This hospital-based analytic cross-sectional study was performed from January 2014 to December 2015. The subjects were hospitalised patients at ICU at Roi Et Hospital, Roi Et Province, Thailand during the study period. The outcome was retrospectively reviewed from medical records and laboratory tests that were performed at the Department of Clinical Microbiology Laboratory, Roi Et Hospital, Roi Et Province, Thailand.

This study was approved by the Ethical Committee (Ref. No. 003/2559, date of issued on November 3^rd, 2017). The power of the study was calculated to be 89.10% [17-19]. The proportion of case and control was 1:2. In total, 141 cases subjects were included in this study, 47 cases and 94 controls. The inclusion criteria for cases were: 1) Patients diagnosed with HAIs by XDR-PA bacteremia infection [21]; 2) Duration of admission more than 48 hours; 3) Aged more than 15 years; 4) Confirmed the diagnosis by physicians with complete laboratory tests. The inclusion criteria for controls were: 1) Hospitalisation more than 48 hours; 2) Aged more than 15 years; 3) No XDR-PA infection. The exclusion criteria were: Patients with incomplete history of treatment and the laboratory tests were excluded.

XDR-PA bacteremia was defined as resistant to all, or almost all of the antimicrobial agents-in particular, susceptible to Colistin [7-10]. The diagnosis of XDR-PA among cases was retrospectively reviewed from the medical records and laboratory-testing database. The methods of XDR-PA laboratory testing were performed using the standards for antimicrobial susceptibility test [21,22] and is described previously by the author in a published study [20].

Demographic variables (gender, age, complications and duration of hospitalised admission) were included in this study. The independent variables were various groups of antimicrobial medication which was treated during their treatment. The first generation of Cephalosporins consisted of Cephalotin and Cephazolin. Second generation Cephalosporins is Cefuroxime, Cefamandole, and Cefoxitin. The third generation of Cephalosporins is Ceftazidime, Cefotaxime, Cefoperazone, and Ceftriaxone. In addition, the Penicillin group consisted of Ampicillin, Amoxicillin, Methicillin, Oxacillin, and Cloxacillin. The Cabapenams group was Imipenem, Meropenem, Ertapenem, and Doripenem. The Aminoglycosides group consisted of Streptomycin, Amikacin, Gentamycin, Kanamycin, Tobramycin, and Spectinomycin. The Quinolones group was Norfloxacin, Ofloxacin, Ciprofloxacin, and Levofloxacin. Finally, the Colistin group included Sulfamethoxazole/Trimethoprim and Tetracyclines (Tetracycline, Doxycycline). Variables of using antimicrobial medication were divided into dichotomous variables such as ‘Yes’ if using antimicrobial medication and ‘No’ if not using antimicrobial medication.

Descriptive statistics was used to describe demographic characteristics. Test of homogeneity between cases and controls was performed by using Pearson’s Chi-square test. The outcome of this study was the occurrence of XDR-PA infection among subjects. Bivariate and multivariate analysis was constructed by using logistic regression analysis to investigate an association between using antimicrobial medication and occurrence of XDR-PA infection. Initially, bivariate analysis was performed by using simple logistic regression for each interested variables and outcome. The factors with the p-value ≤0.25 by the Wald’s test were considered and entered into the initial model of multivariate analysis. Using a backward elimination method, the factors with the p-value of Wald’s test >0.05 were eliminated, respectively. The p-value of the partial likelihood ratio was tested by fitting model. The magnitude of using antimicrobial medication and the treatment of XDR-PA infection by presenting Odds Ratios (OR) with 95 percent confidence intervals (95% CI). The interpretation shows no association if the OR included 1, the more of using antimicrobial medication as if the OR is greater than 1 and the lower of using antimicrobial medication as if the OR is less than 1.

The demographic characteristics among cases and controls revealed that more than a half of them were males (55.32% and 59.57%, p-value=0.056). The average age of cases and controls was 66.51 (SD=14.76) years and 56.52 (SD=18.35) year, (p-value=0.474). However, most of the cases had complications (82.98% and 65.96%, p-value=0.035) and had duration of admission for more than seven days (82.98% and 47.87%, p-value <0.001) [Table/Fig-1]. The principle diagnosis among cases was respiratory tract infection (38.30%) and encephalopathy (19.15%) and among controls were respiratory tract infection (38.30%), encephalopathy (7.45%) and gastrointestinal infection (7.45%) [Table/Fig-2].

Bivariate analysis was constructed by using simple logistic regression. It revealed that there were only two antimicrobial agents that were statistically and significantly associated with the treatment of XDR-PA infection among subjects. Carbapenem was 3.12 times more likely to be administrated for treatment XDR-PA infection (OR=3.12; 95% CI:1.50-6.59). In addition, Ciprofloxacin was 3.28 times more likely to be administrated for treatment of XDR-PA infection (OR=3.28; 95% CI: 1.22–8.84) [Table/Fig-3].

Multivariate logistic regression analysis showed an association of antimicrobial agent administration for treatment of XDR-PA infection after adjustment for sex, duration of hospitalisation, and having complications. It revealed that third generation Cephalosporin was 1.99 times more likely to be administrated for treatment of XDR-PA infection (OR=1.99; 95% CI: 1.22 to 4.13). Carbapenem was 4.66 times more likely to be administrated for treatment of XDR-PA infection (OR=4.66; 95% CI: 2.04 to 10.64). As well, Ciprofloxacin was 3.40 times more likely to be administrated to treat XDR-PA infections (OR=3.40; 95% CI: 1.24 to 9.49) [Table/Fig-3].

An association between antimicrobial agent administration and treatment of XDR-PA infection was found in this study and revealed that third generation Cephalosporin, Carbapenem, and Ciprofloxacin was more likely to be used for treatment of XDR-PA phenotype infections.

The present study was similar to the study by Pinto Pereira LM et al., (Spain, Trinidad) and showed that the prevalence of using third generation Cephalosporins was 9.5 per 1,000 admissions. The most common infections among patients were skin and soft tissue infection (20.3%), respiratory (10.9%), gastrointestinal (10.7%) and urinary (9.3%) tract infections. [23]. The study by Willmann M et al., in Germany found that haematological-oncological patients with XDR-PA bacteremia infection were administrated with Ciprofloxacin (OR=5.53, 95% CI: 1.11-27.53) [4]. Additionally, the study by Gomez-Zorrilla S et al., from Spain revealed that patients with underlying diseases infected with P. aeruginosa were treated with Carbapenems [21]. The study by Samonis G et al., in Greece found that cancer patients infected with XDR-PA had primary bacteremia (65%), central venous catheter placed (28%), respiratory tract infections (22%) and Urinary Tract Infections (UTI) (22%). They were treated with Fluoroquinolones (63.2%), Carbapenems (63.2%), Colistin (21.1%), Penicillins (10.5%), Aminoglycosides (5.3%), respectively [5].

The study by Pena C et al., revealed that patients who had bloodstream infection caused by XDR-PA had underlying conditions- chronic obstructive pulmonary disease (37%), malignancy (26%), diabetes (21%), chronic renal failure (16%). Total 88% of patients with XDR-PA had prior antimicrobial agent administration (Amoxicillin-clavulanic acid (42%), Piperacillin-tazobactam (37%), Fluoroquinolones (50%), Carbapenems (21%), Cephalosporins (18%) and Aminoglycosides (18%) [12]. In addition, the study by Ciofi Degli Atti M et al., (Rome, Italy) revealed that patients who had bacteremia infection and had XDR-PA phenotype infections were resistant to Carbapenems, Cephalosporins, and Penicillins plus β-lactamase inhibitors [13]. The study by Deptula A et al., in Poland showed that patients with bloodstream infection were primarily diagnosed with gastrointestinal tract or intraabdominal infection (11.8%), UTI (11.2%), respiratory tract infection (10.1%), surgical site infection (7.1%) and skin and soft tissue infections (3.6%). The infection associated with the patient department accounted to 31.91% of ICU and 53.5% of medical department. More than a half of bloodstream infection was P. aeurugonosa resistant to Carbapenems (55.6%) [16].

A sensitivity analysis was constructed by using multiple logistic regression analysis in present study. The significant antimicrobial agents including Carbapenem and Ciprofloxacin in bivariate analysis were selected into the multivariate model. Also, the similar confounding factors were accounted into the final model. The model yielded a smaller magnitude and a similar association of using antimicrobial agents and treatment of XDR-PA bacteremia infections as shown in the results. This revealed that Carbapenem and Ciprofloxacin were more likely to be used to treat such P. aeruginosa infection.

Pharmacologically, the property of antimicrobial agents is used to treat bacterial infections. Cephalosporins are the original class of β-lactam antibiotics derived from the fungus Acremonium known as Cephalosporium. The latest Cephalosporins are more effective against gram-negative bacteria compared to previous generation [24]. It generally has a prolonged half-life of action, in particular against P. aeurginosa infection [25,26]. Carbapenems are the broadest spectrum and the greatest potential against bacterial infection in both gram-positive and gram-negative bacteria [27]. Carbapenems are the β lactam antibiotics with efficacy in treatment of severe infections. Additionally, an increasing use of Carbapenems is due to its resistance to Cephalosporin [28]. While Ciprofloxacin is a broad-spectrum quinolone carboxylic acid against a wide range of gram-negative and gram-positive bacteria [29].

The principal diagnosis among patients with XDR-PA is difficult due to underlying diseases and complications, which leads to an increase in the usage and the dosage of antimicrobial agents in treatment. Besides, the doses of antimicrobial agents were unaccounted in the study that leads to dose-response bias.

The present study found that third generation Cephalosporin, Carbapenem and Ciprofloxacin were more likely to be administrated for treatment in XDR-PA infections among patients. Hence, the study suggested that patients infected with XDR-PA phenotype should be intensively treated by antimicrobial agents in order to reduce the severity of disease.