Catheter-associated bacterial flora in patients with benign prostatic hyperplasia: shift in antimicrobial susceptibility pattern

Catheter-associated urinary tract infection (CAUTI) is the most common of all healthcare-associated infections [1]. Despite best care and precautions, long-term catheterization is linked to an almost 100% risk of bacteriuria [2, 3], and up to 80% of catheter-associated strains are resistant to multiple drugs [4, 5]. Permanently or temporarily catheterized patients represent a significant proportion of urological ward patients, and knowledge about resistance patterns is crucial in emergencies when empirical treatment is started. Changes in CAUTI bacterial flora can also be used for monitoring drug resistance in the population because in this patient group, resistant pathogens accumulate as a result of biofilm formation, pressure from antibiotics used for prevention or treatment of CAUTI, and occasional contact with contaminated medical facilities.

Recent European Association of Urology guidelines state that empirical management of CAUTI should be carried out with broad-spectrum antibiotics based on local susceptibility patterns [6]. However, no reports are currently available on the susceptibility of catheterized patients according to the specific reason for catheterization.

Among patients treated at urological departments, the largest group consists of men with urinary retention caused by benign prostatic hyperplasia (BPH) who are admitted for surgical removal of obstruction. In this homogeneous group, we have monitored species composition and resistance patterns since 1994 and have already published on differences after one decade [7]. After another 10-year period, we decided to look again for new trends of resistance in this population. Thus, the primary goal of the current study was to examine the evolution of bacterial species distribution and antibiotic susceptibility patterns of strains isolated from urine of catheterized patients with BPH-related urine retention in a tertiary referral urology department in the last 20 years. The other equally important goal was to consider the necessary changes in the empirical therapy of genitourinary infections in this group of patients. 

A total of 169, 132, and 156 patients hospitalized in 1994–1996 (Period 1), 2004–2006 (Period 2), and 2011–2015 (Period 3), respectively, met inclusion criteria. Bacteriuria was found in 119 (70%) individuals in Period 1, 95 (72%) in Period 2, and 107 (69%) in Period 3. Escherichia coli was the most prevalent uropathogen in all periods (22, 25, 27%), followed by Proteus mirabilis (14%) in the 90-ties, and Enterococcus faecalis in the Periods 2 (14%) and 3 (19%). All significant species are given in Table 1.

Median catheterization time was shortest in Period 1 (2 months), followed by Periods 2 and 3 (3 months). For patients with bacteriuria median catheterization time was longer than for those with negative urine cultures in Periods 1 (2 vs 1.75 months; P < 0.01) and 2 (3 vs 2 months; P < 0.05) but not in Period 3 (3 months in both). Frequently more than one species was isolated from urine of catheterized patients. Two or more species were present in 25 (21%), 37 (39%), and 35 (33%) patients of those with bacteriuria in consecutive Periods. In Periods 2 and 3, catheterization time was positively correlated with the number of isolated strains (rs = 0.3 P < 0.01 and rs = 0.25 P = 0.02 respectively), but no such correlation was found for Period 1 (rs = 0.18 P = 0.06). Gram-positive only bacteriuria was steadily growing. It was found in 10/119 (8%), 11/95 (11.6%), and in 18/107 (17%) patients of those with bacteriuria respectively. Mixed bacteriuria of Gram-negative and Gram-positive organisms was found in 2 (1.7%), 21 (22%) and 21 (19.6%) patients in the consecutive Periods. While all Gram-positive bacteria were susceptible to amox/clav and to imipenem in Period 1, in the following Periods resistant strains of Enterococcus faecium and Staphylococci were already reported, and susceptibility rate dropped by at least 20%. For Gram-negative species significant changes in both directions in susceptibility to various antimicrobial agents were noted. Detailed susceptibility patterns for Gram-negative and Gram-positive species for all Periods are presented in Table 2. The right choice of empirical therapy can be made with use of data presented in Table 3. It shows in vitro susceptibility patterns of whole bacterial flora identified in samples from individual patients to antibiotics and their combinations. Looking at data in this way amikacin was the most effective choice in Period 1 when susceptibility data for some antibiotics were not available for all patients. In the following Periods efficacy of amikacin alone dropped. In years 2004–06 and 2011–15 its combination with amox/clav was effective in around 80% of patients. At the same time imipenem was the most effective single agent (90–95%). Combination of imipenem with vancomycin increased susceptibility of the flora in specimen to maximum 98%.

This report is the first to specifically characterize bacteriuria in patients with urinary retention secondary to BPH. These patients require special attention of urologists because they are common in our departments, undergo transurethral procedures, that bear high risk of infectious complications and also they may be a source of contamination with multidrug resistance strains. Finally, during several weeks with a catheter, men are prone to infective complications, such as epididymitis, prostatitis, and symptomatic lower urinary tract infections (UTIs). As current urine culture is rarely available when complications arise, empirical therapy is usually implemented and knowledge of antimicrobial susceptibility pattern becomes crucial.

Our data show several characteristic trends in the composition and antimicrobial susceptibility of bacterial flora that occurred during the studied period. First, the percentage of Gram-positive bacteria increased significantly. Whereas in the 90’s it could be assumed that Gram-negative bacteria were responsible for vast majority of CAUTI (90% of patients had Gram-negative species only), later the percentage of patients contaminated with Gram-positive only or with mixed flora involving Gram-positive germs increased significantly. Gram-positive strains in Period 1 were few and all susceptible to amox/clav and imipenem. After ten years appearance of Enterococcus faecium and resistant Staphylococci caused that those agents had become less effective. At the same time susceptibility Gram-negative flora to fluoroquinolones decreased significantly while susceptibility to cotrimoxazole and gentamicin increased at the same time. The trend for fluoroquinolones has already been reported extensively and attributed to the enormous use of these antibiotics in medicine and in livestock breeding [9]. The improvement in the effectiveness of cotrimoxazole and gentamicin is probably related to lower popularity of both drugs but this issue was investigated to a lesser extent [10, 11]. Whether these trends are universal or not is hard to conclude as results reported by investigators vary.

Studies on antimicrobial resistance of bacteria causing catheter-associated infections are scarce, and none are focused on this specific group of patients. An analysis by Jiménez-Alcaide et al. included all patients with any type of urinary catheter or stent and symptomatic UTI hospitalized in 2012–2013 in a single urological ward [12]. About a quarter of their group consisted of patients with indwelling catheters on admission who might be considered similar to our patients. Another group analyzed samples of urine collected from unspecific catheterized patients in India with UTI symptoms [4]. Valuable data on CAUTI among patients in US acute care hospitals and inpatient rehabilitation facilities also come from reports from the US Centers for Disease Control and Prevention (CDC) National Healthcare Safety Network. A recent update covered the years 2011–2014 [1] and reported on over 138,000 CAUTI events. Data on bacteriuria in catheterized patients in nursing homes from a single Swedish county were presented in a cross-sectional study from 2011 [13], but in contrast to other studies, this group consisted mainly of asymptomatic women and men – only 7% used antibiotics when the sample was collected. However, age, sex, and time of catheterization (70% catheterized longer than 12 months) make this study significantly different from ours. Some CAUTI studies have focused on a specific bacteria species (e.g., Proteus spp. [14]) and others on specific resistance mechanisms (e.g., extended-spectrum beta-lactamase (ESBL) [15, 16]). In those reports, the groups also were not disease-specific, and time of catheterization was not reported.

In contrast, our study reports on a very well-defined group of men with one urological condition, catheterized for a period of several weeks to months. It also offers a 20-year perspective on changes in microbial composition and antimicrobial resistance in a single center, which is unique among papers on CAUTI. Due to the lack of more similar studies, it was decided to compare our results with the aforementioned analyses. A high rate of Gram-positive bacteria in catheter-associated flora was confirmed in some of the mentioned studies (16% [1], 21% [8], 31% [12]) but not all (2% [4]), which may reflect geographical variability. Gram-positive bacteria reported in our series are highly susceptible to amox/clav. However, in the male genitourinary tract, this antibiotic, as well as ampicillin, is not always effective because of low penetration to prostatic tissue [17, 18]. Instead high penetrability has been confirmed for carbapenems and vancomycin [19, 20]. Moreover, E. faecalis resistance to amox/clav was already reported to be as high as 20% [12]. With an increasing number of vancomycin-resistant enterococci (VRE) - colonized patients in our department (data on file) and reported by other groups (8% [1]), we predict that BPH patients with catheters colonized by these germs will soon appear.

The choice of antibiotic for treatment of Gram-negative species is a much more difficult task. Rising resistance to fluoroquinolones means that this group, which penetrates very well to prostatic tissue, should not be used any more for empirical treatment of complicated UTIs. A significant decrease in resistance to cotrimoxazole still does not mean that this drug can replace fluoroquinolones. Among agents available in oral formulations, none can be used without urine culture. Despite methodological differences, a high resistance to fluoroquinolones and cotrimoxazole is convincingly confirmed by other studies (17% [13], 33% [1], 50% [12]). The ESBL prevalence of 20% in our group presents the same level as US and Spanish hospitalized patients with CAUTI [1, 12] and among European participants in the SMART study [21]. Toner et al. found that an indwelling catheter almost doubles the risk for ESBL-producing strains in urine [22]. Susceptibility to amikacin among Gram-negative species is still high, but significant prevalence of Enterococci makes this drug poor choice if used alone. Carbapenems is the last resort, however, as in the case of VRE, we can predict growth in the number of carbapenemase-producing strains in the near future. At the Spanish urological department, resistance against amikacin was less frequent than against carbapenems (3% vs 10% for E. coli) [12]. The US CDC has reported similar numbers [1]. An older global study on nosocomial UTI showed highest susceptibility to imipenem [23]. A low prevalence of two species, Acinetobacter baumanii and Klebsiella pneumoniae, which are the most common producers of carbapenemase, may explain the difference between our results and those in other reports cited here. Apart from geographical variations, exposure to hospital strains also may be a reason. Although our patients visited only outpatient clinics before they entered the ward for surgery, patients from other studies usually were recruited among already hospitalized populations, some of them after urological procedures. In Swedish nursing homes, carbapenemase-bearing strains were also not reported [13]. Considering antimicrobial resistance, our results can be placed between those obtained from hospitalized patients [1, 12] and patients in Swedish decentralized geriatric care facilities where strict adherence to the Swedish Strategic Programme against Antibiotic Resistance recommendations is a rule.

There are some limitations of this study. The main one is the fact that the data come from a single center. Differences in species composition and susceptibility between different continents, countries and centers result from local patterns of antibiotics use, hygienic conditions, migration, or the time of the average catheterization time. Similar analyses should be performed locally in other centers and promotion of this issue was one of our additional goal. Another weakness is the heterogeneous composition of antibiograms. Therefore, only a few antibiotics were repeatedly evaluated in the three Periods studied. Also, we did not have the original minimum inhibitory concentration data and the standard of susceptibility is different prior to and after 2010.

Although our findings do not refer strictly to the situation in other countries and other clinical conditions, we think that taking into account the trends identified in other studies, the significance of these results remains broadly relevant.

Between 1994 and 2015 substantial rise in the incidence of Gram-positive species in catheter-associated bacterial flora was discovered. This fact, together with fluctuations in antimicrobial susceptibility patterns, make necessary a switch in empirical therapy of CAUTI from Gram-negative targeted agents to therapy effective also to Gram-positive species [24].The choice of drugs in the empirical treatment of genitourinary infection in a male with a catheter was and still is limited to amikacin or carbapenems for Enterobacteriaceae and amox/clav, carbapenems or vancomycin for Enterococci. The final choice depends on the localization and severity of infection as well as renal function. Monitoring of catheter-associated bacteria resistance patterns reflects trends in the outpatient population, and these surveillance efforts should be continued independent from monitoring hospital-acquired infections.