An approach to the microbiological risk ranking of cheeses

Abstract

The Food and Agriculture Organization (FAO) promote risk categorization approaches to assist understanding the public health risk associated with imported foods such as cheeses. A risk categorization should use information from the imported food, importer and exporting country profile to establish risk associated with imported foods. The first step involves assessing the product characteristics whilst step two is intended to consider country specific controls.

In this paper, we describe the first step in the development of a risk ranking approach using product characteristics to rank cheeses based on microbiological risk. The approach is applied to sixty types that may be commonly traded, based on an assessment of UK data on cheese imports. It uses risk factors of milk pasteurization, ripening method, and predicted pathogen growth in the finished cheese. Each risk factor was scored, and the total used to rank the cheeses with respect to the likelihood of them containing viable pathogens at the border, dependent on initial pathogen presence and potential for subsequent growth. The possible range of scores was 0 to 6. After scoring, none scored 0 or 1, and 13 scored 6.

Using international foodborne disease outbreaks and EU Rapid Alert System for Food and Feed (RASFF) notifications as benchmarks, cheese types were assigned one of four qualitative risk levels, from Very Low to High, based on their scores. These qualitative intervals did not align with trade codes, which are not risk-based, as some codes included cheeses from different risk levels. Neither was there a good correlation with Codex cheese classes, although it was noted that some very hard cheeses (e.g., Parmesan) received the lowest risk scores.

When scores were compared with outbreak and RASFF qualitative data there was a good correlation. For example, of those scored, no cheese scoring ≤3 had received a RASFF notification or caused an outbreak. Of the outbreaks associated with cheese types that were scored, 83.3 % of outbreaks implicated cheeses with a score ≥5.

For the sixty cheeses presented here, the data required for scoring were available. However, data may not be available for all cheeses. Where it is necessary to score a cheese that is lacking the necessary data, a read-across approach would be a potential solution, but this would increase associated uncertainty. The use of a standardized approach to risk ranking of cheeses produced consistent risk-based information that can be compared between cheese types but not between two classification systems (Harmonised System and Codex system). This aligns with the fact that neither classification system was designed to correlate with consumer safety.