Methods
The study was retrospectively conducted with anonymized patient data from the ED of an academic tertiary care facility. The local ethics committee of the Medical Faculty of Mannheim has approved the study. Cases were eligible for the study, if they were diagnosed with lower or upper UTI in the ED between January 2013 and June 2015. Lower UTI was defined as dysuria, pollakisuria or positive leucocyte and nitrite in urine in patients with reduced vigilance. Upper UTI was defined by additional flank pain, fever, positive systemic inflammation serum parameters or perinephritic abscess in sonography [
9].
Patient information included the following: demographic parameters (gender, age, residence), laboratory analysis (C-reactive protein, leucocyte count, serum creatinine and glomerular filtration rate (GFR) calculated after Modification of Diet in Renal Disease formula), physical condition at admission (signs of exsiccosis, symptoms at presentation), urine analysis (isolated pathogen, pathogen count, antimicrobial susceptibility testing), comorbidities (diabetes mellitus, indwelling urinary catheter, renal transplantation, dialysis), pre-ED antibiotic treatment within the last 30 days, prior hospitalization within the last 30 days and UTI within the last 12 months.
We assessed patients with clinical symptoms consistent with UTI and positive urine culture. A positive result was defined as colony-forming unit > 104/ml for catheter- or midstream-urine and > 103/ml for single-use-catheter. Urine cultures labeled by the microbiology laboratory as “contamination” or “mixed flora” were excluded.
MDR and extensively drug-resistant (XDR) pathogens were defined according to the European Centre for Disease Prevention and Control [
10]: MDR describes pathogens non-susceptible to at least one agent in three or more antimicrobial categories. XDR describes pathogens fully susceptibly to only two or less antimicrobial categories.
Data were grouped by presence of antibiotic resistances towards antimicrobial substances frequently used in the treatment of UTIs. We chose Ciprofloxacin (Cip), Ceftazidime, Cefpodoxime, Gentamicin, Piperacillin with Tazobactam (Pip/taz) and Cefuroxime for further analysis. Risk factors for carbapenem non-susceptibility were not analyzed as overall non-susceptibility was low. Differences in parameters between groups with or without resistances towards a certain antibiotic were tested for significance with Fisher’s exact testing.
Logistic regression was then performed to identify risk factors for resistances toward antimicrobial substances mainly used in upper UTI treatment, MDR pathogens and pathogens simultaneously resistant to Ciprofloxacin, Pip/taz and Ceftazidime (sCPC). The results were presented with odds ratios and their corresponding 95% confidence interval (CI). The area under the curve (AUC)/c-statistic of a Receiver Operating Characteristic (ROC) analysis was calculated and used to estimate the accuracy of fit of our model. The AUC measures the accuracy for the prediction with an AUC = 1.0 representing perfect prediction.
Susceptibility data were calculated for Ciprofloxacin, Ceftazidime, Cefpodoxime, Cefuroxime, Pip/taz, Gentamicin and Imipenem. Susceptibility was analyzed for the whole patient collective as well as subgroups defined by the number of risk factors present.
All analyses were performed using SAS system for Microsoft version 9.4. (SAS Institute Inc., Cary, North Carolina, USA).
Discussion
With the increase of microbial resistance, empiric therapy recommendations without taking local resistance data into account can lead to inferior treatment results. In our data set treating patients with upper UTI per current treatment guidelines would have led to incorrect antibiotic coverage in nearly 30% of cases [
9]. Early identification of patients at risk of antibiotic resistances and thus therapy failure is an important part of an effective empiric therapy.
In the past risk factors for extended-spectrum-betalactamase producing bacteria in non-hospitalized patients with UTI have for example been identified as recent hospitalizations within 3 months, previous antibiotic usage, age > 60 years, diabetes mellitus, male gender, previous UTI with
Klebsiella spp., residence in long-term care facilities, indwelling urinary catheters, recurrent UTIs and previous fluoroquinolone use [
11,
12]. Analogous results were obtained for fluoroquinolones [
3,
6‐
8,
13]. In a similar setting to ours, Faine et al. evaluated risk factors for MDR pathogens in UTI in the United States [
14]. Their overall MDR rate was only 6.7% in contrast to 36.5% in our population. Faine et al. used a more restrictive MDR definition, which may explains the observed difference in MDR rates and they identified male gender, chronic hemodialysis and nursing home residence as risk factors [
14].
We identified several risk factors for UTI with non-susceptible pathogens in our study. They are similar to those previous identified risk factors such as prior hospitalization within 30 days, residence in nursing homes, recurrent UTI, male gender, renal transplantation, permanently indwelling urinary catheter and prior usage of antibiotics within the last 30 days. Our analysis considered several antibiotics and demonstrates that not all risk factors are associated with non-susceptibility to all antimicrobial categories. For example, residing in a nursing home was the only significant risk factor for infection with MDR pathogens, whereas hospitalization within 30 days and indwelling urinary catheter were significant for Ciprofloxacin resistance. This seems consistent with published data that showed various risk factors associated with different antimicrobial categories in different studies [
3,
13,
15]. A large multicenter, prospective study would be necessary to further explore, which risk factors cause non-susceptibility to certain antimicrobial categories.
Stratification of empiric antibiotic therapy by risk factors can improve pathogen coverage significantly. In our study, susceptibility to standard antibiotics was overall less than 85%. However, when analyzed by presence of risk factors our data showed that patients with no risk factors had > 85% susceptibility for all standard antibiotics. Patients with one risk factor present had still a susceptibility rate of 77.5 to 82.5% for cephalosporins and 80.0% for Pip/taz. Patients with two or more risk factors remained only susceptible to Imipenem, but its susceptibility for Pip/taz, Gentamicin and Ceftazidime was reasonable with 75.0% or 76.8%, respectively.
Interestingly, in our study antibiotic use within the last 30 days was associated with a higher resistance rate in cephalosporins and Ciprofloxacin compared to the beta-lactam penicillin Pip/taz consisted with observations that cephalosporins and fluoroquinolones may increase the resistance rates of bacteria [
16]. Our data suggest that patients with previous antibiotic therapy may be better treated with Pip/taz empirically although the number of patients in this sub-analysis was low in order to draw a statistical sound conclusion.
There are several important limitations to consider in our study. It was a single-center, retrospective data analysis, which depended on the accuracy of history taking by the health care provider on call. Certain published risk factors, such as employment in health care, exposure to farming, family members with multidrug resistant pathogens, ambulant chemotherapy and wound care, had to be excluded from our study as they have not been recorded consistently in all cases. A selection bias towards antibiotic resistance cannot be excluded as not all patients presenting with UTI received a urine culture or had a positive urine culture result. Overall, the rate of antibiotic resistance or MDR pathogens was high compared to similar studies.
However, our results are among the first to show risk factors for antibiotic resistances and MDR pathogens in UTI patients in Germany. In contrast to most studies, we analyzed patients admitted to an emergency department in a large tertiary care hospital and did not solely focus on a single antibiotic substance or on MDR pathogens. Our results show that implementation of risk factors can lead to significant improvement in susceptibility in empirical therapy.
Our study revealed also another important aspect as patients diagnosed with upper UTI in the ED were admitted to the ward with an empiric antibiotic therapy, but the sampling of urine cultures was often referred to the ward nursing team due to logistical reasons. This led to a high number of negative urine cultures as the antibiotic treatment was applied before urine cultures were taken. Studies like these can help not only to identify local resistant patterns or risk factors for resistant bacteria, it also helped us to identify serious organizational problems in the daily routine.
Current treatment guidelines often recommend fluoroquinolones and cephalosporins as treatment options in uncomplicated upper UTI. Both antimicrobial categories should be viewed critical, because they lead to a significant increase of
Clostridium difficile colitis and may further increase the rate to MDR pathogens [
16,
17]. Piperacillin/Tazobactam is less prone to induce antibiotic resistance [
18]. It also inhibits
C. difficile colonization during therapy [
19]. In this regard, Pip/Taz is the better antibiotic choice. Gentamicin can cause severe side effects, such as kidney and inner ear damage and needs extended monitoring in comparison to other antibiotics especially in elderly patients. Therefore, in our institution we do not recommend Gentamicin for empiric therapy in upper UTI infections.
β-Lactamase inhibitors (BLIs) such as Tazobactam play an important role in overcoming β-lactam resistance in Gram-negative bacteria. Because of the emergence of varieties of β-lactamases, their effectiveness has diminished over time. New BLI combinations with broad-spectrum antibiotics are promising for increasing the effectiveness of empiric antibiotic therapy in UTI [
20]. Ceftolozane/tazobactam and ceftazidime/avibactam have been approved for UTI treatment and increase the overall susceptibility to gram-negative bacteria. In particular, ceftazidime/avibactam has been shown to be effective in isolates from UTI patients resistant to Pip/Taz, Cephalosporins and Carbapenems [
21,
22]. In our study, bacteria were not routinely tested for these new ß-lactamase inhibitor combinations as they are regarded as last line therapies in severely ill patients due to highly resistant bacteria.
In our setting, all standard antibiotics had a relatively low overall susceptibility. A simple risk factor based treatment algorithm using cephalosporins in patients without risk factors and Pip/taz in all other patients except for patients residing in nursing homes with recent hospitalization in the previous 30 days increases the antibiotic coverage rate to 86.1%. It can be debated whether the use of Pip/taz in all cases as a broad-spectrum beta-lactam penicillin derivative with an overall susceptibility rate of 80.3% would be much different in its clinical outcome. With its advantages regarding C. difficile colonization and induction of resistances Pip/taz could be the better choice compared to cephalosporins and fluoroquinolones for urinary tract infections.