Four

-

-

-

Salmonella

the amount tested is usually governed by practical considerations. In certain circumstances, however, it may be necessary to test unusually large sample amounts. During the investigation of baby fooa-implicated in 3n Australia-wide outbreak of infant salmonellosis, it was shown that sample amounts of between 400-600 g per kg container had to be examined before the causative organism, Salmonella Bredeney, could be isolated (104). In determining the quantity of sample to be tested, consideration should be given to the nature of the product, tnp pyppcted I@U€dftf contamination, the probRnlp di"triblltion of salmonellae within the product, and the purpos.e of the examination, i.e. routine or invpstigRtion ~

The sample size and the number of sample

units (sub-samples) to be tested in the routine examination of foods for salmonellae is usually defined by a sampling plan. The establishment of statistically based sampling plans for different food categories is discussed in Chapter 3. These sampling plans sometimes require the examination of a large number of sample units to attain a high probability of safety. For high risk foods such as infant formulae, it has been recommended that 6,.2individual 25 g sub-samples should be tested with negative results (205). The US Food and Drug Administration (FDA) recommend sampling to the following plan according to the nature of the food (19): Category I. Foods that would not normally be subjected to a process lethal to Salmonella between the time of sampling and consumption and are intended for consumption by the aged, the infirm and infants (60 analytical units); Category Il. Foods that would not normally be subjected to a process lethal to Salmonella between the time of sampling and consumption (30 analytical units); Category Ill. Foods that would normally be subjected to a process lethal to Salmonella between the time of sampling and consumption (15 analytical units). In most laboratories, routine examination of such large sample numbers is not economically feasible. The compositing or pooling of multiple samples, however, provides a more practical approach to Salmonella control: Huhtanen et al. (185) examined 25 meat and bone meal samples for salmonellae, comparing a si~ sample with ten 30 g samples, and found that there was little difference in isolation efficiency. Silliker and Gabis (342) compared the efficiency of testing sixty 30 g, fifteen 100 g and three 500 g subsamples for the recovery of salmonellae from naturally contaminated dried foods such as chocolate bars, egg products, pepper, desiccated coconut and meat and fish meals. Their results

demonstrated that salmonellae could be detected with equal accuracy and with the same statistical reliability using any of these sample pools. Similar results were obtained in a subsequent study of high moisture foods such as meat, poultry and eggs (343). Compositing of samples is now commonly used in many laboratories throughout the world and is particularly advantageous for foods where positive results are relatively rare. In the laboratories of the FDA, 25 g sub-samples may be composited to a maximum size of 375 g (19). In Australia 375 g composite samples are routinely used for many foods, especially dried milk powders and AS 1766.2.5-1991 allows pooling of up to 15 samples of the same batch of product. Although sample units are usually measured by mass, for some foods it may be more appropriate to examine the external surface of the product only. In this case test samples can be obtained by excising, swabbing or washing (rinsing) a defined surface area. These sampling procedures are particularly applicable to poultry, fish and some types of meat, where salmonellae are unlikely to contaminate the unexposed tissues. Excision of tissues followed by maceration can provide precise data on levels of surface contamination, but is not generally favoured for routine work because of the resultant downgrading of products. While surface swabbing is often preferred because of the ease and speed of manipulation, the level of recovery is frequently inconsistent and incomplete. The 'whole bird rinse' technique described by Surkiewicz et al. (364) had been widely recommended as an effective, non-destructive and practical method for the routine testing of poultry for salmonellae (64, 96, 203, 290, 396). The procedure involves rinsing both the external and internal surfaces of the carcass in an appropriate volume of a suitable diluent for 1-2 min using a sterile plastic bag. The sensitivity of this method has been shown to be related to the volume of rinse fluid examined (73, 364) but rinse volumes of 300, 500 and 1000 mL provide essentially equal recovery efficiency (44). It has also been shown that the sensitivity of the technique may be improved by incubating the entire carcase in the rinse solution (73). This latter procedure, however, has not gained acceptance because ofthe need for additional incubation space and because the product is destroyed. As a general method for examining food surfaces, a rinse technique is probably suitable for most purposes. In Australia, the rinse technique has been accepted as a standard reference method for examining poultry for salmonellae (358). Specific discussion of the issues arising from non-selective pre-enrichment and selective 233

~