Published on October 31, 2014
An Epidemiological Investigation of Porcine-Origin Feed Ingredients and the Occurence of Porcine Epidemic Diarrhea on Midwestern United States Pork Farms - Dr. Eric Neumann, Epi-Insight Liimited, from the 2014 Allen D. Leman Swine Conference, September 15-16, 2014, St. Paul, Minnesota, USA.
More presentations at http://www.swinecast.com/2014-leman-swine-conference-material
1. 1 An epidemiological investigation of porcine-origin feed ingredients and the occurrence of porcine epidemic diarrhea on Midwestern United States pork farms Eric J. Neumann,1 DVM PhD; Matthew A. Ackerman,2 DVM; Chris Troxel,3 BS; Ronny L. Moser,3 PhD 1Epi-Insight Limited, Palmerston North, New Zealand; 2Swine Veterinary Services PC, Greensburg, IN, USA; and 3JBS United Incorporated, Sheridan, IN, USA. Allen D. Leman Swine Conference, St. Paul, Minnesota, USA September 13-16, 2014 Roadmap Background (brief) Bias Objectives Methods Results Discussion
2. 2 Background • PED virus was confirmed for the first time in the U.S. on May 16, 2013 (Canada reported on January 22, 2014) – United States • Investigations into the route of introduction were industry-led and included feed analysis (negative samples from May, June 2013) • Other routes were hypothesized • Invited by Matt Ackerman to investigate risk pathways associated with feed manufacture and delivery • Awareness (vitamin, minerals, proteins) → trace-back problem → bioassay → trace-back → on-going concerns prompted another level of commitment to work – Canada • Empirical association between feeding spray-dried porcine plasma (SDPP) and occurrence of a cluster of early outbreaks Bias and context • Have a consultancy role with JBS United • JBS United is a large animal nutrition company – Like any other suppliers, if there is a problem with their product they want to be the first one to know. – They provided me with full access to their company information and management processes. • Feed can be (cross-)contaminated and therefore can be a vector for PED virus (just like any other product) – Vaccine, AI rod, tube of semen, piece of equipment, people, pigs, birds, trucks, and the list goes on) – Pigs have to eat, what is your risk tolerance?
3. 3 Anecdotal observations • Unpublished data (non-peer-reviewed) – Bioassay studies • Minnesota (inoculated feed and contaminated feed) • Ohio (feed from an infected farm) • SDPP manufacturer (natural and inoculated product) • Canada (and USDA-NVSL) related to Canadian outbreak – General observations • SDPP was PCR(+), technically difficult to confirm by virus isolation – Oral exposure to PCR(+) SDPP = possible infection • Feed formulated with PCR(+) SDPP was PCR (+/-) – Oral exposure to feed = no infection Published studies • Dee et al (2014). – An evaluation of contaminated complete feed as a vehicle for porcine epidemic diarrhea virus infection of naive pigs following consumption via natural feeding behavior: proof of concept. BMC Vet Res, 10(1), 176. – Feed can be a vector for PEDV, even when it has been formulated without something like SDPP • Pasick et al (2014). – Investigation into the Role of Potentially Contaminated Feed as a Source of the First- Detected Outbreaks of Porcine Epidemic Diarrhea in Canada. Transbound Emerg Dis, DOI: 10.1111/tbed.12269. – Confirmed that PCR(+) SDPP can contain infectious PED virus • Opriessnig et al (2014). – Porcine Epidemic Diarrhea Virus RNA Present in Commercial Spray-Dried Porcine Plasma Is Not Infectious to Naïve Pigs. PLoS ONE 9(8), e104766. – Even when SDPP is intentionally spiked with PED, exosure doesn’t necessarily result in infection • No analysis of field data on the risk of exposure!
4. 4 Objective • Objective – To investigate the association between exposure to porcine-origin feed ingredients and the occurrence of PED • Study design – Field-based, retrospective case-control study of Midwestern U.S. pork farms (May 2014) • Null hypothesis – No association between exposure to porcine-origin feed ingredients and occurrence of PED. Methods (1) • Reference population – The premises locations at which prepared feed (or feed ingredients) were delivered by a U.S. licensed manufacturer of feed between May 2013 and August 2014. – Included single-site pork production operations, premises within multi-site pork production enterprises, and in some instances secondary private feed milling facilities. – Two feed manufacturing plants equipped with ingredient tracking systems at both the ingredient- and final batch-levels. – Premises in Indiana, Illinois, Iowa, and Ohio.
5. 5 Methods (2) • Study population – Criteria for case premises • All premises in the reference population with PED outbreaks that had been confirmed by laboratory diagnosis (date of first appearance of clinical signs was also recorded); AND • Data on feed deliveries was available at an ‘ingredient lot level’ from the feed manufacturer’s computerized information system • ASSUMPTION: If feed was the source of infection, then it must have been delivered in the two-weeks prior to the start of the outbreak. Methods (3) • ‘Risk-products’ – SDPP – Spray-dried blood cells (SDBC) – Choice white grease (CWG) – Hydrolyzed porcine peptones (HPEP) • Study population – Criteria for control premises • Risk-product ingredient lot numbers that were delivered to case premises were ‘back-traced’ to the mill. • These lot numbers were then ‘forward-traced’ to identify the complete list of premises in the reference population to which the lot numbers were delivered (but were not infected with PED).
6. 6 Methods (3) • Statistical analysis – The strength of association between each of the animal protein ingredients and the occurrence of PED was determined. • Odds ratios (OR) were the basic statistical test used – Estimated through the construction of 2 x 2 contingency tables. • Lot numbers were nested within ingredients – managed through Mantel-Haenszel adjustment when determined to be necessary. – Also estimated through logistic regression. • Conducted using hierarchical mixed model logistic regression to account for the nesting of lot numbers within each ingredient. Odds ratio de-mystified Odds of the disease in an exposed group, divided by the odds of disease in a non-exposed group Disease POS NEG Exposure POS a b NEG c d Disease POS NEG Exposure POS 1 10 NEG 1 100 OR= a/b c/d Example data: 112 premises OR= 1/10 1/100 = 0.1 0.01 =10 Simple math
7. 7 Results (1) • The strength of association between four porcine-origin feed ingredients (SDPP, HPEP, SDBC, and CWG) and the occurrence of PED was estimated at both an ingredient- and lot-level basis. – In addition, the strength of association between fish meal (FISH) and the occurrence of PED was determined as a reference value for the population. • There were 43 cases and 418 controls represented in the study population. Results (2) SDPP1 lot numbers Lot 1 Lot 2 Lot 3 Lot 4 Lot 5 Lot 6 Unadjusted ORs for SDPP1 lots. Dot represents the OR estimate and length of the bar indicates 95% CI.
8. 8 Results (3) CWG lot numbers Lot 1 Lot 2 Lot 3 Lot 4 Lot 5 Lot 6 Lot 7 Lot 8 Lot 9 Lot 10 Unadjusted ORs for CWG lots. Dot represents the OR estimate and length of the bar indicates 95% CI. Results (4) HPEP Summary ORs for each ingredient. Dot represents the OR estimate and length of the bar indicates the 95% CI. The ORs for SDPP1 and CWG have been adjusted by Mantel-Haenszel technique.
9. 9 Other epidemiologic work • Lowe et al (2014) – 3x as likely for truck to become contaminated if slaughterhouse staff enter truck when unloading when selling pigs – 4x as likely if unload after a positive truck • Morrison et al (2014) – 4x more likely to be positive if you are within 1 mile of PED positive farm – Being a sow farm = 8.8x more likely – Rendering pickup = 7.1x more likely • U of MN – Feed Risk Assessment project Discussion • In some cases, more than one cohort of pigs was represented at a given delivery premises. – Identical situation for both cases and controls, should have minimized any bias • Some premises received more than one lot number of an ingredient, or multiple deliveries of one lot number, or diets were not identical in each case or control – i.e. a ‘dose-effect’ – Assessed, and no effect was detected. However, data was not well-suited to this analysis. • Other risk factors were not assessed due to the retrospective nature of the study. – All outbreaks were included, no attempt to distinguish source of introduction • Only looked ‘porcine-origin ingredients’
10. 10 Conclusions • Ingredients had negligible to very low association with PED in this study • Minor differences among lot numbers for some products – not a consistent issue. • No compelling evidence in this study for recall • Work with reputable suppliers • Work to improve traceability from the ingredients to the pigs – Tracking systems (pigs, product, feed) are available off-the-shelf • Educate clients and production staff, communicate, report – Underpins all PED mitigation strategies