To identify characteristics associated with particular groups of uropathogens in catheter-associated urinary tract infection (CA–UTI) and to develop clinical prediction rules for identifying these groups.
Methods
Demographic, clinical and microbiological data were analysed from patients with CA–UTI. Infections were categorized into enteric Gram-negative rods, nonfermenters, Gram-positive cocci and fungal. Variables were analysed using univariate and multiple logistic regression analyses, and were used to develop clinical prediction rules.
Results
A total of 492 patients were included in the study. Candida species were the most common uropathogens (30.7%), followed by enterococci (17.3%), Escherichia coli (12.0%), Pseudomonas spp. (10.8%), Klebsiella spp. (7.9%) and staphylococci (6.5%). Clinical prediction rules for the bacterial uropathogenic groups showed poor-to-fair discriminatory power, with sensitivities of <40% and specificities of >90%. However, clinical prediction rules showed good discriminatory power for fungal infections, with a sensitivity of 67.3% and a specificity of 78.1%.
Conclusions
Clinical prediction rules developed for identifying specific groups of bacterial uropathogens in patients with CA–UTI had a low sensitivity, whereas those for fungal infections showed good discriminatory power. Further studies to develop more refined and sensitive tools for predicting specific bacterial uropathogens in CA–UTI are warranted.
Catheter-associated urinary tract infection (CA–UTI) is the most common nosocomial infection worldwide, comprising ∼40% of such infections.1,2 The acquisition of CA–UTI is associated with increased morbidity, length of hospital stay and healthcare-associated costs. Therefore, early identification of causative uropathogens and initiation of appropriate antibiotic treatment is of great importance. In contrast to community-acquired, uncomplicated UTI, a wide variety of micro-organisms have been isolated from patients with CA–UTI.3,4 In addition, because uropathogens in CA–UTI tend to show high rates of antibiotic resistance, the selection of appropriate empirical antibiotic treatment may be difficult.5,6 Until microbiological results are available, appropriate selection of antibiotics for CA–UTI depends on known susceptibility patterns of suspected bacterial uropathogens, with third-generation cephalosporins or fluoroquinolones being used for enteric Gram-negative rods, ceftazidime for nonfermenters, ampicillin or vancomycin for enterococci, and methicillin for staphylococci. Both the aetiological diversity and the high rate of resistance in CA–UTI can lead to either the misuse or overuse of antibiotics as empirical treatments. Predicting the presence of specific uropathogens in CA–UTI before microbiological results are available would therefore be useful in selecting appropriate antibiotic treatment strategies.
Clinical prediction rules are tools designed to assist clinicians in making medical decisions when caring for patients.7,8 They usually consist of several predictors derived from multivariable analyses, such as demographic features, clinical characteristics, laboratory results and radiographic findings. Clinical prediction rules may provide a means of predicting the presence of specific uropathogens in patients with CA–UTI. The present study was performed to identify aetiological micro-organisms in patients with CA–UTI, and to characterize factors associated with the major groups of uropathogens in order to develop clinical prediction rules for identifying specific groups of uropathogens.
Patients and methods
Patients
This cohort study was conducted at seven university hospitals in the Republic of Korea (Kangwon National University Hospital, Chuncheon; Yeungnam University Medical Centre, Daegu; Seoul National University Bundang Hospital, Seongnam; Chonnam National University Hospital, Gwangju; Wonju Severance Christian Hospital, Wonju; Inje University Busan-Paik Hospital, Busan; Inje University Sanggye-Paik Hospital, Seoul) between June and August 2012. Urine culture results from the laboratory databases at each hospital were screened to identify patients over 18 years of age who had urine culture performed after 48 h of admission. After a review of their medical records, patients who had a urinary catheter either at the time of urine culture or within the preceding 48 h and had CA–UTI (including catheter-associated asymptomatic bacteriuria) meeting the Infectious Diseases Society of America criteria3 were included in the study. Patients with polymicrobial infections were excluded. Only the first CA–UTI episode in each patient during the study period was considered.
The study protocol was approved by the institutional review boards of each of the seven hospitals involved; the requirement for informed consent from the study participants was waived by the boards.
Data collection
Data were collected using standardized case-report forms as soon as patients were enrolled in the study. Collected data comprised: demographic data; comorbid conditions; presence of urinary tract obstruction or vesico–ureteral reflux; type and duration of urinary catheterization; type and duration of other procedures; use of antibiotics within the preceding 30 days; presence of fever; organisms isolated from urine, blood or other cultures within the preceding 3 months; antibiotic resistance of major uropathogens; immunosuppressive therapy within the preceding 30 days.
Demographic data included age, sex, care site (medical ward, surgical ward or intensive care unit), bed-ridden state and duration of stay at the time of urine culture, which was divided into short- (≤ 14 days) or long-term hospitalization (>14 days). Comorbidity was assessed using the Charlson comorbidity index.9
The type and duration of urinary catheterization was divided into three categories: short-term indwelling catheterization (≤ 14 days); long-term indwelling catheterization (>14 days); intermittent catheterization.10 Data on other procedures included surgery, tracheostomy (including endotracheal intubation), gastrostomy (including nasogastric tube insertion), percutaneous nephrostomy, mechanical ventilation, renal replacement therapy, insertion of a biliary drainage tube, double-J stent or central venous catheter and other invasive procedures. In addition, a postvoid residual urine >100 ml, placement of a cystostomy tube within the preceding 48 h and the presence of chronic renal failure or end-stage renal disease were recorded.
Microbiological data relating to the CA–UTI were collected from all enrolled patients, including the susceptibility of the major uropathogens to antibiotics (third-generation cephalosporins and fluoroquinolones for enteric Gran-negative rods, ceftazidime for nonfermenters, ampicillin and vancomycin for enterococci, and methicillin for staphylococci).
Statistical analyses
Based on the antibiotic classes for empirical treatment using known susceptibility patterns, patients with CA–UTI were divided into four groups, based on the infection type: enteric Gram-negative rods; nonfermenters; Gram-positive cocci; fungal. For univariate analysis, variables in one group were compared with those in all the other three groups combined. Pearson’s χ2-test was used to analyse categorical variables and Student’s t-test for continuous variables. Parameters with a P-value <0.10 on univariate analysis were included in multiple logistic regression analyses, using conditional forward selection to identify factors associated with specific groups of uropathogens. Levels of significance of P <0.10 for inclusion and P >0.05 for exclusion were used in the multiple logistic regression analyses. Goodness-of-fit of the regression model was evaluated using the Hosmer–Lemeshow test.
Clinical prediction rules for predicting specific groups of uropathogens in patients with CA–UTI were developed, based on factors derived from the multiple logistic regression models. The rules were weighted by assigning the nearest whole number points to all factors in proportion to their regression coefficients. These clinical prediction rules were then applied to the whole dataset, and their sensitivities and specificities were calculated from the receiver operating characteristic (ROC) curves. Optimal cut-off values for clinical prediction rule scores were determined using the point at which the positive likelihood ratio was >3. As a surrogate of internal validation while adjusting the model parameters for potential overfitting, 95% confidence intervals of the area under the ROC curve (C-statistic) were calculated by bootstrapping with 1000 replications.
A two-tailed P-value <0.05 was considered to be statistically significant for all analyses. Statistical analyses were performed using IBM SPSS® Statistics software version 20.0 (IBM, Somers, NY, USA) and R software (R Foundation for Statistical Computing, Vienna, Austria, http://www.r-project.org).
Results
Aetiological uropathogens
A total of 560 micro-organisms were isolated from 526 patients found to have CA–UTI during the study period. Of these patients, 34 had polymicrobial infections and so were excluded, leaving a total of 492 suitable patients with data available for analysis.
Candida species were the most common uropathogens (30.7%), followed by enterococci (17.3%), Escherichia coli (12.0%), Pseudomonas spp. (10.8%), Klebsiella spp. (7.9%) and staphylococci (6.5%) (Table 1). Overall, >70% of CA–UTI cases were caused by uropathogens other than enteric Gram-negative rods. A similar pattern of aetiological uropathogenic distribution was observed at all seven of the study hospitals (data not shown). Of the 138 enteric Gram-negative rod isolates, 81 (58.7%) and 74 (53.6%) showed resistance to third-generation cephalosporins and fluoroquinolones, respectively. Of the 73 nonfermenter isolates, 37 (50.7%) were resistant to ceftazidime. Of the 85 enterococcal isolates, 53 (62.4%) and eight (9.4%) showed resistance to ampicillin and vancomycin, respectively. Of the 32 staphylococcal isolates, 20 (62.5%) were methicillin resistant.
Aetiological uropathogens in 492 patients with catheter-associated urinary tract infections caused by single pathogenic agents.
Uropathogen
Total n = 492
Intermittent catheterization n = 14
Short-term indwelling catheterization, ≤14 days n = 344
Long-term indwelling catheterization, >14 days n = 134
At the time of urine culture or within the preceding 48 h, 344 (69.9%) patients had short-term indwelling urinary catheters, 134 (27.2%) patients had long-term indwelling urinary catheters, and 14 (2.9%) patients were intermittently catheterized. Enterococci and staphylococci were more frequently isolated in patients with short-term indwelling urinary catheters than in other patients (P = 0.007 and P = 0.001, respectively), while Candida species were more commonly isolated in those with long-term indwelling urinary catheters (P <0.001) than in those in the other groups.
Demographic and clinical characteristics
Demographic and clinical characteristics of patients with CA–UTI are shown in Table 2. Of the 492 patients, 138 (28.0%) had enteric Gram-negative rod infections, 73 (14.8%) had nonfermenter infections, 122 (24.1%) had Gram-positive cocci infections and 159 (32.2%) had fungal infections.
Demographic and clinical characteristics of patients with catheter-associated urinary tract infection divided according to uropathogenic group.
Data presented as n (%) of patients; Pearson’s χ2-test was used to compare categorical variables and Student’s t-test for continuous variables.
NF, GPC and fungal infections.
GNR, GPC and fungal infections.
GNR, NF and fungal infections.
GNR, NF and GPC.
The median age of the patients was 70 years (interquartile range 59–78 years). The median duration of hospital stay before urine culture was 13 days (interquartile range 7–26 days). A total of 147 (29.9%) patients were admitted to the intensive care unit, and the median duration of stay in the intensive care unit before urine culture was 11 days (interquartile range 5–19 days). Among the 478 patients with indwelling urinary catheters, the median duration of catheterization was 8 days (interquartile range 4–16 days). Of the total 492 patients with CA–UTI, 459 (93.3%) had one or more comorbid conditions, and the median Charlson comorbidity index score was 2 (interquartile range 1–3). Parameters with a P-value <0.10 in Table 2 were included in the multiple logistic regression analyses.
Factors associated with specific uropathogens
Factors that had a P-value <0.10 on univariate analysis and that subsequently showed significance in multiple logistic regression models (to identify factors associated with specific groups of uropathogens in patients with CA–UTI) are presented in Table 3. The Hosmer–Lemeshow test showed the goodness-of-fit of these models to be appropriate (P = 0.580 for the enteric Gram-negative rod group, 0.828 for the nonfermenter group, 0.350 for the Gram-positive cocci group and 0.998 for the fungal infection group).
Factors associated with specific groups of uropathogens on multiple regression analysis in patients with catheter-associated urinary tract infections.
Uropathogenic group
Variable
Adjusted odds ratio
95% CI
β coefficient
P- value
Score points
Enteric Gram-negative rods, n = 138
Female
1.909
1.220, 2.986
0.646
0.005
1
No surgery within preceding 30 days
1.645
1.006, 2.690
0.498
0.047
1
No gastrostomy tube in place within preceding 7 days
2.168
1.160, 4.055
0.774
0.015
2
No antibiotic use within preceding 30 days
2.730
1.560, 4.779
1.004
<0.001
2
Presence of fever
2.230
1.422, 3.496
0.802
<0.001
2
Nonfermenters, n = 73
Admission to surgical ward
1.864
1.108, 3.138
0.623
0.019
1
Postvoid residual urine >100 ml
1.868
1.011, 3.453
0.625
0.046
1
Cystostomy tube in place within preceding 48 h
3.729
1.472, 9.451
1.316
0.006
2
Same species isolated from specimens other than urine within preceding 3 months
2.719
1.350, 5.479
1.000
0.005
2
Gram-positive cocci, n = 122
Age ≤65 years
1.839
1.158, 2.923
0.609
0.010
1
Duration of hospital stay at time of urine culture ≤14 days
3.176
1.942, 5.195
1.156
<0.001
2
No comorbid conditions
3.713
1.670, 8.256
1.312
0.001
2
Not bed-ridden
1.962
1.185, 3.251
0.674
0.009
1
No antibiotic use within preceding 30 days
1.782
1.009, 3.146
0.578
0.046
1
Fungal infections, n = 159
Admission to medical ward
2.681
1.524, 4.716
0.986
0.001
2
Duration of indwelling urinary catheterization at time of urine culture >14 days
2.124
1.311, 3.441
0.753
0.002
1
Presence of comorbid conditions
5.622
1.543, 20.486
1.727
0.009
3
Surgery within preceding 30 days
2.389
1.347, 4.237
0.871
0.003
1
Tracheostomy tube in place within preceding 7 days
2.874
1.739, 4.749
1.056
<0.001
2
Central venous catheter in place within preceding 48 h
1.925
1.192, 3.107
0.655
0.007
1
Antibiotic use within preceding 30 days
10.239
3.000, 34.946
2.326
<0.001
4
Absence of fever
2.580
1.530, 4.348
0.948
<0.001
1
CI, confidence interval.
Use of the clinical prediction rules (derived from the multiple logistic regression models for identifying specific groups of uropathogens in patients with CA–UTI) is summarized in Table 4. After bootstrapping for 1000 repetitions, adjusted C-statistics for the clinical prediction rules were 0.706 (95% confidence intervals [CI] 0.657, 0.755) for the enteric Gram-negative rod group, 0.643 (95% CI 0.574, 0.712) for the nonfermenter group, 0.720 (95% CI 0.669, 0.769) for the Gram-positive cocci group and 0.803 (95% CI 0.764, 0.843) for the fungal group.
Performance of the clinical prediction rule (CPR) score in identifying specific groups of uropathogens in patients with catheter-associated urinary tract infections.
Uropathogenic group
CPR score
Sensitivity
Specificity
Positive predictive value
Negative predictive value
Positive likelihood ratio
Negative likelihood ratio
Range
C-statistic
Cut-off value
Value
95% CI
%
95% CI
%
95% CI
%
95% CI
%
95% CI
Value
95% CI
Value
95% CI
Enteric Gram-negative rods
0–8
0.706
0.656, 0.756
≥4
71.0
62.6, 78.3
56.8
51.4, 62.0
39.0
33.0, 45.4
83.4
78.0, 87.7
1.643
1.400, 1.928
0.510
0.392, 0.666
≥5
47.1
38.6, 55.8
78.0
73.2, 82.1
45.4
37.2, 54.0
79.1
74.4, 83.2
2.138
1.642, 2.783
0.678
0.579, 0.795
≥6
31.9
24.4, 40.4
92.4
89.0, 94.8
62.0
49.6, 73.0
77.7
73.3, 81.5
4.180
2.701, 6.471
0.737
0.658, 0.827
Nonfermenters
0–6
0.643
0.570, 0.716
≥1
74.0
62.2, 83.2
43.7
38.9, 48.6
18.6
14.4, 23.7
90.6
85.5, 94.1
1.313
1.119, 1.541
0.596
0.402, 0.882
≥2
41.1
29.9, 53.2
81.6
77.5, 85.1
28.0
20.0, 37.7
88.8
85.1, 91.7
2.236
1.590, 3.144
0.722
0.595, 0.875
≥3
17.8
10.1, 28.9
95.0
92.3, 98.0
38.2
22.7, 56.4
86.9
83.4, 89.8
3.553
1.863, 6.775
0.865
0.777, 0.963
Gram-positive cocci
0–7
0.719
0.668, 0.771
≥3
71.3
62.3, 79.0
58.9
53.7, 63.9
36.4
30.3, 42.9
86.2
81.1, 90.1
1.736
1.470, 2.049
0.487
0.367, 0.646
≥4
43.4
34.6, 52.7
85.1
81.0, 88.5
49.1
39.4, 58.8
82.0
77.7, 85.6
2.922
2.129, 4.012
0.664
0.568, 0.777
≥5
13.1
7.9, 20.7
97.0
94.6, 98.4
59.3
39.0, 77.0
77.2
73.1, 80.9
4.411
2.105, 9.245
0.895
0.836, 0.960
Fungal infections
0–15
0.803
0.763, 0.844
≥9
94.3
89.2, 97.2
35.4
30.3, 40.9
41.1
36.0, 46.3
92.9
86.6, 96.5
1.461
1.338, 1.596
0.160
0.083, 0.305
≥10
86.8
80.3, 91.5
56.1
50.6, 61.5
48.6
42.7, 54.6
89.9
84.8, 93.5
1.980
1.728, 2.268
0.235
0.157, 0.352
≥11
67.3
59.3, 74.4
78.1
73.2, 82.3
59.4
51.9, 66.6
83.3
78.6, 87.2
3.070
2.439, 3.863
0.419
0.335, 0.524
CI, confidence interval.
Cut-off values for the clinical prediction rule scores that gave a positive likelihood ratio >3 were 6 to identify an enteric Gram-negative rod infection, 3 to identify a nonfermenter infection, 5 to identify a Gram-positive coccal infection, and 11 to identify a fungal infection. Using these cut-off values, the sensitivities of the clinical prediction rule scores for enteric Gram-negative rods, nonfermenters and Gram-positive cocci were <40%, despite their high specificities (Table 4). However, the clinical prediction rule score for fungal infections had a sensitivity of 67.3% and a specificity of 78.1%.
Discussion
The present study results show that the spectrum of aetiological micro-organisms in CA–UTI is very diverse, with no single predominant bacterial uropathogen. These findings were consistent across all the hospitals involved in the study. Other studies have also reported that the aetiological uropathogens causing nosocomial UTI vary considerably.1,11 In contrast to community-acquired UTI, in which Escherichia coli is the single most common pathogen, the aetiological diversity (with no particularly predominant uropathogens) in CA-UTI leads to difficulty in predicting aetiological micro-organisms.4,12
Recent reviews on nosocomial UTI have suggested that patients should be empirically treated with a cephalosporin or a penicillin/β-lactamase inhibitor with antipseudomonal activity, a carbapenem, a fluoroquinolone, aztreonam, or an aminoglycoside.11,13 All of these regimens were mainly targeted at Gram-negative rod or nonfermenter infections. However, the wide spectrum of aetiological micro-organisms (and the absence of individual clinical characteristics able to predict the presence of specific uropathogens) underline the impracticality of using particular antibiotic regimens as empirical treatments for CA–UTI. As a result, some clinical practice guidelines do not specify certain antibiotic regimens for the empirical treatment of CA-UTI, but recommend that optimal treatment should be chosen on the basis of clinical judgement and local antibiotic resistance.3,14
Identification of factors associated with specific uropathogens could help predict the presence of particular aetiological micro-organisms in CA–UTI and aid the selection of appropriate treatment. To our knowledge, few studies have identified factors associated with specific pathogens in CA–UTI.15,16 In a report on nosocomial UTI, the following factors were correlated with the presence of specific uropathogens: previous history of UTI urinary tract obstruction or suprapubic catheterization were correlated with Proteus spp. infections; history of hospitalization, urinary stones or any type of catheterization were correlated with Pseudomonas spp. infections; history of hospitalization or prior use of antibiotics were correlated with Klebsiella spp. infections; prior use of antibiotics or insertion of a ureteral stent were correlated with Candida spp.15 In another study, male sex, duration of intensive care unit stay, use of antibiotics at admission, and transfer from another intensive care unit were associated with nosocomial UTI caused by Pseudomonas spp.16 However, it is important to assess how accurate such factors are in predicting the presence of specific uropathogens in patients with CA–UTI.
In the present study, all variables that had been previously identified or considered to be associated with the occurrence of CA–UTI caused by specific uropathogens were included; the aim was to develop clinical prediction rules, based on factors derived from multiple logistic regression models, to identify specific groups of uropathogens in patients with CA–UTI. The clinical prediction rules for enteric Gram-negative rods, nonfermenters and Gram-positive cocci showed poor-to-fair discriminatory powers, with C-statistics of 0.706 (95% CI 0.656, 0.756), 0.643 (95% CI 0.570, 0.716) and 0.719 (95% CI 0.668, 0.771), respectively. Although cut-off values for clinical prediction rule scores were chosen to give a positive likelihood ratio >3, their sensitivities were <40%, indicating a lack of accuracy as screening tools to predict bacterial groups in patients with CA–UTI. This may be because multifactorial interactions affected the pathogenesis of such infections; in addition, such situations are made more complex by the wider variety of micro-organisms isolated from those with CA–UTI compared with noncatheterized patients or those with community-acquired UTI.11,17 The present study results therefore suggest that appropriate selection of antibiotic regimens for the empirical treatment of CA–UTI should depend on the results of a urine smear and local resistance patterns.
In contrast, clinical prediction rules showed good discriminatory power for fungal infections, with a C-statistic of 0.803 (95% CI 0.763, 0.844). Using a cut-off value ≥11, fungal infections could be predicted with a sensitivity of 67.3% and a specificity of 78.1%. This indicates that a clinical prediction rule score of ≥11 could be used to identify patients with CA–UTI at high risk for the acquisition of fungal uropathogens, suggesting that empirical treatment with antibacterial agents may be postponed in such patients until microbiological results are available.
The present study has a number of limitations. First, the clinical prediction rules were developed to help identify major groups of uropathogens rather than specific species, in patients with CA–UTI. Nevertheless, predicting groups of uropathogens could enable a reduction in the number of antibiotic classes commonly used for empirical treatment; this strategy would also help to facilitate optimal selection of regimens. Secondly, the present study did not differentiate between symptomatic CA–UTI and catheter-associated asymptomatic bacteriuria. However, the inclusion of asymptomatic bacteriuria is unlikely to distort the objective of the present study as it usually precedes symptomatic UTI. Finally, clinical prediction rules for fungal infections require external validation and impact analysis.
In summary, clinical prediction rules were developed for identifying specific groups of uropathogens in patients with CA–UTI. Those for identifying specific bacterial groups had a low sensitivity because of the diverse spectrum of aetiological micro-organisms and the multiple pathogenic factors present in the hospital setting. In contrast, clinical prediction rules for fungal infections showed good discriminatory power. Further studies, to develop more refined and sensitive tools for predicting specific bacterial uropathogens in CA–UTI, are warranted. Until these are available, clinicians should be careful when choosing empirical antibiotic regimens in patients with CA–UTI.
Footnotes
Declaration of conflicting interest
The authors declare that there are no conflicts of interest.
Funding
This work was supported by the 2013 Inje University Research Grant (Baek-Nam Kim).
Acknowledgements
The authors thank Professor Sung-Bin Chon, Seoul National University Hospital, for his useful comments on the statistical analyses.
TambyahPA. Catheter-associated urinary tract infections: diagnosis and prophylaxis. Int J Antimicrob Agents2004; 24(suppl 1): S44–S48.
3.
HootonTMBradleySFCardenasDD. Diagnosis, prevention, and treatment of catheter-associated urinary tract infection in adults: 2009 International Clinical Practice Guidelines from the Infectious Diseases Society of America. Clin Infect Dis2010; 50: 625–663.
4.
BouzaESan JuanRMuñozP. A European perspective on nosocomial urinary tract infections I. Report on the microbiology workload, etiology and antimicrobial susceptibility (ESGNI-003 study). European Study Group on Nosocomial Infections. Clin Microbiol Infect2001; 7: 523–531.
5.
HidronAIEdwardsJRPatelJ. NHSN annual update: antimicrobial-resistant pathogens associated with healthcare-associated infections: annual summary of data reported to the National Healthcare Safety Network at the Centers for Disease Control and Prevention, 2006–2007. Infect Control Hosp Epidemiol2008; 29: 996–1011.
6.
RosenthalVDMakiDGJamulitratS. International Nosocomial Infection Control Consortium (INICC) report, data summary for 2003–2008, issued June 2009. Am J Infect Control2010; 38: 95–104.e2.
7.
WassonJHSoxHCNeffRK. Clinical prediction rules. Applications and methodological standards. N Engl J Med1985; 313: 793–799.
8.
LaupacisASekarNStiellIG. Clinical prediction rules. A review and suggested modifications of methodological standards. JAMA1997; 277: 488–494.
9.
CharlsonMEPompeiPAlesKL. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis1987; 40: 373–383.
10.
MoolaSKonnoR. A systematic review of the management of short-term indwelling urethral catheters to prevent urinary tract infections. JBI Libr Syst Rev2010; 8: 695–729.
11.
WagenlehnerFMCekMNaberKG. Epidemiology, treatment and prevention of healthcare-associated urinary tract infections. World J Urol2012; 30: 59–67.
12.
CullenIMManeckshaRPMcCullaghE. The changing pattern of antimicrobial resistance within 42,033 Escherichia coli isolates from nosocomial, community and urology patient-specific urinary tract infections, Dublin, 1999–2009. BJU Int2012; 109: 1198–1206.
13.
PelegAYHooperDC. Hospital-acquired infections due to Gram-negative bacteria. N Engl J Med2010; 362: 1804–1813.
14.
TenkePKovacsBBjerklund JohansenTE. European and Asian guidelines on management and prevention of catheter-associated urinary tract infections. Int J Antimicrob Agents2008; 31(suppl 1): S68–S78.
15.
JohansenTECekMNaberKG. Hospital acquired urinary tract infections in urology departments: pathogens, susceptibility and use of antibiotics. Data from the PEP and PEAP-studies. Int J Antimicrob Agents2006; 28(suppl 1): S91–S107.
16.
VenierAGLavigneTJarnoP. Nosocomial urinary tract infection in the intensive care unit: when should Pseudomonas aeruginosa be suspected? Experience of the French national surveillance of nosocomial infections in the intensive care unit, Rea-Raisin. Clin Microbiol Infect2012; 18: E13–E15.
