This two-phase study was conducted from August 2011 to July 2012, in a 501-bed University Hospital of Minas Gerais, Brazil. Firstly, a 1:2 ratio case–control protocol was run, in order to identify characteristics associated with colonization or infection by resistant Enterobacteriaceae in samples obtained during the first 48h of hospitalization. Thereafter, in the second phase of the study, the included patients were followed up in order to identify the impact of these microorganisms in some clinical endpoints. Patients at least 18 years old, whose culture requested by the attending physician grew enterobacteria in samples collected in pre-specified days after hospitalization were assessed for potential inclusion. Patients with enterobacteria resistant to carbapenems were excluded from the study. All included patients lived in the metropolitan area of Belo Horizonte, Minas Gerais state.

Eligible patients were screened in the emergency service of our hospital and categorized into one of two groups: (i) cases, referring to those colonized or infected by Enterobacteriaceae resistant to third-generation cephalosporins and, (ii) controls, patients colonized or infected by Enterobacteriaceae susceptible to third-generation cephalosporins. For each case included in the study, two controls were selected sequentially on the same day. Each patient was included in the study only once.

Patients were further categorized as (i) infected patients, if there was a clinical suspicion of active infection leading to prescription of antibiotic therapy, and (ii) colonized patients, if there was no clinical suspicion of infection or if the enterobacteria were isolated from surveillance samples. Surveillance samples were samples collected at the time of hospital admission via swabs, as a routine procedure in patients transferred from other hospitals aiming at detecting the presence of patients colonized with multidrug-resistant bacteria.

Demographic, clinical, and epidemiological data were collected using a dedicated case report form. Data was obtained from interviews and by consulting electronic and printed records. The following variables were collected: age, sex, microbiological data, site of infection, presence of comorbidities (diabetes, chronic renal failure, liver failure, solid tumor, malignant hematological disease, heart failure, and others), known immunosuppression (HIV infection, neutropenia with PMN <500cells/mm3, use of corticosteroids in doses above 15mg/day of prednisone or equivalent, use of other immunosuppressive drugs), hospitalization and previous use of antibiotics, performance of invasive procedures in the last four weeks, use of invasive prosthesis, presence of stoma, recurrent urinary tract infection (UTI), adequacy of initial empiric antibiotic treatment (only for infected patients), hospitalization at the intensive care unit (ICU), response to the antibiotic therapy, all-cause hospital mortality, and hospital length of stay.

Recurrence of urinary tract infection (UTI) was defined as the development of two or more documented episodes in the last six months. For the subgroup of infected patients, empirical treatment was considered adequate when an antimicrobial regimen included an active antibiotic against the isolated enterobacteria, and was initiated at the recommended dose in the first 24h after sample collection. Inadequate antimicrobial treatment included absence of antimicrobial agents indicated for a specific class of microorganisms and administration of an antimicrobial agent to which the isolated microorganism was resistant.10

In this subset of patients, clinical response to the antimicrobial treatment was classified as: “complete or partial resolution”, referring to the patients who presented total or partial improvement of fever, leukocytosis and clinical signs of infection, and “therapeutic failure or uncertain result”, for those who, respectively, showed no decrease in these parameters at all or persisted with symptoms and signs that were not clearly attributable to infection.11

For the infected patients clinical outcomes were ICU admittance during follow-up, all-cause hospital mortality, and length of hospital stay.

Bacterial identification and antimicrobial susceptibility testing, including production of β-lactamases, were carried out in accordance with the recommendations of the Clinical and Laboratory Standards Institute (CLSI).12 Identification and ESBL production were confirmed by using the API 20E® and Vitek II® systems (both from bioMérieux).

Categorical variables were compared by using the Pearson's chi-square test (or Fischer's exact test, as indicated), and presented with the corresponding 95% confidence intervals (CI). Continuous variables were compared using Student's t-test. In cases of non-normal distribution of quantitative variables, non-parametric tests were used, such as Mann–Whitney U test.

Non-conditional multivariate logistic regression analysis was conducted to determine the independent risk factors associated with the presence of resistant enterobacteria. All variables presenting the p-value <0.20 in univariate analysis were included in the multivariate model. Furthermore, the independent risks factors associated with in-hospital death were investigated in a multivariate analysis excluding the colonized patients. Using the same criteria described above (p-value <0.20 in the univariate analysis), the following variables were include in this analysis: age, presence of two or more comorbidities, presence of UTI, admittance to ICU, inappropriate empirical antibiotic therapy, and presence of resistant enterobacteria (cases). For the multivariate models, backward method was used, and the Hosmer and Lemeshow test was applied to assess the models’ adequacy.

Finally, to evaluate the impact of the infections caused by enterobacteria resistant to third-generation cephalosporins on hospital length of stay, a Cox proportional hazards analysis was conducted. The variable “age” was included as a binary variable (greater than or less than 50 years). Discharges by death were censored. For all analyses, a two-tailed value of p<0.05 was considered significant. Statistical analyses were performed by using SPSS version 16.0 software.

During the study period, 238 adult patients had one or more Enterobacteriaceae isolated from samples collected in the first 48h after hospital admission. Of this total, 62 (26.0%) patients had Enterobacteriaceae resistant to third-generation cephalosporins. The flowchart presenting the inclusion procedures is summarized in Fig. 1.

Fig. 1.

Flowchart presenting the procedures for inclusion in the study.

(0.29MB).

Results of univariate analysis are shown in Table 1. Case patients had higher mean age (56.7±19.1 vs. 46.9±19.6, p=0.001), were more frequently male (53.2% vs. 25.8%; p<0.001), and usually presented two or more comorbidities (50.0%vs. 34.7%, p=0.044). Specifically, heart failure (HF) was the only comorbidity significantly more frequent among cases (17.7% vs. 7.3%, p=0.031).

Taking into account the whole sample of patients, there was a predominance of infected patients (81.7%); however, there was significantly more colonized individuals among cases as compared to controls (25.8% vs. 12.9%, P=0.028). Out of all cases, 14.5% were detected through surveillance cultures. For both groups, UTI was the most common site of infection (83.1% in the control group vs. 59.7% in cases, p<0.001). A significant difference was observed in the frequency of species of Enterobacteriaceae between the groups (p<0.001). Escherichia coli was the most frequently species isolated from the controls (79.0%) whereas in the cases prevailed Klebsiella pneumoniae (30.6%) with E. coli (30.6%). Enterobacter sp. and Serratia marcescens were isolated only from cases (24.2% and 6.0%, respectively).

The univariate analysis testing the variables associated with the presence of Enterobacteriaceae resistant to third-generation cephalosporins is summarized in Table 2. The composite presence of invasive prosthesis or tunneled catheters and presence of stoma were significantly more frequent among cases than controls. The main types of stoma observed in these patients were cystostomy (41.2%), colostomy (23.5%), and tracheostomy (17.6%). Finally, previous contact with health care services (performance of invasive procedures and/or prior hospitalization) and previous use of antibiotics were also more frequent in patients with resistant Enterobacteriaceae.

Previous use of cephalosporins in last 90 days, presence of invasive prosthesis or tunneled catheters, hospitalization in last six months, and male gender were independently associated with the presence of Enterobacteriaceae resistant to third generation cephalosporins as shown in Table 3.

We prospectively tested the impact of the presence of enterobacteria resistant to third-generation cephalosporins on some clinical endpoints (Table 4). Inappropriate empirical antibiotic therapy was significantly higher among cases (73.3% vs. 10.3%, p<0.001). Moreover, patients with resistant Enterobacteriaceae were more often admitted to the ICU (p=0.003), had worse clinical response to the antimicrobial therapy (p=0.029), and had higher length of stay (p<0.001). Although the death rate was higher among cases (21% vs. 10.5%), this difference did not quite reach statistical significance (p=0.052).

As shown in Table 5, six variables were associated with death during hospitalization in univariate analysis, but seven variables reached criteria to be included in the multivariate model (p<0.20). From these, only increasing age, hospitalization at the ICU, and a primary site of infection other than UTI proved to be independently associated to all-cause hospital mortality.

Finally, to be older than 50 years (HR: 1.63; 95% CI: 1.18–2.27; p=0.003) and presence of Enterobacteriaceae resistant to third generation cephalosporins (HR: 2.31; 95% CI: 1.59–3.34; p<0.001) were independently associated with longer length of stay.