We estimate attributable fractions, deaths and years of life lost among infants and children ≤2 years of age due to suboptimal breast-feeding in developing countries.

We compare actual practices to a minimum exposure pattern consisting of exclusive breast-feeding for infants ≤6 months of age and continued breast-feeding for older infants and children ≤2 years of age. For infants, we consider deaths due to diarrhoeal disease and lower respiratory tract infections, and deaths due to all causes are considered in the second year of life. Outcome measures are attributable fractions, deaths, years of life lost and offsetting deaths potentially caused by mother-to-child transmission of HIV through breast-feeding.

Developing countries.

Infants and children ≤2 years of age.

Attributable fractions for deaths due to diarrhoeal disease and lower respiratory tract infections are 55% and 53%, respectively, for the first six months of infancy, 20% and 18% for the second six months, and are 20% for all-cause deaths in the second year of life. Globally, as many as 1.45 million lives (117 million years of life) are lost due to suboptimal breast-feeding in developing countries. Offsetting deaths caused by mother-to-child transmission of HIV through breast-feeding could be as high as 242 000 (18.8 million years of life lost) if relevant World Health Organization recommendations are not followed.

The size of the gap between current practice and recommendations is striking when one considers breast-feeding involves no out-of-pocket costs, that there exists universal consensus on best practices, and that implementing current international recommendations could potentially save 1.45 million children's lives each year.

To investigate what factors relate most strongly to breast-feeding duration in order to successfully support breast-feeding mothers.

Prospective birth cohort study using questionnaires, routinely collected weights and health check at age 13 months.

Gateshead, UK.

Parents of 923 term infants born in a defined geographical area and recruited shortly after birth, 50% of whom were breast-feeding initially.

Only 225 (24%) infants were still breast-fed at 6 weeks, although 136 (15%) continued beyond 4 months. Infants in the most affluent quintile were three times more likely to be initially breast-fed (P < 0.001) and five times more likely to still be feeding at 4 months (P = 0.001) compared with infants in the most deprived quintile. A third of breast-fed infants were given supplementary feeds in the maternity unit and this was associated with a 10-fold increase in odds of giving up breast-feeding by discharge (P = 0.001). Frequent feeding was reported as a reason for giving up in 70% of mothers at 6 weeks and 55% at 4 months. Those infants who stopped breast-feeding earliest showed the most rapid weight gain and were tallest at age 13 months. Non-breast-fed infants had 50% more family doctor contacts up to age 4 months (P = 0.005).

Initiation of breast-feeding in urban Britain remains strongly determined by socio-economic background and early cessation seems to be related to frequent feeding and rapid growth as well as a continuing failure to eradicate health practices that undermine breast-feeding. Those infants not receiving breast milk suffered increased morbidity, but the apparent association between breast-feeding duration and growth probably reflects reverse causation.

To present current breast-feeding rates for Pacific infants resident in New Zealand. Reasons for the introduction of complementary liquid foods were also explored.

A longitudinal study using hospital discharge summary records and maternal home interviews undertaken at 6 weeks, 12 and 24 months postpartum. Turnbull's non-parametric survival analysis was used to model exclusive breast-feeding rates.

Auckland, New Zealand.

The cohort comprised 1376 infants at 6 weeks, 1223 infants at 12 months and 1142 infants at 24 months. Exclusive breast-feeding rates at hospital discharge, 6 weeks, 3 and 6 months postpartum were 84% (95% confidence interval (CI): 80–88%), 49% (95% CI: 43–55%), 37% (95% CI: 32–42%) and 9% (95% CI: 7–11%), respectively. Significant ethnic difference existed, with Samoan mothers having higher exclusive breast-feeding rates than Tongan mothers (P = 0.002). The percentage of infants receiving any breast milk at hospital discharge, 6 weeks, 12 and 24 months was 96% (95% CI: 94–97%), 95% (95% CI: 94–96%), 31% (95% CI: 28–34%) and 15% (95% CI: 13–17%), respectively. Again ethnic differences emerged. Common reasons cited for discontinuation of exclusive breast-feeding included uncertainty of breast milk supply (56%), problems with breasts (30%) and difficulties breast-feeding in work or educational environments (26%). However, 691 (50%) mothers sought no advice about their breast-feeding concerns within the first six weeks of life.

Exclusive breast-feeding rates for Pacific infants are ethnically heterogeneous, have declined since the 1990s and fall short of the World Health Organization recommendations. The principal reasons cited for exclusive breast-feeding discontinuation echo those reported over a decade ago.

To describe feeding patterns and mothers' perceptions of desirable feeding practices in low-birth-weight (LBW) infants after hospital discharge in England and to test for the association of inappropriate practices/perceptions with identifiable demographics.

Postal questionnaires from 198 mothers of LBW infants in London, Liverpool and Winchester were analysed regarding infant demographics, mothers' demographics, infant milks used, solid feeding practices and mothers' perceptions of infant feeding practices.

At birth, the median weight, independent of gestational age, was 1.80 kg (range 0.6–2.50 kg) and 37.1% of infants received breast milk exclusively; 83.7% were breast-fed at some stage, a practice favoured more by first-time mothers (90.2%) than multiparous mothers (73.4%) (P = 0.004) as well as by mothers of higher social groups (P> = 0.019). The median age of solid food introduction was 17 postnatal weeks (range 8–36 weeks); the timing correlated strongly with infant birth weight (P < 0.001). A high-fibre diet and a low-fat diet were incorrectly considered important for their infants by 67.1% and 51.6% of mothers, respectively. Regarding a high-fibre diet as important was associated with being a multiparous mother (P = 0.006), while regarding a low-fat diet as important was associated with low social group (P = 0.018). A quarter of mothers did not consider 'plenty of calories' to be important for their infants, reflecting similar attitudes in 1990 for mothers of term infants; this incorrect attitude exhibited an association with being a first-time mother (P = 0.047).

Infant feeding practices were very variable, showing poor concordance with national and international recommendations for term infants. In selecting foods for their infants, many mothers appeared to be applying principles more appropriate to planning a healthy diet for adults. There is an urgent need to develop and implement evidence-based guidelines for feeding LBW infants after hospital discharge, with special attention given to multiparous mothers as well as mothers from lower social groups.

To describe women's views about aspects of infants' diets around the time of weaning, making comparisons with national guidelines.

A survey of women with a 9-month-old child.

Adelaide, South Australia.

Five hundred and five women who joined a longitudinal study during pregnancy.

Sources of information varied, with written material most commonly used (37%). Cows' milk was considered suitable as the main drink for weaned infants by 14% of women. There were divergent views about the suitability of eggs, with many women concerned about allergy. The majority of women (84%) viewed fruit juice as suitable although many qualified their response, often by stating that fruit juice should be diluted. Almost all women considered the amount of sugar mattered, primarily because of tooth decay, and that salt mattered although the reason was often uncertain. It was widely believed (77%) that additives in food could cause health problems, in particular hyperactivity and allergies, and half of the women reported avoiding specific foods because of concerns about allergies. Many women thought that giving their child food that was high in fat would encourage a liking for ‘junk’ food or lead to fatness in adulthood.

There is considerable diversity in the views women express about aspects of infant feeding that have been the subject of guidelines. Further health promotion efforts are needed to achieve greater consistency with recommendations and to address other concerns women have. This will entail greater engagement with parents and shared development of responses.

To describe the process of establishing and implementing a social support infant feeding intervention.

This paper outlines the initial stages of a randomised controlled trial which assessed the effectiveness of a social support intervention on a range of infant feeding outcomes. Details are presented of the processes involved in recruiting, training and supporting a group of volunteers who provided support to the study sample.

Camden and Islington, London, UK.

Initial networking with local agencies and organisations provided invaluable information and contacts. Employing a dedicated volunteer co-ordinator is vitally important in the recruitment, training and support of volunteers. Providing child care and travel expenses is an essential incentive for volunteers with young children. Advertisements placed in local newspapers were the most successful means of recruiting volunteers. Appropriate training is needed to equip volunteers with the necessary knowledge and skills to provide effective support. Particular emphasis in the training focused upon developing the necessary interpersonal skills and self-confidence. The evaluation of the training programme demonstrated that it improved volunteers’ knowledge and reported confidence. The provision of ongoing support is also essential to maintain volunteers’ interest and enthusiasm. The retention of volunteers is, however, a key challenge.

The processes outlined in this paper have demonstrated the feasibility of successfully establishing, implementing and maintaining a community-based social support infant feeding programme. The experiences described provide useful insights into the practical issues that need to be addressed in setting up a social support intervention.

The goal of the present work was to examine the prevalence and associated risk factors of undiagnosed diabetes among urban Moroccan Sahraoui women.

Randomised sample of adult women living in the city of Laayoune in south Morocco who visited public health centres during an immunisation campaign. Body weight, height, waist and hip circumferences, blood pressure, fasting plasma glucose (FPG), triglycerides, dietary intake and physical activity were collected.

Data were obtained on 249 urban women aged 15 years and older, who were not pregnant. Only subjects identified as of Sahraoui origin were eligible for this investigation.

The prevalence of impaired fasting glucose (IFG) was 5.5% and that of undiagnosed diabetes 6.4%. Diabetes and IFG were more common among older and obese women as well as among women with hypertension or a family history of diabetes. In addition, sucrose intakes were higher in women with diabetes than in those with normal FPG. Also, physical activity estimated as the time spent in walking was negatively associated with FPG. Regression analyses showed an independent association of age, obesity, family history of diabetes and triglycerides with diabetes.

The high proportion of unknown diabetes suggests the need for increased diabetes awareness in this population. The data suggest also the involvement of obesity in diabetes and the potential importance of intervention strategies to reduce population adiposity for the prevention and management of cardiovascular risk factors.

To evaluate the relationship between dietary diversity score (DDS) and cardiovascular risk factors in Tehranian adults.

In this population-based cross-sectional study, a representative sample of 581 subjects (295 males and 286 females) aged over 18 years, residents of Tehran, participated. Hypercholesterolaemia, hypertriglyceridaemia and low high-density lipoprotein cholesterol were defined according to the Adult Treatment Panel III guidelines of the National Cholesterol Education Program. Diabetes was defined as fasting plasma glucose concentration ≥ 126 mg dl−1 or 2-h post challenge glucose concentration ≥ 200 mg dl−1. Hypertension was defined on the basis of the sixth report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure. Dietary diversity was defined according to the Diet Quality Index revised.

Mean (±standard deviation) DDS was 6.15 ± 1.02. The probability of having diabetes (odds ratio (OR) among quartiles: 1.45, 1.26, 1.11 and 1.00, respectively; P for trend = 0.04) and hypertriglyceridaemia (OR = 1.41, 1.23, 1.05 and 1.00, respectively; P for trend = 0.04) decreased with increasing quartile of the diversity score for whole grains. The probability of having obesity (OR among quartiles: 1.39, 1.06, 1.03 and 1.00, respectively; P for trend = 0.03), hypercholesterolaemia (OR = 1.46, 1.28, 1.11 and 1.00, respectively; P for trend = 0.03), hypertension (OR = 1.32, 1.17, 1.13 and 1.00, respectively; P for trend = 0.03) and high low-density lipoprotein cholesterol (LDL-C) (OR = 1.25, 1.12, 1.07 and 1.00, respectively; P for trend = 0.04) decreased with increasing quartile of the diversity score for vegetables. The probability of having hypercholesterolaemia, high LDL-C, hypertension and diabetes decreased with quartile of the DDS. But the probability of being obese increased with quartile of the DDS (P for trend = 0.03).

DDS was inversely associated with cardiovascular risk factors in this cross-sectional study. Increased diversity scores of diets, to increase the variety score for vegetables, may be emphasised in programmes attempting to bring about changes in lifestyle.

Cross-sectional studies have suggested an association between glycaemic index (GI) or glycaemic load (GL) and serum lipids. However, no prospective studies have been performed.

To examine whether GI or GL was associated with subsequent changes in serum lipids.

Prospective study with 6 years of follow-up. Overall dietary GI and GL of each participant were assessed from diet history interviews.

Population-based study.

Three hundred and thirty-five healthy men and women aged 35–65 years selected randomly from a larger sample of Danish adults.

In men GI was directly related to changes in total cholesterol (ΔTC), regression coefficient (β) = 0.0044 (95% confidence interval (CI): 0.0008–0.0081) and GL was positively related to changes in low-density lipoprotein cholesterol (ΔLDL), β = 0.1554 (95% CI: 0.0127–0.2982). Furthermore, the relationship between GL and ΔTC was modified by age, being particularly strong for the younger men (P = 0.02). In women the relationship between GI and ΔLDL was modified by age and was stronger for younger rather than older women (P = 0.01). A tendency for a similar interaction was seen for GI and ΔTC (P = 0.09). Associations between GL and ΔLDL and GL and ΔTC were inverse for women with body mass index ≥ 30 kg m−2 (P = 0.03 and 0.04, respectively).

This is the first study to demonstrate that dietary GI and GL are related to 6-year changes in serum lipid levels. However, associations were weak and generally confined to subgroups.

To investigate the relationships of two main physical activity domains (during leisure and at work) with cardiovascular risk factors and eating habits.

Cross-sectional study.

Preventive medicine centre.

In 5478 adults (32% women, aged 20–80 years) who consecutively underwent a standardised health examination, leisure-time physical activity (LTPA; i.e. non-sport leisure and sport activities), occupational physical activity (OPA) and eating habits were assessed by self-administered questionnaires. We analysed sex-specific relationships of LTPA and OPA (in quartiles) with (1) various cardiovascular risk factors and (2) eating habits using analysis of variance and logistic regression, respectively.

In both genders, with and without adjustment for education in addition to age, LTPA was associated negatively with body mass index, body fat, waist circumference, resting heart rate, diastolic blood pressure and triglycerides, and positively associated with high-density lipoprotein cholesterol (all P ≤ 0.005). OPA adjusted for age only was positively associated with most cardiovascular risk factors but these associations were not significant after further adjustment on education (except for waist circumference in women). Age- and education-adjusted LTPA was associated with increased frequency of consumption of fruits (odds ratio (OR) = 2.05, 95% confidence interval (CI) 1.68–2.52 in men; OR = 1.90, 95% CI 1.41–2.05 in women) and vegetables (OR = 1.81, 95% CI 1.48–2.21 in men; OR = 2.22, 95% CI 1.66–2.97 in women).

The data emphasise the favourable associations of LTPA, a modifiable behaviour, with various cardiovascular risk factors and healthy eating habits. The results also suggest that the relationships of OPA with cardiovascular risk factors depend, at least in part, on socio-economic status as reflected by educational level.

The International Physical Activity Questionnaire (IPAQ) was developed to measure health-related physical activity (PA) in populations. The short version of the IPAQ has been tested extensively and is now used in many international studies. The present study aimed to explore the validity characteristics of the long-version IPAQ.

Forty-six voluntary healthy male and female subjects (age, mean±standard deviation: 40.7±10.3 years) participated in the study. PA indicators derived from the long, self-administered IPAQ were compared with data from an activity monitor and a PA log book for concurrent validity, and with aerobic fitness, body mass index (BMI) and percentage body fat for construct validity.

Strong positive relationships were observed between the activity monitor data and the IPAQ data for total PA (ρ = 0.55, P < 0.001) and vigorous PA (ρ = 0.71, P < 0.001), but a weaker relationship for moderate PA (ρ = 0.21, P = 0.051). Calculated MET-h day−1 from the PA log book was significantly correlated with MET-h day−1 from the IPAQ (ρ = 0.67, P < 0.001). A weak correlation was observed between IPAQ data for total PA and both aerobic fitness (ρ = 0.21, P = 0.051) and BMI (ρ = 0.25, P = 0.009). No significant correlation was observed between percentage body fat and IPAQ variables. Bland–Altman analysis suggested that the inability of activity monitors to detect certain types of activities might introduce a source of error in criterion validation studies.

The long, self-administered IPAQ questionnaire has acceptable validity when assessing levels and patterns of PA in healthy adults.

To determine the relative validity and reproducibility of two alcohol intake frequency questionnaires (AFQ-A; AFQ-B), designed to classify subjects according to their alcohol intake level, in Bucaramanga, Colombia.

One hundred and nine randomly selected subjects, aged between 20 and 60 years, completed three 30-day semi-quantitative alcohol intake records (30-DR). The AFQs were applied three months after the last 30-DR. AFQ-A contained 53 items; AFQ-B contained five items, with the alcoholic drinks for AFQ-B selected by Max_r. The correlation and agreement between alcohol intake assessed with the AFQs and the 30-DR were obtained using Pearson's correlation coefficient (r), Lin's concordance correlation coefficient (pC), Spearman's rank correlation coefficient(rS), Bland and Altman's limits of agreement (LOA) and Cohen's weighted kappa statistic (Kw).

The reproducibility of the 30-DR was poor; rS ranged from 0.33 to 0.41. The reproducibility of the AFQs was higher, with rS between 0.50 and 0.73. The agreement(Kw) of the 30-DR and the AFQs was 0.40. The lower and upper LOA were between 56.4% and 11.0%. The AFQs and 30-DR were well correlated. Assessment of relative validity between the two methods yielded r values for alcohol between 0.52 Alcohol consumption and 0.60, which reduced to 0.20–0.29 after energy adjustment.

These AFQs may be useful to rank subjects according to their alcohol Max_r intake. The AFQ-B is easy and quick to apply, and is also highly cost-effective.

To validate intakes of protein, folic acid, retinol and n–3 fatty acids estimated from a food-frequency questionnaire in week 25 of pregnancy (FFQ-25).

Validation was done against a 7-day weighed food diary (FD) and biomarkers of the nutrients in gestation week 32–38.

The FFQ-25 to be validated was used in the Danish National Birth Cohort comprising 101 042 pregnant Danish women, of whom 88 participated in the present validation study.

Estimated intakes of protein, retinol and folic acid did not differ significantly between the two dietary methods, but intake of n–3 fatty acids was one third larger when estimated from the FFQ-25. The intakes estimated from the two dietary methods were all significantly correlated, ranging from 0.20 for retinol intake to 0.57 for folic acid intake. Sensitivities of being correctly classified into low and high quintiles were between 0.22 and 0.77, and specificities were between 0.62 and 0.89. Urinary protein content did not correlate significantly with protein estimated from the FFQ (r = 0.17, P > 0.05), but did with intake estimated from the FD (r = 0.56, P < 0.0001). Erythrocyte folate correlated significantly with the estimated total intake from the FFQ (r = 0.55, P < 0.0001) and the FD (r = 0.52, P < 0.0001). No correlations with plasma retinol were found. Erythrocyte eicosapentaenoic acid (C20:5n-3) correlated significantly with n–3 fatty acids intake estimated from both the FFQ-25 (r = 0.37, P < 0.001) and the FD (r = 0.62, P < 0.0001).

The FFQ-25 gives reasonable valid estimates of protein, retinol and folic acid intakes, but seems to overestimate intake of n–3 fatty acids.

The study aimed to evaluate the dietary vitamin B6 intake and determine the vitamin B6 concentration in blood samples of German vegans.

Cross-sectional study with 33 examination sites all over Germany.

Ninety-three vegans (50 females) with a mean (±standard deviation (SD)) age of 43.7 ± 15.7 years who took no vitamin supplements.

Dietary intake was assed using a semi-quantitative food-frequency questionnaire. Erythrocyte aspartate aminotransferase activity coefficient (EAST-AC) was calculated as the ratio of stimulated (pyridoxal 5'-phosphate added) to unstimulated activity in blood samples that were provided after an overnight fast.

Mean ± SD vitamin B6 intake was 2.83 ± 0.98 mg day−1 and mean ± SD protein intake was 56.6 ± 21.7 g day−1. Of the participants 4% showed vitamin B6 intakes lower than daily recommended intakes for Germany, 16% showed EAST-AC > 1.85, and a further 58% showed EAST-AC of 1.5–1-85. Moderate vegans were affected to a lesser extent than strict vegans. None of the established confounders was a significant predictor of EAST-AC. In logistic regression analyses the contribution of nutriments and cereals to pyridoxine intake was the only predictor of EAST-AC classified as < 1.85 and ≥ 1.85, respectively.

In spite of the high total intake of vitamin B6, an adequate concentration in blood samples could not be realised for a majority of the participants. Due to the health implications of a marginal pyridoxine status, vegans should be encouraged to include foods with a high bioavailability of pyridoxine, such as beans, lentils and bananas, in the daily diet.

To conduct a comprehensive assessment of dietary intakes of nitrites and N-nitrosodimethylamine (NDMA).

A study was conducted within the Spanish cohort of the European Prospective Investigation in Cancer and Nutrition (EPIC) to assess the intake and food sources of these compounds in Spanish adults. The study included 41 446 health volunteers, aged 29–69 years, from Northern and Southern regions. Usual food intake was estimated by in-person interviews using a computerised dietary questionnaire.

The estimated geometric mean was 0.994 mg day−1 for nitrites and 0.114 μg day−1 for NDMA. For both compounds a positive trend in consumption with increasing energy intake was observed. Dietary NDMA was related to age and sex after energy adjustment, while nitrite consumption increased with higher intakes of vitamin C (P < 0.001). The food groups that contributed most to intakes were meat products, cereals, vegetables and fruits for nitrites, and processed meat, beer, cheese and broiled fish for NDMA. Current and past smokers, who had high levels of NDMA from tobacco exposure, were also identified as the highest consumers of dietary NDMA. Furthermore, smokers had low intakes of vitamin C (an inhibitor of endogenous nitrosation).

Intake levels of NDMA and nitrites in a Mediterranean cohort are currently relatively lower than those previously reported, although processed meat, beer and cured cheese still are the most important contributors to NDMA intake.

To examine the efficacy and safety of foods fortified with calcium in the adult population in Finland.

A simulation study based on the FINDIET 2002 Survey, which estimated habitual food consumption, dietary supplement use and nutrient intakes using 48-hour recall and two 3-day food records, and an Internet survey of the consumption of fortified foods and dietary supplements.

Participants of FINDIET 2002 were 25–64 years old from five areas (n = 2007). Participants of the Internet-based survey (n = 1537) were over 15 years of age from all over the country.

If all potentially fortifiable foods were to be fortified with calcium, the proportion of participants with calcium intake below the recommended level (<800 mg day−1) would decrease from 20.3% to 3.0% in men and from 27.8% to 5.6% in women compared with the situation where no foods were fortified. At the same time, the proportion of participants with calcium intake above the tolerable upper intake level (UL, >2500 mg day−1) would increase from 0.6% to 12.7% in men and from 0.1% to 3.8% in women. However, in a probability-based model (11% of all fortifiable foods to be fortified with calcium) the proportion of participants with calcium intake below the recommended level would be 15.7% in men and 23.2% in women. The proportion with intake above the UL in this model would be 1.2% in men and 0.7% in women.

Food fortification would be a relatively effective and safe way to increase the calcium intake of the Finnish adult population.

This paper reports on findings from the ex-post evaluation of the Maewo Capacity Building project in Maewo Island, Vanuatu, which was funded by World Vision Australia.

To examine the extent to which the infrastructure and systems left behind by the project contributed to the improvement of household food security and health and nutritional outcomes in Maewo Island, using Ambae Island as a comparator.

Two-stage cluster survey conducted from 6 to 20 July 2004, which included anthropometric measures and 4.5-year retrospective mortality data collection.

A total of 406 households in Maewo comprising 1623 people and 411 households in Ambae comprising 1799 people.

Household food insecurity, crude mortality rate (CMR), under-five mortality rate (U5MR) and malnutrition prevalence among children.

The prevalence of food insecurity without hunger was estimated at 15.3% (95% confidence interval (CI): 12.1, 19.2%) in Maewo versus 38.2% (95% CI: 33.6, 43.0%) in Ambae, while food insecurity with hunger in children did not vary by location. After controlling for the child's age and gender, children in Maewo had higher weight-for-age and height-for-age Z-scores than children of the same age in Ambae. The CMR was lower in Maewo (CMR=0.47/10 000 per day, 95% CI: 0.39, 0.55) than in Ambae (CMR=0.59/10 000 per day, 95% CI: 0.51, 0.67) but no difference existed in U5MR. The major causes of death were similar in both locations, with frequently reported causes being malaria, acute respiratory infection and diarrhoeal disease.

Project initiatives in Maewo Island have reduced the risks of mortality and malnutrition. Using a cross-sectional 'external control group' design, this paper demonstrates that it is possible to draw conclusions about project effectiveness where baseline data are incomplete or absent. Shifting from donor-driven evaluations to impact evaluations has greater learning value for the organisation, and greater value when reporting back to the beneficiaries about project impact and transformational development in their community. Public health nutritionists working in the field are well versed in the collection and interpretation of anthropometric data for evaluation of nutritional interventions such as emergency feeding programmes. These same skills can be used to conduct impact evaluations, even some time after project completion, and elucidate lessons to be learned and shared. These skills can also be applied more widely to projects which impact on the longer-term nutritional status of communities and their food security.

To assess the effectiveness of a behaviour change approach, with or without financial support, in improving vitamin A (VA) intake and serum retinol concentration through mango and liver consumption by children.

A parallel design (no control area) was used to assess changes in VA intake and serum retinol over a 15-week period.

A pilot study was implemented in the Department of Kokologho, a rural area in central west Burkina Faso. One hundred and fifty children aged 2–3 years were randomly selected and assigned to two treatment groups: PA$$ (promotional activities and financial support) and PA (promotional activities).

The intervention significantly increased (P<0.001) total VA intake by 56% in PA$$ and by 50% in PA. VA intake from liver increased significantly (P<0.001) from 12.7±23.5 to 155.3±56.3 μg retinol activity equivalents (RAE) in PA$$ and from 21.6±29.7 to 135.3±44.9 μg RAE in PA. Changes in VA intake from liver were significantly higher (P=0.004) in PA$$ compared with PA. Mean serum retinol concentration increased significantly by 26% (P<0.001) in PA$$ and 30% (P<0.001) in PA. Changes in serum retinol concentration (0.13 μmol l−1 in PA$$ vs. 0.17 μmol l−1 and in PA) did not differ significantly (P=0.455) between groups over the intervention.

Promotional activities on mango and liver intake effectively increased VA intake and serum retinol concentrations. Although an additional beneficial effect of financial support on liver intake was observed, this did not translate into a further increase in serum retinol concentration.