1. Estimating the best way forward: Expert and farmer evaluations of environmental interventions to reduce human exposure to E. coli O157 Paul Cross, Dan Rigby & Gareth Edwards-Jones  RELU funded ‘Reducing Escherichia coli O157 risk in rural communities’

2. Uncertainty & O157 Management Uncertainty regarding: How people become infected (sporadic infection) Effectiveness of measures Likelihood of measures being adopted

3. Is there a problem? Many potential measures + Absence of hard (e.g. RCT) evidence on measures to reduce risk + A (perceived) need to act = a problem

4. Managing Uncertainty 1. Identify all possible interventions 2. Elicit ‘expert’ opinion on interventions Aim: Identify best candidate interventions Ideally = highly effective + highly practical

5. Managing Uncertainty Which experts?  Experts (effectiveness)  Inter alia; Public Health, Veterinary Microbiology (Food), Microbiology (Agricultural/Environmental/Clinical), Risk Assessment, Business, Land Management  Farmers (practicality)  Members of farmers unions in Wales and Scotland.  Farmers’ markets How to elicit their views?  Novel method: Best Worst Scaling

6. Best-Worst Scaling Market research tool  Possible to carry out  over distance; no face to face; anonymous  Multiple choice based  Scaled, fine resolution results  Allows respondents to rank long lists without the associated cognitive gymnastics (bite-sized chunks)

7. Best-Worst Scaling Most effective Least effective A Encourage Farmers and farm visitors to wash hands following contact with farm animals.  B  Vaccinate cattle to control pathogen colonisation and faecal excretion of E. coli O157.  C  Remove farm animals from proximity of private water supplies (e.g. at least 50m from well, borehole or other private water supply by fencing-off).  D  Prevent children under the age of 11, and other vulnerable groups, coming into contact with animals at petting, or public visitor farms.  E  No application of manure to ready-to-eat crops within 12 months of harvest and 6 months of drilling/planting.

8. Best-Worst Scaling analysis Take all the “most effective” & “least effective” choices 5 item set we gain information on 7of the 10 paired combinations Respondent chooses the two measures with the maximum difference in performance (best and worst) Maximises the ability to predict peoples choices

9. Intervention generation Literature review  Non-systematic  Published and grey literature included  Project members  Opportunity for experts to add interventions to list

10. Respondent sample Results of the expert elicitation Experts (Effectiveness) Round 1  Contacted 53 experts  31(75%) completed survey  Reduced initial list of 99 to 30 Round 2  Contacted 70 experts  41 (60%) completed survey of 30 interventions Famers (Practicality) Round 3  50 in Wales  50 in Scotland

11. CV scores: a measure of agreement? Experts had higher CV scores than farmers for ‘effectiveness’ (p<0.001) and ‘practicality’ (p<0.01) Farmer CV mean scores were very similar between Scotland and Wales. The mean confidence interval for the practicality assessment was 0.86 for farmers and 1.46 for experts (p<0.001) and for effectiveness was 0.94 and 1.35 respectively (p=0.002).

12. Double fencing Vaccination Effectiveness scores

13. Is there consensus? Vaccination (Intervention 19)

14. Is there consensus? Reducing stocking densities by 50%

15. Is there consensus? P &E Scores are widely distributed indicating poor agreement amongst respondents 16. HACCP for manure handling

16. Practicality scores Hand washing Reduce cattle stocking densities by 50%.

17. Scottish and Welsh farmers (practicality)

18. 19: Vaccination of cattle 1: Hand washing 27: Reduce cattle by 50% 12: Septic tank leakage High practicality/ low effectiveness Low practicality/ low effectiveness Low practicality/ High effectiveness High practicality/ High effectiveness Best-Worst Scaling 2 x 2 plots

19. Method suitability Best suited to the evaluation of large sets of standalone measures Best suited to the evaluation of multidimensional measures (effectiveness and practicality) Smaller multi-level sets of measures, identify the ‘best’ bundles (ACA, CBC)

20. Future Applications in other areas  E.g. Cryptosporidium, Salmonella, anti-microbial resistant genes Modelling of interventions  % reduction in disease prevalence Development of MACCs  % reduction/cost

21. Participating institutions Health Protection Agency; Health Protection Scotland; University of Glasgow: School of Veterinary Medicine; Scottish Agricultural Ccollege; Veterinary Laboratory Agency; Scottish Infection Research Network; HPS Colindale; FSA: Microbiological Safety Division; Bioss Scotland; Wageningen University and Research Centre; Liverpool John Moores University: Centre for Public Health; NFUScotland; NFUCymru; Farmers Union Wales; Office of the Chief Veterinary Officer Wales; Meat Promotion Wales; Quality Meat Scotland; Countryside Council Wales

22. Acknowledgements Thank you

23. Top interventions by effectiveness and practicality No. Intervention 1Encourage Farmers and farm visitors to wash hands following contact with farm animals. 4 Prohibit recreational activities (such as walking and camping) to land where manure, slurry or abattoir waste have been applied, or animals and faeces present, in the previous four weeks. 5 Monitoring of private water supplies to identify those with either high indicator counts, or those in areas of high risk. These supplies would need to be treated (e.g. by ozonation, chlorination or ultra-violet treatment). 7Keep livestock and pets out of ready-to-eat crop areas, using fencing for example. 8 No application of manure to land at high risk of direct flow to watercourses (e.g. adjacent to a watercourse, borehole or road culvert, or areas with a dense network of open drains. 9 Locate solid manure heaps and slurry pits at least 50m away from watercourses, field drains and ready-to-eat crops. 13 No slurry or livestock manure to be applied to high risk fields (ie high risk of transport into adjacent areas watercourses (e.g. when soils saturated or frozen, or heavy rain expected). 19Vaccinate cattle to control pathogen colonisation and faecal excretion of E. coli O157.