Abstract
Objective
The incidence of bloodstream infection (BSIs) due to extended-spectrum β-lactamase (ESBL) producing Escherichia coli is increasing worldwide. There is controversy as to whether ESBL production in itself is associated with higher mortality. The aim of this study is to evaluate the impact of ESBL production on mortality in BSIs due to E. coli considering the effect of confounders.
Methods
PROBAC study is a prospective, multicentre, cohort study performed in 26 Spanish hospitals (October 2016–March 2017). All patients with E. coli BSIs were included. The outcome variable was all-cause 30-day mortality. Confounding was controlled by calculating a propensity score (PS) for ESBL production using baseline variables. PS were used as covariable, for matching, for inverse probability of treatment weight analysis and for stratified analysis within the PS quartiles.
Results
A total of 2394 cases were included, of which 322 (13.5%) were ESBL-producing isolates. The frequency of appropriate empirical treatment, in ESBL-producing and non-ESBL-producing isolates, was 53.7% and 92.0%, respectively. Thirty-day mortality was 14.6% in ESBL-producing isolates versus 9.6% in non-ESBL-producing isolates (P = 0.006), for a crude OR of 1.61 (95% CI: 1.14–2.27; P = 0.006). When we adjusted by PS and appropriate empirical treatment, the OR changed to 1.12 (95% CI: 0.75–1.67; P = 0.584). Other PS applications provided similar results.
Conclusion
BSIs due to ESBL-producing E. coli were associated with higher mortality in the crude analyses; however, the estimate of the association is reduced after adjustment for baseline variables and empirical therapy, and is not significant in matched analysis.
Introduction
Escherichia coli is the most common cause of bacterial bloodstream infections (BSIs) in developed countries.1–3 The increased incidence, coupled with rising antimicrobial resistance, made it one of the six microorganisms responsible for the highest number of resistance-attributable deaths in 2019.4 In E. coli, a common resistance mechanism is the production of extended-spectrum β-lactamase (ESBL).5 These enzymes inactivate penicillins and oxy-imino-cephalosporins by hydrolysing the β-lactam ring, thereby rendering the bacteria resistant to these antibiotics. In Spain, ESBL-producing E. coli is the most common multidrug-resistant microorganism.6
The association between mortality and ESBL production remains incompletely understood. Data from big cohorts and systematic reviews provided conflicting results about whether ESBL production in patients with BSIs due to Enterobacterales have increased all-cause mortality compared with non-ESBL producers, mostly due to challenges in controlling the effect of confounders, including patients’ characteristics and delay in administration of active antibiotics.7–11 This is important, because ESBL producers frequently affect more debilitated patients; also, delayed administration of active therapy is more frequent in ESBL producers.12
The aim of this study is to evaluate the impact of ESBL production on mortality in E. coli BSIs, specifically considering the impact of potential confounders, including patients’ features and delay in administering appropriate empirical treatment, taking advantage of data from the PROBAC cohort.
Methods
Study design and participants
This analysis is part of the PROBAC project, a national, multicentre, prospective cohort study carried out in 26 Spanish hospitals between October 2016 and March 2017 (ClinicalTrials.gov identifier: NCT03148769). The design and methods of the PROBAC study were previously detailed.13 In summary, all patients older than 14 years with BSI diagnosis and confirmed microbiologically were included and followed up for 30 days after the blood culture were obtained. The study participants were identified by daily review of blood culture results at each hospital and data collection was performed directly from the patient’s medical records by previously trained investigators supervised by infectious diseases from each participant site. All recorded data were anonymized. Blood cultures were collected, processed and interpreted according to Spanish recommendations.14 Patients for whom limitation of therapeutical efforts had been decided were excluded.
For this analysis, all patients with monomicrobial BSIs due to E. coli included the PROBAC database were eligible. This analysis was reported following STROBE recommendations.15
Variables and definition
The main outcome variable was all-cause 30-day mortality. Explanatory variables were selected based on a previously developed direct acyclic graph (DAG) for analysis of mortality in patients with BSI,16 and included demographics (sex, age, hospital and ward of admission); type and severity of underlying conditions using the age-adjusted Charlson Comorbidity index17; neutropenia (neutrophile count <500 cells/mm3); use of antimicrobials or surgery in the previous month; central venous or urinary catheter and mechanical ventilation in last 48 hours; type of infection acquisition, classified as nosocomial (symptoms starting 48 after hospital admission), healthcare-associated18 and community-acquired; source of BSI defined according to clinical and microbiological data19; presence of sepsis or septic shock at BSI presentation according to SEPSIS-3 criteria20 acute severity according to Pitt score21; antibiotic therapy and in vitro susceptibility of the causative isolate. Appropriate empirical treatment was defined as the administration of an antimicrobial agent with in vitro activity against the bacteria within the first 48 hours. ESBL production was studied in isolates with diminished susceptibility to cephalosporins and confirmed using standard microbiological techniques, including combination disc test or double-disc synergy test, at each participant centre. EUCAST recommendations were used for susceptibility interpretation.22
Missing data were tabulated. Patients with missing data about empirical treatment were excluded from the main analysis but were include in sensitivity analyses (see below).
Statistical analysis
Patient data were described using absolute numbers and proportions for categorical variables, and median values with interquartile range (IQR) for continuous variables. Bivariate associations between exposures and all-cause 30-day mortality were studied by Chi-square or Fisher’s exact tests for categorical variables, and Student’s t-test or Mann–Whitney U-test for continuous variables, as appropriate, and odds ratios (ORs) with a 95% confidence interval (CI) were calculated. The possibility collinearity between variables was evaluated through the calculation of tolerance values and variance inflation factor.
We constructed a non-parsimonious logistic regression model including baseline variables to calculate a propensity score (PS) for ESBL production. The variables included were based on a previously proposed DAG for patients with bacteraemia16 and were: sex, age, individual underlying conditions included in the Charlson index, obstructive biliary or urinary tract disease, neutropenia at BSI presentation, use of central venous catheter, urinary catheter and mechanical ventilation, nephrostomy, surgery and use of antimicrobials in the previous month, type of acquisition and source of infection; acute infection-related severity variables were not included as were considered to be in the pathway between the pathogen and mortality. The discriminatory power of the PS was evaluated calculating the area under the empirical receiver operating characteristic curve (AUROC) with 95% CI and goodness of fit by Hosmer–Lemeshow test. The AUROC of the model was 0.72 (95% CI: 0.69–0.75) and the Hosmer–Lemeshow test showed a P value of 0.944. The PS was used in different ways: as a covariate, as a matching variable for ESBL- and non-ESBL producers (matching by appropriate empirical therapy was also included) using the minimum absolute difference between scores and a maximum tolerance of 5%, and for inverse probability of treatment weight (IPTW) analysis. Finally, we also stratified the cohort into PS quartiles and studied 30-day mortality within each quartile.
Multivariate analyses were performed by logistic regression. Effect modification of adequate empirical treatment and urinary source was also studied. Analyses in the matched cohorts were performed by conditional logistic regression.
Two sensitivity analyses were performed by including all patients with missing data for empirical treatment, considering empirical treatment as appropriate or inappropriate in all of them.
All analyses were performed using the SPSS software (IBM Statistics for Windows, v.25.0, IBM Corporation, Armonk, NY, USA) and the graphics were created with Microsoft Excel (Microsoft Corporation, v.2407, Redmond, WA, USA).
Ethics
This study was conducted according with the Declaration of Helsinki and national and institutional standards. The PROBAC study was approved by the Spanish Medicines Agency and the Ethics Committee of the Hospital Universitario Virgen Macarena with waiver for informed consent due to the observational design (approval reference code: FIS-AMO-201601). Approval was also obtained at each participating centre.
Results
The PROBAC cohort included 6313 patients with BSI, of which 2520 cases were due monomicrobial E. coli BSI; of these, 126 patients (5%) were excluded, three who were under limitation of therapeutic effort and 123 with missing data on empirical therapy (Figure 1).
Figure 1.
Flow diagram of inclusion of patients. LTE, limitation of therapeutic effort.
Of the 2394 cases included in the study, 322 (13.5%) had an ESBL-producing isolate. Patients with ESBL producers were slightly older [median age, 76.5 years (IQR, 66–84)], were more frequently male (53.9%) and had a higher prevalence of several comorbidities. These included myocardial infarction (12.1%), dementia (17.4%), chronic pulmonary disease (16.5%) and moderate to severe kidney disease (18.9%). Recurrent urinary tract infections (UTIs) and urinary catheter use were more frequent among patients with ESBL-producing isolates (15.5% and 19.6%, respectively). The most frequent acquisition type in ESBL producers group was healthcare-associated (46.7%), while community acquisition predominated in the non-ESBL-producing group (57.0%). In both cohorts, the most frequent source of BSI was the urinary tract. The frequency of appropriate empirical treatment against the ESBL-producing and non-ESBL-producing isolates were 53.7% (173) and 92% (1907), respectively (P = <0.001). All demographics, clinical and epidemiological characteristics of patients with ESBL-producing and non-ESBL-producing isolates are shown in Table 1. The potential collinearity between age and the age-adjusted Charlson Comorbidity index was evaluated; the variance inflation factor and tolerance were 1.00, discarding significant collinearity.
Table 1.
Demographics, clinical and epidemiological characteristics of patients with bacteraemia due to E. coli producing and not producing ESBLs
| Variable | ESBL production (n = 322) |
Non-ESBL production (n = 2072) |
P value |
|---|---|---|---|
| Median age in years (IQR) | 76.5 (66–84) | 73 (61–82) | <0.001 |
| Male sex | 172 (53.9) | 1027 (49.9) | 0.182 |
| Ward of admission | |||
| Emergency department | 148 (46.0) | 1181 (57.0) | 0.002 |
| Surgical ward | 31 (9.6) | 144 (6.9) | 0.086 |
| Medical ward | 126 (39.1) | 659 (31.8) | 0.009 |
| Intensive care unit | 17 (5.3) | 88 (4.3) | 0.394 |
| Myocardial infarction | 39 (12.1) | 148 (7.1) | 0.002 |
| Congestive heart failure | 36 (11.2) | 230 (11.1) | 0.966 |
| Cerebrovascular disease | 55 (17.1) | 202 (9.7) | <0.001 |
| Dementia | 56 (17.4) | 210 (10.1) | <0.001 |
| Chronic pulmonary disease | 53 (16.5) | 235 (11.3) | 0.009 |
| Liver disease | 32 (9.9) | 138 (6.7) | 0.033 |
| Diabetes mellitus | 107 (33.2) | 521 (25.1) | 0.002 |
| Moderate or severe kidney insufficiency | 61 (18.9) | 258 (12.5) | 0.001 |
| Cancer | 82 (25.5) | 544 (26.3) | 0.764 |
| Haematological malignancy | 8 (2.5) | 112 (5.4) | 0.025 |
| Neutrophils <500 cells/mm3 | 5 (1.6) | 68 (3.3) | 0.093 |
| Obstructive uropathy | 29 (9.0) | 132 (6.4) | 0.079 |
| Recurrent UTI | 50 (15.5) | 186 (9.0) | <0.001 |
| Obstructive biliary pathology | 27 (8.4) | 126 (6.1) | 0.116 |
| Median age-adjusted Charlson Comorbidity index (IQR) | 2 (1–4) | 2 (0–3) | <0.001 |
| Invasive procedures/devices | |||
| Central venous catheter | 35 (10.9) | 163 (7.9) | 0.069 |
| Urinary catheter | 63 (19.6) | 237 (11.4) | <0.001 |
| Previous surgery | 28 (8.7) | 152 (7.3) | 0.389 |
| Mechanical ventilation | 7 (2.2) | 30 (1.4) | 0.329* |
| BSI acquisition | |||
| Community-acquired | 99 (30.8) | 1179 (57.0) | <0.001 |
| Healthcare-associated | 150 (46.7) | 553 (26.7) | <0.001 |
| Nosocomial | 72 (22.4) | 338 (16.3) | 0.007 |
| Source | |||
| Urinary tract | 218 (67.7) | 1214 (58.6) | 0.002 |
| Biliary tract | 48 (14.9) | 413 (19.9) | 0.033 |
| Non-biliary intra- abdominal | 15 (4.7) | 167 (8.1) | 0.032 |
| Unknown | 12 (3.7) | 155 (7.5) | 0.014 |
| Respiratory tract | 7 (2.2) | 39 (1.9) | 0.723 |
| Vascular | 5 (1.6) | 24 (1.2) | 0.580* |
| Skin and soft tissue | 7 (2.2) | 20 (1.0) | 0.080* |
| Pneumonia | 3 (0.9) | 20 (1.0) | 1.000* |
| Othersa | 7 (2.2) | 20 (1.0) | |
| Median Pitt score (IQR) | 1 (0–2) | 1 (0–2) | 0.006 |
| Median SOFA score (IQR) | 2 (1–4) | 2 (0–4) | 0.002 |
| Sepsis or sepsis shock | 177 (55.0) | 1031 (49.8) | 0.082 |
| Previous antimicrobials | 125 (38.8) | 510 (24.6) | <0.001 |
| Appropriate empirical therapy | 173 (53.7) | 1907 (92.0) | <0.001 |
| 30-day crude mortality | 47 (14.6) | 199 (9.6) | 0.006 |
P values for the comparison of ESBL-producing and non-producing isolates were calculated by Chi-squared except *calculated by Fisher test
SOFA, Sequential Organ Failure Assessment.
aIncludes central nervous system, osteoarticular and others.
Thirty-day crude global mortality was 14.6% (47) in ESBL producers and 9.6% (199) in non-ESBL producers, for a crude OR of 1.61 (95% CI: 1.14–2.27; P = 0.006) (Table 2, Figure 2). However, when the PS and appropriate empirical treatments were included as covariates, the adjusted OR was 1.12 (95% CI: 0.75–1.67; P = 0.584) (Table 2, Figure 2).
Table 2.
Summary of crude and adjusted associations between ESBL production and 30-day mortality
| OR | CI (95%) | P value | |
|---|---|---|---|
| Crude | 1.61 | 1.14–2.27 | 0.006 |
| Including PS and empirical treatment as covariates | 1.12 | 0.75–1.67 | 0.584 |
| PS-matched cohorts | 0.92 | 0.58–1.44 | 0.702 |
| IPTW population, crude | 1.46 | 1.22–1.75 | <0.001 |
| IPTW population, adjusted (logistic regression)a | 1.23 | 0.99–1.52 | 0.056 |
aIncluding variables with significant differences between ESBL production and non-ESBL production in the weighted cohort.
Figure 2.
Forest plot of the crude and adjusted associations between ESBL production isolates and 30-day all-cause mortality.
Then, we matched patients with ESBL-producing and non-producing isolates according to the PS and appropriate empirical treatment; we could match 273 couples (including 84.8% of ESBL producers); 164 couples received appropriate empirical treatment, while 109 couples did not. The matched groups were appropriately balanced for baseline covariates (Table 3). In this case, the association of the ESBL production with 30-day mortality by conditional logistic regression showed an OR of 0.92 (95% CI: 0.58–1.44; P = 0.702) (Table 2, Figure 2). The interactions between adequate empirical treatment and urinary source were not significant.
Table 3.
Matched cases according to propensity scores plus adequacy of empirical treatment
| Variable | ESBL production (n = 273) |
Non-ESBL production (n = 273) |
P value |
|---|---|---|---|
| Median age in years (IQR) | 75 (65–84) | 76 (68–84) | 0.703 |
| Male sex | 147 (53.9) | 150 (54.9) | 0.797 |
| Underlying diseases | |||
| Myocardial infarction | 33 (12.1) | 29 (10.6) | 0.590 |
| Congestive heart failure | 33 (12.1) | 32 (11.7) | 0.895 |
| Cerebrovascular disease | 39 (14.3) | 27 (9.9) | 0.115 |
| Dementia | 42 (15.4) | 38 (13.9) | 0.628 |
| Chronic pulmonary disease | 43 (15.8) | 46 (16.8) | 0.728 |
| Liver disease | 28 (10.3) | 32 (11.7) | 0.584 |
| Diabetes mellitus | 95 (34.8) | 87 (31.9) | 0.468 |
| Moderate or severe kidney insufficiency | 56 (20.5) | 57 (20.9) | 0.916 |
| Cancer | 68 (24.9) | 70 (25.6) | 0.844 |
| Haematological malignancy | 8 (2.9) | 9 (3.3) | 0.805 |
| Neutrophils < 500 cells/mm3 | 5 (1.8) | 5 (1.8) | 1.000 |
| Obstructive uropathy | 28 (10.3) | 18 (6.6) | 0.123 |
| Recurrent UTI | 45 (16.5) | 40 (14.7) | 0.555 |
| Obstructive biliary pathology | 22 (8.1) | 29 (10.6) | 0.303 |
| Median Charlson Comorbidity indexa (IQR) | 2 (1–4) | 2 (1–3) | 0.703 |
| Invasive procedures/devices | |||
| Central venous catheter | 32 (11.7) | 31 (11.4) | 0.893 |
| Urinary catheter | 51 (18.7) | 54 (19.8) | 0.745 |
| Previous surgery | 23 (8.4) | 29 (10.6) | 0.382 |
| Mechanical ventilation | 5 (1.8) | 5 (1.8) | 1.000 |
| BSI acquisition | |||
| Community-acquired | 95 (34.8) | 77 (28.2) | 0.097 |
| Healthcare-associated | 119 (43.6) | 128 (46.9) | 0.439 |
| Nosocomial | 59 (21.6) | 68 (24.9) | 0.362 |
| Source | |||
| Urinary tract | 184 (67.4) | 159 (58.2) | 0.027 |
| Biliary tract | 41 (15.0) | 51 (18.7) | 0.253 |
| Non-biliary intra- abdominal | 13 (4.8) | 19 (7.0) | 0.274 |
| Unknown | 11 (4.0) | 26 (9.5) | 0.011 |
| Respiratory tract | 5 (1.8) | 4 (1.5) | 1.000* |
| Vascular | 4 (1.5) | 5 (1.8) | 1.000* |
| Skin and soft tissue | 6 (2.2) | 3 (1.1) | 0.504* |
| Pneumonia | 3 (1.1) | 3 (1.1) | 1.000* |
| Othersb | 6 (2.2) | 3 (1.1) | |
| Median Pitt score (IQR) | 1 (0–2) | 1 (0–2) | 0.292 |
| Median SOFA score (IQR) | 2 (1–5) | 2 (1–4) | 0.290 |
| Previous antimicrobials | 100 (36.6) | 101 (37.0) | 0.929 |
| Appropriate empirical therapy | 164 (60.1) | 164 (60.1) | 1.000* |
| 30-day crude mortality | 36 (13.2) | 40 (14.7) | 0.621 |
P values for the comparison of ESBL-producing and non-producing isolates were calculated by Chi-squared except *calculated by Fisher test.
SOFA, Sequential Organ Failure Assessment.
aAdjusted for age.
bIncludes central nervous system, osteoarticular and others.
In the IPTW analysis, the two groups showed some differences in variables distributions, as expected, such as solid metastatic cancer, neutropenia, BSI source, among others (Table S1, available as Supplementary data at JAC Online). A bivariate analysis of ESBL production in the IPTW population provided an OR for mortality of 1.46 (CI: 1.22–1.75; P=<0.001). An adjusted analysis in this population including neutropenia, obstructive biliary disease, solid metastatic cancer, BSI source and appropriated empirical treatment provided an OR of 1.23 (CI: 0.99–1.52; P = 0.056) (Table 2, Figure 2).
Finally, we studied mortality stratified by PS quartiles in patients with BSI caused by ESBL-producing and non-ESBL-producing isolates. In ESBL- and non-ESBL producers, mortality was 20% and 7.3% in the first quartile (P = 0.038), 4.4% and 7.4% in the second quartile (P = 0.76), 14.1% and 11.6% in the third quartile (P = 0.57), and 16.7% and 12.1% in the fourth quartile (P = 0.17) (Supplementary Table S4).
Sensitivity analysis including patients with missing empirical treatment variable provided similar results, with significant association between ESBL production and mortality only in crude analyses (full and weighted populations), while significant association was not found in all adjusted analysis, although adjusted analysis in the IPWT population was near significance (Supplementary Tables S2 and S3 and Supplementary Figures S1 and S2).
Discussion
Our study shows that while ESBL production was associated with increased mortality in the crude analysis, this association disappeared when adjusting for the adequacy of empirical treatment and patient baseline characteristics. This suggests that the observed association between ESBL production and mortality may be attributable to the patient characteristics at presentation, rather than the intrinsic effect of the enzyme. Only a trend towards an increased mortality in ESBL producers was observed in the IPWT analysis. We suggest this may be due to insufficient confounding control, as this analysis gives greater weight to patients in the first quartile, which was the only one where ESBL production showed a significant increase in mortality. When this confounding was accounted for, the trend was again attenuated.
The clinical outcome and effect of ESBL-producing isolates on E. coli BSI have been extensively studied. Several studies conducted in diverse geographic areas have evaluated the influence of this resistance mechanism on mortality and hospital length of stay, with some focusing exclusively on specific acquisition types or infection categories. An investigation conducted in Queensland analysed these outcomes in E. coli BSI using a 20-year population-based cohort.9 The 30-day mortality rate reported in that study for patients with ESBL production was 12.8%, which is comparable to our finding of 14.6%; ESBL-producing isolates were associated with a crude higher risk of 30-day mortality. Nevertheless, after conducting a multivariable survival analysis adjusted for age, gender, the Charlson index, BSI acquisition, polymicrobial infection, and source of infection, the significance of the association did not persist. This finding aligns closely with our own results, although it is crucial to note that their analysis did not account for the effect of appropriate empirical therapy and included polymicrobial infections. Similar results were found in an investigation conducted in the North Denmark Region.10 That study analysed community-onset ESBL-producing E. coli and Klebsiella pneumoniae causing BSI and UTIs. ESBL-producing isolates were once again associated with higher mortality in crude analysis, but this association dissipated on adjustment for BSI source and polymicrobial bacteraemia. Also, in community-onset BSI due to E. coli, the association of ESBL and mortality was not significant when inappropriate empirical therapy was considered.
Following these results, several systematic reviews have been published that investigate the attributable mortality related to ESBL production. A meta-analysis including studies comparing outcomes of patients with ESBL producers colonization or infection with non-ESBL producers found an association of ESBL production with mortality (OR = 1.70; 95% CI: 1.15–2.49);7 the association was not significant when only adjusted estimates were included, but the estimation lacked precision (OR:1.67; 95% CI: 0.52–5.39). Specifically in patients with bacteraemia, another meta-analysis found ESBL producers to be associated with increased mortality (OR: 1.70; 95% CI: 1.52–1.90), but the authors recognized that most studies provide unadjusted estimations.8 Finally, in a meta-analysis providing data from studies with adjusted estimations including 7682 patients, ESBL production had a pooled OR of 1.52 (95% CI: 1.15–2.01), and 1.37 (95% CI: 1.04–1.82) when adjusting for appropriate empirical therapy (7229 patients),11 suggesting that an important part of the mortality effect would derive from receiving inappropriate initial therapy. Overall, published studies are heterogeneous in confounders considered and methods used to control their effect.
In our study, we first included potential confounders following a previously developed DAG, which was based on a systematic literature review; then, we used different methods to control the effect of confounders to provide a series of effect estimates. This allowed us to gain a more comprehensive view of the effect of ESBL producers on mortality. As recommended, we did not include mediators such as sepsis or shock.23
This study, however, has several limitations. First, we only included patients with BSI and older than 14 years, so the results could not be generalized to patients with non-invasive E. coli infections and not be extrapolated to children. A second limitation was that we only investigated the effect on mortality and not on other outcomes, such as increased length of hospitalization or increased risk of recurrence. Furthermore, genotypic characterization of the isolates was not performed. Consequently, information regarding the predominant β-lactamases and potential differences between the various genotypes is unavailable. Finally, the variable time to adequate treatment administration was not available as a continuous measure. Instead, this variable was collected dichotomously, limiting our data to whether the patient received appropriate empirical treatment within the initial 48 hours. Some strengths include the high number of cases included, the quality assessment of data by monitoring, the inclusion of variables based on a DAG and the use of several methods for confounding control.
In conclusion, ESBL production in E. coli BSI is associated with higher mortality primarily due to the differences in patient baseline characteristics and empirical therapy. The association is reduced when these factors and empirical treatment are controlled for in the analyses, and is not significant in matched analysis.
Supplementary Material
Acknowledgements
Members of the PROBAC/GEIRAS-SEIMC/SAMICEI Group.
José María Reguera-Iglesias (Hospital Regional Universitario de Málaga, IBIMA, Málaga, Spain), José María Bravo-Ferrer (Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen Macarena; Seville, Spain), Antonio Sánchez-Porto (Servicio de Medicina Interna; Hospital Universitario de Puerto Real, Cádiz, Spain.); Juan Manuel Sánchez-Calvo (Servicio de Enfermedades Infecciosas, Hospital Universitario de Jerez, Jerez de la Frontera, Instituto de Investigación e Innovación en Ciencias Biomédicas de Cádiz (INiBICA), Universidad de Cádiz, Cadiz, Spain.); Isabel Reche (Servicio de Enfermedades Infecciosas, Hospital Universitario Torrecárdenas, Almería, Spain.); Isabel Gea-Lázaro (Servicio de Enfermedades Infecciosas, Hospitalario Universitario de Jaén, Jaén, Spain.); David Vinuesa-García (Servicio de Enfermedades Infecciosas, Hospital Clínico San Cecilio, Granada, Spain.); Inés Pérez-Camacho (Hospital Universitario Poniente, Almería, Spain.); Marcos Guzmán-García (Servicio de Medicina Interna; Hospital Universitario de Puerto Real, Cádiz, Spain.); Esperanza Merino de Lucas (Servicio de Enfermedades Infecciosas, Hospital General Universitario de Alicante, Alicante, Spain; CIBERINFEC, Instituto de Salud Carlos III, Madrid, Spain.)
Contributor Information
Paula Olivares-Navarro, Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen Macarena; Departamento de Medicina, Universidad de Sevilla; Instituto de Biomedicina de Sevilla (IBiS)/CSIC, Seville, Spain; CIBERINFEC, Instituto de Salud Carlos III, Madrid, Spain.
María Teresa Pérez-Rodríguez, Complejo Hospitalario Universitario de Vigo, Galicia Sur Health Research Institute, Vigo, Spain.
Adrián Sousa, Complejo Hospitalario Universitario de Vigo, Galicia Sur Health Research Institute, Vigo, Spain.
Ane Josune Goikoetxea-Agirre, Servicio de Enfermedades Infecciosas, Hospital Universitario de Cruces, Bilbao, Spain.
María José Blanco Vidal, Servicio de Enfermedades Infecciosas, Hospital Universitario de Cruces, Bilbao, Spain.
Antonio Plata, Servicio de Enfermedades Infecciosas, Hospital Regional Universitario de Málaga, IBIMA, Málaga, Spain.
Eva León, Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen de Valme, Seville, Spain.
Luis Buzón-Martín, Unidad de Enfermedades Infecciosas, Hospital Universitario de Burgos, Burgos, Spain.
Ángeles Pulido-Navazo, Servicio de Enfermedades Infecciosas, Hospital General de Granollers, Granollers, Spain.
Lucía Boix-Palop, Servicio de Enfermedades Infecciosas, Hospital Universitario Mútua de Terrassa, Terrassa, Spain.
Pilar Retamar-Gentil, Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen Macarena; Departamento de Medicina, Universidad de Sevilla; Instituto de Biomedicina de Sevilla (IBiS)/CSIC, Seville, Spain; CIBERINFEC, Instituto de Salud Carlos III, Madrid, Spain.
Carlos Armiñanzas-Castillo, CIBERINFEC, Instituto de Salud Carlos III, Madrid, Spain; Servicio de Enfermedades Infecciosas, Hospital Universitario Marqués de Valdecilla, Santander, Spain.
Isabel Fernández-Natal, Servicio de Microbiología Clínica, Complejo Asistencial Universitario de León, León, Spain.
Alfonso Del Arco Jiménez, Grupo de Enfermedades Infecciosas, Servicio de Medicina Interna, Hospital Universitario Costa del Sol, Marbella, Spain.
Alfredo Jover-Sáenz, Unidad Territorial Infección Nosocomial y Política Antibiótica (UTIN), Hospital Universitario Arnau de Vilanova, Lleida, España.
Jonathan Fernández-Suárez, Unidad de Microbiología Clínica, Hospital Universitario Central de Asturias, Oviedo, Spain.
Andrés Martín-Aspas, Unidad de Enfermedades Infecciosas, Servicio de Medicina Interna, Facultad de Medicina, Hospital Universitario Puerta del Mar, Instituto de Investigación e Innovación en Ciencias Biomédicas de Cádiz (INiBICA), Universidad de Cádiz, Cadiz, Spain.
Alejandro Smithson-Amat, Unidad de Medicina Interna, Hospital Esperit Sant, Santa Coloma de Gramenet, Barcelona, Spain.
Alberto Bahamonde-Carrasco, Departamento de Medicina Interna, Hospital de El Bierzo, Ponferrada, Spain.
Clara Natera-Kindelán, Servicio de Enfermedades Infecciosas, Hospital Universitario Reina Sofia, Córdoba, Spain.
Pedro María Martínez Pérez-Crespo, Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen de Valme, Seville, Spain.
Inmaculada López-Hernández, Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen Macarena; Departamento de Medicina, Universidad de Sevilla; Instituto de Biomedicina de Sevilla (IBiS)/CSIC, Seville, Spain; CIBERINFEC, Instituto de Salud Carlos III, Madrid, Spain.
Jesús Rodríguez-Baño, Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen Macarena; Departamento de Medicina, Universidad de Sevilla; Instituto de Biomedicina de Sevilla (IBiS)/CSIC, Seville, Spain; CIBERINFEC, Instituto de Salud Carlos III, Madrid, Spain.
Luis Eduardo López-Cortés, Unidad Clínica de Enfermedades Infecciosas y Microbiología, Hospital Universitario Virgen Macarena; Departamento de Medicina, Universidad de Sevilla; Instituto de Biomedicina de Sevilla (IBiS)/CSIC, Seville, Spain; CIBERINFEC, Instituto de Salud Carlos III, Madrid, Spain.
PROBAC/GEIRAS-SEIMC/SAMICEI:
José María Reguera-Iglesias, José María Bravo-Ferrer, Antonio Sánchez-Porto, Juan Manuel Sánchez-Calvo, Isabel Reche, Isabel Gea-Lázaro, David Vinuesa-García, Inés Pérez-Camacho, Marcos Guzmán-García, and Esperanza Merino de Lucas
Funding
This work was financed by grants from Plan Nacional de I+D+i 2013–2016 [PI16/01432] and CIBERINFEC (CB21/13/00012), Instituto de Salud Carlos III, co-financed by the European Development Regional Fund ‘A way to achieve Europe’, Operative program Intelligent Growth 2014–2020.
Transparency declarations
A.P. has served as speaker for Gillead, ViiV and Pfizer and has received support to attend meeting from Astellas and Angelini. L.B.P. has received financial support from Pfizer to attend meetings. P.R.G. participated as speaker for Menarini and as part of an Advisory board for Advanz. L.E.L.C. has served as a scientific advisor for Angelini and speaker for Gilead, Correvio, Angelini and ViiV. The other authors report no conflicts of interest relevant to this article.
Authors contributions
P.O.N. contributed to the conception and design of the work, data analysis and interpretation, drafted the work and approved the submitted version. L.E.L.C. contributed to the conception and design of the work, data acquisition, substantial revised the work and approved the submitted version. J.R.B. contributed to the conception and design of the work, substantially revised the work and approved the submitted version. M.T.P.R., A.S., A.J.G.A., M.J.B.V., A.P., E.L., L.B.M., A.P.N., L.B.P., P.R.G., C.A.C., I.F.N., A.A.J., A.J.S., J.F.S., A.M.A., A.S.A., A.B.C., C.N.K., P.M.M.P.C. and I.L.H. have contributed to the data acquisition and approved the submitted version. All authors agree to be personally accountable for their own contributions and to ensure that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved and the resolution documented in the literature.
Supplementary data
Supplementary Figures S1 and S2 and Supplementary Tables S1–S4 are available as Supplementary data at JAC Online.
Data availability
The data will be made available upon request for scientific use, subject to the prior signing of an agreement with the authors’ institution.
References
- 1. Bonten M, Johnson JR, van den Biggelaar AHJ et al. Epidemiology of Escherichia coli bacteremia: a systematic literature review. Clin Infect Dis 2021; 72: 1211–9. 10.1093/cid/ciaa210 [DOI] [PubMed] [Google Scholar]
- 2. Nielsen SL, Pedersen C, Jensen TG et al. Decreasing incidence rates of bacteremia: a 9-year population-based study. J Infect 2014; 69: 51–9. 10.1016/j.jinf.2014.01.014 [DOI] [PubMed] [Google Scholar]
- 3. Vostal AC, Grance M, Powers JH III et al. Demographics, epidemiology, mortality and difficult-to-treat resistance patterns of bacterial bloodstream infections in the global US Military Health System from 2010 to 2019: a retrospective cohort study. BMJ Open 2025; 15: e094861. 10.1136/bmjopen-2024-094861 [DOI] [Google Scholar]
- 4. Murray CJ, Ikuta KS, Sharara F et al. Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis. Lancet 2022; 399: 629–55. 10.1016/S0140-6736(21)02724-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. World Health Organization . Global Antimicrobial Resistance and Use Surveillance System (GLASS). World Health Organization, 2022. https://books.google.com/books/about/Global_antimicrobial_resistance_and_use.html?hl=es&id=dHsOEQAAQBAJ. [Google Scholar]
- 6. Peñalva G, Cantón R, Pérez-Rodríguez MT et al. Burden of bacterial antimicrobial resistance among hospitalised patients in Spain: findings from three nationwide prospective studies. Lancet Reg Health Eur 2025; 51: 101220. 10.1016/j.lanepe.2025.101220 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Ling W, Furuya-Kanamori L, Ezure Y et al. Adverse clinical outcomes associated with infections by Enterobacterales producing ESBL (ESBL-E): a systematic review and meta-analysis. JAC Antimicrob Resist 2021; 3: dlab068. 10.1093/jacamr/dlab068 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Shamsrizi P, Gladstone BP, Carrara E et al. Variation of effect estimates in the analysis of mortality and length of hospital stay in patients with infections caused by bacteria-producing extended-spectrum beta-lactamases: a systematic review and meta-analysis. BMJ Open 2020; 10: e030266. 10.1136/bmjopen-2019-030266 [DOI] [Google Scholar]
- 9. Ling W, Paterson DL, Harris PNA et al. Mortality, hospital length of stay, and recurrent bloodstream infections associated with extended-spectrum beta-lactamase-producing Escherichia coli in a low prevalence region: a 20-year population-based large cohort study. Int J Infect Dis 2024; 138: 84–90. 10.1016/j.ijid.2023.11.007 [DOI] [PubMed] [Google Scholar]
- 10. Richelsen R, Smit J, Schønheyder HC et al. Outcome of community-onset ESBL-producing Escherichia coli and Klebsiella pneumoniae bacteraemia and urinary tract infection: a population-based cohort study in Denmark. J Antimicrob Chemother 2020; 75: 3656–64. 10.1093/jac/dkaa361 [DOI] [PubMed] [Google Scholar]
- 11. Rottier WC, Ammerlaan HSM, Bonten MJM. Effects of confounders and intermediates on the association of bacteraemia caused by extended-spectrum β-lactamase-producing Enterobacteriaceae and patient outcome: a meta-analysis. J Antimicrob Chemother 2012; 67: 1311–20. 10.1093/jac/dks065 [DOI] [PubMed] [Google Scholar]
- 12. Schwaber MJ, Carmeli Y. Mortality and delay in effective therapy associated with extended-spectrum beta-lactamase production in Enterobacteriaceae bacteraemia: a systematic review and meta-analysis. J Antimicrob Chemother 2007; 60: 913–20. 10.1093/jac/dkm318 [DOI] [PubMed] [Google Scholar]
- 13. Pérez-Crespo PMM, Lanz-García JF, Bravo-Ferrer J et al. Revisiting the epidemiology of bloodstream infections and healthcare-associated episodes: results from a multicentre prospective cohort in Spain (PRO-BAC study). Int J Antimicrob Agents 2021; 58: 106352. 10.1016/j.ijantimicag.2021.106352 [DOI] [PubMed] [Google Scholar]
- 14. Rodríguez Díaz JC, Guna Serrano R, Larrosa Escartín N et al. Diagnóstico microbiológico de la bacteriemia y la fungemia: hemocultivos y métodos moleculares. Sociedad Española de Enfermedades Infecciosas y Microbiología Clínica (SEIMC). https://www.seimc.org.
- 15. von Elm E, Altman DG, Egger M et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol 2008; 61: 344–9. 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
- 16. Varon B, Palacios-Baena ZR, de Kraker MEA et al. Universal risk factors for mortality in bloodstream infections (UNIFORM): a systematic review and Delphi survey. Clin Microbiol Infect 2024; 30: 453–61. 10.1016/j.cmi.2023.12.030 [DOI] [PubMed] [Google Scholar]
- 17. Charlson ME, Pompei P, Ales KL et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40: 373–83. 10.1016/0021-9681(87)90171-8 [DOI] [PubMed] [Google Scholar]
- 18. Friedman ND, Kaye KS, Stout JE et al. Health care-associated bloodstream infections in adults: a reason to change the accepted definition of community-acquired infections. Ann Intern Med 2002; 137: 791–7. 10.7326/0003-4819-137-10-200211190-00007 [DOI] [PubMed] [Google Scholar]
- 19. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health care-associated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control 2008; 36: 309–32. 10.1016/j.ajic.2008.03.002 [DOI] [PubMed] [Google Scholar]
- 20. Singer M, Deutschman CS, Seymour CW et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016; 315: 801–10. 10.1001/jama.2016.0287 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Paterson DL, Ko WC, Von Gottberg A et al. International prospective study of Klebsiella pneumoniae bacteremia: implications of extended-spectrum beta-lactamase production in nosocomial infections. Ann Intern Med 2004; 140: 26–32. 10.7326/0003-4819-140-1-200401060-00008 [DOI] [PubMed] [Google Scholar]
- 22. The European Committee on Antimicrobial Susceptibility Testing (EUCAST) . Breakpoint tables for interpretation of MICs and zone diameters. Version 6.0. 2016. https://www.eucast.org/ast_of_bacteria/previous_versions_of_documents.
- 23. VanderWeele TJ. Explanation in causal inference: developments in mediation and interaction. Int J Epidemiol 2016; 45: 1904–8. 10.1093/ije/dyw277 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data will be made available upon request for scientific use, subject to the prior signing of an agreement with the authors’ institution.