Dar seafood ppp standard

Risk ranking model employed

While each risk ranking approach has its strengths and could be used as a tool for ranking relative risks due to seafood consumption in Australia, none met the requirement of providing a broad three-tier categorisation of seafood industry sectors which could be used in a through-chain assessment of risk. Specifically, a need was identified for a broad comparative method able to take into account a large number of hazards having widely different adverse health consequences. This report uses a method based on consideration of the elements of risk as defined by Codex: the likelihood (probability of occurrence) and severity of adverse health effects (illness).

In deriving estimates of the likelihood of illness due to the presence of a particular hazard in a seafood commodity, available data on seafood consumption and the prevalence and levels of hazards in seafood in Australia were taken into consideration. However, significant data gaps have militated against completion of a formal quantitative exposure assessment for each hazard/commodity pairing considered. In addition, specific characteristics of each hazard, factors along the supply chain that might influence the final risk at point of consumption, and recent epidemiological data were all taken into account. Estimates of the severity of adverse health effects were based on an accepted international scheme, which was adapted to take into account hazards not originally included in that scheme (for example, heavy metals, algal biotoxins, helminthic parasites).
A decision matrix was developed to provide broad qualitative rankings of public health and safety risks due to food safety hazards associated with seafood sectors or commodities. The matrix combines the estimates of severity and likelihood of adverse health effects to arrive at a relative risk ranking. The method used to estimate the severity and likelihood of adverse health effects and to combine these into risk rankings are described below.

Severity of adverse health effects

The estimate of the severity of adverse health effects caused by a food-borne agent is based on the ranking scheme for food-borne pathogens and toxins described by the International Commission on Microbiological Specifications for Foods (ICMSF) [28]. The ICMSF ranking scheme categorises hazards by the severity of the threat they pose to human health, taking into consideration the:

• 
  likely duration of illness
  
• 
  likelihood of death
  
• 
  potential for ongoing adverse health effects.
  
  
  

The severity of adverse health effects caused by a hazard is ranked as moderate, serious or severe according to the following definitions:

• 
  moderate severity applies when the hazard is likely to cause an illness of short duration with no ongoing adverse health effects (sequelae)
  
• 
  a serious hazard would cause an illness of longer duration, with some chance of ongoing chronic and debilitating effects
  
• 
  a severe hazard would cause illnesses with serious sequelae or significant mortality rates.
  

Under the ICMSF ranking, severe hazards are further divided into those applying to the general population and those applying to specific sub-populations, that is, susceptible individuals (for example, the very young and old, the immunocompromised, and pregnant women and their unborn children). This takes into account those situations where a hazard considered to be of moderate or serious severity to the general population may cause a severe illness in certain susceptible sub-populations. This is reflected in the severity rankings presented in Table 3.

The ICMSF severity ranking scheme was developed to apply principally to microbiological hazards. In this report, the ICMSF approach has been adapted to rank the severity of threat posed by all microbiological and natural contaminant hazards potentially associated with seafood eaten in Australia. Certain chemical contaminants and microbiological hazards not originally ranked by the ICMSF have been added to Table 3 (entries with asterisks) based on information available on the severity of adverse health effects associated with exposure to them. Justification for severity rankings is in the relevant sections of Appendix 4.
In determining the severity ranking for chemical contaminants, particularly heavy metals, it was noted that adverse health effects tend to become evident after prolonged exposure, and are typically chronic in nature. This is in contrast to the situation applying to most microbiological hazards, where a single exposure can lead to illness.
In determining the severity of the threat to human health, no consideration has been given to the likelihood of exposure to the hazard or the probability of occurrence of any illness caused by that hazard. These considerations are encapsulated in the estimate of likelihood of adverse effects, below.

Table 3: Ranking of food-borne hazards by severity of adverse health effects

Severity	Description	Food-borne hazards in seafood
Moderate	Not usually life threatening; no sequelae; normally short duration; symptoms are self-limiting; can include severe discomfort.	Staphylococcus aureus enterotoxin	Histamine
		Enteropathogenic Escherichia coli	Vibrio parahaemolyticus
		Enterotoxigenic E. coli	Zinc*
		V. cholerae non-O1/non-O139	Wax esters*
		Norwalk-like viruses (noroviruses)
Serious	Incapacitating but not life threatening; sequelae infrequent; moderate duration.	Non-typhoid Salmonella spp.	Yersinia spp.
		Non-dysenteric Shigella spp.	Listeria monocytogenes
		Aeromonas hydrophila*	V. vulnificus*
		Hepatitis A virus	Helminthic parasites*
		Algal biotoxins* (DSP, NSP)	Mercury*
		Ciguatoxin*
Severe	Life-threatening or substantial chronic sequelae or long duration.	General population
		S. Typhi S. Paratyphi	Cadmium*
		Shigella dysenteriae	V. cholerae O1/O139
		Enterohaemorrhagic E. coli	Aflatoxins
		Clostridium botulinum neurotoxin	Arsenic*
		Algal biotoxins* (ASP, PSP)	Lead*
		Cadmium*
		Susceptible populations
		L. monocytogenes	V. vulnificus
		Enteropathogenic Escherichia coli and Enterotoxigenic E. coli	Hepatitis A virus
		Mercury

* Hazards not originally listed in the International Commission on Microbiological Specifications for Foods severity ranking table.
Key: ASP = amnesic shellfish poison; DSP = diarrhoetic shellfish poison; NSP = neurotoxic shellfish poison; PSP = paralytic shellfish poison.
See Appendix 4 for discussion of the different severity rankings amongst the algal biotoxins.

Likelihood of adverse health effects

The estimate of the likelihood of adverse health effects takes into consideration relevant available data and information on:

• 
  the link between the hazard and illness due to consumption of the particular seafood (epidemiological data)
  
• 
  the prevalence and concentration or level of the hazard in seafood
  
• 
  patterns of consumption of the specific seafood (that is, frequency of consumption, amount eaten)
  
• 
  the impact of existing regulatory and non-regulatory risk management systems
  
• 
  data and information on the following factors related to the properties of the hazard and the effect of production, processing and handling, particularly in terms of how they might influence hazard levels at the point of consumption:
  

• 
  the capacity for microbiological pathogens to survive or grow in the commodity
  
• 
  any other relevant properties of the hazard (for example, toxigenic or infectious dose)
  
• 
  the probable effect of production, processing and handling on the presence and level of the hazard
  
• 
  the likely effect of consumer handling (including cooking and product shelf life) on hazard levels.
  

Epidemiological data

Australian and overseas epidemiological evidence of food-borne illness resulting from consumption of a particular seafood commodity demonstrates observed risk, and is accorded the highest weighting in considering the likelihood of adverse health effects. Where available, Australian epidemiological data were considered to be of greater relevance than those from overseas. However, it is important to note the limitations of existing epidemiological data on seafood-borne illness in Australia.

The hazards listed as ‘moderate’ in Table 3 result in self-limiting illnesses of short duration and tend to be poorly reported in all epidemiological datasets. This is emphasised by United States data that shows even quite severe illnesses are significantly under-reported or are not traced back to a particular hazard and/or food source [20]. In cases where an epidemiological link is established, the strength of the linkage is expressed as an odds ratio. The strength of that ratio indicates the likelihood that the linkage is true. It is understood that there is a great deal of uncertainty with this form of data.

Prevalence and concentration data

Data from monitoring studies of hazards in seafood were used to demonstrate a clear link between particular hazards and certain seafood commodities. For some commodity/hazard combinations (for example, mercury in finfish, L. monocytogenes in cooked crustacea), there have been sufficient studies conducted to provide a clear picture of the extent of contamination in products available in Australia. For other combinations there is little Australian prevalence and concentration data to draw on.

Other information used to assess the likelihood of adverse health effects included data on the prevalence and level of hazards in imported seafood (for example, data from the Australian Quarantine and Inspection Service Imported Food Inspection Program testing regime) and a database of Australian food recalls maintained by FSANZ.
Information from these sources was taken as being indicative of the potential for the seafood commodity/hazard combination to cause adverse health effects. This type of data is treated cautiously, however, as food standards are conservatively set to protect public health and safety. Therefore, failures to meet regulatory limits may not necessarily indicate the presence of a hazard at a concentration likely to present an immediate risk to public health. However, while the presence of a hazard does not necessarily imply an immediate threat of illness in a consumer of that product, other factors may apply, leading to a public health and safety risk.
The use of such routine inspection data also has other limitations. Not all testing is random, and sample size is often quite low compared to the total food volume. The data therefore have the potential to be biased and can become insensitive for low incidence pathogens, introducing uncertainty which can affect the validity of conclusions based upon them.

Consumption data

Data on the level of consumption of specific seafood commodities in Australia were used to adjust the estimates of likelihood of adverse health effects, to take into account relative exposure to hazards in different seafood commodities in the general population.

Consumption figures were derived from the results of the 1995 National Nutrition Survey of Australia, using FSANZ’s dietary modelling computer program, DIAMOND. The National Nutrition Survey collected data from 13 858 respondents, using a 24-hour food recall method. Data on consumption of seafood commodities are presented as mean and 95th percentile consumption amounts along with the number and percentage of respondents who ate that seafood commodity in the survey period (Appendix 3). Some limited frequency of consumption data were also collected as part of the National Nutrition Survey, but only covered a few broad food categories and were of limited value to this assessment.
Food-borne illnesses due to microbial hazards are almost always related to single exposure episodes or consumption on a single day. As such, the 24-hour recall data provide critical insights into the reach of the likely exposure over age/gender groups, the extent of the exposure in terms of individuals exposed and the dose of the exposure in terms of the portion eaten. Here, the long-term intake is not an important factor, so absence of frequency of consumption data is not a major limitation to the ranking. In this case, only sample size is a limitation for obtaining precise estimates for relatively rare events.
A limitation of the National Nutrition Survey method is that it tends to over-estimate habitual food consumption amounts for high consumers. In particular, for foods that people tend to consume less than once a week (for example, molluscs and crustacea), consumption figures derived from a 24-hour recall may be higher for most consumers than if consumption amounts were averaged over a longer time frame that better reflects habitual consumption of these foods.
For metals and toxins, the likelihood estimates are based on the dietary modelling conducted by ANZFA for the review of the Code [4,5,7].
Statistics on production, import and export of seafood commodities, such as those available from the Australian Bureau of Statistics – the Apparent Consumption of Foodstuffs series – and the Australian Bureau of Agricultural and Resource Economics – the Australian Fisheries Statistics series – were also considered, where appropriate. These data sources provide useful information on the gross amount of seafood available for consumption each year in Australia, but do not contain any insight into the frequency, prevalence or levels of consumption of these commodities within the Australian population such as is provided by the National Nutrition Survey data.

Impact of existing regulatory and non-regulatory risk management systems

The likelihood estimates and resultant risk rankings take into account the effect of the regulatory and non-regulatory mechanisms currently in place to manage seafood safety risks. This approach recognises that the epidemiological data and information on the prevalence and concentration of hazards in seafood are gathered in an environment where those existing risk management mechanisms apply. It also allows some conclusions to be made about the effectiveness of those measures and may point to inadequacies or gaps in the current system.

It is also recognised that changes in management of public health risks associated with seafood have recently taken place, either in response to incremental advances in awareness of food safety issues or in reaction to failures in the previous management systems.
The impact of these changes may not be fully gauged by assessing the burden of food-borne illness. Epidemiological data are an historical record and may not accurately reflect recent changes in the regulatory environment. For example, the effectiveness of the partial, yet ongoing, implementation of a shellfish safety program in New South Wales can only be completely validated over a period of time.
Such ongoing changes in risk management systems introduce a further degree of uncertainty into the final relative risk rankings generated in this report. They also demonstrate the need for risk assessment to be an ongoing, dynamic process, responsive to changes in circumstance and the availability of data.

Properties of the hazard and the effect of production, processing and handling

Factors potentially affecting levels of hazards in seafood at the point of consumption were also considered. Production and processing methodologies, consumer food handling habits, product shelf life and the capacity for the seafood to support the survival or growth of pathogens were taken into account. In specific instances, consideration was given to the potential for cross-contamination of seafood commodities during processing and consumer handling, although data on the incidence of contamination is extremely limited.

Data limitations

In general, limited access to or availability of data presented considerable difficulties for assessing the likelihood of adverse health effects. Insufficient data on the prevalence and levels of microbiological hazards in seafood available for consumption in Australia prevented the undertaking of a quantitative exposure assessment, so the estimates of likelihood of adverse health effects are largely qualitative, and this impacts on the risk rankings derived from them.

A clear distinction is made, however, between those cases where limited data is available (leading to uncertainty in the conclusions) and those where the available data shows lack of, for example, a specific hazard in a specific commodity (evidence which will tend to reduce the ‘likelihood of illness’ rating in that case).
Table 4 indicates the way in which the various forms of data and information were used to rank the likelihood of adverse health effects caused by a hazard into categories of unlikely, likely and very likely. Due to the gaps in data and information being unevenly spread across hazard/commodity pairs, it was necessary to employ a degree of expert opinion/judgement in the likelihood of illness ratings, to bridge the gap between what is indicated by the data and what is plausible, given our knowledge of the hazard, the seafood commodity, its regulatory environment, and its production and processing supply chain up to the point of consumption.
Table 4: Ranking of food-borne hazards by likelihood of adverse health effects

Likelihood	Factors influencing estimation of likelihood of adverse health effects
Unlikely	Little or no evidence that the hazard has caused food-borne illness in Australia of overseas Limited consumption of the commodity by the general population, or consumption primarily by selected sub-populations, and/or Limited or no data demonstrating presence of the hazard in seafood.
Likely	Limited evidence that the hazard has caused food-borne illness in Australia or overseas Eaten periodically Availability of data demonstrating the presence of the hazard in seafood, and/or Availability of evidence from other data sources, for example, Imported Foods Inspection Program, FSANZ recall database, environmental surveillance, etc.
Very likely	Evidence that the hazard is associated with reported incidents of food-borne illness in Australia Widely and/or frequently eaten by the general population Availability of data demonstrating the presence of the hazard in Australian seafood, and/or Availability of significant evidence from other data sources, for example, Imported Foods Inspection Program, FSANZ recall database.

Risk ranking matrix

Estimates of the severity of adverse effects caused by the hazards encountered in seafood, and the likelihood of those effects occurring in the general population – defined as the population of generally healthy adults, irrespective of whether or not they consume seafood – are brought together to provide a ranking of the relative risk associated with the particular hazard/commodity combination using the matrix presented in Table 5.

This approach is consistent with the Codex recommendation that a risk characterisation integrate all of the qualitative and quantitative information available into a soundly-based risk estimate, and mirrors the Codex definition of risk as being a function of the probability and the severity of an adverse health effect. Using the matrix, the relative risk assigned to each commodity or group of commodities is qualitatively ranked as high, medium or low. Each broad risk ranking category can arise from three possible combinations of severity and likelihood estimates, as seen in Table 5.
In some susceptible sub-populations (for example, the very young and old, the immunocompromised, and pregnant women and their unborn children), certain hazards may cause illness of greater severity, or illness may occur at a lower infectious or toxigenic dose, leading to a ranking different to that obtained for the general population. In such cases, the relative risk rankings have been explicitly identified for both general and susceptible populations, to guide subsequent development of risk management strategies. However, the overall relative risk ranking for each seafood commodity or group of commodities in this report is a ‘general population’ risk estimate.
Table 5: A likelihood/severity matrix for ranking food safety risks in seafood

		likelihood of illness
		unlikely	likely	very likely
severity of illness	moderate	low	low	medium
	serious	low	medium	high
	severe	medium	high	high