Correlation between antibiotic use and changes in susceptibility patterns of Pseudomonas aeruginosa in a medical-surgical intensive care unit

responsible for ICU acquired infections (unpublished data). This frequency had lead to a large use of antipseudomonal agents and concomitantly to a decline in antibiotic susceptibility of P. aeruginosa because of its ability to acquire resistance.[5-7] Indeed, many studies had reported the infl uence of previous exposure to antibiotic therapy on the susceptibility pattern of P. aeruginosa.[7-11] This impact was called “collateral damage” from antibiotic prescription to refer to ecological adverse effects of antibiotic consumption which are represented by the emergence of multi-drug resistant organisms via selection or mutation.[12] Introduction Multiple surveillance programmes have reported P. aeruginosa as one of the leading causes of nosocomial infection.[1-3] In our hospital, it represents 19% of micoorganisms causing nosocomial infections[4] and in our intensive care unit, it represents 44.7% of pathogens Research Article


Introduction
[3] In our hospital, it represents 19% of micoorganisms causing nosocomial infections [4] and in Because of the increasing frequency of isolation of P. aeruginosa and the emergence of multi-drug resistant strains in our unit, we had undertaken this epidemiological study in order to study the relationship between the use of antipseudomonal agents and the development of resistance to these drugs.

Materials and Methods
This study was conducted at the medical surgical intensive care unit of the Habib Bourguiba University Hospital (Sfax-Tunisia).Our unit is a 22-bed intensive care unit in a 510-bed tertiary-care teaching hospital that serves as fi rst line medical center for an urban population of one million inhabitants and as a referral center for a larger population coming from south Tunisia.
This study is a retrospective analysis of data collected prospectively.It was conducted over a fi ve year period (January 1 st , 1999 to December 31, 2003) which was divided into 20 quarters.

Antimicrobial usage
Antipseudomonal agents available in our hospital are imipenem, ceftazidime, amikacine, and ciprofl oxacin.Antibiotic utilization data were extracted on a quarterly basis from the inpatient pharmacy computer system and stored in a spreadsheet program (Excel ® ).Usage data was expressed as total grams of antibiotic dispensed per quarter and then converted to daily doses dispensed (DDD) by using the daily doses most frequently prescribed in our unit, which were as follows: imipenem, 2 g; ceftazidime, 3 g; amikacine, 1 g; intravenous ciprofl oxacin, 0.4 g; oral ciprofl oxacin, 1g.

Microbiology and susceptibility data
P. aeruginosa was identifi ed in the laboratory by using standard clinical microbiology methods. [13]Antimicrobial susceptibility was determined by disk diffusion methods according to the recommendations of the National Committee for Clinical Laboratory Standards (NCCLS). [14]An isolate was considered susceptible, intermediate, or resistant according to the criteria of the NCCLS.The isolates with intermediate susceptibility were classifi ed as resistant for analysis.
Susceptibility data for P. aeruginosa were obtained quarterly using a computer based documentation system.The system is adjusted to count not only primary isolates from individual patients, but also to include follow-up isolates if the primary isolates show a different pattern of antibiotic resistance.Duplicate isolates, defi ned as the same bacterial species from the same patient with the same antibiogram, were removed.

Data analysis
Categorical variables were expressed in percentage and continuous variables in means (±SD).Relationships between increasing antibiotic use and the resistance rates of P. aeruginosa were analyzed to determine the likelihood of a correlation between antibiotic utilization and the emergence of resistance.A linear curve regression was performed on relevant variables and the associations of primary interest from the correlation analysis were tabulated, showing correlation coeffi cient (r 2 ) and signifi cance (P).Statistical signifi cance was defi ned as a P-value equal or less than 0.05 for the corresponding correlation coeffi cient (r 2 ).In addition, the associations between consumption and resistance to ceftazidime, imipenem, amikacine and ciprofl oxacin were quantifi ed using non-partial and partial correlation coeffi cients according to Pearson and Spearman.

Results
Over the study period, the mean (±SD) number of patients hospitalized in our unit was 299 ± 20 hospitalizations per quarter (range: 267 and 339 hospitalizations per quarter).The mean number of hospitalization day was 1766 ± 250 hospitalization day per quarter (range, 1374 and 2358 hospitalization day per quarter) and the mean occupation rate in the unit was 88 ± 13% (range, 69 and 119%).

Discussion
Our study shows the high level of resistance of P. aeruginosa against ceftazidime, amikacine, imipenem and ciprofloxacin in our unit.In addition, it shows the high level of use of antipseudomonal agents and confirms the correlation between the evolution of resistance to imipenem or ciprofl oxacine and that of their consumption.
Different studies had reported P. aeruginosa as one of the most frequently isolated microorganisms in intensive care unit [2,3,15] and emphasized its ability to acquire resistance toward antipseudomonal agents mainly to imipenem. [8,9,16,17]ndeed, the resistance rate of P. aeruginosa to imipenem is increasing and can reach 24% in certain institutions [18] rekindling interest in polymixins as a last resort in the treatment of nosocomial infections caused by multidrug resistant P. aeruginosa. [19,20]21] Indeed, Loeffl er et al, [16] found a correlation between the resistance of P. aeruginosa to piperacillin and the consumption of piperacillin (r = 0.73; P < 0.005) or that of piperacillin-tazobactam (r = 0.61; P < 0.05), between the resistance to ceftazidime and the consumption of cephalosporins (r = 0.79; P < 0.001), between the resistance to gentamicin and the consumption of gentamicin (r = 0.64; P < 0.05) or that of aminoglycosides (r = 0.76; P < 0.005).Lepper et al, [9] found a correlation between the consumption of imipenem and the resistance of P. aeruginosa to imipenem, to ceftazidime and to piperacillin-tazobactam.This association existed between the consumption and the resistance during the same month and during the following month.Moreover, Mutnick et al, [21] reported a correlation between the use of meropenem (r = 0.98), ciprofl oxacine (r = 0.92) and ceftazidime (r = 0.83) and the resistance of P. aeruginosa toward these antibiotics.Carmeli et al, [11] in a retrospective study demonstrated that the consumption of imipenem was the independent factor related to the development of resistance of P. aeruginosa (OR = 2.8; IC 95% = 1.2-6.6;P = 0.02) toward piperacillin, imipenem or ciprofl oxacine.In a case-control study, Paramythiotou et al, [8] demonstrated that the resistance of P. aeruginosa to ceftazidime was correlated to the previous consumption of piperacillin or of ticarcillin (P = 0.01) and that the resistance to imipenem was correlated to the previous consumption of imipenem (P = 0.01).El Amari et al, [7] in a retrospective study had looked for the factors correlated with the resistance of P. aeruginosa.Using multivariate analysis, they found that the exposure to any antipseudomonal antibiotic as a monotherapy was associated with an increased risk of subsequent resistance to itself (P = 0.006; OR = 2.5; IC 95% = 1.3-4.8).Troillet et al, [17] demonstrated that a previous exposure to imipenem was statistically correlated to the resistance of P. aeruginosa to imipenem (P = 0.0004; OR: 23.2; IC 95% : 4.1-132.7).All these correlations translate the impact of antibiotic prescription on ecology.In addition, they demonstrate that the resistance of P. aeruginosa to antibiotics mainly to imipenem is associated with previous exposure to the antibiotic under question and that the exposure to an antipseudomonal agent as a monotherapy can lead to a great risk of development of resistance against this drug.
In our study, we found a statistically significant  relationship between the use of imipenem and the resistance of P. aeruginosa to imipenem in the same and in the following quarter; and a statistically signifi cant relationship between the consumption of ciprofl oxacin and the resistance of P. aeruginosa to ciprofl oxacine in the following quarter.This correlation is consistent with many other studies where resistance to imipenem or ciprofl oxacin was found to correlate with their previous use.This consideration justifi es the large effort provided by intensivists to avoid the misusage of antibiotics.Indeed, in many studies the antibiotic prescription was found to be inadequate or abusive in a large part of the cases. [22]ere are three types of epidemiological studies which can potentially link the antibiotic use with the ecological adverse effects. [12]The fi rst type is casecontrol studies, [8,11,17] the second type of study assesses accumulated data on antibiotic use and correlates them with rates of antibiotic resistance [9,10] and the third type assesses an intervention aimed at limiting the use of an antibiotic to decrease the resistance to this antibiotic. [9]ur study's design corresponds to the second type of studies.It analyzes the evolution of antibiotic use and the emergence of resistance in the unit.It provides information about the impact of the overuse of antipseudomonal agents and the benefi cial effect of their restriction on the ecology of an intensive care unit.

Conclusion
Our data support that the large use of imipenem or ciprofl oxacin in intensive care unit may lead to the emergence of imipenem-resistant or ciprofloxacinresistant strains of P. aeruginosa.Thus, they support the concept that antibiotic prescription policy of an intensive care unit has a signifi cant impact on bacterial resistance rates.

Figure 1 :Figure 2 :
Figure 1: Correlation between consumption of imipenem and resistance of P. aeruginosa to imipenem: quarterly resistance rates plotted against quarterly consumption rates during the 20 quarters of the study

Figure 3 :
Figure 3: Non-partial correlation coeffi cients between quarterly imipenem consumption and resistance in the quarter of consumption (designated "0") and the 2 quarters prior to and following consumption.Asterisks indicate statistical signifi cance (P<0.05)

Figure 4 :
Figure 4: Correlation between consumption of ceftazidime and resistance of P. aeruginosa to ceftazidime: quarterly resistance rates plotted against quarterly consumption rates during the 20 quarters of the study

Figure 5 :
Figure 5: Correlation between consumption of amikacine and resistance of P. aeruginosa to amikacin: quarterly resistance rates plotted against quarterly consumption rates during the 20 quarters of the study

Figure 6 :
Figure 6: Correlation between consumption of ciprofl oxacine and resistance of P. aeruginosa to ciprofl oxacine: quarterly resistance rates plotted against quarterly consumption rates during the 20 quarters of the study

Figure 7 :Figure 8 :
Figure 7: Correlation between resistance of P. aeruginosa to imipenem and to ciprofl oxacine: quarterly resistance rates of P. aeruginosa to imipenem plotted against quarterly resistance rates to ciprofl oxacine during the 20 quarters of the study