John Sumner

A simple, spreadsheet-based, food safety risk assessment tool

The development and use of a simple tool for food safety risk assessment is described. The tool is in spreadsheet software format and embodies established principles of food safety risk assessment, i.e., the combination of probability of exposure to a food-borne hazard, the magnitude of hazard in a food when present, and the probability and severity of outcomes that might arise from that level and frequency of exposure. The tool requires the user to select from qualitative statements and/or to provide quantitative data concerning factors that that will affect the food safety risk to a specific population, arising from a specific food product and specific hazard, during the steps from harvest to consumption. The spreadsheet converts the qualitative inputs into numerical values and combines them with the quantitative inputs in a series of mathematical and logical steps using standard spreadsheet functions. Those calculations are used to generate indices of the public health risk. Shortcomings of the approach are discussed, including the simplifications and assumptions inherent in the mathematical model, the inadequacy of data currently available, and the lack of consideration of variability and uncertainty in the inputs and outputs of the model. Possible improvements are suggested. The model underpinning the tool is a simplification of the harvest to consumption pathway, but the tool offers a quick and simple means of comparing food-borne risks from diverse products, and has utility for ranking and prioritising risks from diverse sources. It can be used to screen food-borne risks and identify those requiring more rigorous assessment. It also serves as an aid to structured problem solving and can help to focus attention on those factors in food production, processing, distribution and meal preparation that most affect food safety risk, and that may be the most appropriate targets for risk management strategies.

A semi-quantitative seafood safety risk assessment

As part of a semi-quantitative risk assessment of 10 seafood hazard/product combinations, a risk assessment tool was used to generate a Risk Ranking. The tool is in a spreadsheet software format and provides a risk estimate, which is scaled between 0 and 100, where 0 represents no risk and 100 represents all meals containing a lethal dose of the hazard. A full description of the tool is contained in Ross and Sumner (this issue). Based on their ranking, seafoods in Australia fell into three risk categories. Hazard/product pairs with ranking < 32 included mercury poisoning (Relative Risk = 24), Clostridium botulinum in canned fish (RR = 25), or in vacuum-packed cold-smoked fish (RR = 28), parasites in sushi/sashimi (RR = 31), viruses in shellfish from uncontaminated waters, (RR = 31), enteric bacteria in imported cooked shrimp (RR = 31) and algal biotoxins from controlled waters (RR = 31). It is noted that there have been no documented cases of food-borne illness from any of the above hazard/product pairings in Australia. Those with rankings 32-48 included Vibrio parahaemolyticus in cooked prawns (RR = 37), V. cholerae in cooked prawns (RR = 37), Listeria monocytogenes in cold-smoked seafoods (RR = 39), scombrotoxicosis (RR = 40), V. vulnificus in oysters (RR = 41), ciguatera in the general Australian population (RR = 45), L. monocytogenes in susceptible (RR = 45) and extremely susceptible populations (RR = 47) and enteric bacteria in imported cooked shrimp eaten by vulnerable consumers (RR = 48). Almost all the hazard/product pairs in this category have caused the outbreaks of food poisoning in Australasia. Those hazard/product pairs with rankings >48 included ciguatera from recreational fishing in susceptible areas (RR = 60), viruses in shellfish from contaminated waters (RR = 67) and algal biotoxins from uncontrolled waters in an algal event (RR = 72). There have been significant (>100 cases) food poisoning incidents involving viruses and biotoxins in shellfish, while ciguatera poisoning is prevalent among coastal communities in Australias warmer waters.

Microbiological Quality of Australian Beef

A survey of the microbiological quality of beef carcasses and boneless beef produced in Australia was conducted during the period June to November 1998. Sponge samples were collected from 1,275 carcasses, and meat samples were drilled from 990 cartons of frozen boneless beef. Carcass and boneless beef samples were respectively collected from 21 and 27 establishments that concentrated on export and from 38 and 3 establishments supplying the Australian domestic market of which 31 were very small plants slaughtering no more than 150 cattle equivalents per week. The mean log total viable counts (TVCs) were 2.42/cm2 and 2.52/g for carcasses and boneless meat, respectively. Escherichia coli was detected on 10.3% of carcasses and 5.1% of boneless beef samples and coagulase-positive staphylococci on 24.3% of carcasses and 17.5% of boneless beef. Salmonella was detected on 0.2% of carcasses and 0.1% of boneless beef and E. coli O157:H7 recovered from 0.1% of carcasses but not detected on 990 boneless beef samples. Mean log TVCs/cm2 differed significantly (P < 0.05) between establishment types. They were lower on carcasses from export establishments (2.20) compared with domestic (2.61) and very small plants (3.10). There were no significant differences in prevalence of Salmonella or E. coli O157:H7 between establishment types. Excision samples were taken from 670 carcasses to make comparisons with the first baseline study of Australian meat, carried out in 1993 to 1994. While there were differences in sampling and microbiological techniques between the two studies that require detailed consideration, there were small but significant improvements in several microbiological criteria for carcasses and boneless meat.

A baseline survey of the microbiological quality of chicken portions and carcasses at retail in two Australian states (2005 to 2006).

Raw poultry products were purchased from the retail market place in two Australian states, New South Wales (n = 549) and South Australia (n = 310). The products sampled on a proportional volume basis were chicken portions with the skin off or skin on, in bulk or tray packs, and whole carcasses. They were collected from butcher shops, supermarkets, and specialty stores from urban areas during the winter (2005) and summer (2006) months. The samples were analyzed to determine the prevalence and concentration of Escherichia coli, Salmonella, and Campylobacter spp. in addition to total viable counts. Salmonella was found in 47.7 and 35.5% of retail chicken samples (35.3 and 21.9% were the less virulent Salmonella Sofia), at mean counts of -1.42 and -1.6 log MPN/cm2 in New South Wales and South Australia, respectively. Campylobacter was found in 87.8 and 93.2% of samples at mean counts of 0.87 and 0.78 log CFU/cm2, respectively. In both states in both seasons, the mean total viable count was 5 log CFU/cm2. On whole birds, E. coli was detected in all winter samples and on 92.9 and 85.7% of summer samples in New South Wales and South Australia, respectively; the log of the geometric mean per square centimeter was 0.5 in winter and slightly lower in summer. On chicken portions, E. coli was detected in around 90% of winter samples in both states, and in summer on 75.1 and 59.6% of samples in New South Wales and South Australia, respectively. The log of the geometric mean CFU per square centimeter for E. coli was 0.75 and 0.91 in winter, and 0.66 and 0.5 in summer in New South Wales and South Australia, respectively.

Quantitative risk assessment of Listeria monocytogenes in ready-to-eat meats in Australia.

Listeria monocytogenes is a food-borne pathogen that can contaminate processed meats and has caused outbreaks in several nations in which processed meats were the vehicle. Due to its ecology, the control of this organism in ready-to-eat meats is difficult. As a first step in improving risk management for this product:pathogen pair in Australia, a stochastic simulation model to predict the numbers of L. monocytogenes likely to be consumed in those products under a wide range of scenarios was developed. The predictions are based on data describing initial contamination levels of both lactic acid bacteria and L. monocytogenes, product formulation, times and temperatures of distribution and storage prior to consumption, and consumption patterns. The model was used to estimate the probable numbers of cases of listeriosis due to processed meats in Australia per year. The model predicted that processed meats could be responsible for up to approximately 40% of cases of listeriosis in Australia, a level considered credible by comparison with available epidemiological data. The reliability of the model, as well as data gaps and further research needs, is discussed.

Microbial contamination on beef and sheep carcases in South Australia.

A total of 523 chilled beef and lamb carcases were sampled from four abattoirs and 13 very small plants (VSPs) in South Australia during March 2002 in order to develop a microbiological profile of meat produced for domestic consumption within the State. Aerobic viable counts (AVCs) and Escherichia coli counts were obtained from samples taken by sponge-sampling the muscle-adipose tissue at sites designated for each species in the Microbiological Guidelines to the Australian Standard for Hygienic Production of Meat for Human Consumption (identical with those of the USA Pathogen Reduction: hazard analysis and critical control point (HACCP) systems: final rule). On beef carcases (n=159) mean log AVC/cm(2) was 1.82 and E. coli was detected on 18.8% of carcases (area sampled 200 cm(2)) for which the mean log of the positives was -0.34; for lamb carcases, on which 75 cm(2) was sampled (n=364), corresponding values were 2.59, 36.2% and log(10) 0.27, respectively. There was little difference in mean log AVC/cm(2) of carcases produced at abattoirs and VSPs, 1.72 versus 1.81, respectively, for beef, and 2.80 versus 2.44, respectively, for sheep. Prevalence of E. coli was lower at VSPs, however, with abattoirs having 28.4% for beef and 61.5% for sheep, compared with corresponding values of 4.7% and 18.5% at VSPs. In VSPs, the range of mean log AVC/cm(2) was 0.47-3.16 for beef and 1.63-3.65 for sheep carcases, data which will allow the Controlling Authority to assist plants to improve performance of slaughter and dressing techniques. The present survey is part of an assessment by the State meat authority of the effectiveness of co-regulation of meat hygiene between government and industry.

Microbiological Quality of Australian Sheep Meat

Microbiological quality of sheep carcasses and boneless sheep meat produced in Australia was surveyed during the period June to November 1998. Sponge samples were collected from 917 carcasses, and meat samples were drilled from 467 cartons of frozen boneless meat. Carcass and boneless meat samples were respectively collected from 7 and 10 establishments that concentrated on export, and from 36 and 5 establishments supplying the Australian domestic market of which 31 were very small plants slaughtering cattle and sheep but no more than 1,200 sheep equivalents per week. The mean log total viable counts were 3.55/cm2 and 3.30/g for carcasses and boneless meat, respectively. Escherichia coli was detected on 29.2% of carcasses and 24.5% of boneless meat samples and coagulase-positive staphylococci on 24.1% of carcasses, and 38.6% of boneless meat samples. Salmonella was detected on 0.1% of carcasses and 1.3% of boneless meat samples. E. coli O157:H7 was recovered from 0.7% of carcasses and 1.3% of boneless sheep meat. There were statistically significant differences between establishment types for some microbiological criteria, although there were no significant differences in prevalence of Salmonella or E. coli O157:H7 between establishment types. While there were differences in sampling and microbiological techniques between this study and another conducted in 1993 to 1994 that require detailed consideration, there were small but significant improvements in several microbiological criteria for boneless meat. While data that would allow for comparison of carcass data were not gathered, it is unlikely that improvements in the microbiological quality of boneless sheep meat could accrue without improvements to carcasses.

A national survey of the microbiological quality of beef carcasses and frozen boneless beef in Australia.

The third national baseline microbiological survey of Australian beef carcasses and frozen boneless beef was conducted in 2004. Carcasses (n=1155) sampled at 27 slaughter establishments had a mean aerobic plate count (at 25 degrees C) of 1.3 log CFU/cm2. Escherichia coli was isolated from 8.0% of the cacasses, with a mean count of -0.8 log CFU/cm2 for positive samples. On samples from 24 boning (fabrication) plants (n=1082), the mean aerobic plate count for frozen boneless beef was 1.3 log CFU/g, and the mean count for the 1.8% of samples with detectable E. coli was 1.5 log CFU/g. E. coli O157: H7 was isolated from 1 of 1,143 carcasses and from 0 of 1082 boneless samples. Salmonella was isolated from 0 of 1155 carcasses and from 1 of 1082 samples of boneless product. No Campylobacter spp. were isolated from carcasses or boneless beef. Coagulase-positive staphylococci were isolated from 28.7% of beef carcasses and 20.3% of boneless beef samples, and positive samples had a mean count of 0.3 log CFU/cm2 and 0.8 log CFU/g, respectively.

Microbiological quality of Australian sheep meat in 2004

The third national baseline microbiological survey of Australian sheep carcases and frozen boneless sheep meat was conducted in 2004. Carcases (n=1117) sampled at 20 slaughter establishments were found to have a mean log total viable count (TVC, 25°C) of 2.28 cfu/cm(2) and Escherichia coli was isolated from 43.0% carcases with a mean log 0.03cfu/cm(2) on positive samples. In samples from 10 boning (fabrication) plants (n=560) the mean log TVC for frozen boneless sheep meat was 1.85cfu/g and the mean log count for the 8.2% of samples with detectable E. coli was 1.39cfu/g. E. coli O157:H7 was isolated from 6/1117 carcases and from 1/560 boneless samples. Salmonella was isolated from 0/1117 carcases and from 3/560 samples of boneless product. Campylobacter sp. were isolated from 4/1117 carcases and from 1/560 boneless samples. Coagulase positive staphylococci were isolated from 23.4% to 32.7% of carcases and boneless sheep meat samples, respectively, with positive samples having a mean log count of 0.93cfu/cm(2) and 1.14cfu/g, respectively. The low level of bacteria described here is consistent with a very low risk to human health due to bacterial hazards in Australian sheep meat.

A National Survey of the Microbiological Quality of Retail Raw Meats in Australia

A national survey of the microbiology of meat (ground beef and diced lamb) at the retail level in Australia was undertaken. For ground beef samples (n = 360), the mean aerobic plate count (APC) was 5.79 log CFU/g, and Escherichia coli was detected in 17.8% of samples; the mean population for these positive samples was 1.49 log CFU/g. Enterobacteriaceae were detected in 96.9% of samples (mean for positive samples, 3.01 log CFU/g), and coagulase-positive staphylococci were detected in 28.1% of samples (mean for positive samples, 2.18 log CFU/g). For diced lamb samples (n = 360), the mean APC was 5.71 log CFU/g, and E. coli was detected in 16.7% of samples (mean for positive samples, 1.67 log CFU/g). Enterobacteriaceae were detected in 91.1% of samples (mean for positive samples, 2.85 log CFU/g), and coagulase-positive staphylococci were detected in 22.5% of samples (mean for positive samples, 2.34 log CFU/g). Salmonella was recovered from 4 (1.1%) of the 360 ground beef samples (isolates were Salmonella Typhimurium phage types), and E. coli O157 was recovered from 1 (0.3%) of 357 samples; Campylobacter and Clostridium perfringens were not recovered from any of the 91 and 94 samples tested, respectively. Salmonella was recovered from 2 (0.6%) of the 360 diced lamb samples (serovars were Salmonella Infantis and Salmonella Typhimurium), Campylobacter was recovered from 1 (1.1%) of 95 samples, and C. perfringens was recovered from 1 (1.1%) of 92 samples.