The HACCP and FSO Systems for Food Safety
509
be made more stringent by increasing n (e.g., n = 10, c = 2, m = 10
2 ) or by reducing c (e.g., n = 5,
c = 1, m = 10
2 ). On the other hand, it can be made more lenient for a given size n by increasing c.
Whereas a two-class plan may be used to designate acceptable/unacceptable foods, a three-class plan
is required to designate acceptable/marginally acceptable/unacceptable foods. To illustrate a typical
three-class plan, assume that for a given food product, the standard plate count (SPC) shall not exceed
10
6 /g (M) or be higher than 10
5 /g from three or more of five units examined. The specifications are
thus n = 5, c = 2, m = 10
5 , M = 10
6 . If any of the five units exceeds 10
6 /g, the entire lot is rejected
(unacceptable). If not more than c sample units give results above m, the lot is acceptable. Unlike
two-class plans, the three-class plan distinguishes values between m and M (marginally acceptable).
With either two- or three-class attribute plans, the numbers n and c may be employed to find the
probability of acceptance (P a ) of lots of foods by reference to appropriate tables.
8 The decision to
employ a two-class or three-class plan may be determined by whether presence/absence tests are
desirable, in which case a two-class plan is required, or whether count or concentration tests are
desired, in which case a three-class plan is preferred. The latter offers the advantages of being less
affected by nonrandom variations between sample units and of being able to measure the frequency of
values in the m to M range. The ICMSF report and recommendations
8 should be consulted for further
details on the background, uses, and interpretations of sampling plans. Further information may also
be obtained from Kilsby.
10
Microbiological Criteria and Food Safety
The application of criteria to products in the absence of an HACCP program is much less likely to
be successful than when the two are combined. Thus, microbiological criteria are best applied as part
of a comprehensive program. When criteria are not applied as components of a systematic approach to
food safety or quality, the results are known to be less than satisfactory, as found by Miskimin et al.
11
and Solberg et al.
20 These investigators studied over 1,000 foods consisting of 853 ready-to-eat and
180 raw products. They applied arbitrary criteria for APC, coliforms, and E. coli and tested the efficacy
of the criteria to assess safety of the foods with respect to Staphylococcus aureus, C. perfringens, and
salmonellae. An APC criterion of less than 10
6 /g for raw foods resulted in 47% of the samples being
accepted even though one or more of the three pathogens were present, whereas 5% were rejected
from which pathogens were not isolated, for a total of 52% wrong decisions. An APC of less than
10
5 /g for ready-to-eat foods resulted in only 5% being accepted that contained pathogens, whereas
10% that did not yield pathogens were rejected. In a somewhat similar manner, a coliform criterion of
less than 10
2 /g resulted in a total of 34% wrong decisions for raw and 15% for ready-to-eat foods. The
lowest percentage of wrong decisions for ready-to-eat foods (13%) occurred with an E. coli criterion
of less than 3/g, whereas 30% of the decisions were wrong when the same criterion was applied to
raw foods. Although the three pathogens were found in both types of foods, no foodborne outbreaks
were reported over the 4-year period of the study, during which time more than 16 million meals were
consumed.
19
The above findings represent some initial data from the Rutgers Foodservice Program. After a
17-year experience with modifications in surveillance tests, food audits, laundry evaluations, and
more than 30 million meals served, this HACCP-type system has been very effective.
19 The microbial guidelines employed by this program for raw and ready-to-eat foods are presented in Table 21–3. Of over 1,600 food samples examined over the period 1983–1989, only 1.24% contained
pathogens, with protein salads most often contaminated (4.3%). Among the foods that failed microbial
surveillance were raw vegetables (they had excessive coliform numbers).
19 The Rutgers Foodservice
509
be made more stringent by increasing n (e.g., n = 10, c = 2, m = 10
2 ) or by reducing c (e.g., n = 5,
c = 1, m = 10
2 ). On the other hand, it can be made more lenient for a given size n by increasing c.
Whereas a two-class plan may be used to designate acceptable/unacceptable foods, a three-class plan
is required to designate acceptable/marginally acceptable/unacceptable foods. To illustrate a typical
three-class plan, assume that for a given food product, the standard plate count (SPC) shall not exceed
10
6 /g (M) or be higher than 10
5 /g from three or more of five units examined. The specifications are
thus n = 5, c = 2, m = 10
5 , M = 10
6 . If any of the five units exceeds 10
6 /g, the entire lot is rejected
(unacceptable). If not more than c sample units give results above m, the lot is acceptable. Unlike
two-class plans, the three-class plan distinguishes values between m and M (marginally acceptable).
With either two- or three-class attribute plans, the numbers n and c may be employed to find the
probability of acceptance (P a ) of lots of foods by reference to appropriate tables.
8 The decision to
employ a two-class or three-class plan may be determined by whether presence/absence tests are
desirable, in which case a two-class plan is required, or whether count or concentration tests are
desired, in which case a three-class plan is preferred. The latter offers the advantages of being less
affected by nonrandom variations between sample units and of being able to measure the frequency of
values in the m to M range. The ICMSF report and recommendations
8 should be consulted for further
details on the background, uses, and interpretations of sampling plans. Further information may also
be obtained from Kilsby.
10
Microbiological Criteria and Food Safety
The application of criteria to products in the absence of an HACCP program is much less likely to
be successful than when the two are combined. Thus, microbiological criteria are best applied as part
of a comprehensive program. When criteria are not applied as components of a systematic approach to
food safety or quality, the results are known to be less than satisfactory, as found by Miskimin et al.
11
and Solberg et al.
20 These investigators studied over 1,000 foods consisting of 853 ready-to-eat and
180 raw products. They applied arbitrary criteria for APC, coliforms, and E. coli and tested the efficacy
of the criteria to assess safety of the foods with respect to Staphylococcus aureus, C. perfringens, and
salmonellae. An APC criterion of less than 10
6 /g for raw foods resulted in 47% of the samples being
accepted even though one or more of the three pathogens were present, whereas 5% were rejected
from which pathogens were not isolated, for a total of 52% wrong decisions. An APC of less than
10
5 /g for ready-to-eat foods resulted in only 5% being accepted that contained pathogens, whereas
10% that did not yield pathogens were rejected. In a somewhat similar manner, a coliform criterion of
less than 10
2 /g resulted in a total of 34% wrong decisions for raw and 15% for ready-to-eat foods. The
lowest percentage of wrong decisions for ready-to-eat foods (13%) occurred with an E. coli criterion
of less than 3/g, whereas 30% of the decisions were wrong when the same criterion was applied to
raw foods. Although the three pathogens were found in both types of foods, no foodborne outbreaks
were reported over the 4-year period of the study, during which time more than 16 million meals were
consumed.
19
The above findings represent some initial data from the Rutgers Foodservice Program. After a
17-year experience with modifications in surveillance tests, food audits, laundry evaluations, and
more than 30 million meals served, this HACCP-type system has been very effective.
19 The microbial guidelines employed by this program for raw and ready-to-eat foods are presented in Table 21–3. Of over 1,600 food samples examined over the period 1983–1989, only 1.24% contained
pathogens, with protein salads most often contaminated (4.3%). Among the foods that failed microbial
surveillance were raw vegetables (they had excessive coliform numbers).
19 The Rutgers Foodservice