Foodborne pathogens are an important cause of human disease worldwide. In the last decades, several industrialized countries have made efforts to prevent and control foodborne zoonoses. To identify and prioritize options for control in the food chain, it is important to identify the most important sources of the pathogen and estimate the relative contribution of each source to the burden of human disease. A variety of methods to attribute human illness to specific sources is available, including microbial subtyping, comparative exposure assessment, analysis of studies of sporadic infections, and analysis of data from outbreak investigations. All methods present advantages and limitations, and the usefulness of each depends on the public health questions being addressed.
The overall aim of this thesis was to apply different approaches to attribute human salmonellosis and campylobacteriosis to the responsible sources, compare methods and results, and make recommendations on which approach to choose to answer specific risk management questions. The work is based on six manuscripts. In the first manuscript, concepts and terminology on source attribution are presented, and the remaining manuscripts present the source attribution approaches applied to Salmonella and Campylobacter. The methods were applied to different populations and time periods, and different models attributed illness in Denmark, in Europe or globally.
Two microbial subtyping approaches were applied to attribute human salmonellosis in (1) Denmark, and (2) other European countries. In the first, human Salmonella infections reported in Denmark in a three-year period were attributed to the responsible sources using a three-dimension (3D) microbial subtyping approach that constitutes a methodological improvement of the model developed by Hald et al. (2004). The most important sources of salmonellosis in the country were estimated to be table-eggs, imported chicken and pork, and results showed that travel is the single most important cause of salmonellosis in Denmark. S. Newport, S. Virchow, S. Thompson and S. Enteritidis were found to be the Salmonella serotypes with highest impact on the number of reported cases. The 3D model achieved more robust estimates of the proportion subtype-related factors that describe the ability of different Salmonella subtypes to cause disease and, consequently, of the proportion of disease attributed to each source. In addition, the developed model can be applied to data with a lower discriminatory power (e.g. only serotyping information), which is the only data available in many countries, and thus can be used in several countries throughout the world. In the second approach, the original model as described by Hald et al. (2004) was adapted to Salmonella surveillance data from Sweden, the United Kingdom, the Netherlands, Germany and Poland. The adaptations of the model were successful for the first three countries. The relative importance of sources varied between countries, reflecting differences in data availability, in the coverage of surveillance systems, in the subtypes causing reported disease, and in the epidemiology of salmonellosis in each country.
A comparative exposure assessment approach was applied to attribute human salmonellosis in Denmark. The model investigated exposure through 25 foodborne transmission routes and 18 direct contact routes in the overall Danish population and in four risk groups. The risk groups (RGs) encompassed (1) people consuming raw cow milk or unpasteurized dairy products, (2) people consuming raw goat milk and unpasteurised goat and sheep products, (3) people having daily direct contact with farm animals, and (4) children visiting a petting zoo. Results suggest that individuals belonging to the risk groups were more exposed than the overall population, specifically (in presented order) individuals from RG1 with between 5 and 14 years of age, and other age categories within RG1, RG2 and RG3. Within the overall population, people aged between 25 and 64 years were estimated to be more exposed to Salmonella, and the most important routes of exposure were direct contact with pets, consumption of table eggs, domestic pork, and fruits eaten not peeled.
A systematic review of case-control studies collected and analysed studies of sporadic campylobacteriosis and salmonellosis published throughout the world in the last two decades. The objective of the study was to identify the most important risk factors for the two diseases. Data was extracted from 71 studies, 34 investigating risk factors for human salmonellosis and 37 focusing on campylobacteriosis. We estimated that eating undercooked chicken, travelling abroad and eating raw eggs are the most important risk factors for salmonellosis, whereas travelling abroad, daily contact with a pet and eating undercooked chicken were estimated to be the most important significant risk factors for campylobacteriosis. Sub-analyses of risk factors by geographic region, age group, and study period suggested that the major sources and risk factors of Salmonella and Campylobacter are similar in different populations and time periods, although they vary to some extent in order of importance.
The forth approach applied for source attribution of human foodborne disease was an analysis of data from outbreak investigations. The method used data from investigations of Salmonella and Campylobacter outbreaks reported in European countries in 2005 and 2006, provided by the European Food Safety Authority (EFSA). Reporting of the causative vehicles of outbreaks was not harmonized between and within countries, and therefore we organized the implicated foods in mutually exclusive food categories based on the implicated simple foods (i.e. foods belonging to one single food category) and complex foods (belonging to multiple food categories). We estimated that the most important food sources for human salmonellosis were eggs, meat and poultry-meat, and the majority of the cases of campylobacteriosis were attributed to chicken. For both pathogens, a large proportion of cases could not be linked to any source. Results also revealed regional differences in the relative importance of specific sources.
All methods were discussed and compared in light of data requirements, methods’ characteristics and utility of the results. A comparison of all obtained results revealed that the relative importance of the most important sources as estimated in all methods (four methods for salmonellosis and two for campylobacteriosis) was in agreement, suggesting that all methods constitute useful tools to attribute human foodborne illness. Furthermore, some methods investigated the role of non-food routes of transmission (e.g. direct contact with animals and environmental transmission), and results showed that routes other than foodborne should be considered when identifying and prioritizing control measures to prevent and reduce the human incidence.
Finally, the different approaches revealed useful to answer different public health questions, and the integration of results from more than one method is expected to add insight on the contribution of different sources for the human burden of disease and to strengthen confidence in the results. We present recommendations on how to choose one or more source attribution methods to address different risk management problems. For this purpose, a “source attribution tool-box” based on the experiences from this PhD was developed.