Antimicrobial resistance (AMR)

How was AMR research conducted?

We estimated deaths and disability-adjusted life years associated with and attributable to bacterial AMR for 23 pathogens and 88 pathogen-drug combinations in 204 countries and territories in 2019.

Our process consists of five broad components: 

• Estimate the number of deaths where infection played a role.
• Determine the proportion of deaths attributable to a given infectious syndrome.
• Determine the proportion of deaths attributable to a given pathogen.
• Calculate the percentage of a given pathogen resistant to an antibiotic.
• Estimate the excess death risk or longer infection duration associated with resistance.

Using these components, we estimated disease burden based on two counterfactuals:

• Deaths attributable to AMR (based on an alternative scenario in which all drug-resistant infections were replaced by drug-susceptible infections).
• Deaths associated with AMR (based on an alternative scenario in which all drug-resistant infections were replaced by no infection).

What data sources were used to produce AMR findings?

A variety of data were gathered to inform these estimates, including multiple cause of death data, hospital discharges, minimally invasive tissue sampling, systematic literature reviews, and microbiology lab results from hospitals and national and multinational surveillance systems.

AMR data sources and estimates are available for download on the Global Health Data Exchange (GHDx) – a data catalog created and supported by IHME.

Download AMR estimates

What do these terms mean?

• Deaths attributable to AMR: Deaths attributable to AMR refers to deaths that were directly caused by drug-resistant infections as a result of an ineffective treatment. Under an alternative scenario where all drug-resistant infections were instead drug-susceptible, these deaths would not have occurred.
• Deaths associated with AMR: Deaths associated with AMR refers to deaths that occurred from a drug-resistant infection, but for which AMR may or may not have been the cause. Under an alternative scenario where all drug-resistant infections were replaced by no infection, these deaths would not have occurred.
• Age-standardization: Our AMR analysis uses age-standardized aggregates, which is a technique for comparing populations with different age structures to account for over- or under-representation of age groups in different countries.
• Bug-drug (pathogen-drug) combination: A pathogen-drug combination refers to a patient’s infection with a specific bacterium and that person’s subsequent treatment with an antimicrobial drug, in this case, an antibiotic. We modeled global resistance patterns by using laboratory results documenting resistance in these “bug-drug” combinations around the world.
• “Drug-sensitive infections” refer to those effectively treated with available antibiotics. For these infections, no AMR is observed.
• “Drug-resistant infections” describes those lacking efficacious treatments, i.e., when a pathogen has developed resistance to available antimicrobial therapies.

What are the advantages and disadvantages of the modeling used in this analysis?

The use of modeling in this analysis brings several advantages. By leveraging a wide range of data inputs and expert local collaborators, our approach can produce an estimate for every location for several bug-drug combinations using a consistent methodology. By consequence, regional estimates are directly comparable, and we can determine locations with the greatest burden that require immediate attention. We also provide two plausible counterfactuals representing an upper and lower bound.

The major disadvantage is that these estimates are modeled estimates and not population surveillance data (though such data is included as input data). This, of course, is due to the lack of high-quality population surveillance data for some locations. Without the use of modeling, there would be no estimates for the data-scarce locations. In our analysis, each estimate is coupled with 95% uncertainty intervals (UIs) to measure the degree of uncertainty and the margin of error.

Which pathogens were studied?

We prioritized estimating burden for bug-drug combinations of international public health concern and priority for the WHO member states, as well as the Bill & Melinda Gates Foundation.

Core pathogen-drug combinations

• Escherichia coli – third-generation cephalosporins, fluoroquinolones  
• Klebsiella pneumoniae – third-generation cephalosporins, carbapenems  
• Staphylococcus aureus – methicillin  
• Streptococcus pneumoniae – penicillin  
• Salmonella Typhi & Paratyphi A – multidrug resistance, fluoroquinolones  
• Invasive non-typhoidal Salmonella – fluoroquinolones  
• Shigella species – fluoroquinolones  
• Neisseria gonorrhoeae – third-generation cephalosporins  
• Mycobacterium tuberculosis – isoniazid mono-resistance, rifampicin mono-resistance

Supplementary pathogen-drug combinations

• Acinetobacter baumannii – aminoglycosides, anti-pseudomonal penicillin/beta-lactamase inhibitors, beta-lactam/beta-lactamase inhibitors, carbapenems, third-generation cephalosporins, fourth-generation cephalosporins, fluoroquinolones 
• Citrobacter species – aminoglycosides, anti-pseudomonal penicillin/beta-lactamase inhibitors, carbapenems, third-generation cephalosporins, fourth-generation cephalosporins, fluoroquinolones 
• Enterobacter species – aminoglycosides, anti-pseudomonal penicillin/beta-lactamase inhibitors, carbapenems, fourth-generation cephalosporins, fluoroquinolones, trimethoprim-sulfamethoxazole 
• Enterococcus faecalis – fluoroquinolones, vancomycin 
• Enterococcus faecium – fluoroquinolones, vancomycin 
• Enterococcus species – fluoroquinolones, vancomycin 
• Escherichia coli – aminoglycosides, aminopenicillin, beta-lactam/beta-lactamase inhibitors, carbapenems, trimethoprim-sulfamethoxazole 
• Group A Streptococcus – macrolide 
• Group B Streptococcus – fluoroquinolones, macrolide, penicillin 
• Haemophilus influenzae – aminopenicillin, third-generation cephalosporins 
• Klebsiella pneumoniae – aminoglycosides, beta-lactam/beta-lactamase inhibitors, fluoroquinolones, trimethoprim-sulfamethoxazole 
• Morganella species – third-generation cephalosporins, fourth-generation cephalosporins, fluoroquinolones 
• Neisseria gonorrhoeae – fluoroquinolones  
• Proteus species – aminoglycosides, aminopenicillins, third-generation cephalosporins, fluoroquinolones, trimethoprim-sulfamethoxazole 
• Pseudomonas aeruginosa – aminoglycosides, anti-pseudomonal penicillin/beta-lactamase inhibitors, carbapenems, third-generation cephalosporins, fourth-generation cephalosporins, fluoroquinolones 
• Serratia species – aminoglycosides, anti-pseudomonal penicillin/beta-lactamase inhibitors, carbapenems, third-generation cephalosporins, fourth-generation cephalosporins, fluoroquinolones  
• Staphylococcus aureus – fluoroquinolones, macrolide, trimethoprim-sulfamethoxazole, vancomycin 
• Streptococcus pneumoniae – beta-lactam/beta-lactamase inhibitors, carbapenems, third-generation cephalosporins, fluoroquinolones, macrolide, trimethoprim-sulfamethoxazole 

Who works on AMR research?

IHME conducts research on AMR as part of a global initiative called the Global Research on Antimicrobial Resistance (GRAM) Project. The GRAM Project brings together researchers from around the world to develop the evidence base for understanding one of our most pressing global health challenges, the threat of AMR.

We are grateful for the support of the Department of Health and Social Care (DHSC) using UK aid funding, the Bill and Melinda Gates Foundation and Wellcome Trust, which enables us to continue this work. The views expressed in our publications are those of the authors and not necessarily those of the funders.

To learn more about the project scope and deliverables, visit the GRAM Project website.