Acinetobacter baumannii resistance trends in an ICU from Saudi Arabia over 12 years

Abstract

Background and objectives

Acinetobacter baumannii is a major intensive care unit (ICU)-associated pathogen due to associated extensive drug resistance. Regional data from southern Saudi Arabia remain limited. This study aimed to describe overall and temporal resistance patterns of ICU-associated A. baumannii.

Methods

We conducted a 12-y retrospective study (August 2012– August 2024) at Aseer Central Hospital, Saudi Arabia, analysing Acinetobacter spp. clinical isolates from ICU and non-ICU settings. Species identification and antimicrobial susceptibility testing (AST) were performed using the VITEK® 2 system, with susceptibility interpreted per CLSI M100 guidelines. Carbapenem-resistant A. baumannii (CRAB) and difficult-to-treat resistance phenotypes were defined. Temporal trends were assessed via logistic regression.

Results

Of 3,935 isolates, 96.6% were A. baumannii. ICU-derived isolates (48.5%) showed significantly higher resistance rates across nearly all antibiotic classes compared to non-ICU isolates (P<0.001). Carbapenem resistance rates remained above 90% throughout the study period, while difficult to treat resistance ranged between 65.2 to 96.1%. Logistic regression revealed significant temporal declines in resistance to cefepime, ceftazidime, aminoglycosides, and fluoroquinolones (P<0.001).

Interpretation and conclusions

Acinetobacter baumannii isolates from ICU settings in southern Saudi Arabia exhibit persistently high resistance, including carbapenem resistance and difficult-to-treat resistance. We observed significant temporal declines in resistance to selected agents, particularly cefepime, aminoglycosides, and fluoroquinolones. These findings highlight the critical role of ICU-focused antimicrobial stewardship, enhanced infection control strategies, and regional molecular surveillance to limit further spread.

The global escalation of antimicrobial resistance threatens the efficacy of modern medicine and increases the burden of healthcare-associated infections. Among the most formidable pathogens is Acinetobacter baumannii, a Gram-negative, non-fermenting bacillus recognised for its ability to cause multidrug-resistant (MDR) infections, particularly in critically ill patients.¹ Its resilience in hospital environments, intrinsic resistance mechanisms, and capacity for outbreak propagation make it a major threat in intensive care units (ICUs).^2,3

A. baumannii is increasingly resistant to β-lactams, aminoglycosides, fluoroquinolones, and even last-line agents such as tigecycline and polymyxins. Resistance mechanisms include OXA-type carbapenemases, efflux pumps, porin loss, and mutations in target genes.³ In response, the World Health Organization has designated carbapenem-resistant A. baumannii (CRAB) a priority 1 ‘critical’ pathogen for antibiotic development,⁴ while the Centers for disease control and prevention (CDC) lists it as a serious public health threat.⁵ Beyond CRAB, the concept of difficult-to-treat resistance (DTR)⁶ has emerged as a clinically meaningful phenotype, defined as non-susceptibility to all first-line agents, including β-lactams and fluoroquinolones, thereby restricting therapy to less effective or more toxic alternatives. DTR has been associated with increased mortality and prolonged hospital stays, making it a critical metric for resistance surveillance.

In Saudi Arabia, national surveillance has revealed high CRAB prevalence, particularly in ICUs.^7,8 A study from Riyadh, the capital city located in the central region of Saudi Arabia, reported carbapenem resistance rates exceeding 90%, with frequent detection of resistance genes such as blaOXA-23, blaNDM, and blaPER.⁹ However, data from the southern region, including Aseer province, remains scarce. Almaghrabi et al² highlighted the emergence of MDR A. baumannii bacteraemia in this region, classifying it as an emerging health threat.

Given the disproportionate burden in ICUs and limited regional data, we conducted a 12-y surveillance study in a tertiary hospital in southern Saudi Arabia. We aimed to characterise A. baumannii resistance patterns, temporal trends, and ICU-specific phenotypes, filling a critical gap in local antimicrobial resistance (AMR) surveillance and informing infection control and stewardship priorities.

Methods

Study design and setting: This retrospective study was conducted by the department of Microbiology, Aseer Central Hospital, Abha, a 500-bed tertiary care centre in southern Saudi Arabia, between August 2012 and August 2024. The study was approved by the Institutional Review Board of King Khalid University. A waiver of informed consent was granted, given the retrospective design and use of fully de-identified data. The study dataset was extracted and curated in de-identified form from the hospital information system for research purposes after ethics approval; data processing and analysis were performed between January 2024 and March 2025; all data were fully de-identified, and investigators had no access to personally identifiable information during or after data extraction. The hospital houses a 50-bed ICU. All clinical isolates of Acinetobacter spp. were included in the analysis, with only the first isolate per patient per infection episode considered. This study did not involve longitudinal follow up of individual patients; each isolate represents a single data point per patient per infection episode, and the analysis reflects laboratory-based surveillance trends over time rather than repeated patient-level measurements. However, for cases with suspected bacterial translocation events (e.g., isolation of the same species from a non-blood site followed by bloodstream infection in the same patient), additional isolates from different specimen types were included to capture these clinically relevant events. Paediatric patients under 12 y of age were excluded. Isolates were obtained from both ICU and non-ICU settings and derived from a range of clinical specimens, including respiratory tract samples, urine, blood, wound swabs, abscesses, cerebrospinal fluid (CSF), and other sterile body fluids. Relevant microbiological and demographic data were extracted from the hospital’s electronic health system (EHS).

Microbiological procedures

Specimens were collected aseptically and processed by institutional protocols. Initial culture included blood agar, chocolate agar, and MacConkey agar. Plates were incubated at 35–37°C and examined after 24 and 48 h.

Species identification and antimicrobial susceptibility testing (AST) were conducted using the VITEK® 2 automated system (bioMérieux, Marcy-l’Étoile, France). When required, AST results were verified using disk and gradient diffusion tests (bioMérieux^TM, Marcy l’Étoile, France), which were employed, following the clinical and laboratory standards institute (CLSI) guidelines. Interpretive breakpoints were applied based on the annually updated CLSI M100 standards; the VITEK® 2 system breakpoints were updated annually in accordance with the most current CLSI edition at the time of testing. Non-susceptibility was defined, following CLSI criteria, as the combined proportion of isolates classified as intermediate or resistant to the tested antimicrobial agents.¹⁰ These categories were grouped for surveillance analysis to reflect reduced clinical efficacy and to standardise trend evaluation across the study period. Colistin and cefiderocol were excluded due to a lack of validated testing platforms during the study period. Additionally, tobramycin susceptibility data from 2016 were excluded from trend analysis due to suspected misclassification or underreporting in the laboratory information system during that year.

Resistance phenotype definitions

CRAB was defined as non-susceptibility to ≥1 carbapenem (imipenem or meropenem). DTR was defined as in vitro intermediate or resistant to all first-line antibiotic agents, including all carbapenems, extended-spectrum β-lactams, and fluoroquinolones, based on Kadri et al.⁶

Statistical analysis

Data were analysed using SPSS Statistics version 25.0 (IBM Corp., Armonk, NY, USA). Continuous variables were summarised as means ± standard deviations and compared using independent-samples t-tests after assessment of distributional assumptions. Categorical variables were expressed as frequencies and percentages and compared using the Chi-square test or Fisher’s exact test, as appropriate.

Temporal trends in antimicrobial non-susceptibility were evaluated using binary logistic regression models, with calendar year entered as a continuous independent variable to estimate annual changes in resistance. Model outputs were reported as odds ratios (ORs) with corresponding 95% confidence intervals (CIs). For surveillance interpretation, the annual percentage change was derived from the yearly OR as (OR − 1) × 100. A two-sided P value < 0.05 was considered statistically significant.

To explore potential bacterial translocation events, bloodstream isolates of Acinetobacter baumannii were matched to preceding non-blood isolates obtained from the same patient using unique medical record numbers. Only the temporally nearest prior non-blood isolate with concordant organism identification was included for each case. The frequency of these events was calculated relative to the total number of A. baumannii isolates (N=3,800).

Results

Isolate distribution and patient characteristics

Of 4,131 clinical samples screened, 3,935 Acinetobacter spp. isolates met the inclusion and exclusion criteria. Acinetobacter baumannii accounted for the vast majority (96.6%, n=3,800), while non-baumannii species represented only 3.4% (n=135). Baseline and demographic characteristics of A. baumannii isolates are shown in Table I.

Table I.

Demographic and clinical characteristics of patients with Acinetobacter baumannii isolates (N=3800) and their association with infection settings and risk factors

Characteristic		Acinetobacter baumannii (n=3800); n (%)	P value	Odds ratio (OR) (95% CI)
Gender			<0.001	2.57 (2.38–2.78)
	Male*	2737 (72)
	Female	1063 (28)
Risk factors			<0.001
	Long stay	704 (18.3) Expected 20%	<0.001
	Burn	82 (2.2)
	ESRD	63 (1.7)
	DM	35 (0.9)
	Other	2916 (76.7)
Setting			<0.001	21.1 (18.0–24.7)
	Inpatient (INP)*	3628 (95.5)
	Outpatient (OPD)	172 (4.5)
Hospitalisation ward			<0.001
	Medical ward	985 (25.9)
	Surgical ward	800 (21.1)
	ICU wards	1843 (48.5)

Percentages are calculated based on the total number of Acinetobacter baumannii isolates (n=3800). Chi-square test was used to assess associations between categorical variables. Odds ratios (ORs) with 95% confidence intervals (CI) were calculated where a binary comparison was applicable. “Male” and “Inpatient (INP)” were used as reference categories for OR estimation. For risk factors and hospitalisation ward, individual ORs were not calculated due to multi-category comparisons. Expected proportions are shown where indicated to highlight deviation from anticipated distribution. P < 0.05 was considered statistically significant

^*Reference category for the corresponding odds ratio.

ICU, intensive care unit; ESRD, end-stage renal disease; DM, diabetes mellitus; OPD, outpatient department

Culture source distribution

Respiratory specimens were the predominant source (n=1896; 48.2%), with A. baumannii isolated in 1854 (97.8%) of these cases. Wound samples (n=741; 18.8%), urine (n= 630; 16.0%), and blood (n=308; 7.8%) also showed high proportions of A. baumannii, comprising 713 (96.2%), 596 (94.6%), and 295 (95.8%) of isolates, respectively. In specimens from other sites, including cerebrospinal fluid, abscesses, tissue, and synovial fluid, A. baumannii remained dominant (n=342; 95.0%). Non-baumannii species were infrequent across all specimen types ( Table II).

Table II.

Comparison of culture sources between Acinetobacter baumannii and other Acinetobacter species

Characteristic	Total samples (n=3935); n (%)	Organism
		Acinetobacter baumannii (n=3800); n (%)	Acinetobacter species (n=135); n (%)
Culture source
Respiratory	1896 (48.2)	1854 (97.8)	42 (2.2)
Wound samples	741 (18.8)	713 (96.2)	28 (3.8)
Urine	630 (16.0)	596 (94.6)	34 (5.4)
Blood	308 (7.8)	295 (95.8)	13 (4.2)
Other	360 (9.1)	342 (95.0)	18 (5.0)

Other; includes CSF, deep abscess aspiration, tissue biopsies and synovial fluid

Antimicrobial resistance patterns and temporal trends

Of the 3,800 A. baumannii isolates, 50.8% originated from ICUs. Respiratory samples were more prevalent in ICU settings (58.5% vs. 41.5%, P< 0.001), while urinary and sterile site cultures were more common among non-ICU patients. Bloodstream infections were significantly higher in ICU patients (63.8% vs. 36.2%) ( Table III). Non-susceptibility was significantly higher in ICU isolates across nearly all antibiotic classes, including β-lactams, aminoglycosides, and fluoroquinolones (P<0.001) ( Table IV). At the species level, A. baumannii exhibited significantly higher resistance than non-baumannii species across all tested agents (P<0.001) (Supplementary Table I). Carbapenem non-susceptibility reached 93.4% for imipenem and 82.8% for meropenem in A. baumannii, compared with 68.9% and 36.3% in other species, respectively. Inpatients, representing 95.1% of cases, had significantly higher non-susceptibility across most agents. Meropenem resistance was 84.1% in inpatients compared with 32.6% in outpatients (P<0.001), and this pattern remained when restricting the analysis to A. baumannii isolates (Supplementary Tables II and III). Temporal analysis using logistic regression revealed significant declines in resistance to cefepime, piperacillin–tazobactam, aminoglycosides, and fluoroquinolones (all P< 0.001) (Supplementary Table IV). While ceftazidime resistance declined overall, the trend was not statistically significant in ICU-specific analysis (P=0.794) ( Table V). Among ICU A. baumannii isolates, resistance to imipenem increased significantly (OR=1.109, P=0.013), whereas resistance to gentamicin and tobramycin declined sharply (OR=0.785 and 0.829, respectively; P<0.001) ( Table V, Fig. 1). The annual proportion of ICU A. baumannii isolates classified as CRAB remained above 90% throughout the study period, peaking at 97.1% in 2018, while DTR rates followed similar trends, ranging from 65.2 to 96.1% ( Fig. 2).

Table III.

Distribution of culture sources for Acinetobacter baumannii isolates from ICU and non-ICU wards

Characteristic	Total samples (n=3628); n (%)	Ward		P value
		Non-ICU (n=1785); n (%)	ICU (n=1843); n (%)
Culture source
Respiratory	1834 (50.6)	761 (41.5)	1073 (58.5)	< 0.001
Wound samples	699 (19.3)	371 (53.1)	328 (46.9)	0.023
Urine	481 (13.3)	313 (65.1)	168 (34.9)	< 0.001
Blood	282 (7.8)	102 (36.2)	180 (63.8)	< 0.001
Other	332 (9.2)	238 (71.7)	94 (28.3)	< 0.001

Chi-square test. Other; includes CSF, deep abscess aspiration, tissue biopsies and synovial fluid

Table IV.

Antibiotic comparison: Non-susceptibility rates between ICU (n=1843) and non-ICU (n=1785) in Acinetobacter baumannii isolates

Group	Antibiotic	Tested (n)	Non-ICU non-susceptible (%)	ICU non-susceptible (%)	OR (95% CI)	P value
β-lactams	Ampicillin-sulbactam	1,553	560/736 (76.1)	699/817 (85.6)	1.87 (1.45–2.41)	<0.001
	Cefepime	3,322	1,350/1,639 (82.4)	1,551/1,683(92.2)	2.52 (2.02–3.12)	<0.001
	Ceftazidime	3,414	1,378/1,691 (81.5)	1,574/1,723 (91.4)	2.40 (1.95–2.95)	<0.001
	Imipenem	3,147	1,360/1,481 (91.8)	1,608/1,666 (96.5)	2.47 (1.79–3.40)	<0.001
	Meropenem	3,145	1,231/1,552 (79.3)	1,452/1,593 (91.1)	2.68 (2.17–3.32)	<0.001
	Piperacillin-tazobactam	1,221	512/598 (85.6)	591/623 (94.9)	3.10 (2.03–4.73)	<0.001
Aminoglycosides	Amikacin	2,397	754/1,174 (64.2)	909/1,223 (74.3)	1.61 (1.35–1.92)	<0.001
	Gentamicin	3,449	1,157/1,688 (68.5)	1,365/1,761 (77.5)	1.58 (1.36–1.84)	<0.001
	Tobramycinb	2,216	599/1,083 (55.3)	745/1,133 (65.8)	1.55 (1.31–1.84)	<0.001
Fluoroquinolones	Ciprofloxacin	3,399	1,372/1,668 (82.3)	1,611/1,731 (93.1)	2.90 (2.31–3.62)	<0.001
	Levofloxacin	2,598	971/1,269 (76.5)	1,144/1,329 (86.1)	1.90 (1.55–2.32)	<0.001

All P values were derived from Fisher’s exact test. ^bTobramycin data from 2016 were excluded from the model due to an anomalously low number of non-susceptible isolates. This discrepancy is likely attributable to the underreporting or misclassification of resistant results within the laboratory information system during that year, potentially reflecting inconsistent susceptibility testing or documentation practices.

Table V.

Logistic regression analysis of non-susceptibility trends over time for Acinetobacter baumannii in the ICU by antibiotic class

Group	Antibiotic#	Exp(B) (OR)	APC (%)	95% CI for Exp(B)	P value
β-lactams	Ampicillin-sulbactam	0.909	-9.1	0.829 – 0.998	0.044
	Cefepime	0.945	-5.5	0.894 – 0.999	0.045
	Ceftazidime	0.993	-0.7	0.944 – 1.045	0.794
	Ceftriaxone	1.085	+8.5	0.871 – 1.351	0.466
	Imipenem	1.109	+10.9	1.022 – 1.203	0.013
	Meropenem	0.976	-2.4	0.925 – 1.031	0.388
	Piperacillin-tazobactam	0.945	-5.5	0.845 – 1.058	0.325
Aminoglycosides	Amikacin	0.967	-3.3	0.932 – 1.004	0.084
	Gentamicin	0.785	-21.5	0.755 – 0.816	< 0.001
	Tobramycin	0.803	-19.7	0.769 – 0.839	< 0.001
Fluoroquinolones	Ciprofloxacin	0.870	-13.0	0.820 – 0.922	< 0.001
	Levofloxacin	0.811	-18.9	0.767 – 0.858	< 0.001
Tetracyclines	Minocycline	1.021	+2.1	0.922 – 1.131	0.689

^#Year is entered as an independent variable for each antibiotic

Fig. 1.

Temporal patterns of non-susceptibility to carbapenems, fluoroquinolones and aminoglycosides in ICU Acinetobacter baumannii (2012–2024).

Fig. 2.

Annual trends in CRAB and DTR non-susceptibility among ICU Acinetobacter baumannii isolates (2012–2024).

Evidence of bacterial translocation from non-blood sites to bloodstream

Among 3,800 Acinetobacter baumannii isolates analysed retrospectively, 30 cases (0.79%) showed likely translocation from non-blood sites to the bloodstream, based on matched MRNs and organism identity. The most common preceding source was the respiratory tract (18 cases, 0.47%), followed by urine and skin (4 cases each, 0.11%), other specimens (2 cases, 0.05%), and abscesses (2 cases, 0.05%). Notably, 60% of these events occurred in ICU patients, underscoring the vulnerability of this population to invasive infections.

Discussion

This 12-y surveillance study highlights the sustained burden of Acinetobacter baumannii resistance in a tertiary hospital in southern Saudi Arabia, particularly within ICU settings. Our findings confirm A. baumannii as the dominant species among Acinetobacter spp., accounting for over 96% of isolates and disproportionately affecting critically ill patients.

The high recovery from respiratory and blood specimens, especially in ICU patients, supports the pathogen’s association with ventilator-associated pneumonia and bloodstream infections in high-risk populations.^1,3 Consistent with national data, carbapenem resistance exceeded 90%, reinforcing the pathogen’s classification as a WHO Priority 1 ‘critical’ threat.^4,7

Resistance was significantly higher among ICU isolates across nearly all antibiotic classes. This finding is in line with established literature describing ICUs as epicentres for antimicrobial resistance due to high antibiotic pressure, frequent use of invasive devices, and prolonged patient stays.^3,5 Notably, resistance to last-line agents remains high, further limiting therapeutic options for CRAB infections. Our data show a worrisome rise in the resistance patterns, given the limited options for CRAB infections.

Aminoglycoside resistance in A. baumannii showed a two-phase decline initially after 2018 and more sharply post-2021, likely driven by reduced usage, stewardship interventions, and possible selection against less fit resistant strains. ICU-specific analysis showed increasing imipenem resistance, further narrowing therapeutic options. While CRAB rates remained high, a post-2019 decline in DTR phenotypes was observed. This may reflect indirect effects of the COVID-19 pandemic, including stricter infection control measures, reduced ICU admissions for non-COVID cases, and renewed focus on antimicrobial stewardship. These factors may have temporarily limited the transmission and selection of highly resistant pathogens.

Despite these statistically significant temporal declines in resistance to selected antimicrobial classes, the overall prevalence of CRAB and DTR phenotypes remained persistently high throughout the study period, indicating that these improvements may have limited clinical impact on treatment options in critically ill patients.

Resistance fluctuations likely reflect a combination of empirical prescribing patterns, lapses in antimicrobial stewardship, and clonal dissemination of high-risk lineages. International clones I and II, well-described in global literature, are frequently associated with CRAB and DTR phenotypes.^11,12 The lack of molecular data in this study precludes definitive clone attribution, but the epidemiologic pattern is consistent with clonal expansion events.

Environmental persistence of A. baumannii, combined with possible lapses in surface disinfection or infection prevention infrastructure, may contribute to ongoing transmission. Regional studies from the Aseer province, including reports by Almaghrabi et al,² have previously identified A. baumannii as a growing regional threat, particularly in bloodstream infections. Our findings expand on this work by offering detailed phenotypic and temporal analysis over more than a decade. Species-level comparisons confirmed significantly greater resistance in A. baumannii compared to non-baumannii species, reinforcing the clinical importance of species-level identification in routine diagnostics.

This study’s strengths include its long duration, large sample size, and ICU-specific analyses. The use of CLSI-defined DTR and CRAB phenotypes enhances interpretability and global comparability. However, the lack of molecular typing limits insights into resistance mechanisms and clonal dissemination, and clinical outcomes such as mortality and length of stay were not assessed. In addition, susceptibility to colistin and cefiderocol could not be evaluated because of laboratory constraints during the study period. Furthermore, this study did not systematically distinguish between colonisation and true infection, as individual clinical assessment (e.g., symptoms, imaging, clinical response) was beyond the scope of this retrospective laboratory-based surveillance; this may affect the interpretation of resistance rates, particularly for respiratory and wound isolates. Clinical severity scores such as APACHE II or SOFA were not available in our dataset, precluding correlation of CRAB/DTR phenotypes with infection outcomes or ICU length of stay; future prospective studies incorporating these parameters are warranted. Additionally, the non-baumannii isolates were identified as Acinetobacter spp. by the VITEK® 2 system but were not further speciated to individual species (e.g., A. lwoffii, A. pittii), which would require molecular methods such as rpoB gene sequencing and represents an area for future investigation.

Overall, Acinetobacter baumannii remains a major ICU pathogen in southern Saudi Arabia, with persistently high CRAB and DTR rates and ongoing resistance to last-line agents. Our findings also revealed significant temporal declines in resistance to several key antibiotics such as cefepime, aminoglycosides, and fluoroquinolones, suggesting the possible impact of antimicrobial stewardship or infection control interventions during the study period. These data underscore the urgent need for ICU-focused antimicrobial stewardship, targeted environmental decontamination, and molecular surveillance to monitor and control the spread of high-risk clones.

शोध-संदेश

दक्षिणी सऊदी अरब के आईसीयू (ICU) में एसिनेटोबैक्टर बाउमानी (Acinetobacter baumannii) जीवाणु में कई एंटीबायोटिक दवाओं के प्रति प्रतिरोध अब भी बहुत अधिक है। यह निष्कर्ष अस्पतालों में एंटीबायोटिक दवाओं के विवेकपूर्ण उपयोग, संक्रमण नियंत्रण और नियमित निगरानी को और मजबूत करने की आवश्यकता को रेखांकित करता है।

Author contributions

YS: Conceived and designed the study, developed the methodology, supervised the project, managed project administration, prepared the visualisations; YS, AAG, KB, ASo, AAB, AAS, TA, SAM, SH, AOSB: Contributed to investigation, resources, microbiological data acquisition and curation; YS, AJA, ASo, IMA: Data analysis, validation, interpretation; AJA: Contributed to funding acquisition. All authors contributed to manuscript writing. All authors have read and approved the final printed version of the manuscript.

Financial support and sponsorship

This work was supported by the Deanship of Research and Graduate Studies at King Khalid University through a Small Research Project under grant number (RGP.1/289/46).

Conflicts of Interest

None.

Use of Artificial Intelligence (AI)-Assisted Technology for manuscript preparation

The authors confirm that there was no use of AI-assisted technology for assisting in the writing of the manuscript and no images were manipulated using AI.