1,296
Views
1
CrossRef citations to date
0
Altmetric
Original Articles

An easy-to-use scoring system for predicting bacteraemia with third-generation cephalosporin-resistant Enterobacterales in a low-resistance setting

, & ORCID Icon
Pages 242-248 | Received 13 Jun 2019, Accepted 12 Dec 2019, Published online: 23 Dec 2019

Abstract

Background: The incidence of third-generation cephalosporin-resistant Enterobacterales (3GCR-E) is increasing and a growing number of patients risk receiving inappropriate initial antibiotic treatment. Published scoring systems for predicting 3GCR-E bacteraemia are mostly based on studies from countries with a high incidence. In this study, we aimed to create an easy-to-use scoring system for predicting bacteraemia with these bacteria in a low-resistance setting.

Materials and methods: Factors associated with 3GCR-E were studied retrospectively in a cohort of patients with Enterobacterales bacteraemia using uni- and multivariate analysis. A scoring system was constructed and was validated in a separate cohort of patients with Enterobacterales bacteraemia.

Results: The derivation cohort comprised 625 cases of Enterobacterales bacteraemia. Three variables (previous hospital care abroad, 3GCR-E in a previous blood or urine culture and 3GCR-E in a previous rectal swab culture) were significantly associated with 3GCR-E bacteraemia. A scoring system, where at least one positive parameter equalled a positive score, was studied in the validation cohort, which comprised 675 cases of Enterobacterales bacteraemia. The sensitivity and specificity of the score were 53% and 95%, respectively. Positive and negative predictive values were 38% and 97%, respectively.

Conclusions: This study presents an easy-to-use scoring system for predicting bacteraemia with 3GCR-E. The performance of the score is similar to that of several other, more complicated, scoring systems, developed in countries with higher rates of resistance. The minimal extra effort required to use this new score could facilitate its introduction into clinical routine.

Introduction

The development of antibiotics is regarded as one of the greatest successes in modern medicine. Thanks to antibiotics, both non-severe and life-threatening infections can be treated successfully. However, there is a growing threat to the efficacy of antibiotics in the form of multi-resistant bacteria, such as third-generation cephalosporin-resistant Enterobacterales (3GCR-E). Production of extended-spectrum beta-lactamases (ESBL) is the most common resistance mechanism of 3GCR-E [Citation1]. Even though the frequency of 3GCR-E is lower in Sweden than in most other countries, the problem is growing. According to the Swedish Public Health Agency, 594 cases of 3GCR-E bacteraemia were reported in Sweden in 2017. Escherichia coli (E. coli) is the dominant 3GCR-E, causing 86% of the cases, followed by Klebsiella pneumoniae (K. pneumoniae) causing 10% [Citation2].

In infections with 3GCR-E, effective treatment can be delayed, which may result in severe consequences for the patient. The importance of early effective treatment has been shown in several studies [Citation3,Citation4]. Patients presenting at the emergency department with symptoms of bacteraemia are treated empirically until results of antibiotic susceptibility testing are available. Since third-generation cephalosporins are the most commonly used empiric antibiotics in Sweden, as well as in several other European countries, patients with 3GCR-E infection risk ineffective initial treatment [Citation3,Citation4].

There are several factors to consider when attempting to assess if a patient with symptoms of bacteraemia might be infected with 3GCR-E and previous studies have constructed scoring systems, sometimes with six or more variables and advanced decision trees [Citation5,Citation6]. In addition to being cumbersome to use clinically, most of these studies are from countries with a high incidence of 3GCR-E and high rates of transmission of resistant bacteria in hospitals and nursing homes, which limits the applicability in other settings. However, a recently published Swedish study showed that prior culture with ESBL-producing Enterobacterales, a recent prostate biopsy and prior hospital care abroad were risk factors for community-onset bacteraemia with ESBL-producing Enterobacterales [Citation7]. In the present study, we aimed to create an easy-to-use scoring system for predicting 3GCR-E bacteraemia in a low-resistance setting and to validate the scoring system in a separate patient cohort.

Materials and methods

Study design and setting

This study was conducted at Sahlgrenska University Hospital, a large tertiary care hospital at five different locations in the Gothenburg region, with approximately 1950 beds. The study was designed as a retrospective cohort study with one derivation cohort and one validation cohort.

Inclusion and exclusion criteria

Patients with at least one positive blood culture with E. coli or Klebsiella in 2016–2017 were included. Patients with a positive blood culture in 2016 were included in the derivation cohort, while patients with a positive blood culture in 2017 were included in the validation cohort. More than one positive blood culture within 30 days was considered to be a single episode, while positive blood cultures more than 30 days apart were regarded as separate episodes.

Definitions

Isolates resistant to at least one third-generation cephalosporin were classified as 3GCR-E. No distinction was made between ESBL- and AmpC-producing Enterobacterales. The following factors were analysed as predictors of bacteraemia with 3GCR-E: age, sex, hospitalization at Sahlgrenska University Hospital within 12 months preceding admission, prostate biopsy within 30 days preceding admission, invasive surgical procedure within 30 days preceding admission, urinary catheterization within 30 days preceding admission, residence in a nursing home, hospital-prescribed treatment with fluoroquinolones or second- or third-generation cephalosporins within three months preceding admission, immunosuppressive therapy within three months preceding admission (defined as glucocorticoids (equivalent to prednisolone at 20 mg daily or above for at least four weeks) or any of the following: tacrolimus, sirolimus, cyclosporine, mycophenolate, alkylating agents, immunosuppressive monoclonal antibodies), previous hospital care abroad, hospital- or community-acquired infection based on the onset of symptoms within two days after admission, a previous rectal swab culture with 3GCR-E and a previous urine and/or blood culture with 3GCR-E within two years preceding admission.

Data collection

Patients were identified from records at the Department of Medical Microbiology at Sahlgrenska University Hospital. Data on studied variables were obtained from the electronic patient record system and the laboratory data system.

Ethics

According to Swedish regulations, informed consent was not required for this study. The study was approved by the Swedish Ethical Review Authority: EPM 2019-00365.

Statistical analysis

Continuous variables were analysed with the Mann–Whitney U-test. Categorical variables were analysed with Pearson’s chi-square test. Variables with p values <.2 were included in a multivariate model using backward stepwise logistic regression. p Values ≤.05 were considered to indicate significance. A scoring system based on the results of the multivariate analysis was evaluated in the validation cohort. Statistical analysis was performed using the SPSS version 25.0 statistical software package (IBM Corp., New York, NY).

Culturing procedure

3GCR-E includes both ESBL- and AmpC-producing Enterobacterales. E. coli, K. pneumoniae and K. oxytoca from blood cultures were tested for antibiotic susceptibility on Mueller-Hinton agar according to EUCAST guidelines [Citation8]. Isolates with reduced susceptibility to third-generation cephalosporins were initially tested for the ESBL phenotype, using a double-disc diffusion test (DDT). Isolates with positive DDT were classified as ESBL producers. Isolates that displayed a cefoxitin-zone <19 mm were further tested for the AmpC-phenotype, using cloxacillin–ampicillin synergy testing.

Results

Derivation cohort

A total of 625 episodes of Enterobacterales bacteraemia in 600 patients were included in the derivation cohort. Fifty-seven were caused by 3GCR-E (9%). E. coli was isolated from 482 cultures (77%) and Klebsiella from 143 cultures (23%). ESBL production was demonstrated in 53 isolates, AmpC production in four and combined ESBL and AmpC production in three. The median age was 74 years and the mean age 68 years. The number of cultures collected from men and women were 317 and 308, respectively.

Univariate analysis

Univariate analysis of the derivation cohort yielded the following factors further analysis with multivariate logistic regression (): sex (p = .049), urinary catheter use within 30 days (p = .146), treatment with fluoroquinolones or third-generation cephalosporins within the last three months (p = .077), hospital care abroad (p < .001), ESBL-positive in a rectal swab culture within the last two years (p < .001) and 3GRC-E in blood and/or urine culture within the last two years (p < .001). Age was analysed separately with the Mann–Whitney U-test and was not a significant factor.

Table 1. Univariate analysis with chi-square and Mann–Whitney’s U-test in the derivation cohort of risk factors for bacteraemia with third-generation cephalosporin-resistant Enterobacterales (3GCR-E).

Multivariate analysis

In multivariate logistic regression analysis, the following variables were significantly associated with 3GCR-E bacteraemia: previous hospital care abroad, OR 3.7 (CI 1.3–10.7), 3GCR-E in a rectal swab culture within the last two years, OR 5.3 (CI 1.2–23.9) and 3GCR-E in blood and/or urine culture within the last two years, OR 38.2 (CI 9.9–147.9) ().

Table 2. Multivariate logistic regression analysis of risk factors for third-generation cephalosporin-resistant Enterobacterales (3GCR-E) bacteraemia.

Prediction score

A scoring system for predicting 3GCR-E bacteraemia was constructed from the three variables above (previous blood and/or urine culture with 3GCR-E, previous rectal swab culture with 3GCR-E and previous hospital care abroad). The score was categorized as positive if one or more variables were positive. If all parameters were negative, the score was categorized as negative.

Validation cohort

In the validation cohort, 675 episodes of Enterobacterales bacteraemia in 650 patients were included. Thirty-four were caused by 3GCR-E (5%). E. coli and Klebsiella were isolated from 529 (78%) and 146 (22%) cultures, respectively. ESBL production was demonstrated in 26 isolates, AmpC production in two and combined ESBL and AmpC production in six. The median age was 74 years and the mean age 69 years. The number of cultures from men and women was 311 and 364, respectively.

The difference in 3GCR-E prevalence in the derivation (9%) and validation (5%) cohort was statistically significant (p = .004).

When the scoring model was applied to the validation cohort, 47 episodes had a positive score. Of these, 18 were caused by 3GCR-E. The sensitivity of the model in predicting 3GCR-E bacteraemia was 53% (CI 35–70%), while the specificity was 95% (CI 93–97%). Positive and negative predictive values were 38% (CI 25–54%) and 97% (CI 96–98%), respectively. The impact of more than one positive variable was explored. However, the simultaneous presence of more than positive variable was uncommon, which resulted in a sensitivity of only 15%. Further evaluation of the scoring system with ROC curve analysis was therefore not considered.

Comparison with the Stockholm score

The Stockholm score proposed in the study by Fröding et al. was applied to the validation cohort in this study [Citation7]. The resulting sensitivity and specificity were 56% and 95%, respectively.

Discussion

In this retrospective study of patients with Enterobacterales bacteraemia, we found that a previous culture with 3GCR-E in blood, urine or rectal swab and previous hospital care abroad were predicting factors for bacteraemia with 3GCR-E. We showed that a scoring system based on these factors can be useful to predict 3GCR-E bacteraemia in clinical routine.

The incidence of infections with 3GCR-E has increased greatly in the last few years [Citation9]. The treatment of severe infections caused by multi-resistant bacteria is problematic, due to the risk of ineffective empirical therapy. Previous research has shown that infections with 3GCR-E have significantly worse clinical outcomes [Citation10]. Early administration of effective antimicrobial treatment is of the utmost importance in increasing survival in sepsis and septic shock [Citation11]. In light of this knowledge, multiple studies have evaluated predictive factors for bacteraemia with 3GCR-E in patients with symptoms of Gram-negative sepsis [Citation12–15].

Data on predictive factors in low-resistance countries such as Sweden are limited. The results of international studies are not directly generalizable due to differences in epidemiology, healthcare systems and antibiotic prescription policies. The prevalence of multi-resistant bacteria in Sweden is relatively low, with 7.4% of E. coli being ESBL producers, compared with 10.2% in France and 29.5% in Italy. The corresponding data for ESBL-producing Klebsiella are 5.6% in Sweden, 28.8% in France and 54.6% in Italy [Citation16]. Implementation of scoring systems derived in countries with a higher prevalence of multi-resistant bacteria is therefore problematic.

In the present study, three significant predictors of bacteraemia caused by 3GCR-E were identified. The first was previous hospital care abroad. A Finnish study indicated that hospitalization in countries with a high prevalence of 3GCR-E was a risk factor for colonization by these bacteria and Fröding et al. found a correlation between hospital care abroad and bacteraemia with ESBL-producing Enterobacteriaceae [Citation7,Citation17]. The second predictor was previous isolation of 3GCR-E in a rectal swab culture, which has been identified as a risk factor in previous studies, some of which were conducted in low-resistance settings. The risk of subsequent infection in patients colonized with 3GCR-E appears to decline over time, but currently available data are insufficient to evaluate when the risk of infection with 3GCR-E in previously colonized patients approaches that of patients with no known previous colonization [Citation18–22]. The third predictor was a previous blood and/or urine culture with 3GCR-E. As for previous colonization, a previous infection with 3GCR-E has been a predictor of 3GCR-E bacteraemia in earlier studies [Citation23,Citation24].

The recently published Swedish study by Fröding et al. also aimed to identify risk factors for bacteraemia with resistant gram-negative bacteria [Citation7]. This study investigated bacteraemia by ESBL-producing Enterobacterales only, while our study also included Enterobacterales resistant to third-generation cephalosporins by other mechanisms, mainly AmpC production. As a suspicion of phenotypic, not genotypic, resistance guides the choice of empiric antibiotic therapy at the emergency department, we included all resistant bacteria, irrespective of resistance mechanism. However, as ESBL-producing Enterobacterales make up the majority of 3GCR-E, the studies are similar enough to be comparable. The study by Fröding et al. also found that a prior culture with ESBL-producing Enterobacterales and hospital care abroad were strong risk factors for community-onset bacteraemia with ESBL-producing Enterobacterales. In addition, they found that a recent prostate biopsy was a risk factor. In the derivation cohort of our study, a recent prostate biopsy showed no correlation with a sequential 3GCR-E bacteraemia and was therefore not included in the score. When the Stockholm score (i.e. our score plus recent prostate biopsy) was applied to the validation cohort of our study, the two scoring systems showed similar performance: sensitivity 53% vs. 55% and specificity 95% vs. 95%. The inclusion of data on recent prostate biopsy in the prediction model for 3GCR-E bacteraemia therefore does not seem to improve the model.

The study by Fröding et al. used a different study population. They included admitted patients that were prescribed antibiotics with activity against Gram-negative bacilli, e.g. patients with a clinical suspicion of infection with Gram-negative bacilli. Our study population, on the other hand, was comprised of patients with a verified bacteraemia with Gram-negative bacilli. Both study designs have drawbacks. Including all patients prescribed antibiotics with activity against Gram-negative bacilli might dilute the results as it is well known that many patients are prescribed antibiotics with unnecessarily broad spectrum. On the other hand, it could be argued that including only patients with proven gram-negative bacteraemia means choosing a study population less representative of the real world, as the emergency room doctor has to choose antibiotic treatment without knowing if a patient is bacteremic. Interestingly, the two different types of study population and somewhat different scoring system nevertheless yielded very similar results.

Several studies have designed advanced algorithms that are promoted as being precise in determining the risk of 3GCR-E bacteraemia [Citation6,Citation12,Citation24]. The scoring systems in some of these studies attain somewhat higher sensitivity and specificity than ours, which had 53% sensitivity and 95% specificity in predicting 3GCR-E bacteraemia. However, the NPV of 97% of our scoring system is high enough to be used to guide empirical antibiotic treatment. Non-severely ill patients with a negative score can safely be given empirical antibiotic treatment not covering 3GCR-E. Furthermore, a sophisticated algorithm might be problematic to apply clinically, whereas our scoring system is easy to use in the emergency department. We chose to include only data accessible in the emergency room to increase clinical utility. We did not, for instance, include medical information and prescriptions from primary healthcare providers, which often cannot be accessed at the hospital. Another strength of this study is the validation of our results in a separate cohort to ensure the applicability of the model.

The majority of variables analyzed in this study were not significant predictors of 3GCR-E bacteraemia, in contrast to results in previous studies. We found no association between previous treatment with third-generation cephalosporins or fluoroquinolones and risk of bacteraemia with 3GCR-E. In a recent study, Isendahl et al. found that use of fluoroquinolones or other antibiotics not active against 3GCR-E within the previous 3 months was associated with 3GCR-E bacteraemia [Citation25]. The study compared patients with ESBL-producing Enterobacteriaceae bacteraemia with randomly selected controls. The objection can be raised that previous consumption of antibiotics might increase the risk of both ESBL- and non-ESBL-producing Enterobacteriaceae and it is therefore difficult to compare the results with those of our study. However, the lack information on antibiotics prescribed outside the hospital might have had an impact on the results in our study, as described below.

Other studies have identified an association between patients living in nursing homes and risk of bacteraemia with 3GCR-E [Citation15]. We found no such association. This could be due to good infection control routines in Swedish nursing homes, with a low rate of transmission of resistant bacteria between patients [Citation26].

Limitations to this study should be noted. The material was collected retrospectively from patient medical records and is therefore dependent on correct documentation of the studied variables. Information on the use of a urinary catheter was inconsistently documented and the number of patients with a urinary catheter might have been higher. The association between the use of a urinary catheter and 3GCR-E bacteraemia might therefore have been underestimated. Data on prescription of antibiotics outside Sahlgrenska University Hospital (e.g. from primary healthcare providers) was not included in the study. In some patients, previous use of third-generation cephalosporins or fluoroquinolones might not have been detected, leading to an underestimation of the risk of 3GCR-E bacteraemia associated with these antibiotics. As third-generation cephalosporins are rarely used outside hospital, this lack of data would primarily affect the previous use of fluoroquinolones. Results of previous bacterial cultures taken by primary healthcare providers were not consistently available in the hospital laboratory data system. As a previous culture with 3GCR-E was a highly significant predictor of 3GCR-E bacteraemia, this lack of data would not, however, impact the conclusions of this study. Information on prostate biopsies performed outside the Sahlgrenska University Hospital, by private healthcare providers, was not available in the electronic patient record system and therefore not included in this study. The association between this procedure and the risk of 3GCR-E bacteraemia might therefore be underestimated.

Regarding previous hospital care abroad, these data are registered on the emergency room charts as a yes or no question with no details about which country. Patients treated in countries with a low prevalence of 3GCR-E may have been included in this group. As the variable was a highly significant predictor of 3GCR-E bacteraemia, this possible dilution effect would not alter the overall conclusions of the study.

There was a significant difference in the prevalences of 3GCR-E bacteraemia in the derivation and validation cohort. We have thoroughly reviewed our data but have not been able to locate any systematic error. All data were validated with the Department of Medical Microbiology, confirming that there was a significant decrease in the prevalence of 3GCR-E bacteraemia between 2016 and 2017. We have no explanation for this decrease, but similar variations have occurred in previous years at Sahlgrenska University Hospital.

The number of patients included in this study is similar to that in several other studies performed in settings of higher resistance. A larger study population, especially a larger validation cohort, would nonetheless strengthen the conclusions. Finally, this study was conducted at a large tertiary care university hospital and the results are not directly generalizable to other types of hospitals, even in countries with similar healthcare systems and rates of resistance.

To summarize, a score based on the presence of one of the following: previous culture from blood and/or urine or rectal swab with 3GCR-E or previous hospital care abroad, can be used to predict bacteraemia with 3GCR-E. The validity of these results is supported by another recent Swedish study. Based on the results of this study, it can be recommended that patients with symptoms of gram-negative bacteraemia and at least one of the above variables, should receive empirical treatment covering 3GCR-E, while patients without any of the above variables and who are not severely ill, can be treated empirically without covering 3GCR-E.

Disclosure statement

The authors report no conflict of interest.

Additional information

Funding

This study was supported by grants from the Swedish State under the agreement between the Swedish government and the country councils, the ALF agreement (ALF-70150, ALF-73490).

References

  • Pitout JDD, Laupland KB. Extended-spectrum beta-lactamase-producing Enterobacteriaceae: an emerging public-health concern. Lancet Infect Dis. 2008;8(3):159–166.
  • Folkhälsomyndigheten. Extended spectrum beta-lactamase (ESBL) – sjukdomsstatistik — Folkhälsomyndigheten [Internet]. Available from: https://www.folkhalsomyndigheten.se/folkhalsorapportering-statistik/statistikdatabaser-och-visualisering/sjukdomsstatistik/extended-spectrum-beta-lactamase-esbl
  • Van Aken S, Lund N, Ahl J, et al. Risk factors, outcome and impact of empirical antimicrobial treatment in extended-spectrum β-lactamase-producing Escherichia coli bacteraemia. Scand J Infect Dis. 2014;46(11):753–762.
  • Anesi JA, Lautenbach E, Nachamkin I, et al. Poor clinical outcomes associated with community-onset urinary tract infections due to extended-spectrum cephalosporin-resistant Enterobacteriaceae. Infect Control Hosp Epidemiol. 2018;39(12):1431–1435.
  • Rottier WC, van Werkhoven CH, Bamberg YRP, et al. Development of diagnostic prediction tools for bacteraemia caused by third-generation cephalosporin-resistant Enterobacteria in suspected bacterial infections: a nested case-control study. Clin Microbiol Infect. 2018;24(12):1315–1321.
  • MacFadden DR, Coburn B, Shah N, et al. Decision-support models for empiric antibiotic selection in Gram-negative bloodstream infections. Clin Microbiol Infect. 2019;25(1):108.e1–108.e7.
  • Fröding I, Valik JK, Bolinder L, et al. Prediction of bloodstream infection caused by extended-spectrum β-lactamase-producing Enterobacterales in patients with suspected community-onset sepsis. Int J Antimicrob Agents. 2019;53(6):820–829.
  • EUCAST. Antimicrobial susceptibility testing [Internet]. eucast.org [cited 2019 Dec 17]. Available from: http://www.eucast.org/ast_of_bacteria/
  • Brolund A. Overview of ESBL-producing Enterobacteriaceae from a Nordic perspective. Infect Ecol Epidemiol. 2014;4.
  • Lee C-C, Lee C-H, Hong M-Y, et al. Propensity-matched analysis of the impact of extended-spectrum β-lactamase production on adults with community-onset Escherichia coli, Klebsiella species, and Proteus mirabilis bacteremia. J Microbiol Immunol Infect. 2018;51(4):519–526.
  • Liu VX, Fielding-Singh V, Greene JD, et al. The timing of early antibiotics and hospital mortality in sepsis. Am J Respir Crit Care Med. 2017;196(7):856–863.
  • Falcone M, Tiseo G, Dentali F, et al. Predicting resistant etiology in hospitalized patients with blood cultures positive for Gram-negative bacilli. Eur J Intern Med. 2018;53:21–28.
  • Johnson SW, Anderson DJ, May DB, et al. Utility of a clinical risk factor scoring model in predicting infection with extended-spectrum β-lactamase-producing Enterobacteriaceae on hospital admission. Infect Control Hosp Epidemiol. 2013;34(4):385–392.
  • Tham J, Odenholt I, Walder M, et al. Risk factors for infections with extended-spectrum beta-lactamase-producing Escherichia coli in a county of Southern Sweden. Infect Drug Resist. 2013;6:93–97.
  • Lee C-H, Chu F-Y, Hsieh C-C, et al. A simple scoring algorithm predicting extended-spectrum β-lactamase producers in adults with community-onset monomicrobial Enterobacteriaceae bacteremia: matters of frequent emergency department users. Medicine (Baltimore). 2017;96(16):e6648.
  • ECDC. Surveillance of antimicrobial resistance in Europe – annual report of the European Antimicrobial Resistance Surveillance Network (EARS-Net) 2017. Stockholm; 2018.
  • Khawaja T, Kirveskari J, Johansson S, et al. Patients hospitalized abroad as importers of multiresistant bacteria—a cross-sectional study. Clin Microbiol Infect. 2017;23(9):673.e1–673.e8.
  • Isendahl J, Giske CG, Hammar U, et al. Temporal dynamics and risk factors for bloodstream infection with extended-spectrum β-lactamase-producing bacteria in previously-colonized individuals: National Population-based Cohort Study. Clin Infect Dis. 2019;68(4):641–649.
  • Rottier WC, Bamberg YRP, Dorigo-Zetsma JW, et al. Predictive value of prior colonization and antibiotic use for third-generation cephalosporin-resistant Enterobacteriaceae bacteremia in patients with sepsis. Clin Infect Dis. 2015;60:1622–1630.
  • Ny S, Löfmark S, Börjesson S, et al. Community carriage of ESBL-producing Escherichia coli is associated with strains of low pathogenicity: a Swedish nationwide study. J Antimicrob Chemother. 2017;72(2):582–588.
  • Lindblom A, Karami N, Magnusson T, et al. Subsequent infection with extended-spectrum β-lactamase-producing Enterobacteriaceae in patients with prior infection or fecal colonization. Eur J Clin Microbiol Infect Dis. 2018;37(8):1491–1497.
  • Denis B, Lafaurie M, Donay JL, et al. Prevalence, risk factors, and impact on clinical outcome of extended-spectrum beta-lactamase-producing Escherichia coli bacteraemia: a five-year study. Int J Infect Dis. 2015;39:1–6.
  • Goodman KE, Lessler J, Cosgrove SE, et al. A clinical decision tree to predict whether a bacteremic patient is infected with an extended-spectrum β-lactamase-producing organism. Clin Infect Dis. 2016;63(7):896–903.
  • Augustine MR, Testerman TL, Justo JA, et al. Clinical risk score for prediction of extended-spectrum β-lactamase-producing Enterobacteriaceae in bloodstream isolates. Infect Control Hosp Epidemiol. 2017;38(3):266–272.
  • Isendahl J, Giske CG, Tegmark Wisell K, et al. Risk factors for community-onset bloodstream infection with extended-spectrum β-lactamase-producing Enterobacteriaceae: national population-based case-control study. Clin Microbiol Infect. 2019;25(11):1408–1414.
  • Sundvall P-D, Elm M, Gunnarsson R, et al. Antimicrobial resistance in urinary pathogens among Swedish nursing home residents remains low: a cross-sectional study comparing antimicrobial resistance from 2003 to 2012. BMC Geriatr. 2014;14(1):30.