Defining the model inputs

The model was used to analyse both the restricted and unrestricted estimates for sourcing pigs for export. The unrestricted risk estimate involves all herds and assumes no controls on exports are in place. This unrestricted risk estimate demonstrates the hazard of T. spiralis infection in any domestically produced pig in New Zealand and it

provides the base values from which the effects of the restricted estimates are calculated. The restricted estimate uses the same model as the unrestricted risk, but selected inputs are removed or altered to simulate the different management situations. The restricted risk estimate is the risk that remains once further controls to prevent exporting a pig with T. spiralis have been implemented. The input values stated in

sections 5.2.3.2 and 5.2.3.3 are reported in Appendix 5.1. 5.2.3.1 Risk assessment: Unrestricted estimate

The following two sections describe the release and exposure assessment of the model with parameters from the unrestricted estimate.

Chapter 5 – Risk Assessment 5.2.3.2 Input values for release estimate

R1 and R2 - Selecting a pig from a commercial herd

The sub-variables R1.1 and R2.1 model the number of commercial herds in New Zealand that may carry infected pigs. The outcome of Government testing from 1983 to 2004 on commercial herds provided 95% confidence of detecting T. spiralis at a

prevalence of 0.5% (Clear and Morris, 2004). This testing provided the maximum prevalence value for the between-herd prevalence of commercial herds.

A PERT distribution was used to model the sub-variables R1.1 and R2.1. The negative government testing provided the minimum and most likely values, which were entered as a zero value. As the test is sensitive to detecting disease down to 0.5% we have used a maximum value of 0.005 for confined commercial herds which allows for potential sampling error. Compared to confined herds, non-confined commercial herds had a higher chance of exposure to wildlife and so a higher maximum value of 0.025 was entered.

The sub-variables R1.2 and R2.2 are the within-herd prevalence of T. spiralis for a

randomly selected pig from an infected commercial herd. As no positives have been found in a New Zealand commercial herd the input values for these sub-variables were based on a study of commercially raised pigs from the north eastern United States of America (Gamble et al., 1999). The within-herd prevalence of the 156 sampled herds was up to 10%. From this, the input value of 0.1 was entered as a point estimate for the within-herd prevalence for confined commercial herds (R2.2). However, non-confined herds have unrestricted assess to rodents and wildlife that can aid the transmission of the parasite (Gamble and Bush, 1999) and therefore 0.4 was entered as a point estimate for the within-herd prevalence for non-confined commercial herds (R1.2).

R3 - Selecting a pig from a non-commercial herd

In developed countries Trichinella infection is usually associated with small farm pig

rearing for personal, or local consumption (Joshi et al., 2005; Kapel, 2005; Pozio, 2000). In New Zealand, two investigations have been carried out on infected herds. The first was in 1997 in Rotorua when the discovery of T. spiralis in a pig carcass

initiated an investigation. The whole herd of 31 pigs on the property were slaughtered and tested using the pepsin digest method. However, the prevalence of infection was only assessed for the adult population. This prevalence was 44% (Paterson et al., 1997).

The second herd investigation was carried out in Whangamata in 2001 where 31 adult pigs were tested from the farm and 16 pigs tested positive (Thornton and King, 2004).

This second investigation was used as the input values for a Beta distribution to allow for uncertainty around these values. This uncertainty is because only one herd is considered for this estimate, so a distribution must be used to allow for variations in New Zealand non-commercial herds.

The between herd prevalence for the backyard production systems was modeled as a pert distribution where the input parameters were a minimum of zero, most likely value of 0.002 and a maximum of 0.05. These numbers were based on a number of assumptions. Firstly, the minimum was assumed to be zero because despite large numbers of backyard producers, many of whom kill and cure their own pork and bacon, there have only ever been four human cases in New Zealand. The most likely value was set at 0.002, based on the assumption that the prevalence in these systems is low but likely to higher than that in commercial farms. The maximum was entered into the model as 0.05 as this was considered the worst case scenario.

R4 - Proportion of production

The more intensive confined commercial herds have a higher level of production as the pigs are housed indoors so their environment can be controlled to provide the ideal conditions for optimum growth. Expert opinion was consulted to estimate the contribution of each herd type to national domestic production (Morris pers comm., 2005). Pork from confined herds (R4.2) was considered to contribute to 98% of domestic production and non-confined herds (R4.1) 1% of domestic production. Although there are a very large number of non-commercial herds compared to commercial herds, the number of individual pigs on non-commercial farms (R4.3) is far fewer resulting in a relatively small contribution of 1% to the overall domestic pork production in New Zealand.

Input values for R5

There are two commonly used methods of carcass testing; trichinoscopy and pepsin digest. The first method used was trichinoscopy which involves microscopic examination of a 1 g sample of meat between two plates of glass. The second method is the pepsin digest test where the muscle tissue is digested using pepsin enzymes to

Chapter 5 – Risk Assessment

release the Trichinella larvae which are then examined microscopically. The pepsin

digest test is recommended by the OIE because it has a higher sensitivity than the trichinoscopy method.

The sensitivity of the test was zero for the unrestricted estimate because there was no post mortem test and visual inspection would not detect contaminated carcasses.

Input values for R6

Trichinellosis is often more of a problem when poor food hygiene allows cross- contamination to occur. Cross-contamination at the butchering stage, or incomplete removal of adjoining skeletal muscle could lead to T. spiralis being found in tissues

other than skeletal muscle, for example, offal. However, New Zealand abattoirs have very high meat hygiene standards and strict regulations (Animal Products Act, 1999). These include the requirement that all blades and other butchering equipment are washed between pigs to avoid cross-contamination between carcasses. As a result of these practises the probability of cross-contamination between carcasses is negligible.

The input values for R6 depended on whether the product was pork or offal. Offal has an extremely low risk for being contaminated with >1 larva/g of T. spiralis. This

low risk was modelled using a PERT distribution with a minimum and most likely value of zero, and a maximum value of 0.005.

The input values for pork were entered as a PERT distribution, where the most likely value was taken from a case in 1997 where three of four infected pigs (75%) had larval burdens of >1 larva/g (Paterson et al., 1997). For the pert distribution the minimum value was 0.65, the most likely 0.75, and the maximum value was 0.85.

Input values for R7

Trichinella is tolerant of post-mortem pH changes and is readily passed on to animals

scavenging off the carcass of a dead host several days after death (Nockler et al., 2000; Wakelin and Goyal, 1996).

Therefore a point estimate of 0.99 has been given to R7 because it is highly unlikely that post-mortem changes will affect the survival of T. spiralis in porcine tissues.

Input values for R8

This temperature is consistent with normal freezing of meat and the time of 21 days would be usual for frozen pork to be distributed to point of sale, especially if that was offshore. The risk of viable T. spiralis being present in frozen pork is negligible. Fresh

pork for export would be chilled. Chilling is between -1oC – 8oC which does not sufficiently reduce temperature to affect viability of larvae (Gamble, 1999; Smith, 1975).

Therefore a single probability value of 0.99 was given to the likelihood that T. spiralis would remain viable at the end of the R8 step if it was present at the beginning

because chilling, packaging and transportation of pork is unlikely to affect T. spiralis

infectivity in pork.

5.2.3.3 Input values for entry and release estimate

Input values for L1

New Zealand’s market share of freshly chilled product was estimated by the New Zealand Pork Industry Board to be 5% of Singapore’s total market volume (New Zealand Pork Industry Board, 2003). Singapore’s total market volume of freshly chilled product was taken from the World Trade Atlas for Singapore’s total imports for selected pig products from November 2003 to October 2004 (Singapore Customs, 2004). The values used were 1,381 tonnes of offal and 5,557 tonnes of pork.

Input values for L2

The unrestricted input value was the unrestricted likelihood of entry estimate, as calculated in section 5.2.3.2.

Input values for L3

Different cooking practises are used for the different products which alters the likelihood of T. spiralis larvae remaining viable after preparation for consumption. As

pork was considered to be more thoroughly cooked than offal, offal was allocated a ten times higher risk of carrying viable T. spiralis larva than pork. The point estimates

were derived from expert opinion. This was reflected by allocating a point estimate of 0.01 to pork and 0.1 to offal.

Chapter 5 – Risk Assessment