Foodbook Report

Executive Summary

Outbreaks associated with widely-distributed food products can have important health and economic impacts in Canada. These outbreaks also affect the health of Canadians and may impact the public's confidence in the food safety system. One of the challenges facing the Public Health Agency of Canada (the Agency) in identifying the source of multi-jurisdictional outbreaks has been the availability of food exposure data for the general Canadian population. Similarly, limited food exposure data has impacted the Agency's ability to evaluate risks associated with enteric illness.

To date, foodborne illness outbreak investigators have relied on outdated food exposure data that are not entirely representative of the current Canadian population. In recognizing these challenges, the Agency recommended that a Canadian-specific, consecutive-day food consumption study be conducted to enhance the Agency's ability to investigate and respond to foodborne illness outbreaks across all government levels.

Foodbook is a population-based telephone survey that was conducted in all Canadian provinces and territories over a one-year period with a primary focus on describing what foods Canadians eat over a seven-day period to inform outbreak investigation and response in Canada. The data generated from this study will enhance Canadian public health capacity to take timely and appropriate action in response to foodborne illness outbreaks and will reduce the impact of these events on the health of Canadians.

The Foodbook Report summarizes the food, water and animal exposure information collected during the survey. Foodbook data will be used across federal, provincial and territorial (F/P/T) public health and food safety partners to: inform timely and effective response to foodborne illness outbreaks; determine sources of enteric illness which will support risk assessments and public health interventions to prevent illness; and assist in examining relationships between eating patterns, obesity and socioeconomic status.

The Foodbook study addresses large data gaps that existed for Canadian food, water and animal exposure data for foodborne illness outbreak response investigations. These data are available to F/P/T stakeholders who work together to investigate, control, prevent and understand enteric illness in Canada.

Introduction

Outbreaks associated with widely-distributed food products can have important health and economic impacts in Canada. These outbreaks also affect the health of Canadians and may impact the public's confidence in the food safety system. The Public Health Agency of Canada (the Agency) responds to outbreaks in a timely manner to protect the health of Canadians and to uphold the public trust. A key step in the timely identification of the source of outbreaks is to compare foods eaten by ill individuals affected by the outbreak to foods eaten by the general population. One of the challenges facing the Agency in identifying the source of multi-jurisdictional outbreaks has been the availability of food exposure data for the general Canadian population. Similarly, limited food exposure data has impacted the Agency's ability to evaluate risks associated with enteric illness.

To date, foodborne illness outbreak investigators have relied on food exposure data from the United States' Population Survey Atlas of Exposures^{Footnote 1}  and one regional study in Ontario^{Footnote 2} . These data sources are not entirely representative of current Canadian food exposures and so, more recently, the National Single Day Food Consumption Report^{Footnote 3}  has been used.^Footnotea However, these data have some limitations for outbreak response. They cannot be extended to estimate the consecutive multiple day exposure periods (e.g., 3, 5, or 7 days) needed to compare foods consumed by people experiencing foodborne illness to those consumed by the general population. In addition, the 24-hour recall period limits the ability to capture foods consumed less frequently or rarely consumed.

In order to address these challenges and limitations, the Agency recommended that a Canadian-specific, consecutive-day food consumption study be conducted. These data were collected to enhance the Agency's ability to investigate and respond to foodborne illness outbreaks across all government levels.

Foodbook is a population-based telephone survey that was conducted in all Canadian provinces and territories. It provides essential data on food, animal and water exposure which will be used by the Agency, as well as other federal, provincial and territorial (F/P/T) partners to understand, respond to, control and prevent enteric illness in Canada.

The Foodbook study was developed jointly by the Enteric Surveillance and Population Studies Division (ESPS) and the Outbreak Management Division (OMD) in the Centre for Food-borne, Environmental and Zoonotic Infectious Diseases (CFEZID) of the Public Health Agency of Canada, in consultation with F/P/T partners.  Its primary purpose was to establish a database describing Canadians' exposure to foods over a seven-day period that may serve as sources for foodborne infections.

Foodbook data will be used to:

1. Inform timely and effective response to foodborne illness outbreaks in Canada.
2. Determine sources of enteric illness in Canada which will support risk assessments and help to target public health interventions to prevent illness.
3. Inform food safety education initiatives and health promotion efforts linked to obesity and the determinants of health.

The Foodbook Report summarizes the food, water and animal exposure information collected during the survey. Data are summarized nationally, by province and territory and by a number of demographic characteristics (e.g., gender, income, education level).

Methodology

1. Study Design

Foodbook data were collected in Canada's ten provinces and three territories over a one-year period between 2014 and 2015 using a population-based telephone survey. The study questionnaire was developed in collaboration with F/P/T stakeholders who identified key data gaps within the context of establishing a database of food exposures for outbreak investigations. In addition to food exposures, the study also collected data on drinking and recreational water exposures, animal-related exposures, consumer food safety knowledge and practices, acute gastrointestinal illness, obesity indicators and demographic factors.

All permissions to complete the study were obtained. The study was reviewed and approved by Health Canada and the Agency's Research Ethics Board (REB 2013-0025) as well as the Newfoundland and Labrador Health Research Ethics Authority to meet a unique provincial legal requirement (HREB 13.238).

The sampling frame used to draw telephone numbers consisted of land lines (70% listed, 10% random digit-dialing) and cell phones (20%). Including cell phones ensured better coverage of cell phone-only households and demographic groups more likely to use cell phones exclusively (e.g., younger Canadians).

The recall period for most questions was seven days which aligns with the average incubation period of common foodborne illnesses. Thus, a 'yes' response meant that the respondent had eaten a food item or had an exposure at least once during the seven-day period preceding the call. Refused and 'don't know' responses were recoded to 'missing'.

The survey was pilot tested over a two-week period between April 11 and April 27, 2014. Survey respondents provided feedback about their experience, assessed the Computer Assisted Telephone Interviewing (CATI) programming, identified difficult questions and assessed the effect of the call recording disclaimer.

2. Participant Selection

In order to improve completion rates for younger age groups, when households contained children less than 18 years old, 50% of surveys were conducted with the child who would have the next birthday and 50% were conducted with the adult who would have the next birthday. If there were no children in the household, the survey was conducted with the adult who would have the next birthday. This screening procedure skewed the sample distribution so there were fewer households with adults and children than among the general population. Information on household type (adult/child combination) was thus included in the weight calculations to ensure the weighted sample was representative of the Canadian population.

3. Questionnaire Administration

The surveys were conducted by an independent research company contracted by the Agency. Interviews were administered in English, French, Inuktitut and on-demand verbal translation for other languages. Those who could not speak the supported languages were excluded from the survey. Additionally, in order to ensure that exposures occurred within the province or territory of residence, those who had travelled outside their province or territory of residence during the recall period were excluded. Informed verbal consent was obtained from all study participants and a CATI tool was used to enhance quality control on question-flow and data collection.

Proxy respondents were used in accordance with provincial and territorial laws for participants under a certain age, or for those unable to respond on their own behalf. If the selected individual was under the age of consent, the interview was conducted with the child's parent or legal guardian, or with another adult knowledgeable about the child's activities at the discretion of the parent or legal guardian. Proxy response was also allowed for individuals with medical or activity limitations.

4. Targeted Population Groups

The survey was designed with a target sample size of 11,016 collected evenly over a 12-month period across four age groups (0–9, 10–19, 20–64, 65+ years). All provinces and territories were allocated a minimum sample size of 416 to ensure that estimates from the area and within each of the four key age groups would be of an acceptable quality.^Footnoteb Some provinces were allocated additional samples for the purpose of reducing overall design effects and improving regional analysis within the provinces more frequently involved in multi-jurisdictional outbreaks.

5. Exposures

Major food categories included:

• Vegetables
• Fruit
• Herbs and spices
• Nuts
• Meats
• Fish and shellfish
• Eggs
• Dairy products
• Country foods^Footnotec

Respondents were prompted to include foods eaten raw and/or cooked both inside and outside the home.  Vegetables did not include frozen or canned/jarred vegetables. Fruits did not include frozen, canned/jarred or dried fruits.

Major non-food exposure categories included:

• Drinking water sources
• Contact with recreational water (e.g., lakes, oceans, hot tubs, etc.)
• Contact with household pets and pet food
• Contact with farm animals

6. Weighting

The forward sortation area (FSA, or first three digits of the postal code) was collected for each respondent. The FSA was then converted to the most likely census metropolitan area (CMA). The CMA indicator along with age group, household type, province or territory, the number of people in the household, the number of land lines and cell phones in the household and gender were used to calculate the individual-level survey expansion weight. To create the final weighting variable, a post-stratification step used iterative proportional ranking with available control tables.

7. Analysis

The analysis was performed in Stata 13.0 (Stata Corp., Texas Station, TX) using the survey weight provided by the research company. The survey data file was checked for inconsistencies and estimates were verified against external sources of exposure data, where available.

The analysis in Stata was performed using the expansion weight as the probability weight for mean and variance estimation, via the svyset command and the svy prefix to estimation commands. Primary estimation commands included the "mean" and "proportion" commands.

Results

The full survey (including pilot test) was conducted over a 12-month period from April 11, 2014 to April 9, 2015. Approximately the same number of interviews was completed during each calendar month for a total of 10,942. All sample targets for each province and territory were met (Table 1).

A sample of 515,691 phone numbers from three sources (listed residential numbers, random digit-dialing residential numbers and cell phone numbers) were randomly selected from the total sample. Of these phone numbers, 429,167 were called, of which 119,074 numbers (27.7%) were invalid (e.g., not in service, or modem or fax numbers). Language barriers were identified among 3,894 (0.9%) of respondents and these individuals were not interviewed. The gross refusal rate was 12.7% and the gross response rate and Marketing Research and Intelligence Association (MRIA)^Footnoted response rate were 2.7% and 19.9% respectively.

All consumption estimates included in this report have been weighted so they are representative of the Canadian population.

Food, water and animal exposures as well as the frequency of exposure to select high-risk foods were summarized nationally, by province and territory and by the demographic characteristics presented in Tables 1-9. The data reported in the tables are the results of the Foodbook study and no discussion of the results is presented.

Study Limitations

There are several limitations of the data that should be considered when interpreting the results.

Foodbook data are not representative of populations that were excluded from the study such as individuals who did not have land lines or listed cell phone numbers (e.g., people living in correctional facilities, hospitals, etc.) and respondents who were unable to communicate in the languages available. This may limit the external validity of the study and Foodbook data should not be used to describe exposures explicitly within these populations. Furthermore, Northern Québec, Northern Ontario and the territories (Yukon, Northwest Territories, and Nunavut) have proportionally more residents living in remote rural areas without access to a telephone or a cell phone and therefore, these Canadians are not well-represented.

Although the overall response rate (MRIA) was low, it was consistent with response rates typically found in most commercial telephone surveys in Canada, which range from 10% to 20%^{Footnote 4} . A low response rate may bias results, if those who did not respond had different food exposures than respondents. Several strategies were implemented at the outset to maximize response rates such as: explaining the study to respondents, arranging for call-backs at a more convenient time, and redirecting to a different telephone number to eliminate costs for the respondent. The script and respondent selection process were revised following the pilot. Despite these revisions, there was minimal impact on the response rate compared to the pilot testing period. In order to improve response rates, future surveys may consider a mixed-method sampling strategy that includes both telephone and online components as well as incentive programs.

Bias due to retrospective recall may also impact the results. Respondents were asked about retrospective exposures over a seven-day recall period. It is easier to recall recent exposures (e.g., in the previous 24 hours) than exposures that occurred more than one or two days ago. Although a shorter recall period may have reduced recall bias, a seven-day retrospective recall period was selected because it aligns with most incubation periods for foodborne illness outbreak case interviews. In order to ensure the study data are comparable to outbreak cases, it was critical that participants were asked questions in the same manner as outbreak cases would be asked. Furthermore, a seven-day recall period provides valid estimates of usual exposure within the general population and better estimates of exposure to infrequently-consumed foods that would not be captured using shorter recall periods. It should be noted that the retrospective collection of exposure data is less accurate than prospective collection of data over the same time period. This should be considered when comparing Foodbook data to those collected using recall periods shorter or longer than seven days and/or those using prospective data collection.

Conclusions

The primary purpose of the Foodbook study was to inform timely and effective response to foodborne illness outbreaks in Canada and to maximize the opportunity to address key data gaps identified by the Agency and F/P/T partners.

The Foodbook Report provides rapid access to general population exposure data that will help guide outbreak investigations where the source of the outbreak is not immediately evident. This will help build the weight of evidence required to support the timely removal of contaminated food source(s) from the marketplace. Frequency of exposure to high-risk food items will inform microbial risk assessments and ranking of food-pathogen risk. Water and animal exposure data are provided to inform microbial risk assessments, source attribution analyses and water safety policy at the national level. Finally, the Foodbook Report includes data that will inform Canada-wide cross-disciplinary efforts to elucidate relationships between eating patterns, obesity and socioeconomic status and will strengthen collaboration and leadership in health promotion and illness prevention.

The Foodbook study addresses large data gaps that existed for Canadian food, water and animal exposure data for foodborne illness outbreak response investigations. These data are available to F/P/T stakeholders who work together to investigate, control, prevent and understand enteric illness in Canada.

References

Tables

Table 6: Weighted proportions of food, animal and water exposures in the past seven days, by province/territory and nationally

Table 7: Weighted proportions of food, animal and water exposures in the past seven days, by month