4/29/2009 - Welcome | Iowa State University



The Impact of Pig Health on Human Food-borne Risk

Abstract 1

Situation 1

Overview 1

Animal health and human health relationship 2

Animal health management 2

Modeling 2

Preliminary field studies by this team to estimate potency ratio 4

Lesions at slaughter 7

Further work is needed 8

Objectives 8

Inputs – What we invest 9

Activities - What we do (Materials and Methods) 9

Objective 1 – systems policy model 9

Objective 2 – Estimation of key parameters (relationships) with epidemiologic field studies 11

Activities - Who we reach 14

Outputs 15

Outcomes 16

• Knowledge 16

• Actions 16

• Conditions 16

Assumptions 17

External factors 17

The Impact of Pig Health on Human Food-borne Risk

H. Scott Hurd, James Dickson, Derald Holtkamp, Rodney Baker, Annette O’Connor, Louis A. Cox

Abstract

The long-term goal of this project is to develop information and decision tools for scientists, policy-makers, producers, veterinarians, and students. These tools can be useful in assessing the macro-level impacts of various livestock production practices. Our first objective is to develop a quantitative systems policy model linking the relative impacts of pig health on incidence of human food-borne illness. One of the critical relationships in the model, for which there is a paucity of data, is the relationship between the prevalence of lesions at slaughter and carcass contamination. Therefore, our second objective is to generate information on these relationships, which we will use to parameterize the decision-support tool. For the second objective field data collection will entail two studies implemented with 40 pork producers and two Midwestern abattoirs. The first study will compare contamination rates (Salmonella spp., Campylobacter spp., Escherichia coli Biotype I/II ) and microbial load in lesioned and non-lesioned carcasses. The second study will determine if low herd health is associated with increases in the proportion of lesioned carcasses observed at slaughter. This proposed research is unique in that it will: 1) develop a model, 2) parameterize the model with primary data collected specifically for that purpose, and 3) disseminate the model, data, and findings to decision-makers from farm-to-fork. The “systems” model and data as proposed in this project can alter the entire decision-making paradigm as the public health impact of changing pork production practices is better understood.

Situation

Overview

A long-standing premise of the U.S. food safety inspection system is that healthy livestock are essential for a safe food supply. This premise is the primary motivation for USDA antemortem and gross pathological postmortem inspection. Many groups in society, including politicians, activists, scientists, and stakeholders, are advocating significant changes to livestock production practices. These changes include modification to stocking densities, limitations on antimicrobial use, and requirements for outdoor exposure. Such changes will affect animal health! Will changes in animal health also affect the microbiological quality of resulting meat, and consequently public health?

This food inspection premise is somewhat driven by the “yuck” factor of consuming meat from overtly unhealthy animals. It is unclear whether unhealthy animal products have quantitatively increased microbial contamination (e.g. Salmonella and Campylobacter ). If the latter premise is true, then any new policy or practice that reduces animal health is less desirable as it may reduce public health. Further work is needed to evaluate this premise; providing quantitative data and analysis on the relationship between changes in animal health and resulting changes in human food-borne risk.

Animal health and human health relationship

During antemortem inspection, overtly ill animals will be prevented from entering the slaughter process or be tagged as “suspect”; receiving extra postmortem inspection. However, there is a portion of animals that harbor some pathology or evidence of illness which is not discernable antemortem (Russell, 2003; Adreasen et al., 2001; Hurd et al., 2008). These we will refer to as “subclinically ill.” This subclinical illness could be an active on-going infection or the remains of some tissue damage such as adhesions, abscesses, arthritis, or injections-site lesions (Noyes et al., 1990).

This subclinical illness may increase carcass contamination in a variety of ways. Animals stressed or immune compromised by long-term, low-grade illness are more likely to be infected with food-borne pathogens, especially Salmonella (Johnson and McGlone, 2007). Additionally, animals with abscesses or other significant lesions will require extra trimming or further handling during the slaughter process. This handling may increase the likelihood of cross-contamination (Olsen et al., 2003; Rosenquist et al., 2006). Finally, certain illnesses or conditions may increase the chance of mistakes during the slaughter process. An adhesion, for example, may cause the adherence of intestines to the body cavity. During evisceration, extra force may be required, leading to leakage or spilling of intestinal contents. If leakage occurs, there is about a 40% probability of Salmonella contamination, given that percentage of swine carrying Salmonella in the gastrointestinal tract at the time of slaughter (Hurd et al., 2002; Hurd et al., 2004; Rostagno et al., 2003).

Animal health management

The diseases most likely to cause the above-noted conditions are infectious processes, usually bacterial, such as respiratory disease or peritonitis. Practices that reduce these conditions include adequate ventilation, protection from environmental extremes, good nutrition, preventative medications, vaccinations, biosecurity, and proper stocking densities (Wallgren et al., 1994).

Many groups, in addition to the producer and veterinarian, are now trying to direct and regulate livestock management practices. For example, in 2008 California passed a law regulating the “confinement of certain farm animals ….” (California, 2008). The U.S. Congress recently introduced a bill restricting the use of on-farm antibiotics (Rep. Slaughter, 2009). Some organizations are blaming the April 2009 outbreak of swine flu on intensified pork production (Food and Water Watch, 2009). The alternative to intense confinement swine rearing is extensive outdoor rearing where the environment is more difficult to control. Recent research comparing outdoor and confinement reared pigs showed a higher prevalence of zoonotic pathogens such as Salmonella, Toxoplasma, and Trichinella in pigs raised outdoors (Gebreyes et al., 2008).

Modeling

Some members of this research team, Hurd, Cox, and Dickson, have developed a quantitative policy simulation model to evaluate potential human health risks and benefits from changes in poultry health (Singer et al., 2007). The model predicts the changes in human illness days (campylobacteriosis) per year caused by changes in the number of subclinically ill birds presented to slaughter. This model is foundational to this proposed project. It will be converted from a poultry to a pork application.

The current version of the model is highly aggregate, reflecting the very high-level data that have been available to date. Equation 1 shows one of the four differential equations used to model the change in human illness days (IH) relative to changes in the quantity of subclinically ill animals (IA) presented to slaughter. The model assumes that clinically ill animals would be removed due to antemortem inspection.

dIH / dt = [c + d * IA + e * (1 – IA)] * (1 – IH) – h * IH (1)

The main variables in this model are as follows:

▪ IH(t) = the fraction of the human population of interest that has a specified food-borne illness, such as campylobacteriosis, at any time t. (IH = “ill human” fraction.)

▪ IA(t) = the fraction of servings of a particular food commodity that comes from animals with a specified illness or adverse condition (e.g., airsacculitis or necrotic enteritis) that the animal treatment could help prevent, reduce, or control. (IA = “ill animal” fraction for servings from processed animals. Animals not sent to slaughter and animal carcasses removed during processing are excluded from consideration when IA is calculated, as they presumably do not affect IH.)

▪ c = background human illness from non-meat causes

▪ h = background recovery rate of human illnesses

▪ d = proportion of human illness rate (IH) generated from the consumption of subclinically ill animals

▪ e = proportion of human illness rate per serving from healthy animal population

The change in human illness (dIH / dt) is modeled as a function of the proportion of the illness rate from ill animals (d) and illness rate per serving from healthy animals (e). The ratio of d/e, termed the potency ratio (D) was used to represent the relative pathogen load on poultry carcasses from birds with and without airsacculitis. The parameter D was very uncertain due to dependence on only one publication (Russell, 2003) Therefore, the results were modeled across a range of D from 1 to 10, where 10 represents a 1 log increase in Campylobacter spp. contamination.

As shown in Figure 1, human illness days, even with a low potency ration (D = 2), were very sensitive to small changes in animal illness rates (IAnew). Therefore, it is critical for decision-makers to carefully evaluate and consider the human health impact of policies affecting animal health. Clearly, additional data are needed to better estimate the magnitude of the potency ratio for poultry, beef, and pork production. A critical objective of this proposed project would be to estimate the potency ratio (D for salmonellosis and campylobacteriosis in the Midwest pork production.

[pic]

Figure 1. Changes in the total number of human illness days per year due to Campylobacter expressed as a percentage change from baseline. The percentage change in illness days is shown as a function of D, the potency ratio between chicken servings from ill vs. healthy chickens. All other parameters are held constant at their baseline values. The model was evaluated for different values of animal illness rates (IAnew), the prevalence of ill chickens after intervention.

Preliminary field studies by this team to estimate potency ratio

The previous work by Singer et al. showed the importance of the potency ratio for accurate inferences from the model. Therefore, we have been working to estimate this parameter for pork production. Two studies using primary data collected by this team showed that pig carcasses affected with lesions indicating chronic internal infections were 2 to 5 times more likely to be contaminated at end of slaughter with indicator bacteria and pathogens.

The most recent study, funded by the USDA Formula Funds Program, was a follow-up to the previous study comparing the group frequency of peelouts with the carcass contamination (Hurd et al., 2008). A “peelout” occurs when the pleural and peritoneal lining must be removed due to adhered visceral tissue (e.g. lungs, liver, etc, Figure 2). The study’s purpose was to evaluate lesion scoring and contamination on an individual carcass basis. Due to funding restrictions, the number of carcasses cultured was limited. However, this study provided compelling evidence of a significant and quantifiable association between lesions and carcass contamination. Additionally, it showed that investigators can readily discriminate lesioned versus non-lesioned carcasses on a busy slaughter line.

In one high-speed abattoir (collaborator on this proposed project) we collected swab samples from 358 carcasses. A lesioned carcass was identified immediately after evisceration. A photo was taken and was marked for swabbing at the end of slaughter, just before final rinse (Figure 2). After another 5 to 10 carcasses passed, a non-lesioned carcass was marked and photographed. The viscera from each marked carcass were also photographed.

The photos were later viewed by three board-certified pathologists, and the severity was scored. Each half of the carcass was scored 0-3 based on the amount of adhered visceral tissue. The heart was scored 0-1 based on presence of pericarditis. The three scores were added to obtain a pathology score, which could range from 0-7.

Before taking pictures, the initial call of “lesioned” or “non-lesioned” was determined by a data collector. For analysis purposes, the pathology score was used to redefine lesioned and non-lesioned. Every pig with a total pathology score of higher than 2 was considered lesioned, while pigs with a score of 2 or less were called non-lesioned.

[pic]

[pic]

Figure 2. The lesioned carcass cultured Salmonella positive (upper) and non-lesioned carcass was cultured Salmonella negative (lower)

Table 1 shows Salmonella were recovered from 36 (10.1%) of the 358 carcass samples. Enterococcus were recovered from 39 (10.9%) of the carcasses. Of the 187 lesioned carcasses, 23 (12.3%) were positive for Salmonella, 21 (11.2%) positive for Enterococcus, and 39 (20.9%) were contaminated by either or both. Among the 171 non-lesioned carcasses , positive percentages were 7.6%, 10.5%, and 18.1% for Salmonella, Enterococcus and contamination by either or both, respectively. A contaminated carcass was defined as one positive for either Salmonella and Enterococcus or both. The percentages of Salmonella, Enterococcus, and contaminated positive carcasses out of the carcasses sampled changed frequently across the four replicates. Replicate 1 had the lowest contamination percentages of Salmonella, Enterococcus, and contamination by either or both, while replicate 4 had the highest.

A logistic regression model, with replicate as a covariate, shows the probability of Salmonella positive in lesioned carcasses was 70% higher than in non-lesioned carcasses (OR = 1.7, 95% CI 0.82 – 3.51). The lack of statistical significance (p≤0.05) is likely due to the relatively small sample size. The probability of Enterococcus positive in lesioned carcasses was 19% higher than in non-lesioned carcasses (OR = 1.19, 95% CI 0.61 – 2.31). The probability of contamination by either or both was 25% higher in lesioned carcasses compared to non-lesioned (OR = 1.25, 95% CI 0.73 – 2.12).

Figure 3 shows that in-line data collector can readily discriminate between lesioned and non-lesioned carcasses. The agreement (Kappa value) between initial call (data collector) and different cutoff levels (e.g. > 3, or > 5) of total pathology scores was very good. The total score is the sum of all three pathologists, so it could be as high as 21 when including the heart score, and 18 when excluding the heart score. The initial call did not include observation of the heart, so the agreement was calculated for the resulting total pathology scores, with and without the heart score. The greatest agreement (Kappa=0.952) with initial call was a lesion of greater than 2 without the heart. Agreement was high (Kappa > 0.9) up until total pathology score exceeded > 5. As noted above, the results of the agreement analysis informed our decision regarding how to define a lesioned carcass for purposes of analyzing the correlation between lesions and contamination.

Table 1. Number and percentages of positive carcass swabs for Salmonella, Enterococcus, and contamination by either or both, categorized by replicates of visiting slaughterhouses.

| |Replicate 1 |Replicate 2 |Replicate 3 |Replicate 4 |Total |

|Carcasses swabbed |102 |97 |98 |62 |358 |

|Carcasses Salmonella positive (percentage) |1 (1.0%) |11 (11.3%) |11 (11.2%) |13 (21.0%) |36 (10.1%) |

|Carcasses Enterococcus positive (percentage) |8 (7.8%) |13 (13.4%) |10 (10.2%) |8 (13.0%) |39 (10.9%) |

|Carcasses 1Contaminated positive (percentage) |9 (8.8%) |22 (22.7%) |21 (21.4%) |20 (32.3%) |70 (19.6%) |

1Contaminated = sample positive for either or both of Salmonella and Enterococcus

[pic]

Figure 3. Agreement (Kappa value) between initial call and different cutoff levels of total pathology scores. X axis is the various cutoff definitions for lesioned vs. non-lesioned.

Lesions at slaughter

A remaining question is how well lesions at slaughter reflect animal health status on the farm. A number of studies have investigated the association between gross lung lesions at slaughter and various measures of pig health and performance during the growing period. A report by Morrison et al. in 1986 reviewed 23 studies. A reduction in growth rate and/or increase in feed conversion was associated with pigs that had lung lesions at slaughter in 13 studies, but no association was reported in the other 10 (Morrison et al., 1986). A more recent study (Regula et al., 2000) found the presence of gross lung lesions at slaughter to be associated with lower average daily gain

When pigs are affected by respiratory disease early in the growing period, lung lesions have sufficient time to resolve before slaughter (Andreasen et al., 2001; Noyes et al., 1990; Wallgren et al., 1994). Noyes et al. used radiography to evaluate lung lesions over the life of the pig and found that pneumonia was a dynamic process with lesions progressing and regressing throughout the life of the pig. Noyes did report a statistically significant relationship (p=0.025) between the percentage of affected lung at slaughter and body weight at 180 days of age. The association was much weaker than that between lifetime pneumonia and body weight at 180 days of age (p ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download