3.6 Comparison of border cases among major importers

Each year, thousands of tonnes of products are detained, rejected or destroyed at national borders. The preceding sections have detailed the facts behind border cases in recent years for the three most important importing regions in the world, namely the European Union, North America and Japan.

As the data indicate, there are major differences between the importing regions studied, both in terms of numbers and nature of cases. It is important however to realize that, beyond shear numbers, the type of border case (safety, quality or economic fraud) and its direct macro and microeconomic impacts are different and need to be taken into consideration when comparing the different cases and strategies to reduce them. Unfortunately, the available data do not always enable this type of refined analysis. Recommendations are provided later as to how data collection and dissemination can be modified to improve the analysis.

It should be also noted that beyond differences in control systems, standards and analytical techniques, importing countries may devote different human and financial resources to border controls. This is likely to play an important role in the efficiency of border controls. Unfortunately, this information is not available to enable refinement of the analysis when comparing the different importing control regimes.

3.6.1 Relative frequency of border cases

Figure 1 shows a quite dramatic difference in the absolute numbers of border cases in the various importing countries/regions when shown relative to import quantities. These highlight some important differences, even though it is not possible to determine the absolute quantities of border cases per unit weight of imports.

At first glance, the United States of America has around ten times as many border cases per 100 000 tonnes as the European Union or Japan, and 3-4 times as many as Canada. This should not be taken to indicate necessarily that the United States have a higher performance in border controls or that products exported to the United States have more non conformity problems.

In fact, the data need to be adjusted and substantiated to enable comparisons of performance between the regions studied. Firstly, a high percentage of United States cases end up with the product actually entering the United States after re-examination, sorting, re-packing, new documentation and information or new labelling. During 1999-2001, 78 percent of detained shipments were released for import into the United States (Allshouse et al., 2003). Therefore, only around 22 percent of the cases in the United States can be considered in comparing the different regions as the other 78 percent ended up being accepted. Taking this into account, the United States has now around twice more border cases than the European Union and Japan and 60 to 80 percent as many as Canada (see Figure 1, adjusted United States graphs).

Secondly, the other countries/regions, especially the European Union, use some sort of "prevention at source" approach. Indeed, Chapter 2 explains that the European Union relies on national Competent Authorities (CAs) in exporting countries to examine establishments and products to assess their conformity to European Union requirements prior to shipments. By so doing, several non conformity cases are detected and stopped in the exporting countries. This approach has proven to be more preventative and cost effective than relying only on controls at the border. But it can also be penalizing for well managed seafood companies that cannot export to the European Union because they are in a country that does not have the resources and the capacity to put together a competent authority that meets the European Union requirements. Likewise, Canada and to some extent Japan, have adopted a "prevention at source" approach, although less formalized and less actively in comparison to the European Union. Canada has developed MOU/MRAs with a limited number of countries - Australia, Ecuador, Iceland, Indonesia, Japan, New Zealand, Philippines and Thailand, whereas Japanese importing companies have a long tradition of fielding quality controllers to work at the exporting sites. In both cases, some non conformity cases are eliminated before consignments are shipped.

More and more countries, including the United States (National Academy of Sciences, 2003), are advised to adopt a "prevention at source" approach because of its higher performance and cost effectiveness. However, care must be exercised to ensure that exporting developing countries are assisted in their efforts to build national capacity in safety and quality. This can only be a win-win situation both for the exporter and for the importer. While reducing safety and quality problems, their inherent costs and damages are reduced for exporters. At the same time, resources for control at borders are reduced significantly and target better problem cases, increasing their efficiency. Also, reducing losses due to rejections and detentions should result in greater supply of safe fish and less illnesses due to unsafe foods.

A third difference is the types and methods of control and standards applied at border. Chapter 2 shows that in the different importing countries studied, not only are border checks different (see Table 10), but the analytical techniques used can be different and the criteria or standards applied to judge conformity or non conformity are different (e.g. histamine, Salmonella, Listeria monocytogenes, etc.). Most importantly, these criteria and standards are not always based on fully fledged scientific risk assessments. Not only can this create arbitrary barriers to trade, but it is also costly as safe products may be refused in some regions while unsafe products may be distributed in others. Consequently, there is a need to harmonize the procedures and the standards between these majors markets, using risk assessment methodologies where applicable. This is further developed under section 3.6.7.

3.6.2 Border case patterns and trends

The previous sections broke down the causes of border cases into three main categories - microbial, chemical and other causes.

The breakdown for the four countries/regions covered in this publication is summarized in Figure 2. The differences in the profile of each country are quite obvious, with both the European Union and Japanese border cases being predominately microbial or chemical in origin, while these causes only account for a quarter to a third of border cases in the United States and Canada. The previous sections on each country detail the types of "other causes" for both Canada and the United States. Given the well publicized increase in 2001/2 of chemical (veterinary drugs) contamination from Asia (especially for shrimps), it is interesting to note that this becomes evident in the European Union data, where chemical contamination becomes a dominant category, while for other major importers, this trend is not noticeable. These other regions were importing large quantities of shrimp from Asia during this period, but were clearly handling the imported products differently, or recording the data differently.

Section 3.6.5 discusses in more detail the disparity between the data recorded for the causes of border cases, and the difficulty this poses for comparison between importing regions.

However, the obvious differences highlighted again point to the significant variations in approaches to controls at the borders of the countries being studied. For an exporter, it would be helpful if these procedures were harmonized, and that if they export a product, it should be treated the same way irrespective of who the importing country is. The extra costs imposed on traders by these differences may be significant, but are difficult to quantify due to the absence of relevant data, most importantly the quantities and value of rejected products and costs of controls.

3.6.3 Performance of continents

Again, we have a crude analysis here, but the results do provide a useful reference for discussion. The only two importing regions with full data over the four year period, 1999-2002, to allow comparison of the performance of exporting continents are the European Union and Canada. The Japanese data allow this comparison for the two periods 2000/2001 and 2001/2002 (Table 52). The underlying data are important and readers are referred to Table 23 for the European Union, Table 40 for Japan and Table 50 for Canada.

It is first apparent that there are some significant differences in the "relative" performance of the exporting continents dependent on whether fish is being sent to the European Union, Canada or Japan. This fact alone is worthy of comment. There are two main reasons why this might occur. One, the importing regions apply different criteria for border actions (whether sampling frequencies, limits for contamination levels or other procedures) and/or the exporting continents send different volumes and products (either different risk categories or of varying quality) to the export markets.

TABLE 52
Performance of continents in exporting to the European Union, Canada and Japan

* Central and South.
** 2001 detention figures used are an average 12 month period in Apr 2000-Oct 2001, 2002 figures are from Nov 2001-Oct 2002. See Table 40.

If the latter is the case, and given that the product forms are fairly similar (frozen fish dominates, significant numbers of crustacea, cephalopods, molluscs, etc.) for the European Union and Canada, it would seem that individual exporters must recognize the differences and target their products to suit the market criteria. This certainly does happen, but it is probably more likely that importing regions treat the imports (as a whole) in different ways resulting in different border actions. In the case of Japan, it is possible that the higher risk products that Japan imports may have originated from its neighbouring countries, as the species are similar. However, this is conjecture with the data available.

Being more specific, Oceania ranks as the best exporting region when exporting to the European Union, but ranks very poorly when exporting to Canada and Japan. Africa is the poorest performer in exports to Canada and second poorest to the European Union, however performs quite well in exports to Japan. The poorest performer by some margin in exporting to the European Union is Asia, exacerbated in later years with the veterinary drug issue alluded to in earlier sections. It is also the poorest performer in exports to Japan. However, Asia outperforms both Oceania and the European Union in exporting to Canada, though still only performs moderately. Central and South America perform very well in exports to Canada but less well when exporting to the European Union and Japan. North America is consistently a top performing exporter.

It is not easy to determine the significance of these differences and their causes. It was noted earlier that there seemed to be a tendency for those exporting the least amounts relatively to have more border cases per unit volume, and this certainly applies in the case of exports to Canada, though not in order. However, this does not apply to the European Union, as Oceania is the smallest exporter but is one of the top performers, or Japan, as Asia is the largest exporter, but is a poor performer.

It may be misleading to research in more detail as to why these differences occur, mainly because the importing nations use different procedures (sampling plans, analytical techniques, type of defect) and/or criteria on imports and the products exported differ from one importing region to the other. Again, for the benefits of international trade and ultimately the consumer, it is desirable that the importing rules are harmonized both in terms of the governing legislation and in implementation to enable proper evaluation of performance.

3.6.4 Access to data for border cases

The ease of access to data is variable. Three of the countries/regions in this study hold individual border case information on the internet, making access to data easy, as long as users have Internet access (see section 3.6.6). This initiative is to be applauded, and more countries should follow this example, though the type of information available on the three Web sites was variable. Japanese data is available on a Web site, but it is only held as annual summary tables at a macrolevel (main food commodities and exporting regions) and cannot be queried to individual border actions.

Table 53 compares the border case information available on the individual Web sites for the European Union, the United States and Canada.

As can be seen, the data available are similar. A notable exception is that the European Union reports do not include company details, whereas both the United States and Canadian sites do include these details. Also, the European Union records the importing country (of course, this is irrelevant for Canada or the United States). This information is useful for exporters, as there are differences in the way in which European Union member states perform some border controls for imports of fish and fish products. This has been mentioned earlier.

Missing data from all the Web sites includes Latin names (for identification), actions taken (e.g. re-exported, destroyed, reworked) and, importantly, the quantity and values of the lots in question. The latter data would allow calculations for the costs involved at both the country level and company level. This data would be very useful for policy makers, and for industry also. Presently, most of the economic studies costing implementation of safety and quality requirements can estimate fairly accurately the cost of appraisal (inspection and inspection management, analysis), the cost of prevention (training, maintenance, validation) but lack the data to assess the cost of failure (detention, rejection, scraps, re-work, destruction). The true performance of safety and quality assurance systems can only be seen in light of costs (appraisal, prevention) and benefits (reduction or elimination of failure).

TABLE 53
Border case data available on the World Wide Web

* sometimes this is not specified, e.g. when "all products" is used.; Ö yes. ×: no

Another key field missing in all data is the method of production. This is important as it is useful to be able to differentiate between farmed and wild fish in international trade and therefore in border actions where trade is interrupted (see also section 3.6.9). It is very likely that the causes for border cases will be different, e.g. veterinary drugs for farmed fish and histamine for wild fish (scombroids). In looking at border case data as potential lessons to be learned, it would be useful to provide specific recommendations to fish farmers and to fishermen based on actual border cases.

3.6.5 Type of data recorded for border cases

It is worth looking at the type of data recorded in more detail. Of course for the date and country fields^[15], the data kept are going to be very similar in each database of records (United States, Canada, European Union, etc.). For instance, if Australia is the exporting country this will be noted in the same way in each database. Similarly the date will be understandable in all databases (though the United States and European models are different for date format, but this is a minor issue).

However, and more importantly, the data recorded for the cause of the border case are different between countries/regions, and this makes it much more difficult to make comparisons at both the micro- and macrolevels. The European Union and Canada use free text entries for the cause while the United States use a coding system, choosing from one of over 170 codes to detail the reason why the product was detained.

Given the wide acceptance of HACCP based principles in ensuring food safety worldwide, it would make sense to adopt the approach used in hazard analysis using microbial, chemical and physical hazard categories for the categorization of the causes for border cases. Given the preponderance of other causes outside of the pure safety hazards in border cases, then an "other causes" category could be added, or could be further sub-divided into, say, documentation, labelling, and so on.

Considering the "cause" data in more detail, there are also differences seen in causes of border cases. For example, in microbial terms, specific (and normally pathogenic) bacteria are commonly cited in European Union border case records as the reason for an action taken at the borders, but a lot less so in Canada and the United States and not all in Japan, where indicator organisms are used for taking actions on suspect consignments. Also, there is a preponderance of more rapid and cheap tests e.g. sensory tests in Canada and the United States, rather than analytical tests with definable limits as used in the European Union (e.g. heavy metals, microbial limits). In a very specific example, the can integrity test is used systematically at the borders in Canada but not anywhere else and, not surprisingly, can failure is often cited as a problem at Canadian borders, but not elsewhere.

There is a separate, but important, issue to the use of inspection to control food imports and that is the argument about whether control systems should be limited to safety issues only, and that other quality issues (where public health is not in question) should be left to the market place. Here is not the place to decide upon this issue, but it is an issue that needs to be addressed to provide consistency between importing regions.

Similarly, the species and products are detailed in different ways, using different classifications. For instance, the European Union combines the product, species and cause into one field, while the United States and Canada split the product/species and cause fields, though the product and species fields are combined (Table 54). It would be useful, at little or no extra work when compiling, to record these data separately, i.e. fields for product, species and cause, and to harmonize products and species under defined categories allowing easier comparisons across importing regions. It is also advisable to harmonize this data with import and export data categories, so that relative importance of border cases can be easily obtained. The next chapter will make further recommendations about these and related issues.

TABLE 54
Recording of cause, product and species data in European Union, the United States and Canadian Web sites

TABLE 55
How border case data are held on the World Wide Web

PDF - Portable Document Format. Readable with free Adobe Acrobat software (www.adobe.com).
HTML - Hyper Text Mark up Language. The method used to layout Web pages.

3.6.6 Ease of analysis of data for border cases

Having said that the information is available on the Web sites, the ease with which this information can be manipulated and analysed is different. Table 55 summarizes the way the data are held on the four Web sites. The Canadian reports are by far and away the easiest dataset to transfer to spreadsheets for later analysis, as the data can be easily cut and paste into spreadsheets, and be immediately available for sorting and other database operations. Thus, to compile all the available data from the Canadian Web site into a spreadsheet would take minutes rather than hours. Also, fish data can be pre-selected on the Canadian Web site, separating these data from data on other foods. This is useful.

The data from the European Union and United States Web sites are not so readily usable. To transfer European Union data into a spreadsheet requires some further processing, as the text tool used in the PDF file will copy the data, but pasting into the spreadsheet creates mixed columns, which then need to be re-aligned properly under the headings. This takes some time. Also, non-fish records need to be removed.

The United States data are even more problematic, as the data will only allow cut and paste into a spreadsheet on a word by word (or phrase by phrase) basis, taking some minutes for each record. This would make it a very laborious task to convert even a months worth of records, and would increase the chances of human error in the transfer process. However, fish data can be pre-selected on the Web site, so these data come clean of data on other foods.

Japanese data are restricted to annual summaries (available only for 2000 at the time of writing) of detention data, and is across all foods sources (livestock, marine products, etc.) and does not break down data further to seafood detentions. This is a further area that needs to be harmonized for importing countries.

3.6.7 Requirements for harmonization and equivalency schemes

The present study shows that "the prevention at source" approach is not generalized and that all the major importing countries rely on end product control at the borders despite the deficiencies related to end product sampling and analyses.

Limitations of end product sampling and analyses

The following limitations have been reported for end product control methods (Huss, Ababouch and Gram, 2004):

• The chances of finding a hazard will be variable, but most often very low as explained later. Nevertheless, the laborious and costly work of sampling and testing will give a sensation of "being in control" and create a strong but false sense of security.

• It may take several days before results from end-product testing are available. This makes the method inapplicable for fresh fish and short shelf life products.

• The results are retrospective - if hazards are identified in the end-product testing programme and product needs to be destroyed, then there is significant loss as the production costs and expenses have already been incurred.

• It is costly. Decently equipped laboratories are needed as well as trained personnel. The running costs of a laboratory are high, as are often the costs of products "lost" to testing.

In most cases, there is no test that gives an absolutely accurate result with no false positives and no false negatives. This is certainly the case for many microbiological testing methods.

Furthermore, there are the principles of sampling and the concept of probability to consider. Indeed, the number, size and nature of the samples taken for analysis greatly influence the results. In some instances, it is possible for the analytical sample to be truly representative of the "lot" sampled. This applies to liquids such as milk and water. However, in cases of lots or batches of food such as seafood, this is not the case, and a food lot may easily consist of units with wide differences in microbiological or chemical quality. Even within the individual unit (i.e. a retail pack), the presence of a hazard (pathogen or toxic chemical) can be very unevenly distributed, and the probability of detecting the hazard may be very low. For example, it has been estimated (Mortimer and Wallace, 1988) that for a heterogeneously distributed contamination by Salmonella (at a rate of 5 cells/kg and assuming that the contamination is restricted to 1 percent of the batch), the probability of detecting the hazard by taking 10 samples of 25 g would be lower than 2 percent. This assumes 100 percent effectiveness for the detection test and most are less than 90 percent.

Therefore, even the most elaborate sampling and testing plans of end-products cannot guarantee safety of the product. There is no way to avoid some degree of risk and error in each acceptance and each rejection of lots unless the entire lot is tested, in which case no edible seafood will be left. This is obviously not acceptable both from practical and economic points of view. More worrying is the sense of false security it creates.

What is needed is to promote wider application and recognition, through equivalence schemes, of the "prevention at source" approach, anticipating safety hazards and building safety into the product and the food chain right from the start.

Lack of harmonization of control methods, criteria and standards

Border case data and epidemiological data have indicated that the major safety concerns involving fish and seafood are bacterial pathogens and chemical toxins or contaminants. Yet, there is a major discrepancy between the major importing countries as to how to stop these undesirable pathogens and chemicals from entering the seafood or crossing borders. These discrepancies persist despite the fact that they have been recognized for many years. The following are most relevant to international fish trade and illustrate the magnitude of the discrepancies.

Histamine: Histamine poisoning is a food-borne chemical intoxication occurring a few minutes to several hours following the ingestion of foods that contain unusually high levels of histamine.

Histamine poisoning occurs throughout the world and is perhaps the most common form of toxicity caused by the ingestion of fish. Japan, the United States of America and the United Kingdom are the countries with the highest number of reported incidents, although this possibly implies better reporting on their part. Less frequent incidents have been reported elsewhere in Europe, Asia, Africa, Canada, New Zealand and Australia (Huss, Ababouch and Gram, 2004).

There is uncertainty regarding the threshold toxic concentration of histamine because potentiators of toxicity, such as cadaverine, putrescine and spermine may be present in fish and lower the effective dosage compared with pure histamine. Different fish could contain different potentiators, and the levels of potentiators could also vary considerably from one individual fish to another.

A review (Shalaby, 1996) of the oral toxicity to humans of histamine and other biogenic amines in foods concluded that histamine-induced poisoning can be considered, in general, slight at 8-40 mg/100 g, moderate at > 40 mg/100 g and severe at >100 mg/100 g. Based on an analysis of poisoning episodes, the review suggested the following guideline levels for histamine content in fish:

- < 5 mg/100 g (safe for consumption)
- 5-20 mg/100 g (possibly toxic)
- 20-100 mg/100 g (probably toxic), and
- >100 mg/100 g (toxic and unsafe for human consumption).

Because of the recurrence of histamine poisoning in many parts of the world and the importance of international trade of the concerned fish species, many countries have imposed maximum limits or produced guidelines on histamine levels in traded fish. But these limits are not harmonized and none have been based on a thorough risk assessment.

Thus, the US FDA guidelines, established for tuna, mahi-mahi and related fish, specify 50 mg/100 g (500 ppm) as the toxicity level, and 5 mg/100 g (50 ppm) as the defect action level because histamine is not uniformly distributed in a decomposed fish. Therefore, FDA considers that if 5 mg/100 g is found in one section, there is a possibility that other units may exceed 50 mg/100 g. FDA requires the use of the Association of Official Analytical Chemists (AOAC) fluorometric method.

The European Union requires Competent Authorities to take nine samples from each batch of fish species of the following families: Scombridae, Clupeidae, Engraulidae and Coryphaenidae. These samples must fulfil the following requirements:

- the mean value must not exceed 10 mg/100 g (100 ppm)

- two samples may have a value of more than 10 mg/100 g (100 ppm) but less than 20 mg/100 g (200 ppm)

- no sample may have a value exceeding 20 mg/100 g (200 ppm).

However, fish belonging to these families and which have undergone enzyme ripening treatment in brine may have higher histamine levels but not more than twice the above values. For example, in salted anchovies, a major traded commodity, European Union accepts a mean value as high as 200 to 400 ppm (instead of 100 to 200 ppm required for non ripened products). Examinations must be carried out in accordance with reliable, scientifically recognized methods, such as high-performance liquid chromatography (HPLC).

In Australia and New Zealand, the level of histamine in a composite sample of fish or fish products, other than crustaceans and molluscs must not exceed 10 mg/100 g (100 ppm). A 'composite sample' is a sample taken from each lot, consisting of five portions of equal size taken from five representative samples. This clause, which came into force in October 1994, was under review in 2002, with a proposal to increase the maximum allowable level of histamine in fish and fish products to 20 mg/100 g (200 ppm).

In Canada, the level of histamine in enzyme-ripened products (e.g. anchovies, anchovy paste, fish sauce) should not exceed 20 mg/100 g. For all other scombroid fish products (e.g. canned or fresh or frozen tuna, mackerel, mahi-mahi), samples are collected according to sampling plan 1 (AQL 6.5) for inspection. Any sample exceeding 50 mg/100 g will result in the lot being rejected with no right to re-inspection. The acceptance number is that corresponding to the number for decomposition.

Salmonella: Many countries, especially the major seafood importing countries, view the presence of Salmonella in raw frozen fish and crustacea as a form of adulteration, based on the fact that species of Salmonella are not usually found in clean marine environments and would only be found in products which have been exposed to poor standards of hygiene during handling and processing.

However, more and more fish and crustacea are produced by aquaculture. Aquaculture practices in many countries, especially Asian countries which produce almost 90 percent of world aquaculture fish and crustacea, involve pond fertilization with chicken and animal manure which are a source of faecal organisms. Environmental conditions in fish ponds in the tropics are conducive for growth and proliferation of bacteria such as Salmonella.

Likewise, many studies (Reilly, Twiddy and Fuchs, 1992) have shown that Salmonella and other enterobacteria can be present as part of the natural bacterial flora of water ponds and that specific serotypes of Salmonella can be frequently isolated from cultured shrimp. Usual processing of raw fish and shrimp such as washing, grading, chilling and freezing will not eliminate Salmonella if it is naturally present in cultured fish or shrimp. However, cooking quickly destroys this pathogen.

It is therefore legitimate to question whether a zero tolerance for Salmonella in aquaculture shrimp and fish is justifiable and useful for health protection. A risk assessment will help clarify the issue and remove any injustifiable barrier to the ever increasing trade of aquaculture shrimp and fish.

Vibrio species: Vibrio species are typical of marine and/or estuarine environments and are commonly isolated from fish and crustacea. Most of the species are mesophilic and their numbers tend to increase during the warm seasons. The genus comprises 34 species of which 13 species can cause human disease. Seafood-borne diseases are primarily caused by Vibrio parahaemolyticus, Vibrio vulnificus and Vibrio cholerae. These pathogenic Vibrio spp. are ubiquitous in warm (>15 °C) seawater environment. They can be found at levels of up to 10²-10³ cells/g in shellfish and up to 10⁴-10⁸ cells/g in the intestines of shellfish-eating fish. They are indigenous to the aquatic environment and their presence and numbers are influenced by factors such as temperature, salinity and algal density.

The major importing countries use different Vibrio standards, ranging from absence of V. cholerae (United States of America and Canada) and Vibrio vulnificus (USA) in ready to eat seafoods, to <100/g V. parahaemolyticus in cooked crustacea (European Union) or (Japan) (RTE) to < 10⁴ (FDA) in RTE seafood (see Annexes A.4, A.7, A.10, A.14).

This again has created important trade flow disruptions despite repeated concerns raised by scientists regarding the subjectivity of these standards. For instance, the EC Scientific Committee on Veterinary Measures relating to Public Health (SCVPH) recommended that (EC 2001):

• The practice of judging seafood exclusively based on total Vibrio counts as indicative for the presence of pathogenic vibrios is not appropriate and should be discontinued.

• The practice of judging seafood exclusively based on total V. parahaemolyticus counts without consideration of the virulence factors TDH/TRH (or tdh/trh) is not appropriate and should be discontinued.

• Currently available scientific data do not support setting specific standards or microbiological criteria for pathogenic V. parahaemolyticus and V. vulnificus in seafood. Codes of practice should be established to ensure that GHP has been applied.

Regarding cholera, an ongoing FAO/WHO risk assessment on Vibrio spp. in seafood addresses the issue of Vibrio cholera in warm water shrimp imported by United States of America, Europe and Japan. The justification for taking up this risk assessment is that several millions of tonnes of warm water shrimp are traded annually and this trade is generally adversely affected whenever there are outbreaks of cholera in shrimp-producing tropical countries.

The hazard identification concluded that V. cholerae is a heterogeneous species comprising of more than 200 serotypes. Of these only serotypes O1 and O139 are known to cause cholera. Non-O1/non-O139 serotypes are rarely associated with the sporadic cases of gastroenteritis. Therefore the agents involved in cholera need to be clearly identified as choleragenic V. cholerae.

Furthermore, a series of studies conducted in several countries in Asia during the late 1980s reported an absence of choleragenic V. cholerae in warm water shrimp, making it difficult to predict the distribution of choleragenic V. cholerae in warm water shrimp. On the other hand, frequent testing is done on warm water shrimp at the port of entry in importing countries. The FAO/WHO risk assessment team considered this data. Over 20 000 samples were tested in Japan during 1995-2000. Data on 181 samples were available from the US FDA and findings from a survey of 752 samples were available from Denmark. Of the total of 21 857 samples tested only two samples imported into Japan from India were positive for choleragenic V. cholerae.

This risk assessment has concluded that even if all imported shrimp is consumed without any further cooking, the risk of cholera is about 2-4 cases in 100 years. This is an over-estimate because (a) imported volumes has been taken as edible volume and (b) it is common that shrimp are generally consumed after cooking - this would reduce the bacterial numbers by greater than six logs. Thus the risk would be very low or near to zero. This inference is supported by epidemiological data. Cholera is a reportable disease and good surveillance mechanisms exist in most developed countries importing warm water shrimp. The data show that there are no cases of cholera reported due to imported shrimp in these countries.

Shrimp intended for export is a high value item. It is produced using GHP/HACCP in the producing countries. Adoption of such procedures greatly reduces the risk of contamination of shrimp with choleragenic V. cholerae, and the risk assessment confirms this.

Listeria monocytogenes: Owing to the widespread occurrence of L. monocytogenes, some experts consider that it is extremely difficult (and expensive) to produce ready-to-eat (RTE) foods, including RTE seafood such as smoked fish, without sporadic occurrence of the organism at low levels (FAO, 1999). The dose-response relationships (and resulting risk estimate) indicate that such low levels constitute a very low risk. Yet there is currently no international agreement on "acceptable levels" of L. monocytogenes in seafoods. Some countries, such as the United States, Austria, Australia, New Zealand and Italy, require the absence of L. monocytogenes in 25 g of seafood (referred to as zero tolerance). Other countries (Germany, Netherlands, France) have a tolerance of < 100/g at the point of consumption. Others (Canada, Denmark) have a tolerance of < 100/g for some foods and zero tolerance for others - especially those with extended shelf lives and that can support the growth of L. monocytogenes. In addition, differences exist in the analytical methods adopted by different countries.

A recent FAO/WHO risk assessment on L. monocytogenes in RTE foods has prompted several countries to review food safety objectives regarding this bacterium. The risk assessment concluded that in the case of fish and fishery products, zero tolerance is not always appropriate. This work and other similar studies have led several countries to revise standards on L. monocytogenes in RTE foods. For example, the US FDA has recently called for public comments regarding requests that the agency establish a regulatory limit of 100/g for L. monocytogenes in foods that do not support the growth of the micro-organism.

Chemical contaminants: The present study shows that a major cause of border cases in 2001/2002 was due to chemical residues. This is because at the end of 2001 and during the first months of 2002 several control laboratories in Europe detected trace amounts of chloramphenicol and nitrofurans in imported animal products (e.g. shrimps and chicken). Following the safeguard provisions as foreseen in the European Union regulations for food imports of animal products, some producers and producing countries were temporarily withdrawn from the list of approved exporters and others were forced to rapidly implement drastic measures (e.g. analysis lot by lot). North America and Japan adopted similar controls, although using less sensitive analytical techniques.

In Europe, this increase was triggered mainly by improvements in analytical methods which significantly lowered the levels of detection for residues of these drugs. However, several producers and exporters argued that the products were not produced using these drugs, and that the trace amounts were at such low levels that they could not result from the illicit use of drugs but from environmental contamination. Some also argued that very low levels would pose no risk to consumers.

At the international level (FAO, 2004a), the Joint FAO/WHO Expert Committee on Food Additives (JECFA) is responsible for developing acceptable daily intake (ADI) values and maximum residue levels (MRLs) for veterinary drugs which are compatible with Good Veterinary Practices (GVP). These are adopted by the Codex Alimentarius Commission usually after one round of comments from member countries at the Codex Committee on Veterinary Drugs in Foods (CCRVDF). Due to the nature of toxicity (chloramphenicol, furazolidone) or lack of data (nitrofurazone), JECFA did not establish ADIs for these compounds. Consequently, CAC did not adopt MRLs for these compounds. Recently, JECFA re-considered chloramphenicol and concluded that there was no sufficient evidence that low level contamination of animal products could result from the occurrence in the environment.

The affected exporting countries were forced to take appropriate measures which include, for example, destruction of animal products, tighter controls on the use of illegal drugs, high investments into modern analytical equipment, and training of laboratory personnel. Most countries could meanwhile resume their exports.

It is important to distinguish between zero tolerance and a de minimis limit. The term zero tolerance is used for residues of substances which are considered to be unacceptable at any concentration, whereas concentrations below de minimis limits constitute only a theoretical risk that can be ignored (i.e. the presence below de minimis levels is acceptable).

While development of analytical chemistry continues to lead to significant improvements in the sensitivity of methods applied by research and control laboratories (for chloramphenicol, the limit of detection has decreased since 1970 by five orders of magnitude), problems continue to exist and the impact on trade is considerable. For example, it was recently revealed that trace amounts of a certain substance that had initially been identified as one of the nitrofuran metabolite markers indicating drug use had instead originated from other food ingredients, for example, flour and the packaging of foods.

It should be recognized that at the very low levels at which limits of detection and quantification (LOD/LOQ) are set, the uncertainty of analytical results increases. This needs to be respected when such results are communicated.

More importantly, the increased sensitivity of analytical methods raises the probability of finding trace amounts of substances that may originate from other routes than the administration of veterinary drugs to animals. Such routes could be environmental contamination, cross-contamination at the feed mill or the farm, or contamination from other sources like ingredient or packaging. An inventory of such substances without an ADI/MRL that potentially could cause problems in trade and are used under conditions of good veterinary practices is needed. For such compounds the nature of the existing data and the potential gaps in the data bases supporting their use should be identified and discussed.

Finally, modern analytical equipment is very expensive and requires resources and considerable theoretical knowledge and practical expertise which are not created overnight. The use of modern analytical methods puts considerable burden on the shoulders of control laboratories of exporting countries. Ways and means for proactive capacity building should be considered, instead of the current reactive approach for capacity building which starts after products have been rejected.

Conclusions

It is now universally agreed that food standards and control systems should be scientifically-based using risk assessment methods. While this approach has been used for some time for setting the MRL of pesticides, chemical contaminants and additives, it is relatively new for biological hazards and these unfortunately represent a major concern in fish trade. In fact, international guidelines are still being developed and only a few countries, with significant scientific and financial resources, have been able to initiate fully fledged food microbiological risk assessments. Of these, only a few deal with seafood hazards. To have international value, these risk assessments will need, where appropriate, to incorporate data from different countries and regions where the fish species concerned are produced, traded and/or consumed. Unfortunately, developing countries, which lack the necessary human and financial resources, are not able to contribute adequately despite their important role in international fish trade.

To fill this gap, the thirty-second session of the CCFH identified, in 1999, a list of pathogen-commodity combinations that require expert risk assessment advice. In response, FAO and WHO jointly launched a programme of work. The ad hoc Joint Expert Meetings on Microbiological Risk Assessment, JEMRA, have the objective of providing expert advice on risk assessment of microbiological hazards in foods to their member countries and to Codex. This involved the implementation of a number of activities including the establishment of expert drafting groups to examine four of the 21 pathogen-commodity combinations identified in 1999 as priority issues (Listeria monocytogenes in ready to eat foods, Salmonella spp. in broilers and eggs, Campylobacter in poultry and Vibrio spp. in seafoods). The Salmonella and Campylobacter risk assessments have been recently finalized and the others are in the final stages. Among the other 17 pathogen-commodity combinations identified as priority issues, only Salmonella in fish is relevant to fish trade. Biotoxins in bivalve molluscs have been addressed recently through an international expert consultation called for by the CCFFP. The consultation completed risk assessments regarding the various biotoxins and its main recommendations will be debated during the next session of the CCFFP in 2005. Other similar mechanisms therefore need to be initiated to assess the risk of other pathogenic agents such as histamine, heavy metals, viruses and parasites relevant to fish trade and consumption. But important resources need to be mobilized for this to take place.

In the meantime, the major trading countries/regions are using microbiological criteria as interim measures. These are presented in various annexes to this study. Annex A.4 presents the draft EC regulation on microbiological criteria for foodstuffs that has been in development since 1999. Once approved, these will be used as interim measures awaiting formal risk assessments.

Finally, the present study seems to indicate that control at the borders, despite its limitations and doubtful performance, is given more priority over the "prevention at source" approach using GHP/GMP and HACCP based systems. These systems are presently widely recognized and industry has experimented with them and enjoyed their perceived benefits. Therefore, it seems appropriate that trading partners should be collaborating to build equivalent schemes based on the HACCP approach, using appropriate safeguards to protect consumer health. This way, the resources presently spent on border controls can be used more cost effectively.

3.6.8 Trade and economic implications

While international efforts are focussing on harmonization, several development agencies and donors have been exploring ways and means, both financial and technical, to assist developing exporting countries build national and regional capacity to meet international safety and quality standards. Proper assessment of the extent of assistance needed is key in decision making. Therefore costing the impact of substandard quality and safety products would be of interest not only to producers, processors, quality control authorities and consumers, but also to governments, donors, public health authorities and development agencies. In addition to the economic losses incurred because of fish spoilage, product rejections, detention and recalls and the resulting adverse publicity to an industry and even to a country, fish-borne illnesses cost billions of dollars to the community because of their costly adverse health effects, the loss of productivity and the medical expenses.

Furthermore, risk managers who will be weighing different mitigation options need economic data to assess the cost effectiveness of the different options presented to them. Unfortunately, the detention/rejections data, as they are collected, cannot be exploited to assess the cost of border cases. It is important to have access to such information in future for the reasons mentioned above.

The few studies which have looked at the economic impact of detentions and rejections have addressed it both at the macro- and microlevels. At the macroeconomic level, analysis focused mostly on estimating the economic costs of trade flow disruption cases. Whereas, at the microeconomic level, studies addressed mostly the implementation and maintenance costs of HACCP-based systems and maintenance. Fewer of these studies used the PAF (Prevention-Appraisal-Failure) model.

Macroeconomic level

Macroeconomic studies carried out by FAO (Cato et al., 1998) and others (Casewell, 2001; Allshouse et al., 2003) reported on specific trade flow disruptions that gave rise to international disputes over seafood safety and affected trade opportunities for producers, exporters and importers with the resulting economic impact. These cases have been discussed as to their relevance and technical and scientific justification in section 3.6.7. Following are key examples to illustrate the macroeconomic impact.

In 1997, the European Commission (EC) banned shrimp imports from Bangladesh because processing plants in Bangladesh did not meet EC standards. The estimated net cost of this August-December 1997 ban after considering shipments diverted to other countries was US$14.7 million to the Bangladesh frozen shrimp processing industry. As in many other less developed countries (LDCs), many plants in Bangladesh had difficulty meeting the required quality and safety standards because of a lack of sufficient funds to invest in quality control measures, more adequately trained staff, and expensive equipment. The Bangladesh Department of Fisheries, Fish Inspection and Quality Control had verified and certified compliance for only 20 percent of the seafood processing companies that previously were shipping to the European Union (EU). This ban affirms the apprehension of some LDCs that evolving standards can be a major market access issue.

During the period 1997-1999, Kenya and some other countries surrounding Lake Victoria have faced a series of food safety related restrictions of their fish exports. Salmonella contamination in Nile perch from Kenya in April 1997 led to border testing of all Nile perch consignments. Later, a cholera epidemic in East Africa in December 1997 resulted in a European Commission ban of imports of fish products from Kenya, Mozambique, the United Republic of Tanzania and Uganda until June 1998. The World Health Organization and FAO issued statements that the ban was not scientifically justifiable and the restrictions were lifted in June 1998. For Mozambique alone, the ban resulted in a loss of US$60 000 in trade per month while the ban was in place, equating to about 30 tonnes of fish that were not exported to the European Union market. Following reports of pesticide contamination of fish from Lake Victoria, another round of restrictions began in April 1999 that prohibited all fish exports from Lake Victoria to the European Union. As a result of these events, employment in the sector declined and industrial fish processing companies reduced capacity or closed.

In January 2002, the European Union suspended shrimp and prawn imports (and other products of animal origin) from China because of residues of chloramphenicol, and because of general deficiencies in the Chinese residue control system. Chloramphenicol has been linked to fatal leukemia and anaemia in humans. At the same time, FDA stepped up surveillance for chloramphenicol residues and residues of other unapproved aquaculture drugs in shrimp and crayfish imports from all countries and modified its testing methods so as to be able to detect the antibiotic at 0.3 part per billion or ppb, equal to that of Canada and the European Union. Subsequently, products with detectable levels of chloramphenicol were refused entry into the United States, which temporarily suspended shrimp imports from China.

Although some of these seafood safety incidents appear to have resulted in relatively limited and short-term interruptions of trade and economic impacts, costs could continue to accrue from continued market diversions (i.e., lost market share), loss of momentum in the sector, decreased prices, and reduced capacity due to temporary or permanent plant closures. The above examples illustrate that food safety restrictions can act as barriers to trade as they can for any type of food. Despite the advantages of some developing countries in terms of preferential trading arrangements, food safety incidents can impose costly requirements on developing countries beyond their ability to afford compliance.

Microeconomic level

At the micro-level, the PAF model divides production costs relevant to quality and safety into three categories:

- prevention costs are the costs of any action taken to investigate, prevent or reduce defects and failures. They include the costs of planning and documentation, training, maintenance, personnel incentives;

- appraisal costs are the costs of assessing and recording the quality achieved. These are generally the easiest to measure and include: costs of inspection and control of raw materials, ingredients and packaging, costs of in-plant process inspection, laboratory costs and recording costs;

- failure costs are the costs arising from failure to achieve the quality specified. They can be divided into internal and external costs, depending on whether they are produced within the plant or after the transfer of product ownership to the customer. Internal failure costs include scraps, reprocessing, additional laboratory analysis, extended cold storage, low yield. External failure costs include product rejection, detention, recall, liability, bad publicity, etc.

The PAF model theory demonstrates clearly that the failure costs decrease significantly with an increase in prevention and appraisal expenditure. But, for each process and situation, there is a point at which total quality and safety costs will be at their optimum and any extra expenditure in prevention and/or appraisal will not bring additional improvement.

Public health managers have been more specifically interested in safety implications and their costs for public health. They studied food safety valuation by attempting to measure the effectiveness levels of a food safety public programme and the extent to which this programme achieved its goals. An example would be to measure the effectiveness of a programme designed to educate consumers on the safety of seafood, or of nutritional attributes of seafood, or the cost and benefits of a seafood HACCP programme.

In this respect, two estimation methods have been used: Cost of illness (COI) and the willingness to pay (WTP) method. The COI approach estimates the resources that society will save by avoiding food-borne illness. Social costs include costs to individuals, industry costs and public health surveillance costs. Costs to individuals can be measured through documenting medical costs, income or productivity loss, pain and suffering, leisure time costs, child care costs, risk aversion costs, travel costs, and vocational and physical rehabilitation costs, among others. Industry costs include product recalls, plant closings and cleanups, product liability costs, reduced product demand and insurance administration. Public health surveillance costs include disease surveillance costs, costs of investigating outbreaks and costs of cleanup. The WTP method actually measures peoples' willingness-to-pay for the reduced risk of death or illness in a specified population from consuming food. For example, FDA estimated benefits of implementing the HACCP programmeme for seafood ranging from US$1.435 to US$2.561 billion. This represents total discounted benefits beyond the fourth year after implementation using a discount rate of six percent. Benefits included those derived from safety (cost savings resulting from reduction in illnesses from a variety of hazards), nutrition, increased consumer confidence, expert advice and reduced enforcement costs.

Estimates from the present study

As mentioned earlier, the above-mentioned studies need to be expanded beyond country cases to assess the economic implications of border cases in the main markets. Following is an attempt to estimate the cost of border cases in Japan using data presented in Chapter 2 and available on the MHLW Web site. Unfortunately, similar data were not available from the other importing countries.

Fish and seafood detentions cases in Japan, posted on the Web for 2000, numbered 201 for a total of 445 tonnes, yielding an average of 2.2 tonnes/case. Assuming that for the period covered under the present study, the organizational structure of fish trade to Japan (types of products, container size, shipment routes, means, duration, packaging, etc.) have not seen major changes, this figure was used to estimate 2001 and 2002 volumes and costs of border cases for Japan as follows:

Estimated Volume (in tonnes) of border cases for each product category = Number of cases for the product category x 2.2 (t/case)

Table 56 estimates at 255.2 tonnes and 490.6 tonnes the total volume of Japan border cases respectively for 2001 and 2002. These represent a small fraction (respectively 0.0083 percent and 0.016 percent) of total imports to Japan in 2001 and 2002. They were valued at US$1 159 870 and US$2 230 465 (or 0.009 percent and 0.017 percent) of total import values respectively for 2001 and 2002. For the period 2001-2002, the average cost was estimated at US$4546 per ton detained and US$10 000 per border case.

These costs are much greater than the prevention costs that would have enabled the concerned companies to avoid these border cases. This is confirmed by several studies, compiled by Cato (Cato, 1998), which estimated the costs of implementing GMP and HACCP. In the United States, 1995 cost estimates of HACCP implementation for seafood processing plants averaged US$23 000 the first year and US$13 000 per year the subsequent years. In parallel, prices for seafood were also estimated to increase by less than one percent in the first year and less that 0.5 percent in subsequent years with the larger cost increase expected to decrease consumption by less than 0.5 percent.

Other studies estimated the costs of implementing in the United States of America the HACCP-based Model Seafood Surveillance Programme (MSSP) in the crab industry at US$3 100 per plant or US$0.04 per kg, representing 0.33 percent of processor price. Compliance costs were estimated at US$6 100 per plant. Investment costs averaged US$3 200 for large plants and US$1 700 for small plants. All in all, added cost per kg of product for compliance was US$0.02 for small plants and insignificant for large plants. For molluscan shellfish (oysters, mussels, clams), these costs were estimated at US$5 500 per plant. Annualized compliance costs per kg were estimated at US$0.11 for small plants and US$0.01 for larger plants.

In Bangladesh, costs per kg for the shrimp industry were estimated between US$0.26 and 0.71 for upgrading the plant and implementing HACCP and between US$0.03 and 0.09 for its maintenance. Those were higher than the figures estimates in the United States, mainly because the Bangladesh shrimp industry had to start from scratch and also had more small and medium enterprises than in the United States. It is well established that economy of scale lowers the costs of safety and quality systems in large enterprises. But even though high, these costs represent only 0.31 percent (implementation) and 0.85 percent (maintenance) of 1997 price. (Cato and Lima dos Santos, 1998).

TABLE 56
Estimates of volumes and value of border cases for Japan

More importantly, these costs remain very low in comparison with the cost of border cases estimated in the present work at US$4.55 per kg. Indeed, the per kg costs of implementing and maintaining HACCP or HACCP-based systems would represent between 1.46 percent and 3.4 percent (United States of America) or 6.45 percent to 17.6 percent (Bangladesh) of the costs of border cases. Furthermore, and as stated before, these costs should be considered only as the visible part of the iceberg. The cost of transportation, the resulting adverse publicity, the requirements for systematic physical checks of subsequent shipments, the loss of clients confidence and ensuing market shares, market diversions, loss of momentum, decreased prices, reduced capacity due to temporary or permanent closures, are certainly additional costs with far reaching impact, but unfortunately difficult to quantify.

3.6.9 The case of aquaculture

Aquaculture has been the fastest growing food production sector in many countries for nearly two decades, with an overall growth rate greater than 11.0 percent per year since 1984, compared with 3.1 percent for terrestrial farm animal meat production, and 0.8 percent for landings from capture fisheries. The majority of the food fish production comes from land-based freshwater culture, and in some countries it exceeds that from freshwater capture fisheries.

Production in 2002 reached 51.4 million tons including aquatic plants, with 71 percent from China. Developing countries accounted for around 90 percent of production. All continents showed increases in production during 2000-2002 with the exception of Europe where production remained relatively unchanged (0.1 percent annual decrease).

The rapid growth in aquaculture production and trade has made the sector important to the economies of many countries, especially developing countries, both for food security and trade. Over the years, aquaculture products have helped to stabilize traded fish supplies and to bring down fish prices.

However, aquaculture products have been subject to close scrutiny for their safeness for consumption within international fish trade, i.e. the recent issue on veterinary drugs residues discussed in details previously in the present study. Likewise, aquaculture products are target for Salmonella reduction strategies and there is increasing concern about organic contaminants in farmed fish products. Indeed, the farming of fish high up in the food chain leads to a concentration of contaminants. For example, "fishmeal and fish oil were found to be the most heavily dioxin contaminated feed materials with products from European fish stocks contaminated more heavily than those from South Pacific" (EU, 2000).

Consequently, it would be very beneficial to examine border cases from the perspective of the production method (i.e. farmed fish or captured fish). Unfortunately, the data does not allow this either in the trade data or the border case data. The increasing importance of aquaculture should be considered a good opportunity for better control over the whole food chain. However, this is still not the case for several hazards, most notably the chloramphenicol issue with farmed shrimp. Clearly, the introduction of good aquaculture practices will certainly improve the performance in this respect and it would be very nice to support this with actual and verifiable border case data. This requires that we are able to differentiate the production methods in both trade and border case data. It is now mandatory in the European Union to inform the consumer on the label on the production method, but trade statistics do not differentiate aquaculture from capture fisheries yet.