Antibiotic Resistance Among Gram-Negative Bacilli in US Intensive Care Units:  Implications for Fluoroquinolone Use FREE

Susceptibility data derived from national surveillance can be a barometer for emerging resistance problems.^1- 2 Previously, we published the results of a national intensive care unit (ICU) surveillance study of aerobic gram-negative bacilli collected between 1990-1993.³ The study revealed a rising incidence of ceftazidime-resistant Klebsiella pneumoniae and Enterobacter species in ICUs. The purpose of the current study was to describe national rates of antimicrobial resistance in ICUs between 1994-2000.

The surveillance program has been described previously.³ In brief, each institution agreed to test 100 consecutive gram-negative aerobic isolates recovered from ICU patients. Organisms were identified to the species level. Susceptibility tests were performed with a standardized microtiter minimal inhibitory concentration (MIC) panel (Microscan MKD MIC, Dade International MicroScan, Sacramento, Calif). Participating laboratories used National Committee for Clinical Laboratory Standards–recommended validation of MICs with American Type Culture Collection test strains.³ Piperacillin/tazobactam and cefepime were added to the panel in 1996 and 1998, respectively. Hospitals were categorized based on teaching status and bed size. National fluoroquinolone consumption data were obtained from IMS HEALTH Retail and Provider Perspective (Plymouth Meeting, Pa) in conjunction with its MIDAS database. The data presented are from nonduplicate isolates that we evaluated independently of the research sponsor.

Study variables were subjected to univariate descriptive analysis. Non-normal data were rank transformed before application of parametric testing when appropriate. Pearson and Spearman rank order intercorrelational analyses of year of observation were used to compare resistance rates and fluoroquinolone use. All statistical analyses were performed using SAS release 8.0 for IBM PC Windows and SAS Version 6.14 for mainframe computers (SAS Institute Inc, Cary, NC).

A total of 35 790 isolates were collected during 1994-2000; 77 to 117 ICUs participated per year, representing 43 states plus the District of Columbia. Approximately half of the hospitals repeatedly took part in this program. The majority of institutions were teaching hospitals (85%) and were intermediate to large sized (200-500 beds [59%]; >500 beds [38%]).

Pseudomonas aeruginosa was the most frequently isolated organism (23%) followed by Enterobacter species (14.0%), K pneumoniae (13.6%), and Escherichia coli (11.3%). The remaining 38.1% of isolates included Acinetobacter species (5.8%), Serratia marcescens (5.4%), Stenotrophomonas maltophilia (4.3%), Proteus mirabilis (3.6%), Citrobacter species (2.9%), and Morganella morganii (0.9%).

Most isolates were cultured from the respiratory tract (51.5%), urine (16.0%), blood (13.8%), or wounds (11.8%). Pseudomonas aeruginosa was the most frequent isolate from the respiratory tract (31.6%) and wounds (24.9%); K pneumoniae was the most common blood isolate (20.8%); and E coli was the most frequent urine isolate (35.5%).

Antimicrobial agents could be grouped into 3 broad categories based on overall in vitro activity (Table 1). While the activity of most agents decreased 6% or less over the study period, the overall susceptibility to ciprofloxacin decreased steadily from 86% in 1994 to 76% in 2000. Ciprofloxacin maintained excellent in vitro activity against E coli, but showed reduced activity against other organisms, especially P aeruginosa. Ciprofloxacin resistance did not differ significantly in teaching vs nonteaching hospitals or in hospitals with more than 500 beds vs hospitals with 500 beds or less. The decline in ciprofloxacin susceptibility was associated significantly with increasing national use of fluoroquinolones during the study period (Figure 1). Resistance to ciprofloxacin was associated with cross-resistance to other broad-spectrum antimicrobial agents (Table 2).

We examined susceptibility data for gram-negative bacilli isolated from ICU patients in 43 states and the District of Columbia. As in 1990-1993,³ amikacin (90%) and imipenem (89%) were the most active agents. Our findings (Table 1) suggest that there is an overall chance of 10% or greater that an infecting gram-negative bacillus in an ICU patient will be resistant to any single agent, which may help to explain the trend to increased use of multidrug regimens for initial empirical therapy of suspected nosocomial infection.^1- 6

Our 1990-1993 data identified the rising incidence of ceftazidime-resistant K pneumoniae and Enterobacter species, suggesting the widespread presence of plasmid-mediated extended-spectrum β-lactamases (ESBLs) and of hyperproducers of type 1 chromosomal β-lactamases, respectively. From 1990-1993 to 1994-2000, there was a further decline in the ceftazidime susceptibility for K pneumoniae (93% vs 87%) and Enterobacter species (67% vs 63%), similar to the findings of the most recently published Centers for Disease Control and Prevention National Nosocomial Infections Surveillance data from 1994-1998.⁶

The most alarming trend detected in the current study was the decreasing activity of ciprofloxacin. The overall susceptibility to ciprofloxacin among aerobic gram-negative bacilli declined from 89% in 1990-1993³ to 86% in 1994 to 76% in 2000. The most notable reductions in ciprofloxacin susceptibility were seen with P aeruginosa (89% in 1990-1993³ to 68% in 2000). The declines in activity of ciprofloxacin correlate with a greater than 2.5-fold increase in use of quinolones (ciprofloxacin, levofloxacin, ofloxacin)—popular agents for treating community-acquired pneumonia, urinary tract infections, and skin and soft tissue infections—over the past 10 years (Figure 1).

Cross-resistance has been observed with the newer fluoroquinolones against ciprofloxacin-resistant gram-negative bacteria.⁷ While there have been suggestions that fluoroquinolone resistance also is related phenotypically to the presence of ESBLs,^8- 9 implying that fluoroquinolone resistance could be driven by cephalosporin use, fluoroquinolone resistance has not been linked genetically to resistance to other classes of drugs. However, plasmid-mediated fluoroquinolone resistance has been described recently,¹⁰ and fluoroquinolone use may select for bacteria with heightened antibiotic efflux capability.¹¹ Thus, ciprofloxacin resistance may be associated with limited treatment options for other classes of agents, as observed in our study (Table 2) and other studies.^8,12

As with any large national surveillance study, these findings have limitations. Molecular typing of organisms was not performed; therefore, we cannot exclude the possibility of epidemics or clonal spread of bacteria. However, the large number of study sites makes it unlikely that epidemics in an individual ICU influenced our results. In addition, our analysis did not include or adjust for potentially important confounders such as case mix, prior antibiotic exposure, mechanical ventilation, or ICU length of stay. Although actual antibiotic consumption data for the ICUs under study would allow for a more targeted correlation of drug exposure and resistance, ecologic population data such as ours can provide important support for analyses of resistance trends.¹³ Population-based antibiotic use data may be especially important for the fluoroquinolones since the extensive use of these agents in the community setting may affect hospital resistance rates. In fact, fluoroquinolones are the only class of antibiotics for which resistance has been similar in the ICU and non-ICU setting.²

In conclusion, our findings add to prior surveillance efforts by (1) providing a decade-long perspective, (2) presenting results from 35 790 isolates from ICUs nationwide, (3) documenting the rising incidence of antibiotic-resistant gram-negative bacilli, and (4) comparing these results with fluoroquinolone use trends. This work expands on other research, including a recent international study in 5 European countries documenting a high incidence of reduced antibiotic susceptibility among gram-negative bacteria.¹⁴ We have focused on ciprofloxacin resistance because of the increasingly frequent use of fluoroquinolones for treatment of urinary tract infections and pneumonia in the community and hospital setting. Fluoroquinolones that are not affected by currently circulating resistance mechanisms need to be developed to conserve this class of agents.¹⁵ In the meantime, ongoing surveillance and more judicious use of fluoroquinolone antibiotics¹⁶ will be necessary to limit this downward trend in susceptibility.