Antibiotic resistance of Escherichia coli isolated from uncomplicated UTI in general practice patients over a 10-year period

Most antibiotics in the Netherlands prescribed by general practitioners (GPs) are for uncomplicated acute urinary tract infections (UTI) in women [1, 2]. Escherichia coli (E. coli) is the most frequently isolated microorganism, in up to 80% of UTIs [3, 4]. Different studies report increased resistance of E. coli in UTI, especially to beta-lactam antibiotics and fluoroquinolones alone or in combination [5, 6]. Although antibiotic resistance in the Netherlands is one of the lowest worldwide, the prevalence of resistant microorganisms has been increasing over time. Several studies describe an increasing prevalence of extended spectrum β-lactamase (ESBL)-producing Gram-negative rods and other highly resistant microorganism (HRMO) in the general population [7‐9]. In addition, a high prevalence of resistant microorganisms has been reported in specific populations such as travelers returning from South East Asia [10], immigrants and refugees [11, 12], and patients with hospital-replaced home treatment [13], as well as in livestock and agriculture [14‐17]. Therefore, to make an optimal empiric choice, it is important to determine with regular time intervals the antibiotic susceptibility in populations commonly treated such as women with UTI.

For women with UTI, treatment guidelines are published by the Dutch College of General Practitioners (NHG). The choice of antimicrobial agents in the guidelines is based on national antimicrobial resistance (AMR) data published in Nethmap [8]. These data are derived from microbiological laboratories sending data to the National Institute for Public Health and the Environment. Yearly, this institute in close cooperation with the SWAB (Dutch Working group on Antibiotic policy) reports national data on antibiotic consumption and resistance in humans and animals in Nethmap and MARAN, respectively [8]; based on the NHG guideline, GPs should only submit urine for culture after failure of first-line therapy [18]. However, antimicrobial resistance of uropathogens isolated after therapy failure is not representative for resistance rates in patients who were successfully treated. Since NHG guidelines for treatment of UTIs should be based on the latter population, we determined the AMR among women with complaints of an acute uncomplicated UTI visiting their GPs. Additionally, to determine trends in susceptibility, AMR data of 2014 were compared with the data collected in a similar population in 2004 and 2009 [19, 20].

In this study, we demonstrate that AMR did not increase over the 10-year study period except the resistance to ciprofloxacin and the prevalence of ESBL-producing E. coli. These results are remarkable in this time of worldwide concern on the increase in prevalence of antibiotic resistance [5, 6]. We also showed that the prevalence of E. coli isolated from urine significantly increased in this time period and that fosfomycin was more frequently prescribed at the expense of co-amoxiclav, co-trimoxazole, and fluoroquinolones.

In contrast to the global tendency of increasing AMR [25], we showed only small changes in antibiotic susceptibility over a 10-year time period. Both in Europe and on other continents, AMR has been increasing in bacteria causing UTIs [5, 26‐28]. The Netherlands has always reported relatively low AMR rates and prudent antibiotic use [29]. Antibiotic consumption has been directly related to AMR and the low antibiotic use in the Netherlands may be an explanation for the stable AMR [30‐32]. It should be noted that the cultures were taken from community-dwelling females with uncomplicated UTIs and a presumed low antibiotic use and pressure.

The susceptibility of E. coli to co-amoxiclav increased from 88% and 87% in the 2004 and 2009 studies to 92% in the present one. In the last study, the EUCAST criteria version 4.0 valid from January 2014 was used, whereas the two other studies used the version 1.2. The main difference is the breakpoint for susceptibility for co-amoxyclav being > 16 mm for all infections according to the version 1.2 and > 15 mm for UTI only is mentioned in the 4.0 version. The version 1.2 does not mention the criterium “UTI only”; in the 4.0 version, the breakpoint for all infections is > 18 mm. Due to the different breakpoints used and depending on the type of infections, it is difficult to assess whether the increase in susceptibility is a real increase of just a virtual one. In other European countries, a decrease of susceptibility has been described [26]. In Ireland, the susceptibility of E. coli to co-amoxiclav decreased from 90% in 2005 to 80% in 2014 in community-dwelling patients with UTI [27]. Also in Germany, co-amoxiclav susceptibility decreased from 92% in 2014 to 75% in 2017 [28].

The decreased susceptibility of E. coli to ciprofloxacin was in line with other studies [26‐28]. In the EC0-SENS study, a significant increase of resistance to ciprofloxacin from 2000 to 2014 was observed in different countries: in Germany from 2.2 to 20.2%, in Spain from 14.7 to 30.8%, in Sweden from 0.0 to 7.3%, and in the UK from 0.6 to 15.3% [26]. In Ireland, resistance to ciprofloxacin increased from 5% in 2005 to 11.4% in 2014 [27]. An average increase in resistance per year of 0.40% and 0.26% was found both in community-dwelling and hospitalized patients, respectively [28]. According to the Dutch NHG guidelines, ciprofloxacin is a last resort drug and we showed that prescription rates decreased from 14% in 2004, 10% in 2009, to 5.9% in 2014 [18]. However, only a small effect on susceptibility percentages could be demonstrated; the E. coli susceptibility decreased from 96% in 2004 to 94% in 2014 (Table 3).

The rise in prevalence over time of E. coli was a remarkable finding. Although the microbiological analysis of our last survey was performed in another laboratory, the methods were comparable. The 2009 survey showed already a significant increase in prevalence of E. coli compared to 2004 [20]. The prevalence of E. coli in UTI ranged from 76.7 to 84% in different studies [27, 32, 33]. Cullen et al. also described an increase of E. coli in an 11-year period of uncomplicated UTI in community patients including fluctuation over time [33]. Different explanations were described in the literature, such as rise in global temperature, seasonal peaks, and increasing age of women with symptoms of UTI [34‐36]. In this study, we included patients during 6 months and did not observe influences of these seasons or age for the percentage of E. coli isolates.

In Belgium, Heytens et al. found in a 20-year survey from 1995 to 2015 a higher percentage of E. coli in uncomplicated UTI in post-menopausal women > 55 years of age compared to the age group of 18–54 years, 89.9% versus 78.4%, respectively [36]. Although the mean age in our study was 53.5 years and 57% was 51 years or older, the prevalence of E. coli in the age group 51 years and older was comparable to the younger age group (82% versus 84%, respectively).

Other uropathogens such as K. pneumonia and K. oxytoca were isolated less frequently. Risk factors for uropathogens other than E. coli, like recurrent UTI, pregnancy, intensive care hospitalization, presence of catheters, gender, age, and diabetes mellitus, were exclusion criteria in our study except age and gender [36, 37].

The regional differences in susceptibility of E. coli (Table 1) were in contrast to the regional antibiotic prescription of these agents (Table 2). The Northern and Eastern regions are both adjacent to Germany, where, as shown in the updated ECO-SENS study, resistance to ciprofloxacin increased from 2.2 to 20.2%, and trimethoprim from 22.5 to 36.8% from 2000 to 2014 [26]. Social and economic traffic between the regions and Germany is common [38] and geographical boundaries are not recognized by microorganisms [39]. However, no data is available of E. coli susceptibility in uncomplicated UTI in the community in the cross-border region. We found only limited regional data on antibiotic resistance in Belgium or in Germany. The only available numbers we found in Belgium were described in the relatively small study of De Backer (2008) in the region Gent with 187 patients [40]. He described in his study from 1996 to 2006 a stable prevalence of antibiotic susceptibility in urinary isolates, i.e., a susceptibility of 83.3 to 86% for co-trimoxazole, versus 86% in the South and 83% in the Western part of the Netherlands (Den Heijer 2009). Van den Donk et al. compared the prevalence of resistance in the cross-border region between Belgium, Germany, and the Netherlands among general practitioners, nursing homes, and ICU isolates and found a significant difference in prevalence of co-amoxiclav and ciprofloxacin resistance among all E. coli isolated between the Netherlands and Belgium, i.e., 27% versus 21%, and 23% and 16%, respectively. No significant differences with Germany were reported [41].

We found a regional difference in prevalence of ESBL-positive isolates, where 8 out of 10 isolates all in the age group of over 51 were from the Eastern region, close to the border of Germany. The ECO-SENS study in 2014 reported an ESBL incidence of 10.5% [26]. Van den Donk et al. (2012) showed a cross-border spread of E. coli ST131 and CTX-M type ESBLs and a similar prevalence of resistant E. coli in the Euregio Meuse Rhine [42]. A recent study found the same ESBL CTX-M15 genotype in cross-border hospitals in the Northern part of the Netherlands and in Germany whereas in community isolates from the Netherlands, CTXM1 was found. The results of this small study suggested cross-border spread in clinical isolates. As no data was available from German community strains, no conclusion could be drawn concerning possible cross-border spread in the community [43].

The prevalence of ESBL-positive E. coli isolates increased from 0.1% in 2004 to 1.0% in 2009 and 2.2% in 2014. Several reports also mentioned an increase in ESBL-producing E. coli in community-acquired urinary tract infections. In Germany, an increase was found, from 1.2 to 4.8% in 8 years’ time [26], and more than 12% in a 10-year period in Ireland [27]. Of the different risk factors described, such as recurrent UTI in the past year, (recent) use of β lactam antibiotics or fluoroquinolones, and traveling to Asia, Middle East, and Africa [10, 44], we only had information for recurrent UTI.

The increase in nitrofurantoin and fosfomycin as empirical prescription by the participating GPs was in line with the first and second choice of antibiotic treatment of uncomplicated UTI according to the NHG guidelines [18]. In addition, the prescription of fluoroquinolones, co-trimoxazole, and co-amoxiclav decreased significantly. The increase of fosfomycin prescription was in line with the alteration from third to second preference in 2013 of the national guideline. These data underscore the adherence of GPs to the national guidelines and indicate that national guidelines are valuable in antibiotic stewardship [45].

Although fosfomycin shows promising results of low resistance in UTI isolates [46] recently, a few studies describe less clinical efficacy [47, 48]. Based on the literature and the NHG guidelines, our recommendation is still nitrofurantoin as first choice, followed by fosfomycin and trimethoprim as third place if the isolated pathogen is susceptible. When more clinical studies become available concerning the (decreased) clinical efficacy of fosfomycin, the NHG guidelines might revise the recommendations.

The strength of our study is the national coverage, because the participating general practices are nationally representative for age, gender, regional distribution, and population density [21]. This enables us to compare (regional) differences in antibiotic resistance of uropathogenic E. coli over time and to evaluate whether the prescriptions of the GPs are in line with the recommendations of the NHG guideline for the treatment of uncomplicated UTI. The data we collected in our study is essential for the setup of guidelines for the treatment of uncomplicated UTI. Data in Nethmap are from patients after therapy failure and not representative for successfully treated patients and not suitable for the NHG guidelines [18].

In conclusion, the susceptibility percentages of E. coli to most antimicrobial agents isolated from uncomplicated UTI among women were stable in the Netherlands over 10 years’ time, although the prevalence of E. coli and ESBL-producing E. coli increased. These findings suggest that performing the survey at a regular interval is warranted. The GPs’ empirical prescription treatment was in line with the recommendations of the national NHG guidelines.