A quantitative approach for evaluating current and alternative border control of Salmonella Typhimurium and Enteritidis in imported frozen poultry meat in respect to human risk of salmonellosis

During the last decades, the Middle East and North African countries, including Jordan, have gone through fast development with significant increases in population numbers, pressures on agriculture production, and water resources. In addition, there is a continuous change of preferred dietary and consumption patterns. Simultaneously, the international food trade has grown, negotiated their agreements. According to Food and Agriculture Organization (FAO) study in 2017, there are enormous demands on different food products globally (Bruinsma, 2017). It was expected that poultry meat imports into developing countries would grow 3-4% per year until 2025, while the African and Middle East countries will likely share about 65% of this rise in world poultry imports (Bruinsma, 2017).

Up to date, poultry meat is considered one of the primary sources of disseminating hazardous pathogens, especially non-typhoidal Salmonella (NTS), to consumers, and salmonellosis is the major food safety issue worldwide, including the Middle East and North Africa (MENA) population. In these developing countries, including Jordan, sporadic cases and outbreaks of salmonellosis are common, and World Health Organization (WHO) linked most of the cases to consumption of contaminated poultry meat products. These products can become contaminated with NTS throughout the food chain, from primary production in the poultry farms through slaughtering and cutting plant to consumption.

To ensure the entrance of safer imported poultry meat products to Jordan at the border control, Jordan Food and Drug Administration (JFDA) applied a Salmonella surveillance system for imported poultry meat and poultry products. Border control takes place at ten entry points and has taken place since the 1980s. Since initiated, the border control has been continuously updated, focusing on the size of the picked packages for sampling and the diagnostic method. However, there is no such available tool to evaluate the effectiveness of the current or alternative Salmonella surveillance system in imported poultry meat and poultry products.

In response to the lack of objective and quantitative tools for evaluation of the Salmonella surveillance system, using Jordan as a model country, the research presented in this thesis has the following objectives:

Objective I: Investigating the number of frozen meat imports to Jordan between 2015 and 2019 that was contaminated with microbiological zoonosis.

Objective II: Determining the limit of detection (LOD) of the Rapid’Salmonella method in the surveillance settings.

Objective III: Developing a stochastic Quantitative Microbiological Risk Assessment (QMRA) model for assessing a surveillance system's effectiveness on the risk for the consumers – in this case, salmonellosis due to imported poultry meat and poultry products.

The specific outcome of the research performed during the PhD scholarship is a Quantitative Microbiological Risk Assessment (QMRA) model were i) the occurrence of Salmonella in the poultry meat of different products and export countries, ii) the sample size for microbiological analysis, iii) performance of the diagnostic test (Rapid'Salmonella) and iv) criteria for compliance are integrated. The model includes a consumer phase module, which makes it possible to assess the effectiveness of the surveillance on the likelihood of salmonellosis (the risk) in the consumer population.

This quantitative tool will support decision-makers to have a quantitative, data-driven evaluation of the applied inspection systems – both current and potential systems in the future. This is reliable toward reducing the anticipated risk among consumers attributable to imported poultry meat specifically and imported meat generally.

Chapter 1 of the thesis describes the problem analysis and research hypothesis in depth.

Chapter 2 described the JFDA's active surveillance system for imported poultry meat and poultry products. In the thesis, I considered the surveillance system as two parts - monitoring (assessing the risk) and actions (managing the risk). The monitoring part had three consecutive procedures:

- Sampling procedure for detection of Salmonella in imported poultry meat.

- Applying the Salmonella detection method used in JFDA laboratories.

- Interpretation of laboratory results and report the interpretation in a formal report, including decision complying / non-complying of the batch.

The actions part that initiated according to the results of monitoring which had the following procedures:

- The direct actions on the batch that non-compliant to destroy it or return it to the country of origin or other destinations.

- Proactive actions toward countries where the non-compliant batches originate, such as banning the import or more strict sampling and further detection.

In Chapter 3, I described the main objectives in research activities I have performed, including the used methodologies.

Chapter 4 includes the three manuscripts (manuscript I, II, and III), wherein the core of the research is presented and discussed. Chapter 5 includes the overall discussion and conclusion, bringing the obtained results into a broader context and outlining the work's pros and cons. In chapter 6, based on the obtained results and needs for reducing the risk, I outline the future perspectives especially facing the needs and challenges in developing countries regarding food safety issues.

The manuscript I addresses objective I - Investigating the number of frozen meat imports to Jordan between 2015 and 2019 that was contaminated with microbiological zoonosis. In this study, I investigated the frequency of detected foodborne pathogens in frozen red meat and frozen poultry meat imports to Jordan: Salmonella enterica subsp. enterica ser. Typhimurium and Salmonella enterica subsp. enterica ser. Enteritidis, Listeria monocytogenes, and Escherichia coli (E. coli) O157:H7. The data was obtained from the JFDA's Laboratory Food Examination System (LFES) portal database, and covered the period from 2015 to 2019. This investigation was essential to focus on the most frequent detected hazard in meat imported to Jordan. Overall, the main product that was contaminated with microbial hazards was the poultry product, and the dominant detected pathogen was Salmonella. These results gave the right path to reach both Objectives II and III.

In manuscript II, I studied the limit of detection (LOD) of the laboratory method (Rapid'Salmonella) used in the border control in Jordan. Knowledge of the LOD for Salmonella in frozen imported poultry products into Jordan, and especially for the Salmonella serotypes typically found in the imported products, is crucial when using the estimated LODs in the context of the Jordanian surveillance at border control. The LOD was studied using a spiking experiment. Salmonella Typhimurium and Salmonella Enteritidis field serotypes were spiked on Salmonella-free poultry meat with different concentrations and subsequently were tested using the Rapid'Salmonella method. The observed results (detected/not detected) were utilized to fit the probability function of detection (POD). The POD function gives the likelihood of detection, given concentration in the poultry product. The estimated POD functions were integrated into the QMRA model, assessing the performance of the border control for the risk.

In manuscript III, I presented the Quantitative Microbiological Risk Assessment (QMRA) developed to estimate the quantitative effect on the risk of the current and alternative monitoring systems for Salmonella in imported poultry products, using imports to Jordan as a case. The model is a further development of an existing QMRA model for campylobacteriosis due to poultry meat in European Union (EU) countries. The assessment model consists of a consumer phase module, a risk estimation module, and a risk reduction module. The main model outputs were public health impact expressed as the Minimum Relative Residual Risk (MRRR) given the rejection of batches and the percentage of Batches that do Not Comply with the Microbiological Criteria (BNMC).

Strictly speaking, the MRRR is the relative residual risk given border control compared to the size of the risk if no border control was performed. The overall MRRR is based on the weighted sum of risk for each different poultry product from each country. The weight is the relative amount of imported poultry products from each country. The current border control based on one sample per batch gives an overall MRRR value of 10.5%. Alternative border control based on three samples per batch is 4.1% and five samples per batch 2.6%. Theoretically, the model will always reduce risks for consumers by increasing collected samples from each batch at the border control. However, the effect decreases as the number of samples increases. The added effect of increasing the number of samples is primarily due to detecting batches with reducing contamination. However, because the risk originating from these batches is relatively low, risk reduction is limited.

In an attempt to check the model's validity, the estimated model output parameter BNMC was compared with the observed parameter obtained from the national survey observed data. For country and products, updated data about Salmonella's occurrence in poultry products were available, the estimated and the observed BNMC was at the same magnitude, whereas estimated based on old data or even imputed data were significantly different from the observed BNMC. It was surprisingly difficult to obtain data about hazards in poultry products from other countries.

Based on the experience and results obtained in the studies, to strengthen the possibilities to assess the risk reduction obtained by the border control, I think future studies should focus on:

i) Obtaining more valid and up-to-date data about the occurrence of Salmonella in poultry products. Obtaining updated records, including the current prevalence and occurrence of pathogens in food, is considered vital inputs to QMRA models and an accurate estimate of risk and risk reduction. As an alternative to observational studies, a method for expert elicitation could be applied. A combination of expert elicitation and observational studies is probably a robust and efficient approach to obtaining valid data about the occurrence. In addition, there is a need to facilitate data sharing between private and public sectors, both within and between countries.

ii) It can expect that new laboratory technologies will make it possible to investigate the presence of several pathogens in one sample within the near future. These technologies might also have a higher analytical sensitivity. When assessing the potential performance of these new technologies in surveillance, it is necessary to have a holistic view focusing on the risk because many steps of the process food chain industry are interlinked. The QMRA model presented in this thesis and other similar QMRA models can facilitate this holistic approach.

iii) To increase the transfer of knowledge generated in the risk assessment models to decision-makers, it is necessary to make the application of these models more user-friendly. The focus should be on easiness to feed the models with data and no computer coding in the application.

The most important output of the research presented in this PhD thesis is the model framework estimating the effect of current and alternative monitoring approaches used in the border control of imported meat and microbiological hazards. The quantitative estimate of the effect on the human risk aligns with the focus for the decision-makers designing better border control, and there is no need for "extra interpretation" before the model output can be used as decision support.