Introduction

In this era of aged populations and increasing numbers of immunocompromised patients the incidence of healthcare-associated infection (HAI) has risen dramatically and is most often caused by multidrug-resistant bacteria [1, 2]. For Gram-positive bacteria in the US, Staphylococcus aureus is the most common pathogen causing HAI, followed by enterococci [3]. Among these pathogens, methicillin (oxacillin)-resistant S. aureus (MRSA) and vancomycin-resistant enterococci (VRE) are of greatest concern because the resistant pathogens are associated with extended length of hospital stay, higher cost and greater mortality [1, 4].

Although antibiotic use is the most important risk factor for the emergence of bacterial resistance, the nature of this relationship is complicated. Previous studies which examined the relationship between antimicrobial consumption and antibiotic resistance have had inconsistent findings, which may be due to differences in the resistance profiles and antibiotic prescribing practices in different countries [5–10]. The prevalence of MRSA and VRE was shown to be associated with the use of glycopetides, extended-spectrum cephalosporins, and fluoroquinolones [11–18]. This study investigated the correlation between antibiotic consumption and antimicrobial resistance of MRSA and VRE causing healthcare-associated infection from 2000 to 2009 at a medical center in Taiwan.

Materials and methods

Setting and patients

National Taiwan University Hospital (NTUH) is a 2,500-bed, academically affiliated medical center providing both primary and tertiary care in northern Taiwan. The number of annual patient-days at the hospital increased from 624,675 in 2000 to 751,026 in 2009. Linezolid and fusidic acid were introduced in the hospital formulary in 2002. Some of the data analyzed in this study were included in a previous study [5, 6]. Prior to 2004, no specific and well-established antibiotic control polices were implemented at the hospital. All attending physicians at the hospital can prescribe nearly all antimicrobial agents (except for liposomal amphotericin B, voriconazole, caspofungin, linezolid, and ganciclovir) without the permission of the infection disease specialists.

Antimicrobial agents and consumption

Data on annual consumption (defined daily dose [DDD] per 1,000 patient-days) of glycopetides (vancomycin and teicoplanin), linezolid, fusidic acid, extended-spectrum cephalosporins (cefotaxime, ceftriaxone, flumoxef, ceftazidime, cefempime and cefpirome), β-lactam-β-lactamase inhibitor combinations (ticarcillin-clavulanic acid and piperacillin-tazobactam), carbapenems (imipenem, meropenem and ertapenem), aminoglycosides (amikacin and gentamicin), and fluoroquinolones (ciprofloxacin and levofloxacin) from 2000 to 2009 were obtained from the pharmacy department of the hospital.

Bacterial isolates

To determine the temporal trend of MRSA and VRE causing health-associated infections at NTUH, data on the disk diffusion susceptibilities of S. aureus to oxacillin and of enterococci to vancomycin among isolates recovered from 2000 to 2009 were retrieved from annual summary documents as previously reported [1, 2]. Definitions for health-associated infections followed the National Nosocomial Infection Surveillance guidelines [19]. These isolates were non-duplicate samples, as several isolates of each species from each patient recovered within 7 days were considered as a single isolate. Susceptibility testing for S. aureus and enterococci followed Clinical and Laboratory Standards Institute (CLSI) guidelines [20]. S. aureus ATCC 25923 and E. faecalis ATCC 29212 were used as control strains for routine disk susceptibility testing [20]. Methicillin resistance in S. aureus was routinely screened for by growth of the isolate on a trypticase soy agar plate containing oxacillin 6 mg/L plus 2% NaCl and incubation in ambient air at 35°C for 24 h [20, 21]. Vancomycin resistance in enterococci was further confirmed by growth of the isolate on a brain heart infusion (BHI) agar with vancomycin 6 mg/L and incubation in ambient air at 35°C for 24 h [20]. S. aureus ATCC 29213, S. aureus ATCC 43300, E. faecalis ATCC 29212 and E. faecalis ATCC 51299 were used as control strains.

Statistical analysis

Linear regression analysis was used to analyze the trends of annual consumption of antimicrobial agents and the trends in the nonsusceptible (including intermediate and resistant) rate of S. aureus and enterococci causing all healthcare-associated infections with time. Because of the emergence of severe acute respiratory syndrome (SARS) in 2003, the trends of annual consumption of antimicrobial agents were analyzed with three time periods: 2000–2003, 2004–2009, and 2000–2009, with linear regression analysis. Pearson’s correlation coefficient was used to determine the relationship between annual antibiotic consumption and trends in resistance. In order to study the lag-time effect, the relationship of the annual consumption of antimicrobial agents and the next year antimicrobial resistance was further explored in pairs using linear regression and Pearson’s correlation analysis. A p value <0.05 was considered statistically significant.

Results

Annual antibiotic consumption

In general, the use of each individual antimicrobial agent varied with time period (Table 1 and Fig. 1). During 2000 to 2003, a significant increase in consumption of glycopetides, extended-spectrum cephalosporins, and fluoroquinolones was found while a decrease in consumption was found for aminoglycosides. From 2004 to 2009, only the prescription of carbapenems and aminoglycosides significantly increased and decreased, respectively. Overall, the consumption of extended-spectrum cephalosporins, β-lactam-β-lactamase inhibitor combinations, carbapenems, and fluoroquinolones significantly increased while the consumption of aminoglycosides decreased during the study period. Vancomycin consumption significantly increased during 2003 (r = 0.96, p = 0.03, b = 2.28) and decreased after 2004 (r = 0.84, p = 0.036, b = −1.00). In contrast, the teicoplanin use remained stable during the study period (r = 0.47, p = 0.18, b = 0.28). In addition, the trend in consumption of linezolid insignificantly increased from 2.2 DDD per 1,000 patient-days in 2002 to 66.5 DDD per 1,000 in 2009, while the trend for fusidic acid use remained stable from 2002 to 2009.

Table 1 Annual consumption of several antimicrobial agents at National Taiwan University Hospital, 2000–2009
Full size table
Fig. 1
figure 1

Portions of annual consumption (defined daily dose (DDD) per 1,000 patient-days) of vancomycin, teicoplanin, linezolid and fusidic acid

Full size image

Trend of antimicrobial resistance among S. aureus and enterococci isolates

A total of 4,352 non-duplicate S. aureus and 3,673 enterococci isolates were identified and tested for susceptibility. Table 2 shows the trends in the resistance rate of S. aureus and enterococci isolates. A significant decrease in the resistant rate with time was found for oxacillin- and gentamicin-resistant S. aureus. In contrast, a significant increase in resistance rates was found for vancomycin- and teicoplanin-resistant enterococci.

Table 2 Prevalence trends of Gram-positive bacteria causing healthcare-associated infections not susceptible to the indicated antimicrobial agent at National Taiwan University Hospital, 2000–2009
Full size table

Correlation between antibiotic consumption and rate of resistance

The correlation between the rates of resistant MRSA and VRE causing healthcare-associated infections and the annual consumption of the specific antibiotics are shown in Tables 3 and 4. A significant correlation was found between the increased use of extended-spectrum cephalosporins, β-lactam-β-lactamase inhibitor combinations, carbapenems and the decreased prevalence of MRSA. In contrast, a significant positive correlation was found between the increased use of teicoplanin, extended-spectrum cephalosporins, β-lactam-β-lactamase inhibitor combinations, carbapenems and the increased prevalence of VRE. Decreased prescription of aminoglycosides was correlated with the decreased prevalence of MRSA and the increased prevalence of VRE.

Table 3 Relationship between rates of methicillin-resistant Staphylococcus aureus (MRSA) causing healthcare-associated infection and annual consumption of several antibiotics at National Taiwan University Hospital, 2000–2009
Full size table
Table 4 Relationship between rates of vancomycin-resistant enterococci (VRE) causing healthcare-associated infection and annual consumption of several antibiotics at National Taiwan University Hospital, 2000–2009
Full size table

Discussion

This study evaluated the association between antibiotic consumption and the antimicrobial resistance of MRSA and VRE causing healthcare-associated infection in a medical center in Taiwan during a 10-year period. The first noteworthy finding of this study was that the trend in the consumption of extended-spectrum cephalosporins, β-lactam-β-lactamase inhibitor combinations, carbapenems, and fluoroquinolones significantly increased during the study period; however, the trend for consumption of glycopeptides, with significant increases only during 2000 to 2003, was followed by stable consumption during the next six years. As has been previously documented, the emergence of SARS led to acknowledgement of the need for more stringent infection control measures by physicians and in healthcare-associated infection policy. The revised policies reflected the need for more cautious intervention but not misuse of antibiotics. The slow-down in the increasing use of antibiotics documented in this work thus appears to have resulted from the implementation of infection control intervention measures after the emergence of SARS. In addition, the consumption of aminoglycosides significantly decreased during this period, which is probably attributable to changes in prescription practices due to the notable side effects of these agents.

A second noteworthy finding of this study is that the prevalence of MRSA significantly increased from 71.8% in 2000 to 78.2% in 2002 (peak) and finally decreased to 54.9% in 2009. A similar trend was noted for gentamicin-resistant S. aureus. As stated above, stricter infection control policy after the emergence of SARS and another simple measure—hand washing—have been intensively carried out since 2003. Therefore, the declining prevalence of MRSA could be partly explained by the aggressive policy of infection control and the resulting decreased risk of hand transmission. However, the prevalence VRE and teicoplanin-resistant enterococci significantly increased up to more than 20% during the study period. These differences most likely resulted from the character of the bacteria itself as enterococci colonized the gastrointestinal tract of hospitalized patients. Therefore, more effective infection control interventions continue to be needed to prevent the spread of VRE.

In addition to infection control, we were also interested in the relationship between antibiotic use and the prevalence of MRSA related healthcare-associated infection. In contrast to previous studies [5, 11, 14, 22] we found a negative correlation between the use of extended-spectrum cephalosporins, β-lactam-β-lactamase inhibitor combinations, and carbapenems and the prevalence of MRSA. We demonstrated that the consumption of extended-spectrum cephalosporins and carbapenems reduced the rate of MRSA with a 1-year lag. On the contrary, the prescription of aminoglycosides was positively correlated with the prevalence of MRSA with and without a time lag. Most importantly, all of the MRSA were susceptible to vancomycin and most MRSA remained susceptible to teicoplanin during the 10-year period.

Besides MRSA, the epidemiology of VRE could be affected by the use of antibiotics including third-generation cephalosporins and vancomycin [16]. In this work, we further confirmed that the consumption of teicoplanin, β-lactam-β-lactamase inhibitor combinations, and carbapenems were positively correlated with an increased rate of VRE. In addition, aminoglycosides exerted a protective effect against this resistance, which lasted as long as one year.

This study had one major limitation. Although we found a relationship between antibiotic use and the resistance rate among MRSA and VRE, these results must be interpreted with care. The cause of emergence of drug resistant bacteria in the hospital is complex and selective pressure from widespread use of antimicrobials might only partly explain investigational results. In fact, the prevalence of MRSA and VRE are susceptible to several factors such as infection control measures and hand hygiene, or operational changes in the hospital (such as shorter length of stay because of the shift of procedures or rehabilitation to an outpatient setting, etc.), or even the epidemiology of patients (i.e. shifts towards an ageing population with more chronic diseases over a decade-long period). Their effect was not measured in this study, hence their impact on MRSA and VRE trends remains unknown. Besides, because this was an ecological surveillance, we could not analyze the impact of individual level, duration of exposure to antibiotics or possibility of clonal spreads of resistant bacteria to determine their further correlation with the identified associations.

In conclusion, this 10-year study in an institution provided a profile of the different correlations between the prescription of antibiotics and the resistant rate of MRSA and VRE. Strict implementation of infection control policy including appropriate antimicrobial prescription and simple infection control measures including hand washing are positively correlated with a decreased presence of MRSA in the hospital.