Surveillance of antibiotic-resistant bacteria integrating whole genome sequencing and epidemiological information – a global, systematic review, 2012–2023

Abstract

Background

Epidemiologic surveillance is one pillar in the global efforts to stop the spread of antibiotic-resistant bacteria (ARB) and increasingly includes whole genome sequencing (WGS). However, no review systematically examined the characteristics of such surveillance systems until February 2025. We conducted a systematic review to identify and describe routine surveillance systems of ARB, that integrate WGS and epidemiological information.

Methods

A protocol was developed and published on PROSPERO. We searched 8 databases and Google. Included articles needed to report on a national or subnational surveillance system and include epidemiological data as well as WGS of carbapenem-resistant or carbapenemase-producing Enterobacterales (CRE) or Acinetobacter spp., methicillin-resistant Staphylococcus aureus (MRSA) or vancomycin-resistant Enterococci (VRE). Surveillance system characteristics were analysed narratively.

Results

From 13,142 searched articles, 41 were included. The publications were stratified by pathogen. The strata demonstrated similar WGS approaches with a predominance of short-read sequencing platforms. Typing methods mostly included multilocus sequence typing and single nucleotide polymorphism -based analyses for higher discrimination. Reported cluster thresholds showed a wide range of values within and between species. Across strata, there was often a low level of epidemiological data collection, such as exposure of cases, and integration in cluster analyses. WGS surveillance complemented with at least basic epidemiological data has been reported from all six WHO regions for CRE. However, fewer studies and no studies at all for MRSA and VRE were included from the African, Eastern Mediterranean and South-East Asian regions. The national coverage of surveillance ranged from 36% for MRSA to 56% for CRE.

Conclusion

WGS-based surveillance on ARB is employed worldwide with predominant use of short-read technology and largely consistent typing methods. Substantial variation in genomic clustering thresholds underscores the need for internationally harmonized standards for data analysis. Insufficient epidemiological information or lacking integration in the analysis of WGS data may reduce the ability to infer transmission pathways and identify outbreaks. To fully exploit the potential of WGS-based surveillance systems in this context for public health and contain current and prevent future outbreaks, we thus call for a routine collection of detailed epidemiological data.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13756-026-01821-9.

Introduction

Infections with antibiotic-resistant bacteria (ARB) cause increased morbidity, mortality and healthcare costs. It has been estimated that ARB led to 1.27 million deaths globally in 2019 [1]. Hence, the World Health Organization (WHO) declared antimicrobial resistance, to be “one of the top 10 global public health threats facing humanity” [2]. There are predictions of a ten-fold increase in death due to antibiotic resistant infections by 2050 [3].

In 2015, the WHO launched a global action plan to address antibiotic resistance, aiming to optimize antimicrobial use and strengthen surveillance systems [4]. Surveillance of ARB may be able to detect the emergence and dissemination of new antibiotic-resistant strains. By helping to discern transmission pathways, chains of infection may be stopped more quickly.

Traditional pathogen-based surveillance methods rely on phenotypic antibiotic susceptibility testing (AST) although molecular methods like PCR are also used [5]. Alongside epidemiological information, these methods may support inferences on transmission pathways through the identification of identical resistance patterns and determinants. However, to confirm the genetic relatedness of a suspected outbreak clone, a typing method needs to be deployed. Because of its superior discriminatory power, whole genome sequencing (WGS) has become the gold standard for bacterial strain typing and is increasingly implemented in routine surveillance [6–8].

WGS can provide the complete DNA sequence of a bacterial isolate [9] through the application of different sequencing technologies. In general, available technologies for WGS either generate short (Illumina, Ion Torrent) or long reads (Oxford Nanopore (ONT), Pacific Biosciences). Short-read technologies produce millions of high-quality reads, typically 100–300 base pairs in length, enabling high-throughput sequencing of samples and thus decreasing costs. However, these short reads are generally unable to fully reconstruct genomic structures of interest, including mobile genetic elements such as resistance plasmids. In contrast, long-read technologies can overcome these limitations. Previously, long-read technologies were characterized by higher sequencing error rates as compared to short-read platforms and the limited possibility of multiplexing of samples increased the costs per sequenced isolate [10]. However, recent advancements substantially increased accuracy and affordability of long-read sequencing.

Obtained sequences can be used to reconstruct the genome from scratch (de novo assembly) or be aligned against a reference sequence (mapping). Multiple bacterial genomes can be compared to evaluate their genetic relatedness, which is usually measured in the number of differences between isolates. In incrementing order, multilocus sequence typing (MLST) has the lowest discriminatory power because it relies only on the allelic combination of a few house-keeping genes, whose number depends on the species to be analyzed. In contrast, the core genome (cg)MLST is based on the pairwise allelic comparison of thousands of genes present in all isolates of a species and whole genome (wg)MLST which also includes genes of the accessory genome existing only in a subset of isolates. Single nucleotide polymorphisms (SNP)-based analyses provide the highest resolution to differentiate between closely related isolates.

Irrespective of the typing method, the fewer differences, the more recently the strains are likely to have diverged, which in conjunction with epidemiological information may provide important insights into transmission pathways, cluster affiliation of isolates in outbreak settings, or the emergence and dissemination of pathogens [11].

Therefore, the combination of WGS-based pathogen information along with epidemiological data and reasoning is a promising means to make a difference in the prevention of AMR [11, 12]. However, to our knowledge, a comprehensive overview of existing surveillance systems that integrate WGS with epidemiological information is lacking. In order to provide an important resource for health authorities to develop or evaluate their own systems, this systematic review aims to describe the existence, distribution and characteristics of WGS systems, integrated with epidemiological information, used for surveillance of antibiotic-resistant organisms globally.

Materials and methods

Protocol

A detailed protocol for this systematic review was developed collaboratively by the systematic-reviewing company (SRC) and the Funding Academic Institution (FAI), and signed off prior to any extraction of data. The protocol was published on (PROSPERO 2023 CRD42023456676) (https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023456676), and is summarised in Table 1 below.

Table 1.

Systematic review protocol

What are the characteristics of models for routine surveillance of pathogens with antimicrobial resistance, that utilise Whole Genome Sequencing integrated with epidemiological information?
Population	Inclusion • Human samples containing pathogens with antimicrobial resistance. • Only include studies that consider at least one of the following species:  • Carbapenem-resistant or carbapenemase-producing Enterobacterales (CRE/CPE)  • Carbapenem-resistant or carbapenemase-producing Acinetobacter spp. (CRA/CPA)  • Methicillin-resistant Staphylococcus aureus (MRSA)  • Vancomycin-resistant Enterococci (VRE) Exclusion • Other pathogens / resistances • Studies that ONLY handle food, animal or other environmental samples.
Inclusion criteria relating to content of publications	Any reported models of whole genome sequencing, integrated with epidemiological information, that are used for surveillance of antimicrobial resistance, will be included. The publication must have sufficient detail about methods, to determine that the system described meets the definition of a surveillance system (which will exclude sole outbreak reports). The WHO definition of public health surveillance will be used: ’the continuous, systematic collection, analysis and interpretation of health-related data […] guiding priority-setting and planning and evaluation public health policy and strategies’ [13]. The surveillance system must operate at a regional, national or supra-regional level. There must be detailed description of the population, including epidemiological information (has to include information on patient characteristics, such as age, sex, diagnosis, geographic location, patient location (ward/clinic), specimen type, or clinical outcome). Data covering at least one of the following four pieces of information must be available for inclusion: • Information on any countries that have started to build such systems, or have them already established • Scope of the models: for sporadic outbreak detection, for temporal and/or spatial trends, or routine monitoring of genetic properties • Transmission pathways described. • Visualisation methods applied [trees, timelines, heatmaps, networks and others].
Study types	Inclusion • Full-text primary articles of any form (including grey literature) that include a description of WGS models for surveillance of antimicrobial resistance that are in actual use. Exclusion • Papers solely discussing hypothetical systems that are not currently in use. • Secondary (i.e. review/guideline) articles and abstracts.
Dates	2012 to August 2023 (details see below)
Languages	English, Chinese (Mandarin), French, Spanish and German. Other languages will be excluded.
Stratification	Unconditional Stratification will be applied unconditionallyA for the type of pathogen (the four categories are given in the population section). Conditional Conditional stratificationB will occur for the type of additional environmental, or animal samples used (no additional samples versus additional environmental samples versus additional food samples versus additional animal samples versus other categories that reflect combinations of the additional types of sample [for example, additional food and environmental samples]). If no meta-analysis is used, then these conditional categories for additional samples will become unconditional, and the strata of the review will be formed by permutations of pathogen and additional sample categories.
Synthesis of findings	Meta-analysis will be used if appropriate, but it is envisaged that the most likely form of synthesis will be narrative (tabular).

Abbreviations: WHO = World Health Organization

A. unconditional stratification: stratified ‘up front’, prior to any initial overall meta-analysis, and therefore not conditional on any initial overall meta-analysis I² result.

B. conditional stratification: stratified after initial overall meta-analysis shows an I² value of ≥ 50%

Literature searches

Searches were carried out to identify relevant information on currently used routine surveillance of ARB, that utilise WGS integrated with epidemiological information. The search strategies were developed specifically for each database and the keywords adapted according to their configuration. Searches combined relevant search terms comprising indexed keywords (e.g. Medical Subject Headings (MeSH) and EMTREE) and free text terms appearing in the title and/or abstract of database records. Search terms were identified through discussion with the review team and FAI, by scanning background literature and ‘key articles’ already known to the review team, and by browsing database thesauri.

The search was limited by publication date (2012- August 2023) but not by language, nor publication status.

The following databases were searched for relevant articles:

• MEDLINE and Epub Ahead of Print, In-Process, In-Data-Review & Other Non-Indexed Citations and Daily Update (Ovid): 1946 to 2023/08/21.

• Embase (Ovid): 1974 to 2023/08/21.

• PubMed (https://pubmed.ncbi.nlm.nih.gov/): up to 2023/08/22.

• KSR Evidence (https://ksrevidence.com/): up to 2023/08/30.

• Cochrane Database of Systematic Reviews (CDSR) (Wiley): up to 2023/08/Iss8.

• Cochrane Central Register of Controlled Trials (CENTRAL) (Wiley): up to 2023/07/Iss7.

The following grey literature resources were searched from inception to the present:

• International HTA Database (https://database.inahta.org/): up to 2023/08/29.

• Policy Commons (https://policycommons.net/): up to 2023/08/22.

References in retrieved articles were checked for additional publications. Further to this, a search of the first 100 results retrieved by Google (search terms: ‘whole genome sampling resistance surveillance’ on 08/09/23) was conducted, to ensure that no key grey literature documents were missed.

For all searches undertaken by the SRC Information Team, the main Embase strategy was independently peer-reviewed by a second SRC Information Specialist, based on the Canada’s Drug and Health Technology Agency (CADTH) checklist [14–16].

Details of the search strategies are contained in Additional file 1.

Methods of article selection, quality assessment, and data extraction

Two SRC reviewers (KMcD and MP) independently selected potentially eligible articles from the titles and abstracts based on the protocol.

Eligibility of the full-texts was assessed by one SRC reviewer (MP) based on the protocol. A second (blinded) one (KMcD), re-evaluated 5% of the full-text papers. The two reviewers obtained 100% concordance. Finally, FAI staff checked full-text paper inclusion and found further relevant articles.

One SRC reviewer (MP) extracted the following data in a tailor-made extraction sheet in Word: location and time of the surveillance, type of surveillance (sentinel or comprehensive), coverage (national or subnational), isolate collection methods, target population, epidemiological data collected (age, sex, diagnosis, outcome, suspected place of acquisition, ward/clinic, travel history, hospitalisation history), and WGS methodologies used. There were no outcome data, as this was a descriptive review. Members of the FAI team checked data extraction accuracy.

No appraisal of bias within studies was carried out, since the surveillance methods and not the study results were described in this systematic review.

Data synthesis

The articles were stratified by pathogen (carbapenem-resistant or carbapenemase-producing Enterobacterales (CRE/CPE), carbapenem-resistant or carbapenemase-producing Acinetobacter spp. (CRA/CPA), methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococci (VRE)) as prescribed in the protocol. The extracted characteristics of the underlying surveillance were described and summarised with frequency or range (genetic distance) and median (time) within each stratum. Where adequate, relative frequencies were compared between subgroups (such as WHO region) or summarised across strata. Formal quantitative analysis was not conducted, as this was considered over-analysis and not pre-specified in the protocol. Synthesis was performed by one reviewer (MP), with accuracy checked by the FAI team. The systematic review was written up adhering to PRISMA methodology [17, 18], with the completed PRISMA checklist available in Additional file 2.

Results

Included articles

The systematic search yielded 13,040 abstracts. After removal of duplicates, 7,955 articles remained. The first 100 ‘titles’ from a Google search were added, alongside two further ad-hoc identified abstracts, making a total of 8,057 abstracts/titles for sifting. From the initial sift, 333 full-text articles were retrieved. Of these, 41 articles [19–59] were included in the review. Exclusion reasons for the 292 articles are provided in Additional file 3. Figure 1 provides the PRISMA flow chart.

Fig. 1.

PRISMA flow diagram

Locations, sectors and target pathogens

Surveillance including epidemiologic and WGS data was reported from all six WHO regions (Fig. 2). Earliest and 85% of publications originate from the Western Pacific region (n = 14), European region (n = 11) and region of the Americas (n = 10). No article reported additional environmental or animal samples. Table 2 summarises these articles per pathogen stratum. Five articles included pathogens from several strata [22, 30, 52, 53, 56]. Some articles included target pathogens of this systematic review but had a wider or different pathogen scope (n = 13; e.g. ESBL-producing Enterobacterales), or focussed on specific properties (n = 5; e.g. NDM-5-producing E. coli).

Fig. 2.

Locations of included articles by coverage, pathogen(s) under surveillance and WHO region

Table 2.

Included studies, separated by pathogen stratum

Publication	WHO REGION Country Coverage (type of surveillance) $	Start year End year approx. duration	Target species (sample material)	No. isolates: -study total -belonging to stratum -stratum WGS	Sequencing technology and lab (central, decentral, external)	Typing methods	Availability* of epidemiological information								Epidemiological information integrated with genetic differences (threshold) to infer clusters or transmission pathways	Public availability of sequence data
							Age	Gender	Diagnosis	Outcome	HAI/ CAI	Ward, clinic	Travel	Hosp. history
Stratum 1: Carbapenem-resistant or carbapenemase-producing Enterobacterales
Afolayan, 2021 [20]	AFRICAN Nigeria Subnational (sentinel, tertiary care hospitals)	2016 2018 36 m	Antimicrobial-resistant Klebsiella spp. (clinical: blood, CSF, urine)	141 11 11	Illumina de novo central	MLST, SNP	✓	✓	✓	✗	✓	✓	✗	✗	Time of isolate recovery; others for visualisation (ESBL-Kp: 1–3 SNPs)	Yes
Milenkov, 2021 [46]	AFRICAN Madagascar Subnational (sentinel, maternity wards)	2018 2019 9 m	ESBL- producing Enterobacterales (screening)	170 1 1	Illumina + ONT de novo central	MLST, SNP	✓	✓	✓	✗	✗	✗	✗	✗	Unclear; isolate origin for visualisation (ESBL-Ec: <10 SNPs)	Yes
Karlowsky, 2021 [35]	AMERICAS Canada National (comprehensive, tertiary care hospitals)	2007 2018 144 m	ESBL-positive Escherichia coli and Klebsiella pneumoniae (clinical)	812 23 23	Illumina de novo central	MLST	✓	✓	✗	✗	✗	✓	✗	✗	NR (NA)	Yes
Saavedra, 2021 [51]	AMERICAS Colombia National (comprehensive, NRL)	2013 2017 60 m	Carbapenem-resistant Klebsiella pneumoniae (clinical)	425 425 425	Illumina de novo central	MLST, SNP	✓	✓	✓	✗	✗	✗	✗	✗	Unclear (NR)	Yes
Bulens, 2023 [23]	AMERICAS USA Subnational (sentinel, MuGSI)	2012 2015 36 m	Carbapenem-resistant Enterobacterales (clinical)	1499 1499 12	Illumina de novo central	MLST	✓	✓	✓	✓	✓	✗	✓	✓	NR (NA)	No
Lapp, 2023 [39]	AMERICAS USA Subnational (comprehensive, MuGSI and statewide CRE surveillance)	2016 2018 36 m	Carbapenem-resistant (blaKPC-carrying) Enterobacterales (clinical)	255 255 255	Illumina de novo decentral	MLST, SNP	✓	✓	☑	✗	✗	✓	✗	✓	Comorbidities, geographic (≤ 5, 10, and 15 SNVs)	Yes
Al-Farsi, 2020 [21]	EAST MEDITERRANEAN Oman National (comprehensive)	2015 2015 10 m	Escherichia coli with reduced susceptibility to carbapenems (clinical and screening)	35 35 35	Illumina de novo central	MLST, cgMLST, wgMLST, SNP	✓	✓	✓	✓	✗	✓	☑	✓	Hospital, ward, isolation date, hospitalisation and travel history (Ec: ≤10 SNPs)	Yes
Sonnevend, 2022 [55]	EAST MEDITERRANEAN United Arab Emirates National (sentinel)	2018 2019 9 m	Carbapenem-resistant Enterobacterales (clinical and screening)	504 504 504	Illumina de novo central	MLST, cgMLST	✓	✓	✗	✗	✗	✓	✗	✗	Hospital (Kp: ≤15 allelic differences)	Yes
Do Nascimento, 2017 [28]	EUROPE UK Subnational (comprehensive)	2015 2016 18 m	Enteroaggregative Escherichia coli (clinical: faecal samples)	155 1 1	Illumina mapping central	MLST	✓	✓	✗	✗	✗	✗	☑	✗	NR (NA)	Yes
Roer, 2017 [50]	EUROPE Denmark National (comprehensive)	2014 2015 24 m	Third-generation cephalosporin-resistant Escherichia coli (blood culture)	552 5 5	Illumina de novo central	MLST, SNP	✓	✓	✓	✗	✗	✗	✗	✗	Unclear; isolate origin for visualisation (Ec: ≤10 SNPs)	No
Hammerum, 2020 [32]	EUROPE Denmark National (comprehensive)	2016 2019 44 m	OXA-244-producing Escherichia coli (clinical and screening)	24 24 24	Illumina de novo central	MLST, cgMLST	✓	✓	✗	✗	✗	✓	✓	✓	Travel and hospitalisation history, place of sampling and residence, (Ec: ≤10 allelic differences)	Yes
Taylor, 2021 [56]	EUROPE UK Subnational (sentinel)	2016 2016 6 m	Amikacin-resistant Enterobacterales (clinical and screening)	271 135 135	Illumina de novo central	MLST	✓	✓	✓	✗	✗	✗	☑	✗	NR (NA)	No
Canada-Garcia, 2022 [24]	EUROPE Spain National (sentinel)	2019 2019 4 m	Carbapenemase-producing Klebsiella pneumoniae and Escherichia coli (clinical)	403 403 403	Illumina de novo central	MLST, SNP, cgMLST	✓	✓	✗	✗	✗	✗	✗	✗	Unclear (Kp: ≤10 SNPs, ≤ 5 allelic differences)	Yes
Hans, 2023 [33]	EUROPE Germany National (comprehensive)	2013 2019 84 m	NDM-5-producing Escherichia coli (clinical and screening)	329 329 222	Illumina de novo central	MLST, cgMLST, SNP	✓	✓	✗	✗	✗	☑	✗	✗	Geography (Ec: ≤100 SNPs)	Yes
Ljungquist, 2023 [43]	EUROPE Norway National (comprehensive)	2015 2021 84 m	Carbapenemase-producing Enterobacterales (clinical and screening)	389 389 389	Illumina de novo central	MLST, cgMLST	✓	✓	✗	✗	✗	✗	✓	✗	Hospital, travel history (NR)	Yes
Raffelsberger, 2023 [49]	EUROPE Norway Subnational (comprehensive)	2015 2016 19 m	ESBL-producing Escherichia coli and Klebsiella pneumoniae (screening)	4999 2 2	Illumina de novo external	MLST, SNP	✓	✓	✓	✗	✗	✗	✓	✓	NR (Ec: ≤17 SNPs)	Yes
Nagaraj, 2021 [47]	SOUTH EAST ASIA India National (sentinel)	2013 2019 84 m	Carbapenem resistant Klebsiella pneumoniae (clinical)	344 280 280	Illumina de novo NR	MLST, SNP	✓	✓	✓	✗	✓	✓	✗	✗	Hospital, inpatient/outpatient, age (Kp: ≤3 SNPs)	Yes
Marimuthu, 2017 [44]	WEST PACIFIC Singapore National (comprehensive)	2013 2015 17 m	Carbapenem-resistant Enterobacteriaceae (clinical and screening)	430 430 206	Illumina de novo central	MLST, SNP	✓	✓	✓	✓	✗	✓	✓	✓	Hospital (Ec: ≤58 SNPs, Kp: ≤151 SNPs)	Yes
Sherry, 2019 [52]	WEST PACIFIC Australia Subnational (comprehensive)	2012 2016 60 m	Carbapenemase-producing Enterobacterales (clinical and screening)	361 361 361	Illumina de novo central	MLST, SNP	✓	✓	✗	✗	✓	✓	✓	✓	“Data from concurrent epidemiologic investigations” (Hospital, hospitalisation and travel history) (≤ 23 SNPs)	Yes
Argimon, 2020 [22]	WEST PACIFIC Philippines National (sentinel)	2013 2014 24 m	Carbapenemase- and/or ESBL-producing Klebsiella pneumoniae and Escherichia coli (clinical)	805 ~ 420 ~ 420	Illumina de novo central	MLST, SNP	✓	✗	✗	✗	✓	✓	✗	✗	Hospital, ward isolation date, HAI/CAI, age (threshold unclear)	Yes
Carlos, 2021 [25]	WEST PACIFIC Philippines National (sentinel)	2015 2017 36 m	Carbapenem-, extended spectrum beta-lactam-, and cephalosporin-resistant Klebsiella pneumoniae (clinical)	259 139 139	Illumina de novo central	MLST, SNP	✓	✓	✗	✗	✓	✓	✗	✗	Hospital, ward, sampling time, age, HAI/CAI (threshold unclear)	Yes
Lane, 2021 [38]	WEST PACIFIC Australia Subnational (comprehensive)	2016 2018 36 m	Carbapenemase-producing Enterobacterales (clinical and screening)	402 402 402	Illumina de novo central	MLST, SNP	✓	✓	✓	✓	✗	✓	✓	✓	Hospital, hospitalisation and travel history and eventual additional outbreak investigation (threshold unclear)	Yes
Forde, 2023 [30]	WEST PACIFIC Australia Subnational (comprehensive)	2017 2021 50 m	Carbapenemase-producing Enterobacterales (clinical and screening)	2660 293 293	Illumina de novo central	MLST, SNP	✓	✗	✗	✗	✓	✓	✗	✗	Hospital, HAI/CAI (≤ 5 SNPs/Mb)	Yes
Ko, 2023 [37]	WEST PACIFIC China National (sentinel)	2019 2020 12 m	Carbapenemase-producing Escherichia coli (clinical)	31 31 13	ONT de novo NR	MLST, cgMLST	✓	✓	✗	✗	✗	✗	✗	✗	Unclear (NR)	No
Li, 2023 [42]	WEST PACIFIC China Subnational (sentinel)	2017 2022 72	Carbapenem-resistant Klebsiella pneumoniae (clinical)	662 662 101	Illumina de novo external	MLST, SNP	✓	✓	✗	✗	✗	✗	✗	✗	Unclear, age, sex, isolation region and year for visualisation (NR)	Yes
Stratum 2: Carbapenem-resistant or carbapenemase-producing Acinetobacter spp.
Perovic, 2022 [48]	AFRICAN South Africa Subnational (sentinel)	2017 2019 30 m	Acinetobacter baumannii complex (focus on colistin resistant) (blood culture)	4822 1786 38	Illumina de novo central	MLST, SNP	✓	✓	✓	✓	✗	✓	✗	✓	Hospital and age for visualisation (NR)	Yes
Adams, 2020 [19]	AMERICAS Argentina National (sentinel, NRL)	2014 2017 48 m	NDM-producing Acinetobacter spp. (clinical)	40 40 40	Illumina de novo NR	MLST, SNP	✓	✓	✗	✗	✗	✓	✗	✗	Hospital but no cluster definition (threshold unclear)	Yes
Iovleva, 2022 [34]	AMERICAS USA Subnational (comprehensive for study network)	2017 2018 14 m	Carbapenem-resistant Acinetobacter baumannii (clinical and screening)	155 155 115	Illumina + ONT de novo central	MLST, SNP	✓	✓	✓	✓	✗	✓	✗	✗	Hospital, study site (≤ 10 SNPs)	Yes
Taylor, 2021 [56]	EUROPE UK Subnational (sentinel)	2016 2016 6 m	Amikacin- resistant Acinetobacter baumannii (clinical and screening)	271 16 16	Illumina de novo central	MLST	✓	✓	✓	✗	✗	✗	☑	✗	NR (NA)	No
Argimon, 2020 [22]	WEST PACIFIC Philippines National (sentinel)	2013 2014 24 m	Carbapenemase-producing Acinetobacter baumannii (clinical)	805 108 108	Illumina de novo central	MLST, SNP	✓	✗	✗	✗	✓	✓	✗	✗	Hospital, ward isolation date, HAI/CAI, age (threshold unclear)	Yes
Chilam, 2021 [26]	WEST PACIFIC Philippines National (sentinel)	2013 2014 24 m	Acinetobacter baumannii (focus on carbapenem-resistant) (clinical)	5254 445 117	Illumina de novo central	MLST, SNP	✓	✓	✓	✗	✓	✓	✗	✗	Hospital (threshold unclear)	Yes
Sherry, 2021 [53]	WEST PACIFIC Australia Subnational (unclear)	2017 2017 2 m	Carbapenem-resistant Acinetobacter spp. (clinical)	408 2 2	Illumina de novo central	MLST, SNP	✓	✓	✗	✗	✓	✓	✗	✓	Hospitalisation history, wards, admission dates (≤ 25 SNPs)	Yes
Forde, 2023 [30]	WEST PACIFIC Australia Subnational (comprehensive)	2017 2021 50 m	Carbapenem-resistant Acinetobacter baumannii (clinical and screening)	2660 35 35	Illumina de novo central	MLST, SNP	✓	✗	✗	✗	✓	✓	✗	✗	Hospital, HAI/CAI (≤ 5 SNPs/Mb)	Yes
Stratum 3: Methicillin-resistant Staphylococcus aureus
Lee, 2017 [40]	AMERICAS USA Subnational (comprehensive, Ten primary care clinics)	2007 2015 108 m	Staphylococcus aureus (MRSA and MSSA) (clinical and screening)	144 71 71	Illumina mapping NR	MLST, SNP	✓	✓	✓	✗	✗	✗	✗	✗	Place of sampling (NR)	Yes
Guthrie, 2020 [31]	AMERICAS Canada Subnational (sentinel)	2010 and 2016 24 m	Methicillin-Resistant Staphylococcus aureus (clinical)	426 426 426	Illumina mapping central	MLST, SNP	✓	✓	✓	✓	✓	✓	✗	✗	Age, hospital, HAI/CAI (≤ 10 SNPs)	Yes
Di Gregorio, 2023 [27]	AMERICAS South America International (sentinel)	2019 2019 7 m	Staphylococcus aureus (MRSA and MSSA) (blood culture)	443 182 165	Illumina de novo central	MLST, SNP	✓	✓	✓	✗	✓	✗	✗	✗	Country, HAI/CAI (threshold unclear)	Yes
Toleman, 2019 [57]	EUROPE UK Subnational (comprehensive)	2012 2013 12 m	Methicillin-resistant Staphylococcus aureus (MRSA) (blood culture)	425 425 425	Illumina de novo central	MLST, SNP	✓	✓	✗	✗	✓	✗	✗	✗	NR	Yes
Enger, 2022 [29]	EUROPE Norway National (comprehensive)	2008 2016 108 m	Methicillin-resistant Staphylococcus aureus (clinical and screening)	475 475 108	Illumina de novo central	MLST, SNP	✓	✓	✓	✗	✓	✗	☑	✓	Birth country and HAI/CAI for visualisation (NR)	Yes
van der Mee-Marquet, 2022 [58]	EUROPE France Subnational (sentinel)	2021 2021 8 m	Staphylococcus aureus (MRSA and MSSA) (blood culture)	53 5 5	Illumina NR	MLST, (SNP unclear)	✓	✓	✓	✓	✗	✓	✗	✗	NR	No
Masim, 2021 [45]	WEST PACIFIC Philippines National (sentinel)	2013 2014 24 m	Methicillin-resistant Staphylococcus aureus (clinical)	6211 3549 118	Illumina de novo external	MLST, SNP	✓	✓	✗	✗	✓	✓	✗	✗	Unclear (NR)	Yes
Sherry, 2021 [53]	WEST PACIFIC Australia Subnational (unclear)	2017 2017 2 m	Methicillin-resistant Staphylococcus aureus (clinical and screening)	408 88 88	Illumina de novo central	MLST, SNP	✓	✓	✗	✗	✓	✓	✗	✓	Hospitalisation history, wards, admission dates (≤ 15 SNPs)	Yes
Sherry, 2022 [54]	WEST PACIFIC Australia Subnational (comprehensive)	2017 2018 14 m	Methicillin-resistant Staphylococcus aureus (clinical and screening)	2275 811 811	Illumina de novo central	MLST, SNP	✓	✓	✗	✗	✓	✓	✗	✓	Hospitalisation history, wards, admission dates (≤ 15 SNPs)	Yes
Yamaguchi, 2022 [59]	WEST PACIFIC Japan National (comprehensive)	2010 2018 56 m**	Methicillin-resistant Staphylococcus aureus (clinical: skin and pus)	1770 1770 92	Illumina de novo external	MLST, SNP	✓	✓	✗	✗	✗	✗	✗	✗	Unclear (NR)	Yes
Forde, 2023 [30]	WEST PACIFIC Australia Subnational (comprehensive)	2017 2021 50 m	Methicillin-resistant Staphylococcus aureus (clinical and screening)	2660 620 620	Illumina de novo central	MLST, SNP	✓	✗	✗	✗	✓	✓	✗	✗	Hospital, HAI/CAI (≤ 5 SNPs/Mb)	Yes
Stratum 4: Vancomycin-resistant Enterococci
Kleinman, 2023 [36]	AMERICAS Canada National (sentinel)	2013 2018 72 m	Vancomycin-resistant Enterococcus ST1478 (blood culture)	115 115 115	Illumina de novo central	MLST, SNP	✓	✓	✓	✓	✓	✓	✗	✓	Patient movement data (hospital, ward) (< 20 SNVs)	No
Lee, 2020 [41]	WEST PACIFIC Australia National (sentinel)	2015 2017 36 m	Vancomycin-resistant Enterococcus faecium (blood culture)	1025 537 537	Illumina de novo central	MLST, SNP	✓	✓	✗	✓	✗	✗	✗	✗	Unclear (NR)	Yes
Sherry, 2021 [53]	WEST PACIFIC Australia Subnational (unclear)	2017 2017 2 m	vanA vancomycin- resistant Enterococcus faecium (clinical and screening)	408 64 64	Illumina de novo central	MLST, SNP	✓	✓	✗	✗	✓	✓	✗	✓	Hospitalisation history, ward, admission dates (≤ 25 SNPs)	Yes
Sherry, 2022 [54]	WEST PACIFIC Australia Subnational (comprehensive)	2017 2018 14 m	vanA vancomycin- resistant Enterococcus faecium (clinical and screening)	2275 346 346	Illumina de novo central	MLST, SNP	✓	✓	✗	✗	✓	✓	✗	✓	Hospitalisation history, wards, admission dates (≤ 25 SNPs)	Yes
Forde, 2023 [30]	WEST PACIFIC Australia Subnational (comprehensive)	2017 2021 50 m	Vancomycin-resistant enterococci (clinical and screening)	2660 433 433	Illumina de novo central	MLST, SNP	✓	✗	✗	✗	✓	✓	✗	✗	Hospital, HAI/CAI (≤ 5 SNPs/Mb)	Yes

Abbreviations: bp = base pair; cg = core genome; CSF = cerebrospinal fluid; Ec = Escherichia coli; HAI/ CAI = hospital- /community- acquired infection; Hosp. history = hospitalisation history; Kp = Klebsiella pneumoniae; m = month; med = medium; MLST = multi-locus sequence typing; MuGSI = Multi-site Gram-negative Surveillance Initiative; NR = not reported; PFGE = pulsed field gel electrophoresis; SCCmec = staphylococcal cassette chromosome mec; SNP = single nucleotide polymorphism; target = target pathogen (i.e. MRSA); target WGS = target pathogens submitted for WGS; VNTR = variable number tandem repeat; WGS = whole genome sequencing; WHO = World Health Organization

^$ “Sentinel” refers to surveillance in a selected sample of sites that do not cover all areas in the national or sub-national region; “comprehensive” refers to surveillance that is potentially including all sites within the covered area

*✓ Available; ☑ Available but reportedly < 80% complete; ✗ Not reported/not available

**Intermittent periods from 2010 to 2018: 2010 (February to September), 2012 (February 2012 to January 2013), 2014 (January to December), 2016

(January to December), and 2018 (January to December)

Time and surveillance periods

The first articles were published in 2017 on CRE/CPE and MRSA, and in 2020 for CRA/CPA and VRE with isolates dating back to 2007 and 2013, respectively. Reported surveillance periods varied between 2 and 144 months, with a median of 19 months for MRSA to 36 months for CRE/CPE and VRE (overall 24 months). Surveillance periods of two years or less often resulted from self-contained surveillance or pilot studies [24, 27, 44, 49, 53–56, 58]. Most publications with reporting durations longer than 2 years [19, 20, 23, 25, 29, 30, 32, 33, 35, 36, 38, 39, 41, 43, 47, 48, 51, 52, 59] used datasets from ongoing, continuous surveillance systems.

Summary of findings

Stratum 1: Carbapenem-resistant or carbapenemase-producing Enterobacterales (CRE)

For CRE surveillance, there were 25 publications from 17 countries in all six WHO regions: the African region (Madagascar [46], Nigeria [20]), the region of the Americas (United States of America (USA) [23, 39], Canada [35], Colombia [51]), the South East Asian region (India [47]), the European region (the United Kingdom (UK) [28, 56], Denmark [32, 50], Germany [33], Norway [43, 49], Spain [24]), the Eastern Mediterranean region (United Arabic Emirates (UAE) [55] Oman [21]) and the Western Pacific region (Philippines [22, 25], Australia [30, 38, 52], China [37, 42], and Singapore [44]). National level strategies were reported in 14 publications, operating in Canada [35], China [37], Colombia [51], Denmark [32, 50], Germany [33], India [47], Norway [43], Oman [21], Philippines [22, 25], Singapore [44], Spain [24] and UAE [55]. Eleven of the articles were based on sentinel, rather than comprehensive, surveillance within the subnational or national boundaries (Table 2).

In many cases, collection of data was initiated and coordinated by a national or subnational organisation for AMR surveillance purposes [20–23, 25, 28, 32, 33, 35, 38, 39, 43, 46, 47, 50–52]. These organisations mandated or encouraged participating hospital laboratories to supply relevant isolates, which were then sequenced centrally. In other cases, isolate collection appeared to have an academic or scientific background [24, 30, 37, 42, 44, 49, 55, 56]. Although these two categories did not indicate national or subnational coverage, academic studies tended to include less epidemiological information (63% of academic studies reported ≤ 3 epidemiological variables versus 47% of the reports of ongoing surveillance systems).

Availability and completeness of epidemiological information such as demographic, clinical and exposure data varied widely, with 13/25 studies [22, 24, 28, 30, 33, 35, 37, 42, 43, 46, 50, 51, 55] reporting ≤ 3 epidemiological variables, 7/25 studies [20, 25, 32, 39, 47, 49, 56] reporting 4 to 5, and 5/25 studies [21, 23, 38, 44, 52] reporting ≥ 6 (Table 2).

WGS was conducted for all CRE/CPE isolates in 20 of 25 studies [20–22, 24, 25, 28, 30, 32, 35, 38, 39, 43, 46, 47, 49–52, 55, 56], although this was sometimes unclearly reported. In others, selection criteria were applied, such as additional cephalosporin resistance [23], polymyxin resistance [42], or the availability of geographical information [33]. WGS was typically performed with Illumina technology and de novo assembly (22/25) whereas one study used a mapping approach (Table 2). One study conducted hybrid assemblies using Illumina and ONT reads for plasmid reconstruction [46], another ONT reads only for de novo assemblies [37] (Table 2). All studies employed MLST genotyping and 21/25 also analyses of higher resolution. SNP-based analyses were the method of choice in 17 studies including three, which also applied cgMLST analyses. Four studies conducted only cgMLST and one compared results from cgMLST, wgMLST and SNP-based analyses [21]. Epidemiological information on exposure (travel, ward or hospitalisation history) was given in 19 studies, enabling transmission analysis of genetic clusters. However, only 14 studies integrated this information for transmission or cluster analyses [20–22, 25, 30, 32, 33, 38, 39, 43, 44, 47, 52, 55]. These were generally based on links in sampling time, ward/hospital location [21, 25] or recent travel [32]). For six and five studies, respectively, this epidemiological information was either not reported or not used for cluster analysis. Genomic cluster thresholds were applied in 14 studies including seven that linked genetic relatedness with epidemiological data. In SNP-based analyses, these cut-offs were either species-specific with values ranging up to 100 SNPs for E. coli and up to 151 SNPs for K. pneumoniae or were applied irrespective of the species with thresholds of up to 5 SNPs or 5 SNPs/Mb, which considers the size of the reconstructed genomes. In cgMLST analyses, thresholds defining clonally related isolates were 10 and 15 allelic differences for E. coli and K. pneumoniae, respectively. Overall, 21/25 studies provided sequencing data in publicly available repositories.

Stratum 2: Carbapenem-resistant or carbapenemase-producing Acinetobacter spp. (CRA)

Regarding CRA surveillance, 8 publications arose from 6 countries in four WHO regions: the African region (South Africa [48]), the region of the Americas (USA [34], Argentina [19]), the European region (UK [56]) and the Western Pacific region (Philippines [22, 26] and Australia [30, 53]). Both articles from the Philippines [22, 26] and the Argentinian article [19] were at a national level. Within both national and subnational boundaries, 5/8 articles reported sentinel rather than comprehensive surveillance (Table 2).

In 5/8 publications, collection of data was initiated and co-ordinated for AMR surveillance purposes [19, 22, 26, 34, 48]. In other cases, the isolates appeared to be derived from a more academic organisation [30, 53, 56], which was conducting a study outside the remit of an ongoing surveillance system. All three academic studies were subnational, whereas 3/5 of the surveillance system reports were national.

Availability of epidemiological information ranged from 3/8 publications [19, 22, 30] reporting ≤ 3 epidemiological variables, 4/8 publications [26, 34, 53, 56] reporting 4 to 5, and only one article [48] reporting ≥ 6.

Six studies [19, 22, 30, 34, 53, 56] submitted all of their CRA isolates for WGS, whilst two applied additional selection criteria [26, 48]. Sequencing was exclusively conducted with Illumina technology followed by de novo assembly including one study, which in combination with ONT reads generated high-quality reference genomes for subsequent SNP calling. Besides MLST employed by all eight studies, SNP-based analyses were performed by seven studies whereas one reported no additional analyses. Transmission or clusters were inferred in 5/8 articles [19, 22, 30, 34, 53] using epidemiological data (such as geography [22]) and genetic information. Among these, three studies applied thresholds (i.e. 10 SNPs, ≤ 25 SNPs or ≤ 5 SNPs/Mb) to identify clonally related isolates. Except for one, all studies provided sequence data in public databases [22].

Stratum 3: Methicillin-resistant Staphylococcus aureus (MRSA)

Regarding MRSA surveillance, 11 publications arose from 8 countries and one multi-country collaboration (South America). These were from three WHO regions: the region of the Americas (USA [40], Canada [31] and South America [27]), the European region (UK [57], France [58] and Norway [29]) and the Western Pacific region (Philippines [45], Japan [59] and Australia [30, 53, 54]). National level surveillance was conducted in Norway [29], Japan [59] and the Philippines [45] and involving several countries (international level) in South America [27]. Four publications reported sentinel surveillance (Table 2).

In 5/11 articles [29, 31, 45, 57, 59], data collection was initiated and co-ordinated by an organisation for surveillance purposes. In other cases, the isolates were collected by an academic organisation, operating outside the remit of an ongoing surveillance system [27, 30, 40, 53, 54, 58].

Availability and completeness of epidemiological information varied with 4/11 articles [30, 40, 57, 59] reporting ≤ 3 epidemiological variables, 5/11 studies reporting 4 to 5 [27, 45, 53, 54, 58], and only 2/11 studies [29, 31] reporting ≥ 6.

In 8/11 publications all included MRSA isolates were submitted for WGS. In the others a random subset was selected for WGS [29] or specific criteria applied [45, 59]. WGS always involved short read sequencing using the Illumina technology followed by either de novo assembly (eight studies) or mapping (two studies). Predominant typing methods were MLST in combination with SNP-based analyses. Transmission or cluster analyses were reported in 6/11 publications [27, 30, 31, 40, 53, 54], generally based on the interaction of epidemiological factors (such as healthcare setting [31] or geography [40]). In 4/11 publications, cut-off values of ≤ 10 SNPs, ≤ 15 SNPs or ≤ 5 SNPs/Mb defined genomic clusters (Table 2). Generated sequencing data were submitted to public databases by 10/11 studies.

Stratum 4: Vancomycin-resistant Enterococci (VRE)

For VRE surveillance, there were 5 publications from two countries and WHO regions: the region of the Americas (Canada [36]), and the Western Pacific region (Australia [30, 41, 53, 54]). National level surveillance was conducted in one of the Australian [41] and the Canadian publication [36] that both collected isolates from sentinel sites. The three other Australian articles rather aimed at a comprehensive subnational coverage [30, 53, 54].

In 2/5 articles [36, 41] data collection was initiated and co-ordinated by a governmental surveillance programme. In the others [30, 53, 54], isolates were collected by an academic organisation.

Availability of epidemiological information varied, with 2/5 articles reporting 3 or fewer epidemiological variables [30, 41], 2/5 articles reporting 4 to 5 [53, 54], and one article reporting 6 or more [36].

In all publications all VRE isolates were submitted for centralised Illumina WGS and de novo assembly with subsequent genotyping using MLST and SNP-based analyses. Inferences on transmission or cluster analyses were reported in 4/5 publications [30, 36, 53, 54]. These were generally based on the interaction of epidemiological factors such as location [36], combined with genetic relatedness defined by thresholds of ≤ 20 SNVs, ≤ 25 SNPs or ≤ 5 SNPs/Mb. Sequence data were deposited in public databases by 4/5 articles (Table 2).

Discussion

Surveillance data including genomic and epidemiological information (integrated genomic surveillance, IGS) was published from all WHO regions. Publications on all four pathogen strata were found from the Region of the Americas and the Western Pacific Region. There were many publications included from the European Region, but none from Eastern Europe. Early and frequent publications originated from countries with high per capita gross domestic product, such as Australia, Singapore, USA, Denmark and UK, or were supported by external sequencing or grants, as was the case in studies from the Philippines. IGS was less often reported in the African, Eastern Mediterranean and South-East Asian Regions. Worldwide coverage varied across pathogen strata. Whilst the 25 reports on CRE crossed all 6 WHO regions, coverage was less global for CRA, MRSA and VRE. In particular, no publication on IGS of VRE outside Australia and Canada met our inclusion criteria, similarly for MRSA across three entire WHO regions. However, we have been made aware of additional surveillance reports in the European Region including IGS on MRSA and VRE that were not identified in the systematic literature search (see limitation section) [60–63]. Given the burden of VRE [64, 65] and MRSA [66, 67] worldwide, widespread surveillance is required. International partners and networks like the Centre for Genomic Pathogen Surveillance or the International Pathogen Surveillance Network [68] can help to establish WGS in LMICs [20, 22]. National WGS data can provide valuable insights but should be integrated with epidemiological data to maximize public health impact and cannot replace AST.

There were further differences between pathogen strata. The proportion of national surveillance ranged from 36% for MRSA to 56% for CRE. Nationwide infection control and antibiotic stewardship interventions are estimated to be more effective than facility-based or regional efforts [69]. There was a similar range of proportions opting for a sentinel approach, from 36% in the MRSA stratum to 63% in the CRA stratum. A sentinel surveillance can provide a nationally representative isolate selection, but was often limited to higher-level healthcare or specific subpopulations.

The reported surveillance does not necessarily reflect the underlying surveillance system. First, only 29/41 publications (71%) describe an underlying surveillance system. Second, surveillance was often enhanced for the reported study period, e.g. by collecting additional case information or conducting WGS on top of traditional pathogen surveillance. Third, WGS was often performed retrospectively so that results could not timely inform the involved stakeholders. For real-time actionability, sequencing needs to be performed with a short sample to feedback-time and case data collection performed routinely.

Regarding epidemiological data, many publications merely included basic demographic information and place of sample collection. Between 36% and 52% of publications in each pathogen stratum only included 3 or less of the 8 epidemiological variables that we extracted. Even if available and reported, exposure information was often not integrated with the genomic data and the integrated cluster or transmission analyses tended to be superficially reported or not reported at all. Integrated cluster analysis was particularly lacking in CRE and MRSA (44 and 45% of articles) and less so in VRE (20% of articles). Nonetheless, some studies applied thresholds for cluster identification and integrated epidemiological information on time (of sampling, etc.), place (ward or hospital, etc.) and person (e.g. age or diagnosis) for transmission analyses. This information is crucial to establish the existence of an outbreak and hint towards the source, e.g. specific procedures in the hospital setting or food items in the outpatient setting. The patient outcome needs to supplement the genetic information when assessing virulence. Reasons for the persistent lack of epidemiologic data integration are not covered in this review but might reflect the lack of beneficial experiences with this relatively new technology and lacking governance for integrated surveillance in the context of a historical coexistence of laboratory-based pathogen surveillance and clinical or public health epidemiology. This could be overcome by collaborative efforts between academic research, hospitals and public health authorities (PHAs) to increase genetic and epidemiologic data completeness and showcase the benefits of integrated analysis. On the technical side, genomic surveillance seems to be often built on existing surveillance systems with separated data storage (siloed IT) and detailed clinical/ epidemiological information often only at hospital level. The enhanced capabilities to detect supraregional outbreaks increase the need to share information between different levels and surveillance systems. Exchange of detailed case information on exposure and outcome from hospitals and timely cluster information will improve integrated outbreak analyses on all levels. In practice, relevant epidemiological data should routinely accompany the isolate or sequence or pre-defined forms or interfaces could be used for enhanced surveillance of cluster isolates. Additionally, academic research can refine integrated analyses as several innovative articles show [30, 34, 38, 39, 52] and close potential gaps in routine surveillance for example regarding pathogens or plasmid outbreaks. Interdisciplinary teams and joint training of public health microbiologists and field epidemiologists [70] as well as providing PHAs with sequences or funding for sequencing could increase data integration. One of the included studies [38] has demonstrated a decrease in KPC-2-producing Enterobacterales infections after a comprehensive prospective genomic and epidemiological surveillance and response program for the control of CPE was released. Additionally, the economic impact of genomic surveillance of ARB on national level should be evaluated further [71, 72].

Integration of animal or environmental samples has not been reported by any included publication. Sequencing of antibiotic-resistant bacteria in animal and food needs to be increased and intersectoral collaboration fostered by political stakeholders [73] in order to identify relevant sources of strain dissemination and outbreaks among humans.

WGS methodology was homogenous across strata, mainly using Illumina short-read technology with a preference for de novo assembly. Typing methods mostly included MLST, SNP-based analyses, and cgMLST for CRE. Reported cluster thresholds showed a wide range of values within and between species, reflecting the lack of consensus definitions for clonality between closely related isolates. Across pathogen strata, cut-off values for SNP-based analyses to define genomic clusters ranged ≤ 25 SNPs for CRA, VRE, and MRSA, whereas for CRE a general threshold may not be applicable given the heterogeneity of species. SNP-based analyses typically rely on mapping reads to a given reference. Sequencing errors may appear as false-positive SNPs and mistakenly increase genetic differences. Similarly, the choice of the reference genome profoundly effects the results [74]. A reference sequence phylogenetically too divergent may show genomic characteristics which are not present in the isolates under study. Thus, the absence of SNPs in accessory genomic regions may lead to erroneous homogeneity between isolates, ultimately causing misleading epidemiological inferences. Although beyond the scope of this systematic review, we emphasize the need for international standards, particularly focusing on quality metrics, on how to analyse WGS data, such as percentage of reference sequence covered by isolates in SNP-based analyses. Surveillance systems harmonized in this way worldwide would ensure comparability of data for cross-border evaluation of ARB. Almost all studies in this review shared generated sequence data via public databases which is another important pre-condition for international investigations.

This systematic review is strengthened by the precisely defined and focused protocol. Almost all criteria for inclusion and methodological procedures were decided pre-hoc. The limited initial secondary verification of full-text eligibility assessment and data extraction by the SRC was post-hoc identified as a weakness and therefore supplemented by secondary verification through the FAI. Due to different interpretation of the inclusion criteria ‘detailed description’ of epidemiological data, eight articles [20, 22, 24, 30, 36, 39, 41, 59] were added based on the additional structured full-text screening and suggestion of the FAI after the original paper selection by the SRC, and the initial draft summary. Potential selection bias was minimised by agreement on a common interpretation and final inclusion decision by the SRC.

A risk of bias assessment was deemed irrelevant due to descriptive data. The review did not involve measurement of performance or efficacy and bias can only be manifested by a displacement of a measurement from its ‘true’ state. We were unable to find any useful quality appraisal tools focussing on the reporting quality of papers, even from the Enhancing The Quality and Transparency of Health Research (EQUATOR) network [75], where reporting guidelines were not available for descriptive papers. Formal quality appraisal of the papers was thus omitted. Readers are urged to conduct appropriate checks on relevant studies.

The pre-hoc defined search strategy showed some limitations. After conducting the review, we identified that studies mentioning only carbapenemase-production but not resistance were not found [76, 77]. Also, national surveillance reports were not captured although we searched grey literature sources [60–63]. We anticipated not to include surveillance systems without published results or lacking/incomplete description of the methodology due to the search for published studies and the inclusion criteria. Additionally, inference from the published surveillance results to an underlying surveillance system was not always possible and often challenging. However, the evidence base that has been surveyed is likely to provide an overview of current methodology of IGS systems of ARB with peer-reviewed articles.

For a more comprehensive mapping of all existing IGS systems and their characteristics, we recommend additional screening of the grey literature including national surveillance reports and descriptions of IGS systems, as well as a mapping and direct survey of stakeholders.

Conclusion

Surveillance of CRE, CRA, MRSA and VRE, integrating WGS with epidemiological information, has been reported globally since 2017 and suggest that WGS is increasingly embedded within operational outbreak detection frameworks. Nevertheless, surveillance appears patchy, with potential for broader national and global coverage, in particular for VRE and MRSA. There is general agreement on technical aspects such as sequencing platforms, de novo assembly and sequence typing methods. There is, however, variability in the extent to which epidemiological variables are collected. In many cases, exposure information was not reported or not integrated in the analysis of WGS data. In order to use the potential of WGS to help identify outbreaks and transmission pathways, epidemiological and WGS information have to be adequately integrated into surveillance. This would support Public Health professionals in implementing adequate measures to contain current and prevent future outbreaks.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ARB

Antibiotic-resistant bacteria

AST

Antibiotic susceptibility testing

bp

base pair

CADTH

Canada’s Drug and Health Technology Agency

cg

core genome

CPA

Carbapenemase-producing Acinetobacter spp

CPE

Carbapenemase-producing Enterobacterales

CRA

Carbapenem-resistant Acinetobacter spp

CRE

Carbapenem-resistant Enterobacterales

Ec

Escherichia coli

FAI

Funding Academic Institution

HAI/ CAI

Hospital- /community- acquired infection

Kp

Klebsiella pneumoniae

MeSH

Medical Subject Headings

MLST

Multi-locus sequence typing

MRSA

Methicillin-resistant Staphylococcus aureus

MuGSI

Multi-site Gram-negative Surveillance Initiative

NR

Not reported

ONT

Oxford Nanopore Technology

PHA

Public health authority

PFGE

Pulsed field gel electrophoresis

SCCmec

Staphylococcal cassette chromosome mec

SNP

Single nucleotide polymorphism

SRC

Ssystematic-reviewing company

UAE

United Arabic Emirates

UK

United Kingdom

USA

United States of America

VNTR

Variable number tandem repeat

VRE

Vancomycin-resistant Enterococci

wg

whole genome

WGS

Whole genome sequencing

WHO

World Health Organization

Author contributions

F.R. finalised the protocol, assisted with paper selection and data extraction/synthesis, wrote the final draft of the article and led the project. M. P. drafted the protocol, screened the abstracts, selected publications, extracted data, synthesised data, wrote the first draft of the article and revised the final draft of the article. M. S. finalised the protocol, assisted with paper selection and data extraction/synthesis, and revised the final draft of the article. S. H. finalised the protocol, assisted with paper selection and data extraction/synthesis, and revised the final draft of the article. D. S. finalised the protocol, assisted with paper selection and data extraction/synthesis, and revised the final draft of the article. T. E. assisted with data interpretation and revised the final draft of the article. K. M. screened the abstracts, selected publications, assisted with the writing of the first draft of the article and revised the final draft of the article. C. N. helped to draft the protocol, devised the search strategies, carried out the searches, assisted with the writing of the first draft of the article and revised the final draft of the article. R. W. assisted with the drafting of the protocol, assisted with the writing of the first draft of the article and revised the final draft of the article. J. B. H. assisted with data extraction/synthesis, and wrote the WGS parts of the final draft of the article.

Funding

Open Access funding enabled and organized by Projekt DEAL. Funding was provided by the Robert Koch Institute (WGS_232).

Data availability

All data generated or analysed during this study are included in this published article [and its additional files].

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.