The Center for Disease Dynamics, Economics & Policy

ANTIMICROBIAL RESISTANCE

DATA 

• United States

U.S. resistance indicators come from The Surveillance Network (TSN), an extensive collection of microbiological susceptibility test results used for surveillance of antimicrobial resistance. TSN was established in 1994 and is among the largest databases of its kind. Researchers at the Centers for Disease Control (CDC), U.S. Food and Drug Administration (FDA), and Clinical and Laboratory Standards Institute (CLSI) have drawn on its resources for major scientific publications. Information is submitted on a daily basis from a selected network of some 300 clinical laboratories in nearly 200 zip codes. All laboratories adhere to CLSI standards, and consistency checks and duplicate screenings are performed regularly by the data provider. Participation is voluntary, but laboratories are required to submit all clinical isolates. Sites may vary from year to year; however, the percentage of sites that change is no more than 10% of the sample each year. Although certain states have no participating laboratories, TSN is considered representative at the national and census division levels. 

• Europe

European data come from the European Antimicrobial Resistance Surveillance Network (EARS-Net, formerly EARSS), a network of national surveillance systems providing reference data on antimicrobial resistance for public health purposes from clinical laboratories in more than 30 countries. The 900 participating public health laboratories serve some 1,400 hospitals in Europe and provide services to an estimated population of 100 million. The majority (68%) of laboratories apply CLSI guidelines, with the remainder using EUCAST or national standards. According to the European Center for Disease Control and Prevention (ECDC) which coordinates the network, harmonization efforts between the different national guideline committees have been satisfactory, and discrepancies between guidelines have become less significant (see Limitations section). Validated summary results can be accessed through an interactive database on the ECDC website, in addition to annual print reports. 

• Canada

National and regional data for Canada come from the CANWARD 2009 study. Surveillance is coordinated by the Canadian Antimicrobial Resistance Alliance (CARA), a research group dedicated to the study of medical microbiology and infectious diseases issues, with special interest in antimicrobial usage and infections caused by antimicrobial-resistant pathogens. More than 15 hospitals across Canada submit clinical isolates from emergency room, inpatient, outpatient, and ICU settings. Susceptibility testing is done according to CLSI guidelines at a single site. The results are made available as a web-based interactive portal on the CARA website, along with valuable educational materials.

METHODOLOGY

Unless otherwise specified, resistance rates in the atlas are calculated as the ratio of strictly resistant isolates (R) to the total number of collected isolates in a given region (sample size N). To derive rates for the United States, individual isolate data from TSN are aggregated at the national, census division, and state levels, depending on the atlas component in question. The data are stratified by patient location and outpatient and inpatient results (inpatient includes ICU and nursing home). Resistance rates are reported by drug class. To ensure compatibility across different data sources, individual drugs were used to proxy for resistance to an entire class. Click to view details for GRAM-NEGATIVE or GRAM-POSITIVE species.

LIMITATIONS

• Sample size and representativeness

Non-US data reported in visualizations is collected at the national level under the responsibility of the respective participating laboratories and research organizations. Data from the TSN database is considered representative when aggregated at the census division level as certain states have few or no participating laboratories. Please note that sample size and coverage may vary widely between countries and regions in the visualizations. The number of tested isolates is is included as a variable in all bug-drug animations and can be seen by hovering over the respective region or selecting the SAMPLE SIZE variable from the dropdown menu. Data points where countries reported fewer than 20 isolates have been excluded from these visualizations to avoid outlier values. 

• Global standartization of resistance criteria

Another limitation of our global resistance comparisons may be the different interpretation of resistance criteria applied by clinical laboratories in various parts of the world. Whereas CLSI breakpoints are used to report susceptibility from North America and more than half of the European laboratories in the sample, some countries use their own national criteria or EUCAST breakpoints that may differ from those established by CLSI. 

To justify the pooling and comparison of data across Europe, EARS-Net performs regular external quality assessments of the overall comparability of results between laboratories and countries. The results published in the 2009 annual report show that, in most instances, there is sufficient accuracy to provide good estimates of overall resistance prevalence and trends. Overall concordance between reported results is high (>90%), except in cases of borderline susceptibility – when various guidelines reveal discrepancies in routine testing (e.g., piperacillin-tazobactam susceptibility of ESBL producers) or when a breakpoint has been recently changed (e.g., penicillin susceptibility of Streptococcus pneumoniae). 

• Other limitations of susceptibility data

First, surveillance studies provide information concerning only the site of isolate collection and its susceptibility. Because there are no clinical data to confirm infection, results should not be interpreted as prevalence of actual drug-resistant infections. Second, the TSN database is designed to be representative of U.S. hospitals, but it may not capture the entire distribution of resistance rates in U.S. hospitals. Nevertheless, U.S. results are not likely to be biased by localized epidemics because of the large sample size of the TSN database. Finally, test selection procedures may influence resuts (ascertainment bias). As resistance to an antibiotic becomes established, isolates may be tested for susceptibility to the antibiotic more frequently, either because of official changes in protocol or because treatment failures become more common.