1. Peter Davies, Carie Alexander, Jeein Chung, Sylvia Wanzala College of Veterinary Medicine University of Minnesota Critical appraisal of evidence that low- dose, long term (growth promotion) antimicrobial use augments public health risks from antimicrobial resistant organisms

2. Background and objectives Pressures to restrict antimicrobial use in food animals Particularly low-dose group treatments ‘low hanging fruit’ European changes Science vs. precautionary principle Danish data indicating that AGP use was replaced by increased therapeutic use

3. Aggregate antimicrobial use in Danish swine industry

4. Initial scope Not a systematic review Evidence cited in 4 key documents arguing for removal of AGP use in the USA PAMTA FDA guidance 209 Pew Commission report on Industrial Farm Animal Production Keep Antibiotics Working annotated bibliography Assumption of ‘diligence’

5. Methods Compile reference list for each document FDA-209; PAMTA; PEW; KAW Screen references cited to identify relevant papers Only articles clearly unrelated to AMU or AMR eliminated Obtain original papers and compile in RefWorks database Assess relevance of retained papers Use specific appraisal tools to assess relevant papers

6. Tools for detailed appraisal Design of tools based on Steinberg et al. (2000) and Shamliyan (2010) Check list approach Questions refined iteratively by research team Original plan to use a single too 6 screening questions for relevance Two tools developed Descriptive or analytical papers Reviews or reports

7. Study population 7. Are inclusion/exclusion criteria for selecting subjects described? 8. Is the study population animal or human? 9. If animal, are they swine or other animals? 10. Is the age of the subjects described? 11. Is the housing system described? 12. Is the nutrition of the animals described? 13. Was the study done in an experimental or commercial/clinical setting? Treatments 14. Were multiple groups (treatments) included? 15. Was an untreated control group included? 16. Was the study replicated? Appraisal tool for descriptive studies

8. Antimicrobial Use 17. Was antimicrobial use reported? 18. Was the antimicrobial specified? 19. Was antimicrobial use recorded, or obtained by survey questionnaire? 20. Was the antimicrobial dose reported? 21. Was the antimicrobial route recorded? 22. Was the antimicrobial duration recorded? 23. Were antimicrobial blood (or tissue/urine/other) levels measured? Antimicrobial Resistance 24. Was antimicrobial resistance reported? 25. Was the method of resistance testing described or referenced? 26. Was the resistance description categorical, or by MIC? 27. Was the origin of MIC breakpoints specified? Appraisal tool for descriptive studies

9. Statistical Analysis 28. Was it conducted? 29. Were specific statistical tests indicated? 30. Were multivariate methods used? 31. Were confidence intervals included? Results 32. Does the analysis directly compare antimicrobial resistance in relation to use? 33. Is there comparison of antimicrobial dose and resistance?

10. Appraisal tool for descriptive studies Discussion 34. Were the limitations of the study discussed? 35. Was the relationship of the study population to other populations discussed? 36. Is the use of antimicrobials in food animals in relation to resistance discussed? 37. Is antimicrobial use in animals implicated in the development of resistance? 38. If so, is this with direct evidence or by inference?

11. Appraisal process – review papers Appraisal of review studies included documentation of the number of papers cited number of citations that were original studies Examined abstracts of all citations Book chapters deemed not to be original studies and were not examined further Some references cited in reviews could not be located and were recorded as “cannot find”.

12. Detailed appraisal Selected studies Specifically addressed AMU in food animals in relation AMR Implicated AMU in animals in AMR based on direct evidence Panel of 3 veterinarians Critically read all papers Collectively assess the design, results and conclusions

13. Results Screening - 154 papers likely to provide some information on relationship between AMU in food animals and AMR Analytical (n = 37) Descriptive (n = 48) Reviews (n = 69)

14. Results: screening questions for analytical studies (n = 37)

15. Results of analytical studies (n = 21) Study subjects 5 human subjects only 10 had animal subjects (5 in swine) 6 both human and animal subjects. 9 studies stated criteria for selecting subjects. Setting 19 in clinical/commercial settings 2 in experimental setting. 14 provided data on AMU

16. Results of analytical studies (n = 21) 19 reported measurement of AMR 14 reported MIC method (11 with breakpoints) 4 zone diffusion methods 1 genotyping method Statistical methods 4 with no analysis or method not specified 8 studies only univariate analysis (12 used multivariate approaches) 11 studies without confidence intervals

17. Analytical studies reporting details of AMU (n = 14)

18. Results of analytical studies(n = 21) 12 studies for detailed review 7 of 9 studies that did not compare AMU and AMR implicated use in resistance

19. Quality indicators of studies (n = 12)

20. Summary 48 descriptive studies were collectively uninformative regarding association with AMR of low-dose/long term use of antimicrobials in food animals 1 (0.6%) of 154 relevant cited papers) directly compared a low and therapeutic doses of AM to food animals (poultry) on the prevalence of AMR Measurement and recording of AMU suboptimal in most studies

21. Appraisal of review papers 37 reviews specifically on AMU or AMR All narrative reviews (29) or reports (8) no systematic review One review included search methods and one (different) review specified inclusion and exclusion criteria for studies Only 3 discussed validity in analyzing studies or drawing inferences None discussed potential limitations.

22. Specific organism-antimicrobial pairs cited in 25 review papers

23. Appraisal of review papers 37 sources cited 1,869 publications 1,012 (54.2%) were determined to be original studies providing original data Generally reiterated the most storied examples linking AMU in animals and AMR. Most examples related to antimicrobials used only therapeutically in the USA antimicrobials never used in the USA (avoparcin, nourseothricin).

24. Appraisal of review papers Individually and collectively do not identify studies contrasting impacts of low-dose, long duration AMU relative to other AMU provide evidence of any differential effects among specific modes of AMU in selecting for AMR. no evidence that measures to reduce low-dose, long term AMU that do not reduce aggregate antimicrobial use would provide benefits upon: prevalence of AMR organisms in food animals public health

25. Overall summary Enduring uncertainty over the public health consequences of AMU in animals has frustrated groups who have pursued greater regulation Vast majority of papers cited by key sources contain no primary data Shortcomings in design or reporting common

26. Overall summary Dearth of RCT contributes to difficulty in drawing inferences about this question Available evidence inadequate for meaningful comparison of different modes of AMU in relation to emergence of AMR in pathogens or commensals in commercial swine populations The conservative position is that reducing aggregate use of antimicrobials is the most appropriate goal

27. Removal of Growth Promoters in Denmark

28. Expectations for the Restrictions Reduction of antibiotic use in food animals Minimal impact on production Reduction of antibiotic resistance Animal isolates Human isolates Reduced risk to human health Zoonotic and foodborne pathogens Commensal organisms and animal pathogens

29. Response of Danish producers (2004)  Increased weaning age by more than 3 days  Reduced ration density (protein/energy)  Practice limit feeding where possible  Added ZnO and organic acids to diets  Utilized more therapeutic antimicrobials

30. Aggregate antimicrobial use in Danish swine industry

31. Effects on weaned pig performance

32. DANMAP 2004 Antimicrobial resistance in Salmonella

34. AMR in Salmonella in pigs Danmap 2004/2012

35. Non-susceptibility of in S. pneumoniae isolates from humans in Denmark

36. Human MRSA cases in Denmark (DANMAP 2012)

37. Summary of Danish experience Anticipated decrease in aggregate AMU in food animals not realized Measurable animal health impact Reduction in AMR of zoonotic and commensal isolates from pigs variable AMR in major human pathogens increased Silver lining of public health benefit hard to perceive

38. Can we extrapolate this to the USA Differences in regulations Virtual elimination of vet dispensing (1995) Mandatory herd health consultations per year 12 farm visits per year 35 day limit in prescriptions Changes preceded AGP bans by several years

39. To the future Growth promotant usage being phased out in USA Are we prepared? Earlier weaned pigs More disease challenge (?) – PRRS Less veterinary oversight