View
214
Download
0
Category
Preview:
DESCRIPTION
http://www.cddep.org/sites/default/files/535_vanboeckelglobal_2015.pdf
Citation preview
Global trends in antimicrobial use in food animalsThomas P. Van Boeckela,1, Charles Browerb, Marius Gilbertc,d, Bryan T. Grenfella,e,f, Simon A. Levina,g,h,1,Timothy P. Robinsoni, Aude Teillanta,e, and Ramanan Laxminarayanb,e,j,1
aDepartment of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544; bCenter for Disease Dynamics, Economics & Policy,Washington, DC 20036; cUniversite Libre de Bruxelles, B1050 Brussels, Belgium; dFonds National de la Recherche Scientifique, B1000 Brussels, Belgium;ePrinceton Environmental Institute, Princeton, NJ 08544; fFogarty International Center, National Institutes of Health, Bethesda, MD 20892; gBeijer Instituteof Ecological Economics, 10405 Stockholm, Sweden; hResources for the Future, Washington, DC 20036; iInternational Livestock Research Institute, 00100Nairobi, Kenya; and jPublic Health Foundation of India, New Delhi 110070, India
Contributed by Simon A. Levin, February 18, 2015 (sent for review November 21, 2014; reviewed by Delia Grace and Lance B. Price)
Demand for animal protein for human consumption is risingglobally at an unprecedented rate. Modern animal productionpractices are associated with regular use of antimicrobials, poten-tially increasing selection pressure on bacteria to become resistant.Despite the significant potential consequences for antimicrobialresistance, there has been no quantitative measurement of globalantimicrobial consumption by livestock. We address this gap byusing Bayesian statistical models combining maps of livestockdensities, economic projections of demand for meat products, andcurrent estimates of antimicrobial consumption in high-incomecountries to map antimicrobial use in food animals for 2010 and2030. We estimate that the global average annual consumption ofantimicrobials per kilogram of animal produced was 45 mg·kg−1,148 mg·kg−1, and 172 mg·kg−1 for cattle, chicken, and pigs, respec-tively. Starting from this baseline, we estimate that between 2010and 2030, the global consumption of antimicrobials will increaseby 67%, from 63,151 ± 1,560 tons to 105,596 ± 3,605 tons. Up toa third of the increase in consumption in livestock between 2010and 2030 is imputable to shifting production practices in middle-income countries where extensive farming systems will be re-placed by large-scale intensive farming operations that routinelyuse antimicrobials in subtherapeutic doses. For Brazil, Russia,India, China, and South Africa, the increase in antimicrobial con-sumption will be 99%, up to seven times the projected populationgrowth in this group of countries. Better understanding of the con-sequences of the uninhibited growth in veterinary antimicrobialconsumption is needed to assess its potential effects on animaland human health.
antimicrobials | livestock | mapping | drug resistance | linear regression
Antimicrobials are widely used for disease prevention andgrowth promotion in food animals. In the United States,
antimicrobial use in food animals is estimated to account for∼80% of the nation’s annual antimicrobial consumption (1), asignificant fraction of which involves antimicrobials that areimportant in human medicine in the treatment of common infec-tions and also necessary to perform medical procedures such asmajor surgeries, organ transplantation, and chemotherapy (2).This widespread use of antimicrobials in livestock contrib-
utes––by means of natural selection––to the emergence of anti-microbial-resistant bacteria (ARBs) and has significant publichealth implications: ARBs of animal origin can be transmitted tohumans through the environment (3) and food products (4) andto agricultural workers by direct contact (5). Although direct cau-sality is difficult to establish because of the ecological nature ofantibiotic selection pressure, studies have shown a close associationbetween the prevalence of livestock-associated ARBs in animalsand in humans (6), as well as between the levels of antimicrobialuse in animals at a population level, and the prevalence of ARBs inanimals (7) and in humans (8). A recent study from seven Euro-pean countries (Norway, Sweden, Denmark, Austria, Switzerland,The Netherlands, and Belgium) showed a strong correlation be-tween consumption levels for eight classes of antimicrobials (9) andthe prevalence of antimicrobial-resistant commensal Escherichia
coli in pigs, poultry, and cattle. Several works additionally sug-gested that repeated exposure to low doses of antimicrobialagents––the context in which growth-promoting antimicrobialsand prophylactic are administered––creates ideal conditions forthe emergence and spread of ARBs in animals (10).In low- and middle-income countries, rising incomes have
driven an unprecedented growth in demand for animal protein(11) and, as a result, the global biomass of animals raised forfood now exceeds the global biomass of humans (12). In Asia,daily animal protein intake grew from 7 grams per capita per dayto 25 grams per capita per day (12) between 1960 and 2013 whilethe proportion of the diet coming from rice and wheat pro-gressively decreased, primarily among higher-income adults (13).To meet this demand, countries such as Brazil, Russia, India,China, and South Africa (BRICS) have shifted toward highlycost-efficient and vertically integrated intensive livestock pro-duction systems. Because these production systems necessitateantimicrobials to keep animals healthy and maintain productivity,rising incomes in transitioning countries are effectively driving anincrease in antimicrobial consumption and thereby antimicrobialresistance. Meanwhile, multiresistant ARBs have been isolated infood animals in BRICS countries (14, 15) and throughout the de-veloping world where the use of antimicrobials for growth pro-motion remains largely unregulated (16).The challenges of the nutritional transition to animal protein-
based diets and the rise of antimicrobial resistance are thusclosely linked: The use of antimicrobials as growth promoters
Significance
Antimicrobials are used in livestock production to maintainhealth and productivity. These practices contribute to the spreadof drug-resistant pathogens in both livestock and humans, pos-ing a significant public health threat. We present the first globalmap (228 countries) of antibiotic consumption in livestock andconservatively estimate the total consumption in 2010 at 63,151tons. We project that antimicrobial consumption will rise by67% by 2030, and nearly double in Brazil, Russia, India, China,and South Africa. This rise is likely to be driven by the growthin consumer demand for livestock products in middle-incomecountries and a shift to large-scale farms where antimicrobialsare used routinely. Our findings call for initiatives to preserveantibiotic effectiveness while simultaneously ensuring foodsecurity in low- and lower-middle-income countries.
Author contributions: T.P.V.B. and R.L. designed research; T.P.V.B., C.B., and A.T. per-formed research; C.B. and M.G. contributed new reagents/analytic tools; T.P.V.B. andA.T. analyzed data; and T.P.V.B., C.B., M.G., B.T.G., S.A.L., T.P.R., A.T., and R.L. wrotethe paper.
Reviewers: D.G., International Livestock Research Institute; and L.B.P., Translational Ge-nomics Research Institute.
The authors declare no conflict of interest.1To whom correspondence may be addressed. Email: thomas.van.boeckel@gmail.com,slevin@princeton.edu, or ramanan@cddep.org.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1503141112/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1503141112 PNAS Early Edition | 1 of 6
AGRICU
LTURA
LSC
IENCE
S
and therapeutics to support the growing demand for meat isplacing ever greater selection pressure for resistant strains ofbacteria to evolve. Whereas trends in antibiotic consumption inhumans are now being tracked in most high-income and somemiddle-income countries through databases on antibiotic sales(17, 18), antimicrobial consumption in livestock has receivedcomparatively little attention. Expert opinion suggests that globalconsumption of antimicrobials in animals is twice that of humans(19). However, the underlying data from the veterinary sectorsupporting these claims are weak and lack standardization.Without reliable evidence to estimate global antimicrobial con-sumption in livestock, the links between antimicrobial consump-tion and resistance patterns are poorly quantified, and efforts andpolicies to optimize antibiotic use in animals are poorly targeted.This study addresses these gaps in our understanding of anti-
microbial use in animals. We use statistical models combiningmaps of livestock densities and reports of antimicrobial consump-tion in high-income countries to estimate and map the globalconsumption of antimicrobials in food animals for 2010. We thenproject antimicrobial consumption trends for the year 2030 basedon growth projections of the consumption of livestock products.
ResultsOverall Antimicrobial Consumption Trends. Global consumption ofantimicrobials in food animal production was estimated at 63,151(±1,560) tons in 2010 and is projected to rise by 67%, to 105,596(±3,605) tons, by 2030. Two thirds (66%) of the global increase(67%) in antimicrobial consumption is due to the growing numberof animals raised for food production. The remaining third (34%)is imputable to a shift in farming practices, with a larger pro-portion of animals projected to be raised in intensive farmingsystems by 2030. In Asia alone, as much as 46% of the increasein antimicrobial consumption by 2030 is likely due to shifts inproduction systems. By 2030, antimicrobial consumption in Asiais projected to be 51,851 tons, representing 82% of the currentglobal antimicrobial consumption in food animals in 2010.In 2010, the five countries with the largest shares of global
antimicrobial consumption in food animal production wereChina (23%), the United States (13%), Brazil (9%), India (3%),and Germany (3%) (Fig. 1). By 2030, this ranking is projected tobe China (30%), the United States (10%), Brazil (8%), India(4%), and Mexico (2%). Among the 50 countries with the largestamounts of antimicrobials used in livestock in 2010, the fivecountries with the greatest projected percentage increases inantimicrobial consumption by 2030 are likely to be Myanmar(205%), Indonesia (202%), Nigeria (163%), Peru (160%), andVietnam (157%). China and Brazil are among the largest con-sumers of antimicrobials currently but are not the countries with
the most rapid projected increases in antimicrobial consumption.This indicates that these two countries have already initiated ashift toward more intensified livestock production systems usingantimicrobials to maintain animal health and increase productivity.Antimicrobial consumption for animals in the BRICS coun-tries is expected to grow by 99% by 2030, whereas their humanpopulations are only expected to grow by 13% over the sameperiod (20).
Consumption by Type of Livestock. The global estimates of anti-microbial consumption presented in this study are based onspecies-specific coefficients of antimicrobial consumption perpopulation correction unit (PCU). Using a Bayesian regressionframework, we estimated the posteriors distributions for thesecoefficients for intensively farmed animals (Fig. 2). The mean ofthe posterior for antimicrobial consumption in cattle was gen-erally lower (45 mg/PCU) than for chickens (148 mg/PCU) andpigs (172 mg/PCU). The difference in Bayes’ factors between acomplete regression model including cattle, chickens, and pigsand regression models including two types of animals were, re-spectively, 1.32 for a model including just chicken and pigs, 78for a model including just cattle and pigs, and 1.72 × 109 for amodel including just cattle and chickens. This indicates thatdropping chicken and pigs from the regression models resultedin a significant loss of predictive power to estimate the overallantimicrobial consumption, and that the number of pig PCUsbest explained the differences in overall antimicrobial consump-tion between countries. The higher dispersion of the posteriordistribution of chicken production compared with that of pig pro-duction suggests that intensive chicken production showed a widerrange of intensity of antimicrobial use across countries than didpork production.
Geographical Patterns. Antimicrobial consumption displayed im-portant geographic heterogeneity across continents. In Southand Southeast Asia, antimicrobial consumption hotspots includethe southeast coast of China, Guangdong and Sichuan provinces,(Fig. 3, Top), the Red River delta in Vietnam, the northernsuburbs of Bangkok, and the south coast of India and the cities ofMumbai and Delhi. In the Americas, the highest consumption ofantimicrobials was observed in the south of Brazil, the suburbsof Mexico City, and midwestern and southern United States. Theonly notable hotspots of antimicrobial consumption in Africawere the Nile delta and the city of Johannesburg and its sur-rounding townships. The uncertainty bounds associated with thespatial predictions of antimicrobial consumption are presentedin Fig. 3 (Bottom). In general, the SDs of the coefficients ofantimicrobial consumption per PCU were moderate in regionswhere intensive farming practices are common and food animalsare densely populated. Higher uncertainty in the model predictionwas observed in Central Asia, Ethiopia, Canada, and eastern In-dia, for example.When disaggregated by food animal species, the geographical
distributions of antimicrobial consumption display distinct spa-tial patterns according to regional production patterns. Fig. S1shows these patterns within the European Union. Most of theantimicrobial consumption associated with chicken productionis found in Flanders (Belgium), The Netherlands, the BritishMidlands, Brittany (France), and the Po Valley (Italy). Con-sumption in pork production is largely concentrated in northernGermany, Denmark, The Netherlands, northern France, north-ern Belgium, Madrid and the autonomous region of Cataluña inSpain, and the Po Valley. Comparatively, the geographic inten-sity of antimicrobial consumption in cattle production was lowacross Europe because of the lower use of antimicrobials per PCUand the lower animal densities characteristic of cattle, comparedwith chicken and pig.
Fig. 1. (A) Largest five consumers of antimicrobials in livestock in 2010.(B) Largest five consumers of antimicrobials in livestock in 2030 (projected).(C) Largest Increase in antimicrobial consumption between 2010 and 2030.(D) Largest relative increase in antimicrobial consumption between 2010 and2030. CHN, China; USA, United States; BRA, Brazil; DEU, Germany; IND, India;MEX, Mexico; IDN, Indonesia; MMR, Myanmar; NGA, Nigeria; PER, Peru;PHL, Philippines.
2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1503141112 Van Boeckel et al.
In Asia, antimicrobial consumption in chicken and pigs is ex-pected to grow by 129% and 124%, respectively, by 2030 (Fig. 4).However, the total acreage of areas where antimicrobial con-sumption is currently greater than 30 kg·km−2 will grow by 4%for pork and 143% for chicken. This has potentially importantlogistical implications for surveillance programs to track the emer-gence of ARBs over larger portions of land. The extreme growthin consumption for chickens is primarily the result of the expan-sion of this sector in India alone, where areas of high consumption(30 kg·km−2) are expected to grow 312% by 2030.
DiscussionIn this study, we use statistical models to map the global con-sumption of antimicrobials in food animals for 2010 and 2030.This is the first study that we are aware of that attempts to quantifyantimicrobial consumption in food animals at a global scale.As with any model-based study, our analysis is subject to
assumptions and limitations. Data on antimicrobial use in live-stock are scarce, stemming from both the lack of publicly fundedsurveillance systems and the reluctance of food animal pro-ducers, animal feed producers, and veterinary pharmaceuticalcompanies to provide comprehensive reports of antimicrobialconsumption or sales. For this study, estimates of antimicrobialconsumption could be obtained for only 32 countries, all of whichwere high income. These data were first interpolated amongother high-income countries and subsequently extrapolated toestimate antimicrobial consumption in intensive production sys-tems of low- and middle-income countries. This modeling strat-egy was necessarily chosen as a result of the lack of systematic andreliable reports on antimicrobial sales in middle- and low-incomecountries. The underlying assumption implicit to this modelingstrategy is that because they are highly standardized, intensivefarming operations use similar quantities of antimicrobials acrosshigh-, middle-, and low-income countries. Additionally, severalEuropean countries used to train the statistical model have ex-perienced declining sales of antimicrobials for animal consump-tion, ranging from 0.4% to 28% between 2010 and 2011 (21), andseveral Organization for Economic Co-operation and Devel-opment (OECD) countries are currently engaged in initiativesaiming at reducing antibiotic use in livestock production (22).Additionally, patterns of antimicrobial consumption in middle-income and high-income countries differ in many respects. Inmost instances, the absence of clear legislative framework on theuse of antimicrobials in livestock production in most middle- andlow-income countries may result in increased irrational con-sumption. Our estimates of antibiotic consumption in 2010 maythus represent an overestimation of the current consumption levels.However, several of our assumptions may also result in under-estimating the global antimicrobial consumption. First, among thecountries used to train the statistical models, 25 are subject to
a ban on antimicrobial use for growth promotion, and two aresubject to a partial ban (Australia and New Zealand) (16). Second,the United States, where withdrawal of antimicrobials for growthpromotion is voluntary, was excluded from the model-fitting pro-cedure to prevent its comparatively larger statistical weight fromartificially increasing the significance of the linear regression. In-dividual US state estimates could have helped overcome theseissues, but these were not available. As a result, we likely under-estimate consumption of antimicrobials in livestock in the UnitedStates. Additionally, because this information was missing in themajority of countries used to train the statistical models, this studydoes not evaluate antimicrobial consumption on a compound-specific basis. Finally, the total volume figures do not account forchoices of drugs, potential differences in drug potencies, resistanceselection pressures, or use for treatment in human medicine.A potential caveat of this study is that ionophores—compoundsthat are used only in animals—are reflected in our estimates of theglobal antimicrobial consumption because these are generallypooled with medically important antimicrobials in national reportsof total antimicrobial sales. As a result, our estimates of antimi-crobial consumption may not always reflect differences in expo-sures to antimicrobials among countries.The introduction of a binary distinction between extensively
and intensively raised livestock masks a variety of productionprocesses prevalent globally but is relatively well documented forpoultry production systems in Asia (Fig. S2) (23) and was re-cently validated for Thailand (24). For pigs, production systemsof intermediate size (semiintensive) may represent a nonnegligibleshare of production, but these were not treated as an individualcategory for this study. Our simplifications were chosen over morespeculative and arbitrary modeling assumptions that could haveintroduced additional uncertainty and potentially affected theprojected trends for growth in antimicrobial consumption. Theestimates presented in this study should therefore be seen as con-servative estimates of antimicrobial consumption in food pro-duction, barring major changes in the global regulatory frameworkof these substances over the next 2 decades.Globally, intensive livestock farming has increased food pro-
duction at a low cost per unit produced, but perhaps at an un-recognized price paid in increased antimicrobial resistance.Linking antimicrobial consumption in animals to drug-resistantinfections of humans is inherently complex owing to the ecologicalnature of the selection pressure for drug-resistant pathogens aswell as the existence of indirect routes of transmission throughthe environment. However, in recent years, a growing body of
Antimicrobial (mg) / PCU
Pro
b. d
ensi
ty
0 100 200 300
00.
02
Fig. 2. Posterior distributions for estimates of antimicrobial consumption incattle, chickens, and pigs in OECD countries.
Log10 [(mg/pixel)+1]004567891
Std. Dev. mg/PCUN a
0 -20
30
40
50
75 015 00
20 00
- 1- 4- 5- 6- 7- 8- 9- 10
0 - 11
o Dat
20- 30
- 40
- 50
- 75
- 150 - 2
0 - 5
Fig. 3. Global antimicrobial consumption in livestock in milligrams per 10 km2
pixels (Top) and average SD of estimates of milligrams per PCU (Bottom).
Van Boeckel et al. PNAS Early Edition | 3 of 6
AGRICU
LTURA
LSC
IENCE
S
evidence has accumulated that strengthens the hypothesis that theroutine (25) use of antimicrobials in intensive animal productionsystems constitutes a waste of natural resources (26)—antimicro-bial effectiveness—that is of crucial importance in human medi-cine. Intensive farming practices have not only been associatedwith antimicrobial resistance in animals, humans, and meat butalso with numerous other livestock diseases such as highly path-ogenic avian influenza H5N1 (27) and porcine reproductive andrespiratory syndrome (28). Beyond animal health and antimicro-bial resistance, other negative externalities associated with poorlyregulated intensive farming include water and soil pollution (29),loss of biodiversity (30), and decline of meat nutritional quality(31). All of these have severe consequences that potentially out-weigh long-term benefits of increased productivity.Mapping the antimicrobial consumption in livestock provides
a baseline estimate of its global importance. Similar mappingexercises have been conducted for other major public healthissues, such as malaria (32) or tuberculosis (33). Leading acomparable initiative for antimicrobial consumption providesseveral insights. First, it provides an objective data-driven esti-mate of the potential magnitude of antimicrobial consumption atthe global scale, whereas previous estimates were based on ex-pert opinions (34). Second, it identifies regions at higher risk ofemergence of drug-resistant pathogens—places where surveil-lance and intervention efforts should be targeted. Third, thisbaseline estimate can be used to evaluate the progress (35) offuture antimicrobial stewardship efforts. Finally, this approachcan be adapted to predict antimicrobial consumption in the fu-ture using updated maps of livestock, and thus continuouslyupdate projections for the evolution of antimicrobial resistancein livestock and humans.If regulatory action is not taken, our projections suggest that
global antimicrobial consumption in food animals will grow byat least 67% by 2030. This corresponds to a compound annual
growth rate of 2.60%, a rate comparable to the compound an-nual growth rate for consumption of antibiotics in humans forthe period 2000–2010 (2.84%) (18) but almost threefold theprojected annual growth rate of the human population (0.98%)from 2010 to 2030. In 2001, Wise estimated the annual antimi-crobial market at 100,000–200,000 tons (34), but Wise’s meth-odology is unknown to us. Under the plausible hypothesis thatglobal human antimicrobial consumption is likely smaller or atbest equivalent to animal consumption, and that human con-sumption grew by 36% between 2000 and 2010 (17), we find thatour estimate for antimicrobial consumption in 2010 (63,151 tons)is surprisingly consistent with the expert opinion estimate (34).Finally, up to one third of the increase in antimicrobial consump-tion in animals between 2010 and 2030 will be imputable to a shifttoward intensive production systems where antimicrobials areused routinely in subtherapeutic doses for disease prevention andgrowth promotion, rather than for disease treatment (16, 36, 37).In 2010, China was the largest antimicrobial consumer for
livestock, and we estimate that its livestock industry will use up to30% of the global antimicrobial production by 2030. Anothercountry contributing to a large share of the overall growth inantimicrobial consumption in food animals, if current trendscontinue, is India––a country already confronted with antibioticoveruse in human medicine and an extremely high (and increasing)prevalence of ARBs (e.g., ∼95% of adults in India carry bacteriaresistant to β-lactam antimicrobials) (38). Widespread resistancemay be more consequential for India than for other countries be-cause India’s bacterial disease burden is among the highest in theworld, and therefore antimicrobials play a critical role in limitingmorbidity and mortality (39). Currently, India has no regulatoryprovisions for the use of antimicrobials in cattle, chicken, and pigsraised for domestic consumption, nor do the majority of middle-income countries for which substantial growth in antimicrobialconsumption over the next 15 y is predicted (16). Recent studiesin various regions of India have discovered antimicrobial resi-dues in food animal products (such as chicken meat and milk)(40), indicating that antibiotic use in food animal production iswidespread and current regulation is nonexistent for domesticproduction. Limiting antimicrobial consumption in both humansand livestock may present a formidable challenge for Indianpublic health authorities, but it might also be an opportunity forthe country to take a regional lead in tackling this problem. Forinstance, neighbors such as Pakistan, Bangladesh, Nepal, and SriLanka are likely to be guided and influenced by regulatory actionin India, given the interconnectedness of the region’s pharma-ceutical commerce (41).The role of aquaculture has not been investigated in the
present study. However, this industry may represent a large shareof the antimicrobial consumption for an increasing number ofcountries in Southeast Asia. For instance, studies of antimicro-bial consumption in fish farming in Chile (42) and shrimp farmingin Vietnam (43) demonstrate that aquaculture is associated withextremely high rates of antimicrobial consumption per PCU [upto 1,400 mg/PCU reported for salmon farming in Chile (42)]. Asthe aquaculture industry grows (44) and shifts toward more-efficient production systems, it could constitute a major source ofantimicrobial contamination of the aquatic environment over thecoming decades.The analysis presented here is based on the very limited
available evidence on antimicrobial consumption in livestockproduction. We provide somewhat crude estimates of presentand projected antimicrobial consumption in food animals in2030. Our estimates of absolute values for the global antimi-crobial consumption should therefore be viewed with caution.Antimicrobial consumption levels in middle- and low-incomecountries were extrapolated from consumption levels in intensiveproduction systems in high-income countries. This methodologymay result in significant uncertainties when evaluating the global
A
B
Antimicrobial (Kg)No Data
0 - 10
10 - 30
30 - 250
> 250
> 30 in 2030
Fig. 4. Antimicrobial consumption in chickens (A) and pigs (B) in 2010.Purple indicates new areas where antimicrobial consumption will exceed30 kg per 10 km2 by 2030.
4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1503141112 Van Boeckel et al.
antimicrobial consumption level in livestock production. How-ever, working under the hypothesis that our methodology issubject to a systematic error over the period 2010–2030, weproject that, in relative terms, the consumption of antimicrobialsin food animals will grow significantly by 2030 owing to an im-portant increase in demand for meat in middle- and low-incomecountries. This indicates a potentially growing contribution ofthese countries to the global burden of antimicrobial resistance.In fast-growing Asian countries, this will constitute a seriouschallenge because these countries are currently experiencing themost rapid increase in demand for meat products (45), but re-gulations on antimicrobial use (for the domestic market) are stilllacking and surveillance information on antimicrobial consump-tion is either nonexistent or not publicly available. Despite itsregional nature, this process will unavoidably drive a global in-crease in the prevalence of ARBs through trade and transportnetworks, with potentially important consequences for humanhealth. This rise of ARBs seems potentially reversible for certaincompounds: Withdrawal of antimicrobials for growth promotionin several European countries led to a decrease in the prevalenceof ARBs, but the duration needed for reversal is unclear and thesubject of ongoing investigation (7, 46).Given the potential costs of inaction, this study, among others,
calls for urgent and concerted action in all countries, which isneeded to limit the overuse and abuse of antimicrobials in foodanimal production (2, 18, 47). These actions should include(i) implementation of a publicly funded international surveil-lance network of antimicrobial consumption in food animals incountries undergoing rapid intensification in the livestock sector,(ii) collaboration with veterinary drug manufacturers and animalfeed producers to cross-validate estimates of consumption withsales data, (iii) implementation of an international agenda toharmonize regulatory frameworks among countries, and (iv) theultimate phasing out of antimicrobial use for growth promotion,based on the successful experience in the European Union andthe new biological (48, 49) and economic (50, 51) evidencechallenging the purported benefits of antimicrobial use in foodanimal production.
MethodsIn the absence of systematic and harmonized data on antimicrobial con-sumption in livestock, we use indirect means to estimate antimicrobialconsumption (in milligrams of active ingredient per kilogram of animal) forcattle, pigs, and chickens raised in both extensive and intensive farmingsystems in 228 countries. In our study, intensive production refers to highinput–high output systems that, compared with extensive systems (backyardproduction), achieve greater economies of scale and efficiency while alsopossibly using mechanized labor, operating with high animal densities, andusing specialized breeds with rapid weight gain and high feed conversionratios. The coefficients calculated for each type of livestock and for eachsystem were subsequently applied to high-resolution maps of livestock pop-ulation densities to predict the geographic distribution of antimicrobial con-sumption in food producing animals for the years 2010 and 2030.
Antimicrobials. Data on antimicrobial consumption in food animals wereobtained from government veterinary agencies, agriculture ministries, sci-entific reports and publications, and personal communications with academicresearchers; all data are included in Table S1. Data were collected by con-tacting relevant government ministries or agencies and through a systematicsearch of studies on PubMed. The search terms and a detailed description ofthe data collection process are given in Table S1. The majority of countrieseither do not collect or do not release data on veterinary antimicrobialconsumption. Across countries, data on antimicrobial consumption werefound in three formats: (i) estimates of overall antimicrobial consumption,(ii) estimates of consumption by livestock species, and (iii) estimates ofconsumption per PCU disaggregated by species type (e.g., chickens, cattle,and pigs). PCUs are used to compare population and production of differenttypes of livestock across countries and correspond to 1 kg of living orslaughtered animal (21). For example, a herd of 10 pigs each weighting 100 kgcorresponds to 1,000 PCUs. Assuming that antimicrobial consumption inchickens, cattle, and pigs represents the majority of antimicrobial consumption
in food-producing animals, the total consumption of antimicrobials was cal-culated for each country by pooling the estimates collected in case ii or bymultiplying the per PCU figure by the total national PCU for each type oflivestock in case iii. When data could not be obtained for the reference year2010, the antimicrobial estimates obtained for another year were adjustedusing the ratio of overall antimicrobial consumption between 2010 and thecorresponding year. Estimates of total antimicrobial consumption could beobtained for 32 countries, including 28 member states of the OECD and fourcandidate-OECD countries (Cyprus, Latvia, Lithuania, and Bulgaria).
Animal Census. To calculate estimates of antimicrobial consumption per PCUthat could be applied at the pixel level to generate total antimicrobialconsumption maps, we estimated national PCUs as a function of the numberof living animals. Thus, total PCUs in a country or a pixel for livestock type k inthe production system s were defined as follows:
PCUk,s = Ank,s ·�1+nk,s
�·�
Yk
RðCW=LWÞ,k
�
where An k is the number of living animals, nk,s is the number of productioncycles in each production system (extensive or intensive), Y is the quantityof meat per animal (carcass weight) obtained for each country fromFAOSTAT, and RðCW=LWÞ,k is the killing-out percentage (or dressing per-centage)—that is, the ratio of carcass weight to live weight of an animal—obtained from literature estimates (52). The last term of this equation canbe interpreted as the animal weight reconstructed from country-specificproductivity figures.
To reflect differences in productivity, distinct values were used for thenumber of production cycles in extensive (nc,Ext ) and intensive ( nc,Int ) pro-duction systems. Working under the assumption that extensive farmingrepresents the bulk of livestock production in low-income countries, nExt wasestimated from the median number of production cycles in the quartile ofcountries characterized by the lowest gross domestic product (GDP) percapita (World Bank estimates). This value was considered identical in allcountries on the basis that backyard productivity displays little variabilityacross low-income countries (the ratio of SD to the mean in the lower GDPper capita quartile was 0.65 for cattle, 0.44 for chickens, and 0.91 for pigs).The number of production cycles in intensive systems was calculated byimputation as nc,Int = ðS− nExt ·AnExtÞ=AnInt , where S is the total number ofanimals slaughtered in 2010.
Statistical Models. Antimicrobial consumption per PCU for each type oflivestock in extensive and intensive systems was estimated using Bayesiansregressions through a three-step procedure described in SI Text and Fig. S3.
Mapping Predictions. To generate spatially explicit estimates of antimicrobialuse, the estimates of consumption per PCU in each country weremultiplied bythe corresponding PCU values in each pixel. The uncertainty of the mappedprediction was quantified by weighting the respective SD for each livestockcoefficient by the relative PCU for each pixel. Maps of chickens and pigs,disaggregated by extensive and intensive production systems for the year2010, were obtained from Gilbert et al. (24). For cattle, maps of populationdensities disaggregated between extensive and intensive were generatedfrom the total population densities obtained from Robinson et al. (53). Weused a threshold of five cattle head per kilometer to allocate animals to amap of intensively raised cattle. The map of extensively raised cattle wasgenerated by subtracting the intensively raised from the total number ofhead in each pixel. Livestock densities in 2030 were estimated based on aprojection of meat consumption in 2030 (45). Assuming a constant value forthe compound annual growth rate from 2000 to 2030, this value was used tocalculate the ratio of meat consumption R30/10 between 2010 and 2030,
R30=10 = An2030
An2010=�An2030
An2000
�23
where An is the number of animals in each pixel. The projected livestockdensities for 2030 were estimated by multiplying animal densities per pixelin 2010 by R30/10. For pigs and chickens, we used the model developed byGilbert et al. (24) that quantifies the country-level proportion of extensivelyraised chicken and pig stock as a function of GDP per capita in purchasingpower parity. On the national level, the proportion of chickens or pigs raisedintensively showed a good correlation with this metric because the de-velopment of cost-efficient, large-scale farms typically requires substantialinvestments and influx of capital.
Van Boeckel et al. PNAS Early Edition | 5 of 6
AGRICU
LTURA
LSC
IENCE
S
The proportion of animals raised in extensive production systems wascalculated by replacing the current GDP values in the model by those of theyear 2030, obtained from linear projection of GDP estimates from the In-ternational Monetary Fund for the period 1980–2018. For cattle, in the ab-sence of reliable global estimates for shift in production structure, theproportion of animals raised extensively and intensively was assumed to beidentical for 2010 and 2030. All maps were resampled at 0.08333 decimaldegree resolution (∼10 kilometers at the equator).
ACKNOWLEDGMENTS. Data collection was supported by the Organization forEconomic Co-operation and Development (SRM 500026083). B.T.G. and T.P.V.B.were supported by the Science and Technology Directorate, Department ofHomeland Security, Contract HSHQDC-12-C-00058. B.T.G. and S.A.L. weresupported by the Research and Policy for Infectious Disease Dynamics programof the Science and Technology Directorate, Department of Homeland Security.B.T.G. was supported by the Bill & Melinda Gates Foundation and by the FogartyInternational Center, National Institutes of Health. S.A.L., A.T., and R.L. weresupported by Princeton University through its Grand Challenges Program.
1. Food and Drug Administration (2010) CVM Updates - CVM Reports on AntimicrobialsSold or Distributed for Food-Producing Animals (Food Drug Admin, Silver Spring. MD).Available at www.fda.gov/AnimalVeterinary/NewsEvents/CVMUpdates/ucm236143.htm.Accessed March 10, 2015.
2. Laxminarayan R, et al. (2013) Antibiotic resistance-the need for global solutions.Lancet Infect Dis 13(12):1057–1098.
3. Graham JP, Evans SL, Price LB, Silbergeld EK (2009) Fate of antimicrobial-resistantenterococci and staphylococci and resistance determinants in stored poultry litter.Environ Res 109(6):682–689.
4. Price LB, Johnson E, Vailes R, Silbergeld E (2005) Fluoroquinolone-resistant Cam-pylobacter isolates from conventional and antibiotic-free chicken products. EnvironHealth Perspect 113(5):557–560.
5. Smith TC, et al. (2013) Methicillin-resistant Staphylococcus aureus in pigs and farmworkers on conventional and antibiotic-free swine farms in the USA. PLoS ONE 8(5):e63704.
6. Vieira AR, et al. (2011) Association between antimicrobial resistance in Escherichia coliisolates from food animals and blood stream isolates from humans in Europe: Anecological study. Foodborne Pathog Dis 8(12):1295–1301.
7. Aarestrup FM (2005) Veterinary drug usage and antimicrobial resistance in bacteria ofanimal origin. Basic Clin Pharmacol Toxicol 96(4):271–281.
8. Schwarz S, Kehrenberg C, Walsh TR (2001) Use of antimicrobial agents in veterinarymedicine and food animal production. Int J Antimicrob Agents 17(6):431–437.
9. Chantziaras I, Boyen F, Callens B, Dewulf J (2014) Correlation between veterinaryantimicrobial use and antimicrobial resistance in food-producing animals: A report onseven countries. J Antimicrob Chemother 69(3):827–834.
10. You Y, Silbergeld EK (2014) Learning from agriculture: Understanding low-dose an-timicrobials as drivers of resistome expansion. Front Microbiol 5:284.
11. Tilman D, Balzer C, Hill J, Befort BL (2011) Global food demand and the sustainableintensification of agriculture. Proc Natl Acad Sci USA 108(50):20260–20264.
12. FAOSTAT. Available at faostat.fao.org. Accessed March 10, 2015.13. Guo X, Mroz TA, Popkin BM, Zhai F (2000) Structural change in the impact of income
on food consumption in China, 1989–1993. Econ Dev Cult Change 48(4):737–760.14. Silva NCC, et al. (2013) Molecular characterization and clonal diversity of methicillin-
susceptible Staphylococcus aureus in milk of cows with mastitis in Brazil. J Dairy Sci96(11):6856–6862.
15. Zhu Y-G, et al. (2013) Diverse and abundant antibiotic resistance genes in Chineseswine farms. Proc Natl Acad Sci USA 110(9):3435–3440.
16. Maron DF, Smith TJ, Nachman KE (2013) Restrictions on antimicrobial use in foodanimal production: An international regulatory and economic survey. Global Health9:48.
17. Goossens H, Ferech M, Vander Stichele R, Elseviers M; ESAC Project Group (2005)Outpatient antibiotic use in Europe and association with resistance: A cross-nationaldatabase study. Lancet 365(9459):579–587.
18. Van Boeckel TP, et al. (2014) Global antibiotic consumption 2000 to 2010: An analysisof national pharmaceutical sales data. Lancet Infect Dis 14(8):742–750.
19. Aarestrup F (2012) Sustainable farming: Get pigs off antibiotics. Nature 486(7404):465–466.
20. World Bank. Available at data.worldbank.org. Accessed March 10, 2015.21. European Medicines Agency (2013) Sales of Veterinary Antimicrobial Agents in 25 EU/
EEA Countries in 2011: Third ESVAC Report (Eur Med Agency, London). Available atwww.ema.europa.eu/ema/index.jsp?curl=pages/regulation/document_listing/document_listing_000302.jsp. Accessed March 10, 2015.
22. Department of Health (2013) UK Five Year Antimicrobial Resistance Strategy 2013 to2018 (Dep Health, London). Available at https://www.gov.uk/government/publications/uk-5-year-antimicrobial-resistance-strategy-2013-to-2018. Accessed March 10, 2015.
23. Van Boeckel TP, Thanapongtharm W, Robinson T, D’Aietti L, Gilbert M (2012) Pre-dicting the distribution of intensive poultry farming in Thailand. Agric Ecosyst Environ149:144–153.
24. Gilbert M, et al. (2015) Mapping the global distribution of intensively farmed chickenand pigs. PLoS ONE, in press.
25. Sarmah AK, Meyer MT, Boxall ABA (2006) A global perspective on the use, sales,exposure pathways, occurrence, fate and effects of veterinary antibiotics (VAs) in theenvironment. Chemosphere 65(5):725–759.
26. Laxminarayan R (2007) Extending the Cure: Policy Responses to the Growing Threat ofAntibiotic Resistance. (Resour Future, Washington, DC).
27. Van Boeckel TP, et al. (2012) Improving risk models for avian influenza: The role ofintensive poultry farming and flooded land during the 2004 Thailand epidemic. PLoSONE 7(11):e49528.
28. Chung WB, Lin MW, Chang WF, Hsu M, Yang PC (1997) Persistence of porcine re-productive and respiratory syndrome virus in intensive farrow-to-finish pig herds. CanJ Vet Res 61(4):292–298.
29. Gerber P, Chilonda P, Franceschini G, Menzi H (2005) Geographical determinants andenvironmental implications of livestock production intensification in Asia. BioresourTechnol 96(2):263–276.
30. Tscharntke T, Klein AM, Kruess A, Steffan-Dewenter I, Thies C (2005) Landscape per-spectives on agricultural intensification and biodiversity—Ecosystem service manage-ment. Ecol Lett 8:857–874.
31. Sami AS, Augustini C, Schwarz FJ (2004) Effects of feeding intensity and time on feedon performance, carcass characteristics and meat quality of Simmental bulls. Meat Sci67(2):195–201.
32. Hay SI, Guerra CA, Tatem AJ, Noor AM, Snow RW (2004) The global distribution andpopulation at risk of malaria: Past, present, and future. Lancet Infect Dis 4(6):327–336.
33. Dye C, Scheele S, Dolin P, Pathania V, Raviglione MC (1999) Consensus statement.Global burden of tuberculosis: estimated incidence, prevalence, and mortality bycountry. WHO Global Surveillance and Monitoring Project. JAMA 282(7):677–686.
34. Wise R (2002) Antimicrobial resistance: Priorities for action. J Antimicrob Chemother49(4):585–586.
35. Gething PW, et al. (2014) Declining malaria in Africa: Improving the measurement ofprogress. Malar J 13:39.
36. El-Lethey H, Huber-Eicher B, Jungi TW (2003) Exploration of stress-induced immu-nosuppression in chickens reveals both stress-resistant and stress-susceptible antigenresponses. Vet Immunol Immunopathol 95(3-4):91–101.
37. Silbergeld EK, Graham J, Price LB (2008) Industrial food animal production, antimi-crobial resistance, and human health. Annu Rev Public Health 29:151–169.
38. Walsh TR, Weeks J, Livermore DM, Toleman MA (2011) Dissemination of NDM-1positive bacteria in the New Delhi environment and its implications for human health:An environmental point prevalence study. Lancet Infect Dis 11(5):355–362.
39. Ganguly NK, et al.; Global Antibiotic Resistance Partnership (GARP) - India WorkingGroup (2011) Rationalizing antibiotic use to limit antibiotic resistance in India. IndianJ Med Res 134:281–294.
40. Kakkar M, Rogawski L (2013) Antibiotic Use and Residues in Chicken Meat and MilkSamples from Karnataka and Punjab, India: Research Scheme 34 (Public Health Found,New Delhi).
41. Basnyat B (2014) Antibiotic resistance needs global solutions. Lancet Infect Dis 14(7):549–550.
42. Cabello FC (2006) Heavy use of prophylactic antibiotics in aquaculture: A growingproblem for human and animal health and for the environment. Environ Microbiol8(7):1137–1144.
43. Le TX, Munekage Y, Kato S (2005) Antibiotic resistance in bacteria from shrimpfarming in mangrove areas. Sci Total Environ 349(1-3):95–105.
44. The World Bank (2013) Fish to 2030: Prospects for Fisheries and Aaquaculture (WorldBank, Washington, DC). Available at documents.worldbank.org/curated/en/2013/12/18882045/fish-2030-prospects-fisheries-aquaculture. Accessed March 10, 2015.
45. Robinson TP, Pozzi F (2011)Mapping Supply and Demand for Animal-Source Foods to2030, Animal Production Health Working Paper (Food Agric Org, Rome), No 164.
46. Cogliani C, Goossens H, Greko C (2011) Restricting antimicrobial use in food animals:Lessons from Europe. Microbe 6(6):274−279.
47. Woolhouse M, Farrar J (2014) Policy: An intergovernmental panel on antimicrobialresistance. Nature 509(7502):555–557.
48. Allen HK, Levine UY, Looft T, Bandrick M, Casey TA (2013) Treatment, promotion,commotion: Antibiotic alternatives in food-producing animals. Trends Microbiol21(3):114–119.
49. Diarra MS, Malouin F (2014) Antibiotics in Canadian poultry productions and antici-pated alternatives. Front Microbiol 5:282.
50. Graham JP, Boland JJ, Silbergeld E (2007) Growth promoting antibiotics in food an-imal production: An economic analysis. Public Health Rep 122(1):79–87.
51. Teillant A, Laxminarayan R (2015) Economics of antibiotic use in U.S. swine andpoultry production. Choices 30(1):1–11.
52. Warriss PD (2010) Meat Science: An Introductory Text. (CABI, Wallingford, UK),2nd Ed.
53. Robinson TP, et al. (2014) Mapping the global distribution of livestock. PLoS ONE 9(5):e96084.
6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1503141112 Van Boeckel et al.
Supporting InformationVan Boeckel et al. 10.1073/pnas.1503141112SI Text
Protocol S1: Three-Step Procedure to Calculate Coefficientsof Antimicrobial Consumption per Population CorrectionUnitsStep 1. OECD countries with published estimates of overall an-timicrobial consumption were used to train a linear regressionmodel to predict the overall consumption of antimicrobials inmilligrams in other OECD countries (AmOECD) as a function ofthe log-transformed stock of animals (Fig. S3). The model wasweighted by the total size of the livestock sector (PCUs) in eachcountry. To avoid overestimating antimicrobial consumption, weexcluded the United States from the training set, since it is knownto have uncharacteristically high consumption of antimicrobialscompared with other OECD countries.
Step 2. A Bayesian linear regression model was fitted to the totalconsumption of antimicrobials AmOECD, to estimate con-sumption per PCU for each type of livestock in intensive pro-duction systems in 37 countries (all OECD countries, as well asfour candidate-OECD countries, except the United States). Theprior distributions used for the regression coefficients were as-
sumed to be Gaussian, with means and SDs corresponding to theanimal-specific antimicrobial consumption per PCU that could beobtained for 10 countries (Australia, Belgium, Denmark, France,Japan, Korea, The Netherlands, New Zealand, Sweden, and theUnited Kingdom). The relative importance of cattle, chickens, orpigs in explaining the variability in total antimicrobial con-sumption was quantified using the difference in the Bayes factorvalue between a full model including all animal types and a modelexcluding the animal type of interest.
Step 3. For each country, a value for antimicrobial consumptionper PCU in intensive production systems was sampled from theestimated posteriors distribution. The corresponding coefficientsfor extensive systems were calculated using a randomly sampledscaling ratio ranging from 0.05 to 0.5 times the value sampled forintensive systems. The procedure was repeated 50 times for eachcountry to compute an SD and average coefficient value for eachtype of livestock. An overall confidence interval for global an-timicrobial consumption in food animals was calculated bycombining the SD of the antimicrobial consumption predicted ineach country in relation to the share of each country in the globalantimicrobial consumption.
Antimicrobial (Kg)
No anim
al
0 - 1010
- 30
30- 25
0
> 250
A B C
Fig. S1. Antimicrobial consumption in livestock in the European Union in 2010 for cattle (A), chickens (B), and pigs (C).
Fig. S2. Distribution of farm sizes in Thailand. The bimodal distribution shows that farms are distributed in two distinct groups corresponding to differentproduction systems: A large number of farms have a median size of 10 birds, whereas another large group of farms host between 5,000 and 10,000 birds.
Van Boeckel et al. www.pnas.org/cgi/content/short/1503141112 1 of 5
Fig. S3. Association between national antimicrobial consumption obtained from national reports (Table S1) and population correction units (kilograms ofmeat) in OECD countries in 2010 for cattle, chicken, and pigs. Gray dots represent the total production, and red dots represent intensive production alone.
Van Boeckel et al. www.pnas.org/cgi/content/short/1503141112 2 of 5
Table
S1.
Datausedto
inform
themodel
presentedin
this
study
Country
Cattle,
mg/PCU
Chicke
ns,
mg/PCU
Pigs,
mg/PCU
Annual
sales
cattle,kg
Annual
sales
chicke
ns,kg
Annual
sales
pigs,kg
Totalan
nual
sales,kg
Yea
rDatasource
URL
Australia
––
–13
3,30
0*40
6,40
010
4,20
064
4,00
020
09–20
10APV
MA
arch
ive.ap
vma.gov.au
/publications/reports/docs/
antimicrobial_sales_report_m
arch
-201
4.pdf
Austria
––
––
–63
,000
2010
EMA
www.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
––
26.1
†–
––
2010
Trau
ffleret
al.(1)
veterinaryrecord.bmj.com/content/ea
rly/20
14/07/
22/vr.10
2520
.full.pdf+
htm
lBelgium
––
–29
9,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
––
–24
,800
26,079
156,86
920
09–20
10BelVet-SAC‡
www.belve
tsac.ugen
t.be/pag
es/home/Belve
tSAC_
report_2
013%
20finaa
l.pdf
Bulgaria
––
––
––
46,200
§20
11EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Cyp
rus
––
––
––
57,200
§20
11EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Can
ada
––
––
––
1,45
0,77
6{20
12Corresponden
cewithJean
Szko
tnicki
(CAHI)
–
Cze
chRep
ublic
––
––
––
71,000
2010
EMA
www.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Den
mark
––
–11
9,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
––
–14
,636
477#
100,52
711
5,01
020
10DANMAP
www.dan
map
.org/∼/m
edia/Projekt%20
sites/
Dan
map
/DANMAP%
20reports/Dan
map
_20
10.ashx
Estonia
––
––
––
7,60
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Finland
––
––
–13
,000
2010
EMA
www.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
38.4
k20
10Correspondan
cewithDr.Han
nele
Nau
holz
(ETT
)
–
Fran
ce–
––
997,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
19.11
107.53
147.37
––
–20
10ANSE
S**
https://w
ww.anses.fr/sites/default/file
s/docu
men
ts/ANMV-Ra-Antibiotiques20
10.pdf
German
y–
––
––
–2,00
2,00
0§20
11EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Hungary
––
––
––
206,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Icelan
d–
––
––
–90
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Irelan
d–
––
––
–96
,000
2010
EMA
www.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Italy
––
––
––
1,92
8,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Van Boeckel et al. www.pnas.org/cgi/content/short/1503141112 3 of 5
Table
S1.
Cont.
Country
Cattle,
mg/PCU
Chicke
ns,
mg/PCU
Pigs,
mg/PCU
Annual
sales
cattle,kg
Annual
sales
chicke
ns,kg
Annual
sales
pigs,kg
Totalan
nual
sales,kg
Yea
rDatasource
URL
Latvia
––
––
––
6,60
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Lithuan
ia–
––
––
–16
,000
2010
EMA
www.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Japan
––
–59
,490
89,782
506,54
965
5,82
0.50
2010
MAFF
††
www.m
aff.go.jp
/nva
l/iya
kutou/han
baidak
a/pdf/h22
han
baidak
abessatu.pdf
Korea
––
–57
,443
204,47
258
1,50
784
3,42
220
10QIA
‡‡
lib.qia.go.kr/Se
arch
/Detail/9
0953
?key
=%
EC%
B6%
95%
EC%
82%
B0%
20%
ED%
95%
AD%
EC%
83%
9D%
EC%
A0%
9C%
20%
EB%
82%
B4%
EC%
84%
B1%
EA%
B7%
A0%
20%
EA%
B0%
90%
EC%
8B%
9C%
EC%
B2%
B4%
EA%
B3%
84%
20%
EA%
B5%
AC%
EC%
B6%
95Th
eNetherlands
––
–46
1,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
––
–19
3,92
042
,740
174,70
041
1,37
020
10Bondtet
al.(2)§§
www.scien
cedirect.co
m/scien
ce/article/
pii/S0
1675
8771
2002
346
New
Zealan
d–
––
8,00
024
,000
24,000
57,043
2010
–20
11MPI
{{www.foodsafety.govt.nz/industry/acvm/docu
men
ts/
reports.htm
Norw
ay–
––
––
–6,30
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Poland
––
––
––
479,00
020
11EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Portugal
––
––
––
181,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Slova
kia
––
––
––
–20
11EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Slove
nia
––
––
––
8,40
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Spain
––
––
––
1,74
6,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
Swed
en–
––
13,000
2010
EMA
www.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
––
14.5
##
–3,78
020
09SV
Awww.sva
.se/upload
/Red
esign20
11/Pd
f/Om_
SVA/publik
ationer/Swed
res_Svarm20
13.pdf
––
–44
.5kk
14,117
2010
SVA
www.sva
.se/upload
/Red
esign20
11/Pd
f/Om_
SVA/publik
ationer/1/Sva
rm20
10.pdf
Switze
rlan
d–
––
––
–58
,121
***
2010
SwissM
edic
https://w
ww.swissm
edic.ch/m
arktueb
erwachung/
0013
5/00
136/00
181/index
.htm
l?lang=en
United
Kingdom
456,00
020
10EM
Awww.ema.eu
ropa.eu
/docs/en_G
B/docu
men
t_lib
rary/Rep
ort/201
3/10
/WC50
0152
311.pdf
620
420
411
,000
302,00
029
9,00
039
0,00
020
10VMD†††
https://w
ww.vmd.defra.gov.uk/pdf/VARSS.pdf
United
States
––
––
–13
,542
,030
2011
FDA‡‡‡
www.fda.gov/download
s/Fo
rIndustry/UserFee
s/Anim
alDrugUserFee
ActADUFA
/UCM33
8170
––
––
–1,27
0,00
020
06–20
09Apleyet
al.(3)§§§
www.nppc.org/w
p-content/upload
s/Sw
ine-in-fee
d-use-estim
ates.pdf
Van Boeckel et al. www.pnas.org/cgi/content/short/1503141112 4 of 5
Antibioticsalesforuse
inother
poultry
birdsareincluded
inmost
ofthech
icke
n-specificdata(i.e.,“Chicke
ns,mg/PCU”or“Annual
salesch
icke
ns”)presentedin
thistable.A
lthoughthismay
potentially
resultin
ove
restim
ationofan
tibioticuse
inch
icke
nsonly,w
ebelieve
thisap
proximationisreasonab
leco
nsideringch
icke
nproductiongen
erally
represents
themajority
ofpoultry
productionin
most
countries.Chicke
n-specific
salesdatawereobtained
forBelgium,D
enmark,
Finland,a
ndSw
eden
only.T
hedataco
llectionsearch
procedure
invo
lved
both
contactingreleva
ntgove
rnmen
tministriesan
dag
encies
regardingav
ailable
datasources
andasystem
atic
search
ofex
istingpublished
studiesonPu
bMed
.In
theinitialstag
esofoursearch
forspecies-specific
sales/co
nsumptiondata,
weco
ntacted
theag
encies
listedas
datasources
forthe25
Europea
nco
untriespresentedin
theEu
ropea
nSu
rveilla
nce
ofVeterinaryAntimicrobialConsumptionreport
(4)as
wellas
allve
terinarymed
icines
authorities
listedbytheHea
dsofMed
icines
Agen
cies
web
site
(forfurther
inform
ation,seewww.hma.eu
/vdirectory).Allnational
reportslistedin
Annex
6oftheES
VACreport
wereaccessed
forspecies-specific
data(4).Fo
rtheremainingOEC
Dco
untriesnotlistedin
theES
VACreport,
asystem
atic
search
forthereleva
ntgove
rnmen
tministryorag
ency
was
perform
edonGoogle
usingthefollo
wingterm
sin
conjunctionwithaco
untrynam
e:“(M
inistryORAgen
cy)AND
(Agricu
lture
ORHea
lth),”
“VeterinaryMed
icines
(Directorate
OR
Authority),”
“Med
icines
Directorate,”
“Drug
Controlle
r,”“Fo
od
and
Drug
Administration,”
“VeterinaryAssociation,”
“Fo
od
Safety,”
and
“(A
ntimicrobialOR
Antibiotic)
(ConsumptionORSa
les)Su
rveilla
nce.”
Wethen
contacted
ministries/ag
encies
from
search
resultsan
dwould
either
receivedatadirectlyfrom
theag
ency/m
inistryco
ntacted
,beinform
edthat
dataarenotav
ailable,
orbedirectedto
another
ministry/ag
ency.A
limitationofthissearch
procedure
was
theuse
ofEn
glishsearch
term
swhen
man
yco
untries’ministries/ag
encies
listtheirnam
esin
nativelanguag
es.H
owev
er,thereleva
nt
agen
cies
forallOEC
Dco
untrieswereco
ntacted
regardingtheav
ailability
ofdata.
Thesearch
forex
istingpublished
studiesonPu
bMed
usedthefollo
wingterm
sin
conjunctionwithaco
untrynam
e:“Veterinary
(AntibioticORAntimicrobial)(ConsumptionORSa
les)”an
d“(A
ntibioticORAntimicrobial)(ConsumptionORSa
les)AND(FoodAnim
alsORLive
stock).”
ANSE
S,Agen
ceNational
deSé
curité
Sanitaire
(France);APV
MA,
Australia
nPe
sticides
andVeterinaryMed
icines
Authority
(Australia
);BelVet-SAC,B
elgianVeterinarySu
rveilla
nce
ofAntibacterial
Consumption(Belgium);CAHI,Can
adianAnim
alHea
lthInstitute
(Can
ada);D
ANMAP,
Dan
ishProgram
forsurveilla
nce
ofan
timicrobialco
nsumptionan
dresistan
cein
bacteriafrom
anim
als,
foodan
dhuman
s(D
enmark);EM
A(orES
VAC),Eu
ropea
nMed
icines
Agen
cy(orEu
ropea
nSu
rveilla
nce
of
VeterinaryAntimicrobialC
onsumption)(Eu
rope);E
TT,A
ssociationforAnim
alDisea
sePrev
ention(Finland);FD
A,U
SFo
odan
dDrugAdministration(U
nited
States
ofAmerica);M
AFF,M
inistryofAgricu
lture,Forestry
and
Fisheries(Jap
an);MARAN,M
onitoringofAntimicrobialR
esistance
andAntibioticUse
inAnim
alsin
theNetherlands(N
etherlands);M
PI,M
inistryforPrim
aryIndustries
(New
Zealan
d);QIA,A
nim
alan
dPlan
tQuaran
tine
Agen
cy(Korea);SVA(orSV
ARM),National
VeterinaryInstitute
(orSw
edishVeterinaryAntimicrobialR
esistance
Monitoring)(Sw
eden
);Sw
issM
edic,SwissAgen
cyforTh
erap
euticProducts(Switze
rlan
d);VMD,V
eterinary
Med
icines
Directorate
(United
Kingdom).
*Sales
dataforcattle
andshee
pwerepooledtogether––therefore,thisfigure
likelyrepresents
anove
restim
ationofuse
incattle.T
otala
nnual
salesdataincludes
salesforfood-producingan
imalsonly
(sales
for
use
inco
mpan
ionan
imalswereex
cluded
).†Use
dataweregiven
inmilligramsan
tibioticper
biomass(kilo
grams)
produced
per
year.In
thestudyfrom
which
this
datapointwas
extracted,electronic
drug
application
reco
rdsfrom
farm
ersfrom
75co
nve
ntional
pig
farm
swererevisedan
dch
ecke
dfortheirplausibility.Th
eregistereddrugam
ounts
wereve
rified
byco
mparingthefarm
ers’reco
rdswithve
terinarians’dispen
sary
reco
rds.
‡Sp
ecies-specific
annual
salesdatawereex
trap
olatedusingamodel
based
ondatafrom
Persoonset
al.(5),Callenset
al.(6),an
dPa
rdonet
al.(7).
§To
talan
nual
sales(inkilogramsactive
ingredient)
for20
10was
back-calculatedfrom
2011
ESVAC(Europea
nMed
icines
Agen
cy)databecau
se20
10datawerenotav
ailable.Th
iscalculationwas
perform
edby
dividingthe20
11ES
VACdatabythemed
ianoftheratioof20
11ES
VACdatato
2010
ESVACdataforallco
untrieswithdataav
ailable
for20
11[i.e.,1/med
ian(ESV
AC20
11/ESV
AC20
10)].
{ Totalan
nual
salesdataincludes
salesforfood-producingan
imalsonly
(sales
foruse
inco
mpan
ionan
imalswereex
cluded
).#Annual
salesdataforch
icke
nsincludes
only
broile
rs,breed
ingan
drearingbroile
rs,an
dlaye
rs––thisfigure
does
notincludean
yother
poultry
birds.
k Annual
salesdataforch
icke
nsincludes
only
broile
rs––thisfigure
does
notincludean
yother
poultry
birds.
**Datagiven
inmilligramsofan
timicrobialper
kilogram
liveweightoffoodan
imal
trea
tedan
dtherefore
iseq
uivalen
tto
milligramsper
PCU.
††Data
weredisag
gregated
into
thean
nual
salesvo
lume-specific
antibiotics.Th
erefore,totalsalesvo
lumeforallan
tibiotics
was
calculated
bysumming
across
allan
tibiotics.Dataon
antihelmintics
and
antiprotozo
alswas
also
colle
cted
butnotincluded
inthesefigures.
‡‡Dataweretran
slated
byChan
g-YongChoioftheResea
rchInstitute
forAgricu
lture
andLife
Science
atSe
oulNational
University.
§§Sp
ecies-specifican
nual
salesdatawerederived
from
use
dataco
llected
inaMARANsample
survey
.Totala
nnual
salesrefers
only
tocattle,p
oultry,a
ndpigs.Plea
senote
that
thedataforcattle
isthesum
ofuse
for“Cattle”
(29,93
0kg
)an
d“Vea
lCalve
san
dYoungBee
f”(163
,990
kg).
{{MPI
presentedsalesdataforpoultry
andpigspooledtogether.Th
esedatapoints
weredisag
gregated
into
species-specificco
nstan
tsbyassumingaco
nstan
tco
nsumptionper
PCU
togen
eratespecies-specific
coefficien
tsusedin
thepriors.
##Dataarepresentedas
milligramsper
kilogram
slau
ghteredpig.T
oco
nve
rtunitsto
mg/PCUliv
ean
imal,thisdatapointwas
multiplie
dby1/0.7where0.7represents
theratioofliv
eweightto
carcassweightfor
pigs.In
other
words,1kg
ofliv
epig
was
raised
forev
ery0.7kg
ofpig
mea
tproduced.
kkAnnual
salesdataforch
icke
nsincludes
only
broile
rs––this
figure
does
notincludean
yother
poultry
birds.“In
2010
,thetotalsaleswhereGallusgalluswas
given
asspecieswas
44.5
kgofwhich78
%were
aminopen
icillins”
(SVARM
2010
).**
*Totalan
nual
salesdataincludesalesforfood-producingan
imalsonly
(sales
foruse
inco
mpan
ionan
imalswereex
cluded
).†††VMD
presentedmilligramsper
PCU
dataforpigsan
dpoultry
pooledtogether.Annual
salesdatabyspecieswas
presentedin
thefollo
wingcategories:“CattleOnly
Products,”“PigOnly
Products,”“Po
ultry
Only
Products,”“Pigan
dPo
ultry
Combined
Products,”an
d“MultiS
peciesProductsin
FoodAnim
alsOnly.”
“Po
ultry
Only”an
d“Pigan
dPo
ultry
Combined
”product
salesweresummed
toyieldoursalesestimate
forpoultry.W
eperform
edthean
alogousprocedure
toderiveoursalesestimateforpigs.W
eex
cluded
salesof“MultiS
peciesProductsin
FoodAnim
als”
inan
attemptto
preve
ntove
restim
ationofan
tibioticuse.
Species-specific
consumptionwas
derived
from
thisdatabyassumingtheco
efficien
tsforco
nsumptionper
PCU
wereiden
tical.
‡‡‡To
talan
nual
salesdataincludes
salesforfood-producingan
imalsonly
(sales
foruse
inco
mpan
ionan
imalswereex
cluded
).Th
eFD
Apresented20
11salesdatafordomesticco
nsumptiondisag
gregated
by
antimicrobialclass.Th
erefore,ourestimateisthesum
ofsalesofallan
timicrobialclassesfordomesticco
nsumptiononly.
§§§Apleyet
al.(3)
presenteddataas
“National
Estimates
ofTo
talK
ilogramsofSw
ineIn-Fee
dAntimicrobialsforAllProductionCyclesin
aYea
rbyAntimicrobialan
dRea
son.”
Therefore,ourestimateisthesum
across
allan
timicrobials
andreasons.
1.Trau
fflerM,Griesbacher
A,Fu
chsK,KöferJ(201
4)Antimicrobialdruguse
inAustrian
pig
farm
s:Plau
sibility
checkofelectronic
on-farm
reco
rdsan
destimationofco
nsumption.VeterinaryRecord
175(16
):40
2.2.
BondtN,JensenVF,
Puister-JansenLF
,va
nGeijlswijk
IM(201
3)Comparingan
timicrobialex
posure
based
onsalesdata.
Prev
Vet
Med
108(1):10–
20.
3.ApleyMD,Bush
EJ,MorrisonRB,Singer
RS,
SnelsonH
(201
2)Use
estimates
ofin-fee
dan
timicrobialsin
swineproductionin
theUnited
States.Fo
odbornePa
thogDis9(3):272
–27
9.4.
Europea
nMed
icines
Agen
cy(201
3)Sa
lesofve
terinaryan
timicrobialag
ents
in25
EU/EEA
countriesin
2011
-Th
irdES
VACRep
ort.Ava
ilable
at:www.ema.eu
ropa.eu
/ema/index
.jsp?curl=pag
es/reg
ulation/docu
men
t_listing/docu
men
t_listing_
0003
02.jsp.AccessedMarch
10,20
15.
5.Pe
rsoonsD,et
al.(201
2)Antimicrobialuse
inBelgianbroile
rproduction.Prev
Vet
Med
105(4):320
–32
5.6.
CallensB,et
al.(201
2)Prophylactican
dmetap
hylactican
timicrobialuse
inBelgianfatten
ingpig
herds.Prev
Vet
Med
106(1):53–
62.
7.Pa
rdonB,et
al.(201
2)Prospective
studyonquan
titative
andqualitativean
timicrobialan
dan
ti-inflam
matory
druguse
inwhiteve
alcalves.JAntimicrobChem
other
67(4):10
27–10
38.
Van Boeckel et al. www.pnas.org/cgi/content/short/1503141112 5 of 5
Recommended