Trends in yearly prevalence of third-generation cephalosporin and fluoroquinolone resistant Enterobacteriaceae infections and antimicrobial use in Spanish hospitals , Spain , 1999 to 2010

A Asensio (aasensio.hpth@salud.madrid.org)1, T Alvarez-Espejo2, J Fernandez-Crehuet3, A Ramos2, J Vaque-Rafart4, C Bishopberger5, M J Hernández Navarrete6, F Calbo-Torrecillas7, J Campayo8, R Canton9, on behalf of the Estudio de Prevalencia de las Infecciones Nosocomiales en España (EPINE) Working Group10 1. Department of Preventive Medicine, Hospital Universitario Puerta de Hierro, Madrid, Spain 2. Department of Internal Medicine, Hospital Universitario Puerta de Hierro, Madrid, Spain 3. Department of Preventive Medicine, Hospital Universitario Virgen de la Victoria, Universidad de Málaga, Málaga, Spain 4. Department of Preventive Medicine, Hospital Vall d ́Hebron, Barcelona, Spain 5. Department of Preventive Medicine, Hospitales El Escorial/Guadarrama, San Lorenzo del Escorial, Madrid, Spain 6. Department of Preventive Medicine, Hospital Universitario Miguel Servet, Zaragoza, Spain 7. Department of Preventive Medicine, Hospital Universitario Carlos Haya, Málaga, Spain 8. Department of Preventive Medicine, Hospital General Universitario Reina Sofía, Murcia, Spain 9. Department of Microbiology , CIBER in Epidemiology and Public Health (CIBERESP), Hospital Universitario Ramón y Cajal and Instituto Ramón y Cajal de Investigación Sanitaria (IRYCIS), Madrid, Spain 10. Estudio de prevalencia de las infecciones nosocomiales en los hospitales españoles (EPINE), http://www.vhebron.net/ac/ preventiva/epine.htm. The members of the network are listed at the end of the article.


Article published on 6 October 2011
Escherichia coli, Klebsiella pneumoniae, and Enterobacter spp.are a major cause of infections in hospitalised patients.The aim of our study was to evaluate rates and trends of resistance to third-generation cephalosporins and fluoroquinolones in infected patients, the trends in use for these antimicrobials, and to assess the potential correlation between both trends.The database of national point prevalence study series of infections and antimicrobial use among patients hospitalised in Spain over the period from 1999 to 2010 was analysed.On average 265 hospitals and 60,000 patients were surveyed per year yielding a total of 19,801 E. coli, 3,004 K. pneumoniae and 3,205 Enterobacter isolates.During the twelve years period, we observed significant increases for the use of fluoroquinolones (5.8%-10.2%,p<0.001), but not for third-generation cephalosporins (6.4%-5.9%,p=NS).Resistance to third-generation cephalosporins increased significantly for E. coli (5%-15%, p<0.01) and for K. pneumoniae infections (4%-21%, p<0.01) but not for Enterobacter spp.(24%).Resistance to fluoroquinolones increased significantly for E. coli (16%-30%, p<0.01), for K. pneumoniae (5%-22%, p<0.01), and for Enterobacter spp.(6%-15%, p<0.01).We found strong correlations between the rate of fluoroquinolone use and the resistance to fluoroquinolones, third-generation cephalosporins, or co-resistance to both, for E. coli (R=0.97,p<0.01,R=0.94, p<0.01, and R=0.96, p<0.01, respectively), and for K. pneumoniae (R=0.92,p<0.01,R=0.91, p<0.01, and R=0.92, p<0.01,

Introduction
Enterobacteriaceae are a major cause of infections in hospitalised patients [1].Among them, the most frequent are Escherichia coli, Klebsiella pneumoniae, and Enterobacter spp.[2], with E. coli being the most frequent cause of bacteraemia, and community and hospital-acquired urinary tract infections (UTI) [3].Bacteria of the genus Klebsiella frequently colonise the gastrointestinal tract, but can also be associated with opportunistic infections, including those of the urinary and respiratory tract.These organisms can spread rapidly among patients, mostly via the hands of hospital personnel, causing nosocomial outbreaks [4].Within the Klebsiella genus, K. pneumoniae is the second leading cause of gram-negative bloodstream infections after E. coli [3].In the family of the Enterobacteriaceae, Enterobacter spp.are also frequent pathogens that can cause opportunistic infections in hospitalised patients, especially in those who are immunocompromised or have mechanical ventilation support [1].
Fluoroquinolones and third-generation cephalosporins are classes of antibiotics frequently used for the treatment of infections caused by all of these organisms [1].As a result of their extensive use, a continuous decline of the therapeutic effectiveness of these antimicrobial agents has been observed.This was predicted and seems unavoidable [5].
The aim of our study was to evaluate the rates and trends of resistance to different antimicrobials, particularly fluoroquinolones and third-generation cephalosporins, in the main Enterobacteriaceae causing nosocomial and community-acquired infections in hospitalised patients in Spain over a twelve year period (1999-2010), the trends in use of these antimicrobials within this period, and to assess a potential correlation between both trends.

Methods
Since 1990, point prevalence study series of nosocomial and community-acquired infections among patients hospitalised in acute care facilities have been conducted in Spain (Estudio de prevalencia de las infecciones nosocomiales en los hospitales españoles -EPINE study).Each year in May, acute care hospitals in Spain are requested to voluntarily join the EPINE prevalence study.In some of the regions it is compulsory to conduct this study or any other prevalence audit.Participating hospitals fill a standardised questionnaire on each hospitalised patient as well as provide overall data on the hospital and the hospital's wards.Every participating hospital designates a hospital contact point as the person responsible for the survey at the hospital level and for communicating with the EPINE executive board.

Infection diagnosis relied on Centers for Disease
Control and Prevention (CDC) case-definitions for nosocomial infections [6].Infections that met the CDC criteria but developed outside of the hospital or in the first 48 hours upon admission were categorised as community-acquired infections.In addition to information on the source of the infection and the associated microorganisms, the patient forms collected from the hospitals included demographic data (age and sex), use of antimicrobial (as the number of patients receiving any antibiotic per 100 patients on the day of the survey), type of ward (general medical as opposed to a surgical, intensive care, paediatric or obstetric ward), and size of the hospital as measured by the number of beds (small: less than 200 beds, medium: 200-500 beds, large: more than 500 beds).The survey was performed by trained doctors, nurses and, in some hospitals, medical students.All the forms were revised for inconsistency and a percentage of them (10-15%) were validated by the hospital contact point after revising medical records.Hospital validated forms were sent to an independent central analysis unit.A hospital report was sent back to every participating hospital to avoid possible disagreements before final integration of the collected results in a centralised database.We focused our analysis on resistance and antimicrobial use during the period from 1999 to 2010.
Standard microbiological procedures were used by different institutions in their corresponding microbiology departments or laboratories that followed accreditation or certification procedures following local health authorities' requirements.Interpretive criteria (breakpoints) for susceptible, intermediate and resistant categories were those included in the Clinical and Laboratory Standards Institute guidelines [7,8].Resistant isolates included both intermediate and resistant isolates as reported by the microbiological laboratory to the clinician.As not all the microbiology laboratories tested the same antibiotics in susceptibility testing, resistance rates were expressed as percentage of isolates that were resistant to fluoroquinolones (ciprofloxacin or levofloxacin) or thirdgeneration cephalosporins (ceftriaxone, cefotaxime or ceftazidime).No further characterisation of resistance mechanisms was recorded.Comparisons of characteristics of infections were made by Pearson's chi-square test.To determine significant trends over time of resistance rates and proportion of use, the Cochran-Armitage test was used.For correlation of antimicrobial use and the annual prevalence rates, Spearman correlation coefficient and regression coefficient were calculated.A p value <0.05 was considered significant.All calculations were performed with Stata/SE 9.0 statistical software.

Results
Between 1999 and 2010 an average of 265 hospitals per year (increasing from 233 in 1999 to 287 in 2010) participated in the EPINE survey yielding a sample of almost 60,000 hospitalised patients per year.Of a total of 303 hospitals participating in the survey at any given year, 230 took part in the entire twelve-year series.
During the period from 1999 to 2010 a total of 19,801 E. coli, 3,004 K. pneumoniae and 3,205 Enterobacter spp.isolates -including mainly E. cloacae (72%) and E. aerogenes (18%) species -causing nosocomial or community-acquired infections were recorded.The main characteristics of the corresponding infections for the period from 1999 to 2010 are displayed in Table 1.More than half of the infections occurred in patients aged over 65 years.E. coli infections were slightly more frequent in females, in contrast with K. pneumoniae or Enterobacter spp.infections (p values<0.001).Overall, about three quarters of infections were identified in patients hospitalised in medical or surgical wards but for Enterobacter spp.infections the ratio medical/surgical ward was inversely (0.64) related to the ratio for E. coli or K. pneumoniae (1.42 and 1.56, respectively).Intensive care unit (ICU) infections accounted for 14% and 17% of K. pneumoniae and Enterobacter spp.infections respectively, in contrast to only 7% of E. coli infections.UTI was the most common localisation of infection for E. coli and for K. pneumoniae but not for Enterobacter spp.. Bloodstream infection represented 11.3%, 11.4% and 14% of the infection locations for each microorganism.

Escherichia coli infections
The characteristics of patients and hospitals where third-generation cephalosoporin resistance was found are shown in Table 2. Rates of resistance by type of ward ranged from 4.4% in gynecology and obstetrics to 13.3% in ICU patients, and by localisation of infection from 8.7% for UTI to 14.7% for respiratory infections.Information on fluoroquinolone resistance can also be found in table 2. Fluoroquinolone resistant E. coli infections were detected in patients from both sexes and all age groups seemed to be affected (although there appeared to be lower rates in age groups younger than 16 years old).Fluoroquinolone resistant E. coli could be community-acquired, but was more frequent in nosocomial infections, and in the largest hospitals.There were differences in the rates of fluoroquinolone resistance related to the type of ward and site of infection ranging from 5.4% in paediatric to 31.4% in medical wards and 32.6% in other hospital wards, and from 22.9% for surgical site wounds to 30.9% for respiratory tract infections (Table 2).

Klebsiella pneumoniae infections
Nosocomial infections as well as hospital size and type of ward, were associated with third-generation cephalosporin resistance.No difference for fluoroquinolone resistance was found by sex of the patients.High rates of fluoroquinolone resistant K. pneumoniae were found significantly associated to nosocomial infections, and the rates of fluoroquinolone resistance increased with the size of the hospital, or age of the patients, especially in patients older than one year.For the type of ward, a high rate of resistance was found in patients admitted to ICU (23.1%), and the lowest rates of resistance were found in paediatric (2.9%).For localisation of infection, UTI showed the highest rate of resistance (Table 2).

Enterobacter species infections
Nosocomial acquisition, ICU and pediatric wards, as well as bloodstream infections and UTI showed the highest prevalence of Enterobacter spp.resistance to

Trends in antimicrobial use and resistance
Antimicrobial use (number of patients receiving antibiotic/100 admitted patients) increased steadily for fluoroquinolones from 5.8% in 1999 to 10.2% in 2010 (p<0.001),but no significant trend was observed for third-generation cephalosporins use (from 6.4% in 1999 to 5.9% in 2010) (Figure 1).
Proportion of resistance to third-generation cephalosporins increased significantly for E. coli infections (from 4.6% to 15.4%, p<0.01) and for K. pneumoniae infections (from 3.9% to 20.8%, p<0.01) with a more pronounced increase since year 2004.Nevertheless thirdgeneration cephalosporins resistance for Enterobacter spp.did not show a trend, and ranged from 22.5% to 31% with a mean rate of 24.3% (Figure 2).
Proportion of resistance to fluoroquinolones increased significantly for E. coli infections (from 15.9% to 29.6%, p<0.01), and to a minor extent for Enterobacter spp.
Proportion of co-resistance to fluoroquinolones and third-generation cephalosporins increased significantly for E. coli infections (from 1.6% to 11.3%, p<0.01), for Enterobacter spp.(from 4.8% to 9.5%, p<0.01), and to a higher extent for K. pneumoniae infections (from 0.8% to 14.4%, p<0.01) and more pronounced during the period from 2004 to 2007 (Figure 4).
We found strong correlations for rates of E. coli resistant to fluoroquinolones, resistant to third-generation cephalosporins, or co-resistant to third-generation cephalosporins and fluoroquinolones, by the rate of use of fluoroquinolones (R=0.97,p<0.01;R=0.94, p<0.01; and R=0.96, p<0.01, respectively) (Figure 5A).

Percentage of patients using an antimicrobial
Year of the survey

Third-generation cephalosporins Fluoroquinolones
Use of fluoroquinolones or third-generation cephalosporins is defined as the number of patients receiving either type of antimicrobial per 100 patients on the day of the annual survey.
Also we found strong correlations for rates of K. pneumoniae resistant to fluoroquinolones, to third-generation cephalosporins, or co-resistant to third-generation cephalosporins and fluoroquinolones, by the rate of use of fluoroquinolones (R=0.92,p<0.01;R=0.91, p<0.01; and R=0.92, p<0.01, respectively) (Figure 5B), but these correlations could not be found by the use of third-generation cephalosporins.No significant correlations could be found for Enterobacter spp.resistant to third-generation cephalosporins and use of thirdgeneration cephalosporins or of fluoroquinolones.This was also the case for Enterobacter spp.resistant to fluoroquinolones and use of third-generation cephalosporins or fluoroquinolones.

Discussion
In the present study, we observed an increase in rates of resistance to third-generation cephalosporins and fluoroquinolones in the past twelve years in the main Enterobacteriaceae causing infections in hospitalised patients in Spain.This high rate of antibiotic resistance could increase the risk of inappropriate empirical therapy in hospitalised patients with potentially serious infections.In Spain, fluoroquinolones and thirdgeneration cephalosporins can be used empirically for both nosocomial and community-acquired pneumonias and UTIs.The increase in resistance to fluoroquinolones and third-generation cephalosporins nevertheless differed among Enterobacteriaceae representatives.
Enterobacter spp.(mainly E. cloacae) displayed the highest rates of resistance to third-generation cephalosporins and the rates did not increase over time, but by 2010 E. coli and K. pneumoniae isolates were 3.3 and 5.3 times more frequently resistant respectively than in 1999.The difference in resistance for Enterobacter spp.could be due to the fact that the main mechanism of resistance to third-generation cephalosporins for Enterobacter spp. is overproduction of AmpC betalactamases, whereas for E. coli and K. pneumoniae the main resistance mechanism is extended-spectrum beta-lactamases (ESBL) that has been related to antibiotic usage [9].
Overall, an increase in resistance to fluoroquinolones was observed for all three microorganisms reaching levels of resistance of 30%, 22%, and 15%, for E. coli, K pneumoniae, and Enterobacter spp.respectively.This tremendous increase in resistance was encompassed over the same time period by a 75% increase in the use fluoroquinolones in the hospitalised population.Furthermore, strong correlations were found for the use of fluoroquinolones inside the hospital and resistance to fluoroquinolones, to third-generation cephalosporins, or co-resistance to both groups of the latter antimicrobials for E. coli and K. pneumoniae.
From a microbiological point of view, the increase of  resistance to fluoroquinolones has been associated with mutations in the topoisomerases [10] and more recently by the acquisition of plasmid mediated fluoroquinolones resistance genes such as qnr, aac(6`)-Icr or qep particularly in ESBL producing isolates [11].Although the use of third-generation cephalosporins did not increase over the time period surveyed, use of this antimicrobial was maintained.The continuing use of third-generation cephalosporins could have contributed to an increase in the rate of resistance of E. coli and K. pneumoniae to third-generation cephalosporins.In the hospital environment, the increase of fluoroquinolone use may also have allowed cephalosporinresistant Enterobacteriaceae to acquire additional resistance to fluoroquinolones.A co-selection process of co-resistance might have occurred [12].
In recent years, other studies have observed Spain as one of the countries in Europe with the highest use of fluoroquinolones outside the hospital [13].This may have a role in co-selection and co-resistance, and potential clonal expansion influencing the results of this study.The observed differences in third-generation cephalosporin resistance among Enterobacteriaceae representatives could be related to the main resistance mechanism in the different studied bacteria (ESBL in E. coli and K. pneumonaie or AmpC hyperproduction in Enterobacter spp.) [14].Not surprisingly, rates of resistance to third-generation cephalosporins were higher for ICU infected patients than on average, but the highest rates of fluoroquinolone resistance were found outside the ICUs (Table 2) probably related to the widespread use of this class of antimicrobial outside the ICUs.
It is important to note that resistance to antimicrobials can differ depending of the site of the infection.For E. coli the main resistance to third-generation cephalosporins was found in respiratory tract infections (14.7%), infections that are rarely caused by E. coli outside the critical care units [15], a much higher rate than that for bloodstream infections (9.7%, p<0.001).Nevertheless, for Enterobacter spp. the highest rates of third-generation cephalosporin resistance were found for bloodstream infections (29.1%).Regarding resistance to fluoroquinolones the highest resistance in E. coli was for respiratory tract infections, while for K. pneumoniae and Enterobacter spp. was for UTI.It is important to keep in mind these differences when comparing to rates derived from a single site of infection.
It is also of note that for all these enterobacterial infections more than 50% (56% for E. coli and 41% and 36% for K. pneumoniae and Enterobacter spp., respectively) are community-acquired infections, that 88% of them are not bloodstream infections, and that looking for resistance only in nosocomial or bloodstream infections, in studies of antimicrobial resistance, can omit an important and valuable information for the development of prevention strategies.
Regarding co-resistance to both fluoroquinolones and third-generation cephalosporins, Enterobacter spp., E. coli and K. pneumoniae increased significantly their rates of co-resistance with time, but the most pronounced increase could be observed for K. pneumoniae reaching rates of co-resistance of almost 15% in the 2007-2010 period.It has been reported that many enterobacteria strains resistant to third-generation cephalosporins are also resistant to fluoroquinolone, and that this co-resistance could be mediated by plasmids harbouring both quinolone resistance and ESBL genes or presence of different plasmids harbouring these resistance genes [16].It is remarkable that the greatest increase in co-resistance for these microorganisms, and to some extent for the resistance to fluoroquinolones and to third-generation cephalosporins, started around year 2004.These facts do not seem to be related to the antimicrobial use of fluoroquinolones or third-generation cephalosporins in hospitals, neither to any other fact known by us, and remains to be elucidated.It should be interesting to study the relationship of these facts with the consumption of other antimicrobial, such as amoxicillin-clavulanic, broadly used in Spain, since new formulations, for oral and parenteral routes, with increased dosage started to be marketed in Spain at the end of 2003.
To which extent rates of resistance derived from point prevalence studies can be overestimated when these rates of resistance are derived from an augmented length of infectious state attributable to antimicrobial resistance, needs further evaluation.When comparing data from our study to those reported by the European Antimicrobial Resistance Surveillance Network (EARSnet, formerly EARSS) for the period 2007-2009 in Spain, only slight differences were found (values for EARS-net and EPINE: non-suceptible E. coli to fluoroquinolones 32.3% vs 28.7%, or to third-generation cephalosporins 9.5% vs 10.7%, non-susceptible K. pneumoniae: to fluoroquinolones 17.2% vs 20.8%, to third-generation cephalosporins 11.1% vs 16.0%) [17].Nevertheless, it should be taken into account that EARS-net records data to a net of national surveillance systems only on invasive infections, from voluntary clinical laboratories in Spain (between 29 to 33 and between 14 to 33 laboratories reporting for E. coli and K. pneumoniae, respectively), as opposed to the data from around 265 hospitals per year in EPINE, and that huge variations are found among laboratories reflecting wide local variations in rates of resistance [18].
There are some limitations for our study.As hospitals participated on a voluntary or compulsory basis, but were not selected at random, representativeness applies to regions where the study is mandatory but can not be warranted at the national level.On the other hand, this prevalence series represents more than half of the population hospitalised in acute care centres in Spain on a given day, and most data come from hospitals that have regularly participated in the survey every year so concerns about representativeness are diminished.Another limitation of our work is related to the non-experimental nature of the study so that to which extent this increase in antimicrobial resistance can be attributable to the increase in the use of fluoroquinolones cannot be concluded from the results of this ecological study but those results points out to this direction.One of the concerns about generalising our data is seasonality.As the survey was performed every year during May, seasonal variations in time could not be assessed.However, the fact that we performed the survey in the same season each year, although precluding a study of seasonality, allowed us to measure trends.Finally, point prevalence studies prove to be useful to monitor trends on rates of resistance and antimicrobial use at a national level, and can provide valuable information for comparisons among European Union Member States.Furthermore the information on antimicrobial use and resistance gathered by national prevalence surveys, linked to patient and infection characteristics, not focused exclusively on invasive infections, can complement that of much rigorous databases, from the microbiological point of view, such as EARS-net and could help to develop national strategies to prevention.

Figure 4
Figure 4Annual rates of Enterobacteriaceae co-resistant to fluoroquinolones and third-generation cephalosporins, Spain, 1999-2010

Figure 5
Figure 5Correlation between annual rate of fluoroquinolone use and resistance to fluoroquinolones, third-generation cephalosporins, or fluoroquinolones and third-generation cephalosporins, Spain, 1999-2010

Table 1
Main characteristics of patients infected by Escherichia coli, Klebsiella pneumoniae, or Enterobacter species in Spanish hospitals,Spain, 1999Spain,  -2010 (n=672,362)    (n=672,362) a Numbers in parentheses are specific for Enterobacter cloacae.b Patients hospitalised who, at the time of the survey, presented a nosocomial infection acquired during a previous admission in the same or another hospital.

Table 2
Prevalence of resistance to third-generation cephalosporins and fluoroquinolones by microorganism and by characteristics of patient, infection, and hospital, Spain, 1999-2010 a Numbers in parentheses are for Enterobacter cloacae.b Patients hospitalised who, at the time of the survey, presented a nosocomial infection acquired during a previous admission in the same or other hospital.