The study aim was to determine the prevalence of food addiction (FA) in individuals with type 2 diabetes and to assess the association between FA and type 2 diabetes. MEDLINE, EMBASE, Web of Sciences, Latin American and Caribbean Literature in Health Sciences, ScienceDirect, Scopus and PsycINFO were searched until November 2024. This study was registered with PROSPERO (CRD42023465903). Cross-sectional studies, case–control, cohorts and clinical trials that were carried out with individuals with type 2 diabetes, regardless of age and sex, were included. The complete data extracted included the prevalence, OR and/or risk ratio of FA, the number of individuals evaluated, age, sex, weight, presence of co-morbidities, age of participants and FA symptoms. A pooled prevalence of FA of 30 % (95 % CI (18, 44) with estimated predictive interval (0; 85); I2 = 99·51; 12 studies; 15 947 participants) was identified. For the associations between FA and type 2 diabetes, we found a grouped crude OR value of 2·35 (95 % CI (1·71, 2·98)). The pooled OR adjusted for age and sex was 2·60 (95 % CI (1·77, 3·42)). Finally, the OR adjusted for age, sex and BMI was 2·01 (95 % CI (1·39, 2·64)). The results of the meta-analyses showed a high prevalence of FA in individuals with type 2 diabetes and that the associations between these two conditions remained even after adjustment for age, sex and BMI, although with a high heterogeneity among individual estimates.

Although research on the relationship between lean body mass and blood pressure (BP) has been inconsistent, most studies reported that measures of lean body mass are associated with a higher risk of hypertension. We explored relationships between body composition (fat and skeletal muscle mass) and BP in 1162 young adult African women. Dual-energy X-ray absorptiometry-derived measures of whole-body, central and arm fat mass were associated with higher systolic and diastolic BP, while leg fat percentage was associated with lower systolic and diastolic BP. However, only the associations with diastolic BP remained after adjusting for appendicular skeletal muscle mass (ASM). ASM was associated with higher systolic and diastolic BP, before and after adjusting for whole-body fat percentage and visceral adipose tissue. While there was no overlap in targeted proteomics of BP and body composition, REN was lower in the elevated BP than the normal BP group and was inversely associated with diastolic BP (false rate discovery adjusted P< 0·050). Several proteins were positively associated with both visceral adipose tissue and ASM (LEP, FABP4, IL6 and GGH) and negatively associated with both visceral adipose tissue and ASM (ACAN, CELA3A, PLA2G1B and NCAM1). NOTCH3, ART3, COL1A1, DKK3, ENG, NPTXR, AMY2B and CNTN1 were associated with lower visceral adipose tissue only, and IGFBP1 was associated with lower ASM only. While the associations between body fat and BP were not independent of skeletal muscle mass, the associations between muscle mass and BP were independent of overall and central adiposity in young adult African women. Future interventions targeting muscle mass should also monitor BP in this population.

Adherence to healthy diet principles and to cardiopreventive medication, both key behaviours in CVD prevention, is known to differ between women and men. Whether these adherence behaviours are differentially related among women and men has never been thoroughly assessed. The objective was to assess gender differences in the association between adherence to healthy diet principles and to cardiopreventive medication in adults free of CVD. This cross-sectional study included 268 women and 204 men from the CARTaGENE cohort (Québec, Canada) who were using antihypertensive and/or cholesterol-lowering medication. Adherence to healthy diet principles was assessed using the Alternate Healthy Eating Index (AHEI, %), calculated from a validated FFQ assessing diet in the 12-month preceding its completion. Medication adherence was assessed using the daily pharmacotherapy possession rate (DPPR, %), calculated from prescription claim data over the same 12-month period. In multivariable-adjusted analyses, an inverse association between AHEI and DPPR was observed among men (βAHEI for 10 % increment in DPPR = –0·65 %; 95 % CI −1·28 %, −0·03 %; P = 0·04), while it tended to be positive among women (β = 0·44 %; 95 % CI −0·11 %, 1·00 %; P = 0·12; Pgender×DPPR = 0·01). The negative association between AHEI and DPPR was stronger among men who never smoked or used cholesterol-lowering medication only. Among women, the positive association was stronger and statistically significant among those with obesity or using ≥ 3 medications simultaneously. Association between adherence to healthy diet principles and to cardiopreventive medication differs between women and men, with men potentially facing greater challenges in achieving optimal complementarity between these two behaviours.

Increased ultra-processed food (UPF) is associated with adverse health outcomes. However, with limitations in UPF evidence, and partial overlap between UK front-of-package labelling (FOPL) and degree of food processing, the value of food processing within dietary guidance is unclear. This study compared food and drink from the UK National Diet and Nutrition Survey (NDNS) database based on micronutrient content, Nova classification and FOPL. The aim was to examine the micronutrient contributions of UK food and drink to UK government dietary micronutrient recommendations for adult females and males, aged 19–64 years, based on the degree of food processing and FOPL. NDNS items were coded into minimally processed food (MPF), processed culinary ingredients, processed food (PF) and UPF, and FOPL traffic lights. MPF, PF and UPF provided similar average contributions per 100 g to micronutrient recommendations. Per 100 kcal, MPF provided the greatest average contribution (14·4 % (interquartile range (IQR): 8·2–28·1)), followed by PF (7·7 % (IQR: 4·6–10·9) and then UPF (5·8 % (IQR: 3·1–9·7)). After adjusting for healthy/unhealthy items (presence of 1+ red FOPL), MPF had higher odds of an above-average micronutrient contribution per 100 kcal than UPF (OR: 5·9 (95 % CI 4·9–7·2)) and PF (OR: 3·2 (95 % CI 2·4–4·2)). MPF were more likely to provide greater contributions to micronutrient recommendations than PF or UPF per 100 kcal. These findings suggest that UPF or PF diets are less likely to meet micronutrient recommendations than an energy-matched MPF diet. The results are important for understanding how consumers perceive the healthiness of products based on FOPL.

This interventional single-centre prospective open-label study aims to evaluate the effects of a vegan diet, compared with a vegetarian and omnivorous diet, on metabolic parameters, insulin sensitivity, and liver and kidney steatosis in healthy adults. The study included fifty-three omnivorous participants aged 18–40 years, BMI 18–30 kg/m2, without any chronic disease, chronic medication use, active smoking or significant alcohol consumption. All participants were omnivorous at baseline and selected to continue an omnivorous diet or transition to a vegetarian or vegan diet, with follow-up over 6 months. Anthropometric measurements, biochemical parameters and liver and kidney steatosis were assessed at baseline and after six months using MRI-proton density fat fraction. Primary outcomes included changes in liver and kidney steatosis, while secondary outcomes were alterations in anthropometric and biochemical markers. Among fifty-three participants, eighteen followed an omnivorous diet, twenty-one adopted a vegetarian diet and fourteen transitioned to a vegan diet. Dietary interventions did not result in statistically significant changes in BMI, fat mass, fat percentage or muscle mass over 6 months. However, statistically significant improvements in systolic and diastolic blood pressure, favouring the vegan diet, were observed. We aimed to control for potentially confounding variables to ensure the reliability of these findings. We have demonstrated a better decline in steatosis at the lower kidney pole, the total hilus and the Liver 6 index in vegans. We demonstrated that a plant-based diet is associated with improvements in several metabolic parameters and may reduce liver and kidney steatosis.

Low vegetable consumption among school-age children and adolescents puts them at risk of micronutrient malnutrition and non-communicable diseases. There is a dearth of synthesised literature on vegetable intake and interventions to promote increased consumption among this age group in West Africa. This study pooled evidence on vegetable consumption and interventions to promote vegetable consumption among school-age children and adolescents (6–19 years) in West Africa. Quantitative and qualitative studies from 2002 to 2023 were electronically searched in PubMed, African Journals Online (AJOL) and Google Scholar databases. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses system was adhered to in reporting this review (PROSPERO ID: CRD42023444444). The Joanna Briggs Institute critical evaluation tool was used to appraise the quality of studies. Forty (40) studies met the search criteria out of n 5080 non-duplicated records. Meta-analysis was not possible due to high heterogeneity. Low vegetable consumption expressed in frequency or amounts was recorded among school-age children and adolescents in the reviewed studies. Intervention studies were mostly among adolescents; the most common type of intervention was the use of nutrition education. Insufficient evidence and high heterogeneity of studies reflect the need for more high-quality interventions using globally identified standards but applied contextually. School-age children appear to be an under-served population in West Africa with regard to nutrition interventions to promote vegetable consumption. There is a need for multi-component intervention studies that encourage vegetable consumption as a food group. Gardening, parental involvement, gamification and goal setting are promising components that could improve the availability, accessibility and consumption of vegetables.

Oxidative stress is present in chronic obstructive pulmonary disease (COPD); however, the effect of increased dietary antioxidants on reducing COPD risk remains unclear. The aim of this study was to investigate the association of the composite dietary antioxidant index (CDAI) with COPD in adults. This study conducted a cross-sectional investigation using data from the National Health and Nutrition Examination Survey (NHANES) spanning from 2015 to March 2020 to explore the association between CDAI and COPD in adults. This study included 9295 participants. Three logistic regression models (crude model, partially adjusted model and fully adjusted model) and restricted cubic spline (RCS) curves were utilised to assess the association between CDAI levels and COPD risk. Subsequently, a two-sample Mendelian randomisation (MR) was employed to analyse the causal impact of antioxidant levels within CDAI on the occurrence of COPD. CDAI levels were inversely associated with COPD after adjusting for confounders (OR = 0·97, 95 % CI 0·95, 1·00), and the association was linear (P < 0·001), and the results of the RCS showed that CDAI was linearly correlated with COPD occurrence (P < 0·001). MR analysis revealed a causal relationship between vitamin C and COPD occurrence (OR = 0·99, 95 % CI 0·98, 1·00, P < 0·05). Our study indicates that dietary sources of antioxidants may reduce the risk of COPD occurrence, and the results of the MR analysis further show that vitamin C is causally associated with a reduced risk of COPD occurrence. However, further exploration is needed to understand how antioxidants prevent COPD.

Colorectal cancer (CRC), the third most common cancer globally, causes over 900 000 deaths annually. Although vitamin D is observed to have potential anti-carcinogenic properties, research findings on its preventable effect against CRC remain inconclusive. Notably, different subsites within the colon and rectum may be associated with distinct risk factors. While some studies have explored this relationship with circulating 25-hydroxyvitamin D (25(OH)D), the results remain contradictory. Our study employed a nested case–control design, involving 775 CRC cases matched with 775 cancer-free controls based on age, region of living and the time of blood sampling. The study was conducted within the Norwegian Women and Cancer post-genome cohort, which comprises approximately 50 000 women. We measured pre-diagnostic circulating plasma 25(OH)D status 5–13 years before diagnosis. Adjustment variables were based on self-administered questionnaires and included BMI, physical activity level, smoking, intake of processed meat, calcium, alcohol and fibre. An increase of 5 nmol/l in 25(OH)D reduced the risk of proximal colon cancer by 6 % (OR = 0·94, 95 % CI 0·89, 0·99). Furthermore, a sensitivity analysis revealed a 62 % increased risk among the women with 25(OH)D levels below 50 nmol/l compared with sufficient levels, ≥ 50 to < 75 nmol/l (OR = 1·62, 95 % CI 1·01, 2·61). No association was found with CRC, colon or distal colon cancer. We observed a subsite-specific association between 25(OH)D and CRC, highlighting the need for further investigation to elucidate the potential underlying mechanisms and clinical implications.

Low iron (Fe) stores at birth may adversely influence child cognitive and motor development. The aims of this study were to assess cord blood Fe levels and explore maternal and neonatal factors associated with Fe status. Cord blood specimens (n 46) were obtained from the BC Children’s Hospital BioBank in Vancouver, Canada. The primary outcome was cord plasma ferritin, measured using sandwich-ELISA. Predictors of interest included maternal age, gestational age, gravidity, infant sex, birth weight and delivery method. Median (interquartile range (IQR)) maternal age and gestational age at delivery was 33·5 (29·3–35·8) years and 36·5 (30·0–39·0) weeks, respectively, and 44 % of infants were female. Median (IQR) cord ferritin was 100·4 (75·7–128·9) µg/l, and 26 % had low Fe status (ferritin <76 µg/l). Among preterm deliveries, a 1-week increase in gestational age was associated with a 6·22 (95 % CI (1·10, 9·52)) µg/l increase in median cord ferritin. However, among term deliveries, a negative trend was observed (–2·38 µg/l per week of gestation (95 % CI (–34·8, 0·78))), indicating a potential non-linear relationship between gestational age and cord ferritin. Female term infants had higher cord ferritin compared with males (β (95 % CI): 30·3 (18·4, 57·9) µg/l), suggesting sex-specific differences in Fe transfer, acquisition and utilisation. Cord ferritin was higher with vaginal deliveries compared with caesarean sections (β (95 % CI): 39·1 (29·0, 51·5) µg/l). Low Fe status may be a concern among infants in Canada; however, further research is needed to inform appropriate thresholds to define optimal Fe status in cord blood.

Indigenous peoples are often not routinely included in iodine programmes because of language barriers and remote access and may thus be at higher risk of iodine deficiency disorders, which could adversely impact their quality of life. We conducted this cross-sectional study in the remote Pwo Karen community of Thailand to determine the urinary iodine concentration of school-aged children and women of reproductive age and investigate the iodine content in household salt. We measured urinary iodine concentration in spot urine samples from healthy school-aged children and women of reproductive age, administered a questionnaire, estimated daily iodine intake and collected household salt samples to determine salt iodine concentration. The median urinary iodine concentration (range) of school-aged children (n 170) was 192 (136–263) µg/l, which was significantly higher than women of reproductive age (n 306) (147 (89–233) µg/l) (P < 0·001). The estimated daily iodine intake in school-aged children and women of reproductive age were 135 and 195 μg/d, respectively. The median (range) iodine concentration in rock and granulated salts consumed in the households were 2·32 (0·52–3·19) and 26·64 (20·86–31·01) ppm, respectively. Surprisingly, the use of iodised salt and the frequency of seafood consumption were NS predictors of urinary iodine concentration in these two groups. Our data suggest that school children and women of the Pwo Karen community have sufficient iodine intake, indicating the Thai salt iodisation programme is effectively reaching even this isolated Indigenous community. Sentinel surveys of remote vulnerable populations can be a useful tool in national iodine programmes to ensure that programme coverage is truly universal.

In this study, nine isonitrogenous experimental diets containing graded levels of carbohydrates (40 g/kg, 80 g/kg and 120 g/kg) and crude lipids (80 g/kg, 120 g/kg and 160 g/kg) were formulated in a two-factor (3 × 3) orthogonal design. A total of 945 mandarin fish with similar body weights were randomly assigned to twenty-seven tanks, and the experiment diets were fed to triplicate tanks twice daily for 10 weeks. Results showed that different dietary treatments did not significantly affect the survival rate and growth performance of mandarin fish. However, high dietary lipid and carbohydrate levels significantly decreased the protein content of the whole body and muscle of cultured fish. The lipid content of the whole body, liver and muscle all significantly increased with increasing levels of dietary lipid, while only liver lipid level was significantly affected by dietary carbohydrate level. Hepatic glycogen content increased significantly with increasing dietary carbohydrate levels. As to liver antioxidant capacity, malondialdehyde content increased significantly with increasing dietary lipid or carbohydrate content, and catalase activity showed an opposite trend. Superoxide dismutase activity increased significantly with increasing levels of dietary lipid but decreased first and then increased with increasing dietary carbohydrate levels. Additionally, the increase in both dietary lipid and carbohydrate levels resulted in a significant reduction in muscle hardness. Muscle chewiness, gumminess and shear force were only affected by dietary lipid levels and decreased significantly with increasing dietary lipid levels. In conclusion, considering all the results, the appropriate dietary lipids and carbohydrate levels for mandarin fish were 120 g/kg and 80 g/kg, respectively.

Ultra-processed foods (UPF), defined using the Nova classification system, are associated with increased chronic disease risk. More recently, evidence suggests the UPF subgroup of whole-grain breads and cereals is in fact linked with reduced chronic disease risk. This study aimed to explore associations of cardiometabolic risk measures with Nova UPF intake v. when foods with ≥ 25 or ≥ 50 % whole grains are excluded from the definition. We considered dietary data from the Australian National Nutrition and Physical Activity Survey 2011–2012. Impacts on associations of UPF intake (quintiles) and cardiometabolic risk measures were analysed using regression models. The median proportion of UPF intake from high whole-grain foods was zero for all quintiles. Participants in the highest Nova UPF intake quintile had significantly higher weight (78·1 kg (0·6)), BMI (27·2 kg/m2 (0·2)), waist circumference (92·7 cm (0·5)) and weight-to-height ratio (0·55 (0·003)) compared with the lowest quintile (P< 0·05). Associations were the same when foods with ≥ 25 and ≥ 50 % whole grains were excluded. Adjusted R-squared values remained similar across all approaches for all outcomes. In Australia, high whole-grain foods considered UPF may not significantly contribute to deleterious cardiometabolic risk associations. Until conclusive evidence on Nova UPF is available, prioritisation should be given to the nutrient density of high whole-grain foods and their potential contribution to improving whole-grain intakes and healthful dietary patterns in Australia.

The COVID-19 pandemic has presented multifaceted challenges globally, impacting adolescent health. Among these, food security and nutrition are intertwined closely with mental health outcomes. In Indonesia, with its diverse socio-economic landscape, these interconnections may have been exacerbated by the pandemic. This study investigated the relationship between food security, nutrition and adolescent mental health in Indonesia during COVID-19. Longitudinal data were collected from 511 adolescent boys and girls in 2021–2022 in Gunungkidul district, Yogyakarta. Food security was measured using the Household Food Insecurity Access Scale (HFIAS), and the validated Kessler-10 Psychological Distress Scale (K10) was used to measure adolescent depression. Multivariate linear regression and linear mixed-effects regression were employed to explore associations between these variables, while adjusting for sex, age, pubertal status and household income. Overall, food insecurity score was positively associated with depressive symptoms (β: 0·72, 95 % CI 0·52, 0·92), while BMI z-score was inversely associated (β: −0·31, 95 % CI 0·68, −0·03). We found an increase in strength of association between food insecurity and depressive symptoms over time (moderately food-insecure: β: 1·36 (95 % CI −0·10, 2·83) to 4·63 (95 % CI 2·17, 7·09); severely food-insecure: β: 1·89 (95 % CI 0·36, 3·41) to 3·30 (95 % CI 1·50, 5·10). Enhancing food access, improving nutritional status and providing mental health support are crucial components of adolescent health.

Research on nutraceutical and dietary interventions in psychiatry has grown substantially, but progress is hindered by methodological inconsistencies and limited reporting standards. To address this, the International Society for Nutritional Psychiatry Research presents the first guidelines on clinical trial design, conduct, and reporting for future clinical trials in this area. Recommendations were developed using a Delphi process including eighteen researchers with considerable clinical trial expertise and experience in either methodology, nutraceutical, or dietary interventions in psychiatry. These guidelines provide forty-nine recommendations for clinical trial design and outcomes, five for trial reporting, and seven for future research priorities. The recommendations included in these guidelines are designed to inform both nutraceutical and dietary clinical trial interventions in Nutritional Psychiatry. Common themes include an emphasis on the importance of a multidisciplinary research team and integration of co-design processes into the conduct and design of clinical research, methods to improve transparency and replicability of trial outcomes, and measures to address common biases in nutrition trials. Furthermore, we provide recommendations for future research including examining a greater variety of nutraceutical and dietary interventions, scalable delivery models, effectiveness and implementation studies, and the need to investigate these interventions in the prevention and management of less studied psychiatric conditions (e.g. schizophrenia and bipolar disorder). Recommendations included within these guidelines are intended to improve the rigor and clinical relevance of ongoing and future clinical trials in Nutritional Psychiatry.

While there is evidence that long-chain n-3 PUFA supplementation benefits mood, the extent to which a single high dose of n-3 PUFA can induce acute mood effects has not been examined. The present study investigated whether a single dose of a DHA-rich powder affects self-reported mood in middle-aged males during elevated cognitive demand. In a randomised, double-blind, placebo-controlled trial with a balanced crossover design, twenty-nine healthy males (age M = 52.8 years, sd = 5.3) were administered a powder (in a meal) containing 4·74 g n-3 PUFA (DHA 4020 mg; EPA 720 mg) or placebo in random order on two different testing days separated by a washout period of 7 ± 3 d. Participants completed mood assessments before and after completing two cognitive test batteries at baseline and again 3·5–4·0 h following the consumption of the active treatment or placebo. While completion of the cognitive test batteries increased negative mood, differential effects for alertness (P = 0·008) and stress (P = 0·04) followed consumption of the DHA-rich powder compared with placebo. Although alertness declined when completing the cognitive batteries, it was higher following consumption of the DHA-rich powder compared with placebo (P = 0·006). Conversely, stress was lower following consumption of the DHA-rich powder relative to placebo, though this difference only approached significance (P = 0·05). Overall, results from this pilot study demonstrate that a single high dose of n-3 PUFA may deliver acute mood benefits following elevated cognitive demand in healthy middle-aged males.

The Mediterranean-Dietary Approach to Stop Hypertension (DASH) Intervention for Neurodegenerative Delay (MIND) diet is a dietary pattern designed to prevent cognitive decline. Dietary adherence is assessed with the MIND diet scoring system, which is currently based on the American diet and serving sizes. It is known that serving sizes and consumed food products differ between countries. Existing literature lacks reporting on food products included within the MIND diet and weight or volume equivalents corresponding to MIND diet servings, impeding accurate comparisons across studies. This study sought to overcome these limitations by evaluating MIND food products consumed in the Dutch context and developing a scoring system based on consumed quantities in weight or volume amounts rather than in standard serving amounts. The third objective was to modify an existing Dutch brief FFQ (Eetscore-FFQ) to evaluate adherence to the MIND diet. We translated nine of the fifteen MIND food groups directly to grams and volumes using the United States Department of Agriculture measurement conversion table. For the remaining food groups, we employed indirect translation to align them as closely as possible to the original MIND diet. These translated quantities in weight and volumes amounts were subsequently rounded to the nearest Dutch household measures, resulting in the culturally adapted MIND-NL diet scoring. The development of the MIND-NL-Eetscore-FFQ, comprising seventy-two food items (forty-one questions), is described. Our adaption approach is reproducible and can be used to customize the MIND diet scoring system to other cultures.

Growing evidence indicates a link between diet and depression risk. We aimed to examine the association between an inflammatory diet index and depression utilising extensive data from UK biobank cohort. The energy-adjusted dietary inflammation index (E-DII) was calculated to quantify the potential of daily diet, with twenty-seven food parameters utilised. The E-DII scores were classified into two categories (low v. high) based on median value. To mitigate bias and ensure comparability of participant characteristics, propensity score matching was employed. To ascertain the robustness of these associations, sensitivity analyses were conducted. Subgroup analyses were performed to evaluate the consistency of these associations within different subpopulations. Totally, 152 853 participants entered the primary analyses with a mean age of 56·11 (sd 7·98) years. Employing both univariate and multivariate logistic regression models, adjustments were made for varying degrees of confounding factors (socio-demographics, lifestyle factors, common chronic medical conditions including type 2 diabetes and hypertension). Results consistently revealed a noteworthy positive correlation between E-DII and depression. In the context of propensity score matching, participants displaying higher E-DII scores exhibited an increased likelihood of experiencing incident depression (OR = 1·12, 95 % CI: 1·05, 1·19; P = 0·000316). Subgroup analysis results demonstrated variations in these associations across diverse subpopulations. The E-value for the point-estimate OR calculated from the propensity score matching dataset was 1·48. Excluding individuals diagnosed with type 2 diabetes or hypertension, the findings consistently aligned with the positive association in the primary analysis. These findings suggested that consumption of a diet with higher pro-inflammatory potential might associated with an increase of future depression risk.

The UK government launched a two-component sugar-reduction programme in 2016, one component is the taxation of sugar-sweetened beverages, the Soft Drinks Industry Levy, and the second is a voluntary sugar reduction programme for products contributing most to children’s sugar intakes. These policies provided incentives both for industry to change the products they sell and for people to change their food and beverage choices through a ‘signalling’ effect that has raised awareness of excess sugar intakes in the population. In this study, we aimed to identify the relative contributions of the supply- and demand-side drivers of changes in the sugar density of food and beverages purchased in Great Britain. While we found that both supply- and demand-side drivers contributed to decreasing the sugar density of beverage purchases (reformulation led to a 19 % reduction, product renewal 14 %, and consumer switching between products 8 %), for food products it was mostly supply-side drivers (reformulation and product renewal). Reformulation contributed consistently to a decrease in the sugar density of purchases across households, whereas changes in consumer choices were generally in the opposite direction, offsetting benefits of reformulation. We studied the social gradient of sugar density reduction for breakfast cereals, achieved mostly by reformulation, and found increased reductions in sugar purchased by households of lower socio-economic status. Conversely, there was no social gradient for soft drinks. We conclude that taxes and reformulation incentives are complementary and combining them in a programme to improve the nutritional quality of foods increases the probability of improvements in diet quality.