Previous Page Table of Contents Next Page


Traditionally, three principal means have been used by governmental agencies and food processors to control microorganisms in food as listed by ICMSF (1988). These are (a) education and training, (b) inspection of facilities and operations and (c) microbiological testing. These programmes have been directed toward developing an understanding of the causes and consequences of microbial contamination and to evaluate facilities, operations and adherence to good handling practices. Although these are essential parts in any food control programme, they have certain limitations and shortcomings. The rapid turnover of personnel means that education and training must be a continuing exercise, which is rarely the case. As far as inspection of facilities and operations is concerned, this is often carried out with reference to various guidelines such as codes of practice, food control laws etc. These documents often fail to indicate the relative importance of the various requirements, and often these requirements are stated in very unprecise terms such as “satisfactory”, “adequate”, “acceptable”, “suitable”, “if necessary” etc. This lack of specificity leaves the interpretation to the inspector, who may place too much emphasis on relatively unimportant matters and thus increase costs without reducing hazards.

Microbiological testing also has some limitations as a control option. These are constraints of time, as results are not available until several days after testing as well as difficulties related to sampling, analytical methods and the use of indicator organisms. These problems will be discussed in more detail below followed by the description of a modified approach aiming at a preventive quality assurance programme.

Estimation of bacterial numbers in food is frequently used in the retrospective assessment of microbiological quality or to assess the presumptive “safety” of foods. This procedure requires that samples are taken of the food, microbiological tests or analyses are performed and the results evaluated - possibly by comparing with already established microbiological criteria. There are serious problems related to all steps in these procedures.


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 that can be sufficiently well mixed.

In cases of “lots” or “batches” of food this is not the case since a lot may easily consist of units with wide differences in the microbiological quality. A number of factors therefore must be considered before choosing a sampling plan (ICMSF 1986). These include:

A sampling plan (Attributes plan) can be based on positive or negative indications of a microorganism. Such a plan is described by the two figures “n” (number of sample units drawn) and “c” (maximum allowable number of positive results). In a 2-class attributes sampling plan, each sample unit is then classified into acceptable or non-acceptable. In some cases the presence of an organism (i.e. Salmonella) would be unacceptable. In other cases, a boundary is chosen, denoted by “m”, which divides an acceptable count from an unacceptable. The 2-class sampling plan will reject a “lot” if more than “c” out of “n” samples tested are unacceptable.

In a 3-class sampling plan “m” separates acceptable counts from marginally acceptable counts and another figure “M” is indicating the boundary between marginally acceptable counts and unacceptable counts.

The safety which can be obtained with such sampling plans depends on the figures chosen for “c” and “n”. This can be illustrated with the so-called operating characteristic curves which are demonstrating the statistical properties of such plans (Figure 4.1).

Figure 4.1.

Figure 4.1. Operating characteristic curves for different sample sizes (n) and different criteria of acceptance (c) for 2-class attributes plan (ICMSF 1986).

Figure 4.1 shows that the greater the number of defective units (Pd), the lower is the probability of acceptance (Pa) of that lot. It is further demonstrated that high value of “n” and low value of “c” reduces the risk of accepting lots with same number of defective units. However, even the strictest sampling plans used are no great assurance of safety. Following those sampling plans recommended for infant formulas (n=60, c=0) involve testing 1.5 kg of food and even then there is a 30% risk of accepting product with 2% of sample units contaminated with Salmonella.

It is evident that even the most elaborate sampling of end-products cannot guarantee safety of the product.

It might be argued that although sampling and examination of samples may provide little assurance, it is still worthwhile in situations where there is no jurisdiction over handling and processing practices (such as for lots presented for acceptance at ports of entry). Even if only a fraction of the substandard consignments are found, the psychological effect on exporting companies is high.

In order to increase the relevance of sampling and testing, the International Commission on Microbiological Specifications for Foods (ICMSF) has introduced the concept of relating the stringency of the sampling plan to the degree of hazard of the food (ICMSF 1986). Thus the hazard may vary from a condition of no health hazard but only of utility (case 1–3) through low indirect health hazard (case 4–6) to moderate (case 7–12) and severe direct health hazards (case 13–15). In case of moderate or severe hazards, a 2-class attributes sampling plan is normally used. When the health hazard is low and in application of microbiological guidelines, a 3-class plan is suggested. For example, atypical 2-class plan with n = 5 and c = 0 requires that 5 sample units be tested and the lot would be rejected if one of the five sample units was defective. Table 4.1 shows the sampling plans and recommended microbiological limits suggested by ICMSF (1986) for seafood products.

The sampling plans applied by the Food and Drug Administration (FDA) for seafoods have been discussed and evaluated by a large Committee on Seafood Safety (Ahmed 1991). It is concluded that these sampling plans provide relatively little safety to the public and that increasing the sample size is not a reasonable solution. Even if testing methods for dangerous microorganisms, toxins and contaminant chemical were fully available and completely reliable, it is very clear that statistical uncertainties associated with lot sampling makes this an unreliable method for ensuring safety of food products. It is finally recommended by this Committee (Ahmed 1991) that suppliers of seafood to U.S. should be required to employ a Hazard Analysis Critical Control Point system, in order to obtain a high level of assurance and real-time control at the processing level.

Table 4.1. Sampling plan and recommended microbiological limits for seafood (ICMSF 1986).
ProductTestCasePlan ClassncLimit per gram or per cm2
Fresh and frozenAPC113535×105107
fish; cold smoked fishE. coli435311500
Precooked breadedAPC23525×105107
fishE. coli535211500
Frozen rawAPC1353106107
crustaceansE. coli435311500
Frozen cookedAPC23525×105107
crustaceansE. coli535211500
 S. aureus8250103-
Cooked, chilled,APC2352105106
and frozenE. coli635111500
Fresh and frozenAPC32505×105-
bivalve molluscsE. coli625016-

1) APC = Aerobic Plate Count (preferably carried out at 21–25°C on a nutrient rich, non-selective agar.


A number of microbiological tests of fish and fish products are used by industry for contractual and internal purposes and by authorities to check that the microbiological status is satisfactory. The purpose of these examinations is to detect for pathogenic bacteria (Salmonella, V. parahaemolyticus, Staphylococcus aureus, Listeria monocytogenes, E. coli) or for organisms which are possible indications of fecal contamination (E. coli) or other types of general contamination or poor manufacturing practices (coliform bacteria, fecal streptococci, aerobic plate count (APC).

Microbiological tests are generally costly, time-consuming and require a lot of manual labour but rapid automated tests are becoming available and being given accreditation. Consequently the number of samples which can be examined is limited. Furthermore, it should be emphasized again that a negative test for specific pathogens in a food sample is no guarantee that the whole lot is free of these pathogens. Thus only a very limited degree of safety can be obtained by microbiological testing. There are other limitations for some of these tests.

Total Viable Count (TVC) or Aerobic Plate Count (APC) is defined as the number of bacteria (cfu/g) in a food product obtained under optimal conditions of culturing. Thus the TVC is by no means a measure of the “total” bacterial population, but only a measure of the fraction of the microflora able to produce colonies in the medium used under the conditions of incubation. Thus it is well known that the temperature during incubation of plates influences greatly on the number of colonies developing from the same sample. As an example, the TVC may vary by a factor 10–100 when iced fish is sampled and plates are incubated at 20°C and 37°C respectively. Furthermore, the TVC does not differentiate between types of bacteria and similar levels of TVC may therefore be found although the biochemical activity of the bacteria may vary widely in the food. Also, high counts as a result of microbial growth are much more likely to cause defects in foods than are similar levels caused by recent gross contamination. TVC is therefore of no value in assessing the present state of sensory quality.

A TVC is meaningless as a quality index for products in group C and F (see Section 5.1.3) as a large population of non-spoilage lactic acid bacteria normally develop in these products. TVC is of very doubtful value in the examination of frozen fish products. An unknown and uncontrolled kill or damage of the bacteria may have taken place during freezing and cold storage. A very low “total” count may therefore lead to false conclusions about the hygienic quality of the product. Tests for TVC may be useful for measuring the conditions of the raw material, effectiveness of procedures (i.e. heat treatment) and hygiene conditions during processing, sanitary conditions of equipment and utensils and time x temperature profile during storage and distribution. However, to be useful and for correct interpretation of results a thorough knowledge of handling and processing conditions prior to sampling is essential.

E. coli: The natural habitat for this organism is the intestines of human and vertebrate animals. In temperate waters this organism is absent from fish and crustaceans at the time of capture (except in grossly polluted waters). Moreover, fish and shellfish should always be held at temperatures below those which support growth. This organism is therefore particularly useful as indicator of contamination (small numbers) or mishandling such as temperature abuse in product handling (large numbers). Contamination of food with E. coli implies a risk that one or more of enteric pathogens may have gained access to the food. However, failure to detect E. coli does not assure the absence of enteric pathogens (Mossel 1967, Silliker and Gabis 1976).

Recent investigations have shown that E. coli and fecal coliform bacteria can be found in unpolluted warm tropical waters and that E. coli can survive indefinitely in this environment (Hazen 1988, Fujioka et al. 1988, Toranzos et al. 1988). These studies also revealed that there was no correlation between presence or absence of fecal coliforms, total coliforms and virus. Thus, in the tropics E. coli or fecal coliforms are not reliable as indicators of recent biological contamination or sewage effluent discharge into the aquatic receptor. This point should be taken into consideration when microbiological standards are applied to fish products from tropical countries.

The resistance of E. coli to adverse physical and chemical conditions is low. This makes E. coli less useful as indicator organisms in examination of water and frozen or otherwise preserved fish products. Thus it is well established that enteric viruses survive much longer than E. coli in sea water (Melnick and Gerba 1980) and that E. coli is less resistant than Salmonella in frozen products (Mossel et al. 1980).

Fecal coliforms: This group of bacteria is often used in microbiological criteria instead of E. coli in order to avoid the lengthy and costly confirmation tests for E. coli. These organisms are selected by incubating an inoculum derived from a coliform enrichment broth at higher temperatures (44°C – 45.5°C). Thus, the group of fecal coliforms has a higher probability to contain organisms of fecal origin and hence indicating fecal contamination. Apart from being more rapid (- and less specific), a test for fecal coliform suffers the same limitations as described for E. coli. It should also be observed that the recently described pathogenic E. coli 0157:H7 does not grow at 44°C on all the selective media normally used for enumeration of E. coli (see Section 3.1.2).

Fecal streptococci or enterococci: It is now well established that fecal streptococci are not a reliable index of fecal contamination. Many foods and fish products contain these organisms as a normal part of their flora, and they are also able to establish themselves and persist in a food processing plant. Most are salt tolerant and may grow at 45°C as well as at chill temperatures (7–10°C). Unlike E. coli, they are relatively resistant to freezing, which makes them potentially useful as indicator organisms for evaluating plant hygiene during processing of frozen food.

Staphylococcus aureus: This organism is included in a number of microbiological criteria. Enumeration of this organism presents no problem. Spread plating on Baird-Parkers egg yolk medium and incubation for 30 hours at 37°C is a most reliable method. Positive cultures need to be confirmed by testing for coagulase activity.

The natural reservoir for S. aureus is human skin, hair and superficial mucous membranes (nose), while it is not a part of the normal flora on fish and fish product. Presence of large numbers indicate the possible presence of enterotoxin and/or faulty sanitary or production practice. Small numbers are to be expected in products handled by humans. It should be emphasized that S. aureus grows poorly in competition with large numbers of other microorganisms. For this reason, a test for S. aureus is only relevant for fish products which have received a bactericidal treatment, i.e. a heat treatment during processing. A test for toxin should be included if growth of S. aureus is suspected.


A microbiological criterion is a standard, against which comparison and evaluation of own data can be made. A microbiological criterion may have either mandatory or advisory status. The various types of criteria have been defined by a subcommittee on microbiological criteria established by the U.S. National Research Council (FNB/NRC 1985):

Microbiological criteria may be helpful in assessing the safety and shelf-life of foods, the adherence to established Good Manufacturing Practice (GMP) and the suitability of food for a particular purpose. The various criteria will therefore often include both values for pathogenic bacteria or their toxins and indicator organisms.

It was further recommended by the subcommittee (FNB/NRC 1985) that a microbiological criterion should include the following components:

Microbiological criteria should only be established when there is a need for it, and when it can be shown to be effective and practical. A number of factors should be considered as listed in FNB/NRC (1985). These include evidence of a hazard, nature of the product and the associate microflora, effect of processing, the state in which the food is distributed, the manner in which it is ultimately prepared for consumption, and whether reliable and practical methods of detection are available at a reasonable cost. A microbiological standard should be considered only when:

Microbiological guidelines or reference values (Mossel 1982) are established as a result of surveys carried out during processing in a number (8–10) of factories where GMP is applied. Initially, all significant details of GMP are checked by visual inspection, instrumental methods or bacteriological tests. When everything is found to be in order, at least 10 samples from every checkpoint from every factory are drawn and examined. Distribution curves of the data obtained are prepared and used as a basis for establishment of reference values as suggested by Mossel (1982) (see Figure 4.2).

Figure 4.2.

Figure 4.2. Distribution plot of the result of microbiological surveys on a given type of food (Mossel 1982).
Φ- 95th percentile
n - Reference Value proper
N - maximal count to be expected
under conditions of GMP
cfu - colony-forming units
MID - Minimal Infectious Dose
MSL - Minimal Spoilage Level

The selection of values for n and N may vary according to the food involved and the local situation. As a general rule, N is one log cycle higher than n and one log cycle lower than MID or MSL. If Φ approaches MID or MSL too closely, improvement of the manufacturing technique is required. However, a certain tolerance has to be built into the reference values. The area between n and N is the “alert” area and the customary tolerance for non-pathogenic organisms is that no more than 2 out of 10 samples are expected in this range and none should show a cfu/g value over 10 times the reference value.

Microbiological guidelines are useful in determining the degree of control during processing and the conditions during distribution and storage. Thus the microbiological guideline can easily be incorporated in a HACCP-system (see Section 5.1) where they are useful as reference values in the monitoring work.

Also, microbiological specifications used in commercial transactions should be based on relevant background data and should fill a need. The currently used microbiological criteria applied to fish and fishery products by the members of the European Community together with Canada, Japan and USA (- these countries collectively import over 90% of the fish that is traded) are compiled by FAO (1989). The tests required are listed in Table 4.2.

It is apparent that the requirements of the microbiological criteria as listed above are not always considered in current practices for fish and fish products. Most of the standards listed in the FAO-circular (FAO 1989) are incomplete, unnecessary , unrealistic and should be reconsidered. In most cases only microbiological limits are specified and all the other components in a criterion are not considered. By careful evaluation of all aspects related to, for instance, fresh and frozen fish products which are intended to be heated before consumption, it is clear that these products do constitute neither a health risk nor a serious quality problem. The major problem related to these products is concerned with the possible presence of biotoxins. Thus, there is no need or justification for a microbiological criterion. Similarly, a large population of harmless lactic acid bacteria develop in lightly salted and cold smoked fish which makes a microbiological standard based on aerobic plate count (APC) meaningless. The inclusion of counts for S. aureus in microbiological standards for raw products with a large associate flora is also meaningless as already mentioned (Section 4.2).

A more realistic approach is taken by the ICMSF (1986) as shown in Table 4.1. Test for S. aureus is only recommended for cooked products, and E. coli is generally used as indicator for fecal contamination of all product types. However, the grouping of products in Table 4.1. is unscientific. Cold smoked fish are grouped with fresh and frozen fish although their microbiological ecology is vastly different, while frozen raw crustaceans form a group on its own although being microbiologically very similar to fresh and frozen fish. It is suggested that fish products should be grouped as shown in Section 5.1.3.

The microbiological limits recommended by ICMSF (1986) should be regarded as part of microbiological guidelines and useful mainly in the control of GMP. However, there is little or no evidence that these criteria have contributed significantly to the prevention of outbreaks of diseases attributed to these products. In view of the differences in the microbiological contamination of fish and crustaceans from various parts of the world, it is doubtful, whether such criteria are universally applicable.

Table 4.2. Microbiological tests included in the Microbiological Standards and Regulations of Some European Countries, Japan and USA. Belgium, Canada, Denmark, Germany, Greece and Portugal have no microbiological standards for fish and fish products. Data from FAO (1989).
 ItalyFranceLuxembourgNetherlandsUnited KingdomSpainUSAJapan
Raw fish, fillets, 1,2,7,10,11*)1,3,7,10,11  1,2,5,6,7,10 1,2
fresh/frozen       (6)
pasteurized 1,2,7,10,11      
non-pasteurized 1,2,7,10,11      
Smoked salmon 1,2,7,10,11      
raw 1,3,7,111,3,7,11   1,6,10 
cooked 1,3,7,111,3,7,11 1,6,7,10   
cooked and 1,3,7,101,3,7,10     
peeled 11117,10    
raw6,7  6,7 1,6,7 1,6
pre-cooked6,71,3,7,10,111,3,7,10,11  1,7,8,9,10  

The figures refer to tests for:
1. Aerobic plate count (TVC)
2. Coliforms
3. Fecal coliforms
4. Fecal streptococci
5. Enterococci
6. E. coli
7. Salmonella
8. Shigella sp.
9. Total enterobacteriaceae
10. Staphylococcus aureus
11. Anaerobic sulfite red.

In conclusion, it can be stated that there are no practical systems for providing safety or assurance of safety and normal shelf-life of fish products by microbiological end-product testing. Point-of-entry testing of fish products must generally be considered as an inefficient means of retrospective assessment of processing, transport and storage conditions. For this reason other methods should be used to assure a reasonable degree of protection of both consumer and producer against risks associated with microbiological activity. Apart from being useless as hygienic measures irrelevant criteria can still be of consequence by imposing unnecessary costs, introducing non-tariff barriers to trade and inducing a false sense of security.

However, microbiological criteria may be useful as a means for assessing the effectiveness of a quality assurance programme (HACCP) particularly as apart of a verification programme. This will be discussed in more detail in Section 5.1.3, but it cannot be overemphasized that microbiological criteria per se are totally inadequate.

By January 1st 1993 the single European market was established. The EEC Council Directive 91/493/EEC (EEC 1991b) is laying down the health conditions for the production and the placing on the market of fishery products. The directive gives provisions for laying down criteria for organoleptic quality, parasites, chemical checks (TVB-N, histamine and chemical contaminants) and microbiological analysis, including sampling plans and methods of analysis. So far, there are only criteria for histamine content of fish (9 samples must be taken from each batch. The mean value must not exceed 100 ppm, 2 samples may have a value of > 100 ppm, but < 200 ppm, no samples may have > 200 ppm) and microbiological criteria for cooked, ready-to-eat shrimp and crabmeat, where the following standards apply:

  1. Salmonella sp. - not to be detected in 25 g (n = 5, c = 0)

  2. S. aureus (cfu/g) m = 100, M = 1000 (n = 5, c = 2)

  3. either thermotolerant coliforms (44°C) (cfu/g), m = 10, M = 100, (n=5, c=2) or E. coli (cfu/g), m = 10, M = 100, (n=5, c = 1).

See page 55 for explanation of n, c, m and M.

Furthermore the following microbiological guidelines apply for the same product.

Total viable count (aerobic, 30°C):

Whole product: m = 10,000, M = 100,000 (n = 5, c = 2)

Products without shell, not including crabmeat: m = 50,000 M = 500,000 (n =5, c=2)

Crabmeat: m = 100,000, M = 1,000,000 (n = 5, c=2).

For live, bivalve molluscs the requirements are listed in EEC-Council Directive 91/492/EEC of 15 July 1991 (EEC 1991a) as shown below:

Requirements concerning live bivalve molluscs

Live bivalve molluscs intended for immediate human consumption must comply with the following requirements:

  1. The possession of visual characteristics associated with freshness and viability, including shells free of dirt, an adequate response to percussion, and normal amounts of intravalvular liquid.

  2. They must contain less than 300 fecal coliforms or less than 230 E. coli per 100 g of flesh and intravalvular liquid based on a 5-tube 3-dilution MPN-test or any other bacteriological procedure of equivalent accuracy.

  3. They must not contain Salmonella in 25 g of flesh.

  4. They must not contain toxic or objectionable compounds occurring naturally or added to the environment.

  5. The upper limit as regards the radionucleide contents must not exceed the limits for foodstuffs as laid down by the Community.

  6. The total paralytic shellfish poison (PSP) content in the edible parts of molluscs must not exceed 80 μg per 100 g.

  7. The customary biological testing methods must not give a positive result to the presence of diarrhetic shellfish poison (DSP) in the edible parts of the molluscs.

  8. In the absence of routine virus testing procedures and the establishment of virological standards, health checks must be based on faecal bacteria counts.

Public health control

The public health control system must check, among other things, the microbiological quality of the live bivalve molluscs and the possible presence of toxin-producing plankton in the water and biotoxins in the molluscs. The sampling used for the control of toxins must be carried out in two steps:

  1. Monitoring: Periodic sampling organized to detect changes in the composition of the plankton containing toxins and the geographical distribution thereof. Information leading to a suspicion of accumulation of toxins in mollusc flesh must be followed by:

  2. Intensive sampling: The number of sampling points and the number of samples is increased and at the same time toxicity tests are introduced.

Previous Page Top of Page Next Page