# Choosing the Right Resistance Surveillance Database ## Quick Answer - Match the database to your specific monitoring objective by evaluating geographic scope, specimen types, and reporting standards before committing to a system. - GLASS provides standardized global human clinical isolate data, while EARS-Net offers European invasive isolate surveillance with distinct de-duplication rules. - No single database covers all One Health domains, and pharmacovigilance reports do not reliably replicate conventional surveillance findings. ## Understanding Antimicrobial Resistance Surveillance Databases Antimicrobial resistance surveillance databases are structured repositories that collect, store, and disseminate susceptibility testing results, resistance phenotype data, and associated epidemiological metadata. These systems enable public health professionals, researchers, and laboratory personnel to track resistance trends, detect emerging threats, and evaluate the effectiveness of stewardship interventions. The World Health Organization defines antimicrobial resistance as the ability of microorganisms to resist the effects of drugs that once treated them effectively, making surveillance a cornerstone of global stewardship efforts and public health policy development. The selection of an appropriate surveillance database depends on multiple factors including the target pathogen, specimen types, geographic coverage, data standards, and analytical capabilities. Different databases employ distinct methodologies for isolate collection, de-duplication, susceptibility testing, and data reporting. Understanding these methodological differences is essential for interpreting resistance data correctly and making valid comparisons across time periods or geographic regions. Laboratory professionals must recognize that surveillance databases are not interchangeable. Each system was designed with specific objectives, whether tracking hospital-acquired infections, monitoring community pathogens, or supporting national reporting obligations. The choice of database influences data quality, comparability, and the conclusions that can be drawn from resistance trend analyses. A database that excels at tracking invasive bloodstream infections may be poorly suited for monitoring community-acquired urinary tract infections, and vice versa. ## Core Principles of Resistance Surveillance ### Standardized Susceptibility Testing Reliable surveillance depends on consistent antimicrobial susceptibility testing methods across participating laboratories. The Clinical and Laboratory Standards Institute publishes standards-based methodologies that define testing procedures, quality control requirements, and interpretive criteria for susceptibility testing. Laboratories contributing data to surveillance systems must adhere to recognized testing standards to ensure that minimum inhibitory concentration values and categorical interpretations are comparable across sites and over time. Standardization extends beyond the testing method itself to include culture conditions, inoculum preparation, media selection, and incubation parameters. Variations in any of these factors can alter susceptibility results and introduce bias into surveillance data. Laboratories should document their testing protocols and participate in external quality assessment programs to verify ongoing accuracy. The American Society for Microbiology provides professional education and laboratory practice resources that support workforce development in clinical microbiology, helping ensure that personnel performing susceptibility testing maintain current competency. ### Isolate Selection and De-duplication Surveillance systems employ different rules for selecting which isolates to include in their databases. De-duplication refers to the removal of repeated isolates from the same patient to prevent overrepresentation of persistent infections. The World Health Organization Global Antimicrobial Resistance Surveillance System requires selecting only the first isolate of a given bacterial species per patient per surveillance period per specimen type, with stratification by age group, gender, and infection origin. A comparative study of de-duplication methods between GLASS and the Japan Nosocomial Infections Surveillance system demonstrated that the number of patients tabulated per specimen type was consistently smaller when using the JANIS method, which removes repeated isolates of the same species within 30 days regardless of specimen type while distinguishing isolates with changes in resistance phenotype. Despite these differences, the percentage of resistant Escherichia coli isolates differed by less than three percent between the two methods for any antibiotic tested in both outpatient and inpatient settings. This finding suggests that while de-duplication rules affect isolate counts, resistance percentages may be relatively robust to these methodological variations. ### Data Quality and Validation Surveillance databases depend on accurate species identification and reliable susceptibility results. Reference culture collections provide authenticated bacterial strains that laboratories can use for quality control and method validation. The American Type Culture Collection maintains characterized strains with documented phenotypic properties that support reproducible testing and verification of laboratory procedures. Data validation procedures vary among surveillance systems. Some databases perform automated checks for improbable results, while others rely on laboratory-level quality assurance before data submission. Researchers should understand the validation mechanisms of each database to assess data reliability appropriately. The presence of documented quality assurance programs, participation in external quality assessment schemes, and transparent reporting of data completeness metrics all contribute to the overall reliability of a surveillance database. ## Major Surveillance Databases Compared ### WHO Global Antimicrobial Resistance Surveillance System GLASS represents the World Health Organization's standardized approach to global antimicrobial resistance surveillance. This system collects data from participating countries using standardized case definitions, specimen types, and reporting formats. GLASS focuses on human clinical isolates from blood, urine, stool, and other specified specimen sources, with particular attention to priority pathogens including Escherichia coli, Klebsiella pneumoniae, Staphylococcus aureus, and Streptococcus pneumoniae. The GLASS methodology emphasizes national coordination and laboratory capacity building. Participating countries designate national reference laboratories and follow standardized protocols for isolate collection, susceptibility testing, and data submission. The system supports trend analysis and international comparison while acknowledging variations in laboratory capacity across member states. For researchers seeking global comparability and standardized case definitions, GLASS provides a framework that facilitates cross-country comparisons, though the depth of data may vary depending on the participating country's surveillance infrastructure. ### European Antimicrobial Resistance Surveillance Network EARS-Net operates under the coordination of the European Centre for Disease Prevention and Control, collecting data from hospital laboratories across European Union member states and other participating countries. This network focuses on invasive isolates from blood and cerebrospinal fluid, providing a specific perspective on severe infections requiring hospitalization. The EARS-Net methodology includes defined de-duplication rules and standardized susceptibility testing protocols. Participating laboratories submit data on priority pathogen-antimicrobial combinations, enabling annual reporting of resistance percentages and trend analyses. The network's focus on invasive isolates distinguishes it from systems that include all clinical specimen types. Researchers studying severe, hospital-associated infections in Europe will find EARS-Net particularly relevant, while those interested in community-acquired infections or non-invasive specimens may need to supplement EARS-Net data with other sources. ### NCBI Pathogen Detection The National Center for Biotechnology Information maintains a pathogen detection system that integrates genomic and epidemiological data for foodborne pathogens and other clinically significant bacteria. This database links whole-genome sequencing data with isolate metadata, enabling phylogenetic analysis and resistance gene detection at the molecular level. NCBI Pathogen Detection serves a different purpose than phenotypic surveillance databases. While traditional surveillance systems report susceptibility testing results, this genomic database supports cluster detection, outbreak investigation, and the identification of resistance determinants. Researchers interested in resistance mechanisms instead of phenotypic trends may find this database more appropriate for their objectives. The database's utility depends on the submission of isolates for sequencing, which may introduce selection bias, as not all clinical isolates undergo genomic characterization. ### Pharmacovigilance Databases Pharmacovigilance systems such as VigiBase for human adverse event reports and EudraVigilance Veterinary for animal products have been proposed as supplementary sources for antimicrobial resistance surveillance. A comparison of antibiotic resistance reports in pharmacovigilance databases against conventional surveillance data found consistent statistically significant differences in the distribution of cases across antibiotic classes. The analysis identified 26,086 reports from 91 countries in VigiBase and 1,010 cases from 18 countries in EudraVigilance Veterinary. The distribution of antibiotic resistance cases differed significantly between pharmacovigilance and conventional surveillance for at least one Anatomical Therapeutic Chemical class grouping. These findings indicate that pharmacovigilance data cannot substitute for dedicated resistance surveillance systems, though they may provide complementary information about adverse outcomes associated with resistant infections. Researchers should treat pharmacovigilance data as a supplementary signal detection tool instead of a primary surveillance source. ## At a Glance | Database | Geographic Scope | Primary Specimen Types | De-duplication Method | Data Type | Best Suited For | |----------|------------------|----------------------|----------------------|-----------|-----------------| | GLASS | Global | Blood, urine, stool, specified clinical specimens | First isolate per patient per period per specimen type with stratification | Phenotypic susceptibility | National and international resistance trend comparison | | EARS-Net | Europe | Invasive isolates from blood and cerebrospinal fluid | Defined network-specific rules | Phenotypic susceptibility | Hospital-acquired severe infection monitoring | | NCBI Pathogen Detection | Global | Foodborne and clinical isolates with genomic data | Genomic cluster-based | Whole-genome sequencing and resistance genes | Outbreak investigation and resistance mechanism analysis | | Pharmacovigilance (VigiBase) | Global | Adverse event reports | Report-based | Spontaneous reporting | Supplementary signal detection, not primary surveillance | ## Selecting a Database for Specific Monitoring Needs ### Defining Surveillance Objectives The first step in database selection involves clarifying the specific monitoring objective. Researchers tracking community-acquired urinary tract infections require different data than those investigating hospital-acquired bloodstream infections. Public health authorities preparing national reports may need comprehensive data across multiple specimen types, while researchers studying resistance mechanisms may prioritize genomic databases. The surveillance objective determines the required data elements, geographic coverage, and temporal resolution. A study examining regional resistance trends requires consistent data collection over multiple years, while an outbreak investigation demands real-time or near-real-time data access. These different requirements point toward different database choices. For example, a researcher investigating the molecular basis of carbapenem resistance in Enterobacterales would find NCBI Pathogen Detection more useful than a purely phenotypic surveillance system, while a public health authority tracking national resistance trends would rely more heavily on GLASS data. ### Matching Pathogen and Specimen Priorities Each surveillance database emphasizes particular pathogens and specimen types. GLASS includes a defined list of priority pathogens with specified antimicrobial panels for susceptibility testing. EARS-Net focuses on invasive isolates, providing data specifically relevant to severe infections. NCBI Pathogen Detection includes isolates with genomic data, which may not represent the full spectrum of clinical isolates in a region. Laboratory professionals should verify that the database of interest includes their target pathogens and specimen types. A database that does not collect data on a specific pathogen cannot support conclusions about that pathogen's resistance patterns, regardless of the database's overall size or reputation. For instance, a researcher studying resistance in Acinetobacter baumannii should confirm that the chosen database includes this organism and collects data from relevant specimen types before proceeding with analysis. ### Evaluating Data Accessibility Data access policies vary substantially among surveillance databases. Some systems provide open access to aggregated data through public portals, while others require formal data sharing agreements or restrict access to participating institutions. Researchers should review data use agreements and publication requirements before committing to a particular database. The level of data granularity also varies. Some databases provide isolate-level data with detailed metadata, while others report only aggregated resistance percentages. Isolate-level data supports more sophisticated analyses including risk factor assessment and multivariable modeling, but may require additional data use approvals. Researchers should weigh the analytical benefits of isolate-level data against the administrative burden of obtaining access approvals. ## Practical Implementation Steps ### Step 1: Document the Surveillance Question Write a formal surveillance question that specifies the target pathogen, antimicrobial agents of interest, patient population, specimen types, geographic region, and time period. This documentation guides database selection and prevents scope creep during the analysis phase. A well-defined surveillance question also facilitates communication with database coordinators and data providers. ### Step 2: Inventory Available Data Sources Create a table listing candidate databases with their geographic coverage, pathogen scope, specimen types, data access policies, and update frequency. Include both conventional surveillance systems and supplementary sources such as pharmacovigilance databases when relevant to the surveillance question. This inventory provides a structured basis for comparing options and documenting the selection process. ### Step 3: Assess Methodological Compatibility Review the susceptibility testing standards, de-duplication rules, and data validation procedures for each candidate database. Verify that the testing methodology aligns with your laboratory's protocols or that you can interpret results appropriately given methodological differences. The Clinical and Laboratory Standards Institute provides standards-based methodologies that define testing procedures and interpretive criteria, which can serve as a reference point for assessing methodological compatibility. ### Step 4: Evaluate Data Quality Indicators Examine available quality indicators for each database, including participation in external quality assessment programs, data completeness metrics, and validation procedures. Databases with documented quality assurance programs provide more reliable data for research and public health decision-making. Reference culture collections such as the American Type Culture Collection provide authenticated strains that support quality control verification. ### Step 5: Test Data Accessibility Download sample data or explore the query interface for each candidate database before committing to a full analysis. Verify that the data structure supports your planned analytical approach and that the level of detail meets your requirements. This testing phase can reveal practical issues such as data format limitations, incomplete metadata, or restricted fields that may not be apparent from documentation alone. ### Step 6: Document Selection Rationale Record the reasons for selecting a particular database, including the surveillance question, methodological compatibility, data quality assessment, and accessibility considerations. This documentation supports reproducibility and provides context for interpreting results. The selection rationale should be included in any publications or reports that use the surveillance data. ## Records and Measurements ### Essential Records for Database Selection Maintain documentation of the surveillance question, candidate database inventory, methodological comparison, and selection rationale. Record the date of data access and the specific data version or extraction date, as surveillance databases update regularly and analyses must reference the exact data snapshot used. Document any data transformations applied during analysis, including filtering criteria, de-duplication steps, and handling of missing or indeterminate susceptibility results. These records support result verification and enable other researchers to reproduce the analysis. The level of documentation should be sufficient for an independent researcher to understand what data was used and how it was processed. ### Measurements for Data Quality Assessment Quantify data completeness by calculating the proportion of isolates with complete susceptibility results for each antimicrobial agent. Assess temporal consistency by examining the number of isolates reported per year and identifying any gaps or abrupt changes that might indicate reporting artifacts instead of true resistance trends. Compare resistance percentages across databases for overlapping pathogens and regions to identify potential discrepancies. The comparison of de-duplication methods between GLASS and JANIS demonstrated that methodological differences can produce small but measurable differences in resistance percentages, highlighting the importance of understanding each database's specific rules. When discrepancies exceed expected methodological variation, investigate potential causes such as differences in patient populations, testing practices, or data validation procedures. ### Database Comparison Metrics Table | Metric | GLASS | EARS-Net | NCBI Pathogen Detection | VigiBase | |--------|-------|----------|------------------------|----------| | Data granularity | Isolate-level with patient stratification | Isolate-level from hospital laboratories | Sequence-level with isolate metadata | Report-level adverse events | | Update frequency | Annual reporting cycles | Annual reporting cycles | Continuous as sequences are submitted | Continuous as reports are filed | | Access model | Country-coordinated submission | Network participation required | Open genomic data portal | Restricted research access | | Primary antimicrobial data | Phenotypic susceptibility | Phenotypic susceptibility | Genotypic resistance determinants | Adverse event reports with resistance terms | | Quality assurance | National reference laboratory coordination | External quality assessment participation | Sequence quality metrics | Spontaneous reporting system | ## Common Failure Patterns ### Inappropriate Cross-Database Comparisons Researchers frequently compare resistance percentages across databases without accounting for methodological differences. Variations in de-duplication rules, specimen types, testing standards, and patient populations can produce apparent resistance differences that reflect methodology instead of true epidemiological variation. The study comparing GLASS and JANIS de-duplication methods found differences in patient counts per specimen type, though resistance percentages differed by less than three percent for E. coli. To avoid this failure, document the methodological characteristics of each database before making comparisons and interpret resistance differences in light of these factors. When possible, use databases with similar methodologies for direct comparisons, or apply appropriate adjustments to account for methodological differences. ### Ignoring Specimen Type Differences Surveillance databases that include multiple specimen types may combine isolates with different clinical significance. Bloodstream isolates generally represent invasive disease, while urine isolates may include asymptomatic colonization. Combining these specimen types without stratification can obscure clinically meaningful resistance patterns. Researchers should stratify analyses by specimen type whenever the surveillance question involves clinical significance or infection severity. EARS-Net's focus on invasive isolates from blood and cerebrospinal fluid provides a clear example of how specimen type selection shapes the interpretation of surveillance data. ### Overlooking Temporal Changes in Testing Standards Susceptibility testing interpretive criteria change over time as new evidence emerges. Resistance percentages from different time periods may not be directly comparable if breakpoints have changed. Researchers should verify that the database provides information about the interpretive criteria applied to historical data. When analyzing multi-year trends, check whether the database documents breakpoint changes and consider analyzing data within periods of consistent interpretive criteria. The Clinical and Laboratory Standards Institute regularly updates its standards, and these updates can affect the classification of isolates as susceptible, intermediate, or resistant. ### Assuming Completeness of Genomic Databases Genomic surveillance databases such as NCBI Pathogen Detection depend on isolates submitted for sequencing, which may not represent the full spectrum of clinical isolates in a region. Submission biases can affect the apparent prevalence of resistance genes and the phylogenetic diversity captured in the database. Researchers using genomic databases should recognize that the absence of a resistance gene in the database does not confirm its absence in the clinical population. The presence of a resistance gene also does not always correlate with phenotypic resistance, as gene expression, mutation effects, and regulatory mechanisms influence the relationship between genotype and phenotype. ### Misinterpreting Pharmacovigilance Signals Pharmacovigilance databases capture spontaneous reports that reflect reporting behaviors and adverse event detection instead of systematic surveillance. The comparison of pharmacovigilance data with conventional surveillance found consistent statistically significant differences in the distribution of antibiotic resistance cases, indicating that these systems capture different phenomena. Researchers should not use pharmacovigilance data to estimate resistance prevalence or trends. ## Quality Controls and Laboratory Standards ### External Quality Assessment Participation Laboratories contributing to surveillance databases should participate in external quality assessment programs to verify the accuracy of their susceptibility testing results. These programs distribute characterized strains with known susceptibility profiles, allowing laboratories to compare their results against expected values and identify systematic errors. External quality assessment participation provides an objective measure of laboratory performance and supports the reliability of surveillance data. Laboratories should maintain records of their participation and performance to document data quality for surveillance submissions. ### Reference Strain Verification The American Type Culture Collection maintains authenticated reference strains with documented susceptibility characteristics. Laboratories should use these strains for routine quality control of susceptibility testing methods and for verifying the performance of new reagent lots or equipment. Reference strain verification should be documented and reviewed regularly to ensure ongoing accuracy. The use of authenticated strains with known phenotypic properties supports reproducible testing and provides a baseline for comparing results across time periods and laboratory personnel. ### Biosafety Considerations Laboratory work with antimicrobial-resistant bacteria requires appropriate biosafety practices. The Biosafety in Microbiological and Biomedical Laboratories publication from the CDC and NIH provides risk assessment guidance and containment recommendations for work with infectious agents. Laboratories should conduct risk assessments for the specific pathogens they handle and implement appropriate biosafety levels. Biosafety considerations extend beyond the laboratory to include specimen transport, waste disposal, and decontamination procedures. The Biosafety in Microbiological and Biomedical Laboratories framework supports the safe handling of resistant organisms throughout the surveillance workflow, from specimen collection to data submission. ### Professional Competency The American Society for Microbiology provides professional education and laboratory practice resources that support workforce development in clinical microbiology. Laboratories should ensure that personnel performing susceptibility testing and contributing to surveillance databases have appropriate training and maintain current competency. Ongoing professional development supports accurate testing, proper data interpretation, and adherence to evolving standards. Laboratories should document personnel training and competency assessments as part of their quality management systems. ## Evidence Limitations and Interpretation ### Surveillance Data Represent Testing Practices Surveillance databases reflect the testing practices of participating laboratories, which may vary in their decisions about which isolates to test and report. Some laboratories test all clinical isolates, while others test only isolates from specific specimen types or patient populations. These testing practices influence the resistance percentages reported in surveillance databases. Researchers should consider whether testing practices in the database align with their surveillance question. A database that includes only isolates from patients with suspected resistant infections will report higher resistance percentages than one that includes all clinical isolates, even when the underlying bacterial populations are identical. ### Under-Reporting and Selection Bias Participation in surveillance systems is voluntary in many regions, and laboratories with limited resources may under-report or submit data inconsistently. The absence of data from certain regions or facility types can bias resistance estimates and limit the generalizability of surveillance findings. Researchers should assess the representativeness of the database relative to their target population and acknowledge limitations in generalizability when interpreting results. The World Health Organization acknowledges variations in laboratory capacity across member states participating in GLASS, which affects the completeness and comparability of submitted data. ### Pharmacovigilance Data Limitations Pharmacovigilance databases capture adverse event reports instead of systematic surveillance data. The comparison of pharmacovigilance and conventional surveillance data found consistent differences in antibiotic distribution across databases, indicating that pharmacovigilance reports do not replicate conventional surveillance findings. These systems may serve as supplementary signal detection tools but cannot replace dedicated resistance surveillance. The analysis of VigiBase identified 26,086 reports from 91 countries, while EudraVigilance Veterinary identified 1,010 cases from 18 countries. The distribution of antibiotic resistance cases differed significantly between pharmacovigilance and conventional surveillance for at least one Anatomical Therapeutic Chemical class grouping, supporting the conclusion that these data sources capture different phenomena. ### Genomic Data Interpretation Challenges Whole-genome sequencing data can identify resistance genes, but the presence of a resistance gene does not always correlate with phenotypic resistance. Gene expression, mutation effects, and regulatory mechanisms influence the relationship between genotype and phenotype. Researchers using genomic databases should verify resistance gene predictions against phenotypic susceptibility data when available. The integration of genomic and phenotypic data provides a more complete picture of resistance than either approach alone. NCBI Pathogen Detection links whole-genome sequencing data with isolate metadata, supporting phylogenetic analysis and resistance gene detection, but researchers should recognize the limitations of genomic prediction and the potential for submission bias in genomic databases. ## Safety and Regulatory Context ### Laboratory Biosafety Requirements Working with antimicrobial-resistant bacteria requires adherence to established biosafety guidelines. The Biosafety in Microbiological and Biomedical Laboratories publication provides the foundational risk assessment framework for clinical and research laboratories. Institutions should implement containment practices appropriate for the risk group of the organisms being handled. Biosafety considerations should be integrated into the surveillance workflow, from specimen collection and transport to laboratory testing and data submission. The Biosafety in Microbiological and Biomedical Laboratories framework supports risk assessment, containment, and decontamination principles that protect laboratory personnel and the surrounding community. ### Data Use and Confidentiality Surveillance databases may contain patient-level data subject to privacy protections. Researchers must comply with applicable data use agreements and confidentiality requirements when accessing and analyzing surveillance data. Aggregated data generally presents fewer privacy concerns than isolate-level data with detailed patient metadata. Researchers should review data use agreements carefully and ensure that their planned analyses comply with all restrictions. When working with isolate-level data, researchers should minimize the collection and retention of identifying information and follow institutional data security requirements. ### Professional Escalation Criteria Laboratory professionals should escalate concerns about surveillance data quality to appropriate authorities when they identify systematic errors, unexplained data gaps, or discrepancies that could affect public health decision-making. Escalation may involve contacting the database coordinator, national reference laboratory, or public health authority responsible for the surveillance system. Escalation is appropriate when data quality issues could lead to incorrect conclusions about resistance trends or inappropriate treatment recommendations. Timely escalation supports the integrity of surveillance systems and protects public health decision-making. ## Treatment Context for Resistant Infections ### Clinical Implications of Resistance Surveillance Resistance surveillance data informs treatment guidelines and clinical decision-making. The ESCMID guidelines for infections caused by multidrug-resistant Gram-negative bacilli address targeted antibiotic treatment for third-generation cephalosporin-resistant Enterobacterales and carbapenem-resistant organisms. These guidelines were developed through systematic review of randomized controlled trials and observational studies, with recommendations classified by GRADE methodology. The guidelines examined head-to-head comparisons between individual antibiotics and between monotherapy and combination therapy regimens. The primary outcome was all-cause mortality at 30 days, with secondary outcomes including clinical failure, microbiological failure, development of resistance, relapse, adverse events, and length of hospital stay. Many comparisons were addressed in small observational studies at high risk of bias, limiting the certainty of evidence for specific treatment recommendations. ### Enteric Fever Treatment Considerations The Cochrane Review of cephalosporins for enteric fever provides context for resistance surveillance in typhoid and paratyphoid fever. The World Health Organization recommends treatment with azithromycin, ciprofloxacin, or ceftriaxone due to widespread resistance to older first-line antimicrobials. Resistance patterns vary by location and change over time, with fluoroquinolone resistance in South Asia often precluding ciprofloxacin use and extensively drug-resistant strains emerging in Pakistan. The review noted that susceptibility to older first-line antimicrobials such as chloramphenicol has reappeared in some areas. This observation underscores the importance of ongoing resistance surveillance to detect changing susceptibility patterns and inform treatment recommendations. Surveillance data that captures these shifts in susceptibility can directly influence clinical decision-making and public health policy. ## One Health Surveillance Considerations ### Human and Animal Data Integration Antimicrobial resistance surveillance spans human medicine, animal health, and environmental monitoring. The comparison of pharmacovigilance and conventional surveillance data across One Health domains identified no suitable data sources for environmental health, highlighting gaps in current surveillance coverage. Researchers interested in environmental resistance should recognize that existing databases may not capture this domain adequately. The absence of environmental surveillance data represents a significant limitation in understanding the full scope of antimicrobial resistance transmission. Researchers should acknowledge this gap when designing studies and interpreting findings from human and animal surveillance systems. ### Cross-Sectoral Data Comparability Data from human and animal surveillance systems may not be directly comparable due to differences in testing standards, specimen types, and reporting practices. The joint EFSA and ECDC report provides integrated human and animal resistance data for Europe, but researchers should understand the methodological differences between the human and animal components. When comparing resistance data across sectors, document the methodological characteristics of each data source and interpret differences in light of these factors. Direct comparisons without methodological adjustment can produce misleading conclusions about cross-sectoral resistance transmission. ## Practical Assessment Framework ### Database Suitability Checklist Evaluate each candidate database against the following criteria: - Geographic coverage matches the surveillance question - Target pathogens included in the database scope - Specimen types align with the monitoring objective - Susceptibility testing standards are documented and compatible - De-duplication rules are understood and appropriate - Data access policies permit the planned analysis - Data quality indicators are available and acceptable - Update frequency supports the intended monitoring timeline Each criterion should be assessed systematically and documented to support the selection decision. A database that fails one or more criteria may still be suitable if the limitations are understood and acknowledged in the analysis. ### Comparative Analysis Approach When comparing resistance data across databases, document the methodological differences that could affect results. Create a comparison table showing the de-duplication method, specimen types, testing standards, and patient populations for each database. Interpret resistance differences in light of these methodological factors instead of attributing all differences to true epidemiological variation. The comparison of GLASS and JANIS de-duplication methods provides a useful example of how methodological differences can be quantified and assessed. The study found that the number of patients per specimen type differed between methods, but resistance percentages for E. coli differed by less than three percent for any antibiotic tested, suggesting that resistance percentages may be relatively robust to de-duplication methodology. ## Building a Practical Database Selection Workflow ### Define the Minimum Viable Dataset Before Comparing Systems The most common selection error is starting with a database and then trying to fit the surveillance question to its available fields. Reverse this process by defining the minimum viable dataset first. Write down the specific data elements your analysis requires, including pathogen species, specimen source, patient age group, infection origin, antimicrobial agents tested, susceptibility testing method, and geographic identifiers. Then assess whether each candidate database contains all of these elements at the required level of granularity. A database that lacks even one essential field cannot support the intended analysis, regardless of its overall size or reputation. For example, a study on resistance trends in pediatric urinary tract infections requires age stratification and urine specimen data. A database that reports only aggregate resistance percentages without age breakdowns fails the minimum viable dataset test. Document the required fields in a checklist format and score each database against the checklist before considering secondary factors such as data accessibility or analytical tools. ### Map the Surveillance Question to Database Design Features Each surveillance database was designed with specific analytical purposes that shape its data structure. GLASS emphasizes national coordination and standardized case definitions, making it suitable for cross-country comparisons of priority pathogens. EARS-Net focuses on invasive isolates from blood and cerebrospinal fluid, which supports analysis of severe hospital-associated infections but excludes community-acquired infections from non-sterile sites. NCBI Pathogen Detection integrates genomic data with isolate metadata, enabling resistance gene detection and phylogenetic analysis that phenotypic databases cannot provide. Create a mapping table that links your surveillance question components to database design features. For each question component, identify which database features are essential, which are desirable, and which are irrelevant. This mapping exercise forces explicit consideration of how database design affects analytical possibilities. A researcher studying carbapenem resistance mechanisms in Enterobacterales requires genomic data and would find NCBI Pathogen Detection essential, while a public health authority tracking national resistance trends for policy reporting would prioritize GLASS data. ### Apply a Weighted Scoring System for Database Comparison A weighted scoring system provides a structured method for comparing databases when multiple factors influence the selection decision. Assign weights to each evaluation criterion based on its importance to your specific surveillance question. Common criteria include geographic coverage, pathogen scope, specimen type coverage, data granularity, update frequency, data access policies, quality assurance documentation, and analytical tool availability. Score each database on a defined scale for each criterion, then multiply the score by the criterion weight and sum the results. This approach prevents subjective impressions from dominating the selection process and creates a documented rationale that can be reviewed and challenged. The weighting should reflect the surveillance question, not general preferences. A researcher prioritizing real-time outbreak detection would weight update frequency heavily, while a researcher conducting multi-year trend analysis would weight temporal consistency and breakpoint documentation more heavily. ### Validate Database Selection With a Pilot Data Extraction Before committing to a full analysis, conduct a pilot data extraction from the selected database. Download a limited dataset covering your target pathogen, geographic region, and time period. Verify that the data structure matches your analytical requirements and that the fields contain the expected values. Check for unexpected coding schemes, missing metadata, or formatting issues that could complicate analysis. The pilot extraction also provides an opportunity to test the database query interface and assess practical usability. Document any difficulties encountered during the extraction process, including response times, export format limitations, or restrictions on the volume of data that can be downloaded at once. These practical considerations can affect the feasibility of the full analysis even when the database content is appropriate. ### Establish a Data Quality Verification Protocol After selecting a database, implement a verification protocol to confirm data quality before drawing conclusions. Start by checking internal consistency within the dataset. Verify that isolate counts match the reported denominators and that resistance percentages are calculated correctly from the underlying data. Cross-check a sample of records against original laboratory reports when possible to confirm accurate data entry. Compare your extracted data against published reports from the same database to identify discrepancies. If the database publishes annual summaries, verify that your extraction reproduces the published resistance percentages for the same time period and geographic scope. Discrepancies may indicate extraction errors, data version differences, or undocumented changes in database methodology that require investigation before proceeding with analysis. ### Document Methodological Assumptions and Limitations Create a written record of all methodological assumptions made during database selection and analysis. This documentation should include the surveillance question, minimum viable dataset requirements, database comparison scores, pilot extraction results, and data quality verification findings. Record the specific data version or extraction date, as surveillance databases update regularly and analyses must reference the exact data snapshot used. The documentation should also note any limitations identified during the selection process, including gaps in geographic coverage, missing specimen types, or methodological differences from other databases. These limitations should be acknowledged in any publications or reports that use the surveillance data. Transparent documentation of limitations supports appropriate interpretation of results and enables other researchers to understand the context of your findings. ### Establish a Review Schedule for Database Reassessment Surveillance databases evolve over time, with changes in participating laboratories, testing standards, data submission requirements, and analytical tools. A database that was appropriate for your surveillance question at the time of selection may become less suitable as these factors change. Establish a regular review schedule to reassess whether the selected database continues to meet your monitoring needs. The review should include checking for updates to the database methodology, changes in data access policies, and new data elements that may have been added. Also monitor for changes in your own surveillance question that may require different data sources. Document the review findings and update the selection rationale when significant changes occur. This ongoing reassessment ensures that your surveillance approach remains aligned with your monitoring objectives over time. ### Record System for Database Selection Decisions Maintain a structured record system that captures all decisions made during the database selection process. The record should include the date of each decision, the individuals involved, the criteria considered, and the rationale for the final selection. This record supports reproducibility and provides a reference point for future reassessment. The record system should also track data access agreements, data version information, and any correspondence with database coordinators. This information supports proper citation of data sources and facilitates resolution of any questions about data provenance that may arise during analysis or publication. A complete record system transforms the selection process from an informal decision into a documented, auditable procedure. ## Frequently Asked Questions ### How do de-duplication methods affect resistance percentages? De-duplication methods influence the number of isolates included in surveillance data but may have limited impact on resistance percentages. The comparison of GLASS and JANIS methods found that the number of patients per specimen type differed between methods, but resistance percentages for E. coli differed by less than three percent for any antibiotic tested. Researchers should understand the de-duplication rules of each database but may find that resistance percentages are relatively robust to these methodological differences. ### Can pharmacovigilance databases replace conventional surveillance? Pharmacovigilance databases cannot replace conventional surveillance systems. The comparison of VigiBase and EudraVigilance Veterinary data against GLASS and EFSA-ECDC surveillance found consistent statistically significant differences in the distribution of antibiotic resistance cases. Pharmacovigilance data may provide supplementary information about adverse outcomes but does not replicate conventional surveillance findings. ### What data does NCBI Pathogen Detection provide? NCBI Pathogen Detection integrates genomic and epidemiological data for foodborne pathogens and other clinically significant bacteria. The database links whole-genome sequencing data with isolate metadata, supporting phylogenetic analysis and resistance gene detection. This database serves different purposes than phenotypic surveillance systems and is best suited for outbreak investigation and resistance mechanism analysis. ### How should researchers handle missing data in surveillance databases? Researchers should document the extent and pattern of missing data in their analysis. Missing susceptibility results may reflect testing decisions, laboratory capacity limitations, or data submission issues. Sensitivity analyses can assess the impact of missing data on resistance estimates, and researchers should acknowledge these limitations when interpreting results. ### What quality controls support reliable surveillance data? External quality assessment participation, reference strain verification, and adherence to standardized testing methods support reliable surveillance data. The Clinical and Laboratory Standards Institute provides standards-based susceptibility testing methodologies, while the American Type Culture Collection provides authenticated reference strains for quality control. The American Society for Microbiology offers professional education resources for laboratory workforce development. ### How do resistance patterns inform treatment guidelines? Resistance surveillance data informs treatment recommendations by documenting current susceptibility patterns and trends. The ESCMID guidelines for multidrug-resistant Gram-negative bacilli and the Cochrane Review of cephalosporins for enteric fever both rely on evidence that includes resistance surveillance context. Treatment recommendations must account for regional resistance patterns, which change over time and require ongoing surveillance. ### What are the limitations of genomic resistance prediction? Genomic data can identify resistance genes, but the presence of a gene does not always predict phenotypic resistance. Gene expression, mutation effects, and regulatory mechanisms influence the relationship between genotype and phenotype. Researchers should verify genomic predictions against phenotypic susceptibility data when available and recognize that genomic databases may not represent the full spectrum of clinical isolates. ### How should laboratories select a surveillance database for participation? Laboratories should evaluate their capacity to meet the data submission requirements of each surveillance system, including susceptibility testing standards, data format specifications, and reporting timelines. Participation in surveillance systems requires ongoing commitment to quality assurance and data submission. Laboratories should also consider the value of surveillance data for their local infection control and stewardship programs. ## Related Bacteriology Guides - [Salmonellosis in Poultry: Serovar Surveillance, Antimicrobial Resistance, and Egg Safety](/knowledge/bacteria/avian-bacteria/salmonellosis-poultry-serovar-surveillance-antimicrobial-resistance-egg-safety) - [Antimicrobial Resistance in Livestock-Associated Staphylococcus aureus: Genomic Epidemiology and One Health Implications](/knowledge/bacteria/livestock-bacteria/antimicrobial-resistance-livestock-associated-staphylococcus-aureus-genomic-epidemiology-one-health) - [Biological Foundation Models for Antimicrobial Peptide Discovery in Veterinary Pathogens](/knowledge/bacteria/general/biological-foundation-models-for-antimicrobial-peptide-discovery-in-veterinary-pathogens) - [Salmonellosis in Poultry: Public Health and Flock Management](/knowledge/bacteria/avian-bacteria/salmonellosis-in-poultry) - [Aeromonas hydrophila in Aquaculture: Virulence Factors and Antimicrobial Resistance](/knowledge/bacteria/aquatic-bacteria/aeromonas-hydrophila-aquaculture-virulence-factors-antimicrobial-resistance) ## References and Further Reading - [Biosafety in Microbiological and Biomedical Laboratories](https://www.cdc.gov/labs/bmbl/index.html). CDC and NIH. - [Microbiology Resources](https://asm.org/). American Society for Microbiology. - [Antimicrobial Resistance](https://www.who.int/health-topics/antimicrobial-resistance). World Health Organization. - [Bacteriological Analytical Manual](https://www.fda.gov/food/laboratory-methods-food/bacteriological-analytical-manual-bam). U.S. Food and Drug Administration. - [Antimicrobial Susceptibility Testing](https://clsi.org/standards/products/microbiology). Clinical and Laboratory Standards Institute. - [Culture Collections](https://www.atcc.org/). American Type Culture Collection. - [NCBI Pathogen Detection](https://www.ncbi.nlm.nih.gov/pathogens). National Center for Biotechnology Information. - [European Society of Clinical Microbiology and Infectious Diseases (ESCMID) guidelines for the treatment of infections caused by multidrug-resistant Gram-negative bacilli (endorsed by European society of intensive care medicine).](https://pubmed.ncbi.nlm.nih.gov/34923128). Clinical microbiology and infection : the official publication of the European Society of Clinical Microbiology and Infectious Diseases, 2022. - [Treatment of enteric fever (typhoid and paratyphoid fever) with cephalosporins.](https://pubmed.ncbi.nlm.nih.gov/36420914). The Cochrane database of systematic reviews, 2022. - [A comparison of antibiotic resistance reports in pharmacovigilance databases and conventional surveillance across "One Health".](https://pubmed.ncbi.nlm.nih.gov/42221622). Frontiers in public health, 2026. - [Comparison of de-duplication methods used by WHO Global Antimicrobial Resistance Surveillance System (GLASS) and Japan Nosocomial Infections Surveillance (JANIS) in the surveillance of antimicrobial resistance.](https://pubmed.ncbi.nlm.nih.gov/32589639). PloS one, 2020. > This article is educational and does not replace validated laboratory procedures, clinical diagnosis, institutional biosafety review, or public-health guidance.