Priority assessment of antimicrobial-resistant pathogens in the Republic of Korea: a multicriteria decision analysis

Summary

Background

Priority assessment of antimicrobial-resistant pathogens is essential for resource allocation and national risk management. This study aims to establish a priority list of antimicrobial-resistant pathogens to strengthen the national response to antimicrobial resistance (AMR) in the Republic of Korea (ROK).

Methods

Antimicrobial-resistant pathogens were ranked throughout a multicriteria decision analysis using a preference-based survey based on PAPRIKA (Potentially All Pairwise Rankings of all possible Alternatives) method. This process involved identifying relevant criteria to evaluate pathogen priority and collecting data on disease burden for each antimicrobial-resistant pathogen.

Findings

A literature review was conducted to develop a candidate list of antimicrobial-resistant pathogens (n = 56) and risk assessment criteria (n = 14). Seventeen pathogens and eight criteria were selected via a Delphi survey of experts (n = 31). Incidence, case fatality, five-year trend of AMR, and nonfatal health burden was set as quantitative indicators, while transmissibility, preventability, treatability, and antibiotic development pipeline were qualitative indicators. The pathogens were ranked using 56 expert responses from the PAPRIKA survey, and categorized them into three priority groups: emergency response, preferred management, and monitoring. The emergency response group included carbapenem-resistant (CR) Klebsiella pneumoniae, CR Acinetobacter baumannii, and vancomycin-resistant Enterococcus faecium.

Interpretation

This priority list represents the first comprehensive assessment of AMR pathogens of greatest concern to the Korean healthcare system. Our findings may inform country-specific policy decisions, particularly regarding infection prevention and control and AMR research and development.

Funding

This work was supported by a fund (Grant No. 2024ER220500) provided by the Korea National Institute of Health.

Research in context.

Evidence before this study

We searched PubMed, Embase, and KoreaMed, as well as reference lists of relevant articles and grey literature, for studies published from Jan 1, 2010, to Jan 31, 2025, without language restrictions, using combinations of terms related to “antimicrobial resistance”, “priority pathogens”, “burden”, and “multicriteria decision analysis”.

Previous studies have estimated the global and regional burden of antimicrobial resistance using systematic analyses of surveillance data, hospital datasets, and modelling approaches. These studies consistently show that antimicrobial resistance is a major contributor to mortality and disability worldwide, with substantial heterogeneity across regions and pathogens. Key pathogens contributing to the burden include Escherichia coli, Staphylococcus aureus, Klebsiella pneumoniae, Streptococcus pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa, although the relative contribution of each pathogen varies by region and healthcare context.

Although these studies provide important estimates of disease burden, they are often based on aggregated modelling approaches and are not designed to directly inform prioritisation of pathogens for national-level policy decision making. In particular, existing evidence is limited in its ability to integrate multiple dimensions of risk—including epidemiological burden, transmissibility, preventability, treatability, and future antibiotic pipeline—within a structured and transparent prioritisation framework. Furthermore, country-specific prioritisation efforts that incorporate local surveillance data and healthcare system context remain scarce, particularly in the Western Pacific region.

Added value of this study

This study provides a national priority assessment of antimicrobial-resistant pathogens in the Republic of Korea using a multicriteria decision analysis framework that integrates systematically collected evidence, national surveillance data, and expert consensus. By combining quantitative indicators (including incidence, mortality, non-fatal burden, and resistance trends) with qualitative assessments (including transmissibility, preventability, treatability, and antibiotic development pipeline), this study offers a comprehensive and context-specific prioritisation.

The use of a transparent weighting process based on the PAPRIKA method, along with sensitivity analyses, enhances the robustness, reproducibility, and interpretability of the prioritisation framework. This study represents one of the few applications of a structured, data-informed prioritisation approach at the national level in the Western Pacific region.

Implications of all the available evidence

The findings of this study provide an evidence-based framework to inform national public health strategies for antimicrobial resistance, including prioritisation of surveillance activities, infection prevention and control policies, antimicrobial stewardship, and research and development investments. When considered alongside existing global prioritisation efforts, these results highlight the importance of adapting priority pathogen lists to local epidemiological contexts to ensure efficient and targeted use of limited resources.

Implementation of such context-specific prioritisation frameworks may enhance the effectiveness of national AMR responses by aligning public health investments and interventions with the highest-risk pathogens. Future research should focus on updating prioritisation with more comprehensive data and evaluating its impact on policy implementation and health outcomes.

Introduction

Antimicrobial resistance (AMR) has emerged as a leading threat to global public health in the 21st century.¹ Initiatives designed to combat AMR are currently being implemented at local, national, and international levels. At the global level, the World Health Organization (WHO) and the World Organization for Animal Health have proposed a framework to optimize antimicrobial use and standardize surveillance programs for antimicrobial-resistant pathogens.² Following the WHO’s endorsement of the Global Action Plan on AMR in 2015, member states have been urged to develop their own country-specific, One Health-based national action plan on AMR.³

Given the complexity and global scale of AMR, limited resources should be allocated based on prioritization of antimicrobial-resistant pathogens of concern to strengthen public health policies and improve implementation of policy-based strategies. The WHO and the Centers for Disease Control and Prevention (CDC) have announced updated priority lists of antimicrobial-resistant pathogens to support antibiotic research and development and AMR containment strategies.^4,5 Although these priority lists are important contributions to the existing knowledge on AMR, they are insufficient for fully understanding local AMR burdens and identifying and targeting the highest-priority pathogens in various national contexts.

Public health decision making is complex and typically involves trade-offs between multiple, often conflicting, values and strategic management of finite resources. To address these multifaceted challenges, multiple criteria decision analysis (MCDA), an analytical technique that structurally integrates data-driven objective indicators with expert judgments, offers a convenient and systematic approach to optimal prioritization.⁶ Notably, the Potentially All Pairwise Rankings of all Possible Alternatives (PAPRIKA) technique, designed to derive the weights of each criterion through intuitive and clear dichotomous choices, has been used as an alternative that overcomes the limitations of existing MCDA methodologies, which can cause cognitive overload for respondents or produce inconsistent results.⁷

This study was conducted to develop a priority list of antimicrobial-resistant pathogens in the Republic of Korea (ROK) based on expert opinion and current evidence, thereby providing the basis for the development and implementation of a national action plan on AMR.

Methods

In 2017, the WHO developed the Bacterial Priority Pathogens List (BPPL) to guide antibiotic R&D and inform AMR surveillance and control.⁸ The updated methodology of the 2024 WHO BPPL proved reproducible and applicable to our systematic prioritization of AMR risks in the ROK.⁴ We additionally incorporated analyses of national AMR surveillance data from the Korean CDC and clinical data collected from five medical institutions. Prioritization of antimicrobial-resistant pathogens followed a multistage process: pathogen candidate selection, criteria selection and definition, evidence extraction and data synthesis, weighting of the criteria, scoring and ranking of the pathogens, sensitivity analysis of criteria weighting, and expert review⁹ (Supplementary Fig. S1). While the 2024 WHO BPPL focused on pathogens requiring novel antibiotics, this assessment aimed to provide a broader tool to support national AMR policy. The intended use of the priority list was not formally predefined at the outset, and the scope was refined as the study progressed.

Pathogen selection

A total of 62 pathogen-antibiotic resistance combinations were identified based on prioritized lists of multidrug-resistant pathogens provided by the WHO, CDC, the Europe CDC, and Public Health Agency of Canada (PHAC).^4,5,9,10 The candidate pool was derived from the WHO BPPL, WHO Fungal Priority Pathogen List (FPPL), CDC, ECDC, and PHAC prioritizations, which collectively address antimicrobial resistance in bacterial and fungal pathogens. Antiviral resistance was not considered in our list as diagnostic accessibility is low and data are relatively limited. A candidate list of 56 pathogens was then refined through a Delphi survey of a multidisciplinary expert panel comprising infectious disease specialists, laboratory medicine physicians, microbiologists, veterinarians, epidemiologists, and public health professionals (Supplementary Data S1). Notably, pathogen selection was conducted prior to any data collection to minimize the influence of data source availability on the candidate pathogen list. We conducted a two-round online Delphi survey via email for pathogen selection. In addition to the initial 56 candidate pathogens reviewed in the first round, seven additional pathogens and resistance phenotypes proposed by the expert panelists during the first-round survey were incorporated into the second-round Delphi assessment (Supplementary Data S1).

Selection and definition of prioritization criteria

The following 14 criteria were identified based on previous reports from the WHO, CDC, ECDC, and PHAC: mortality, case fatality, incidence, prevalence, 10-year trend of AMR, nonfatal health burden, transmissibility, preventability, treatability, antibiotic development pipeline, public/political attention, clinical impact of untreated infections, potential sequela, and economic burden.^4,5,9,10 The members of the working group selected eight criteria suitable for the Korean context: incidence, case fatality, five-year trend of AMR, nonfatal health burden, transmissibility, preventability, treatability, and antibiotic development pipeline. The applicability of these risk assessment criteria in the ROK and the need for additional criteria were discussed based on expert opinions collected through a Delphi survey.

The operational definitions, data sources, and scoring rules for each of the eight criteria are provided in Table 1 and Supplementary Data S2. The scoring system for each criterion was established by one of the eight working group members. Each scoring system was validated by a second working group member, with disagreements discussed among all the members. The entire working group reached a consensus on any issues after the validation stage. Notably, case fatality was operationalized as all-cause in-hospital mortality among patients with an infection caused by the target pathogen, rather than as mortality directly attributable to AMR pathogens. For preventability, the effectiveness of each measure was graded based on the documented evidence of its ability to control the transmission of the target pathogen, drawn from relevant guidelines and literature, and finalized by the working-group consensus (Supplementary Data S2). The criterion of the antibiotic development pipeline captures the forward-looking dimension of prioritization, namely the expected future availability of therapeutic options, in contrast to preventability reflecting the current preventive capacity.

Table 1.

Prioritization criteria, definitions, and scoring system.

Criterion	Definition	Scoring system	Survey score	Preference value
Case fatality	All-cause in-hospital mortality among patients with infection caused by antimicrobial-resistant pathogens	>30%	High	18.24
		21–30%	Medium–High	14.59
		11–20%	Medium	10.94
		5–10%	Medium–Low	5.47
		<5%	Low	0
Incidence	Incidence of cases per 1 million population	>10,000 cases per 1 million population	High	13.83
		5001–10.000 cases per 1 million population	Medium–High	11.05
		1001–5000 cases per 1 million population	Medium	8.28
		100–1000 cases per 1 million population	Medium–Low	4.14
		<100 cases per 1 million population	Low	0
Non-fatal health burden	Years lived with disability (YLD) attributable to infections caused by each antimicrobial-resistant pathogen	>10.00 YLD per 1 million population	High	14.38
		1.01–10.00 YLD per 1 million population	Medium–High	11.58
		0.51–1 YLD per 1 million population	Medium	8.79
		0.11–0.5 YLD per 1 million population	Medium–Low	4.39
		<0.1 YLD per 1 million population	Low	0
Trend of resistance	5–10-year trend of resistance (proportion of resistant isolates among all identified pathogens). Trend data were obtained from systematic literature review, Kor-GLASS surveillance data, KARMS data (secondary care and long-term care hospitals), and tertiary hospital clinical microbiology datasets.	Increasing trend in ≥3 indicators	Level 5	14.10
		Increasing trend in 2 indicators	Level 4	10.96
		Increasing trend in 1 indicator	Level 3	7.83
		Stable trend across all indicators	Level 2	3.92
		Decreasing trend in ≥1 indicator without increasing trends in the other indicators	Level 1	0
Transmissibility	Evidence of transmission of the antimicrobial-resistant pathogen across multiple pathways: ①Human-to-human transmission: person-to-person transmission in healthcare and community settings, including outbreak potential ②Intersectoral transmission: transmission between humans and the animal, food, and environmental sectors	Scores were assigned based on the combination of outbreak capability (OC) and transmission potential (TP). See footnotea	High	16.41
			Medium-High	12.79
			Medium	9.15
			Medium–Low	4.58
			Low	0
Preventability	The existence and effectiveness of preventive strategies designed to reduce transmission of the target antimicrobial-resistant pathogen and the associated disease burden.	IPC measures •Effective and sufficient for controlling transmission of the target pathogen (+2 points) •Recommended, and in use, but effectiveness is partial or applicability is limited in certain settings or populations (+1 point) •Not universally recommended due to limited efficacy/feasibility (0 point)	>5 points: High	0
		Decolonization/Chemoprophylaxis •Existing and effective for the general population (+2 points) •Existing and partly effective but restricted to high-risk populations (+1 point) •Not existing or noneffective (0 point)	5 points: Medium–High	1.93
			4 points: Medium	3.84
		Public health interventions in community settings •Existing and effective OR not needed (+2 points) •Existing and partly effective (+1 point) •Not existing or noneffective (0 point)	3 points: Medium–Low	4.93
			<3 points: Low	6.02
Treatability	Availability and quality of antibiotic treatment options for infections caused by the target antimicrobial-resistant pathogen (limited to agents approved and available for clinical use in Korea)	Number of first-line option(s) recommended by guidelines (+2 points for each option)	≥12 points: High	0
		Availability of alternative option(s) for the most typical infectious syndrome •Number of options available or options available but with poor toxicity profiles or recommended only in combination (−1 point for each option) •Option(s) with poor toxicity profiles (0 point for each option) •At least one alternative with acceptable toxicity profile (1 point for each option)	10–11 points: Medium–High	2.89
		Availability of oral option(s) (+1 point for each option)	8–9 points: Medium	5.78
		Availability of OPAT option(s) (+1 points for each option)	6–7 points: Medium–Low	7.69
		Available option(s) approved or tested for pediatric populations (+1 points for each option)	≤5 points: Low	9.62
		National insurance-based accessibilityb (+1 points for each option)
Pipeline	The extent to which the global antibiotic pipeline is expected to meet the therapeutic needs for each target antimicrobial-resistant pathogen over the next 5–7 years, including agents not yet introduced in Korea (based on data from the 2023 WHO Antibacterial Agents in Clinical Development report, ClinicalTrials.gov, and grey literature sources)	Number of newly approved antibiotics and candidates against the target antimicrobial-resistant pathogen •phase 1 (+1 point for each candidate) •phase 2 (+2 points for each candidate) •phase 3 (+3 points for each candidate) •Ongoing market authorization application and/or new drug application (+4 points for each candidate) •Newly approved (2018–April 2025) (+5 point for each candidate)	>47 points: Likely	0
			47–34 points: Possible	5.32
			<34 points: Unlikely	7.39

IPC, infection prevention and control; KARMS, Korean Antimicrobial Resistance Monitoring System; Kor-GLASS, Korean Global Antimicrobial Resistance Surveillance System; OC, outbreak capability; OPAT, outpatient parenteral antibiotic; TP, transmission potential (between human and animal, food, and environmental sectors); YLDs, years lived with disability.

Note: For preventability, treatability, pipeline scores were calculated by summing all applicable components for each pathogen and categorized according to cutoffs. Treatability assessment was restricted to antibiotic agents approved and available for clinical use in the Republic of Korea. Pipeline assessment included agents approved, under regulatory review, or in clinical development between 2018 and April 2025.

^aOC (outbreak capability): Well documented = sustained outbreak records in multiple settings; Moderately documented = limited outbreak reports; Poorly documented = no or anecdotal evidence. Outbreak capability was assessed by combining domestic and global outbreak data to reflect both the Korean context and available international evidence (Supplementary Data 2.5). TP (transmission potential): High = documented intersectoral transmission; Moderate = possible intersectoral transmission; Low = primarily human-to-human only. Score categories by OC/TP combination: Well documented/High = High; Well documented/Moderate or Moderately documented/High = Medium–High; Poorly documented/High, Well documented/Low, or Moderately documented/Moderate = Medium; Moderately documented/Low or Poorly documented/Moderate = Medium–Low; Poorly documented/Low = Low.

^bAccessibility based on National Health Insurance reimbursement status; agents with indication-restricted reimbursement were considered non-reimbursable outside the approved indication.

Evidence extraction and data synthesis

Baseline data for the eight selected criteria were collected to determine the disease burden for each antimicrobial-resistant pathogen. Incidence, case fatality, five-year trend of AMR, and nonfatal health burden were set as quantitative measures, while transmissibility, preventability, treatability, and antibiotic development pipeline were defined as qualitative measures.

For quantitative indicators, data were collected through systematic review of published literature, supplemented with national surveillance data from the Korea Global Antimicrobial Resistance Surveillance System (Kor-GLASS) and Korean Antimicrobial Resistance Monitoring System (KARMS) covering 2019–2023, electronic medical records from five tertiary academic medical centers (2015–2024), or expert input from working group members^11,12 (Supplementary Table S2.0). Data generation periods varied by pathogen and data source; for pathogens where laboratory detection was recently established, data were limited to the period from which reliable detection was possible. For qualitative indicators, data were collected through literature search of peer-reviewed and grey literature, and the scoring framework was inherently structured such that limited or absent evidence resulted in a lower score, as each category was anchored to the presence and strength of documented evidence.

ROK-specific factors were incorporated into the evidence framework according to indicator type. The four quantitative criteria were scored based on numerical intervals derived from ROK-specific published literature, national surveillance data, and EMR data. The four qualitative criteria were scored on ordinal scales based on systematic searches of PubMed, Embase, the Cochrane Library, and ClinicalTrials.gov, reflecting intrinsic pathogen characteristics. For treatability, only agents approved and available in the ROK were included.

Weighting of criteria

In scoring each pathogen, each criterion was assigned a specific weight to reflect its relative importance to risk prioritization in Korean perspective. The final weights of the assessment criteria were determined through consensus of the AMR experts, using a blinded preference-based survey designed with the 1000 minds software according to the PAPRIKA method.¹³ The authors convened a panel of multidisciplinary experts categorized into infectious disease specialists, non-infectious disease specialist physicians, and non-clinical experts (including microbiologists, infection control nurses, AMR researchers, and public health policymakers) with expertise in AMR.

The respondents answered intuitionally to a chain of simple questions based on their expertise. One question addressed only two criteria for two pathogens at a time, involving a trade-off while keeping other criteria constant. Pathogen names were blinded to decrease bias and enhance reliability.

Scoring and ranking of the pathogens

Using a weighted sum approach for ranking, the score for each criterion was multiplied by the weight assigned to that criterion and summed across all the criteria to calculate the total risk score, which ranged from 5 (lowest priority) to 95 (highest priority). Pathogens were classified into three tiers based on a combination of score distribution, within-group risk profile homogeneity, and practical resource allocation considerations. The final priority list of the antimicrobial-resistant pathogens was compared with the 2024 WHO BPPL.⁴

Agreement and sensitivity analysis

For agreement analyses, respondents were stratified by institutional affiliation and years of clinical and research experience, as these variables were hypothesized to explain heterogeneity in weighting preferences. Agreement was assessed using coefficients of variation (CVs) and Kendall’s W. The CV is the relative ratio of the standard deviation to the mean value, indicating the degree of dispersion or concentration of expert responses around the mean. A lower CV indicates relatively more similarity in expert perceptions, whereas a higher CV indicates a relatively greater diversity of perspectives and interpretations of a given criterion. Kendall’s W, a value between 0 and 1, measures the degree of agreement in ranking criteria within a specific group of experts. A value closer to 1 indicates that a specific group of experts share similar judgments about the relative priority of each criterion.

A sensitivity analysis was performed to consider other weightings. The analysis was conducted to evaluate differences in priority assessment perspectives between clinical experience-focused and research experience-focused groups. The impact of years of professional experience on the total risk score for each pathogen was also evaluated.

Ethical approval and consent to participate

This study was approved by the Institutional Review Board of Korea University Anam Hospital (IRB No. 2026AN0105) on March 13, 2026. Written informed consent was obtained from all participants before their participation.

Funding

This work was supported by a fund (Grant No. 2024ER220500) provided by the Korea National Institute of Health. The funding source had no role in the study design, data collection, data analysis, data interpretation, decision to publish, or manuscript preparation.

Results

Pathogen selection and prioritization criteria

Of the 56 candidate pathogens evaluated in the first-round Delphi survey (n = 38; response rate: 61.3%), 14 met the CVR threshold (≥0.30) included as candidate list. In the second round (n = 31), three additional pathogens met the CVR threshold (≥0.33) bringing the final candidate list to 17 pathogens selected for prioritization (Table 2). Following expert review, eight criteria were selected for inclusion in the MCDA framework (Table 1). Candida auris, recently reclassified as Candidozyma auris, was retained under its former name throughout the analyses and manuscript to ensure consistency with current surveillance and indexing.¹⁴

Table 2.

Summary of the assessment of each pathogen against the eight criteria using the collected evidence.

Tier	Final Risk Score (SD)	Pathogen	Case fatality	Incidence	Non-fatal health burden	Trend of resistance	Transmissi-bility	Prevent-ability	Treatability	Antibiotic development pipeline
1	85.94 (6.35)	Klebsiella pneumoniae, carbapenem-resistant	High	Medium–High	Medium–High	Level 5	High	Medium–Low	Low	Likely
1	80.58 (7.02)	Acinetobacter baumannii, carbapenem-resistant	High	High	Medium–Low	Level 3	High	Medium–Low	Low	Possible
1	79.87 (7.07)	Enterococcus faecium, vancomycin-resistant	High	Medium–Low	Low	Level 3	Medium	Medium–Low	Low	Unlikely
2	70.36 (8.38)	Pseudomonas aeruginosa, carbapenem-resistant	Medium-High	Medium–High	Low	Level 4	Medium-High	Low	Low	Possible
2	69.66 (10.39)	Candida auris, antifungal-resistant	High	Medium–Low	Low	Level 3	High	Low	Low	Unlikely
2	65.05 (10.43)	Escherichia coli, third-generation cephalosporin-resistant	Medium	High	Medium–High	Level 4	Medium-High	Medium–Low	High	Likely
2	64.94 (10.68)	Pseudomonas aeruginosa, ceftolozane-tazobactam resistant	High	Medium–Low	Low	Level 3	Medium-High	Medium–Low	Low	Unlikely
2	61.3 (11.52)	Enterococcus faecalis, vancomycin-resistant	High	Medium–Low	Low	Level 3	Medium	Medium–Low	Low	Unlikely
2	57.53 (8.48)	Escherichia coli, carbapenem-resistant	Medium-High	Medium	Low	Level 4	Medium	Medium–Low	Low	Likely
2	57.15 (11.74)	Enterobacteriaceae, ceftazidime-avibactam resistant	Medium-high	Low	Low	Level 3	Medium–High	Medium–Low	Low	Unlikely
3	57.11 (9.46)	Klebsiella pneumoniae, third-generation cephalosporin-resistant	Medium-high	High	Low	Level 4	Medium–high	Medium–Low	High	Likely
3	55.61 (9.09)	Mycoplasma pneumoniae, macrolide-resistant	Low	High	Medium–high	Level 2	Medium	Low	Medium	Possible
3	50 (8.66)	Enterobacter spp., carbapenem-resistant	Medium-High	Medium–Low	Low	Level 2	Medium–High	Medium–Low	Low	Likely
3	49.65 (9.71)	Staphylococcus aureus, methicillin-resistant	Medium-High	High	Low	Level 1	High	Medium–high	Medium–High	Likely
3	44.73 (7.64)	Mycobacterium tuberculosis, drug-resistant	Medium	Medium–low	Low	Level 2	High	Medium–high	High	Unlikely
3	28.36 (7.44)	Streptococcus pneumoniae, third-generation cephalosporin-resistant	Low-medium	Medium–low	Low	Level 2	Medium–Low	Medium–low	High	Possible
3	24.8 (9.4)	Staphylococcus aureus, vancomycin-resistant	Low	Low	Low	Level 2	Medium Low	Low	Medium–High	Unlikely

Evidence extraction and data synthesis

Using MCDA, we selected 17 pathogens for prioritization and evaluated them against the eight predefined criteria based on evidence from literature review, surveillance data, and clinical data from the five participating institutions or expert opinion (Supplementary Table S2.0). Table 2 presents the MCDA matrix, which illustrates how each pathogen is evaluated against the eight criteria. Detailed descriptions of the criteria definitions, scoring frameworks, and data sources used for the evaluation are provided in Supplementary Data S2. The literature search strategies, process of evidence synthesis, and additional methodological details supporting the MCDA are also presented in the Supplementary material.

Weighting of criteria

Of 100 multidisciplinary experts, 56 participated in the blinded preference-based survey to determine the final weight (%) for the eight criteria and the preference values (score) for each level of the criteria (response rate: 56%). For preference values calculated using the PAPRIKA algorithm, a higher preference value indicates a relatively greater impact on the importance of each criterion’s level. The weights and the preference values for the eight criteria are presented in Fig. 1 and Table 1, respectively.

Fig. 1.

Comparison of the pooled criteria weights (derived from the PAPRIKA^a survey) for the Republic of Korea priority pathogen list versus those for the 2024 WHO bacterial priority pathogen list (BPPL). Radar chart comparing pooled weights assigned to prioritization criteria based on PAPRIKA survey results for the Republic of Korea (ROK) and the WHO 2024 Bacterial Priority Pathogen List. Solid lines represent ROK weights, and dashed lines represent WHO weights. Percentages indicate normalized criteria weights that sum to 100% for each framework. Inter-rater agreement among respondents was lower for ROK (Kendall’s W = 0.286) than for WHO (Kendall’s W = 0.871). Spearman rank correlation coefficients reflect within-panel inter-rater agreement for each framework independently: ROK panel (ρ = 0.82) and WHO panel (ρ = 0.90). ^aPAPRIKA (Potentially All Pairwise Rankings of all Possible Alternatives) is a robust decision-making approach for systematic evaluation and ranking of all conceivable pairwise alternatives to ensure comprehensive decision analysis. The method is a structured, thorough means of comparing and prioritizing diverse alternatives in making complex decisions.

Comparison of the criteria weights between the 2024 WHO BPPL and the list developed in the present study revealed differences. Treatability, case fatality, and AMR trend had the highest weights in the 2024 WHO BPPL, whereas case fatality, transmissibility, and nonfatal health burden had the highest weights in our priority list. The criteria that differed by more than five ranks between the two lists were treatability and transmissibility, which were considered important only by the WHO and the ROK, respectively⁴ (Fig. 1).

Scoring and ranking of the pathogens

Total risk scores ranged from 24.8 to 85.9 (Fig. 2). Pathogens were classified into three tiers based on score distribution, within-group homogeneity, and resource allocation considerations. The top three pathogens (79.9–85.9 points) all shared very high case fatality, very low treatability, and very high transmissibility. Ranks 4–10 scored 57.2–70.4 points, and ranks 11–17 ranged from 24.8 to 57.1 points. The distribution of the pathogens across priority tiers was as follows: Tier 1 (requires emergency response), three pathogens (17.6%); Tier 2 (requires preferred management), seven pathogens (41.2%); and Tier 3 (requires monitoring), seven pathogens (41.2%).

Fig. 2.

Final prioritization scores of antimicrobial-resistant pathogens in the Republic of Korea based on the results of the PAPRIKA analysis. The stacked bars represent the weighted contributions of eight evaluation criteria to the final prioritization score of each pathogen: incidence, nonfatal health burden, trend of antimicrobial resistance, transmissibility, preventability, treatability, case-fatality, and antibiotic development pipeline. Final scores were calculated using criteria weights derived using the PAPRIKA method. Values in parentheses indicate the standard deviation (SD) of expert scores. Background shading denotes the three priority tiers: Tier 1 (emergency response), Tier 2 (preferred management), and Tier 3 (monitoring). Tier boundaries were determined based on natural discontinuities in the score distribution and within-group risk profile homogeneity.

Comparison of pathogen prioritization between the 2024 WHO BPPL and our list is presented in Supplementary Fig. S3. Although the total number of pathogens and the classification of groups differed between the two lists, carbapenem-resistant K. pneumoniae (CRKPN) and carbapenem-resistant A. baumannii complex (CRAB) were included in the most important groups in both lists. Vancomycin-resistant Enterococcus faecium (VREF), the remaining one of the three most important pathogens in the ROK, was included in the secondary group in the 2024 WHO BPPL.

Agreement and sensitivity analysis

Expert agreement was highest for case fatality (CV = 39%) and weakest for preventability (CV = 103%), treatability (CV = 76%), and antibiotic development pipeline (CV = 86%) (Table 3). Regarding Kendall’s W, the agreement level for all respondents (n = 56) was W = 0.286, indicating a weak level of agreement (P < 0.001). Subgroup analyses showed higher agreement among university hospital-based experts (W = 0.361) compared to those from other institutions (W = 0.128, non-significant), and among experts with less than 10 years of experience (W = 0.434) compared to those with 21 or more years (W = 0.325). By specialty, agreement was moderate among nonclinical specialists (W = 0.427) and weak among infectious disease specialists (W = 0.334), non-infectious disease specialists (W = 0.284), and physicians (W = 0.342) (Supplementary Table S3.1).

Table 3.

PAPRIKA-derived criteria weights and levels of consensus assessed using coefficients of variation.

Criterion	Mean (%)	SD	CV (%)	Level of consensus
Case fatality	18.24	7.16	39	Moderate
Transmissibility	16.41	7.2	44	Weak
Non-fatal health burden	14.38	7.28	51	Weak
Trend of resistance	14.1	6.73	48	Weak
Incidence	13.83	8.72	63	Very weak
Treatability	9.62	7.36	76	Almost no consensus
Pipeline	7.39	6.37	86	No consensus
Preventability	6.02	6.2	103	No consensus

PAPRIKA = Potentially All Pairwise Rankings of all Possible Alternatives; SD = standard deviation; CV = coefficient of variation.

Note: CV was calculated as (SD/mean) × 100.

Consensus levels were defined as follows: CV < 30%, strong consensus; 30–50%, moderate.

Sensitivity analyses demonstrated high rank stability across all four weighting scenarios, with Spearman rank correlations between the base scenario (S1) and alternatives ranging from 0.975 to 0.988. Among 17 pathogens, 11 showed no rank change; the maximum rank shift was three positions (Mycoplasma pneumoniae and C. auris, ranks 8th–11th). All three Tier 1 pathogens retained their top positions across all scenarios (Supplementary Table S3.3).

Discussion

In this study, we established a ROK-specific priority list of antimicrobial-resistant pathogens within the healthcare system, rather than a fully population-based assessment. In addition, we developed the first platform for priority assessment of antimicrobial-resistant pathogens in the ROK. A key methodology of this platform was the use of MCDA, which provides a structured, explicit, and transparent approach to multidimensional decision making. Accordingly, this priority list of antimicrobial-resistant pathogens needs to be updated regularly to reflect the constantly evolving epidemiological trends of AMR and supplemented with more data on the disease burden of each antimicrobial-resistant pathogen in the broader community and primary care settings.

We developed our AMR priority list using MCDA with the PAPRIKA method, which has been applied to rank national treatment and reimbursement priorities and inform hospital formularies.15, 16, 17 MCDA systematically evaluates multiple criteria by integrating objective data from peer-reviewed literature and surveillance systems with structured expert input.⁷ The PAPRIKA method, based on pairwise ranking, reduces respondent burden compared to approaches requiring simultaneous comparison of multiple alternatives. In our framework, the eight criteria were categorized as quantitative indicators, assessed using numerical surveillance system and literature data, and qualitative indicators, scored on ordinal scales based on expert judgment and published evidence, thereby enabling the integration of heterogeneous data types within a single-weighted scoring model.

Despite these methodological strengths, panelist agreement in the PAPRIKA survey was relatively low (Kendall’s W = 0.286), with high variability for preventability (CV = 103%), treatability (CV = 76%), and antibiotic development pipeline (CV = 86%), likely reflecting differences in AMR expertise and the inherent directional ambiguity of these criteria. For example, high preventability may be interpreted either as a reason to prioritize a pathogen or as a sign of lower urgency. However, the practical impact on final rankings was limited: these criteria received the lowest pooled weights (6.0%, 9.6%, and 7.4%, respectively), and sensitivity analyses showed that 11 of 17 pathogens had no rank change across alternative weighting scenarios, with all three Tier 1 pathogens retaining their top positions (Supplementary Table S3.3). Spearman rank correlation among respondents’ final pathogen rankings was 0.82, indicating stronger agreement on overall prioritization than on individual criterion weighting. Future assessments should incorporate explicit criterion guidance and a structured pre-survey workshop to align interpretive frameworks.

Our list has several differences compared to previous priority lists, such as the 2024 WHO BPPL, 2022 WHO FPPL, the 2019 CDC list, and the updated 2025 PHAC list.^4,5,9,10,18 First, our study was designed with a comprehensive public health objective—supporting AMR control policies— rather than a single specific purpose such as guiding R&D, informing public health investments, or streamlining surveillance. Second, our candidate pathogens included novel and emerging resistant pathogens: ceftolozane/tazobactam-resistant P. aeruginosa, vancomycin-resistant Enterococcus faecalis, ceftazidime/avibactam-resistant Enterobacteriaceae, macrolide-resistant M. pneumoniae, and vancomycin-resistant S. aureus, though data limitations for these recently emerging pathogens represent an inherent constraint. Third, the ROK-specific data on antibiotic accessibility or the epidemiology of multidrug-resistant microorganisms led to differences in assessment criteria and candidate antimicrobial-resistant pathogens. Finally, our list was broadly evaluated and included bacteria, fungi, and Mycobacterium tuberculosis. In particular, C. auris, the only fungal pathogen in our list, was assigned to Tier 2 of our framework, while it was ranked as the highest-priority pathogen in the 2022 WHO FPPL.

Comparison with the 2024 WHO BPPL revealed differences in criterion weighting, reflecting differences in objectives. The WHO BPPL prioritized treatability (20%) to guide R&D investment, while our study prioritized case fatality (18.2%) and transmissibility (16.4%) to reflect national public health and clinical needs. The four highest-weighted criteria in the WHO BPPL — treatability, mortality, healthcare burden, and 10-year resistance trend — collectively represented 49.7% of total weight, compared to 63.1% for our top four criteria: case fatality, transmissibility, nonfatal health burden, and five-year AMR trend.⁴ Notably, treatability ranked first in the WHO BPPL but sixth in our framework.

The Tier 1 (emergency response) group in the present study included three pathogens; carbapenem-resistant K. pneumoniae (CRKPN), carbapenem-resistant A. baumannii, and vancomycin-resistant E. faecium (VREF). Both CRKPN and CRAB were also classified as a “critical” group in the 2024 WHO BPPL and an “urgent hazard” in the 2019 CDC. These pathogens have been recognized as the most critical antimicrobial-resistant pathogens worldwide, sharing the triple high-risk characteristics of very high fatality, very low treatability, and very high transmissibility. VREF was included in the “high” group by the WHO but was upgraded to Tier 1 (requires emergency response) in the present study. Our findings reflect the unique characteristics of the ROK, where the incidence rate of VREF is relatively high and transmission occurs predominantly in long-term care facilities.19, 20, 21

This study has some limitations. First, the Delphi response rate of 61.3% may have introduced non-response bias, though the panel of 38 experts exceeded the commonly recommended minimum. Second, evidence was limited for some criteria and pathogens, with scoring relying partly on expert opinion and data from five institutions, which may limit representativeness of the national AMR epidemiology. Furthermore, the case fatality criterion reflects the all-cause in-hospital mortality, rather than any directly attributable mortality, potentially overrepresenting pathogens influencing severely ill patients in tertiary care. Additionally, while KARMS includes some primary care isolates, Korean AMR surveillance remains predominantly tertiary care-based, and findings should be refined as nationally representative data become available. Future efforts should establish systematic strategies for obtaining reliable disease burden data and may explore context-specific prioritization by healthcare setting or patient population to provide more targeted AMR guidance. Also consistent with the 2024 WHO BPPL, healthcare-associated outbreak potential, community transmission, and intersectoral transmission were all captured within a single transmissibility criterion, rather than scored separately. As was acknowledged in the 2024 WHO BPPL, this approach may not fully distinguish pathogens whose transmission risk is concentrated in a single setting. Third, the specific purpose of this priority pathogen list was not clearly defined when we commenced this study. While the 2024 WHO BPPL focuses on pathogens requiring novel antibiotics, our study aimed to provide a broader tool to support public health AMR policy decision-making. Aspects of such as criterion weighting and pathogen prioritization may vary depending on the intended use of the list. Finally, this assessment was limited to bacterial and fungal pathogens; antiviral resistance was excluded as it requires distinct surveillance systems and evidence frameworks, and should be addressed in future priority assessments in the ROK.

In the present study, we established the first systematic MCDA-based priority list of antimicrobial-resistant pathogens in the ROK that could inform national strategies for AMR containment. Our findings provide a coordinated prioritization platform combining structured evidence and expert input via MCDA, allowing for convenient and regular updates. However, effective utilization of this platform requires continuous high-quality data on the disease burden of antimicrobial-resistant pathogens across all healthcare settings, including community and primary care, from a stable public health system.

Contributors

Conceptualization and design: Y. K. Y. Acquisition, analysis, and interpretation of data: S. Y. Y., E. Y. B., J. H. Y., H. S., Y. J. C., S. J. C., H. S. O., J. W. S., J. C., G. N., J. Y. K., and Y. K. Y. Statistical analysis: S. Y. Y., E. Y. B., and J. H. Y. Drafting of the manuscript: Y. K. Y., S. Y. Y., J. H. Y., and J. Y. K. Critical review and revision of the manuscript: All authors. Y. K. Y., S. Y. Y., and J. H. Y. accessed and verified the underlying data. Y. K. Y. was responsible for the decision to submit the manuscript.

Data sharing statement

Data supporting the findings of this study are available from the corresponding author upon request.

Declaration of interests

The authors declare no competing interests.