR data (36 patients), those with eBSI had a slightly lower eGFR at 1 year compared to those who did not have eBSI on multivariable linear regression, but the difference did not reach statistical significance (mean difference: −3.42 mL/min, 95% CI: −7.51 to 0.67, p 0.10).

4. Discussion

This is the largest study evaluating the risk factors and outcomes of eBSI in KTRs, focusing on the previously reported high burden period for BSI. Incidence of eBSI was 5.3% occurring at a median time of 21 days (IQR: 12–42.5) posttransplant. The focus on early infections following transplantation provides an opportunity to understand the effects of the donor and recipient index hospitalizations and the more highly controlled perioperative period on infection risk and outcomes.

Consistent with prior studies, most eBSIs were of urinary origin and due to GNB and Enterococcus species. Drug resistance was present in 40% with the same percentage of Enterobacteriaceae producing ESBL and all E. faecium being vancomycin resistant. Globally, colonization and infection due to ESBL and carbapenemase producing Enterobacteriaceae has increased in the KTRs population over the past decades [5, 6, 13, 14]. Our current perioperative prophylaxis protocol does not cover these organisms though our eBSI incidence alone would not justify broadening coverage in all recipients. It is also not clear from our data that routine perioperative antibiotics would effectively mitigate the infections described here. We found no association between positive donor urine culture and eBSI, possibly because recipients are screened and preemptively treated when donor urine cultures are reported as positive. The high percentage of ESBL producing Enterobacteriaceae does suggest that recipients with early sepsis may benefit from empiric coverage for such organisms. However, such a strategy needs to be balanced with the evolving risk for carbapenem resistance [6]. Effects on graft and patient outcomes for eBSI with and without ESBL strains is beyond the scope of this study but is worth additional investigation.

Our study identified relatively few significant factors associated with eBSI likelihood. Longer durations of index admission for KTRs are likely to reflect more complicated courses, though we were unable to identify specific factors that would link the longer hospitalizations to eBSI. We explored if return to the OR, critical care stay and presence of central line were associated with higher risk for eBSI, as those are commonly associated with longer index admission duration. None of those were statistically significant. Contrary to other studies, we did not find that DGF was independently associated with BSI though it may have contributed to longer index hospitalization. We also did not find that diabetes, cold ischemic time, time on dialysis before transplant, dialysis modality before transplant, immunosuppression regimens, and recipient age were associated with eBSI as previously reported. These may be more relevant to later BSI or other eras of kidney transplantation [2, 3, 8, 9].

There was no association between the duration of recipient Foley or ureter stents and eBSI. However, longer duration of surgical drain maintenance was associated with eBSI. Of interest, the duration of the surgical drain did not appear to be linked specifically to surgical site infection, and its causative association with eBSI overall is unclear. The decision to insert a subcutaneous drain was at the discretion of the transplant surgeon and the drain was typically removed once output consistently falls below 30 cc/day. Re‐evaluation of the protocol for standardized patient selection for insertion and the timing of drain removal, particularly regarding early discontinuation, may serve as a target for further optimization in clinical practice.

We observed that eBSI is significantly associated with graft loss (within 3 years), and patient mortality (within 1 year). eBSI frequently leads to alterations in immunosuppression, commonly withholding antimetabolites. Hypotension and hemodynamic instability accompanying infectious episodes can affect graft function, at least temporarily. Our study did not assess those clinical factors. In addition, the most common source of BSI was a urinary tract infection, a known risk factor for allograft rejection and reduced graft function [15, 16]. We did not account for allograft rejection in the graft function model. Graft loss was significant at 3 years posttransplantation, but did not reach statistical significance at 1 year. We believe that the initial insult from eBSI does affect the graft function, evidenced by lower eGFR at 1 year, ultimately leading to earlier graft failure than expected. The half‐life of a deceased and living donor transplant is 11 and 20 years, respectively [17]. Graft failure, even at 3 years, is early.

This study is limited by its single center population that may limit broader applicability. Second, there is a risk of confounding that we tried to mitigate using multivariable analysis, but the risk for residual confounders remains. Third, missing data from patients who were lost to follow‐up poses statistical bias. Fourth, we collected data from the first episode of bacteremia and did not assess whether recurrent episodes would affect graft function any differently. We also did not adjust for clinical sepsis or shock, rejection after eBSI, which may affect the graft function.

Our findings demonstrate that early bloodstream infections remain a major clinical challenge with meaningful consequences for graft survival in KTRs. A substantial proportion was caused by antibiotic resistant bacteria, highlighting future challenges in managing and preventing eBSI. In this contemporary cohort, prolonged surgical drain use and extended recipient hospitalization emerged as important risk factors for eBSI, highlighting modifiable perioperative practices that may reduce infection risk.

Author Contributions

Mathew Duck: data curation, resources. Christine E. Koval: writing – review and editing, conceptualization, methodology, project administration, data curation, supervision, resources. Lana Hasan: investigation, writing – review and editing, writing – original draft, methodology, formal analysis, data curation. Joshua Augustine: conceptualization, writing – review and editing, supervision, resources. Alvin Wee: resources, supervision, writing – review and editing. Richard Fatica: supervision, resources, writing – review and editing. Nabin K. Shrestha: conceptualization, methodology, formal analysis, writing – review and editing.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Information: tid70278‐sup‐0001‐SuppMat.docx