Hospital-Based Surveillance and Resistance Index Analysis of Antimicrobial Resistance Trends: A Three-Year Study from a Tertiary Hospital in Iran (2021–2023)

Introduction

Antimicrobial resistance (AMR) has emerged as one of the gravest threats to global health, responsible for an estimated 4.95 million deaths in 2019, including 1.27 million directly attributable to drug-resistant bacterial infections.¹ The World Health Organization (WHO) has repeatedly warned that without urgent action, the world is heading toward a post-antibiotic era in which common infections may once again become lethal. Gram-negative pathogens particularly Acinetobacter baumannii (A. baumannii), Pseudomonas aeruginosa (P. aeruginosa), and members of the Enterobacterales are among the most critical organisms prioritized by WHO for new antibiotic development.²

Hospital-based surveillance data provide essential insights into local epidemiology and guide empiric therapy.³ Yet, despite global initiatives such as the WHO Global Antimicrobial Resistance and Use Surveillance System (GLASS), significant data gaps persist in low- and middle-income countries, including Iran.^4,5 Most available reports from Iran describe overall resistance percentages but rarely assess temporal trajectories of resistance within individual hospitals or introduce aggregate metrics that capture resistance dynamics beyond static antibiograms.^6,7

The COVID-19 pandemic further altered antibiotic prescribing practices worldwide, with a surge in broad-spectrum antibiotic use even in patients without confirmed bacterial infections. Although our dataset begins in 2021, understanding resistance trends in the post-COVID era remains critical for recalibrating empiric therapy and strengthening antimicrobial stewardship programs (ASPs) interventions.⁸

In addition to conventional resistance reporting, there is an increasing need for metrics that translate microbiological data into clinically actionable insights. In this context, the Resistance Index (RI) provides a composite measure of cumulative resistance burden across pathogens and antibiotics. When integrated with temporal trend analysis (eg, slope-based changes), RI can support practical decision-making by identifying high-risk pathogens, informing empirical therapy adjustments, prioritizing surveillance targets, and generating early warning signals for emerging resistance patterns. This approach facilitates a more structured integration of routine antibiogram data into ASPs and infection prevention and control (IPC) strategies. Overall, this analytic framework offers a pragmatic and scalable solution for hospitals particularly in resource limited settings to enhance early detection of resistance trends, optimize escalation or de-escalation strategies, and strengthen coordinated ASP and IPC interventions.⁹

Methods

Study Design and Setting

This retrospective, single-center study was conducted using microbiological data collected between January 2021 and December 2023. To avoid overrepresentation bias, only the first isolate per patient per pathogen within the study period was included in the analysis. Repeated isolates from the same patient were excluded unless they represented a different pathogen or a clinically distinct episode, where identifiable. Data were extracted from the microbiology laboratory information system of Shahid Hasheminejad Hospital in Mashhad, Iran. Antimicrobial susceptibility results (AST) were interpreted in accordance with the latest guidelines of the Clinical and Laboratory Standards Institute (CLSI).¹⁰ During the study period, IPC practices in the hospital included routine microbiological surveillance, implementation of contact precautions for patients infected or colonized with multidrug-resistant organisms, environmental cleaning and disinfection protocols, and adherence to standard infection control guidelines in high-risk units such as intensive care and surgical wards.

In addition, ASP activities were partially implemented, including guideline-based empirical therapy recommendations, routine microbiology reporting to clinical teams, and restricted use of selected broad-spectrum and reserve antibiotics. Prescription of last-line agents such as colistin and linezolid required approval from an infectious diseases specialist under the supervision of the institutional stewardship committee. However, no fully integrated, real-time electronic decision-support system for antimicrobial prescribing was available during the study period.

Specimen Collection and Processing

Specimens were processed according to established protocols from the Clinical Microbiology Procedures Handbook (5th ed.) and the Infectious Diseases Society of America (IDSA)/American Society for Microbiology (ASM) laboratory utilization guide.^11,12 Primary inoculation was performed on blood agar (Blood Agar Base with 5% sheep blood; Oxoid, Thermo Fisher Scientific, UK), MacConkey agar (HiMedia, India), and, when indicated, chocolate agar (BD BBL, USA). For susceptibility testing, Mueller–Hinton agar (Oxoid, UK) with a depth of 4±0.5 mm and pH 7.2–7.4 was used. Plates were incubated at 35 ± 2°C for 16–18 hours under ambient atmospheric conditions. Validity (internal validity) was assured through standardized workflows, clear acceptance criteria, and harmonized diagnostic algorithms. To evaluate reproducibility, approximately 10% of AST plates (~300) were independently re-read by a second microbiologist blinded to the original results. Interobserver agreement for S/I/R categorization was evaluated using weighted Cohen’s kappa (κ). The level of agreement was high (Po = 0.93; Pe = 0.50; κ = 0.86; 95% CI 0.81–0.91), indicating excellent reliability in line with WHO and CLSI quality assurance standards.^13,14

Bacterial Identification

Organisms were identified by Gram staining, colony morphology, and primary biochemical reactions (catalase, oxidase), with confirmatory testing by standard biochemical panels. Escherichia coli (E. coli) was identified as Gram-negative rods, lactose-fermenting, oxidase-negative, indole-positive, citrate-negative, and motile; Klebsiella pneumoniae (K. pneumoniae) as lactose-fermenting, oxidase-negative, indole-negative, urease-positive, citrate-positive, and non-motile; Enterobacter spp. as lactose-fermenting, oxidase-negative, and positive for amino-acid decarboxylases (LDC/ODC, species-dependent); P. aeruginosa as non-lactose-fermenting, oxidase-positive, pigment-producing (pyocyanin) with growth at 42°C; A. baumannii as non-lactose-fermenting, oxidase-negative, non-motile coccobacilli; and Staphylococcus aureus (S. aureus) as Gram-positive cocci in clusters, catalase-positive and coagulase-positive, with methicillin resistance determined by a cefoxitin 30 µg disk rather than oxacillin.¹⁵ Quality assurance included the routine use of reference strains to verify key phenotypes and test performance: E. coli ATCC 25922 and K. pneumoniae ATCC 700603 for Enterobacterales biochemical profiles, P. aeruginosa ATCC 27853 for oxidase and 42°C growth verification, A. baumannii ATCC 19606 for non-fermenting Gram-negative coccobacilli characteristics, and S. aureus ATCC 25923 for catalase, coagulase, and cefoxitin disk confirmation, in adherence to CLSI recommendations.¹⁰

Antimicrobial Susceptibility Testing (AST)

AST was performed using the Kirby–Bauer disk diffusion method on Mueller–Hinton agar in accordance with CLSI M02 guidelines. Inocula were standardized to a 0.5 McFarland suspension using a densitometer (BioSan, Latvia), inoculated by swabbing in three directions onto Mueller–Hinton agar plates (Oxoid, Thermo Fisher Scientific, UK), and incubated at 35 ± 2°C for 16–18 h. Inhibition zone diameters were measured with calipers and interpreted according to CLSI breakpoints.¹⁰

For Gram-negative isolates, the following antibiotic disks (Oxoid, Thermo Fisher Scientific, UK; 6 mm) were tested at standard potencies: ampicillin (AMP, 10 µg), ampicillin/sulbactam (SAM, 10/10 µg), piperacillin/tazobactam (TZP, 100/10 µg), ceftazidime (CAZ, 30 µg), ceftriaxone (CRO, 30 µg), cefepime (FEP, 30 µg), imipenem (IPM, 10 µg), meropenem (MEM, 10 µg), amikacin (AMK, 30 µg), gentamicin (GEN, 10 µg), ciprofloxacin (CIP, 5 µg), levofloxacin (LEV, 5 µg), trimethoprim/sulfamethoxazole (SXT, 1.25/23.75 µg), nalidixic acid (NAL, 30 µg), and nitrofurantoin (NIT, 300 µg). NAL and NI results were reported only for isolates obtained from the urinary tract. NAL was retained in this study as part of the routine antimicrobial susceptibility testing panel for urinary Enterobacterales (excluding Salmonella and Shigella), primarily for longitudinal surveillance and comparability with historical datasets. Its use does not reflect current therapeutic recommendations, and results should be interpreted strictly within an epidemiological and laboratory surveillance context rather than as guidance for clinical treatment.

For Gram-positive isolates, the panel included penicillin G (PEN, 10 U), cefoxitin (FOX, 30 µg; used for MRSA detection), erythromycin (ERY, 15 µg), azithromycin (AZM, 15 µg), clindamycin (CLI, 2 µg), and trimethoprim/sulfamethoxazole (SXT, 1.25/23.75 µg). Vancomycin (VAN) was tested using agar dilution according to CLSI.¹⁰

RI and Data Analysis

Resistance rates (%R) were calculated annually for each species–antibiotic combination. A composite resistance index (RI), defined as the mean %R across all antibiotics tested for a given species, was used to represent cumulative resistance pressure.¹² Temporal trends were evaluated using linear regression models applied to annual %R values (2021–2023) to estimate slope and statistical significance (p-values). Given the limited number of time points, slope estimates were considered exploratory and used primarily to describe the direction of change rather than to infer causality.

Pairwise comparisons between years (2021 vs 2022, 2022 vs 2023, and 2021 vs 2023) were performed using the Z-test for proportions. A two-tailed p-value < 0.05 was considered statistically significant. No adjustment for multiple comparisons (eg, Bonferroni or false discovery rate correction) was applied, as the analyses were exploratory and intended to generate hypotheses regarding emerging resistance patterns.

Statistical analyses were conducted using SPSS version 27 (IBM Corp., Armonk, NY, USA). Results are presented in tables and figures. Continuous variables are summarized as means ± standard deviations and categorical variables as frequencies and percentages.

Ethical Considerations

The study protocol was reviewed and approved by the Ethics Committee of Mashhad University of Medical Sciences, Mashhad, Iran (Approval No: IR.MUMS.IRH.REC.1403.052). Given the retrospective nature of the study, the requirement for informed consent was waived by the Ethics Committee. Strict confidentiality of patient information was maintained. All data were anonymized prior to analysis, and no patient identifiers were included. The study was conducted in accordance with the principles of the Declaration of Helsinki.¹⁶

Results

Distribution of Samples

Between 2021 and 2023, a total of 38,514 clinical specimens were processed, of which 3109 (8.1%) yielded positive bacterial growth. The distribution of specimen types varied significantly across the study period (p<0.001, Table 1). Urine specimens accounted for the majority but declined progressively from 71.7% in 2021 to 46.0% in 2023 (p<0.001). Cerebrospinal fluid (CSF) samples decreased sharply, from 8.5% in 2021 to 0.8% in 2023 (p<0.001). Blood specimens remained relatively stable (39.2% in 2021 vs 40.6% in 2023, p=0.421), whereas respiratory tract specimens increased from 3.4% in 2021 to 5.2% in 2023 (p=0.041). Other specimen types showed only minor fluctuations without statistical significance (6.8% vs 7.4%, p=0.648).

Table 1 Annual Distribution of Collected Specimens by Type (2021–2023)

Pathogen Distribution

The annual distribution of culture positivity differed significantly across the study period (Figure 1), rising from 7.1% in 2021 to 9.2% in 2022, before declining to 7.9% in 2023 (p<0.001). Similarly, the total number of specimens collected each year varied significantly (p<0.001).

Figure 1 Annual frequency of culture-positive samples (2021–2023). Bars represent the number of culture-positive samples with percentages of total specimens shown above each bar. Horizontal brackets indicate pairwise comparisons between years (Z-test). Statistical annotation: p <0.001 = ***, p <0.01 = **, p <0.05 = *, n.s. = not significant.

Among culture-positive isolates, E. coli remained the predominant pathogen but demonstrated a significant decline in relative frequency, decreasing from 45.0% in 2021 to 29.7% in 2023 (p=0.02). A. baumannii showed the opposite pattern, increasing steadily from 17.6% to 26.8% over the three years (p=0.03). K. pneumoniae displayed a rising trend from 13.5% in 2021 to 20.6% in 2023, although the change did not reach statistical significance (p=0.06). In contrast, the proportions of P. aeruginosa (9.4–12.2%), S. aureus (8.0–8.8%), and Enterobacter spp. (5.0–6.3%) fluctuated modestly across years without statistically significant variation (all p>0.05). Overall, significant temporal changes were confined to E. coli and A. baumannii, with other major pathogens remaining stable in distribution (Table 2).

Table 2 Annual Distribution of Culture Positivity and Pathogen Frequency (2021–2023)

Antimicrobial Resistance Patterns

Between 2021 and 2023, several statistically significant shifts in antibiotic resistance were observed across major pathogens. Among Gram-negative pathogens, E. coli remained the most frequently isolated species but demonstrated persistently high resistance to third-generation cephalosporins (CRO, CAZ, and FEP) and fluoroquinolones (CIP, LEV), with only a modest reduction in SXT resistance noted in 2023 (Figure 2A). K. pneumoniae showed a progressive increase in resistance, most notably to NAL, which rose from 44.4% in 2021 to 79.5% in 2023 (p=0.01), while resistance to carbapenems (IPM, MEM) remained stable at >50% (Figure 2B). A. baumannii displayed the most concerning profile, with CRO resistance increasing from 93.9% to 100% (p=0.03) and CIP resistance from 81.6% to 97.5% (p=0.01), culminating in near pan-resistance (Figure 2C). In contrast, P. aeruginosa resistance levels were relatively stable, with consistently high resistance to CAZ, CIP, and TZP, and no significant temporal trends in carbapenem resistance (Figure 2D). Among Gram-positive pathogens, S. aureus demonstrated persistent methicillin resistance, but a significant reduction in SXT resistance was observed, falling from 55.6% in 2021 to 0% in 2023 (p=0.03), while VAN susceptibility was retained (Figure 2E). Finally, Enterobacter spp. exhibited fluctuating resistance patterns, with a significant decline in PEN resistance from 100% in 2021 to 57.1% in 2023 (p=0.01), whereas resistance to cephalosporins and carbapenems remained elevated but stable (Figure 2F).

Figure 2 Heatmaps of antimicrobial resistance profiles across major bacterial pathogens (2021–2023). Each heatmap depicts annual resistance rates (%) for the antibiotics tested. Color intensity reflects the level of resistance, with darker green indicating lower resistance and yellow indicating higher resistance. (A) Escherichia coli; (B) Klebsiella pneumoniae; (C) Acinetobacter baumannii; (D) Pseudomonas aeruginosa; (E) Staphylococcus aureus; (F) Enterobacter spp.

Temporal Resistance Trends

Between 2021 and 2023, regression analysis of antimicrobial resistance patterns across major bacterial pathogens revealed heterogeneous temporal trends. In A. baumannii, multiple antibiotics demonstrated increasing resistance trends over time, although these patterns should be interpreted as descriptive changes across a limited three-year period rather than definitive long-term epidemiological trajectories. AMK resistance increased by +8.55% per year (p<0.001) and SAM by +11.2% per year (p<0.001). Additional significant rises were observed for FEP (+6.9%/year, p<0.001), CRO (+3.05%/year, p=0.001), CIP (+7.95%/year, p<0.001), GEN (+26.9%/year, p<0.001), IPM (+5.9%/year, p<0.001), LEV (+5.15%/year, p<0.001), TZP (+5.45%/year, p<0.001), and SXT (+9.3%/year, p<0.001), while other agents remained stable.

Enterobacter spp. showed a different pattern, with significant decreases across multiple drugs: AMP (−42.5%/year, p<0.001), CIP (−10.6%/year, p=0.008), ERY (−11.1%/year, p=0.001), LEV (−12.85%/year, p<0.001), PEN (−21.45%/year, p<0.001), and VAN (−13.75%/year, p=0.001). In contrast, NIT resistance increased significantly (+31.3%/year, p<0.001).

In E. coli, significant increases were detected for AMK (+3.95%/year, p<0.001), SAM (+9.45%/year, p<0.001), CZO (+6.9%/year, p<0.001), CAZ (+8.3%/year, p<0.001), GEN (+11.75%/year, p<0.001), and NIT (+15.5%/year, p<0.001). Conversely, significant decreases were found for AMP (−11.0%/year, p<0.001), CIP (−5.55%/year, p=0.003), IPM (−9.45%/year, p<0.001), and MEM (−3.85%/year, p=0.008).

K. pneumoniae displayed both increasing and decreasing resistance trends. A strong upward trend was observed for NAL (+17.55%/year, p<0.001), along with a modest increase in SXT (+4.4%/year, p=0.042). Significant decreases were identified for AMK (−24.35%/year, p<0.001), AMP (−37.5%/year, p<0.001), FEP (−10.2%/year, p<0.001), CAZ (−8.35%/year, p<0.001), CIP (−6.05%/year, p=0.009), IPM (−34.75%/year, p<0.001), MEM (−27.8%/year, p<0.001), NIT (−4.25%/year, p=0.006), and TZP (−18.1%/year, p<0.001).

In P. aeruginosa, CAZ resistance rose significantly (+7.95%/year, p=0.002), whereas significant decreases were found for GEN (−18.85%/year, p<0.001), IPM (−8.7%/year, p<0.001), and MEM (−9.8%/year, p<0.001). Other agents, including AMK, SAM, FEP, and LEV, remained stable, and CIP showed no directional change.

Finally, S. aureus exhibited pronounced reductions in AMK (−47.5%/year, p<0.001), FOX (−25.75%/year, p<0.001), NIT (−30.0%/year, p<0.001), and SXT (−27.8%/year, p<0.001). In contrast, significant increases were observed for CLI (+11.15%/year, p=0.006) and GEN (+33.25%/year, p<0.001), while resistance to AZM, CIP, ERY, and PEN remained unchanged (Table 3). In Enterobacter spp., E. coli, K. pneumoniae, P. aeruginosa, and S. aureus, both increases and decreases in resistance rates were observed for selected antibiotics. However, given the short observation window, these variations may reflect short-term or local fluctuations in resistance patterns rather than stable temporal trends. Therefore, all slope estimates are reported as exploratory indicators of directionality rather than confirmatory measures of long-term resistance evolution.

Table 3 Temporal Slopes of Antimicrobial Resistance Among Major Bacterial Pathogens (2021–2023)

Resistance Index

Analysis of organism distribution trends demonstrated pronounced shifts in pathogen frequencies between 2021 and 2023 (Figure 3A). E. coli was consistently the leading isolate but declined markedly from 45.0% in 2021 to 29.7% in 2023. In contrast, A. baumannii exhibited a continuous increase, rising from 17.6% to 26.8% during the same period, while K. pneumoniae also expanded from 13.5% to 20.6%. Enterobacter spp. showed a gradual decrease from 6.3% to 5.0%. P. aeruginosa remained relatively stable, fluctuating around 9–12%, and S. aureus maintained a steady distribution near 8–9% across all years.

Figure 3 Temporal dynamics of bacterial distribution and composite RI. (A) Organism distribution trends, showing the relative share of each major pathogen among total positive cultures during 2021–2023. (B) RI trends, representing mean resistance across all tested antibiotics for each species during 2021–2023.

RI trends revealed striking differences among species (Figure 3B). A. baumannii demonstrated the highest burden, with composite resistance exceeding 90% in all years, underscoring its critical role as a multidrug-resistant pathogen. P. aeruginosa also showed persistently high resistance, ranging from 68% to 79%. Moderate indices were observed for K. pneumoniae and E. coli (50–70%), though both exhibited gradual downward trends by 2023. Notably, S. aureus displayed a sharp reduction, with its RI decreasing from 68.1% in 2021 to 44.0% in 2023, whereas Enterobacter spp. remained relatively stable at intermediate levels (60–63%).

Discussion

Over the three-year surveillance period, clear shifts in the epidemiology of antimicrobial resistance were identified. E. coli remained the most frequently isolated organism but declined in relative frequency, while A. baumannii steadily increased and carried the highest cumulative resistance burden. P. aeruginosa showed consistently elevated resistance, and K. pneumoniae demonstrated progressive increases in resistance to several key agents. Together, these findings mirror global surveillance trends and highlight the sustained threat of multidrug-resistant (MDR) and extensively drug-resistant (XDR) Gram-negative bacilli, which are prioritized by the WHO as “critical” pathogens.²

Our results align with the 2024 IDSA guidance, which recommends carbapenems as the treatment of choice for serious infections caused by extended-spectrum β-lactamase-producing Enterobacterales (ESBL-E) and supports selective use of novel β-lactam/β-lactamase inhibitor (BL/BLI) combinations or cefiderocol for carbapenem-resistant Enterobacterales (CRE) and difficult-to-treat resistant P. aeruginosa (DTR-PA).¹⁷ Of particular relevance, although not tested in our local AST panel, sulbactam–durlobactam has recently been approved and shown superior efficacy to colistin in the treatment of carbapenem-resistant A. baumannii (CRAB), directly addressing the near pan-resistance observed in our cohort.¹⁸ These therapeutic advances highlight how local surveillance data can be translated into evidence-based ASP decisions.

From an antimicrobial stewardship perspective, the integration of regression slopes with composite resistance indices provides a structured and actionable framework for guiding empiric therapy. Rather than serving solely as descriptive metrics, these indicators can directly inform clinical decision-making. For example, pathogens demonstrating consistently high RI values (eg, >70%) or significant upward resistance slopes may warrant escalation of empirical regimens or restriction strategies, whereas declining trends may support safe de-escalation once susceptibility data become available. In this context, the RI functions as a practical tool for prioritizing high-risk pathogens and optimizing antibiotic selection.

In addition to guiding treatment decisions, this approach enables operational interventions at the institutional level. Rapid increases in resistance can serve as early warning signals, prompting targeted IPC measures, enhanced screening strategies, or focused stewardship interventions. Embedding RI- and AST-based alerts into routine microbiology reporting systems or electronic health records could further support real-time clinical decision-making and improve responsiveness to emerging resistance threats.

Equally important is the IPC perspective. Non-fermenters such as CRAB and DTR-PA are increasingly associated with environmental reservoirs particularly in intensive care settings. Recent studies have shown that targeted interventions, including sink removal, waterless patient care, and strict device management, may reduce transmission when implemented alongside standard contact precautions.^19,20 Integrating surveillance-derived signals into predefined IPC bundles may further enhance outbreak detection, strengthen prevention strategies, and improve resource allocation.

Beyond the ICU setting, the implementation of standardized IPC guidelines across all hospital wards is essential to reduce the transmission of multidrug-resistant organisms. Such measures may also indirectly reduce the need for escalation to last-line antimicrobial agents, thereby supporting ASP objectives and preserving the effectiveness of reserve antibiotics. In this context, hand hygiene and alcohol-based hand disinfection remain the cornerstone of infection prevention strategies and represent the most effective and universally applicable IPC measures across all clinical settings.

Methodological strengths of this study include the use of a standardized AST panel, species–drug level trend analysis with robust statistical validation, and the application of a composite RI. While the Drug Resistance Index (DRI) has been criticized for oversimplifying antibiotic effectiveness by combining resistance rates and consumption data,²¹ in our study it was applied only as a supplementary metric rather than a standalone outcome. The absence of local antibiotic utilization data precluded calculation of a weighted DRI, but the core analyses were anchored in established indices (%R, slope-based trends, and RI), ensuring clinical interpretability while minimizing the risk of misinterpretation.

To address these gaps, we propose an integrated AST–IPC framework (Figure 4) that transforms cumulative antibiogram data into real-time, actionable interventions. This framework operationalizes periodic data extraction in accordance with the CLSI M39 guideline,²² followed by calculation of resistance metrics, application of predefined thresholds for escalation or de-escalation, and linkage with guideline-based therapeutic recommendations. However, considerable variability persists in antibiogram preparation across institutions, underscoring the importance of CLSI M39 as a universal reference for standardization. Therefore, it is imperative that ASPs collaborate closely with prescribers and microbiology laboratories to ensure effective dissemination, education, and utilization of antibiograms. As third-party surveillance platforms become increasingly integrated with institutional electronic medical records (EMRs), opportunities arise to harness machine learning for the development of dynamic, real-time “smart antibiograms” capable of integrating longitudinal resistance data, MIC trends, and patient demographics to inform precision-guided antimicrobial therapy and infection control strategies.²³

Figure 4 Integrated roadmap for transforming AST data into actionable ASP–IPC strategies. Abbreviations: AST, antimicrobial susceptibility testing; CLSI, Clinical and Laboratory Standards Institute; RI, resistance index; DRI, dynamic resistance index; ASP, antimicrobial stewardship program; IPC, infection prevention and control; ESBL-E, extended-spectrum β-lactamase–producing Enterobacterales; CRE, carbapenem-resistant Enterobacterales; DTR-PA, difficult-to-treat resistant P. aeruginosa; CRAB, carbapenem-resistant A. baumannii; BL/BLI, β-lactam/β-lactamase inhibitor; ICU, intensive care unit.

Recent evidence further supports the integration of artificial intelligence into antimicrobial stewardship frameworks. A recent meta-analysis demonstrated that AI- and machine learning based models can outperform traditional risk scoring systems in identifying resistant pathogens and optimizing antimicrobial therapy, particularly in terms of sensitivity and negative predictive value.²⁴ In addition, AI-driven approaches have shown significant potential in the management of infectious diseases in vulnerable populations, such as elderly and multimorbid patients, by integrating clinical, microbiological, and patient-specific factors into personalized decision-making processes.²⁵ These advances highlight the potential for combining RI-based surveillance metrics with AI-supported clinical decision-making systems to further enhance the precision and impact of antimicrobial stewardship and infection control strategies.

A limitation of this study is the relatively short observation period (three years), which limits the robustness of time-series analyses. In particular, regression-based slope estimates derived from only three data points should be interpreted with caution, as they may be sensitive to year-to-year fluctuations and may not fully capture long-term resistance trajectories. Furthermore, the lack of formal adjustment for multiple comparisons increases the risk of type I error; therefore, statistically significant findings should be interpreted as exploratory and hypothesis-generating rather than confirmatory. Further research is warranted to identify optimal strategies for integrating cumulative antibiogram data into clinical workflows and to elucidate their direct, measurable impact on patient outcomes.

Conclusion

In conclusion, this study demonstrates that integrating temporal trend analysis with a composite RI provides a scalable, evidence-based framework for transforming cumulative microbiology data into more actionable epidemiologic insights. The persistently high resistance observed in A. baumannii and the sustained high resistance of P. aeruginosa underscore the need for careful reassessment of empiric therapy protocols and reinforcement of IPC measures. Conversely, the observed downward resistance trends in E. coli and K. pneumoniae highlight potential opportunities for more data-informed de-escalation strategies. Building upon these findings, the implementation of an integrated ASP–IPC model, aligned with CLSI M39 principles and supported by real-time “smart antibiograms”, may enable improved linkage between resistance surveillance, antimicrobial utilization, and clinical decision-making. Expanding surveillance beyond a three-year window and incorporating antibiotic consumption metrics into a localized DRI could further enhance predictive accuracy and strengthen stewardship strategies for antimicrobial resistance containment.