4.1. Quantitative Microbial Risk Assessment (QMRA) for E. coli O157:H7
In recent years, many countries have applied QMRA to assess the risk of foodborne illnesses in individuals consuming contaminated foods. The results assist governments in developing management policies to improve food safety. However, there is currently no comprehensive QMRA research conducted for specific groups and foods in Taiwan. Owing to differences in the dietary habits of the Taiwanese population, varied results may be observed in comparison to the risk assessments conducted in other countries for foodborne illnesses caused by ingesting E. coli O157:H7 in beef products.
Comparatively, Kiermeier et al. investigated the incidence of foodborne illness from consuming hamburgers made from imported Australian beef in fast food restaurants in the USA (2.2 × 10−6
–1.5 × 10−5
]. This rate was similar to our observed in Taiwanese males aged 19–65 years from consuming rare beef (i.e., 2.8 × 10−5
) with an exceedance risk of 0.05. On the other hand, the incidence of foodborne illness from consuming medium beef with an exceedance risk of 0.05 in males aged 19–65 years was 1.3 × 10−7
, which is also lower than that which Kiermeier et al. reported [6
]. In practice, when importing the beef from the USA, contaminated beef should be removed from the supply chain after microbiological testing in import or export inspection; thus, it is expected that this incidence is lower. Moreover, there were no E. coli
O157:H7 contamination data for intact beef or steak. Therefore, we referred to the contamination data published for ground beef. Although this tends to overestimate the bacterial contamination of real beef cuts, it is rational to use the data for a conservative estimate in risk assessment.
We use exceedance risk for knowing the possibilities that may occur in different specific morbidity scenarios. When the exceedance risk was 0.05, the following results were observed: (1) The incidence of foodborne illness from consuming rare beef in males aged 19–65 years was 2.8 × 10−5. Therefore, the incidence rate of foodborne illness was >1 in 10,000 individuals and the probability of occurrence was 5%. (2) The incidence of foodborne illness from consuming medium beef in males aged 19–65 years was 1.3 × 10−7. Thus, the incidence rate of foodborne illness was >2 in 1 million individuals and the probability of occurrence was 5%. (3) The incidence of foodborne illness from consuming rare beef in females aged 19–65 years was 9.3 × 10−6. Hence, the incidence rate of foodborne illness was >10 in 10,000 individuals and the probability of occurrence was 5%. (4) The incidence of foodborne illness from consuming medium beef in females aged 19–65 years was 7.8 × 10−8. Thus, the incidence rate of foodborne illness was >8 in 10 million individuals and the probability of occurrence was 5%.
Comparison of the risk calculated in the present study with that reported by Cassin et al., only when the exceedance risk was 0.05 that the incidence of foodborne illness from consuming rare and medium beef in the Taiwanese population is higher than that recorded in the Canadian population caused by ingesting E. coli
O157:H7 in beef patties [17
]. In other words, when consuming medium beef in the Taiwanese population had a 95% chance, the incidence of foodborne illness for every 100 million serving beef contaminated with E. coli
O157:H7 was under 1.3 × 10−8
. Our results were also lower than risk assessments annual illnesses from ground beef consumption in the United States (1.0 × 10−6
]. Moreover, the incidence of foodborne illness from consuming medium-well beef (near zero) was lower than that noted among the Canadian population.
In present report, under the most conservative scenario, there was approximately a 5% probability that the incidence of foodborne illness from consuming rare steak would be >2.8 × 10−5 (95% probability will under 2.8 × 10−5) and the highest incidence of foodborne illness was 1.43%; however, the probability of occurrence was only 0.01%, in other words, 99.9% of illness will be under 1.43%. There was approximately a 5% probability that the incidence of foodborne illness from consuming medium beef would be >1.3 × 10−7. The highest incidence of foodborne illness was 8.21 × 10−6; however, the probability of occurrence was only 0.01%.
Cassin et al. studied the risk of foodborne illness caused by ingesting E. coli
O157:H7 in beef patties in Canada [17
]. Their results showed that the incidence of foodborne illness from consuming one beef patty was 1.0 × 10−22
–1.0 × 10−2
. The average incidence of foodborne illness for each beef patty in adult and child populations was 5.1 × 10−5
and 3.7 × 10−5
, respectively. The sensitivity analysis showed that the amount of E. coli
O157:H7 bacteria in bovine feces exhibited the highest correlation (~0.6), followed by host susceptibility (~0.6), and carcass contamination (~0.3). In our study, the highest correlation was estimated dose (~0.909), followed by a contamination rate of beef with E. coli
O157:H7 (~0.307), and consumption rate (~0.263). The estimated dose of E. coli
O157:H7 bacteria had the most impact on the mean number of illnesses in our model.
In our study, the scenario uncertainty could be caused by the following conditions (1) Assessment of beef imported from the USA only: the domestic production of beef accounts for only 5% of the total beef sold, and E. coli O157:H7 has not been found in domestic beef products. USA beef accounts for the largest proportion of the imported beef (approximately 30%). Therefore, only beef products made from beef imported from the USA were assessed. (2) Transportation temperature and time: under actual conditions, beef may be refrigerated when imported and the transportation times may vary. However, the present study assumed that changes in the number of E. coli O157:H7 in raw beef were not significant, regardless of whether it was frozen or refrigerated during transportation. (3) Cooking conditions: the assumptions of this study regarding the thickness, weight, and cooking temperature and time for each piece of beef may differ from the actual conditions in restaurants. (4) Cross-contamination: the risk of cross-contamination after cooking beef was not considered and may be underestimated. (5) Consumption rate: the present study only calculated individuals who ate beef (consumer-only) and excluded a consumption rate = 0. Therefore, it is not to assess the actual food intake in the whole population. In addition, Taiwanese individuals do not consume beef frequently which may have resulted in an overestimation of the consumption rate. The present study assumed that the daily consumption rate was the consumption rate for each incidence. However, this may differ from the actual daily consumption of beef.
The parameter uncertainties contained in the present study included the following aspects: (1) The number of contaminated bacteria: the amount of E. coli
O157:H7 contamination was based on the research hypotheses proposed by Cassin et al. [17
] rather than the actual test results for E. coli
O157:H7 in fresh beef sold in the USA or Taiwan. (2) Contamination rate: the contamination rate of E. coli
O157:H7 is derived from domestic inspections by the USDA rather than inspections of fresh beef imported into Taiwan from the USA. (3) Dose–response relationship: in the present study, the dose–response relationship was based on the adult population [17
]. The actual dose–response relationship may differ among individuals with low immunity, such as children and the elderly.