Purpose: To evaluate the effect of different sampling programs on the risk of E. coli O157 infection from the consumption of hamburgers made from Australian manufacturing meat.
Methods: A total of 10,000 contaminated lots of 700×27.2kg cartons of Australian manufacturing meat were simulated in production of 100g patties for home consumption. Each simulated lot was sampled using a Binomial-Poisson-Lognormal approach, where the
- Binomial distribution gives the probability of sampling contaminated cartons;
- Poisson distribution (rate λ) gives the probability of detecting E. coli O157 in a carton; and
- Lognormal distribution gives variability in the average rate of E. coli O157 contamination (λ).
Sampling plans included N-60 (testing five samples from 12 cartons each), N-90 (testing five samples from 18 cartons each), N-120 (testing five samples from 24 cartons each) and ICMSF n=60 (testing five 50cm2 samples from 12 cartons each). Lots in which contamination was detected were removed from the simulation and the resulting infection rate (per 100g serving) was calculated.
Results: The results indicate that compared with no sampling, which had a risk of E. coli O157 infection of 2.83×10-7, sampling using current N-60 reduced this risk to 2.58×10-7, while sampling using N-90 and N-120 had limited further impact on the risk, which were 2.53×10-7 and 2.51×10-7, respectively. Sampling more surface area per carton (ICMSF n=60) yielded a risk of 2.44×10-7. In contrast, cooking to 68°C reduced the risk to 7.30×10-11.
Significance: While sampling helps detect highly contaminated lots, it does not provided effective public health protection from the consumption of undercooked burgers. Increased testing beyond the current sampling plan provides marginal additional public health benefit.