Ten years of Foodbook

Abstract

Background: Enteric illnesses are a preventable cause of morbidity and healthcare utilization in Canada. To support public health and epidemiological activities, Foodbook was launched in 2014 by the Public Health Agency of Canada to collect representative information on food, water, and animal exposures, food safety knowledge, burden of gastrointestinal illnesses, and sociodemographic information. The aim of this overview was to identify how this valuable data source has been used in the past decade since its launch.

Methods: Peer-reviewed and grey literature were identified by applying the search term “Foodbook” to two academic databases and two grey literature sources, respectively. Citations were screened against eligibility criteria. Study information, including study characteristics, module of Foodbook data used, and how Foodbook data was used was extracted and synthesized in tabular format.

Results: A total of 27 articles were identified in the published literature that utilized Foodbook survey data in their analyses. The most common use was for outbreak investigations. In addition, Foodbook has been used to describe food, water, and animal exposures, determine food safety knowledge and practices of Canadians, estimate the burden of acute gastrointestinal illness, and evaluate data collection methods for foodborne illnesses.

Conclusion: By summarizing its use, the authors aim to encourage broader use of this publicly available data source to inform health protection and promotion activities to reduce the burden of enteric illnesses in Canada.

Introduction

Foodborne illnesses are a preventable cause of morbidity and healthcare utilization in Canada, with approximately 4 million annual episodes ^{Footnote 1}. While symptoms are often self-limiting, domestically acquired infections result in over 11,000 hospitalizations and 200 deaths each year ^{Footnote 2}. In Ontario alone, foodborne illnesses contribute to an estimated 137,000 primary care visits, 40,000 emergency department visits, and 6,200 hospitalizations annually, representing an important source of preventable healthcare utilization ^{Footnote 3}.

Foodbook is a population-based survey developed, conducted, and funded by the Public Health Agency of Canada (PHAC) to collect representative information on food, water, and animal exposures to support response to enteric illness outbreaks, inform health protection activities, and identify associations with sociodemographic factors and food safety behaviours ^{Footnote 4}. The first version of Foodbook (Foodbook 1.0) was conducted in all provinces and territories over a one-year period, from April 2014 to April 2015. The sampling frame consisted of cellphones (20%) and landlines (70% listed numbers plus 10% random digit dialing). Interviews were conducted in English, French, and Inuktitut and on demand translation was available for other languages. In addition to sociodemographic data and information on seven-day food consumption, information was also collected on other risk factors, including drinking and recreational water exposures, animal exposures, and food safety knowledge, as well as burden of acute gastrointestinal illnesses ^{Footnote 4}. Those who travelled outside the province or territory during the recall period were excluded. In the first cycle, 10,942 interviews were completed, with an overall response rate of 19.9%. A Foodbook 1.0 public use microdata file (PUMF) is available through the Government of Canada Open Government Portal ^{Footnote 5}.

Data collection for the second version (Foodbook 2.0) was completed from January 2023 to January 2024, online and by telephone, and the second report was released in July 2024 ^{Footnote 6}. This provides a timely opportunity to review how Foodbook data has been utilized to date. This article offers a brief synthesis describing the use of Foodbook data to advance our understanding of the epidemiology of enteric illnesses in the Canadian population, as available in the published literature.

Methods

Peer-reviewed articles were identified by applying the search term “Foodbook” to two academic databases, PubMed and Web of Science, and two grey literature sources, namely Google Scholar and the PHAC website. No date or language limiters were applied. The initial search was conducted in July 2023 and updated in June 2024. Citations were screened according to the following inclusion criteria: 1) Foodbook was used as a data source and 2) the study was peer-reviewed. The initial list of included citations was then reviewed by a PHAC epidemiologist responsible for Foodbook to check for completeness. Study information, including study characteristics, module of Foodbook data used, and how Foodbook data was used was extracted and synthesized in tabular format.

Results

In total, 27 articles were included that used Foodbook for enteric illness-related studies. These studies were broadly categorized into five purposes: outbreak investigations (n=11), exposure types (n=9), food safety knowledge and practices (n=3), methods development (n=3), and burden of acute gastrointestinal illnesses (n=1). An overview of these studies, including the modules of Foodbook utilized, a brief description of how the data was used, and the main findings is provided in Table 1.

Discussion

The most common use of Foodbook is to support outbreak investigations. There are several examples of outbreak investigations successfully using Foodbook food exposure data as reference values to which the reported exposures of cases are compared, including a Salmonella Typhimurium outbreak linked to pre-prepared tofu ^{Footnote 7}, Salmonella Newport infections linked to onions ^{Footnote 8}, cyclosporiasis linked to sugar snap peas ^{Footnote 9}, an Escherichia coli outbreak linked to raw milk Gouda-style cheese ^{Footnote 10}, Salmonella infections of multiple serovar types linked to sprouted chia seed powder ^{Footnote 11}, and Salmonella Enteritidis infections linked to frozen, uncooked, processed chicken ^{Footnote 12}.

In an Escherichia coli outbreak investigation, comparisons were made to the Foodbook reference values and significant differences were observed; however, the source of the outbreak was determined to be kimchi, which is not included in the Foodbook data ^{Footnote 13}. In another study, reference values for food exposures were used during a Salmonella Montevideo outbreak ^{Footnote 14}. While the source was later attributed to contaminated plumbing at a restaurant, it was suggested that the plumbing could have been contaminated by chicken, and indeed, cases were significantly more likely to have consumed chicken at a restaurant compared to the reference values provided by Foodbook.

An investigation into a cyclosporiasis outbreak utilized Foodbook for both reference values and for recruiting controls for a case-control study ^{Footnote 15}. This was made possible by creating a control bank of Foodbook participants who consented to being contacted to assist with future investigations. The authors highlight that having access to this control bank allowed for a case-control study to be conducted in a timely and cost-effective manner.

Two outbreaks of Salmonella Typhimurium utilized Foodbook as part of their investigations, where hedgehogs ^{Footnote 16} and feeder rodents ^{Footnote 17} were eventually identified as the respective causes, demonstrating that Foodbook can be used to investigate outbreaks with animal sources.

Characterizing different exposures

Food exposures

Collection of food exposure information has elucidated the eating patterns of Canadians. For example, ultra-processed foods (UPF) are commonly consumed, with these items being consumed at least weekly by 99.0% of respondents ^{Footnote 18}. Younger age and higher body mass index (BMI) were associated with UPF consumption in both males and females. Fast-food was consumed at least weekly by 48.0% of respondents, with highest fast-food consumption among men and those of younger age ^{Footnote 19}. Sex-specific multivariable logistic regression models highlighted differences between males and females. For example, women in central Canada (compared to the territories), and men with an income of $30,000 to $80,000 (versus a higher income) had higher fast-food consumption ^{Footnote 19}.

Foodbook respondents from Yukon (YT), Northwest Territories (NT) and Nunavut (NU), were also asked questions regarding consumption of country or traditional food items, referring to “food that is trapped, fished, hunted, harvested, or grown for subsistence or medicinal purposes, outside of the commercial food chain” ^{Footnote 34}. Consumption of country food during the previous seven days varied by territory, ranging from 60.7% of respondents in NT to 77.5% in NU ^{Footnote 20}. Overall, this food-specific data can support nutrition and food security research, in addition to outbreak investigations ^{Footnote 20}.

Consumption of foods considered high-risk for enteric illness, such as unpasteurized milk, cheese, or juice, is common among Canadians, with approximately 94% of respondents consuming at least one high-risk food per week, and more than half consuming three or more foods ^{Footnote 21}. Importantly, knowledge of high-risk foods did not influence consumption, suggesting that improving food safety practices may be of greater benefit in reducing the risk of illness associated with these foods.

Water exposures

One study analyzed Canadians’ exposure to drinking and recreational water ^{Footnote 22}. While most Canadians use municipal drinking water (68.5%), 10.8% of respondents indicated that a private well was their primary source of drinking water; however, this figure was as high as 40.2% to 55.6% in the Maritime provinces. Private well use was also highest among rural residents. Further, 18.8% of respondents used bottled water as their primary source, and residents of Saskatchewan (SK) were more likely to consume bottled water ^{Footnote 22}. Regarding recreational water, children aged 0–9 years were most exposed. Unsurprisingly, recreational water exposure was highest across all age groups during the summer months, during which up to 30% of respondents reported recreational water exposure. Together, this data informs private well water testing recommendations and messaging, as well as enhanced public health messaging during the summer months, with specific messaging aimed at parents and caregivers given the higher level of recreational water use among children ^{Footnote 22}.

Animal exposures

A descriptive analysis of animal exposure data showed that most respondents (63.4%) had contact with an animal, animal food or waste, or an animal habitat within the previous seven days ^{Footnote 23}. Animal contact was highest among children, an at-risk group for more serious enteric illness. Overall, dogs and cats were shown to be the most common animal exposure. Given that companion animals are often an overlooked source of zoonotic infections, this analysis provides the basis for continued awareness campaigns to reduce the risk of illness and improved surveillance for identifying outbreaks.

Comparative exposure assessments

Comparative exposure assessments have been conducted utilizing Foodbook to characterize different exposure routes to hepatitis E from pigs or pork ^{Footnote 24}, Campylobacter ^{Footnote 25} and Salmonella ^{Footnote 26}. In the first study, the number of Canadians consuming pork liver supported the hypothesis that a relatively low number of Canadians are at high risk of contracting hepatitis E via this route ^{Footnote 24}. Foodbook was used more extensively in the Campylobacter and Salmonella studies. In both studies, chicken meat was the highest source of foodborne exposures, and the authors discuss how interventions could reduce risk posed by these pathogens ^{Footnote 26}.

Food safety knowledge and practices

Foodbook has been used to explore the food safety knowledge and practices of Canadians ^{Footnote 27}, including those at higher risk for severe enteric illness, namely, children ^{Footnote 28} and seniors ^{Footnote 29}. While most Canadians reported cleaning their hands after handling meat and taking appropriate measures to avoid cross contamination, the use of food thermometers and awareness of specific high-risk foods was low, highlighting the need for improved messaging and education ^{Footnote 27}.

Most seniors followed instructions on food labels and refrigerated leftovers within two hours ^{Footnote 29}. Following instructions on food labels was significantly associated with food safety storage practices. Interestingly, there was no association between knowledge that seniors are at an increased risk of foodborne illness and safe food storage practices, highlighting that food safety knowledge is not an accurate proxy for food safety practices among Canadian seniors ^{Footnote 29}.

Among households with children, higher earners, and those living in urban areas were less likely to wash their hands with soap after handling raw meat ^{Footnote 28}. The practice of separating food in the refrigerator to avoid cross contamination was more likely to be practiced by female caregivers and single-child households. Education level was inversely associated with food safety practices, as those with a bachelor's degree were less likely to refrigerate food within two hours and use a food thermometer. These findings demonstrate key demographic subgroups that may benefit from targeted public health messaging relating to food safety practices.

Methods development in foodborne illness research

Using the Foodbook telephone survey data as a comparator group, the utility of online surveys for collecting food exposure information was investigated by two studies ^{Footnote 30Footnote 31}, which found that online surveys present an efficient, accurate, and lower cost approach for data collection, which may be particularly useful during outbreak investigations when speed of response is important. Differences in exposure information collected online and by telephone may partially be explained by differing recall periods (14-day for the online survey versus 7-day for Foodbook) and changing food preferences ^{Footnote 30}.

The impact of recall periods on reported food exposures and their subsequent utility in identifying outbreak sources was examined by one study, which found no overall difference in reported consumption of most foods when comparing three- and seven-day recall periods ^{Footnote 32}. However, when data from a Salmonella Infantis outbreak was compared to the reference values for both groups, only the three-day recall period group showed a significant result for the identified source of the outbreak, suggesting that a three-day recall period is preferred when investigating illnesses with a short incubation period or where the source is a commonly consumed item ^{Footnote 32}.

A study discussed earlier, which utilized Foodbook data to explore fast food consumption, also demonstrated that question type can influence the reported consumption of fast foods ^{Footnote 19}. In the Foodbook survey, there are three questions relating to fast food consumption; in one question, no examples of fast food were given, whereas in two more detailed questions, examples of specific fast-food types were provided. Respondents who answered the broader question reported less consumption of fast foods compared to those answering either of the two more detailed questions ^{Footnote 19}.

Estimating the burden of acute gastrointestinal illness

Analysis of Foodbook data provided the first estimate of acute gastrointestinal illness burden in Canada, with an estimated 19.5 million episodes each year ^{Footnote 33}. Differences in burden were demonstrated (albeit not significantly) between provinces and territories, highlighting the need to collect nationally representative data. This information, coupled with demonstrated seasonal patterns, can be used in the design of locally tailored public health interventions, and in healthcare services planning ^{Footnote 33}.

Limitations

Foodbook 1.0

The Foodbook 1.0 report lists several limitations, including potential non-response bias due to the exclusion of individuals without a telephone and those unable to communicate in one of the survey languages. This may have disproportionately impacted residents of Northern Canada, where there is a larger proportion of individuals living in remote communities without reliable access to a telephone, affecting representativeness of the data and resultant findings. Other types of bias outlined in the report include recall bias and low response rate ^{Footnote 4}.

Several of the included articles identified additional limitations of Foodbook, including recall bias ^{Footnote 21Footnote 22Footnote 23Footnote 33} and non-response bias ^{Footnote 18Footnote 19Footnote 20Footnote 21Footnote 22Footnote 27Footnote 28Footnote 33}. Several studies discussed the potential for social desirability bias, including in relation to consumption of UPFs ^{Footnote 18}, fast-food ^{Footnote 19}, high risk foods ^{Footnote 21} and food handling practices ^{Footnote 28Footnote 29}. The lack of demographic variables, including ethnicity, pregnancy status, and health status (i.e., whether an individual is considered high risk for enteric illness) also limits exploration of food consumption in particular groups ^{Footnote 20Footnote 21Footnote 27}. The small sample size for some sub-populations, such as parents and caregivers, was also highlighted as a limitation ^{Footnote 28}, and the availability of data regarding education level ^{Footnote 22}. Further, the sample size of respondents selected to participate in the online survey sub-study ^{Footnote 31} and three-day recall period sub-study ^{Footnote 32} were highlighted as limitations.

The utility of Foodbook exposure data is also limited by the specific foods included in the study ^{Footnote 10Footnote 13Footnote 14Footnote 18Footnote 21}. This includes lack of information relating to serving size or frequency of intake, which is of particular importance when attempting to quantify exposure to particular food groups ^{Footnote 18}. For one study, the seven-day recall period selected by Foodbook presented a limitation as this did not match with case interviews, where a 14-day period was chosen to align with the incubation period for cyclosporiasis ^{Footnote 15}. In terms of animal exposure, Foodbook did not capture data regarding where the animal exposure occurred, meaning that it is not possible to draw conclusions about high-risk animal environments ^{Footnote 23}. Lastly, data for the food safety practices module were collected from November to April, and this was identified as a limitation given the potential for food safety practices to vary over time ^{Footnote 27}.

Foodbook 2.0

Foodbook 2.0 includes several key updates. First, a dual sampling frame was utilized with 75% of the sample derived from a list of mailing addresses and the remainder from listed telephone numbers ^{Footnote 6}. To increase the number of children participating, after the fifth month of data collection in households with children, a child was selected to participate 100% of the time. Ethnicity data was also collected as part of Foodbook 2.0.

Additional exposure data included information on special diets, food shopping habits, and information relating to types of water and animal exposures. We also note that additional food items identified as potential sources of enteric illness are included in Foodbook 2.0, expanding the utility of Foodbook as a resource for use in outbreak investigations ^{Footnote 6}.

Conclusion

Foodbook is a valuable data resource for food, water, and animal exposures, food safety behaviours, and socio-demographic factors that increase risk of enteric illnesses in Canada. This resource has supported multiple public health activities, most predominantly, outbreak investigations. In this commentary, we have focused on the use of Foodbook as described in the published literature. However, it is recognized that other uses of Foodbook may not have resulted in publications. By synthesizing published studies, the aim is to increase the visibility of Foodbook as an openly accessible resource for conducting epidemiological studies. In particular, the release of Foodbook 2.0 provides an opportunity to assess how patterns in food, water, and animal-related exposures have changed over time and to update associations with socio-demographic and behavioural factors. Moreover, there is an opportunity to utilize Foodbook to inform food safety and health promotion activities. Lastly, linkage to other data sources, application of a health equity lens to assess differential risk across key socio-demographic groups, and novel use of methodological tools, such as artificial intelligence, will provide further insights to reduce enteric illness burden in Canada.

Authors' statement

• HG — Data curation, writing–original draft
• JH — Data curation
• LEG — Conceptualization, writing–original draft, writing–review and editing, supervision

The content and view expressed in this article are those of the authors and do not necessarily reflect those of the Government of Canada.

Competing interests

The authors have no conflicts of interest to declare.

ORCID numbers

• Heather Grieve — 0000-0001-6905-4718
• Jillian Macleod — 0009-0007-8740-6263
• Lauren E Grant — 0000-0001-6549-0205

Acknowledgements

None.

Funding

HG is funded by the Public Health Agency of Canada Infectious Disease and Climate Change Fund program (Grant Number 2324-HQ-000026 held by LEG). JH was supported by an Undergraduate Research Assistantship from the University of Guelph.

• Previous Page
• Table of contents
• Next page