A content and quality analysis of free, popular mHealth apps supporting 'plant-based' diets.

There has been an increased emphasis on plant-based foods and diets. Although mobile technology has the potential to be a convenient and innovative tool to help consumers adhere to dietary guidelines, little is known about the content and quality of free, popular mobile health (mHealth) plant-based diet apps. The objective of the study was to assess the content and quality of free, popular mHealth apps supporting plant-based diets for Canadians. Free mHealth apps with high user ratings, a high number of user ratings, available on both Apple App and GooglePlay stores, and primarily marketed to help users follow plant-based diet were included. Using pre-defined search terms, Apple App and GooglePlay App stores were searched on December 22, 2020; the top 100 returns for each search term were screened for eligibility. Included apps were downloaded and assessed for quality by three dietitians/nutrition research assistants using the Mobile App Rating Scale (MARS) and the App Quality Evaluation (AQEL) scale. Of the 998 apps screened, 16 apps (mean user ratings±SEM: 4.6±0.1) met the eligibility criteria, comprising 10 recipe managers and meal planners, 2 food scanners, 2 community builders, 1 restaurant identifier, and 1 sustainability assessor. All included apps targeted the general population and focused on changing behaviors using education (15 apps), skills training (9 apps), and/or goal setting (4 apps). Although MARS (scale: 1-5) revealed overall adequate app quality scores (3.8±0.1), domain-specific assessments revealed high functionality (4.0±0.1) and aesthetic (4.0±0.2), but low credibility scores (2.4±0.1). The AQEL (scale: 0-10) revealed overall low score in support of knowledge acquisition (4.5±0.4) and adequate scores in other nutrition-focused domains (6.1-7.6). Despite a variety of free plant-based apps available with different focuses to help Canadians follow plant-based diets, our findings suggest a need for increased credibility and additional resources to complement the low support of knowledge acquisition among currently available plant-based apps. This research received no specific grant from any funding agency.

Accelerating the Classification of NOVA Food Processing Levels Using a Fine-Tuned Language Model: A Multi-Country Study.

The consumption and availability of ultra-processed foods (UPFs), which are associated with an increased risk of noncommunicable diseases, have increased in most countries. While many countries have or are planning to incorporate UPF recommendations in their national dietary guidelines, the classification of food processing levels relies on expertise-based manual categorization, which is labor-intensive and time-consuming. Our study utilized transformer-based language models to automate the classification of food processing levels according to the NOVA classification system in the Canada, Argentina, and US national food databases. We showed that fine-tuned language models using the ingredient list text found on food labels as inputs achieved a high overall accuracy (F1 score of 0.979) in predicting the food processing levels of Canadian food products, outperforming traditional machine learning models using structured nutrient data and bag-of-words. Most of the food categories reached a prediction accuracy of 0.98 using a fined-tuned language model, especially for predicting processed foods and ultra-processed foods. Our automation strategy was also effective and generalizable for classifying food products in the Argentina and US databases, providing a cost-effective approach for policymakers to monitor and regulate the UPFs in the global food supply.

The Effects of Ontario Menu Labelling Regulations on Nutritional Quality of Chain Restaurant Menu Items-Cross-Sectional Examination.

Restaurant foods are associated with excessive energy intake and poor nutritional quality. In 2017, the Healthy Menu Choices Act mandated food service establishments with ≥20 outlets in Ontario to display the energy content on menus. To examine the potential impact of menu labelling, nutrition information for 18,760 menu items were collected from 88 regulated and 53 unregulated restaurants. Descriptive statistics were calculated for serving size, energy, saturated fat, sodium and total sugars. Quantile regression was used to determine the differences between regulated and unregulated restaurants. The energy content of menu items from regulated restaurants (median (95% CI): 320 kcal (310, 320)) was significantly lower than those from unregulated restaurants (470 kcal (460, 486), p < 0.001). Saturated fat, sodium and total sugars were significantly lower in regulated restaurants (4 g (4, 4), 480 mg (470, 490) and 7 g (6, 7), respectively) than in unregulated restaurants (6 g (6, 6), 830 mg (797, 862) and 8 g (8, 9), respectively, p < 0.001). This study showed that menu items from regulated restaurants had smaller serving size, lower levels of energy and nutrients of public health concern compared to those from the unregulated restaurants, suggesting potential downstream beneficial effects of menu labelling in lowering caloric content and nutrients of public health concern in foods.

The estimated dietary and health impact of implementing the recently approved 'high in' front-of-package nutrition symbol in Canada: a food substitution scenario modeling study.

Background: Front-of-pack labeling (FOPL) has been identified as a cost-effective policy to promote healthy food environments and to help consumers make healthier food choices. Consumer surveys report that after implementation of mandatory 'high in' FOPL symbols between 30 and 70% of consumers choose or were willing to choose products with fewer 'high in' symbols. Health Canada has recently published FOPL regulations that will require prepackaged food and beverages that meet or exceed thresholds for sodium, total sugars, or saturated fat to display a 'high in' FOPL nutrition symbol. Objectives: The aims were to estimate the potential (1) dietary impact of substituting foods with similar foods that would display at least one less 'high in' symbol, and (2) the number of diet-related noncommunicable disease (NCD) deaths that could be averted or delayed due to estimated dietary changes. Methods: Baseline and counterfactual intakes of sodium, total sugars, saturated fats, and energy were estimated among Canadian adults (n = 11,992) using both available days of 24 h-recall data from the 2015 Canadian Community Health Survey-Nutrition (CCHS). Similar foods to those reported in CCHS that would display at least one less 'high in' symbol (n = 239) were identified using a Canadian branded food composition database. Based on current FOPL consumer research, identified foods were substituted for 30, 50, and 70% of randomly selected CCHS-Nutrition adult participants and for all adult participants. Potential health impacts were estimated using the Preventable Risk Integrated ModEl. Results: Mean dietary reductions of between 73 and 259 mg/day of sodium, 2.0 and 6.9 g/day of total sugars, 0.2 and 0.5 g/day of saturated fats, and 14 and 46 kcal/day of energy were estimated. Between 2,148 (95% UI 1,913-2,386) and 7,047 (95% UI 6,249-7,886) of deaths due to diet-related NCDs, primarily from cardiovascular diseases (70%), could potentially be averted or delayed if Canadians choose products with fewer 'high in' symbols. Conclusion: Results suggest that FOPL could significantly reduce sodium and total sugar intakes among Canadian adults, the consequences of which could avert or delay an important number of diet-related NCD deaths. These findings provide relevant data to support the importance of the impending FOPL regulations.

Development of the Canadian Eating Practices Screener to assess eating practices based on 2019 Canada's Food Guide recommendations.

In 2019, Health Canada released a new iteration of Canada's Food Guide (2019-CFG), which, for the first time, highlighted recommendations regarding eating practices, i.e., guidance on where, when, why, and how to eat. The objective of this study was to develop a brief self-administered screener to assess eating practices recommended in the 2019-CFG among adults aged 18-65 years. Development of the screener items was informed by a review of existing tools and mapping of items onto 2019-CFG recommendations. Face and content validity were assessed with experts in public health nutrition and/or dietary assessment (n = 16) and individuals from Government of Canada (n = 14). Cognitive interviews were conducted with English-speaking (n = 16) and French-speaking (n = 16) adults living in Canada to assess face validity and understanding of the screener items. While some modifications were identified to improve relevance or clarity, overall, the screener items were found to be relevant, well-constructed, and clearly worded. This comprehensive process resulted in the Canadian Eating Practices Screener/Questionnaire court canadien sur les pratiques alimentaires, which includes 21 items that assess eating practices recommended in the 2019-CFG. This screener can facilitate monitoring and surveillance efforts of the 2019-CFG eating practices as well as research exploring how these practices are associated with various health outcomes.

Nutritional quality and price of plant-based dairy and meat analogs in the Canadian food supply system.

There has been an increased consumer interest and public health emphasis on plant-based protein foods, resulting in a rise in the availability of highly processed plant-based analogs. The objectives of this study were to assess the nutritional quality and the price of plant-based dairy and meat analogs compared to their respective animal-derived products and to examine the association between processing levels and the nutritional quality among these products. Using a branded food composition database, products in cheese, yogurt, milk, and meat categories were examined (n = 3231). Products were categorized as plant-based analogs versus animal-derived products using the ingredient list. Products were examined for their nutrient content, overall nutritional quality using the Food Standards Australia New Zealand nutrient profiling model, price, and processing levels using the NOVA classification. All plant-based analogs had lower protein and higher total carbohydrate, sugar, and fiber content compared to their respective animal-derived products. Compared to their respective animal-derived products, plant-based milk and meat analogs had lower energy, total fat, and saturated fat content; plant-based yogurt and meat analogs had lower sodium content; and all plant-based dairy analogs had lower calcium content. Plant-based cheese and yogurt analogs were more expensive than animal-based products; however, there was no significant difference among milk and meat products. There was no association between processing levels and overall nutritional quality among dairy and meat products. Plant-based analogs may be part of a healthy and affordable diet to reduce the intakes of nutrients of concern; however, additional compositional guidelines and/or labeling may be needed to highlight the differences in the levels of nutrients to encourage.

Examining the diet quality of Canadian adults and the alignment of Canadian front-of-pack labelling regulations with other front-of-pack labelling systems and dietary guidelines.

Introduction: Canada promulgated mandatory front-of-pack labelling (FOPL) regulations in 2022, requiring pre-packaged foods meeting and/or exceeding recommended thresholds for nutrients-of-concern (i.e., saturated fat, sodium, sugars) to display a "high-in" nutrition symbol. However, there is limited evidence on how Canadian FOPL (CAN-FOPL) regulations compare to other FOPL systems and dietary guidelines. Therefore, the objectives of the study were to examine the diet quality of Canadians using the CAN-FOPL dietary index system and its alignment with other FOPL systems and dietary guidelines. Methods: Nationally representative dietary data from the 2015 Canadian Community Health Survey-Nutrition survey (n = 13,495) was assigned dietary index scores that underpin CAN-FOPL, Diabetes Canada Clinical Practice (DCCP) Guidelines, Nutri-score, Dietary Approaches to Stop Hypertension (DASH) and Canada's Food Guide (Healthy Eating Food Index-2019 [HEFI-2019]). Diet quality was examined by assessing linear trends of nutrient intakes across quintile groups of CAN-FOPL dietary index scores. The alignment of CAN-FOPL dietary index system compared with other dietary index systems, with HEFI as the reference standard, was examined using Pearson's correlations and к statistics. Results: The mean [95% CI] dietary index scores (range: 0-100) for CAN-FOPL, DCCP, Nutri-score, DASH, and HEFI-2019 were 73.0 [72.8, 73.2], 64.2 [64.0, 64.3], 54.9 [54.7, 55.1], 51.7 [51.4, 51.9], and 54.3 [54.1, 54.6], respectively. Moving from the "least healthy" to the "most healthy" quintile in the CAN-FOPL dietary index system, intakes of protein, fiber, vitamin A, vitamin C, and potassium increased, while intakes of energy, saturated fat, total and free sugars, and sodium decreased. CAN-FOPL showed moderate association with DCCP (r = 0.545, p < 0.001), Nutri-score (r = 0.444, p < 0.001), and HEFI-2019 (r = 0.401, p < 0.001), but poor association with DASH (r = 0.242, p < 0.001). Slight to fair agreement was seen between quintile combinations of CAN-FOPL and all dietary index scores (к = 0.05-0.38). Discussion: Our findings show that CAN-FOPL rates the dietary quality of Canadian adults to be healthier than other systems. The disagreement between CAN-FOPL with other systems suggest a need to provide additional guidance to help Canadians select and consume 'healthier' options among foods that would not display a front-of-pack nutrition symbol.

Adherence to a priori dietary patterns in relation to obesity: results from two cycles of the Canadian National Nutrition Survey.

OBJECTIVE: To test whether adherence to the Mediterranean diet, the Dietary Approaches to Stop Hypertension (DASH) or a dietary pattern in-line with the 2015-2020 Dietary Guidelines for Americans (DGA) was associated with obesity. DESIGN: 24-h dietary recall data from the Canadian Community Health Survey (CCHS)-Nutrition, 2004 and 2015 cycles, were analysed. Diet quality index scores were computed for the Mediterranean-Style Dietary Pattern Score (MSDPS), a DASH index and the 2015 Dietary Guidelines for Americans Adherence Index (DGAI). Higher scores indicated greater adherence. Association between scores and obesity was examined using logistic regression, adjusting for age, sex, physical activity, smoking status, sequence of dietary recall and alcohol and energy intake. SETTING: Canada (excluding territories and the institutionalised population). PARTICIPANTS: Canadian adults (≥ 18 years), non-pregnant and non-breast-feeding; 11 748 from CCHS 2004 and 12 110 from CCHS 2015. The percentage of females in each sample was 50 %. RESULTS: Mean MSDPS, DASH and DGAI scores were marginally but significantly higher in CCHS 2015 than in CCHS 2004. Those affected by obesity obtained lower scores for all indexes in CCHS 2004 (OR 10th v. 90th percentile for DASH: 2·23 (95 % CI 1·50, 3·32), DGAI: 3·01 (95 % CI 1·98, 4·57), MSDPS: 2·02 (95 % CI 1·14, 3·58)). Similar results were observed in CCHS 2015; however, results for MSDPS were not significant (OR 10th v. 90th percentile for DASH: 2·45 (95 % CI 1·72, 3·49), DGAI: 2·73 (95 % CI 1·85, 4·03); MSDPS: 1·30 (95 % CI 0·82, 2·06)). CONCLUSION: Following DASH or the 2015-2020 DGA was associated with a lower likelihood of obesity. Findings do not indicate causation, as the data are cross-sectional.

Child-appealing packaged food and beverage products in Canada-Prevalence, power, and nutritional quality.

BACKGROUND: Children are frequently exposed to marketing on food packaging. This study evaluated the presence, type and power of child-appealing marketing and compared the nutritional quality of child-appealing vs. non-child-appealing Canadian packaged foods and examined the relationship between nutrient composition and marketing power. METHODS: Child-relevant packaged foods (n = 5,850) were sampled from the Food Label Information Program 2017 database. The presence and power (# of techniques displayed) of child-appealing marketing were identified. Fisher's Exact test compared the proportion of products exceeding Health Canada's nutrient thresholds for advertising restrictions and Mann Whitney U tests compared nutrient composition between products with child- /non-child-appealing packaging. Pearson's correlation analyzed the relationship between nutrient composition and marketing power. RESULTS: 13% (746/5850) of products displayed child-appealing marketing; the techniques used, and the power of the marketing varied ([Formula: see text] 2.2 techniques; range: 0-11). More products with child-appealing packaging than with non-child appealing packaging exceeded Health Canada's thresholds (98% vs. 94%; p < .001). Products with child-appealing packaging (vs. non-child-appealing) were higher in total sugars (median: 14.7 vs. 9 g/RA; p < .001) and free sugars (11.5 vs. 6.2 g/RA; p < .001), but lower in all other nutrients. There was weak overall correlation between marketing power and nutrient levels. Results varied by nutrient and food category. CONCLUSIONS: Unhealthy products with powerful child-appealing marketing displayed on package are prevalent in the food supply. Implementing marketing restrictions that protect children should be a priority.

Canadian Free Sugar Intake and Modelling of a Reformulation Scenario.

Recommendations suggest limiting the intake of free sugar to under 10% or 5% of calories in order to reduce the risk of negative health outcomes. This study aimed to examine Canadian free sugar intake and model how intakes change following the implementation of a systematic reformulation of foods and beverages to be 20% lower in free sugar. Additionally, this study aimed to examine how calorie intake might be impacted by this reformulation scenario. Canadians' free sugar and calorie intakes were determined using free sugar and calorie data from the Food Label Information Program (FLIP) 2017, a Canadian branded food composition database, and applied to foods reported as being consumed in Canadian Community Health Survey-Nutrition (CCHS-Nutrition) 2015. A "counterfactual" scenario was modelled to examine changes in intake following the reformulation of foods to be 20% lower in free sugar. The overall mean free sugar intake was 12.1% of calories and was reduced to align with the intake recommendations at 10% of calories in the "counterfactual" scenario (p < 0.05). Calorie intake was reduced by 3.2% (60 calories) in the "counterfactual" scenario (p < 0.05). Although the overall average intake was aligned with the recommendations, many age/sex groups exceeded the recommended intake, even in the "counterfactual" scenario. The results demonstrate a need to reduce the intake of free sugar in Canada to align with dietary recommendations, potentially through reformulation. The results can be used to inform future program and policy decisions related to achieving the recommended intake levels of free sugar in Canada.

Estimating Canadian sodium intakes and the health impact of meeting national and WHO recommended sodium intake levels: A macrosimulation modelling study.

BACKGROUND: Cardiovascular diseases (CVDs) are the second leading cause of total deaths in Canada. High blood pressure is the main metabolic risk factor for developing CVDs. It has been well established that excess consumption of sodium adversely affects blood pressure. Canadians' mean sodium intakes are well above recommended levels. Reducing dietary sodium intake through food reformulation has been identified as a cost-effective intervention, however, dietary sodium intake and the potential health impact of meeting recommended sodium intake levels due to food reformulation have not been determined in Canada. OBJECTIVE: This study aimed to 1) obtain robust estimates of Canadians' usual sodium intakes, 2) model sodium intakes had foods been reformulated to align with Health Canada's sodium reduction targets, and 3) estimate the number of CVD deaths that could be averted or delayed if Canadian adults were to reduce their mean sodium intake to recommended levels under three scenarios: A) 2,300 mg/d-driven by a reduction of sodium levels in packaged foods to meet Health Canada targets (reformulation); B) 2,000 mg/d to meet the World Health Organization (WHO) recommendation; and C) 1,500 mg/d to meet the Adequate Intake recommendation. METHODS: Foods in the University of Toronto's Food Label Information Program 2017, a Canadian branded food composition database, were linked to nationally representative food intake data from the 2015 Canadian Community Health Survey-Nutrition to estimate sodium intakes (and intakes had Health Canada's reformulation strategy been fully implemented). The Preventable Risk Integrated ModEl (PRIME) was used to estimate potential health impact. RESULTS: Overall, mean sodium intake was 2758 mg/day, varying by age and sex group. Based on 'reformulation' scenario A, mean sodium intakes were reduced by 459 mg/day, to 2299 mg/day. Reducing Canadians' sodium intake to recommended levels under scenarios A, B and C could have averted or delayed 2,176 (95% UI 869-3,687), 3,252 (95% UI 1,380-5,321), and 5,296 (95% UI 2,190-8,311) deaths due to CVDs, respectively, mainly from ischaemic heart disease, stroke, and hypertensive disease. This represents 3.7%, 5.6%, and 9.1%, respectively, of the total number of CVDs deaths observed in Canada in 2019. CONCLUSION: Results suggest that reducing sodium intake to recommended levels could prevent or postpone a substantial number of CVD deaths in Canada. Reduced sodium intakes could be achieved through reformulation of the Canadian food supply. However, it will require higher compliance from the food industry to achieve Health Canada's voluntary benchmark sodium reduction targets.

Nutrient intakes and top food categories contributing to intakes of energy and nutrients-of-concern consumed by Canadian adults that would require a 'high-in' front-of-pack symbol according to Canadian labelling regulations.

Canada recently mandated front-of-pack (FOP) labelling regulations, where foods meeting and/or exceeding recommended thresholds for nutrients-of-concern (i.e., saturated fat, sodium, and sugars) must display a 'high-in' FOP nutrition symbol. However, there is limited research on the amounts and sources of foods consumed by Canadians that would require a FOP symbol. The objective was to examine the intakes of nutrients-of-concern from foods that would display a FOP symbol and to identify the top food categories contributing to intakes for each nutrient-of-concern. Using the first day 24-hour dietary recall from the nationally representative 2015 Canadian Community Health Survey-Nutrition (CCHS), Canadian adults' intakes of nutrients-of-concern from foods that would display a FOP symbol was examined. Foods were assigned to 1 of 62 categories to identify the top food categories contributing to intakes of energy and nutrient-of-concern that would display a FOP symbol for each nutrient-of-concern. Canadian adults (n = 13,495) consumed approximately 24% of total calories from foods that would display a FOP symbol. Foods that would display a FOP symbol for exceeding thresholds for nutrients-of-concern accounted for 16% of saturated fat, 30% of sodium, 25% of total sugar, and 39% of free sugar intakes among Canadian adults. The top food category contributing intakes of each nutrient-of-concern that would display a FOP symbol were nutrient-specific: Processed meat and meat substitutes for saturated fat; Breads for sodium; and Fruit juices & drinks for total and free sugars. Our findings show that Canadian FOP labelling regulations have the potential to influence the intakes of nutrients-of-concern for Canadian adults. Using the findings as baseline data, future studies are warranted to evaluate the impact of FOP labelling regulations.

Qualitative analysis of front-of package labeling policy interactions between stakeholders and Health Canada.

Background: Front-of-package labelling regulations proposed by Health Canada in their Healthy Eating Strategy (2016) were finally passed in 2022, but remain unimplemented. This study analyzed interactions that occurred between stakeholders and government related to this policy proposal to identify key themes and policy implications. Methods: A qualitative framework analysis was conducted on publicly available documents for stakeholder correspondences related to front-of-package that occurred between 2016 and 2019 in Health Canada's Meetings and Correspondence on Healthy Eating database. Five sequential steps were applied: familiarization, identifying a thematic framework, indexing, charting, and mapping and interpretation. A complex systems (i.e., a dynamic system with multiple interconnecting components) lens was incorporated in the final step to deepen the analysis. Results: Hundred and seventy-three documents were included, the majority from industry stakeholders (n = 108, 62.4%). Three overarching themes were identified: industry trying to control the agenda and resist regulation; questioning the evidence supporting the policy and its impact on the agri-food industry; and dismissing the need and effectiveness of the policy. Incorporating a complex system lens found industry and non-industry stakeholders held markedly different perspectives on how cohesive the system defined by the front-of-package labelling policy was, and the policy impact on its stability. Economic and opportunity costs were the main trade-offs, and symbol misinterpretation considered an unintended consequence by industry. Finally, some stakeholders argued for wider policy scope incorporating more products, while others requested a narrower approach through exemptions. Conclusion: Interactions with industry stakeholders on health food policy proposals require careful consideration, given it may suit their interests to generate delays and policy discordance. Explicitly setting out the principles of engagement and actively encouraging non-industry stakeholder representation provides a more balanced approach to policy consultation and development.

Estimating the dietary and health impact of implementing front-of-pack nutrition labeling in Canada: A macrosimulation modeling study.

Background: Front-of-pack labeling (FOPL) has been identified as a cost-effective policy to promote healthy diets. Health Canada has recently published FOPL regulations that will require food and beverages that meet or exceed set thresholds for sodium, sugars, or saturated fat to display a 'high in' symbol on the front of the package. Although a promising measure, its potential impact on dietary intakes and health have not yet been estimated in Canada. Objective: This study aims to estimate (1) the potential dietary impact of implementing a mandatory FOPL among Canadian adults; and (2) the number of diet-related non-communicable disease (NCD) deaths that could be averted or delayed due to these estimated dietary changes. Methods: Baseline and counterfactual usual intakes of sodium, total sugars, saturated fats, and calories were estimated among Canadian adults (n = 11,992) using both available days of 24 h recalls from the 2015 Canadian Community Health Survey-Nutrition. The National Cancer Institute method was used to estimate usual intakes, and adjusted for age, sex, misreporting status, weekend/weekday, and sequence of recall. Estimated counterfactual dietary intakes were modeled from reductions observed in experimental and observational studies that examined changes in sodium, sugars, saturated fat, and calorie content of food purchases in the presence of a 'high in' FOPL (four counterfactual scenarios). The Preventable Risk Integrated ModEl was used to estimate potential health impacts. Results: Estimated mean dietary reductions were between 31 and 212 mg/day of sodium, 2.3 and 8.7 g/day of total sugars, 0.8 and 3.7 g/day of saturated fats, and 16 and 59 kcal/day of calories. Between 2,183 (95% UI 2,008-2,361) and 8,907 (95% UI 8,095-9,667) deaths due to diet-related NCDs, mostly from cardiovascular diseases (~70%), could potentially be averted or delayed by implementing a 'high in' FOPL in Canada. This estimation represents between 2.4 and 9.6% of the total number of diet-related NCD deaths in Canada. Conclusion: Results suggest that implementing a FOPL could significantly reduce sodium, total sugar, and saturated fat intakes among Canadian adults and subsequently prevent or postpone a substantial number of diet-related NCD deaths in Canada. These results provide critical evidence to inform policy decisions related to implementing FOPL in Canada.

How Many Diet-Related Non-Communicable Disease Deaths Could Be Averted or Delayed If Canadians Reduced Their Consumption of Calories Derived from Free Sugars Intake? A Macrosimulation Modeling Study.

Free sugars are a major source of calories in diets and contribute to the burden of many non-communicable diseases (NCDs). The World Health Organization (WHO) recommends reducing free sugars intake to less than 10% of total energy. This study aimed to estimate the number of diet-related NCD deaths which could be averted or delayed if Canadian adults were to reduce their calorie intake due to a systematic 20% reduction in the free sugars content in foods and beverages in Canada. We used the Preventable Risk Integrated ModEl (PRIME) to estimate the potential health impact. An estimated 6770 (95% UI 6184-7333) deaths due to diet-related NCDs could be averted or delayed, mostly from cardiovascular diseases (66.3%). This estimation would represent 7.5% of diet-related NCD deaths observed in 2019 in Canada. A 20% reduction in the free sugars content in foods and beverages would lead to a 3.2% reduction in calorie intake, yet an important number of diet-related NCD deaths could be averted or delayed through this strategy. Our findings can inform future policy decisions to support Canadians' free sugars intake reduction, such as proposing target levels for the free sugars content in key food categories.

Natural language processing and machine learning approaches for food categorization and nutrition quality prediction compared with traditional methods.

BACKGROUND: Food categorization and nutrient profiling are labor intensive, time consuming, and costly tasks, given the number of products and labels in large food composition databases and the dynamic food supply. OBJECTIVES: This study used a pretrained language model and supervised machine learning to automate food category classification and nutrition quality score prediction based on manually coded and validated data, and compared prediction results with models using bag-of-words and structured nutrition facts as inputs for predictions. METHODS: Food product information from University of Toronto Food Label Information and Price Database 2017 (n = 17,448) and University of Toronto Food Label Information and Price Database 2020 (n = 74,445) databases were used. Health Canada's Table of Reference Amounts (TRA) (24 categories and 172 subcategories) was used for food categorization and the Food Standards of Australia and New Zealand (FSANZ) nutrient profiling system was used for nutrition quality score evaluation. TRA categories and FSANZ scores were manually coded and validated by trained nutrition researchers. A modified pretrained sentence-Bidirectional Encoder Representations from Transformers model was used to encode unstructured text from food labels into lower-dimensional vector representations, followed by supervised machine learning algorithms (i.e., elastic net, k-Nearest Neighbors, and XGBoost) for multiclass classification and regression tasks. RESULTS: Pretrained language model representations utilized by the XGBoost multiclass classification algorithm reached overall accuracy scores of 0.98 and 0.96 in predicting food TRA major and subcategories, outperforming bag-of-words methods. For FSANZ score prediction, our proposed method reached a similar prediction accuracy (R2: 0.87 and MSE: 14.4) compared with bag-of-words methods (R2: 0.72-0.84; MSE: 30.3-17.6), whereas structured nutrition facts machine learning model performed the best (R2: 0.98; MSE: 2.5). The pretrained language model had a higher generalizable ability on the external test datasets than bag-of-words methods. CONCLUSIONS: Our automation achieved high accuracy in classifying food categories and predicting nutrition quality scores using text information found on food labels. This approach is effective and generalizable in a dynamic food environment, where large amounts of food label data can be obtained from websites.

Comparing the nutritional composition of foods and beverages in the Canadian Nutrient File to a large representative database of Canadian prepackaged foods and beverages.

BACKGROUND: Nutrient information used to code dietary intakes in the Canadian Community Health Survey (CCHS) may not be reflective of the current Canadian food supply and could result in inaccurate evaluations of nutrient exposures. OBJECTIVE: To compare the nutritional compositions of foods in the CCHS 2015 Food and Ingredient Details (FID) file (n = 2,785) to a large representative Canadian database of branded food and beverage products (Food Label Information Program, FLIP) collected in 2017 (n = 20,625). METHOD: Food products in the FLIP database were matched to equivalent generic foods from the FID file to create new aggregate food profiles based on FLIP nutrient data. Mann Whitney U tests were used to compare nutrient compositions between the FID and FLIP food profiles. RESULTS: In most food categories and nutrients there were no statistically significant differences between the FLIP and FID food profiles. Nutrients with the largest differences included: saturated fats (n = 9 of 21 categories), fiber (n = 7), cholesterol (n = 6), and total fats (n = 4). The meats and alternatives category had the most nutrients with significant differences. CONCLUSION: These results can be used to prioritize future updates and collections of food composition databases, while also providing insights for interpreting CCHS 2015 nutrient intakes.

Top Sodium Food Sources in the American Diet-Using National Health and Nutrition Examination Survey.

Reducing population-level sodium intake can reduce hypertension, an important preventative strategy to lower the risk of cardiovascular diseases, the leading cause of death in the United States. Considering that most dietary sodium is derived from prepackaged foods, this study quantitatively estimates the proportion contribution and mean sodium intake from key food category contributors to total sodium intake in the US population. Data from the 2017-2018 National Health and Nutrition Examination Survey, which collected interviewer-administered 24 h dietary recalls from Americans (n = 7081), were analyzed. Based on the average proportion contributed, the top 15 sources of sodium were identified overall and by age/sex, poverty-income and race/ethnicity. More than 50% of US population-level dietary sodium intake was contributed by: pizza (5.3%); breads, rolls and buns (4.7%); cold cuts and cured meats (4.6%); soups (4.4%); burritos and tacos (4.3%); savoury snacks (4.1%); poultry (4.0%); cheese (3.1%); pasta mixed dishes (2.9%); burgers (2.5%); meat mixed dishes (2.5%); cookies, brownies and cakes (2.4%); bacon, frankfurters and sausages (2.4%); vegetables (2.2%); and chicken nuggets (1.5%), with the results remaining consistent among population subgroups. The results identified the top sources of sodium in the American population overall, as well as in key population subgroups, which can inform policies and programs aimed at reducing sodium intake.