A Machine Learning Approach to Predict the Added-Sugar Content of Packaged Foods.

BACKGROUND: Dietary guidelines recommend limiting the intake of added sugars. However, despite the public health importance, most countries have not mandated the labeling of added-sugar content on packaged foods and beverages, making it difficult for consumers to avoid products with added sugar, and limiting the ability of policymakers to identify priority products for intervention. OBJECTIVE: The aim was to develop a machine learning approach for the prediction of added-sugar content in packaged products using available nutrient, ingredient, and food category information. METHODS: The added-sugar prediction algorithm was developed using k-nearest neighbors (KNN) and packaged food information from the US Label Insight dataset (n = 70,522). A synthetic dataset of Australian packaged products (n = 500) was used to assess validity and generalization. Performance metrics included the coefficient of determination (R2), mean absolute error (MAE), and Spearman rank correlation (ρ). To benchmark the KNN approach, the KNN approach was compared with an existing added-sugar prediction approach that relies on a series of manual steps. RESULTS: Compared with the existing added-sugar prediction approach, the KNN approach was similarly apt at explaining variation in added-sugar content (R2 = 0.96 vs. 0.97, respectively) and ranking products from highest to lowest in added-sugar content (ρ = 0.91 vs. 0.93, respectively), while less apt at minimizing absolute deviations between predicted and true values (MAE = 1.68 g vs. 1.26 g per 100 g or 100 mL, respectively). CONCLUSIONS: KNN can be used to predict added-sugar content in packaged products with a high degree of validity. Being automated, KNN can easily be applied to large datasets. Such predicted added-sugar levels can be used to monitor the food supply and inform interventions aimed at reducing added-sugar intake.

Projected effects on salt purchases following implementation of a national salt reduction policy in South Africa.

OBJECTIVE: To assess the contribution of different food groups to total salt purchases and to evaluate the estimated reduction in salt purchases if mandatory maximum salt limits in South African legislation were being complied with. DESIGN: This study conducted a cross-sectional analysis of purchasing data from Discovery Vitality members. Data were linked to the South African FoodSwitch database to determine the salt content of each food product purchased. Food category and total annual salt purchases were determined by summing salt content (kg) per each unit purchased across a whole year. Reductions in annual salt purchases were estimated by applying legislated maximum limits to product salt content. SETTING: South Africa. PARTICIPANTS: The study utilised purchasing data from 344 161 households, members of Discovery Vitality, collected for a whole year between January and December 2018. RESULTS: Vitality members purchased R12·8 billion worth of food products in 2018, representing 9562 products from which 264 583 kg of salt was purchased. The main contributors to salt purchases were bread and bakery products (23·3 %); meat and meat products (19 %); dairy (12·2 %); sauces, dressings, spreads and dips (11·8 %); and convenience foods (8·7 %). The projected total quantity of salt that would be purchased after implementation of the salt legislation was 250 346 kg, a reduction of 5·4 % from 2018 levels. CONCLUSIONS: A projected reduction in salt purchases of 5·4 % from 2018 levels suggests that meeting the mandatory maximum salt limits in South Africa will make a meaningful contribution to reducing salt purchases.

The nutritional content of children's breakfast cereals: a cross-sectional analysis of New Zealand, Australia, the UK, Canada and the USA.

OBJECTIVE: To compare the Nutrition Information Panel (NIP) content, serving size and package size of children's ready-to-eat breakfast cereals (RTEC) available in five different Western countries. DESIGN: NIP label information was collected from RTEC available for purchase in major supermarket chains. Kruskal-Wallis, Mann-Whitney U and χ2 tests were applied to detect differences between countries on manufacturer-declared serving size, total energy (kJ), total protein, fat, saturated fat, carbohydrate, total sugar, Na and fibre content. The Nutrient Profiling Scoring Criterion (NPSC) was used to evaluate the number of products deemed to be 'unhealthy'. SETTING: Supermarkets in Australia, Canada, New Zealand, the UK and the USA. PARTICIPANTS: Children's breakfast cereals (n 636), including those with and without promotional characters. RESULTS: The majority of children's RTEC contained substantial levels of total sugar and differences were apparent between countries. Median sugar content per serving was higher in US cereals than all other countries (10·0 v. 7·7-9·1 g; P < 0·0001). Median fat and saturated fat content were lowest in Australia and New Zealand RTEC, while the Na content of RTEC was 60-120 % higher in the USA and Canada than in Australia and the UK (all P ≤ 0·01). CONCLUSIONS: Across all countries, there was a high proportion of RTEC marketed for children that had an unhealthy nutrient profile. Strategies and policies are needed to improve the nutrient value of RTEC for children, so they provide a breakfast food that meets nutrition guidelines.

Association between the Urinary Sodium to Potassium Ratio and Blood Pressure in Adults: A Systematic Review and Meta-Analysis.

While sodium and potassium are individually important for blood pressure (BP) regulation, the relative contribution of sodium to potassium intake has not been sufficiently investigated. This study aimed to evaluate the association between urinary sodium to potassium ratio (UNa: K) and systolic and diastolic BP in adults. A systematic review (PROSPERO; CRD42016035296) was conducted and was reported according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Three scientific databases (MEDLINE, Scopus, Web of Science) were searched to March 2020 while reference lists of included articles were further hand-searched. Randomized controlled trials (RCT), cohort and cross-sectional studies that assessed 24-h urinary excretion in adults were included. Data from eligible studies were extracted and summarized. Random effects meta-analysis was conducted on RCT data to assess standardized mean differences (SMD) in systolic and diastolic BP according to 24-h UNa: K. Thirty-nine studies were included. Meta-analysis of 5 RCTs found a lower UNa: K ratio to be associated with a significantly greater reduction in systolic and diastolic BP compared with a higher UNa: K ratio [SMD: -1.09 (95% CI: -1.91, -0.28) mmHg and -1.42 (95% CI: -2.24, -0.59) mmHg, respectively]. Heterogeneity between RCTs was observed in systolic and diastolic BP (I2 = 97%, P < 0.0001 and I2 = 98%, P < 0.0001, respectively). The current body of evidence demonstrates that a lower 24-h UNa: K ratio is associated with lower BP in adults. Dietary strategies to achieve an increase in potassium while at the same time lowering sodium would be beneficial in lowering BP.

Sodium Content and Labelling Completeness of Packaged Foods and Beverages in Kenya.

Increased consumption of unhealthy processed foods, particularly those high in sodium, is a major risk factor for cardiovascular diseases. Nutrition information on packaged foods can help guide consumers toward products with less sodium, however the availability of nutrition information on foods sold in Kenya is currently unknown. The aims of this study were to estimate the proportion of packaged foods and beverages displaying nutrition information for sodium and determine the amount of sodium in packaged foods available for sale in Kenya. Data was collected in 2019 from five retail supermarkets in Nairobi. The availability of sodium information provided on packaged products and the sodium content were recorded. As secondary analyses, we compared sodium content labelling of products in Kenya by manufacturing location and the sodium content of products available in Kenya and South Africa. A total of 6003 packaged products in 56 food categories were identified. Overall, 39% of products displayed sodium content, though the availability of labelling varied widely between food categories, with coverage in main categories ranging from 0% (yoghurts and yoghurt drinks) to 86% (breakfast cereals). Food categories with the highest median sodium content were herbs and spices (9120 mg/100 g), sauces (1200 mg/100 g) and meat alternatives (766 mg/100 g) although wide variabilities were often observed within categories. Imported products were more likely to provide information on sodium than locally produced products (81% compared to 26%) and reported higher median sodium levels (172 mg/100 g compared to 96 mg/100 g). Kenyan products reported a higher median sodium content than South African products in six categories while South African products had higher median sodium in 20 categories, with considerable variation in median sodium content between countries in some categories. These findings highlight considerable potential to improve the availability of sodium information on packaged products in Kenya and to introduce reformulation policies to reduce the amount of sodium in the Kenyan food supply.

Monitoring and implementation of salt reduction initiatives in Africa: A systematic review.

This systematic review aims to document salt consumption patterns and the implementation status and potential impact of salt reduction initiatives in Africa, from studies published between January 2009 and November 2019. Studies were sourced using MEDLINE, Embase, Cochrane Library electronic databases, and gray literature. Of the 887 records retrieved, 38 studies conducted in 18 African countries were included. Twelve studies measured population salt intake, 11 examined salt level in foods, 11 assessed consumer knowledge, attitudes, and behaviors, 1 study evaluated a behavior change intervention, and 3 studies modeled potential health gains and cost savings of salt reduction interventions. The population salt intake studies determined by 24-hour urine collections showed that the mean (SD) salt intake in African adults ranged from 6.8 (2.2) g to 11.3 (5.4) g/d. Salt levels in foods were generally high, and consumer knowledge was fairly high but did not seem to translate into salt lowering behaviors. Modeling studies showed that interventions for reducing dietary sodium would generate large health gains and cost savings for the health system. Despite this evidence, adoption of population salt reduction strategies in Africa has been slow, and dietary consumption of sodium remains high. Only South Africa adopted legislation in 2016 to reduce population salt intake, but success of this intervention has not yet been fully evaluated. Thus, rigorous evaluation of the salt reduction legislation in South Africa and initiation of salt reduction programs in other African countries will be vital to achieving the targeted 30% reduction in salt intake by 2025.

Variations in Sugar Content of Flavored Milks and Yogurts: A Cross-Sectional Study across 3 Countries.

BACKGROUND: The consumption of dairy products is encouraged at all life stages as a nutrient-rich component of the diet. However, many milk and yogurt products, particularly flavored varieties, may contain large amounts of free sugar. OBJECTIVES: The aim of this paper was to evaluate the availability and sugar content of flavored milks and yogurts in supermarkets across 3 countries: Australia, England, and South. METHODS: Nutrition information for flavored milks and yogurts was collected by trained researchers and supplemented by crowd-sourced data from a smartphone application. Data were extracted in April 2018 and 3724 milk and yogurt products were available for analysis. Mean sugar concentrations were compared across countries with the use of ANOVA followed by Tukey's post-hoc pairwise comparisons. Sugar concentrations were compared with the UK's "green" traffic-light classifications. RESULTS: Approximately 74% (n = 2753) of all products were flavored. Flavored products contained nearly twice the average total sugar content of unflavored products, with substantial variability: mean total sugar was 9.1 g/100 mL (range: 4.3-15.0 g/100 mL) and 11.5 g/100 g (range: 0.1-22.6 g/100 g) for flavored milks and yogurts, respectively. Free sugars contributed an estimated 41% and 42% of total sugar in milks and yogurts, respectively. Flavored milks in England had ∼0.7 g/100 mL higher total sugar on average compared with Australia and South Africa (P ≤ 0.04), whereas flavored yogurts in South Africa had the lowest average total sugar (∼2 g/100 g lower than England and Australia; P < 0.001). Less than 4% of flavored products would receive a "green" rating under the UK traffic-light labeling scheme. CONCLUSIONS: In Australia, England, and South Africa, flavored milks and yogurts are highly prevalent in the food supply and contain significantly higher concentrations of total and added sugars than unflavored products.

Effect of individualised dietary advice for weight loss supplemented with walnuts on blood pressure: the HealthTrack study.

BACKGROUND/OBJECTIVES: In addition to weight-loss, healthy dietary patterns and lower sodium intakes can help reduce blood pressure (BP), but individualised dietary advice may be necessary to achieve these effects. This study aimed to examine the impact of individualised dietary advice on BP in the intensive phase of a weight-loss trial. SUBJECTS/METHODS: Secondary analysis of baseline and 3-month data from the HealthTrack randomised controlled trial (n = 211). Participants were randomly assigned to one of three dietary advice groups: general advice (control), individualised advice (intervention group, I), or intervention group supplemented with 30 g walnuts/day (IW). Resting BP and 24-h urine sodium and potassium were measured. Dietary intake was evaluated through diet history interviews. RESULTS: Unadjusted SBP reduced significantly in all groups (IW and I groups P < 0.001; control group P = 0.002) and DBP in IW and I groups (P < 0.001). Compared to controls, the reductions in BP were 3-4 mmHg greater in the I and IW groups, but this only reached significance for DBP in the I group (-3.3 mmHg; P = 0.041). After controlling for age, sex, medication, weight-loss, physical activity and smoking, only the IW group showed a significant association between SBP reduction and increased urinary potassium (ß = -0.101, P = 0.044), decreased sodium:potassium ratio (ß = 2.446, P = 0.037) and increased consumption of seed and nut products and dishes (ß = -0.108, P = 0.034). CONCLUSIONS: Dietary patterns with distinctive foods and lower sodium:potassium ratios may enhance the effects of weight-loss on BP. The patterns were best achieved with individualised dietary advice and food supplements.

Associations between Dietary Patterns and Blood Pressure in a Clinical Sample of Overweight Adults.

BACKGROUND: Dietary pattern analysis provides important evidence revealing diet-disease relationships. It may be especially useful in areas less well researched, such as diet and hypertension in clinical populations. OBJECTIVE: The aim of this study was to identify the association between dietary patterns and blood pressure (BP) in a sample of overweight adults volunteering for a clinical trial for weight loss. DESIGN: This cross-sectional analysis used baseline data from the HealthTrack study, a 12-month randomized controlled trial. Dietary intake was evaluated with 4-day food records. PARTICIPANTS/SETTING: Participants were 328 adults recruited from the Illawarra region of New South Wales, Australia, between May 2014 and April 2015. MAIN OUTCOME MEASURES: Resting BP and 24-hour urine sodium and potassium were measured. STATISTICAL ANALYSIS: Dietary patterns were derived by principal component analysis from 21 food groups. Multiple regression analysis was performed to assess the association between the extracted dietary patterns and BP. RESULTS: The participants' mean age was 43.6±8.0 years, mean body mass index was 32.4±4.2, and mean systolic BP/diastolic BP was 124.9±14.5/73.3±9.9 mm Hg. Six major dietary patterns were identified: "nuts, seeds, fruit, and fish," "milk and meat," "breads, cereals, and snacks," "cereal-based products, fats, and oils," "alcohol, eggs, and legumes," and "savoury sauces, condiments, and meat." The "nuts, seeds, fruit, and fish" dietary pattern was significantly and inversely associated with systolic BP (F [7,320]=15.248; P<0.0005; adjusted R2=0.234 and diastolic BP (F [7,320]=17.351; P<0.0005; adjusted R2=0.259) and sodium-to-potassium ratio (F [7,320]=6.210; P<0.0005; adjusted R2=0.100). CONCLUSIONS: A dietary pattern rich in nuts, seeds, fruit, and fish was inversely associated with blood pressure in this clinical sample. The findings suggest that current dietary guidelines are relevant to an overweight clinical population and support the value of dietary pattern analysis when exploring the diet-disease relationship.

Relationship between sodium and potassium intake and blood pressure in a sample of overweight adults.

OBJECTIVE: The aim of this study was to examine the relationship between sodium and potassium intakes and blood pressure (BP) in a clinical sample. METHODS: Secondary analysis of baseline data from 328 participants (mean age: 43.6 ± 8 y, mean body mass index [BMI]: 32.4 ± 4.2 kg/m2, mean systolic BP [SBP]/diastolic BP [DBP]: 124.9 ± 14.5/73.3 ± 9.9 mm Hg) of the 12-mo HealthTrack randomized controlled weight loss trial was conducted. Resting BP and 24-h urine sodium and potassium were measured. Dietary intake was evaluated with 4-d food records and self-reported diet histories. RESULTS: Urinary sodium was positively correlated (Spearman's rho) with SBP (r = 0.176; P = 0.001) and DBP (r = 0.150; P = 0.003). The ratio of sodium to potassium was positively correlated with SBP (r = 0.1; P = 0.035). Urinary sodium (F [4,323] = 20.381; P < 0.0005; adjusted R2 = 0.231) and sodium-to-potassium ratio (F[4,323] = 25.008; P < 0.0005; adjusted R2 = 0.227) significantly predicted SBP after controlling for age, sex, BMI, and hypertension medication use. Dietary sodium and potassium significantly predicted urinary sodium (B = 0.33, t = 4.032, P < 0.01) and potassium (B = 0.67, t = 8.537, P < 0.01) excretion, respectively, after adjustment for energy and BMI. Median dietary sodium intake was 3197 mg/d and median dietary potassium intake was 2886 mg/d. Cereal-based products and dishes were the major contributors (22%) to total sodium intake. CONCLUSIONS: In the present study, a high dietary sodium intake and high sodium-to-potassium ratio predicted high SBP. This suggests a need to focus dietary advice on reduction of sources of sodium and increasing sources of potassium in weight loss interventions to improve BP control.

Dietary Patterns and Blood Pressure in Adults: A Systematic Review and Meta-Analysis of Randomized Controlled Trials.

Hypertension is a major risk factor for developing cardiovascular disease, stroke, and kidney disease. To lower blood pressure (BP), several lifestyle changes are recommended such as weight loss, exercise, and following a healthy diet. Investigating the effect of single nutrients may have positive results, but food is consumed as part of a whole diet, resulting in nutrient interactions. The aim of this systematic review and meta-analysis was to assess the effect of dietary patterns on BP in adults. Studies that were published between January 1999 and June 2014 were retrieved using Scopus, Web of Science, and the MEDLINE database. Seventeen randomized controlled trials were included in the meta-analysis. The results suggest that healthy dietary patterns such as the Dietary Approaches to Stop Hypertension diet, Nordic diet, and Mediterranean diet significantly lowered systolic BP and diastolic BP by 4.26 mm Hg and 2.38 mm Hg, respectively. These diets are rich in fruit, vegetables, whole grains, legumes, seeds, nuts, fish, and dairy and low in meat, sweets, and alcohol. Lifestyle factors such as exercise and weight loss in combination with dietary changes may also reduce BP. Further research is needed to establish the effect of dietary patterns on BP in different cultures other than those identified in this review. The review was registered on PROSPERO (International prospective register of systematic reviews) as CRD42015016272.