This retrospective cross-sectional study was conducted at Al-Mouwasat University Hospital, a tertiary academic center in Damascus, Syria, over a 24-month period (January 2022–December 2023). Data were extracted from archived microbiology records; no patient intervention was involved.

We included all patients with clinically suspected RTIs who had positive bacterial cultures from respiratory specimens meeting microbiological significance criteria (≥10 ⁵ CFU/mL for quantitative cultures). Eligible specimens comprised sputum, bronchial lavage, and bronchoalveolar lavage fluid. Only the first isolate per infectious episode was retained to avoid duplication bias. Specimens yielding only commensal flora, negative cultures, or incomplete susceptibility data were excluded.

Specimens were processed per standard protocols. Sputum adequacy was verified by Gram stain (<25 squamous epithelial cells/LPF). Significant isolates were identified using conventional biochemical panels. Mycobacterium tuberculosis (identified by acid-fast staining) was excluded from susceptibility testing; all non-mycobacterial isolates (n = 309) underwent full antimicrobial susceptibility testing (AST).

AST was performed by Kirby-Bauer disk diffusion on Mueller-Hinton agar (supplemented where appropriate) and interpreted per current Clinical and Laboratory Standards Institute (CLSI) breakpoints. ²⁰ A 20-agent panel spanning ten antibiotic classes was tested, including beta-lactams, carbapenems, fluoroquinolones, aminoglycosides, glycopeptides, oxazolidinones, and colistin. Isolates with intermediate susceptibility were classified as resistant for analytical purposes. ESBL production was phenotypically confirmed by double-disk synergy testing.

Multidrug resistance (MDR) was defined as acquired non-susceptibility to ≥3 antimicrobial categories, per standardized international criteria. ²⁰ Extensive drug resistance (XDR) and pan-drug resistance (PDR) were defined accordingly. The total number of resistant categories per isolate was also calculated.

Demographic, specimen, and microbiological data were manually abstracted and verified. Categorical variables were summarized as frequencies; continuous variables as means (±SD). Pathogen–specimen associations were assessed by chi-square (or Fisher’s exact) tests. Pairwise resistance correlations were evaluated using Spearman’s rank coefficient (ρ), with heatmap visualization. Analyses were performed in Python 3.12 and SPSS v26; p < 0.05 denoted statistical significance.

The Institutional Review Board of Al-Mouwasat University Hospital approved this study (approval reference provided). Patient informed consent was waived due to the retrospective, anonymized nature of the data. The study adhered to Declaration of Helsinki principles.

A total of 349 unique patients presenting with culture-confirmed respiratory tract infections (RTIs) were enrolled over a continuous two-year active surveillance period (January 2022 – December 2023) at Al-Mouwasat University Hospital, a tertiary academic referral center operating under conditions of sustained conflict-related resource constraint in Syria. This cohort, drawn exclusively from microbiologically verified infections at a single institution, provides a clinically representative and ecologically valid snapshot of the prevailing bacterial landscape in a high-burden, low-resource setting. The cohort demonstrated a pronounced male predominance, with 225 male patients (64.5%) versus 124 female patients (35.5%) ( Figure 1), yielding a male-to-female ratio of 1.81:1. This disparity is attributable to a convergence of epidemiological forces characteristic of conflict-affected populations, including heightened male exposure to conflict-associated trauma, the greater prevalence of smoking-related chronic obstructive pulmonary disease, and documented asymmetries in healthcare-seeking behavior within the Syrian socio-demographic context. Importantly, univariate sex-stratified analysis revealed no statistically significant association between patient sex and the isolation of any individual pathogen (χ ² range: 0.001–1.694; all p > 0.05), indicating that the observed gender imbalance reflects background epidemiological patterns rather than pathogen-specific host susceptibility differences.

Expectorated sputum specimens constituted the predominant sample type, accounting for 286 of 349 samples (82.0%), with the remaining 63 samples (18.0%) consisting of bronchial lavage and bronchoalveolar lavage (BAL) fluid ( Figure 2). This distribution is consistent with the diagnostic conventions of tertiary-care respiratory medicine, wherein non-invasive sputum collection represents the standard first-line approach and invasive lower-airway sampling is preferentially reserved for diagnostically ambiguous presentations or mechanically ventilated patients.

Microbiological analysis of the 349 culture-positive specimens identified twelve distinct bacterial species or genera, with a clear hierarchical pattern dominated by Gram-negative Enterobacteriaceae. The complete pathogen distribution is presented in Table 1 and illustrated in Figures 3 and 4.

Enterobacteriaceae collectively accounted for 201 isolates (57.6%), a proportion markedly exceeding those reported from high-income country surveillance networks, where atypical and Gram-positive pathogens typically dominate community-acquired pneumonia etiology. Enterobacter spp. emerged as the single most prevalent organism (n = 117; 33.5%), a finding that departs substantially from Western epidemiological norms and reflects the converging pressures of prolonged hospitalization, immunosuppression from comorbid conditions, prior broad-spectrum antibiotic exposure, and the systematic erosion of infection control infrastructure intrinsic to conflict-affected healthcare systems. Klebsiella pneumoniae ranked second (n = 84; 24.1%), a pathogen increasingly recognized as a sentinel species for the global emergence of carbapenem-resistant Enterobacteriaceae. Mycobacterium tuberculosis was identified in 40 specimens (11.5%), a prevalence substantially exceeding global incidence estimates for Syria (~127 per 100,000 population) and reflecting both the referral function of a tertiary-care center and the epidemiological consequences of prolonged conflict, mass population displacement, and the near-total disruption of national tuberculosis control programs. All M. tuberculosis isolates were excluded from antimicrobial susceptibility testing (AST), which was performed exclusively on the remaining 309 non-mycobacterial isolates. Gram-positive pathogens were considerably less prevalent. Staphylococcus aureus (n = 21; 6.0%) and Streptococcus pneumoniae (n = 17; 4.9%) constituted the most frequently isolated Gram-positive species, followed by Streptococcus pyogenes (n = 7; 2.0%) and Enterococcus spp. (n = 3; 0.9%). The relative underrepresentation of S. pneumoniae compared to high-income country data likely reflects the preponderance of healthcare-associated and hospital-acquired infections in this tertiary cohort. Eight isolates (2.3%) were phenotypically confirmed as extended-spectrum beta-lactamase (ESBL) producers, ²¹ a conservative estimate representing a minimum burden given the possibility that routine screening may not have captured all producers.

Antimicrobial susceptibility testing was performed on all 309 non-mycobacterial isolates using a standardized panel of twenty antimicrobial agents spanning ten pharmacological classes, with interpretation according to Clinical and Laboratory Standards Institute (CLSI) breakpoints. The resistance landscape was profoundly concerning across essentially all pathogen–antibiotic combinations, with the exception of colistin, which maintained meaningful—if diminishing—activity against most Gram-negative organisms. A heatmap summarizing overall resistance rates across major pathogens is presented in Figure 5.

Streptococcus pneumoniae (n = 17)

The susceptibility profile of S. pneumoniae in this cohort represents arguably the most extraordinary single finding of this investigation, constituting a departure from global surveillance benchmarks of exceptional magnitude. Universal resistance was documented to vancomycin (17/17; 100%) ( Table 2; Figure 6) and near-universal resistance to linezolid (16/17; 94.1%) — both agents representing the primary and secondary pharmacological pillars of treatment for beta-lactam-intolerant pneumococcal disease in contemporary practice globally. Macrolide resistance was similarly near-total (azithromycin/erythromycin/clarithromycin: 16/17; 94.1%). The practical consequence of this resistance profile is the therapeutic nullification of the two most established salvage regimens for serious pneumococcal infections in essentially all isolates from this collection.

Residual activity was confined exclusively to colistin (10/17 susceptible; 58.8%) and the aminoglycosides amikacin and gentamicin (9/17; 52.9%) — antibiotic classes with well-recognized pharmacodynamic limitations in the pulmonary compartment and not conventionally indicated as primary therapy for pneumococcal pneumonia. These findings mandate urgent molecular characterization (PCR-based resistance gene profiling; whole-genome sequencing) and confirmatory susceptibility testing by reference broth microdilution, as discussed in detail later.

Enterobacter spp. (n = 117)

As the single most prevalent pathogen in this cohort, the resistance profile of Enterobacter spp. carries disproportionate clinical weight in defining the therapeutic landscape at this institution. The detailed susceptibility data are presented in Table 3, with the corresponding graphical representation in Figure 7. Colistin retained the highest susceptibility rate at 69.2% (81/117), followed by aminoglycosides at 57.3% (67/117). Crucially, even these figures — representing the best available options — indicate that approximately one-third and two-fifths of isolates, respectively, are unresponsive to these last-resort and secondary agents.

Resistance rates across clinically critical antibiotic classes were severe: macrolides 76.1% (89/117), fluoroquinolones 70.1% (82/117), carbapenems (meropenem/imipenem) 60.7% (71/117), and beta-lactam combinations 52.1–71.8%. Carbapenem resistance approaching 61% in Enterobacter — a genus already prone to third-generation cephalosporin resistance through chromosomal AmpC beta-lactamase induction — indicates acquisition of higher-order resistance determinants, most plausibly carbapenemase enzymes of class B (metallo-beta-lactamases) or class D (oxacillinases), consistent with regional epidemiological trends.

Klebsiella pneumoniae (n = 84)

K. pneumoniae displayed a resistance trajectory comparable to or exceeding that of Enterobacter spp. Colistin demonstrated the highest susceptibility rate (61/84; 72.6%) ( Table 4; Figure 8), though this leaves more than one-quarter of isolates without an effective last-resort agent. Carbapenem resistance was documented in 62 isolates (73.8%) — a figure approaching rates previously reported only in hyperendemic settings such as certain tertiary hospitals in Greece, Italy, and the Arabian Peninsula, and representing a serious institutional alert. Macrolide resistance reached 85.7% (72/84) and linezolid resistance 81.0% (68/84), the latter likely reflecting intrinsic pharmacodynamic rather than acquired genetic mechanisms given the structural inactivity of oxazolidinones against Gram-negative organisms.

Staphylococcus aureus (n = 21)

Against the backdrop of near-universal multidrug resistance among Gram-negative organisms, S. aureus presented a comparatively more favorable susceptibility profile, as summarized in Table 5 and depicted graphically in Figure 9. Vancomycin retained activity in 85.7% of isolates (18/21), and aminoglycoside susceptibility was 81.0% (17/21). Fluoroquinolone susceptibility was approximately 60%, and macrolide resistance was documented in 52.4% (11/21) of isolates. However, the critical and globally alarming finding within this group was the detection of vancomycin resistance in 3 of 21 isolates (14.3%). Whether representing vancomycin-intermediate S. aureus (VISA) or fully vancomycin-resistant S. aureus (VRSA), a resistance rate of 14.3% would be unprecedented in any published institutional surveillance series and mandates immediate confirmatory testing by broth microdilution or E-test, alongside whole-genome sequencing to determine underlying resistance mechanisms (van gene acquisition versus thickened cell-wall-mediated MIC elevation).

Acinetobacter spp. (n = 14)

Acinetobacter spp. displayed the most extreme resistance phenotype among all Gram-negative pathogens, consistent with this genus’s global reputation for progressive therapeutic intractability. As shown in Table 6 and Figure 10, colistin was the sole agent maintaining reliable activity (12/14; 86.5%), while all other antibiotic classes — including carbapenems — demonstrated susceptibility rates not exceeding 10.8–29.7%. This profile, with an MDR rate of 100%, is consistent with extensively drug-resistant (XDR) or pan-drug-resistant (PDR) phenotypes increasingly reported from conflict-zone and under-resourced healthcare facilities across the Middle East and North Africa region.

Remaining pathogens

Among Haemophilus spp. (n = 8), universal macrolide resistance (100%) was observed. Moderate susceptibility was retained to tobramycin, imipenem, colistin, and meropenem (50% each), with other antibiotic classes demonstrating susceptibility rates of approximately 25%. All 13 Pseudomonas aeruginosa isolates (100%) were susceptible to aminoglycosides — a noteworthy finding that may reflect the comparatively limited prior therapeutic exposure of this organism to aminoglycosides at this institution. ESBL-producing isolates (n = 8) showed good susceptibility to aminoglycosides (7/8; 87.5%) and colistin (6/8; 75.0%), with near-universal resistance to all beta-lactam classes, as expected given their phenotype. Among the smaller groups, all three Enterococcus spp. isolates were fluoroquinolone-susceptible, and all three Mycoplasma spp. isolates were susceptible to fluoroquinolones, tazobactam, and tobramycin, with expected intrinsic resistance to cell-wall-active beta-lactams. A comparative overview of antibiotic resistance rates across the five key drug classes for all major respiratory pathogens in this cohort is presented in Figure 11, which underscores the consistently low colistin resistance among Gram-negative organisms relative to the near-universal resistance observed for macrolides and carbapenems.

Multidrug resistance (MDR), defined as acquired resistance to at least one agent in three or more distinct antimicrobial categories per international consensus criteria, ²⁰ was identified in 270 of 309 evaluable non-mycobacterial isolates (87.4%). This rate substantially exceeds most contemporary surveillance figures from Middle Eastern and North African tertiary-care hospitals — which typically range from 40–70% — and approaches those documented only in the highest-burden institutional settings globally. The mean number of antibiotic classes to which an isolate was resistant was 5.3 (standard deviation ±2.1), spanning a range of 0 to 10 classes. The breadth of resistance was particularly alarming: 81 isolates (26.2%) demonstrated resistance to nine antibiotic classes simultaneously ( Figures 13 and 14), and 16 isolates (5.2%) were resistant to all ten tested classes — a profile meeting criteria for pan-drug resistance (PDR) under international definitions. Conversely, complete susceptibility across all tested antimicrobials was observed in only 39 isolates (12.6%), an uncommonly low proportion that reflects the pervasive and sustained selective antibiotic pressure within this institution. The MDR burden varied substantially by pathogen. Acinetobacter spp. (100%), K. pneumoniae (96.4%), and Enterobacter spp. (91.5%) demonstrated the highest MDR prevalence among quantitatively significant groups. S. pneumoniae and P. aeruginosa both reached 100% MDR when classified by resistance to ≥3 antibiotic classes, while S. aureus (47.6%) and Mycoplasma spp. (33.3%) showed comparatively lower, though still elevated, MDR prevalence. These distributions are detailed in Table 7 and illustrated in Figure 12.

To characterize the structural architecture of resistance relationships across the isolate collection, Spearman’s rank correlation analyses (ρ) were performed between binary resistance profiles (resistant = 1; susceptible = 0) for all antibiotic pairs, applied to all non-mycobacterial isolates (n = 309) and to the Enterobacteriaceae subgroup (n = 201) specifically. Statistical significance was set at p < 0.05 (two-tailed). The complete correlation matrix for Enterobacteriaceae is presented in Figure 15 and Table 8.

Strongest Correlations: Perfect Intra-Class Co-Resistance (ρ = 1.000)

The strongest resistance associations identified across the entire dataset were uniformly intra-class co-resistances — concurrent resistance to multiple agents within the same pharmacological class. These yielded perfect Spearman correlations (ρ = 1.000; p < 0.001) without exception.

All beta-lactam agents (penicillin, amoxicillin) correlated perfectly with all cephalosporins (ceftriaxone, cefotaxime, cefuroxime; ρ = 1.000), indicating that resistance to any single agent within this spectrum predicts with absolute certainty resistance to all others. This finding is mechanistically coherent: a single genetic event — characteristically the acquisition of a broad-spectrum AmpC or ESBL enzyme — confers simultaneous resistance across the entire penicillin–cephalosporin spectrum, rather than through independent resistance acquisitions for each agent.

Within the fluoroquinolone class, ciprofloxacin and levofloxacin resistance correlated perfectly (ρ = 1.000), reflecting their shared mechanistic basis through gyrA/parC target mutations or efflux pump overexpression. Aminoglycoside co-resistance between amikacin and gentamicin, and macrolide co-resistance among azithromycin, erythromycin, and clarithromycin, similarly demonstrated ρ = 1.000. These perfect intra-class correlations confirm the mechanistic validity of the resistance data and are biologically anticipated from first principles.

Cross-Class Correlation: Ceftriaxone–Meropenem Co-Resistance (ρ = 0.72, p < 0.001)

The most clinically consequential cross-class correlation identified was between ceftriaxone (third-generation cephalosporin) and meropenem (carbapenem) resistance among Enterobacteriaceae isolates (n = 201; Spearman’s ρ = 0.72; p < 0.001). This strong positive association indicates that resistance to these two antibiotic classes ( Figure 16) — representing consecutive escalation rungs in the treatment ladder for serious Gram-negative infections — is largely co-selected or co-acquired in this population.

The clinical substrate of this correlation is made explicit by the contingency data: 92 of 201 Enterobacteriaceae isolates (45.8%) were resistant to both ceftriaxone and meropenem simultaneously, while only 44 isolates (21.9%) were susceptible to both. Twenty-seven isolates (13.4%) were ceftriaxone-resistant but meropenem-susceptible — consistent with a classic ESBL phenotype without carbapenemase co-production — while 38 (18.9%) were meropenem-resistant but ceftriaxone-susceptible, an atypical pattern potentially attributable to porin loss combined with pre-existing AmpC expression in the absence of an ESBL. All 8 phenotypically confirmed ESBL-producing isolates were simultaneously resistant to at least one carbapenem, placing them in the XDR category. From a therapeutic standpoint, this strong co-resistance association signals that the concurrent loss of both cephalosporins and carbapenems — leaving colistin and aminoglycosides as the principal remaining options — is not an exceptional event in this collection, but rather the predominant phenotypic pattern in nearly half of all Enterobacteriaceae isolates (45.8%).

Weakest Correlations: Pharmacological Independence of Colistin (ρ ≤ 0.13)

In direct contrast to the strong intra-class and cross-class correlations described above, the weakest resistance associations throughout the entire dataset consistently and reproducibly involved colistin. Colistin resistance showed essentially no meaningful correlation with fluoroquinolone resistance (ρ = 0.038; p = 0.508), cephalosporin resistance (ρ = 0.047; p = 0.411), beta-lactam resistance (ρ = 0.047; p = 0.411), or tetracycline resistance (ρ = 0.087–0.098; p > 0.05). Even with aminoglycosides (ρ = 0.117; p = 0.041) and carbapenems (ρ = 0.130; p = 0.023), while nominally achieving statistical significance owing to the large sample size, the correlation coefficients were biologically trivial.

This pharmacological independence of colistin resistance from all other resistance mechanisms is mechanistically interpretable at a molecular level. Colistin exerts its bactericidal activity through direct disruption of the lipid A component of lipopolysaccharide (LPS) in the Gram-negative outer membrane. Colistin resistance is mediated primarily through lipid A modifications — principally via plasmid-borne mcr gene-encoded phosphoethanolamine transferases, or chromosomal mutations in pmrA/B, mgrB, or phoPQ regulatory systems. These mechanisms are entirely distinct from the beta-lactamase enzymes, efflux pump systems, and target-site mutations that underlie resistance to all other tested antibiotic classes. This mechanistic orthogonality means that even isolates harboring the most complex pan-resistant phenotypes retain a probability of colistin susceptibility essentially equivalent to that of pan-susceptible isolates — a genuinely independent therapeutic window. The practical implication for empirical therapy is direct and clinically important: knowledge that a patient’s isolate is resistant to ceftriaxone, carbapenems, fluoroquinolones, and aminoglycosides provides no useful predictive information regarding colistin efficacy. Colistin susceptibility must be assessed individually for each isolate, and its preservation as a last-resort agent depends critically on strict stewardship — specifically, restricting its clinical use to confirmed cases in which no viable alternative exists, to prevent the dissemination of mcr-mediated colistin resistance across genetic backgrounds already harboring carbapenemase genes, a combination that would render affected isolates truly untreatable.

Chi-square analyses were conducted ( Table 9) to evaluate whether specific pathogens demonstrated statistically significant preferential associations with either sputum (n = 286) or bronchial lavage (n = 63) specimens — a finding that would carry direct implications for understanding topographic localization of infection and for optimizing diagnostic specimen selection strategies. For the majority of pathogens examined, no statistically significant associations with specimen type were detected (all p > 0.05). K. pneumoniae was isolated from 23.1% (66/286) of sputum samples and 28.6% (18/63) of bronchial lavage samples (χ ² = 0.87; p = 0.35). Acinetobacter spp., S. aureus, and S. pneumoniae all demonstrated non-significant specimen-type associations ( p = 0.19–1.00), suggesting broadly equivalent topographic distribution for these organisms. In contrast, Enterobacter spp. demonstrated a statistically significant preferential isolation from bronchial lavage specimens compared to sputum (47.6% vs. 34.6% of lavage and sputum samples, respectively; χ ² = 0.68–6.01; p = 0.014 in the original analysis). This finding is biologically plausible: as opportunistic Gram-negative pathogens, Enterobacter species are more commonly implicated in genuine lower respiratory tract infections — including ventilator-associated pneumonia and healthcare-associated bronchopneumonia — where invasive sampling is required for adequate diagnostic recovery, rather than in upper airway colonization detectable in expectorated sputum. This distribution suggests that bronchial lavage specimens may selectively capture clinically significant Enterobacter lower respiratory tract disease, with important implications for diagnostic sensitivity when the clinical presentation is consistent with lower-lobe involvement or ventilatory compromise.

Several resistance findings in this study transcend the boundaries of what is currently documented in the global literature. An explicit and rigorous critical appraisal of these findings is essential to differentiate genuine epidemiological novelty from potential methodological artifacts — a distinction with profound implications for how these data should be interpreted and acted upon.

Universal Vancomycin and Near-Universal Linezolid Resistance in S. pneumoniae. The documentation of 100% vancomycin resistance across all 17  S. pneumoniae isolates is, to our knowledge, without precedent in any peer-reviewed surveillance study in the global literature. Vancomycin has maintained near-universal bactericidal activity against S. pneumoniae since its introduction into clinical practice, and clinically confirmed vancomycin resistance in this species has never been reported. Similarly, 94.1% linezolid resistance contradicts all available global data. Three mechanistic explanations must be considered and rigorously evaluated: (i) a genuine clonal outbreak of S. pneumoniae harboring novel acquired resistance determinants (e.g., cfr or optrA for linezolid; van genes transferred from enterococci for vancomycin) would constitute a microbiological event of international significance; (ii) a systematic methodological error in disk diffusion testing — such as incorrect inoculum density, improper disk storage, or inadvertent application of CLSI breakpoints designated for other streptococcal species — could artifactually generate these results; or (iii) specimen misidentification, where optochin-susceptible colony morphology could mask the identity of phenotypically similar streptococcal species with different intrinsic resistance profiles. Confirmatory retesting by reference broth microdilution, molecular species identification by 16S rRNA sequencing, and resistance gene profiling by PCR or whole-genome sequencing are urgently required before these findings can be definitively interpreted.

Vancomycin-Resistant S. aureus (VRSA) at 14.3%. The three vancomycin-resistant S. aureus isolates (14.3%) documented in this series, if confirmed by reference methodology, would constitute a prevalence far exceeding any figure previously published in an institutional or national surveillance study. Global VRSA surveillance has documented only isolated cases in the United States, Iran, India, and a small number of other countries since the first confirmed report in 2002. A rate of 14.3% within a single institution would be globally exceptional. While the conflict-disrupted healthcare environment — characterized by deficient infection control, suboptimal antimicrobial stewardship, and cross-contamination risk — theoretically provides conditions conducive to VRSA emergence and nosocomial transmission, the possibility of a technical artifact, particularly MIC classification at the resistance/intermediate breakpoint boundary or E-test versus disk diffusion discordance, cannot be excluded and must be systematically investigated by gold-standard susceptibility assays and molecular characterization before clinical or public health decisions are based on this finding.

Across all Gram-negative pathogens examined in this study, colistin demonstrated the most consistent and highest susceptibility rates. Susceptibility ranged from 58.8% in S. pneumoniae to 86.5% in Acinetobacter spp., spanning 69.2% for Enterobacter spp. and 72.6% for K. pneumoniae. For the most therapeutically intractable pathogen in this collection — Acinetobacter spp. — colistin was functionally the sole active antibiotic, with all alternatives demonstrating susceptibility rates below 30%.

The statistical independence of colistin resistance from all other resistance mechanisms (ρ < 0.13 for all cross-class correlations involving colistin; none biologically meaningful) carries dual implications. Therapeutically, the absence of co-resistance means colistin remains a viable option even for XDR and PDR isolates that have exhausted all other antibiotic classes. Evolutionarily, it means that the intensive selection pressures that have driven resistance to beta-lactams, fluoroquinolones, aminoglycosides, and macrolides — the agents most heavily used in clinical practice — have not simultaneously co-selected colistin resistance through linked genetic mechanisms. This genetic independence is the principal reason why colistin has retained efficacy despite near-universal resistance to essentially every other antibiotic class in this collection.

However, this therapeutic window is both finite and actively eroding. Colistin resistance rates of 27.4–41.2% observed in K. pneumoniae and S. pneumoniae, respectively, signal progressive reserve attrition. The potential emergence of plasmid-borne colistin resistance mediated by mcr genes represents a direct and existential threat to this genetic independence, given the capacity of mcr to disseminate horizontally across diverse genetic backgrounds including those already harboring carbapenemase genes — a combination that would render affected isolates truly pan-drug-resistant and pharmacologically untreatable with currently licensed antibiotics. The preservation of colistin efficacy must therefore be regarded as a public health priority requiring strictly enforced stewardship protocols, mandatory susceptibility testing before each clinical use, and restricted formulary access.

Integrating the pathogen distribution data, individual susceptibility profiles, MDR burden quantification, and multivariable correlation analyses, a coherent and clinically actionable hierarchical architecture of resistance determinants emerges from this dataset.

The first and highest tier of resistance association comprises intra-class co-resistances (ρ = 1.000), which are mechanistically coupled through single-gene acquisition events. For prescribing practice, this means that demonstrated susceptibility to any single agent within a pharmacological class reliably predicts susceptibility to all agents within that class, and conversely, resistance to one agent predicts universal class-level resistance.

The second tier is constituted by the strong cross-class ceftriaxone–meropenem association among Enterobacteriaceae (ρ = 0.72; p < 0.001), reflecting the co-acquisition of ESBL and carbapenemase resistance determinants on mobile genetic elements. This strong correlation identifies a specific and clinically catastrophic resistance phenotype — the concurrent loss of both third-generation cephalosporins and carbapenems — that is already the majority phenotype among Enterobacteriaceae at this institution (45.8% of isolates).

The third and most clinically critical tier, paradoxically defined by the absence of correlation, involves colistin. The pharmacological independence of colistin from all other antibiotic classes (all cross-class ρ < 0.13; most p > 0.05) constitutes the single most operationally important finding from the correlation analysis. It means that colistin represents a genuinely independent therapeutic axis, mechanistically uncompromised by the same resistance pathways that have simultaneously disabled beta-lactams, fluoroquinolones, aminoglycosides, and macrolides in the majority of isolates from this collection.

From an institutional and public health perspective, the convergence of findings in this study — 87.4% overall MDR prevalence, near-exclusive Gram-negative dependence on colistin monotherapy, emergence of exceptional resistance phenotypes in both Gram-positive and Gram-negative key pathogens, and a conflict-constrained institutional environment — defines an acute antimicrobial resistance crisis demanding immediate and coordinated response. The implementation of comprehensive antimicrobial stewardship programs, systematic infection prevention and control bundles for high-risk ICU and mechanically ventilated patients, rigorous molecular surveillance incorporating whole-genome sequencing, and—where feasible—the adoption of novel therapeutic platforms (bacteriophage therapy, cefiderocol, novel beta-lactamase inhibitor combinations) are not discretionary institutional improvements but constitute an urgent collective imperative.

All analyses were conducted on primary institutional surveillance data (N = 349 isolates). Statistical methods employed: Spearman’s rank correlation coefficient (ρ) for pairwise antibiotic resistance associations; Pearson chi-square test for categorical specimen-type associations; descriptive statistics (means, standard deviations, frequencies, proportions) for demographic and pathogen distribution data. Significance threshold: p < 0.05 (two-tailed). Statistical computing: Python 3.12 (scipy, pandas, numpy, matplotlib). Isolates with incomplete susceptibility data were excluded from relevant pairwise correlation analyses.

The dataset can be found in the ZENODO repository, under the title “Excell Sheet 17” ( https://doi.org/10.5281/zenodo.18875489).

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0 or CC0).