Surveillance study of bloodstream infections, antimicrobial use, and resistance patterns among intensive care unit patients: A retrospective cross-sectional study.

Background: Bloodstream infections (BSIs) are one of the most critical illnesses requiring intensive care unit (ICU) admission. This study assessed patterns of antimicrobial use and resistance in ICU patients with BSIs.
Methods: Inpatients admitted to the ICU and who received at least one antimicrobial agent between January 1, 2017, and December 31, 2019, were included in the study. Electronic patients' medical records were used to collect patients' demographic, clinical, and microbiological data.
Results: A total of 1051 patients were enrolled in the study, where 650 patients (61.84%) were treated with three or more antimicrobial agents. The most frequently used antimicrobials were piperacillin/tazobactam followed by teicoplanin, meropenem, and levofloxacin. The most predominant multidrug-resistant pathogens were Acinetobacter baumannii, followed by Escherichia coli, Methicillin-resistant Staphylococcus aureus (MRSA), Klebsiella pneumonia, and Pseudomonas aeruginosa. Conclusions: The administration of the antimicrobials among ICU patients was highly based on a combination of three or more broad-spectrum agents. MDR pathogens were found to be highly prevalent among ICU patients with BSI. Therefore, we suggest recommending that hospital policies should apply the antimicrobial stewardship protocols, infection control, and implement antimicrobial de-escalation protocol to reduce the harm pressure of antimicrobial resistance. Copyright:

INTRODUCTION

Patients admitted to the intensive care unit (ICU) usually suffer from different medical conditions and their prognosis varies based on the severity of illness.[1]

Antimicrobial therapy (AMT) is one of the management strategies needed for ICU patients with sepsis. AMT should be selected appropriately; because ineffective or inappropriate AMT may lead to harmful outcomes, including the development of multidrug-resistant (MDR) organisms which resulted in longer hospital length-of-stay, longer ICU LOS, and increased morbidity and mortality.[23456] Broad-spectrum antimicrobials are not necessary to be administered for all patients; selected patients may require this extended coverage of antimicrobials, including multi-organ failure, invasive catheters, previous healthcare exposure, recent antibiotic use, history of resistant infection, and immunosuppression.[27] Therefore, the appropriate selection of empiric AMT should be based on the patient's specific factors and the location source of the infection.

Furthermore, the high spread of antimicrobial resistance (AMR) causes global threats worldwide.[89] The literature reported that AMR increases the risk of morbidity and mortality[101112131415] as well as the length of hospital stay, hospital cost, and charges.[16171819]

Multi-drug resistance (MDR) is defined as resistance to at least three antimicrobial classes.[20] Infections with MDR pathogens are increasingly common among ICU patients, with reports ranging from 50%[21] to 80%.[22]

This study aimed to assess the types of antimicrobial agents used among ICU patients with sepsis as well as the AMR pattern and predictor variables associated with MDR pathogens.

METHODS

Study design and setting

This retrospective study was conducted among ICU patients who received systemically at least one antimicrobial agent from January 1, 2017, to December 31, 2019. The study was conducted at King Abdullah University Hospital (KAUH), a tertiary care hospital in Jordan. The study was approved by the Institutional Review Board (IRB) at KAUH and informed consent was waived.

Data collection

Electronic patients’ medical records (.iSOFT Visalia [iSOFT Systems Inc., Aldershot, UK]) and charts in KAUH were used to obtain demographic and clinical information for each patient. Any patient who presents with multiple episodes of ICU admission within 1 year period was included as a single participant using the first episode, and other episodes were excluded. In addition, patients with incomplete information in their medical records and charts were excluded from the study. The demographic and clinical information is composed of four parts:

Part 1: Patient demographic characteristics (age, gender, weight, height, body mass index [BMI], length of hospital stay, length of ICU stay).

Part 2: Patient general health status (smoking, the presence of co-morbidities such as (hypertension, diabetes mellitus, myocardial infarction, atrial fibrillation, congestive heart failure, pulmonary diseases, chronic kidney disease, end-stage renal disease, cerebrovascular disease, solid tumors, lymphoma, leukemia, dementia, and prior major surgery), recent invasive procedure within 48 h of admission such as (bronchoscopy, central venous catheter, chest tubes, surgery, mechanical ventilation, catheterization previous hospitalization within 90 days of positive blood culture, transferring from another hospital to KAUH and previous antimicrobials administered within 90 days before hospital admission).

Part 3: Events that occurred during hospitalization (the primary ward/unit admission, administration of vasopressors, blood transfusion, type of nutrition support, pathogens isolated from body sites other than blood, and antimicrobials that were given during ICU stay).

Part 4: Events that occurred after obtaining the blood culture (empiric antimicrobials, definitive antimicrobials, pathogens were obtained from a blood test, the sensitivity test results, the mortality within 14 days and 30 days of positive blood culture, and the complications of the infection).

Microbiology testing

The VITEK II system (bioMerieux, Balmes-les-Grottes, France) identified the isolates during the study period. Antimicrobial susceptibility testing was conducted by the microdilution method on the VITEK II system. The Clinical and Lab Standard Institute breakpoints were used to define the susceptibility profile to the antimicrobial agents tested for the study period, as reported by the microbiology laboratory.[23] The susceptibility tests were checked for each patient and estimated the prevalence of MDR micro-organisms and the associated predictor variables.

Definitions

Bloodstream infections (BSIs) were defined as positive blood cultures with simultaneous signs and symptoms of infection. Line-related infection is a serious infection of the bloodstream that occurs when germs enter the body through a tube (central line) placed in a vein to deliver nutrients and medicine.[24] Sepsis and septic shock were defined based on the ACCP/SCCM criteria.[25] Blood culture contamination was defined as the recovery of normal skin flora from a single blood culture.[26] Among the clinical outcomes of the study, any antimicrobial that was given in the period between collecting the blood sample and obtaining the susceptibility test result was considered empiric therapy. On the other hand, any antimicrobial prescribed after obtaining the result of the susceptibility test was considered definitive therapy. To assess the appropriateness of empiric therapy, two main points must be met: the empiric drug therapy was given within 24 h of blood sample collection PLUS the infecting pathogen is sensitive to at least one of the given antimicrobial agents according to the susceptibility test results. Similarly, definitive therapy deems appropriate if it fulfilled two criteria: prescribed within 24 h of the susceptibility test results PLUS the infecting pathogen is sensitive to at least one of the administered antimicrobial agents according to the susceptibility test results. Multi-drug resistance (MDR) was defined as resistance to at least three antimicrobial classes.[27]

Statistical analysis

Statistical analysis was performed using SPSS (Statistical Package for the Social Sciences, version 23, IBM Corp., Armonk, New York, USA). Descriptive analysis was presented as the mean and standard deviation for continuous data, whereas frequencies and percentages were used to summarize categorical data. Demographic data and risk factors were analyzed using Chi-square test and independent t-test to look for differences between MDR and nonMDR groups. Multivariable logistic regression analysis (adjusted for age and gender) was used to identify independent risk factors associated with MDR in ICU patients with BSIs. All tests performed were two-tailed tests of significance and a P < 0.05 was considered significant.

RESULTS

Demographic and clinical characteristics

A total of 1051 patients were enrolled in this study; all of them had received at least one antimicrobial agent and were admitted to the ICU during the study period.

The demographic and clinical characteristics are presented in Table 1. The mean age and BMI of the study participants were 60.2 ± 19.3 and 27.6 ± 7.2, respectively. Most of them were male (54.5%), nonsmokers (83.4%), known cases of HTN (55.6%) and DM (45.8%), admitted to the ICU (86.4%) as primary ward admission with nosocomial infection (51.2%) of a respiratory focus site (34.3%). In addition, around one-third (36.1%) and one-fifth (22%) of the study participants were hospitalized within 90 days before admission and transfer to KAUH, respectively. Only 185 patients underwent invasive procedures within 48 h before admission.

Table 1

Demographic and clinical characteristics of 1051 patients admitted to the intensive care unit

Characteristics	n (%)
Gender
Male	573 (54.5)
Female	478 (45.5)
Smoking
Smoker	174 (16.6)
Not smoker	877 (83.4)
Co-morbidities
Hypertension	584 (55.6)
Diabetes mellitus	481 (45.8)
Myocardial infarction	176 (16.8)
Atrial fibrilation	67 (6.4)
Congestive heart failure	119 (11.3)
Pulmonary disorders	64 (6.1)
Chronic kidney disease	81 (7.7)
End stage renal disease	74 (7.0)
Cerbrovascular disease	151 (14.4)
Solid tumor	154 (14.6)
Lymphoma	14 (1.3)
Leukemia	12 (1.2)
Dementia	11 (1.1)
Previous surgery	52 (4.9)
Others	360 (34.3)
Invasive procedures within 48 h of admission
Broncoscopy	3 (0.3)
Venous catheter	9 (0.9)
Chest tube	6 (0.6)
Arterial catheter	11 (1.1)
Mechanical ventilation	84 (8.0)
Foleys catheter	55 (5.2)
Pig tube	17 (1.6)
Primary ward admission
ICU	908 (86.4)
Surgery	47 (4.5)
Medical	73 (6.9)
CCU	14 (1.3)
Oncology	9 (0.9)
Previous hospitalization (within 90 days)
Yes	379 (36.1)
No	672 (63.9)
Hospital transfer
Yes	231 (22)
No	820 (78)
Acquisition site
Nosocomial	538 (51.2)
Health care-associated	141 (13.4)
Community-acquired	372 (35.4)
Infection focus site
Respiratory	360 (34.3)
Genitourinary	191 (18.2)
Line-related	47 (4.5)
Gastrointestinal	80 (7.6)
Biliary	5 (0.5)
SSTI	69 (6.7)
CNS	68 (6.5)
Unknown	231 (22)
Age (mean±SD)	60.2±19.3
BMI (mean±SD)	27.6±7.2

SSTI: Skin and soft-tissue infection, CNS: Central nervous system, ICU: Intensive care unit, CCU: Coronary care unit, BMI: Body mass index, SD: Standard deviation

The pattern of antimicrobial use and distribution of microorganisms

During ICU stay, a total of 153 patients (14.6%) were treated with antimicrobial monotherapy, 248 patients (23.6%) were treated with dual AMT, whereas most of the patients (650 patients; 61.8%) were treated with three or more antimicrobial agents.

The patterns of antimicrobial use during ICU stay are illustrated in Table 2. During ICU stay, more than half of patients (69.9%, 65%, and 63%) were treated with glycopeptides (teicoplanin and vancomycin), penicillins (mostly piperacillin/tazobactam), and carbapenems (predominantly meropenem), respectively. Fluoroquinolones (40.9%), cephalosporins (30.4%), and aminoglycosides (15.6%) were also commonly used. Only 255 patients out of 1051 (24.3%) have received antifungal agents, primarily fluconazole (15.4%).

Table 2

Pattern of antimicrobial use during intensive care unit stay for 1051 patients

Antimicrobial agents	n (%)
Glycopeptides	735 (69.9)
Teicoplanin	466 (44.3)
Vancomycin	269 (25.6)
Penicillins	683 (65.0)
Piperacillin/tazobactam	653 (62.1)
Amoxicillin/clavulanic acid	19 (1.8)
Ampicillin	8 (0.8)
Amoxicillin	3 (0.3)
Carbapenems	662 (62.99)
Meropenem	425 (40.44)
Imipenem	222 (21.12)
Ertapenem	15 (1.43)
Fluoroquinolone	430 (40.9)
Levofloxacin	348 (33.1)
Ciprofloxacin	82 (7.8)
Cephalosporins	319 (30.4)
Cefazoline	150 (14.3)
Ceftriaxone	146 (13.9)
Cefuroxime	17 (1.6)
Cefixim	3 (0.3)
Cefotaxim	3 (0.3)
Aminoglycosides	164 (15.6)
Gentamycin	100 (9.5)
Amikacin	63 (6.0)
Tobramycin	1 (0.1)
Oxazolidinone	37 (3.5)
Linezolid	37 (3.5)
Glycylcyclin	14 (1.3)
Tigecycline	14 (1.3)
Macrolides	10 (1)
Clarithromycin	8 (0.8)
Azithromycin	1 (0.1)
Erythromycin	1 (0.1)
Tetracyclins	7 (0.7)
Doxycycline	7 (0.7)
Miscellaneous	285 (27.1)
Colistin	145 (13.8)
Metronidazole	120 (11.4)
TMP/SMX	15 (1.4)
Rifampin	5 (0.5)
Anti-fungal agents	255 (24.3)
Fluconazole	162 (15.4)
Caspofungin	41 (3.9)
Anidulafungin	39 (3.7)
Nystatin	36 (3.4)
Voriconazole	11 (1.1)
Amphotracin B	2 (0.2)

TMP/SMX: Trimethoprim-sulfamethoxazole

Regarding BSI among ICU patients, Gram-negative bacteria represented half of the isolated pathogens (22/44, 50%), (14/44, 31.8%) were Gram-positive bacteria, and only eight pathogens (18.2%) were fungi. Most of our study participants were infected with only one pathogen (882 out of 1051; 83.9%), whereas the rest were infected with poly-microbial pathogens as follow: 132 patients (12.6%) were infected with two pathogens, 31 patients (3%) were infected with three pathogens, four patients (0.4%) were infected with four pathogens, and only two patients (0.2%) were infected with five pathogens. The susceptibility test was done for 39 types of pathogens in 1265 positive blood cultures. Only 378 patients were included with 442 susceptibility tests (34.9%) that were included in this section of the study. In addition, 823 (6.1%) tests were excluded as they fulfilled the definition of contamination explained in the method section.

The types of pathogens that infected our study participants are summarized in Table 3. Our patients were infected more frequently with Gram-negative bacteria compared to Gram-positive bacteria and fungi. The predominant gram-negative bacteria that infected our study participants were Escherichia coli, followed by Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa. On the other hand, the most predominant Gram-positive bacteria were MRSA, followed by Staphylococcus aureus, Enterococcus faecalis, Enterococcus faecium, and Streptococcus pneumonia. Regarding fungal infections, they account for the lowest percentage compared to bacterial infections, where Candida albicans and Candida glabrata are the most frequent fungal infections detected among our study patients.

Table 3

Distribution of microorganisms isolated from 1051 patients admitted to the intensive care unit

Microorganism	n (%)
Gram-negative bacteria (n=298)
E. coli	87 (29.2)
K. pneumoniae	86 (28.9)
A. baumannii	69 (23.2)
P. aeruginosa	17 (5.7)
P. mirabilis	8 (2.7)
E. cloacae	5 (1.7)
E. aerogenes	4 (1.3)
Salmonella species	3 (1.0)
M. morganii	2 (0.7)
P. stutzeri	1 (0.4)
Others	16 (5.4)
Gram-positive bacteria (n=119)
MRSA	43 (36.1)
S. aureus	29 (24.4)
E. faecalis	17 (14.3)
E. faecium	10 (8.4)
S. pneumoniae	9 (7.6)
S. milleri	5 (4.2)
S. epidermidis	2 (1.7)
Others	4 (3.4)
Fungi (n=25)
C. albicans	6 (24)
C. glabrata	6 (24)
C. tropicalis	4 (16.7)
C. parapsilosis	4 (16.7)
C. lipolytica	2 (8)
C. famata	1 (4)
C. dubliniensis	1 (4)
C. kefyr	1 (4)

E. coli: Escherichia coli, K. pneumonia: Klebsiella pneumonia, A. baumannii: Acinetobacter baumannii, P. aeruginosa: Pseudomonas aeruginosa, P. mirabilis: Proteus mirabilis, E. cloacae: Enterobacter cloacae, E. aerogenes: Enterobacter aerogenes, M. morganii: Morganella morganii, P. stutzeri: Pseudomonas stutzeri, S. aureus: Staphylococcus aureus, E. faecalis: Enterococcus faecalis, E. faecium: Enterococcus faecium, S. pneumonia: Streptococcus pneumonia, S. milleri: Streptococcus milleri, S. epidermidis: Staphylococcus epidermidis, C. albicans: Candida albicans, C. glabrata: Candida glabrata, C. tropicalis: Candida tropicalis, C. parapsilosis: Candida parapsilosis, C. lipolytica: Candida lipolytica, C. famata: Candida famata, C. dubliniensis: Candida dubliniensis, C. kefyr: Candida kefyr, MRSA: Methicillin-resistant-S. aureus

Antimicrobial resistance profile

Resistance profiles for Gram-negative bacteria according to electronic microbiology records are depicted in Table 4. The resistance profile of E. coli was observed to be high for ampicillin/sulbactam (62.5%), third and fourth generations’ cephalosporines (60.8%–70.3%), and fluoroquinolones (56.9%–69.2%). The rest of the antibiotics were somewhat active on E. coli, including colistin, tigecycline, imipenem, and meropenem. When going through K. pneumonia, it was a more resistant pathogen compared to E. coli. More than 70% of resistance was observed against ampicillin/sulbactam, ticarcillin/clavulanic acid, third and fourth generations’ cephalosporines, and gentamycin. In comparison, the least resistant antibiotics were imipenem (9.9%), tigecycline (16.2%), and amikacin (16.7%). A. baumannii was the predominant resistant pathogen among our study patients were more than 75% of this isolate resistant to almost all antibiotics except for colistin (4.4%) and tigecycline (33.3%) which were shown to be the most susceptible antibiotics to A. baumannii. Regarding P. aeruginosa, one-third of the isolates were resistant to imipenem (33.3%), one-fifth were resistant to levofloxacin (20%). Moreover, one-third of P. aeruginosa isolates were sensitive to tigecycline, although it lacks the activity on P. aeruginosa.

Table 4

Antimicrobial resistance profile for Gram-negative bacteria isolated from intensive care unit patients*

Antibiotics	E. coli (%)	K. pneumonia (%)	A. baumannii (%)	P. aeruginosa (%)
Piperacillin/tazobactam	3/24 (12.5)	5/15 (33.3)	36/36 (100)	2/12 (16.7)
Ampicillin/sulbactam	30/48 (62.5)	29/38 (76.3)	23/25 (92)	5/5 (100)
Ticarcillin/clavulanic acid	19/39 (48.7)	44/57 (77.2)	43/43 (100)	4/17 (23.5)
Meropenem	3/78 (3.9)	18/67 (26.9)	65/67 (97.0)	1/17 (5.9)
Imipenem	2/79 (2.5)	8/81 (9.9)	61/62 (98.4)	5/15 (33.3)
levofloxacin	41/72 (56.9)	48/83 (57.8)	38/39 (97.4)	1/5 (20.0)
Ciprofloxacin	9/13 (69.2)	1/2 (50)	30/30 (100)	2/12 (16.7)
Moxifloxacin	40/70 (57.1)	48/83 (57.8)	20/23 (86.9)	0/1 (0)
Ceftriaxone	49/76 (64.5)	60/83 (72.3)	36/36 (100)	4/5 (80)
Ceftazidime	8/13 (61.5)	2/2 (100)	32/32 (100)	3/12 (25)
Cefepime	48/79 (60.8)	62/85 (72.9)	65/65 (100)	3/17 (17.7)
Cefixime	52/74 (70.3)	61/80 (76.3)	26/26 (100)	4/4 (100)
Gentamycin	7/22 (31.8)	8/11 (72.7)	26/31 (83.9)	3/12 (25)
Amikacin	3/24 (12.5)	3/18 (16.7)	3/4 (75)	2/12 (16.7)
Tobramycin	3/9 (33.3)	1/2 (50)	25/30 (83.3)	4/12 (33.3)
Colistin	0/53 (0)	16/57 (28.1)	3/68 (4.4)	0/16 (0)
Tigecycline	1/74 (1.4)	12/74 (16.2)	13/39 (33.3)	2/3 (66.7)

*The percentages shown in the table represent the resistance of each pathogen. E. coli: Escherichia coli, K. pneumonia: Klebsiella pneumonia, A. baumannii: Acinetobacter baumannii, P. aeruginosa: Pseudomonas aeruginosa

As shown in Table 5, more than three-quarters of resistant strains of S. aureus (MRSA) were susceptible to ciprofloxacin (83.3%), and about half of these isolates were susceptible to erythromycin (51.2%). Although clindamycin is practically used as anti-MRSA, nearly half of MRSA covers both isolates of isolates were resistant to clindamycin (47.6%). Concerning Enterococcus species, they showed similar resistance profile except for ampicillin/sulbactam (16.7% with E. faecalis vs. 100% with E. faecium), quinupristin/dalfopristin (91.7% with E. faecalis vs. 25% with E. faecium) and imipenem (22.2% with E. faecalis vs. 100% with E. faecium). Even though vancomycin covers both isolates of Enterococcus species, E. faecium showed to be more resistant to vancomycin (33.3%) compared to 6.7% with E. faecalis. Moreover, teicoplanin is not effective against E. faecium, but 66.7% of E. faecium were susceptible to teicoplanin. More importantly, Trimethoprim-Sulfamethoxazole (TMP/SMX) was highly resistant to E. faecalis (100%). Finally, S. pneumonia is highly resistant against erythromycin (80%) and clindamycin (66.7%). In addition, half of these isolates were resistant to imipenem (50%), and 80% were highly susceptible to rifampin.

Table 5

Antimicrobial resistance profile for gram-positive bacteria isolated from intensive care unit patients*

Antibiotics	S. aureus (MRSA) (%)	S. aureus (MSSA) (%)	E. faecalis (%)	E. faecium (%)	S. pneumonia (%)
Vancomycin	0/43 (0)	0/27 (0)	1/15 (6.7)	3/9 (33.3)	0/9 (0)
Tiecoplanin	0/42 (0)	0/29 (0)	2/17 (11.8)	3/9 (33.3)	0/8 (0)
Ampcillin/sulbactam	1/1 (100)	1/1 (100)	1/6 (16.7)	3/3 (100)	Not done
Penicillin G	1/1 (100)	2/2 (100)	ND	ND	0/7 (0)
Oxacillin	42/42 (100)	0/29 (0)	ND	ND	1/1 (100)
Imipenem	17/17 (100)	0/14 (0)	2/9 (22.2)	3/3 (100)	1/2 (50)
Ciprofloxacin	3/18 (16.7)	3/14 (21.4)	2/2 (100)	3/3 (100)	Not done
Cefoxitine	42/43 (97.7)	0/27 (0)	ND	ND	1/1 (100)
Cefotaxim	ND	ND	ND	ND	1/5 (20)
Ceftriaxone	ND	0/2 (0)	ND	ND	0/7 (0)
Linezolid	0/42 (0)	0/28 (0)	0/16 (0)	0/9 (0)	0/6 (0)
Tigecycline	0/38 (0)	0/25 (0)	0/15 (0)	0/8 (0)	0/3 (0)
Erythromycin	21/43 (48.8)	4/28 (14.3)	15/17 (88.2)	8/8 (100)	4/5 (80.0)
Clindamycin	20/42 (47.6)	5/29 (17.2)	5/5 (100)	3/3 (100)	4/6 (66.7)
TMP/SMX	6/35 (17.1)	0/25 (0)	3/3 (100)	1/1 (100)	0/2 (0)
Rifampin	3/39 (7.7)	0/26 (0)	ND	ND	1/5 (20)
Quinupristin/dalfopristin	0/1 (0)	0/1 (0)	11/12 (91.7)	2/8 (25)	Not done

*The percentages shown in the table represent the resistance of each pathogen. S. aureus: Staphylococcus aureus, E. faecalis: Enterococcus faecalis, E. faecium: Enterococcus faecium, S. pneumonia: Streptococcus pneumonia, TMP/SMX: Trimethoprim-Sulfamethoxazole, ND: Note done, MRSA: Methicillin-resistant-S. aureus, MSSA: Methicillin-susceptible-S. aureus

Prevalence of MDR and predictor variables

As mentioned previously, susceptibility tests were conducted for 378 patients, a total of 273 patients of them (72.2%) were shown to be infected with MDR pathogens. The most frequently MDR pathogens infected our study patients were A. baumannii was 67/69 (97.1%), E-coli was 76/87 (87.4%), MRSA was 37/43 (86%), K. pneumonia was 73/86 (84.9%), P. aeruginosa was 10/17 (58.8%), Enterococcus species 12/27 (44.4%), S. pneumonia 3/9 (33.3%), C. albicans was 2/6 (33.3%), S. aureus (MSSA) 4/29 (13.8%). Table 6 represents the predictor variables which associated with MDR pathogens.

Table 6

Predictor variables for multi drug resistance among 378 intensive care unit patients

Variable	Non-MDR (n=105), n (%)	MDR (n=273), n (%)	P
Gender
Male	60 (57.1)	146 (53.5)	0.560
Female	45 (42.9)	127 (46.5)
Age group
18-64	57 (54.3)	143 (52.4)	0.782
≥65	48 (45.7)	130 (47.6)
BMI
<30	62/76 (81.6)	135/204 (66.2)	0.020*
≥30	14/76 (18.4)	69/204 (33.8)
Hospital stay	13 (5-25)	21 (10-34)	0.010*
ICU stay (days)
≤7	70 (66.7)	119 (43.6)	<0.001*
>7	35 (33.3)	154 (56.4)
Smoking	89 (84.8)	228 (83.5)	0.860
Total comorbidities
0	7 (6.7)	23 (8.4)	0.490
1-4	92 (87.6)	229 (83.9)
≥5	6 (5.7)	21 (7.7)
Prior catheterization	8 (7.6)	28 (10.3)	0.425
Previous hospitalization	43 (41.0)	115 (42.1)	0.803
Hospital transfer	19 (18.1)	47 (17.2)	0.858
Acquisition site
Nosocomial	28 (26.7)	137 (50.2)	<0.001*
Health care-associated	24 (22.9)	50 (18.3)
Community acquired	53 (50.5)	86 (31.5)
Source of the infection site
Respiratory	30 (28.6)	100 (36.6)	0.131
CNS	5 (4.8)	9 (3.3)	0.506
GU	21 (20.0)	69 (25.3)	0.263
Line related	18 (17.1)	19 (7.0)	0.003*
GI	8 (7.6)	25 (9.2)	0.619
SSTI	6 (5.7)	19 (7.0)	0.425
Pathogens from other body sites	62 (59.1)	193 (70.7)	0.026*
14 days death	50 (47.6)	135 (49.5)	0.713
30 days death	54 (51.4)	146 (53.5)	0.082
Previous antibiotic use	31 (29.5)	81 (29.7)	0.476
Sepsis	63 (60.0)	163 (59.7)	0.823
Severe sepsis	14 (13.3)	64 (23.4)	0.028*
Septic shock	36 (34.3)	97 (35.5)	0.792

*: Statistical significance, BMI: Body mass index, ICU: Intensive care unit, CNS: Central venous system, GU: Genito-urinary, GI: Gastro-intestinal, SSTI: Skin and soft-tissue infection, MDR: Multidrug resistance

As shown in Table 6, obese patients (BMI ≥30) were significantly more susceptible to be infected with MDR isolates, while nonobese patients (BMI <30) were more associated with non-MDR isolates (P = 0.020). In addition, a longer hospital stay increases the risk of being infected with MDR isolates (P = 0.009) as well as the ICU station for more than 1 week increases the risk of MDR infections (P < 0.001). Moreover, around 50% of patients with MDR infections (137 patients) were infected with nosocomial infections, whereas around 50% of patients with non-MDR infections were infected with community-acquired infections. However, regarding healthcare-associated infections, comparable percentages between both groups slightly tilted toward the non-MDR group (22.9% vs. 18.3%). Furthermore, line-related infections were significantly caused by non-MDR isolates, and only 7% of the patients were infected with MDR isolates. Those who were positive cultures of body sites other than blood concurrently with BSI were significantly susceptible to be infected with MDR pathogens (P = 0.026). Finally, 23.4% of the patients with MDR-BSI were complicated with severe sepsis (P = 0.028). The 14-day and 30-day mortality rates were also analyzed, they shown to be higher among the MDR group, but they were not statistically significant between both groups (P = 0.713, 0.082, respectively). Finally, a multivariable analysis using logistic regression model identified the following as independent risk factors associated with MDR in ICU patients BSIs: obesity (odds ratio [OR]: 2.1, 95% confidence interval [CI] [1.03–4.31], P = 0.0384), line-related catheter (OR: 3.3, 95%CI [1.25–8.89], P = 0.0155), and nosocomial infections acquisition (OR: 2.21, 95% CI [1.13-4.43], P = 0.0212).

DISCUSSION

The current study is the first study in Jordan to investigate the pattern of antimicrobial use and resistance among ICU patients. During ICU stay, all participants were treated with at least one antimicrobial agent; more than half were treated with three and more antimicrobial agents.

This study was conducted in a tertiary care hospital, the top five antimicrobials used were: piperacillin/tazobactam, teicoplanin, meropenem, levofloxacin, vancomycin, and imipenem. This is usually attributable to the degree of ASP implementation, types and severity of the infections, and the availability of selected antimicrobials in hospital formulary. In general, critically ill patients admitted to the ICU are susceptible to several types of infections. However, only 169 patients (16.1%) in this study had a polymicrobial infection. It is important to emphasize that the administration of broad-spectrum therapy is not always necessarily recommended and not as important as administering AMT actively against the most likely pathogens. Special attention should be given to each patient's risk factors and the most likely pathogen based on the infection sources of BSI before selecting the appropriate AMT.

In this study, we found that the distribution of pathogens was different from previous studies in developing and developed countries. Concerning Gram-negative bacteria, the most frequent pathogens in this study were E. coli, followed by K. pneumonia, A. baumannii, and P. aeruginosa. Similarly, an earlier study assessed the susceptibility data of predominant ICU pathogens in France, Germany, Italy, Canada, and the United States reported that E. coli was the most frequent Gram-negative species isolated from infections in the ICU and P. aeruginosa being common in three (USA, Canada, France) of the five countries.[27] In a contradicting manner, most BSI among ICU patients in Turkey and Bangladesh were caused by P. aeruginosa species.[2829] In the same region, the common Gram-negative infection in Saudi Arabia was Acinetobacter species, followed by P. aeruginosa, E. coli, and K. pneumonia.[30]

In our study, the most common Gram-positive bacteria isolated from our patients were MRSA, followed by MSSA, E. faecalis, E. faecium and S. pneumonia.

Consistent with previous studies, fungal infections were less common to infect our study participants compared to bacterial infections where the most predominant fungal infections were C. albicans and C. glabrata followed by Candida tropicalis.[2931]

In terms of MDR, this study showed a high prevalence of MDR pathogens where almost three-quarters of our ICU patients were infected with MDR pathogens. The prevalence of MDR organisms diverges broadly among regions and countries over the world. However, the isolation of MDR organisms among ICU patients had been reported to be high. That could be attributed to the fact that ICU patients usually undergo invasive surgical procedures, prolonged hospitalization, and long-term antimicrobial use.

In this study, several variables were identified to be associated with MDR at univariate level including obesity, longer hospital stay, longer ICU stay, infections with nosocomial infections, line-related infections, severe sepsis, and isolation of pathogens in other body sites. Among these findings, only line-related infections, nosocomial infections, and obesity were identified as independent variables. It is well known that nosocomial infections prone higher risk for MDR pathogens. Previously, six risk factors of MDR Gram-negative bacteria in the ICU were identified by meta-analysis; male gender, operation procedure, central venous catheter, mechanical ventilation, previous antibiotic therapy, and length of ICU stay.[32] The implementation of antimicrobial stewardship (ASP) practices, infection control, and medical staff prevention care is needed to reduce the spread of MDR isolates One of the negative findings in this study was the lack of difference of MDR and mortality at 30 and 15 days. This is usually not expected as infections with MDR pathogens usually versatile and associated with poor prognosis, however, the interindividual variation in severity of the illness can be key factor in this equation.

The literature showed different resistance distributions of Gram-negative and Gram-positive bacteria. Among this study's patients, E. coli showed a high resistance profile against ampicillin/sulbactam, third and fourth generations’ cephalosporines, and fluoroquinolones. Similarly, some studies reported that E. coli was resistant to fluoroquinolones[293033] in addition to third and fourth generations cephalosporines. K. pneumonia showed resistance of more than 70% against beta-lactam/beta-lactamase agents, and third-fourth generations cephalosporines which were similar to previous findings in other countries.[223334] In this study, more than 75% of A. baumannii isolates were resistant to almost all tested antimicrobials except for colistin and tigecycline. Consistently, other studies revealed that more than 90% of A. baumannii were resistant to most of the tested antibiotics.[2234] These findings ensure that A. baumannii is a common MDR leaving fewer therapeutic options for clinicians including colistin or tigecycline. Although colistin was shown to be active on A. baumannii, a study in Turkey expected that colistin may consider as pan-drug resistant.[35] Finally, despite imipenem and levofloxacin being anti-pseudomonal agents, one-third and one-fifth of isolated Pseudomonal spp in this study were resistant to them, respectively.

For Enterococcus species, one-third of isolated E. faecium in this study were resistant to vancomycin and two-third were susceptible to teicoplanin. In addition, all the isolated E. faecalis were resistant to TMP/SMX.

The present study showed that S. pneumonia is highly resistant to erythromycin, clindamycin, and imipenem but susceptible to rifampin. Consistent results were reported by Beheshti et al. where erythromycin and clindamycin were highly resistant,[36] but Al-Tawfiq reported that only 25% of S. pneumonia were resistant to erythromycin.[37] The dissemination of multi-resistant clones and different patterns in the use of AMT, infection control, and epidemiological features may contribute to these huge variations of resistant phenotypes for Gram-negative and Gram-positive bacteria.

This study had several limitations. First, it was a retrospective design that might pose hidden biases. There were limitations in data accessibility and availability, such as the severity of illness score (for example, APACHE II score), which was not incorporated because of unavailable data another example is missing in the susceptibility tests resulted in low rate susceptibility tests (36%) which it may affect the resistant pattern. Second, our study was a single-center, and the results may not be applied to other settings. Finally, medical records did not identify obvious reasons for the delay in AMT.

CONCLUSIONS

The present study investigated the types of antimicrobial use and the AMR pattern among ICU patients in a Jordanian tertiary care hospital. The administration of antimicrobials among ICU patients was highly based on a combination of three or more agents. The MDR prevalence was high among the patients predominantly with A. baumannii, E-coli and MRSA. ASP programs should target ICU patients to optimize antimicrobial management and control modifiable factors in this high-risk group.

Research quality and ethics

This study was approved by the Institutional Review Board / Ethics Committee at King Abdullah University Hospital (Approval # 167/2020; Approval date Mar 31, 2020). The authors followed the applicable EQUATOR Network (http://www.equator-network.org/) guidelines, specifically the STROBE Guidelines, during the conduct of this research project.

Financial support and sponsorship

This study was supported by a grant number: 20200248 from the Deanship of Research at Jordan University of Science and Technology, Irbid, Jordan.

Conflicts of interest

There are no conflicts of interest.