To understand the relationship between vitamin D intake and serum 25-hydroxyvitamin D (25(OH)D) levels in a sample of Portuguese adolescents.

Cross-sectional evaluation carried out in the 2003/2004 school year. Vitamin D intake was assessed by an FFQ and 25(OH)D was measured in a fasting blood sample.

Public and private schools in Porto, Portugal.

Adolescents aged 13 years (n 521) enrolled at school (EPITeen cohort).

Both mean (sd) intake and serum 25(OH)D level were far below the recommended, 4·47 (2·49) µg/d and 16·5 (5·7) ng/ml, respectively. A significant difference in serum level was found according to season, with lower values in winter than summer (14·8 (4·6) v. 17·3 (5·9) ng/ml, P<0·001). Vitamin D intake was weakly correlated with serum 25(OH)D (r=0·056, P=0·203).

Dietary vitamin D and serum 25(OH)D levels were positively but weakly correlated and the error was higher among those with higher serum 25(OH)D concentration. Our results support the need for strategies that promote increase of the most important food sources of vitamin D to reduce the high prevalence of low vitamin D status.

Low and high birth weight have been associated with increased risk of type 2 diabetes and CVD. Diet could partly mediate this association, e.g. by intra-uterine programming of unhealthy food preferences. We examined the association of birth weight with diet in Finnish children.

Birth weight standard deviation score (SDS) was calculated using national birth register data and Finnish references. Dietary factors were assessed using 4 d food records. Diet quality was defined by the Finnish Children Healthy Eating Index (FCHEI).

The Physical Activity and Nutrition in Children (PANIC) study.

Singleton, full-term children (179 girls, 188 boys) aged 6–8 years.

Birth weight was inversely associated (standardized regression coefficient β; 95 % CI) with FCHEI (−0·15; −0·28, −0·03) in all children and in boys (−0·27; −0·45, −0·09) but not in girls (−0·01; −0·21, 0·18) after adjusting for potential confounders (P=0·044 for interaction). Moreover, higher birth weight was associated with lower fruit and berries consumption (−0·13; −0·25, 0·00), higher energy intake (0·17; 0·05, 0·29), higher sucrose intake (0·19; 0·06, 0·32) and lower fibre intake (−0·14; −0·26, −0·01). These associations were statistically non-significant after correction for multiple testing. Children with birth weight >1 SDS had higher sucrose intake (mean; 95 % CI) as a percentage of energy intake (14·3 E%; 12·6, 16·0 E%) than children with birth weight of −1 to 1 SDS (12·8 E%; 11·6, 14·0 E%) or <−1 SDS (12·4 E%; 10·8, 13·9 E%; P=0·036).

Higher birth weight may be associated with unhealthy diet in childhood.

To assess produce availability, quality and price in a large sample of food stores in low-income neighbourhoods in California.

Cross-sectional statewide survey.

Between 2011 and 2015, local health departments assessed store type, WIC (Supplemental Nutrition Program for Women, Infants, and Children)/SNAP (Supplemental Nutrition Assistance Program) participation, produce availability, quality and price of selected items in stores in low-income neighbourhoods. Secondary data provided reference chain supermarket produce prices matched by county and month. t Tests and ANOVA examined differences by store type; regression models examined factors associated with price.

Large grocery stores (n 231), small markets (n 621) and convenience stores (n 622) in 225 neighbourhoods.

Produce in most large groceries was rated high quality (97 % of fruits, 98 % of vegetables), but not in convenience stores (25 % fruits, 14 % vegetables). Small markets and convenience stores participating in WIC and/or SNAP had better produce availability, variety and quality than non-participating stores. Produce prices across store types were, on average, higher than reference prices from matched chain supermarkets (27 % higher in large groceries, 37 % higher in small markets, 102 % higher in convenience stores). Price was significantly inversely associated with produce variety, adjusting for quality, store type, and SNAP and WIC participation.

The study finds that fresh produce is more expensive in low-income neighbourhoods and that convenience stores offer more expensive, poorer-quality produce than other stores. Variety is associated with price and most limited in convenience stores, suggesting more work is needed to determine how convenience stores can provide low-income consumers with access to affordable, high-quality produce. WIC and SNAP can contribute to the solution.

To examine the trend in social inequality in low intake of vegetables among adolescents in Denmark from 2002 to 2014 using occupational social class (OSC) as socio-economic indicator.

Repeated cross-sectional school surveys including four waves of data collection in 2002–2014. The analyses focused on absolute social inequality (difference between high and low OSC in low vegetable intake) as well as relative social inequality (OR for low vegetable intake by OSC).

The nationally representative Health Behaviour in School-aged Children (HBSC) study in Denmark.

The study population was 11–15-year olds (n 17 243).

Low intake of vegetables was defined as less than weekly intake measured by food frequency items. OSC was measured by student reports of parents’ occupation. The proportion of participants who reported eating vegetables less than once weekly was 8·9 %, with a notable decrease from 11·9 % in 2002 to 5·9 % in 2014. The OR (95 % CI) for less than weekly vegetable intake was 2·28 (1·98, 2·63) in the middle compared with high OSC and 3·12 (2·67, 3·66) in the low compared with high OSC. The absolute social inequality in low vegetable intake decreased from 2002 to 2014 but the relative social inequality remained unchanged.

The study underscores that it is important to address socio-economic factors in future efforts to promote vegetable intake among adolescents. The statistical analyses of social inequality in vegetable intake demonstrate that it is important to address both absolute and relative social inequality as these two phenomena may develop differently.

To determine the prevalence of and modifiable factors associated with prediabetes in the Comcáac Indians.

Cross-sectional study where prediabetes was defined using fasting plasma glucose, 2 h plasma glucose and glycated Hb (HbA1c). Physical, anthropometric and biochemical measurements, medical record, socio-economic, dietary and physical activity information were collected. The modifiable factors associated with prediabetes were assessed using multiple logistic regression.

Comcáac Indian communities of Punta Chueca and El Desemboque, Sonora, Mexico.

Adults (n 227) aged 20 years or older.

The sex- and age-adjusted prevalence (95 % CI) of prediabetes was 47·1 (40·8, 53·5) % in the overall population; age-adjusted prevalence was 47·3 (35·6, 59·0) % in men and 46·7 (39·1, 54·3) % in women. The modifiable factors associated with a risk of prediabetes (OR; 95 % CI) were light-intensity physical activity (per 1 h/week increase: 1·04; 1·01, 1·07) and insulin resistance (homeostasis model assessment of insulin resistance score >6·1 v. <4·1: 4·62; 1·37, 15·51). Increased consumption of a traditional dietary pattern based on fish and seafood, low-fat cereals, fruits and vegetables was a protective factor (0·49; 0·31, 0·79). All variables were modelled together and adjusted for age and sex.

The high prediabetes prevalence found in the Comcáac community is alarming because it represents a large number of people who are at risk for type 2 diabetes. The identification of modifiable factors associated with prediabetes that are specific to this population may be useful for designing effective strategies to prevent prediabetes.

Dietary protein plays a role in counteracting age-related muscle loss. However, limited long-term data exist on protein intake and markers of cardiometabolic health, which tend to deteriorate with age.

Prospective cohort study. FFQ-derived protein intake (g/d) and cardiometabolic markers were assessed up to five times across 20 years. Markers included systolic (SBP) and diastolic (DBP) blood pressures, circulating lipids (total, HDL and LDL cholesterol; TAG), estimated glomerular filtration rate (eGFR), fasting glucose (FG), weight and waist circumference (WC). Mixed models accounting for repeated measures were used to estimate adjusted mean annualized changes in outcomes per quartile category of protein.

Framingham Heart Study Offspring cohort, USA.

Participants (n 3066) with 12 333 unique observations, baseline (mean (sd)) age=54·0 (9·7) years, BMI=27·4 (4·9) kg/m2, 53·5 % female.

In fully adjusted models, there were favourable associations (mean (se)) of total protein with annualized changes in SBP (lowest v. highest intake: 0·34 (0·06) v. 0·04 (0·06) mmHg, P trend=0·001) and eGFR (−1·03 (0·06) v. −0·87 (0·05) ml/min per 1·73 m2, P trend=0·046), unfavourable associations with changes in FG (0·013 (0·004) v. 0·028 (0·004) mmol/l, P trend=0·004) and no associations with weight, WC, DBP or lipids. Animal protein was favourably associated with SBP and unfavourably with FG and WC; plant protein was favourably associated with FG and WC.

The present study provides evidence that protein intake may influence changes in cardiometabolic health independent of change in weight in healthy adults and that protein’s role in cardiometabolic health may depend on the protein source.

In a representative sample of US adults, we investigated the associations of nutrient patterns (NP) with bone mineral density (BMD) and fractures.

Cross-sectional.

US community-based National Health and Nutrition Examination Survey (NHANES).

Participants with measured data on dietary intake and BMD from 2005 to 2010 were included. Principal components analysis was used to identify NP. BMD was measured using dual-energy X-ray absorptiometry. ANCOVA, adjusted logistic and linear regression models were employed, accounting for the complex survey design and sample weights.

We included a total of 18 318 participants, with 47·0 % (n 8607) being men. The mean age was 45·8 years with no sex difference. Three NP emerged, explaining 55·9 % of the variance in nutrient consumption. Multivariable-adjusted linear regressions revealed significant inverse associations between the ‘high-energy’ NP (rich in carbohydrates and sugar, total fat and saturated fat) and total femur, femoral neck, trochanter and intertrochanter BMD (β coefficient: −0·029, −0·025, −0·034 and −0·021, respectively, all P<0·001), while there were significant associations between the ‘nutrient-dense’ NP (rich in vitamins, minerals and fibre) and ‘healthy fat’ NP (high dietary PUFA and MUFA) and BMD at total femur, femoral neck, trochanter and intertrochanter (all P<0·001). In adjusted logistic regression models, the odds of hip, wrist or spine fractures did not vary significantly across NP quartiles.

Nutrient-dense and healthy fat NP are associated with higher BMD at various bone sites, while the high-energy NP is inversely associated with BMD measures.

The goal of the present study was to use a methodology that accurately and reliably describes the availability, price and quality of healthy foods at both the store and community levels using the Nutrition Environment Measures Survey in Stores (NEMS-S), to propose a spatial methodology for integrating these store and community data into measures for defining objective food access.

Two hundred and sixty-five retail food stores in and within 2 miles (3·2 km) of Flint, Michigan, USA, were mapped using ArcGIS mapping software.

A survey based on the validated NEMS-S was conducted at each retail food store. Scores were assigned to each store based on a modified version of the NEMS-S scoring system and linked to the mapped locations of stores. Neighbourhood characteristics (race and socio-economic distress) were appended to each store. Finally, spatial and kernel density analyses were run on the mapped store scores to obtain healthy food density metrics.

Regression analyses revealed that neighbourhoods with higher socio-economic distress had significantly lower dairy sub-scores compared with their lower-distress counterparts (β coefficient=−1·3; P=0·04). Additionally, supermarkets were present only in neighbourhoods with <60 % African-American population and low socio-economic distress. Two areas in Flint had an overall NEMS-S score of 0.

By identifying areas with poor access to healthy foods via a validated metric, this research can be used help local government and organizations target interventions to high-need areas. Furthermore, the methodology used for the survey and the mapping exercise can be replicated in other cities to provide comparable results.

Young children lack basic skills related to recognizing the types of foods they consume and dietary surveys often rely on parents’ response. The present study aimed to evaluate how well parents of children aged from 3 months to 10 years perceive images of portions of foods commonly consumed by young children.

Pre-weighed, actual food portions (n 2314) were shown to the study participants who were asked to indicate the picture that corresponded to the food in view. Mean differences between picture numbers selected and shown were estimated and compared using unpaired t tests or Tukey–Cramer pairwise comparisons.

Real-time testing of parents’ perception of food images presenting portion sizes consumed by children up to 10 years old.

A convenience sample of 138 parents/caregivers of young children (69 % females).

Individuals selected the correct or adjacent image in about 97 % of the assessments. Images presenting amorphous solids (i.e. pies and pastries with a filling), liquid or semi-liquid dishes (i.e. soups, porridges, fruit and vegetable purées) were more prone to bias. There was no indication that personal characteristics (gender, age, educational background, age, number of offspring) were associated with differences in the way parents/caregivers perceived the food pictures.

Food pictures may not be appropriate to quantify the intake of liquid, semi-liquid or amorphous solid foods in surveys addressing young children and studies evaluating their performance as food portion anchors should ensure the inclusion of several and various amorphous foods in the assessment.

Folic acid supplementation and folate-rich diets are recommended for women of childbearing age worldwide to prevent congenital anomalies. We aimed to determine the current status of folic acid supplementation among pregnant Japanese women and identify means to increase the intake of these supplements.

Cross-sectional study.

A total of 1862 pregnant women who consulted the perinatal centre from September 2014 to December 2015 completed a questionnaire concerning knowledge about folic acid, sources of information and the use of folic acid supplements.

Osaka Medical Center and Research Institute for Maternal and Child Health (Japan).

In our study population, only 20·5 % of pregnant women took folic acid supplements periconceptionally even though 70·4 % knew about the protective effect of folic acid. A multivariate analysis demonstrated that age ≥35 years (OR=2·80; 95 % CI 1·24, 6·29) and knowledge of the benefits of folic acid (OR=2·64; 95 % CI 1·92, 3·62) were associated with periconceptional folic acid use, and multiparity was negatively associated with such use. Compared with those who took folic acid supplements periconceptionally, women who did not take supplements received information through passive and less interactive media.

Although folic acid awareness was relatively high among pregnant Japanese women, folic acid supplementation before conception was insufficient. To increase the intake of folic acid supplements in countries in which foods are not fortified with folic acid, an effective public health approach promoting behavioural change is necessary for women of reproductive age.

To identify home environment factors associated with toddler dietary behaviours using ecological momentary assessment (EMA).

Home environment and toddler’s diet were assessed by mothers through EMA (random beeps over ≤8 d and a brief survey). Dietary outcomes were fruit/vegetable consumption, eating episode (‘snack’ v. ‘meal’) and sugar-sweetened beverage (SSB) consumption. Home environment factors included interacting with mother, eating alone/with others, eating in a high chair/chair at the table, watching television and movement/translocation. Multilevel logistic mixed-effects regression models assessed both within- (individual toddlers across time) and between- (toddlers-on-average) subject effects.

Low-income mother–toddler dyads (n 277).

Urban and suburban Maryland, USA.

EMA captured eating/drinking episodes for 249/277 (89·9 %) toddlers (883 eating episodes, 1586 drinking episodes). Toddlers-on-average were more likely (adjusted OR, P value) to eat fruit/vegetables when not moving around (0·43, P=0·043), eat with the television off (0·33, P<0·001) and eat in a high chair/chair (3·38, P<0·001); no within-subject effects were shown. For eating episodes, both toddlers-on-average and individual toddlers were more likely to eat snacks when not in a high chair/chair (0·13, P<0·001 and 0·06, P<0·001, respectively) and when eating alone (0·30, P<0·001 and 0·31, P<0·001, respectively). Also, individual toddlers were more likely to eat snacks when moving around (3·61, P<0·001). Toddlers-on-average were more likely to consume SSB when not in a high chair/chair (0·21, P=0·001), eating alone (0·38, P=0·047) or during a snacking episode (v. a meal: 3·96, P=0·012); no within-subject effects shown.

Factors in the home environment are associated with dietary behaviours among toddlers. Understanding the interplay between the home environment and toddler diet can inform future paediatric dietary recommendations.

An index of biomarkers derived from dietary factors (diet–biomarker-related index) identifies foods and nutrients that encompass physiological potentials and provides scientific evidence for dietary patterns that increase the risk of disease associated with specific biomarkers. Although men and women have different dietary patterns and physiological characteristics, sex is not often considered when investigators develop a diet–biomarker-related index. We aimed to review whether epidemiological studies developed diet–biomarker-related indices in a sex-specific way.

We systematically searched for epidemiological studies that developed diet–biomarker-related indices, including (i) biomarker prediction indices that include dietary factors as explanatory variables and (ii) dietary patterns to explain biomarker variations, in the PubMed and EMBASE databases. We qualitatively reviewed the sex consideration in index development.

We identified seventy-nine studies that developed a diet–biomarker-related index. We found that fifty-four studies included both men and women. Of these fifty-four studies, twenty-nine (53·7 %) did not consider sex, eleven (20·3 %) included sex in the development model, seven (13·0 %) considered sex but did not include sex in the development model, and seven (13·0 %) derived a diet–biomarker-related index for men and women separately. A list of selected dietary factors that explained levels of biomarkers generally differed by sex in the studies that developed a diet–biomarker-related index in a sex-specific way.

Most studies that included both men and women did not develop the diet–biomarker-related index in a sex-specific way. Further research is needed to identify whether a sex-specific diet–biomarker-related index is more predictive of the disease of interest than an index without sex consideration.

To examine the associations of individual and food environmental factors with fruit and vegetable (F&V) intake in a city in a low-to-middle-income country (LMIC).

Cross-sectional.

Representative sample of the Brazilian Primary Care service known as the Health Academy Program (HAP) in Belo Horizonte, a Brazilian city.

Using a conceptual model as a guide, individual and food environment data were obtained through: (i) face-to-face interviews with participants aged 20 years or older; and (ii) F&V food store audits. A broad set of individual, household, and community and consumer nutrition environment variables was investigated. Multilevel linear regression was used to quantify area-level variations in F&V intake and to estimate associations with the factors.

Eighteen HAP centres were selected and 2944 participants and 336 food stores were included. F&V intake varied between contexts, being higher in areas with better socio-economic conditions and food store quality, such as specialised F&V markets. Individual-level factors, including age, income, food insecurity, stage of change, self-efficacy and decisional balance, were significantly associated with F&V intake. After controlling for individual-level characteristics, greater F&V intake was also associated with higher quality of food stores.

In one of the first studies to comprehensively assess the food environment in an LMIC, individual-level factors accounted for the largest variation in F&V intake; however, the food environment was also important, because area-level variables explained 10·5 % of the F&V intake variation. The consumer nutrition environment was more predictive of healthy eating than was the community nutrition environment. The findings suggest new possibilities for interventions.

(i) To determine the current state of online grocery shopping, including individuals’ motivations for shopping for groceries online and types of foods purchased; and (ii) to identify the potential promise and pitfalls that online grocery shopping may offer in relation to food and beverage purchases.

PubMed, ABI/INFORM and Google Scholar were searched to identify published research.

To be included, studies must have been published between 2007 and 2017 in English, based in the USA or Europe (including the UK), and focused on: (i) motivations for online grocery shopping; (ii) the cognitive/psychosocial domain; and (iii) the community or neighbourhood food environment domain.

Our search yielded twenty-four relevant papers.

Findings indicate that online grocery shopping can be a double-edged sword. While it has the potential to increase healthy choices via reduced unhealthy impulse purchases, nutrition labelling strategies, and as a method to overcome food access limitations among individuals with limited access to a brick-and-mortar store, it also has the potential to increase unhealthy choices due to reasons such as consumers’ hesitance to purchase fresh produce online.

Additional research is needed to determine the most effective ways to positively engage customers to use online grocery shopping to make healthier choices.

To investigate the psychometric properties, validity and reliability of a newly developed measure of food insecurity, the Household Food and Nutrition Security Survey (HFNSS), among an Australian population.

Cross-sectional study.

Metropolitan areas of Melbourne, Australia, identified as very high, high or medium vulnerability in the 2008 Vulnerability Assessment for Mortgage, Petrol and Inflation Risks and Expenditure index.

A convenience sample of 134 adults (117 females and fifteen males, aged over 18 years).

Rasch modelling and factor analysis identified four items for exclusion. The remaining items yielded excellent reliability among the current sample and assessed three underlying components: the adult experience of food insecurity (component one), initial/periodic changes to children’s food intakes (component two) and progressive/persistent decreases in children’s food intakes (component three). Compared with the widely used US Department of Agriculture Food Security Survey Module, the HFNSS identified a significantly higher proportion of food insecurity; this is likely due to the HFNSS’s identification of food insecurity due to reasons other than (and including) limited financial access.

The HFNSS may be a valid and reliable tool for the assessment of food insecurity among the Australian population and provides a means of assessing multiple barriers to food security beyond poor financial access (which has been identified as a limitation of other existing tools). Future research should explore the validity and reliability of the tool among a more representative sample, as well as specifically among vulnerable population subgroups.

A direct way to evaluate food-based dietary guidelines is to assess if adherence is associated with development of non-communicable diseases. Thus, the objective was to develop an index to assess adherence to the 2013 Danish food-based dietary guidelines and to investigate the association between adherence to the index and risk of myocardial infarction (MI).

Population-based cohort study with recruitment of participants in 1993–1997. Information on dietary intake was collected at baseline using an FFQ and an index ranging from 0 to 6 points was created to assess adherence to the 2013 Danish food-based dietary guidelines. MI cases were identified by record linkage to the Danish National Patient Register and the Causes of Death Register. Cox proportional hazards models were used to estimate hazard ratios (HR) of MI.

Greater areas of Aarhus and Copenhagen, Denmark.

Men and women aged 50–64 years (n 55 021) from the Diet, Cancer and Health study.

A total of 3046 participants were diagnosed with first-time MI during a median follow-up of 16·9 years. A higher Danish Dietary Guidelines Index score was associated with a lower risk of MI. After adjustment for potential confounders, the hazard of MI was 13 % lower among men with a score of 3–<4 (HR=0·87; 95 % CI 0·78, 0·96) compared with men with a score of <3. The corresponding HR among women was 0·76 (95 % CI 0·63, 0·93).

Adherence to the 2013 Danish food-based dietary guidelines was inversely associated with risk of MI.

The Newest Vital Sign© (NVS) was developed in the USA to measure patient health literacy in clinical settings. We adapted the NVS for use in Canada, in English and French, and created a computerized version. Our objective was to evaluate the reliability of the Canadian NVS as a self-administered computerized tool.

We used a randomized crossover design with a washout period of 3–4 weeks to compare health literacy scores obtained using the computerized version with scores obtained using the standard interviewer-administered NVS. ANOVA models and McNemar’s tests assessed differences in outcomes assessed with each version of the NVS and order effects of the testing.

Participants were recruited from multicultural catchment areas in Ontario and Nova Scotia.

English- and French-speaking adults aged 18 years or older.

A total of 180 (81 %) of the 222 adults (112 English/110 French) initially recruited completed both the interviewer-NVS and computer-NVS. Scores for those who completed both assessments ranged from 0 to 6 with a mean of 3·63 (sd 2·11) for the computerized NVS and 3·41 (sd 2·21) for the interview-administered NVS. Few (n 18; seven English, eleven French) participants’ health literacy assessments differed between the two versions.

Overall, the computerized Canadian NVS performed as well as the interviewer-administered version for assessing health literacy levels of English- and French-speaking participants. This Canadian adaptation of the NVS provides Canadian researchers and public health practitioners with an easily administered health literacy assessment tool that can be used to address the needs of Canadians across health literacy levels and ultimately improve health outcomes.

The present study aimed to identify food patterns among 2–9-year-olds and investigate sociodemographic, anthropometric and behavioural predictors of less healthy dietary patterns.

Cross-sectional study. Parents of 2–9-year-olds completed an FFQ and factor analysis was applied to identify dietary patterns. Parents also completed questionnaires assessing sociodemographic, anthropometric and behavioural characteristics of parents and children, including parental feeding practices.

Participants were recruited from private schools of Campinas and São Paulo, SP, Brazil, between April and June 2014.

Parents of 2–9-year-olds (n 929).

Two dietary patterns emerged: ‘traditional food’ and ‘ultra-processed food’. Lower maternal education (OR=2·05, P=0·010) and higher maternal weight status (OR=1·43, P=0·044) were associated with a greater likelihood of the ultra-processed food pattern. Lower perceived parental responsibility for adequacy of food group intake (OR=2·41, P=0·020), and lower scores on the parental feeding practices of ‘Healthy Eating Guidance’ (OR=1·83, P<0·001) and ‘Monitoring’ (OR=2·52, P<0·001), were also associated with the presence of this pattern, as was higher child’s screen use during mealtimes (OR=1·61, P=0·004).

The present study is the first to evaluate associations between less healthy dietary patterns of Brazilian 2–9-year-olds and parental feeding practices. Our findings highlight sociodemographic, anthropometric and behavioural factors within families that could be used to target tailored policies to at-risk populations.

To assess ethnicity- and age-modified associations between mid-upper arm circumference (MUAC) and mortality in Nepalese children and whether sociodemographic factors explain these associations.

Secondary data analysis of children followed until 3 years of age. Estimated mortality hazard ratios (HR) for MUAC<11·5cm (recommended cut-off for identifying severe acute malnutrition among children ≥6 months old) compared with ≥11·5cm in younger (<6 months) and older children (≥6 months) of Pahadi and Madhesi ethnicity, adjusting for sex, socio-economic status (SES) and mother’s education using Cox proportional hazard models.

Sarlahi, Nepal (21 October 2001–2 February 2006).

Children (n 48 492) enrolled in the Nepal Nutrition Intervention Project, Sarlahi-4.

Among children aged ≥6 months, MUAC<11·5 cm was associated with increased risk of mortality in both Pahadis (HR=4·01; 95 % CI 1·42, 11·76) and Madhesis (HR=5·60; 95 % CI 3·87, 8·11) compared with those with MUAC≥11·5 cm, after adjusting for sex, SES and maternal literacy. Among children <6 months old, MUAC<11·5 cm was not associated with mortality in Pahadis with (HR=1·12; 95 % CI 0·72, 1·73) or without adjusting (HR=1·17; 95 % CI 0·75, 1·18) as compared with Madeshis (adjusted HR=1·76; 95 % CI 1·35, 2·28).

Among older children, MUAC<11·5 cm is associated with subsequent mortality in both ethnicities regardless of other characteristics. However, among children aged <6 months, it predicted mortality only among Madhesis, while sociodemographic factors were more strongly associated with mortality than MUAC<11·5cm among Pahadis.