T. Gerhardt et G. Hildebrandt, Routine microbiological sampling and analysis of ground meat for quality control purposes - practical experience and statistical conclusions, ARCH LEBENS, 51(1), 2000, pp. 13-21
The microbiological status of ground beef was routinely assessed at two pro
duction facilities over a period of at least one year. Five re-plicate samp
les were tested per day in accordance with European meat hygiene directives
. In light of the high workload it was of great interest to decide whether
the sample number could be reduced by combining the 5 single samples per lo
t to a pool sample without causing a significance loss of information. Conc
erning the "aerobic total plate count" as the main criterion the findings s
howed that:
Analysing a pool sample and using 3m reads in 129 of 131 lots of producer A
to the same decision (acceptance/rejection) as 5 random samples in connect
ion with the official 3-class-sampling plan (3(m)/c(m) = 2; M/c(M) = 0). In
the case of producer B there have been 3 opposite decisions in 319 lots de
spite of a very heterogeneous structure of the microbiological counts. More
defective lots were observed on Mondays. No other trends could be observed
.
Obviously the microbiological results of 5 samples/day are only used for do
cumentation and not for corrective actions or improvement of the hygienic s
tatus, ii there are no severe objections against the bacteriological qualit
y. Because the relatively poor information does not justify the analytical
workload a modification of the official sampling plan is proposed:
Following the principles of moving sum procedure the 5 random samples of mi
nced meat per lot are combined to a pool sample and tested against the limi
t 3m as long as no more than 2 plate counts exceed the limit 3m in the last
11 days of production.