Unregulated serving sizes on the Canadian nutrition facts table – an invitation for manufacturer manipulations

Background Serving sizes on the Nutrition Facts table (NFt) on Canadian packaged foods have traditionally been unregulated and non-standardized. The federal government recently passed legislation to regulate the serving sizes listed on the NFt. The objective of this study was to compare the serving sizes on food product NFts to the recommendations in the 2003 Nutrition Labelling regulation (Schedule M) reference amounts, the Canadian Food Inspection Agency (CFIA) ranges, and Canada’s Food Guide recommendations. An additional objective was to determine if food and beverage products that report smaller serving sizes have a higher calorie density, compared to similar products with a larger serving size. Methods Data for 10,487 products were retrieved from the 2010 Food Label Information Program (FLIP) database and categorized according to Schedule M categories. Correlations between calorie density and manufacturer stated serving size were tested and the proportion of products meeting recommendations were tabulated. Results 35% of products had serving sizes on the NFt that were smaller than the Schedule M reference amount and 23% exceeded the reference amount. 86% of products fell within the CFIA’s recommended serving size ranges; however, 70% were within the lower-half of the range. Several bread and juice categories exceeded CFG’s recommendations, while several dairy product categories were smaller than the recommendations. Of the 50 Schedule M sub-categories analyzed, 31 (62%) exhibited a negative correlation between serving size and calorie density. Conclusion While most products fell within the CFIA’s recommended serving size ranges, there was a tendency for products with a higher calorie density to list smaller serving sizes. Electronic supplementary material The online version of this article (doi:10.1186/s12889-017-4362-0) contains supplementary material, which is available to authorized users.


Background
In recent years there has been a substantial increase in the prevalence of obesity in Canada. Presently 62.1% of Canadian adults are overweight, and 25% are obese [1]. The rise in obesity has been paralleled by the consumption of excess calories, partially due to increased portion sizes [2].
The Nutrition Facts table (NFt) is mandated to appear on nearly all packaged foods sold in Canada [3]. The serving size stated on the Nutrition Facts table determine the nutrient levels that will be reported on that label (for example, a smaller serving size reports fewer calories, while a larger serving size reports more calories). Traditionally, the serving sizes stated on the NFt on packaged foods sold in Canada were not standardized and could be determined by manufacturers, unlike in the United States, where the FDA regulates serving sizes [4]. Therefore, food companies could decide the serving size, and thus the number of calories a consumers sees when looking at a Nutrition Facts table. In other countries and jurisdictions, such as the EU, UK, and Australia, nutrition information is listed per 100 g to enable comparisons among similar products [5,6]. The Canadian NFt does not feature nutrition information per 100 g.
Research has demonstrated that the reported serving sizes on NFts are often smaller than the portions typically consumed [7]. This suggests that food companies may be intentionally trying to reduce the reported calories on the nutrition label by using smaller serving sizes [8]. Additionally, research has demonstrated that using different serving sizes on the NFts of similar products, confuses consumers and makes comparisons among similar foods difficult. As a result, consumers have difficulty determining the energy content per serving and per package, and cannot accurately calculate calorie content when there is more than one serving per container [9]. Furthermore, anticipated guilt from consumption, purchase intentions, and choice behaviour, can be influenced by serving size manipulations, and may disproportionately influence weight-conscious consumers who are concerned about calories, but not serving size [8].
In Canada, two important government bodies i) Health Canada and ii) the Canadian Food Inspection Agency (CFIA) are responsible for Canadian food labelling regulations and public governance. Health Canada is responsible for administering the provisions of the Food and Drugs Act (FDR) that relate to public health, safety and nutrition [10]. Whereas the CFIA provides all federal inspection services related to food and enforces the food safety and nutritional quality standards established by Health Canada, i.e. responsible for the administration and enforcement of the Consumer Packaging and Labelling Act related to food [11]. The CFIA regulates the consistency, completeness and accuracy of the labelling and packaging of consumer goods. These regulations are intended to provide a fair and competitive marketplace by prohibiting deceptive labelling or advertising practices.
Reference amounts for the serving size on Nutrition Facts tables have been established by Health Canada and are set out in Schedule M of the Food and Drug Regulations (FDR) (B.01.001) [12]. Traditionally, these reference amounts were mandatory only as the basis for calculating the compositional criteria that manufacturers must meet for nutrient content claims and health claims [3]. For products without any nutrient content claims and health claims, the CFIA recommends manufacturers follow the range of serving sizes set in the CFIA Guide to Food Labelling and Advertising (CFIA guide), however, these ranges are not mandatory and only serves as a reference for manufacturers to stay within the recommended ranges [13]. In comparison, in the United States, standardized serving sizes used on the Nutrition Facts table have been regulated by the Food and Drug Administration (FDA) for more than 20 years and are required to conform to the Reference Amount Customarily Consumed (RACC) defined in section 101.12(b) of the food labelling regulations [14].
It was suggested that standardizing the serving sizes reported on the Canadian NFt could be an important policy intervention to help consumers make informed healthy food choices [3]. In December 2016, changes were made to the Food and Drug regulations in Canada that now require food manufacturers to use similar serving sizes for similar products [10,15]. However, this new regulations will not be fully implemented until 2021. By modifying serving sizes to be more consistent and listing realistic measures, it is expected that Canadians will be more easily able to compare similar foods and make it easier to understand how many calories and nutrients they are consuming. This study was initiated before the new legislation was announced. Therefore, the objective of this study was to compare the serving sizes on food product NFTs to the recommendations in the 2003 Nutrition Labelling regulation (Schedule M) reference amounts, the CFIA recommended ranges, and Canada's Food Guide (CFG) recommendations. The goal was to determine the number of foods that currently adhere to the voluntary Canadian FDR Schedule M serving size recommendations (reference amount) as well as the CFIA recommended serving ranges. Comparing serving sizes on food product NFTs to the CFG recommendations is needed to investigate the consistency between the serving sizes recommended in regulatory documents versus consumer education tools for healthy eating. Our second aim was to determine if food and beverage products that have a higher calorie density report a smaller serving size on the NFt, when compared to similar products with a larger serving size. Overall, the aim is for these results to shed light on the potential benefits of the new nutrition labelling changes, to be implemented on Canada, over the next five years.

Data collection
This was a cross-sectional analyses of the serving size and calories listed on the NFt on 10,487 packaged foods from Canadian grocery stores. Canadian food package label information, as reported on the NFt, was retrieved from the 2010 Food Label Information Program (FLIP) database at the University of Toronto [16]. All data were collected between March 2010 and April 2011 from outlets of the four largest grocery chains in Canada (Loblaws, Metro and Sobeys in Ontario) and one major western Canadian grocery retailer (Safeway, in Alberta). These chains represented approximately 75% of the market share of grocery food products sold in Canada; therefore, most national and private label branded food products were collected. A total of 10,487 unique food products were in the FLIP database. Additional details concerning the construction of the FLIP 2010 database can be found elsewhere [16]. . The reference amount is a specific regulated quantity of food (measured in grams) and it is meant to represent the portion that would typically be eaten by an individual at one sitting, but is not required to be used by manufacturers on the NFt (Additional file 1).
The CFIA guide provides a range of suggested serving sizes within each of the Schedule M subcategories to guide manufacturer determined serving sizes [3]. The ranges are meant to give manufacturers flexibility when determining the appropriate serving size to disclose on a product's NFt, however, manufacturers are not required to follow these serving size reference amounts. Use of reference amounts are only mandatory as the basis for determining eligibility of a food to carry nutrient content claims and health claims.
In order to compare manufacturer stated serving sizes to a standardized serving size, schedule M reference amounts were assigned to each food product based on the sub-category that best matched the product's description. To ensure that food items were categorized consistently, data were checked by a second independent reviewer. In any case of discrepancies, the CFIA was contacted to verify categorizations. A description of the Schedule M subcategories and the food products within each subcategory can be found in Additional file 1.

Inclusion and exclusion criteria
Of the 153 schedule M sub-categories, all sub-categories with at least 50 unique food items were included in this analysis leaving a total of 7494 foods for analysis. For categories with less than 50 food items, sample size might be too small to reflect all existing products across the country. Thus, excluding those categories might help reduce selection bias.
Data analysis Descriptive statistics were calculated for the serving size and calorie content listed on the NFt (according to the manufacturer stated serving size). The proportion of products with serving sizes that were less than, equal to, or greater than the reference amount listed in Schedule M were tabulated. The proportions of products with serving sizes below, within, and above the range of recommended serving sizes set out in the CFIA Guide were also tabulated. Additionally, when a product's serving size was within the CFIA range of recommended serving sizes, the proportion of products in the lower half and upper half of the range was calculated.
Each product's calorie density (calculated as calories per 100 g and calories per reference amount) were calculated.
For each product, scatter plots for the calories per reference amount in comparison to the food product's stated serving size were created to study the association between calorie density and serving size. Correlations between calorie density and serving size were tested using Pearson correlation.
The sign test was used to detect differences between the calories per serving and calories per reference amount, within each food category. All statistical analyses were performed using Statistica, version 10 (Tulsa, OK). A p-value <0.05 was considered significant.

Results
Fifty schedule M sub-categories containing 7494 food products were analysed. The 50 sub-categories included in this study are listed in Table 1.
Comparison of serving sizes in relation to schedule M reference amounts Table 1 compares the manufacturer stated serving sizes reported on the NFt with the Schedule M reference amount and CFIA recommended serving size ranges. 35% of products had serving sizes that were lower than the reference amount in schedule M, 42% of products had serving sizes that were consistent with the reference amount, and 23% exceeded the reference amount. In nine categories, (representing 18% of all categories) more than 70% of products had serving sizes that were smaller than the reference amount. The nine categories were 'French toast, pancakes, and waffles'; 'Pies, tarts, cobblers, turnovers' ; 'Quark, fresh cheese and fresh dairy desserts'; 'Yogurt'; 'Dairy desserts, frozen'; 'Dressings for salad'; 'Marine and fresh water animals'; 'Fruit, fresh, canned or frozen' and 'Meat and poultry with sauce'. Furthermore, in an additional twelve categories, (representing 24% of all categories) more than half of the products had serving sizes that were smaller than the reference amount.
Comparison of serving sizes in relation to CFIA recommended serving size ranges When compared to the CFIA recommended serving size ranges, 10% of products had manufacturer stated serving sizes that were smaller than the recommended range, 86% were within the recommended range, and 4% were larger than the recommended range. However, among products whose serving size fell within the recommended range, 70% fell within the lower-half of the recommended range, while 30% fell within the higher half of the recommended range.

Comparison of serving sizes in relation to Canada's food guide
Only a limited number of categories could be compared to the Canada Food Guide recommended serving sizes due to different food categorization systems. In addition, Table 1 The proportion of products with manufacturer stated serving sizes that are equal to, less than, or greater than the Schedule M reference amounts and recommended ranges The CFIA guide suggests recommended serving size ranges for each Schedule M food category, but manufacturer's adherence to these ranges is not mandatory [13] the recommended serving sizes in CFG were primarily based on cooked food portions, whereas the serving size on food product NFts were based on raw food portions, this further limited the number of categories that could be compared. In 'bread'; 'bagels, tea biscuits, scones'; and 'juices, nectars and fruit drinks' , 80-90% of products had manufacturer stated serving sizes that were higher than those recommended by Canada's food guide. Meanwhile in 'cheese ' and 'quark, fresh cheese and fresh dairy desserts' , 90-99% of products had manufacturer stated serving sizes that were smaller than the Canada Food Guide Recommended Serving Sizes.
Comparison of calorie content currently listed on the NFt versus the reference amount

Correlation between calorie density and serving size
There was a significant negative correlation between serving size and calorie density in 31 categories (62% of categories) (Table 3). In 22 of these categories (44%), the negative correlation was significant (p < 0.05). 'Juices, nectars and fruit drinks' showed the strongest negative correlation (−0.9, p < 0.0001) while 'Croutons' , 'Quark, fresh cheese and fresh dairy desserts' , 'Plant-based beverages' , 'Measurable combination dishes' , 'Not measurable combination dishes' and 'Major main entree with sauce' also showed a significant negative correlations ranging from −0.4 to −0.6 (p < 0.001).

Discussion
This study demonstrates that 35% of Canadian food products had manufacturer stated serving sizes that were lower than the Schedule M reference amount. While many products fell within the CFIA recommended serving size ranges (which are quite large), the majority (70%) were within the lower half of the range. Furthermore, in the majority of food categories, products with a smaller manufacturer stated serving size tended to have a higher calorie density. Therefore, the lack of regulated serving sizes on the NFt on packaged foods in Canada has led to a tendency for food manufacturers to state smaller serving sizes and consequently display lower calorie levels, particularly in high calorie density foods. Collectively, these results suggest that there is an urgent need to regulate and standardize the NFt serving sizes, as the current unregulated system has led to a large proportion of food products with a higher calorie density to report smaller serving sizes, which can be misleading to consumers. These findings are concerning because it has been shown that only knowledgeable consumers will be motivated to spend time analyzing nutrition information accurately and few are able to do the calculations necessary to compare products with different serving sizes [13]. These results also illustrated the very wide range of serving sizes (some as high as ten-fold) within categories, used by manufacturers in Canada. Health Canada consumer research has shown that consumers find it difficult to compare products, particularly when different serving sizes are used on the Nft [13]; thus consumers may be falsely led to believe that they are consuming fewer calories, when in fact, they are simply eating less food. Data illustrate that the current non-standardized serving size system in Canada is confusing and can lead to dramatic underestimation of calorie intakes [7,17]. Additionally, this is worrisome, because research has highlighted that certain consumers, such as those who are sensitive to potentially negative nutrients (such as calories), as well as those with less knowledge of nutrition, are likely to be most susceptible to serving size manipulations [18].
This study also illustrates the need to update the serving size recommendations and ranges outlined in Schedule M, to be more in line with the serving size recommendations in Canada's Food Guide. For example, in the 'juices, nectars and fruit drinks' category, most of the product's serving sizes were in agreement with the reference amount, yet greater than 84% of products exceeded the recommended serving size in CFG. This finding illustrates the disparities between the serving sizes recommended in regulatory documents versus consumer education tools for healthy eating. Therefore, while schedule M and the CFIA make their recommendations based on what is typically consumed, this may not reflect what is recommended in Canada's Food Guide. Using the amount typically consumed, rather than the recommended serving sizes, as the criteria for labelling, may in fact, promote increased serving sizes and food intakes and contribute to the increasing rate of obesity in Canada.  In two instances the p-value is significant but there is no difference in the median, this is due to the fact that signficance was determined according to the sign-test Standardizing serving sizes as well as aligning them with recommended servings in Canada's Food Guide, is only one potential solution to this problem. For example, in the food regulations set out by the Food Standards Australia New Zealand, products are required to present nutrient levels both per serving size and per 100 g/mL using a dualcolumn system, thus enabling comparisons amongst products irrespective of their serving sizes [17]. The EU similarly avoids the need to regulate serving sizes by reporting nutrient levels per 100 g [19]. Interestingly, the "Labelling Logic Review" in Australia, recommended that serving sizes be removed from the Nutrition Information Panel (NIP), aiming to simplify requirements for the mandatory NIP and reduce the regulatory burden on industry [20]. However, no further work has been be undertaken on this recommendation due to the perceived lack of benefit [21]. The Public Health Association of Australia (PHAA) stated that removing the serving size column would not solve the problem of consumer confusion and recommended that the only approach to dealing with the inconsistency in serving sizes is to mandate serving sizes within food categories, as is being currently implemented in Canada [15]. Furthermore, the more fundamental question is, what types of nutrition label information actually assists consumers to make healthier food choices? For example, Roberto and Khandpur had suggested package design might also help educate consumers about appropriate serving sizes by having markers on the outside of food packaging that denote serving size amounts; or having clear indicators of pre-portioned servings in the package design [22]. Not to mention, effective consumer education is an essential co-requirement to enable consumers to understand the valuable information on the NFt.
This study evaluated a large number of foods from a wide variety of food categories. Limitations include the fact that Schedule M serving sizes were not available for a number of sub-categories. In addition, our study only investigated calories, and did not analyze other nutrient levels in relation to the manufacturer stated serving size. Hunter et al. noted that discrepancies in serving-size are often attributed to the use of food products for different purposes [23], thus a higher serving size could be advantageous if the manufacturer inflates the content of a healthy nutrient. Our study did not investigate other factors that could motivate serving size manipulations.