Tiago Cravo Oliveira
Michael Padget
Stemming the Superbug Tide
Just A Few Dollars More
OECD Health Policy Studies
Chapter 3. Trends in antimicrobial resistance in OECD countries
Abstract
Antimicrobial resistance (AMR) proportions have evolved very differently across countries and antibiotic-bacterium combinations in the last decade. This chapter looks at the challenges involved in defining and measuring resistance internationally, along with the methodology used to estimate historic and future resistance proportions. The chapter presents predicted resistance proportions for 52 countries for 2015, along with the rate of change since 2005 (averaged across eight priority antibiotic-bacterium pairs). The chapter then provides data on the projected resistance proportions across the same countries up to 2030. Factors contributing to the wide variability in the predicted resistance proportions between antibiotic-bacterium pairs within and across different countries are explored. The chapter concludes by looking at the problem of increasing resistance to second and third-line antimicrobials and highlights steps needed to achieve better empirical research and more targeted policy actions.
The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities. The use of such data by the OECD is without prejudice to the status of the Golan Heights, East Jerusalem and Israeli settlements in the West Bank under the terms of international law.
Note by Turkey:
The information in this document with reference to “Cyprus” relates to the southern part of the Island. There is no single authority representing both Turkish and Greek Cypriot people on the Island. Turkey recognises the Turkish Republic of Northern Cyprus (TRNC). Until a lasting and equitable solution is found within the context of the United Nations, Turkey shall preserve its position concerning the “Cyprus issue”.
Note by all the European Union Member States of the OECD and the European Union:
The Republic of Cyprus is recognised by all members of the United Nations with the exception of Turkey. The information in this document relates to the area under the effective control of the Government of the Republic of Cyprus.
Key findings
Antimicrobial resistance (AMR) in high-priority bacteria has increased in the last decade in a majority of countries. It is estimated that average resistance proportions for eight antibiotic-bacterium combinations, across OECD countries, grew from 14% in 2005 to 17% in 2015. Resistance could grow further to 18% by 2030 if current trends continue into the future. These figures mask significant variation across countries and regions, as well as antibiotic-bacterium pairs.
In OECD countries, in 2015, the highest average resistance proportions predicted (around 35% in Turkey, Korea and Greece) were seven times higher than the lowest proportions (around 5% in Iceland, the Netherlands and Norway). Outside OECD countries, in India, the People’s Republic of China and the Russian Federation, for example, average rates were estimated to be in excess of 42%. In certain countries and with certain antibiotic-bacterium combinations, four out of every five infections were due to resistant bacteria. The estimated average resistance proportions in G20) countries and OECD Key Partners were 30% and 41% respectively.
The majority of countries in the OECD, European Union (EU) and G20 are estimated to have experienced both reductions and increases in resistance proportions for different antibiotic-bacterium pairs between 2005 and 2015, suggesting marked differences in the evolution of resistance across countries, bacteria and antibiotic agents.
While growth in drug resistance proportions could slow down in the coming decades, there are reasons for concern. The recent progress in reducing methicillin-resistant Staphylococcus aureus (MRSA) could be halted, or even reversed in some countries, and growth of resistance in difficult to treat enterococcal infections is projected to accelerate. Consumption of second-line antibiotic therapies will grow if current trends persist, potentially further promoting resistance and limiting options for treatment. With populations growing and ageing, the number at risk of infection will also increase.
Differences in resistance proportions across countries and antibiotic-bacterium combinations are likely associated with differences in antimicrobial consumption, infection prevention and control, as well as use of health care services, but not only. Differences in measurement methods, as well as in probable correlates and drivers (including animal health, sanitation, migration, urbanisation, etc.) are likely relevant, but a lack of internationally comparable and reliable data on these factors hinders empirical research and robust policy-making. Surveillance and monitoring of all facets of AMR are needed.
3.1. Measuring and understanding international data on AMR is challenging
AMR is an umbrella term that includes many different types of drug resistance. While reports typically refer to broader classes of microorganisms and antimicrobial agents (e.g. third-generation cephalosporin-resistant Enterobacteriaceae), resistance is, in practice, defined and measured at the level of a specific microorganism (e.g. Klebsiella pneumoniae) and a specific antimicrobial drug (e.g. ceftriaxone, which is a third-generation cephalosporin). Ensuring data on AMR are internationally comparable is difficult for a number of reasons including the existence – and varying use – of multiple drugs, standards, guidelines, methods and equipment in different parts of the world (see Box 3.1). Cross-country differences in how individual-level results are aggregated to produce population-level measures (e.g. proportions, incidence) further limit comparability.
The development and more widespread use of guidelines have facilitated the collection and aggregation of international data on AMR. This collation and reporting of cross-country estimates can promote further discussion around how to tackle the significant difficulties that remain. The Center for Disease Dynamics, Economics & Policy (CDDEP), for example, started collating data on resistance proportions in 2010 and publishing them in a web-based set of visualisations called ResistanceMap1. The current version provides data on resistance proportions for up to 49 countries for different combinations of up to 12 organisms and 17 antibiotic groups as well as antibiotic consumption for different groups of agents for up to 75 countries, from 2000 to 2015. ResistanceMap1 provides country-level yearly resistance data (number of isolates tested and proportion found to be resistant) primarily collected from international surveillance networks, like EARS-Net and CEASAR.2
The antibiotic-bacterium combinations included in this analysis were selected based on: a) significance of the burden of disease in OECD countries and the EU, both in the health care sector and the community; b) policy priority for OECD, G20 and EU countries (Cecchini, Langer and Slawomirski, 2015[1]); c) data availability (namely availability from CDDEP and networks such as EARS-Net); and d) inclusion in the World Health Organization (WHO) global priority list of antibiotic-resistant bacteria to guide research and development of new antibiotics (WHO, 2017[2]).
According to the latest figures from the surveillance networks included in ResistanceMap1 and OECD analyses (see Box 3.2) the unweighted average of estimated resistance proportions, across eight specific antibiotic-bacterium combinations, was 17% in OECD countries in 2015 (see Table 3.1). Iceland, Netherlands and Norway had the lowest predicted average resistance proportions, around 5%, while in Turkey, Korea and Greece more than 35% of infections were estimated to be due to resistant bacteria, on average, across all eight antibiotic-bacterium combinations. India, China and the Russian Federation all had average resistance proportions in excess of 42% and, for some antibiotic-bacterium pairs, over 80% of infections were from resistant bacteria.
Box 3.1. The challenge of defining and measuring drug resistance
International comparisons of AMR typically rely on data from national and international surveillance networks which aggregate antimicrobial susceptibility test results from multiple laboratories. Results are often presented in annual surveillance reports such as those published by the European Centre for Disease Prevention and Control (ECDC) (2017[3]), WHO (2014[4]) and the Australian Commission on Safety and Quality in Health Care (2017[5]). Multiple factors can affect the accuracy or comparability of the statistics reported in these publications, including:
In the community: Antimicrobial susceptibility test results may never reach a surveillance network if patients: 1) do not seek care, 2) self-medicate, for example with leftover antibiotics (OECD, 2017[6]), or 3) seek treatment from informal care providers. Given that severe infections (such as those caused by resistant bacteria) should eventually lead to formal health care utilisation and testing, it is likely that the number of susceptible infections in networks is conservative and resistance proportions are overestimated.
In the clinic/hospital: Even when patients seek care from a formal health care provider, test results may never make it to a surveillance network. A patient with an infection caused by a bacterium susceptible to antimicrobial treatment may not be tested if the physician prescribes treatment and the symptoms subside. Again, this should lead to an underestimation of infections from bacteria that are susceptible to treatment and an overestimation of resistance proportions. Even if a test is ordered, the clinic/hospital may work with a laboratory that is not part of a surveillance network. The sample may also be contaminated leading to inaccurate test results.
In the laboratory: Different laboratories may use different testing standards, guidelines and equipment, all of which could potentially affect the final outcome (i.e. whether a bacterium is considered resistant). Technical decisions that could affect the outcome include the choice of growth media, measurement method (e.g. phenotypic or genotypic), incubation conditions, and precision of different concentrations of antimicrobial agents (Turnidge and Paterson, 2007[7]; Hindler and Stelling, 2007[8]). Even phenotypic test results from the same laboratory can be categorised differently depending on the choice of breakpoints used (Kassim et al., 2016[9]; van der Bij et al., 2012[10]; Wolfensberger et al., 2013[11]).
In the surveillance network: Some surveillance networks only report results for certain types of samples (ECDC, 2017[3]) while others report on all samples (WHO, 2017[12]). Furthermore, some networks discard repeated measurements (i.e. tests that are conducted more than once a year when patients have recurring infections). Studies that include all tests generally find higher resistance proportions than those that do not (Hindler and Stelling, 2007[8]).
There is no gold standard in the definition and measurement of AMR (unlike, for example, with blood pressure) making many of the decisions described above equally valid. Little is known about the direction and magnitude of some of these biases (Turnidge and Paterson, 2007[7]) making it very difficult to adjust published estimates. Surveillance networks help achieve comparability in that they set common standards and guidelines, but it is important to understand how different networks pool their data.
Table 3.1. Resistance proportions for eight priority antibiotic-bacterium pairs in 2015
Note: * indicates country is missing more than 50% of observations, across all eight antibiotic-bacterium pairs for 2015. The colour scheme is based on a two-point scale (minimum in light grey, maximum in blue, points in between coloured proportionally). Countries (and country groupings) are sorted top to bottom from highest to lowest average resistance proportions (across antibiotic-bacterium combinations). Antibiotic-bacterium combinations are sorted left to right from highest to lowest average resistance proportions (across countries). FREC: fluoroquinolone-resistant E. coli, VRE: vancomycin-resistant E. faecalis and E. faecium, 3GCREC: third-generation cephalosporin-resistant E. coli, CRKP: carbapenem-resistant K. pneumoniae, 3GCRKP: third-generation cephalosporin-resistant K. pneumoniae, CRPA: carbapenem-resistant P. aeruginosa, MRSA: methicillin-resistant S. aureus, PRSP: penicillin-resistant S. pneumoniae.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1.
Resistance proportions in countries outside the OECD (estimated at an average 29%, across all eight antibiotic-bacterium combinations) were significantly higher than those in member states in 2015. Members of the EU had average rates (18%) comparable to OECD countries and member states of the G20 had predicted resistance proportions of 30%. The highest average resistance proportions, an estimated 41%, were in key partners of the OECD (i.e. countries that contribute to the OECD's work in a sustained and comprehensive manner but are not formal members).
Box 3.2. Methodology used to estimate historic and future resistance proportions
Historical and future resistance proportions were estimated using a combination of statistical techniques and models seeking to make use of as much publicly available, internationally comparable, data as possible, while explicitly accounting for uncertainty in the underlying data, models and assumptions. The approach sought to fill data gaps using best guesses from theoretically-hypothesised and empirically-tested relationships with correlates.
Resistance and consumption data were collected from ResistanceMap1. The primary sources of resistance data in ResistanceMap1 are public and private laboratory networks – such as EARS-Net – that routinely collect antimicrobial susceptibility test results. Data on antibiotic consumption were sourced from IMS Health’s MIDAS and XPonent databases. Data on a wide range of topics (from health and sanitation to agricultural and livestock production) were collected from the World Bank, WHO, the Food and Agriculture Organization (FAO), the United Nations World Population Prospects (UN WPP), the United States Department of Agriculture, and the OECD’s own databases.
Producing complete estimates of resistance proportions for eight antibiotic-bacterium combinations (third-generation cephalosporin-resistant E. coli, fluoroquinolones-resistant E. coli, penicillin-resistant S. pneumoniae, methicillin-resistant S. aureus, carbapenem-resistant K. pneumoniae, third-generation cephalosporin-resistant K. pneumoniae, carbapenem-resistant P. aeruginosa, and vancomycin-resistant E. faecalis and E. faecium) in 52 countries from 2000 to 2030, including uncertainty, involved the following procedures:
1. multiple imputation of missing historical values using all relevant information from observed relationships with correlates
2. forecasting of potential correlates of resistance proportions using external forecasts (e.g. UN WPP) and internal forecasts of antibiotic consumption using exponential smoothing with an additive damped trend
3. forecasts of resistance proportions using an ensemble of three models, equally weighted, to capture different aspects of the underlying phenomena (a mixed-effects linear regression, exponential smoothing with an additive damped trend, and a random forest)
4. incorporation of uncertainty from various sources, namely imputations of missing values, model selection and specification, and some model parameters. The final estimates are described using means and 95% uncertainty intervals (e.g. see Figure 3.8).
Forecasts do not incorporate future policy actions or interventions.
3.2. Predicted resistance proportions increased between 2005 and 2015 in a majority of countries and for most antibiotic-bacterium combinations
Between 2005 and 2015, predicted resistance proportions for eight antibiotic-bacterium combinations in OECD countries, increased, on average, by 3 percentage points from 14% in 2005 to 17% in 2015 (see Table 3.3). In seven countries (Switzerland, United Kingdom, Japan, Belgium, Germany, Iceland and Canada), resistance proportions went down, on average across all antibiotic-bacterium-country combinations, by 2.5 percentage points. In the majority of countries, however, average resistance proportions across all eight antibiotic-bacterium pairs increased, by as much as 13 percentage points in OECD countries (e.g. Slovak Republic and Italy), but potentially even more in Brazil, the Russian Federation and China (around 17 percentage points). However, these averages mask significant variation within countries across antibiotic-bacterium combinations.
Despite average reductions in a few countries, it is estimated that in no country have resistance proportions for all eight antibiotic-bacterium combinations gone down between 2005 and 2015. In contrast, in eight countries (Brazil, China, Peru, Argentina, Colombia, Saudi Arabia, Israel and the Russian Federation) resistance proportions increased for all eight antibiotic-bacterium pairs. However, for the majority of countries, both increases and reductions were predicted, in some cases quite extreme.
In France, for example, the proportion of Klebsiella pneumoniae (K. pneumoniae) resistant to third-generation cephalosporins increased by an estimated 27 percentage points (from 5% to 32%, a growth rate of 540%) while the proportion of Streptococcus pneumoniae (S. pneumoniae) resistant to penicillin went down by a predicted 13 percentage points (from 36% to 23%, a growth rate of -36%) over the same ten years. In Mexico while the proportion of Staphylococcus Aureus (S. aureus) resistant to oxacillin (i.e. MRSA) went down by a predicted 21 percentage points (from 52% to 31%, a growth rate of -40%), the proportion of Escherichia Coli (E. coli) resistant to third-generation cephalosporins increased by an estimated 18 percentage points (from 40% to 58%, a growth rate of 45%).
Using resistance proportions in 2005 as the base value, growth rates (change in resistance proportions as a percentage of the 2005 base proportions) across countries and antibiotic-bacterium combinations are very heterogeneous. Across OECD countries, compared to 2005 values and using averages of country growth rates, resistance proportions for MRSA went down the most, by an estimated 17%, while the proportion of E. coli resistant to third-generation cephalosporins increased by a predicted 222% between 2005 and 2015. Among OECD countries, Italy and the United States had some of the biggest predicted growth rates in resistance proportions, more than 140% higher in 2015 compared to 2005, on average across all eight antibiotic-bacterium combinations. Resistance proportions in non-OECD members Croatia and Romania were more than 200% higher than estimates for ten years before. Belgium and Portugal had the biggest reductions in estimated resistance proportions compared to 2005, 10% and 2% respectively. Across all countries and antibiotic-bacterium combinations, predicted resistance proportions in 2015 were between 100% lower and 1 200% higher than in 2005, a very significant range of variation.
It is likely that differences in baseline resistance proportions and rates of change across countries and antibiotic-bacterium combinations are associated with differences in antimicrobial use, infection prevention and control, as well as the use of health care services (ECDC, 2017[3]), not to mention differences in measurement. However, the problem goes beyond the human health sector with links to the animal sector and the environment. The list of potential correlates and drivers of resistance is long, ranging from human and animal health and sanitation, agricultural and livestock production, urbanisation and population density, migration and trade, economic growth and governance, and population structure. Studies seeking to empirically test the direction and magnitude of these relationships have been hampered by two broad yet intertwined challenges: a lack of data and a limited understanding of how all these factors are related, to each other and to drug resistance.
Table 3.2. Percentage point changes in resistance proportions for eight priority antibiotic-bacterium combinations between 2005 and 2015 in select countries
Note: * indicates country is missing more than 50% of observations, across all eight antibiotic-bacterium pairs, for both 2005 and 2015. The colour scheme is based on a two-point scale (minimum in light grey, maximum in blue, points in between coloured proportionally). Countries (and country groupings) are sorted top to bottom from highest to lowest average resistance proportions (across antibiotic-bacterium combinations). Antibiotic-bacterium combinations are sorted left to right from highest to lowest average resistance proportions (across countries). FREC: fluoroquinolone-resistant E. coli, VRE: vancomycin-resistant E. faecalis and E. faecium, 3GCREC: third-generation cephalosporin-resistant E. coli, CRKP: carbapenem-resistant K. pneumoniae, 3GCRKP: third-generation cephalosporin-resistant K. pneumoniae, CRPA: carbapenem-resistant P. aeruginosa, MRSA: methicillin-resistant S. aureus, PRSP: penicillin-resistant S. pneumoniae.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1.
Comprehensive, comparable and reliable data on many of these factors are lacking. Even ignoring the difficulties in ensuring surveillance data are comparable across countries or regions (see Box 3.1), many low and middle-income countries either do not have national surveillance systems in place or surveillance is conducted in private hospitals and the resulting data not shared with international organisations. In high-income countries, where surveillance data from the human health sector are more widely available, there are still regional gaps and statistics in the animal sector are limited. While there has been progress, for example in determining what should be measured and monitored, important data gaps remain, making it difficult to set priorities and identify policy options (Wernli et al., 2017[13]).
A second difficulty is that drivers of resistance can interact, forming a complex web of relationships in which the same variable can have both direct and indirect, as well as non-linear, effects on the emergence and spread of resistance. The empirical literature is full of examples of these intricate associations. For example, it is widely acknowledged that both the underuse and overuse of antibiotics can lead to drug resistance (Mendelson et al., 2016[14]), which would suggest a non-linear relationship. Out-of-pocket spending in low and middle-income countries is positively correlated with AMR, a relationship that Alsan and colleagues (2015[15]) posit is due to high co-payments inducing patients to seek treatment from less well-regulated private providers with financial incentives to prescribe inappropriately. Collignon and co-authors (2015[16]) find that governance and corruption at the national-level are better predictors of antibiotic resistance than economic output or even antibiotic consumption, while Rönnerstrand and Lapuente (2017[17]) show that corruption is, in fact, a good predictor of antibiotic consumption.
3.3. The emergence and spread of AMR is a complex phenomenon with multiple interrelated causes and drivers
There is wide variation in the predicted resistance proportions for 2015 (Table 3.1) and rates of change between 2005 and 2015 (Table 3.2) across countries and antibiotic-bacterium combinations. For the reasons previously mentioned, it is difficult to say what factors are behind these differences but it is possible to identify broad patterns of drug resistance, not only by sex and age, but also by level of antibiotic consumption and economic development.
Across six antibiotic-bacterium combinations for which surveillance data are available from EARS-Net, the estimated incidence rates of all infections (i.e. the sum of infections caused by resistant and susceptible bacteria) exhibit a similar age pattern: children less than one year old and adults over 50 years old are more likely to be infected (see Figure 3.1). The population over 80 years old exhibit the highest rates of infection. This pattern is likely associated with the development of immune responses with age, starting with an immature immune system at birth, which develops with time and then declines in old age (Simon, Hollander and McMichael, 2015[18]). Furthermore, infants and the elderly are at a higher risk of being admitted to hospital, and thus at a higher risk of acquiring a hospital infection.
Figure 3.1. Incidence rates for six bacteria and proportion that are resistant to specific antibiotics, by age and sex, in all EARS-Net countries combined in 2015
Note: Incidence rates estimated using total number of infections in the EU/EEA (from EARS-Net) and age-specific population from the United States Census Bureau.
Source: OECD analyses of surveillance data from EARS-Net.
Men are significantly more likely than women to become infected. This is in line with other studies that have found men are more susceptible than women to many infections, a difference that has been tentatively associated with sex steroid hormones. These hormones are believed to both lower immunocompetence in men and affect genes and behaviours that influence their susceptibility and resistance to infection (Klein, 2000[19]).
While there are clear differences in incidence of infections across age and gender, the proportion of infections that are caused by bacteria resistant to antibiotics exhibit no discernible age-sex-pattern across different antibiotic-bacterium combinations (see Figure 3.1). With the exception of third-generation cephalosporin-resistant E. coli, resistance proportions in men and women are similar. Older people seem to be more likely to become infected with S. aureus resistant to antibiotic treatment (methicillin in this case) than younger people, while the reverse is true for those infected with P. aeruginosa (for which the antibiotic is carbapenems). Differences in resistance proportions by age may be due to both measurement and underlying factors. For example, resistance proportions may be higher for older populations due to increased exposure to antimicrobials and health care, both of which are risk factors for drug resistance. Similarly, younger people might put off care for less severe infections so that once an antimicrobial susceptibility test is conducted there is a higher chance that resistant bacteria will be found. The interplay between these and other factors might explain the age patterns observed for S. aureus and P. aeruginosa.
For six bacteria, both the incidence of infections and the proportion resistant to antimicrobials, show notably different patterns across body sites (see Figure 3.2). Differences in the number of certain types of infections stem from the characteristics of bacteria (e.g. whether they are predominately found in health care or community settings, their transmission mechanisms, what parts of the body they naturally colonise, etc.). For example, E. coli and K. pneumoniae are found in the human gut and are a common cause of urinary tract infections, while S. pneumoniae is often carried in the human respiratory tract or sinuses and is a leading cause of pneumonia (see Box 2.3 in Chapter 2)
There are also differences in the proportions resistant to antibiotic treatment (see Figure 3.2). As with age and sex, these differences could be due to both measurement biases and underlying resistance mechanisms. Drug concentrations can differ between different body sites yet clinical breakpoints used in many antimicrobial susceptibility tests are based on bloodstream concentrations (Turnidge and Paterson, 2007[7]), potentially adjusted in some cases where concentrations are known to be different (e.g. in the urinary tract). Differences in concentration could affect both measurement and underlying resistance.
As previously reported by Albrich et al., (2004[20]), there is a positive correlation between antibiotic consumption and the proportion of infections that are due to bacteria resistant to antibiotic treatment for certain antibiotic-bacterium combinations in specific years (see the panels for S. pneumoniae and S. aureus in Figure 3.3). However, the existence of outliers suggests antibiotic consumption is not the only factor associated with resistance. In Mexico, MRSA is higher than expected given how low the consumption of broad-spectrum penicillins is. In Belgium, the proportion of S. pneumoniae resistant to penicillins is much lower than expected given the relatively high level of consumption, yet this is likely associated with the use of different breakpoints, as explained by Goossens and colleagues (2013[21]), a reminder of how much measurement issues matter. Even when total antibiotic consumption is used in place of specific antibiotic groups, these outliers persist (data not shown). For carbapenem-resistant P. aeruginosa, the relationship between total antibiotic consumption and resistance appears to be non-existent (see Figure 3.3). Non-linear relationships between consumption and resistance proportions were explored but not found (data not shown). Other studies have also found a lack of a relationship between consumption and resistance for certain antibiotic-bacterium combinations (ECDC/EFSA/EMA, 2017[22]).
Figure 3.2. Incidence rates for six bacteria and proportion resistant to specific antibiotics, by type of infection, in all EARS-Net countries in 2015
Note: FREC: fluoroquinolone-resistant E. coli, VRE: vancomycin-resistant E. faecalis and E. faecium, 3GCREC: third-generation cephalosporin-resistant E. coli, CRKP: carbapenem-resistant K. pneumoniae, 3GCRKP: third-generation cephalosporin-resistant K. pneumoniae, CRPA: carbapenem-resistant P. aeruginosa, MRSA: methicillin-resistant S. aureus, PRSP: penicillin-resistant S. pneumoniae.
Source: OECD analyses of data from (Cassini et al., 2018[23]).
Figure 3.3. Associations between antibiotic consumption and resistance proportions for four antibiotic-bacterium combinations in 2015
Source: OECD analyses of data from surveillance networks included in ResistanceMap1.
While the patterns shown in Figure 3.3 could be partly due to measurement issues, there are at least two other non-mutually-exclusive explanations. One is that it is not so much the quantity of antibiotics prescribed but the appropriateness of consumption that is associated with drug resistance (Zilberberg et al., 2017[24]). This would be in line with observations that both low and high consumption of antibiotics can be associated with higher resistance, and might potentially explain why countries with low consumption of antibiotics exhibit such a high proportion of infections from resistant bacteria. The second potential explanation is that antibiotic consumption is merely one of a number of determinants and correlates of drug resistance.
There are, as previously mentioned, many factors thought to be associated with drug resistance, from infection prevention and control, to antibiotic consumption in the animal sector (see Box 3.3), access to health care, trade in goods and services, and migration, to mention just a few. Some of these factors could be associated with drug resistance in some, but possibly not all, countries and for certain antibiotic-bacterium combinations only. Some of these relationships are illustrated in Figure 3.5.
Box 3.3. Antimicrobial consumption in the animal sector
Globally, the bulk of antimicrobials is, in fact, not consumed by humans, but rather given to animals, mostly food-producing animals such as poultry and cattle. In the United States, it was estimated that around 80% of all antibiotic consumption is in the animal sector (Van Boeckel et al., 2015[25]). In the 28 EU/EEA member states that collect both animal and human consumption data, OECD analyses indicate 70% of the active substance of antimicrobials was sold for use in food-producing animals (ECDC/EFSA/EMA, 2017[22]). Antimicrobial agents are used in food-producing animals for a number of reasons including treating sick animals and preventing the spread of infectious diseases, but also, in some countries, to increase growth rates and feed efficiency (Cecchini, Langer and Slawomirski, 2015[1]). Demand for animal protein is rising worldwide driving up the consumption of antimicrobials in the livestock sector. If current trends continue and no effective policy action is put in place, between 2010 and 2030, the global consumption of antimicrobials in food-producing animals is projected to increase by about 67% (Van Boeckel et al., 2015[25]).
Figure 3.4. Associations between combined antibiotic consumption in animals and humans, and proportion of coli
Note: it is considered the average human being weighs 62.5 kg.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1, ECDC and ESVAC.
Antimicrobial consumption in animals favours AMR in ways that are similar to consumption in humans, so that more consumption may lead to more resistance. When animals develop resistant infections these may spread to humans through food, the environment (e.g. water and soil), or through direct contact between animals and humans (Tang et al., 2017[26]). Figure 3.4 illustrates the positive correlation that exists between combined consumption of antibiotics in the human and animal sectors and the proportion of infections in humans that are resistant to specific antibiotic treatments. The association between consumption and antibiotic-resistant E. coli is visibly stronger when taking into account animal use, besides human consumption only, even though recent analyses have found no direct link between antibiotic use in animals and resistance in E. coli in humans (ECDC, EFSA and EMA, 2017[27]).
Efforts to measure drug resistance should include measurement of antimicrobial consumption and resistance in animals. While monitoring and surveillance in the animal sectors are currently limited, international organisations are increasingly promoting not only measurement (ECDC/EFSA/EMA, 2017[22]), but also good governance in the veterinary sector, adoption of international standards, and antimicrobial stewardship (OIE, 2016[28]).
Per capita gross domestic product (GDP) is well correlated with the proportion of P. aeruginosa resistant to carbapenems (countries with a higher GDP per capita have lower proportions of infections caused by resistant bacteria), but less well correlated with rates of third-generation cephalosporin-resistant E. coli (data not shown). Resistance proportions for third-generation cephalosporin-resistant E. coli are more strongly associated with agricultural value added as a percentage of GDP (the higher the agricultural value added the higher the proportion of resistant infections).
In line with previous research (Alsan et al., 2015[15]), out-of-pocket health spending (as a percentage of total health expenditure) is positively associated with proportion of third-generation cephalosporin-resistant K. pneumoniae, but less well correlated with penicillin-resistant S. pneumoniae (relationship not shown). The number of doctors’ consultations per capita is well correlated with the proportion of S. aureus resistant to methicillin when disregarding observations for Mexico and Portugal, which have higher than expected resistance proportions given the low number of consultations in these countries. Mexico is often an outlier in these relationships, exhibiting higher than expected resistance proportions given levels of antibiotic consumption, doctors’ appointments and agricultural value added.
Another factor that has been shown to correlate well with drug resistance is governance (see Figure 3.6). The World Bank Worldwide Governance Indicators capture the views of thousands of stakeholders globally on six dimensions of governance: voice and accountability; political stability and absence of violence/terrorism; government effectiveness; regulatory quality; rule of law; and control of corruption (Kaufmann, Kraay and Mastruzzi, 2011[29]). The average country scores for these six dimensions are consistently well correlated with resistance proportions for six antibiotic-bacterium combinations, a relationship that is generally in agreement with Collignon and co-authors’ study (2015[16]) of the connection between poor governance and AMR. As illustrated below in Figure 3.6, the average score of the Worldwide Governance Indicators does a better job of explaining high resistance proportions in Mexico than the number of doctor appointments, value added from agriculture or even antibiotic consumption.
The average Worldwide Governance Indicators score is better correlated with resistance proportions than any other factor, including consumption of antibiotics (it is likely that the Worldwide Governance Indicators scores are associated antimicrobial use and misuse as well as infection and prevention control). However, the relationship between governance and drug resistance also has outliers. Ghana and Thailand exhibit much lower than expected rates of antibiotic-resistant infections (carbapenem-resistant P. aeruginosa in Ghana and carbapenem-resistant K. pneumoniae in Thailand) than suggested by their average governance scores, again suggesting there are multiple factors at play.
Figure 3.5. Associations between potential correlates of drug resistance and resistance proportions for four antibiotic-bacterium combinations in 2015
Note: Agriculture corresponds to ISIC divisions 1-5 and includes forestry, hunting, and fishing, as well as cultivation of crops and livestock production. The R-squared statistic for the bottom-right panel (S. aureus) excludes outliers Portugal and Mexico.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1, and data from World Bank and OECD databases.
Figure 3.6. Associations between average scores of worldwide governance indicators and proportion of infections caused by antibiotic-resistant bacteria in 2015
Source: OECD analyses of data from surveillance networks included in ResistanceMap1 and World Bank data.
While this analysis is merely illustrative (and while correlation is not causation), it suggests that even though some dimensions (e.g. governance, GDP, out-of-pocket health expenditure, value-added agriculture, among others) may be associated with AMR rates, there is no single major correlate or driver of drug resistance. The emergence and spread of AMR is a complex phenomenon with multiple interrelated causes and consequences. It exhibits many of the characteristics of complex adaptive systems (Plsek and Greenhalgh, 2001[30]), such as non-linear relationships, feedback loops and co‑evolution of multiple systems (e.g. human, animal and environment). For such a complex phenomenon, simple linear cause-effect relationships, even when part of multivariate models, will have limited value (Rutter et al., 2017[31]). As Wernli and colleagues (2017[13]) suggest, more and better data need to be collected and aggregated to inform more sophisticated analyses of determinants and correlates of AMR.
3.4. Should current trends continue into the future, resistance proportions will grow moderately in most countries, but there will be significant heterogeneity
The consequences of growing drug resistance, in terms of the future health and economic burden of AMR, have been used as a call for action. Some analyses concluded that, by 2050, the world could lose (Adeyi et al., 2017[32]; O'neill, 2016[33]): 10 million lives every year, 3.8% of its annual GDP, USD 1.2 trillion annually in additional health care expenditures, and an additional 28.3 million people to extreme poverty. These figures are based on hypothetical scenarios of the evolution of incident infections and the proportion resistant to antimicrobial treatment (KPMG LLP, 2014[34]). Motivated by the availability of internationally comparable data on resistance proportions (from surveillance networks included in ResistanceMap1) and by calls to make estimates more empirically driven (de Kraker, Stewardson and Harbarth, 2016[35]), the OECD projected resistance proportions for eight antibiotic-bacterium combinations for 52 countries (including members, accession and key partners of the OECD, as well as members of the G20 and countries in the European Economic Area) up to 2030 (see Box 3.2 for a description of the forecasting methodology used).
In OECD countries, average resistance proportions across eight antibiotic-bacterium combinations are estimated to have increased from 14% (range: 3-33%) in 2005 to 17% (range: 3-39%) in 2015, and may go up further to 18% (range: 6-40%) by 2030 if current trends in resistance, and correlates of resistance, continue into the future, and no policy actions are taken (see Figure 3.7). Growth in resistance proportions is thus projected to continue, though moderately, in OECD countries and beyond, under a business-as-usual scenario. Between 2005 and 2015, the predicted proportion of infections from resistant bacteria grew much faster than projected for the period 2015-2030, however these averages mask significant heterogeneity across countries and antibiotic-bacterium combinations.
Of the 52 countries in the analyses, the average resistance proportions for eight antibiotic-bacterium combinations could increase in 37 countries and decrease in 13 countries (see Table 3.3). Despite average reductions in a few countries, no country is projected to reduce resistance proportions for all eight antibiotic-bacterium combinations. On the contrary, five countries (Bulgaria, Luxembourg, Iceland, Slovenia and Denmark) could see resistance proportions increase for all eight combinations. The majority of countries could see both increases and reductions. Resistance proportions could increase in 64% of country-antibiotic-bacterium combinations (on average, proportions could grow in these country-antibiotic-bacterium combinations by 42%, when taking 2015 rates as base values), and could decrease in 36% of combinations (on average, proportions could drop by 9%, compared to proportions in 2015).
Figure 3.7. Average proportion of infections caused by bacteria resistant to antimicrobial treatment for eight antibiotic-bacterium combinations in 2005, 2015 and 2030
Note: * indicates country is missing more than 50% of observations, across all eight antibiotic-bacterium pairs, between 2005 and 2015. For countries on the left of this graph, resistance proportions will be higher in 2030, compared to 2015. For countries on the right, rates will be lower in 2030. Otherwise, countries are sorted left to right based on ascending resistance proportions in 2015.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1.
In OECD countries, growth rates of resistance proportions for third-generation cephalosporin-resistant E. coli and K. pneumoniae, using the average of country-specific growth rates, are projected to slow down significantly from 163% between 2005 and 2015 to 16% between 2015 and 2030. On the other hand, the proportion of infections caused by vancomycin-resistant E. faecium and E. faecalis are projected to grow at a rate of 31% between 2015 and 2030 in OECD countries, compared to a growth rate of 5% between 2005 and 2015 (the same trend is projected in G20 countries). Different regions could see very different progressions. While OECD countries could see proportions of penicillin-resistant S. pneumoniae grow faster in the coming decades, the opposite is true for EU countries. Conversely, rates of MRSA in OECD countries should continue to decline (if at a slower rate) but are expected to actually grow in EU countries between 2015 and 2030. Growth in carbapenem-resistant K. pneumoniae is projected to intensify in the coming years across all regions but especially in G20 countries and OECD key partners.
Table 3.3. Percentage point changes in resistance proportions for eight priority antibiotic-bacterium combinations between 2015 and 2030
Note: * indicates country is missing more than 50% of observations, across all eight antibiotic-bacterium pairs, between 2005 and 2015. The colour scheme is based on a two-point scale (minimum in light grey, maximum in blue, points in between coloured proportionally). Countries (and country groupings) are sorted top to bottom from highest to lowest average resistance proportions (across antibiotic-bacterium combinations). Antibiotic-bacterium combinations are sorted left to right from highest to lowest average resistance proportions (across countries). FREC: fluoroquinolone-resistant E. coli, VRE: vancomycin-resistant E. faecalis and E. faecium, 3GCREC: third-generation cephalosporin-resistant E. coli, CRKP: carbapenem-resistant K. pneumoniae, 3GCRKP: third-generation cephalosporin-resistant K. pneumoniae, CRPA: carbapenem-resistant P. aeruginosa, MRSA: methicillin-resistant S. aureus, PRSP: penicillin-resistant S. pneumoniae.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1.
Predicted historic and future resistance proportions for different antibiotic-bacterium combinations in selected OECD countries from 2000 to 2030 are illustrated in Figure 3.8. The heterogeneity in resistance proportions across countries, antibiotic-bacterium combinations and years, as well as the evolution of resistance proportions over time, is striking. The panels also illustrate data gaps and their impact on uncertainty in the estimation.
When considering all country-antibiotic-bacterium combinations together, resistance proportions could grow on average 23% between 2015 and 2030. However, this average masks significant heterogeneity in growth rates across countries and antibiotic-bacterium pairs. In OECD countries, the largest projected growth rate in resistance proportions is 50% for penicillin-resistant S. pneumoniae in the Slovak Republic (an increase of 14 percentage points, from 28% in 2015 to 42% in 2030). However, in the same country, resistance proportions for carbapenem-resistant P. aeruginosa could go down by 10 percentage points (from 58% to 49%, a growth rate of -16%). The largest projected reduction in resistance proportions is -27% for carbapenem-resistant K. pneumoniae in Italy (a drop of 10 percentage points, from 36% to 26%). However, the resistance proportions for third-generation cephalosporins-resistant E. coli in Italy could go up by 6 percentage points (from 31% to 37%, a growth rate of 20%).
Figure 3.8. Predicted historic and future resistance proportions for eight priority antibiotic-bacterium combinations in selected OECD countries from 2000 to 2030
Note: Points are original data, lines are means and bands are 95% uncertainty intervals.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1.
3.5. Growing resistance to third-line treatments and difficult-to-treat microorganisms are a concern for the coming years
Growth of resistance proportions for fluoroquinolone-resistant E. coli and third-generation cephalosporin-resistant E. coli and K. pneumoniae are projected to slow down from 2015 to 2030. In part, this waning growth could be due to a slowing of growth rates in the consumption of fluoroquinolones and third-generation cephalosporins in a large majority of countries for the period 2015-2030 compared to 2005‑15. Furthermore, consumption of third-generation cephalosporins could decrease in half of the countries analysed here. However, the use of fluoroquinolones could still grow by 27%, on average across all countries, by 2030. More than a fifth of E. coli in OECD countries, and close to half in G20 countries, were already resistant to fluoroquinolones in 2015, and resistance proportions for this antibiotic-bacterium combination are projected to keep growing up to 2030. The proportions of E. coli and K. pneumoniae infections resistant to third-generation cephalosporins are also projected to increase (see Figure 3.9).
Figure 3.9. Average proportions of colipneumoniaepneumoniae
Note: Stripes indicate average resistance for these antibiotic-bacterium combinations is projected to decrease up to 2030.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1.
These trends are concerning as both third-generation cephalosporins and fluoroquinolones are generally used as second-line treatments. Growing resistance to these antimicrobials could lead to increased use of third-line treatments like carbapenems, promoting the emergence and spread of carbapenem-resistant bacteria (ECDC, 2017[3]). WHO (2017[2]) considers the urgency of need for new antibiotics to treat carbapenem and third-generation cephalosporin-resistant Enterobacteriaceae (including E. coli and K. pneumoniae) is critical, and while proportions of carbapenem-resistant K. pneumoniae were relatively low in 2015, they are projected to increase (see Figure 3.10).
Figure 3.10. Trends in antimicrobial resistance in selected regions and country groups among high-priority antibiotic-bacterium combinations, by line of treatment
Note: Data were normalised to average antimicrobial resistance in 2005 (equal to 100) for each treatment line (e.g. a value of 150 for resistance to second-line treatments in 2015 in G20 countries means that resistance to second-line treatments is 50% higher than it was in 2005 in G20 countries). Resistance to first-line treatments is defined as the average of the proportions of penicillin-resistant S. pneumoniae and MRSA. Resistance to second-line treatments is the average of the proportions of E. coli and K. pneumoniae resistant to third-generation cephalosporins and of E. coli resistant to fluoroquinolones. Resistance to third-line treatments is defined as the proportion of K. pneumoniae resistant to carbapenems.
Source: OECD analyses of data from surveillance networks included in ResistanceMap1.
In EU28 and G20 countries, as well as OECD key partners, resistance proportions to third-line treatments are now growing faster than those of second-line treatments (markedly so in the EU28). The only options that will be left to treat infections with bacteria resistant to third-line treatments like carbapenems will be older antimicrobial agents with lower efficacy, such as polymyxins (e.g. colistin), or combination therapy. In certain countries with high resistance proportions, like Greece and Italy, resistance to polymyxins is already emerging with potentially disastrous consequences, as these remain one of the few last options for treatment (ECDC, 2015[36]).
Resistance among difficult to treat microorganisms like Enterococci (e.g. E. faecalis and E. faecium) and P. aeruginosa is also worrisome. These bacteria are intrinsically resistant to several antimicrobial agents and are challenging to contain in health care settings (ECDC, 2015[36]). Close to two-thirds of P. aeruginosa were already resistant to carbapenems in Romania and the Slovak Republic in 2015. In some countries with already high resistance (e.g. Greece, Turkey or Bulgaria), the proportion of carbapenem-resistant P. aeruginosa is projected to rise. Vancomycin-resistant E. faecalis and E. faecium is projected to rise in 36 countries while consumption of glycopeptides (e.g. vancomycin) could grow in 40 countries. Growing consumption might not be a problem in itself (as discussed in previous sections), but it is important to balance access to antimicrobial therapies with prudent and appropriate use (stewardship). As discussed in Chapter 5, stringent interventions for infection control are also essential to prevent further emergence and spread of resistance.
Proportions of MRSA dropped considerably across the majority of OECD countries between 2005 and 2015 but further progress could be slow, based on projected trends up to 2030. In some EU countries, reductions in AMR rates could be reversed altogether. Moreover, in the Slovak Republic, Hungary, the Russian Federation and the Czech Republic, the proportions of resistant bacteria have actually increased in recent years and should continue to rise in the coming decades. In some countries (e.g. South Africa, Croatia and Bulgaria), reductions in resistance proportions could be reversed, in line with projected growth in the consumption of penicillins. Despite recent trends, MRSA remains high in certain countries and the transfer of health care-associated clones into the community suggests comprehensive cross-sector policies are needed (ECDC, 2017[37]).
Conclusion: More and better data needed for robust policy making
The proportion of infections that are resistant to antimicrobial therapies will continue to grow, if moderately, in a majority of OECD, EU and G20 countries in the coming decades, if current trends in antimicrobial consumption, economic and population growth, and health spending continue into the future, and no policy action is taken. While some progress has been achieved in slowing down, or even reducing, resistance for MRSA, projections indicate these gains could be halted or reversed in the coming years, if no effective action is put in place. For other pathogens, the growth in resistance proportions of the last 10 to 15 years will likely persist, even if more moderately. There is significant variation across countries, bacteria and antibiotics, suggesting previous forecasting efforts assuming a single growth rate are not corroborated by available data. The reasons behind this variation remain, unfortunately, difficult to pinpoint with any certainty.
Data-driven models of AMR have limited explanatory power, as illustrated by wide confidence intervals (see Figure 3.8). There are important gaps in both data and understanding and these are naturally intertwined. The data that do exist are potentially biased and often not comparable due to challenges in defining and measuring drug resistance (see Box 3.1). More primary research and data collection are needed on the many facets of AMR, from the human health sector, to food and veterinary sectors, to travel, governance and general economic development. There are already some suggestions for what should be monitored, from access to sanitation and safe drinking water, to non-prescription availability and misuse of antibiotics, as well as the economic and health burden of drug resistant infections (Wernli et al., 2017[13]). There are innovative approaches to data collection such as citizen science initiatives in which samples are crowd-sourced from populations directly (Freeman et al., 2016[38]), and some pharmaceutical companies collect global surveillance data (Ashley et al., 2018[39]). Initiatives such as the WHO Global Antimicrobial Resistance Surveillance System (GLASS) and surveillance networks like EARS-Net are important as they set guidelines and standards that are shared by all participating partners. Countries should continue to promote the collection and dissemination of data on all aspects related to AMR.
More and better data are essential to producing high-quality estimates that, in turn, can inform robust policies in addressing growing drug resistance. Still, it is important not to delay action while surveillance and monitoring systems are being set up. While, as discussed in Chapter 5, a number of OECD countries have already put in place policies to tackle inappropriate consumption, as populations grow and age, more people will be at risk of infection and more antimicrobials will be consumed. Policy needs to go further, as the costs of inaction may be very significant as discussed in following chapter.
References
[32] Adeyi, O. et al. (2017), Drug-resistant infections : a threat to our economic future (Vol. 2) : final report (English), http://documents.worldbank.org/curated/en/323311493396993758/final-report (accessed on 16 January 2018).
[20] Albrich, W., D. Monnet and S. Harbarth (2004), “Antibiotic selection pressure and resistance in Streptococcus pneumoniae and Streptococcus pyogenes.”, Emerging infectious diseases, Vol. 10/3, pp. 514-7, http://dx.doi.org/10.3201/eid1003.030252.
[15] Alsan, M. et al. (2015), “Out-of-pocket health expenditures and antimicrobial resistance in low-income and middle-income countries: an economic analysis”, The Lancet Infectious Diseases, Vol. 15/10, pp. 1203-1210, https://doi.org/10.1016/S1473-3099(15)00149-8.
[39] Ashley, E. et al. (2018), “An inventory of supranational antimicrobial resistance surveillance networks involving low- and middle-income countries since 2000”, Journal of Antimicrobial Chemotherapy, Vol. 73/7, pp. 1737-1749, http://dx.doi.org/10.1093/jac/dky026.
[5] Australian Commission on Safety and Quality in Health Care (2017), AURA 2017: Second Australian report on antimicrobial use and resistance in human health, Australian Commission on Safety and Quality in Health Care, Sydney, https://www.safetyandquality.gov.au/wp-content/uploads/2018/01/AURA-2017-Second-Australian-report-on-Antimicrobial-Use-and-Resistance-in-human-health.pdf (accessed on 16 January 2018).
[23] Cassini, A. et al. (2018), “Attributable deaths and disability-adjusted life-years caused by infections with antibiotic-resistant bacteria in the European Union and the European Economic Area in 2015: a population-level health estimate”, Lancet Infectious Disease, http://dx.doi.org/10.1016/S1473-3099(18)30605-4.
[1] Cecchini, M., J. Langer and L. Slawomirski (2015), “Antimicrobial Resistance in G7 Countries and Beyond: Economic Issues, Policies and Options for Action”, OECD, Paris, http://www.oecd.org/els/health-systems/Antimicrobial-Resistance-in-G7-Countries-and-Beyond.pdf (accessed on 16 January 2018).
[16] Collignon, P. et al. (2015), “Antimicrobial Resistance: The Major Contribution of Poor Governance and Corruption to This Growing Problem”, PLOS ONE, Vol. 10/3, p. e0116746, http://dx.doi.org/10.1371/journal.pone.0116746.
[35] de Kraker, M., A. Stewardson and S. Harbarth (2016), “Will 10 Million People Die a Year due to Antimicrobial Resistance by 2050?”, PLOS Medicine, Vol. 13/11, p. e1002184, http://dx.doi.org/10.1371/journal.pmed.1002184.
[3] ECDC (2017), Antimicrobial resistance surveillance in Europe 2016. Annual Report of the European Antimicrobial Resistance Surveillance Network (EARS-Net)., ECDC, Stockholm, http://dx.doi.org/10.2900/296939.
[37] ECDC (2017), TESSy -The European Surveillance System: Antimicrobial resistance (AMR) reporting protocol 2017: European Antimicrobial Resistance Surveillance Network (EARS-Net) surveillance data for 2016., ECDC, Stockholm, https://ecdc.europa.eu/sites/portal/files/documents/EARS-Net-reporting-protocol-2017.pdf (accessed on 12 February 2018).
[36] ECDC (2015), Antimicrobial resistance surveillance in Europe 2015. Annual Report of the European Antimicrobial Resistance Surveillance Network (EARS-Net)., ECDC, Stockholm, https://ecdc.europa.eu/sites/portal/files/media/en/publications/Publications/antimicrobial-resistance-europe-2015.pdf (accessed on 13 February 2018).
[22] ECDC/EFSA/EMA (2017), “ECDC/EFSA/EMA second joint report on the integrated analysis of the consumption of antimicrobial agents and occurrence of antimicrobial resistance in bacteria from humans and food‐producing animals”, EFSA Journal, Vol. 15/7, http://dx.doi.org/10.2903/j.efsa.2017.4872.
[27] ECDC, EFSA and EMA (2017), ECDC/EFSA/EMA second joint report on the integrated analysis of the consumption of antimicrobial agents and occurrence of antimicrobial resistance in bacteria from humans and food-producing animals Joint Interagency Antimicrobial Consumption and Resistance Analysis (JIACRA) Report, http://dx.doi.org/10.2903/j.efsa.2017.4872.
[38] Freeman, S. et al. (2016), “Crowdsourced Data Indicate Widespread Multidrug Resistance in Skin Flora of Healthy Young Adults.”, Journal of microbiology & biology education, Vol. 17/1, pp. 172-82, http://dx.doi.org/10.1128/jmbe.v17i1.1008.
[21] Goossens, M., B. Catry and J. Verhaegen (2013), “Antimicrobial resistance to benzylpenicillin in invasive pneumococcal disease in Belgium, 2003-2010: the effect of altering clinical breakpoints.”, Epidemiology and infection, Vol. 141/3, pp. 490-5, http://dx.doi.org/10.1017/S0950268812001057.
[8] Hindler, J. and J. Stelling (2007), “Analysis and Presentation of Cumulative Antibiograms: A New Consensus Guideline from the Clinical and Laboratory Standards Institute”, Clinical Infectious Diseases, Vol. 44/6, pp. 867-873, http://dx.doi.org/10.1086/511864.
[9] Kassim, A. et al. (2016), “Comparison of Clinical Laboratory Standards Institute and European Committee on Antimicrobial Susceptibility Testing guidelines for the interpretation of antibiotic susceptibility at a University teaching hospital in Nairobi, Kenya: a cross-sectional study”, Annals of Clinical Microbiology and Antimicrobials, Vol. 15/1, p. 21, http://dx.doi.org/10.1186/s12941-016-0135-3.
[29] Kaufmann, D., A. Kraay and M. Mastruzzi (2011), “The Worldwide Governance Indicators: Methodology and Analytical Issues”, Hague Journal on the Rule of Law, Vol. 3/02, pp. 220-246, http://dx.doi.org/10.1017/S1876404511200046.
[19] Klein, S. (2000), “The effects of hormones on sex differences in infection: from genes to behavior”, Neuroscience & Biobehavioral Reviews, Vol. 24/6, pp. 627-638, https://doi.org/10.1016/S0149-7634(00)00027-0.
[34] KPMG LLP (2014), The global economic impact of anti-microbial resistance, KPMG LLP, https://home.kpmg.com/content/dam/kpmg/pdf/2014/12/amr-report-final.pdf (accessed on 16 January 2018).
[14] Mendelson, M. et al. (2016), “Maximising access to achieve appropriate human antimicrobial use in low-income and middle-income countries”, The Lancet, Vol. 387/10014, pp. 188-198, https://doi.org/10.1016/S0140-6736(15)00547-4.
[6] OECD (2017), Tackling Wasteful Spending on Health, OECD Publishing, Paris, https://doi.org/10.1787/9789264266414-en.
[28] OIE (2016), The OIE Strategy on Antimicrobial Resistance and the Prudent Use of Antimicrobials, OIE, http://www.oie.int/fileadmin/Home/eng/Media_Center/docs/pdf/PortailAMR/EN_OIE-AMRstrategy.pdf (accessed on 14 February 2018).
[33] O'neill, J. (2016), Tackling drug-resistant infections globally: Final report and recommendations, https://amr-review.org/sites/default/files/160525_Final%20paper_with%20cover.pdf (accessed on 13 February 2018).
[30] Plsek, P. and T. Greenhalgh (2001), “Complexity science: The challenge of complexity in health care.”, BMJ (Clinical research ed.), Vol. 323/7313, pp. 625-8, http://www.ncbi.nlm.nih.gov/pubmed/11557716 (accessed on 16 January 2018).
[17] Rönnerstrand, B. and V. Lapuente (2017), “Corruption and use of antibiotics in regions of Europe”, Health Policy, Vol. 121/3, pp. 250-256, http://dx.doi.org/10.1016/J.HEALTHPOL.2016.12.010.
[31] Rutter, H. et al. (2017), “The need for a complex systems model of evidence for public health.”, Lancet, Vol. 390/10112, pp. 2602-2604, https://doi.org/10.1016/S0140-6736(17)31267-9.
[18] Simon, A., G. Hollander and A. McMichael (2015), “Evolution of the immune system in humans from infancy to old age.”, Proceedings. Biological sciences, Vol. 282/1821, p. 20143085, http://dx.doi.org/10.1098/rspb.2014.3085.
[26] Tang, K. et al. (2017), “Restricting the use of antibiotics in food-producing animals and its associations with antibiotic resistance in food-producing animals and human beings: a systematic review and meta-analysis”, The Lancet Planetary Health, Vol. 1/8, pp. e316-e327, https://doi.org/10.1016/S2542-5196(17)30141-9.
[7] Turnidge, J. and D. Paterson (2007), “Setting and revising antibacterial susceptibility breakpoints.”, Clinical microbiology reviews, Vol. 20/3, pp. 391-408, table of contents, http://dx.doi.org/10.1128/CMR.00047-06.
[25] Van Boeckel, T. et al. (2015), “Global trends in antimicrobial use in food animals”, Proceedings of the National Academy of Sciences, Vol. 112/18, pp. 5649-5654, http://dx.doi.org/10.1073/pnas.1503141112.
[10] van der Bij, A. et al. (2012), “Clinical breakpoint changes and their impact on surveillance of antimicrobial resistance in Escherichia coli causing bacteraemia”, Clinical Microbiology and Infection, Vol. 18/11, pp. E466-E472, http://dx.doi.org/10.1111/j.1469-0691.2012.03996.x.
[13] Wernli, D. et al. (2017), “Antimicrobial resistance: The complex challenge of measurement to inform policy and the public”, PLOS Medicine, Vol. 14/8, p. e1002378, http://dx.doi.org/10.1371/journal.pmed.1002378.
[12] WHO (2017), Global antimicrobial resistance surveillance system (GLASS) report: early implementation 2016-2017, World Health Organization, Geneva, http://apps.who.int/iris/bitstream/10665/259744/1/9789241513449-eng.pdf?ua=1 (accessed on 12 February 2018).
[2] WHO (2017), Prioritization of pathogens to guide discovery, research and development of new antibiotics for drug-resistant bacterial infections, including tuberculosis., World Health Organisation, Geneva, http://www.who.int/medicines/areas/rational_use/PPLreport_2017_09_19.pdf?ua=1 (accessed on 12 February 2018).
[4] WHO (2014), Antimicrobial resistance: global report on surveillance 2014, http://apps.who.int/iris/bitstream/handle/10665/112642/9789241564748_eng.pdf;jsessionid=7896DA46F16714263D42D26A092CB051?sequence=1 (accessed on 09 October 2018).
[11] Wolfensberger, A. et al. (2013), “Change of Antibiotic Susceptibility Testing Guidelines from CLSI to EUCAST: Influence on Cumulative Hospital Antibiograms”, PLoS ONE, Vol. 8/11, p. e79130, http://dx.doi.org/10.1371/journal.pone.0079130.
[24] Zilberberg, M. et al. (2017), “Carbapenem resistance, inappropriate empiric treatment and outcomes among patients hospitalized with Enterobacteriaceae urinary tract infection, pneumonia and sepsis”, BMC Infectious Diseases, Vol. 17/1, p. 279, http://dx.doi.org/10.1186/s12879-017-2383-z.
Notes
← 1. ResistanceMap aggregates data from international surveillance networks like EARS-Net, CEASER, GLASS, and others, which in turn aggregate data from national surveillance networks.
← 2. The European Antimicrobial Resistance Surveillance Network (EARS-Net) is the largest publicly funded system for antimicrobial resistance surveillance in Europe. Participating countries include Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Romania, Slovak Republic, Slovenia, Spain, Sweden, and the United Kingdom. The Central Asian and Eastern European Surveillance of Antimicrobial Resistance (CAESAR) network includes all countries of the WHO European Region that are not part of EARS-Net, namely: Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Georgia, Kazakhstan, Kyrgyzstan, Montenegro, the Republic of Moldova, the Russian Federation, Serbia, Switzerland, Tajikistan, the former Yugoslav Republic of Macedonia, Turkey, Turkmenistan, Ukraine and Uzbekistan, as well as Kosovo (in accordance with United Nations Security Council resolution 1244 (1999)).