To evaluate the associations between sociodemographic and psychosocial characteristics and food label (FL) use in US adults.

Data from the 1994–1996 Continuing Survey of Food Intakes by Individuals and the Diet and Health Knowledge Survey were used. High socio-economic status (SES) was defined as >high school education and poverty–income ratio (PIR) >350 %, low SES as <high school level or PIR <130 %. Dietary intakes were assessed using 24 h recalls.

Metropolitan statistical area-central city, -suburban, and rural areas in the USA.

US adults (n 2797; 1460 men, 1337 women) aged 20–64 years.

Approximately 80 % of Americans reported using FL, including checking the nutrition panel, list of ingredients, short phrases, serving size, or health benefits. Only 26 % used all FL information. Compared with white women of higher SES, white men, black men and women with lower SES were 77–90 % less likely to use FL. Rural residents were 40 % less likely (OR = 0·60; 95 % CI 0·42, 0·86). Participants with good nutrition knowledge, perceptions and beliefs were twice as likely to check FL for nutrient content of foods (OR = 2·28; 95 % CI 1·53–3·40). Those who were unaware of diet–disease relationships were less likely to use FL (OR = 0·53; 95 % CI 0·32–0·85). Among overweight/obese Americans (BMI ≥ 25 kg/m2), those who perceived their weight ‘about right’ were 51 % less likely to use FL than those perceiving themselves as overweight.

Men, especially black men, women of low SES, rural residents and overweight Americans with inaccurate self-perception of body weight are less likely to use FL and should be targeted for increased intervention.

As a means of empowering consumers, nutrition labelling has become a widely discussed topic. Simplicity and uniformity of labelling systems are regarded as the prevailing demands from the consumer side. In the present study, we analyse the effects of the traffic light signposting scheme on consumers’ food choices.

In an online survey, respondents first rated the understandability of the traffic light signposting scheme. In a following conjoint experiment, they indicated which products they would select as the healthiest of the presented products, based on the nutritive information provided by the traffic light signposting scheme.

A major German university.

In total 2002 undergraduate students participated in the survey. Two-thirds (69 %) of the respondents were female and the majority of the respondents (70 %) were between 18 and 24 years old. Seventy-seven per cent of the participants indicated that they had a higher level of education.

Overall, the participants rated the understandability of the traffic light nutrition signposting scheme fairly high (5·9 out of 7). Sugar and fat were found to be the most important attributes of the scheme. Participants placed greater emphases on a change in a product's nutrient characteristic from ‘amber to ‘red’ compared with a change from ‘green’ to ‘amber’.

Our results confirm the signalling effect of colour coding as it helps reduce the complexity of decision making. Our findings shed new light on the ongoing discussion concerning appropriate and efficient nutrition labelling and provide interesting insights for further research as well as implications for public policy making.

Dietary sodium reduction is a cost-effective public health intervention to reduce chronic disease. In response to calls for further research into front-of-pack labelling systems, we examined how alternative sodium nutrition label formats and nutrition claims influenced consumers’ choice behaviour and whether consumers with or without a diagnosis of hypertension differed in their choice patterns.

An anonymous online experiment in which participants viewed ten choice sets featuring three fictitious brands of baked beans with varied label formats and nutritional profiles (high and low sodium) and indicated which brand in each set they would purchase if shopping for this product.

Participants were recruited from New Zealand's largest online nationwide research panel.

Five hundred people with self-reported hypertension and 191 people without hypertension aged 18 to 79 years.

The addition of a front-of-pack label increased both groups’ ability to discriminate between products with high and low sodium, while the Traffic Light label enabled better identification of the high-sodium product. Both front-of-pack formats enhanced discrimination in the presence of a reduced salt claim, but the Traffic Light label also performed better than the Percentage Daily Intake label in moderating the effect of the claim for the high-sodium product.

Front-of-pack labels, particularly those with simple visual cues, enhance consumers’ ability to discriminate between high- and low-sodium products, even when those products feature nutrition claims.

A few studies link out-of-home eating to higher energy consumption, overweight and obesity in both adults and children. The present study was undertaken to investigate the nutritional value of meals available in a university cafeteria, in order to develop a target nutritional tool to help consumers make a more conscious nutritional choice.

A cross-sectional study.

In a university cafeteria in Pavia, northern Italy, the recipes and ingredients of each meal served during the whole year were obtained from the cooks. Energy, protein, fat, carbohydrate and fibre contents were computed for each meal standardized portion. Thirteen pyramid figures, subdivided into three coloured levels, were used to depict the energy and nutrient content of each meal.

Four hundred randomly selected customers were interviewed on the cafeteria nutritional proposal.

Foods available in the cafeteria consisted of 216 items and were distributed in the pyramids according to their energy content: the lowest ones at the bottom (green level) and the highest ones at the top (red level), passing through an orange level in the middle. Energy values ranged from 460 kJ (110 kcal) for a portion of dressed vegetables to 5021 kJ (1200 kcal) for a pizza. The depicted pyramids were displayed in the cafeteria, so that customers could choose their meal according to its nutritional value. The meals’ nutritional content information was perceived very helpful for customers’ nutritional choices.

Availability of nutrition information in the cafeteria was well accepted by the customers who could plan their meals according to a more balanced diet.

To investigate the purchase and use of fortified foods, and to explore and compare background characteristics, food consumption and nutrient intakes among users and non-users of voluntarily fortified foods in Finland.

A study based on the National FINDIET Survey 2007 (48 h recall), which included also a barcode-based product diary developed to assess the type, amount and users of voluntarily fortified foods. Logistic regression analysis was employed to investigate associations between background characteristics and the use of fortified foods.

Randomly chosen subgroup of 918 adult participants in the National FINDIET 2007 Survey.

Men and women aged 25–64 years from five regions.

The product group of voluntarily fortified foods purchased in the highest volume was yoghurts (44 % of the weight of all fortified food), followed by fruit drinks (36 %). The only characteristics independently associated with the use of voluntarily fortified foods were age (older people used them less commonly) and the consumption of fruit and vegetables (participants with the highest consumption used them more commonly). Users of fortified foods had higher consumption of yoghurt, juice drinks and ready-to-eat breakfast cereals (women only) than non-users, and lower consumption of boiled potatoes (men only).

Use of voluntarily fortified foods is associated with high consumption of fruit and vegetables but not with other health-related behaviours. The use of voluntarily fortified foods does not seem to even out the differences in nutrient intake among Finnish adults.

To estimate the effect of the US Supplemental Nutrition Assistance Program (SNAP) on the food security (consistent access to adequate food) of recipients, net of the effect of the self-selection of more food-needy households into the programme.

The food security of current SNAP recipients and recent leavers is compared in cross-sectional survey data, adjusting for economic and demographic differences using multivariate logistic regression methods. A similar analysis in 2-year longitudinal panels provides additional control for selection on unobserved variables based on food security status in the previous year.

Household survey data collected for the US Department of Agriculture by the US Census Bureau.

Households interviewed in the Current Population Survey Food Security Supplements from 2001 to 2009.

The odds of very low food security among households that continued on SNAP through the end of a survey year were 28 % lower than among those that left SNAP prior to the 30-d period during which food security was assessed. In 2-year panels with controls for the severity of food insecurity in the previous year, the difference in odds was 45 %.

The results are consistent with, or somewhat higher than, the estimates from the strongest previous research designs and suggest that the ameliorative effect of SNAP on very low food security is in the range of 20–50 %.

To explore associations between household food security and home gardening, use of soya and pressure cooker ownership in low-income households affected by HIV/AIDS in Aurangabad, India.

Cross-sectional pilot study which assessed household food security using the validated US Department of Agriculture's food security core-module questionnaire. Questions were added to explore household environment, education, occupation, home gardening, use of soya and pressure cooker ownership. Households with very low v. low food security were compared using logistic regression analysis, controlling for confounding by socio-economic status.

Aurangabad is an urban setting situated in a primarily agricultural dependent area. The study was carried out in 2008, at the peak of the global food crisis.

Adult caregivers of children affiliated with the Network of People Living with HIV/AIDS in Aurangabad.

All except for one of 133 households were identified as food insecure (99·2 %). Of these households, 35·6 % had to cut size or skip a meal in the past 30 d. Households that cut meal size due to cooking fuel shortages were more likely to have very low food security (OR = 4·67; 95 % CI 1·62, 13·44) compared with households having no cooking fuel shortages. Owning a pressure cooker was shown to be protective against very low food security after controlling for confounding by socio-economic status (OR = 0·27; 95 % CI 0·11, 0·64).

Only pressure cooker ownership showed a protective association with low household food security. Pressure cookers save household fuel costs. Therefore, future interventions should explore pressure cookers as a sustainable means of improving household food security.

The present study aimed to assess complementary feeding practices and identify the potential risk factors associated with inappropriate complementary feeding in Indonesia for a nationally representative sample of births from 2004 to 2007.

The data source for the analysis was the 2007 Indonesia Demographic and Health Survey. Multiple logistic regression was performed to analyse the factors associated with complementary feeding, using individual-, household- and community-level determinants.

Indonesia.

Children (n 4604) aged 6–23 months.

Multivariate analysis revealed that infants from poor households were significantly less likely to be introduced to complementary feeding (adjusted odds ratio, AOR = 4·32; 95 % CI 1·46, 12·80) and meet the minimum dietary diversity (AOR = 1·76; 95 % CI 1·16, 2·68). Mother's education (AOR for no education in dietary diversity = 1·92; 95 % CI 1·09, 3·38; AOR for no education in meal frequency = 2·03; 95 % CI 1·13, 3·64; AOR for no education in acceptable diet = 3·84; 95 % CI 2·07, 7·12), residence and decreased age of the infant were negatively associated with minimum dietary diversity, minimum meal frequency and an acceptable diet. Infants aged 6–11 months were also significantly less likely to meet minimum dietary diversity (AOR = 6·36; 95 % CI 4·73, 8·56), minimum meal frequency (AOR = 2·30; 95 % CI 1·79, 2·96) and minimum acceptable diet (AOR = 2·27; 95 % CI 1·67, 3·09). All geographical regions compared with Sumatra were more likely to give the recommended meal frequency and an acceptable diet to breast-fed children.

Public health interventions to improve complementary feeding should address individual-, household- and community-level factors which significantly influence the introduction of complementary feeding. Complementary feeding intervention programmes in Indonesia should ensure that restraints on families with low socio-economic status are addressed. Infants aged 6–11 months and mothers with low education levels may also need special focus. Promotion strategies should also target the health-care delivery system and the media.

To determine whether pre-pregnancy BMI influences breast-feeding practice.

Retrospective epidemiological study.

University Hospital Leuven, Catholic University Leuven, Belgium.

Two hundred women (median age 29 years, interquartile range (IQR) 4; 52 % nulliparae) were grouped into four categories according to pre-pregnancy BMI using WHO cut-offs.

The incidence of intention and initiation of breast-feeding was significantly lower in underweight (64 %) and obese women (68 %) compared with normal weight (92 %) and overweight women (80 %). Initiation was also related to parity (OR = 0·582; 95 % CI 0·400, 0·846), but not to gestational weight gain, method of delivery or hypertensive disorders. Fifty-two per cent of underweight, 70 % of normal weight and 56 % of overweight women were exclusively breast-feeding their infant during the first month of life. This incidence was significantly lower in the obese group (34 %; P = 0·030). Only 40 % of all infants were exclusively breast-fed at 3 months of age, with the lowest prevalence among women with obesity (P = 0·0 0 1). The median duration of any breast-feeding in the obese group (1·8 months, IQR 3·4) was significantly shorter than in the underweight (3·0 months, IQR 3·1), normal weight (3·0 months, IQR 2·4) and overweight group (3·0 months, IQR 3·5; P = 0·024). Reasons given for ceasing breast-feeding in the obese group were maternal complications (29 %), insufficient milk supply (23 %), sucking problems (21 %) and work resumption (21 %).

Breast-feeding practice in the total population, but especially among women with obesity, fell short of global WHO recommended standards. Policy initiatives and local interventions should continue to support breast-feeding, but also prevent maternal obesity.

The purpose of the present research was to investigate the epidemiological profile of the metabolic syndrome (MetS) and to explore its potential dietary, behavioural and socio-economic determinants among European adults residing in Luxembourg.

Cross-sectional, population-based ORISCAV-LUX survey.

European adults aged 18–69 years residing in Luxembourg.

A total of 1349 Europid adults, who participated in the ORISCAV-LUX survey, were included in the study. The prevalence of MetS was estimated according to the Revised-Adult Treatment Panel (R-ATPIII) criteria. Multivariate logistic regression was used to identify the dietary, behavioural and socio-economic factors independently associated with MetS.

The overall prevalence of MetS was 24·7 % with significant gender difference (18·5 % for women v. 30·8 % for men, P < 0·0001). Age, male gender, primary level of education, physical inactivity, family history of diabetes and hypertension and inadequate protein intake were identified as significant determinants of MetS, after adjusting for other socio-economic, family medical history and lifestyle factors.

MetS is a common condition among Europid adults in Luxembourg and increases dramatically with age, in both genders. Several dietary, socio-economic and behavioural factors explain the disparity observed. These findings highlight the importance of a comprehensive approach to MetS encompassing dietary, lifestyle and socio-economic aspects, both in clinical and community settings.

To investigate the impact of eating behaviour traits on central obesity, prediabetes and associated major dietary food patterns.

Assessment of eating behaviour was based on the revised German version of the Three-Eating Factor Questionnaire using cross-sectional and longitudinal data of a feasibility study in employees. Data on lifestyle and nutrition were obtained by validated self-administered questionnaires. Baseline characteristics were analysed by the univariate χ2 test or the Mann–Whitney test. To quantify correlations linear regression analysis was used.

The Delay of Impaired Glucose Tolerance by a Healthy Lifestyle Trial (DELIGHT), which investigated measures to prevent type 2 diabetes mellitus in 2004–2008.

Employees (21–64 years, 127 men, 157 women) with elevated waist circumference (men ≥94 cm, women ≥80 cm) of five medium-sized companies in northern Germany.

At baseline (T0), BMI but particularly waist circumference showed a strong inverse correlation with flexible control (P < 0·0001) and a positive correlation with disinhibition (P < 0·0001) and rigid control (P = 0·063). Flexible control was also significantly inversely related to fasting plasma glucose (P = 0·040), energy intake (P < 0·0001), intake of meat and meat products (P = 0·0001), and positively associated with intake of fruit and vegetables (P < 0·0001) at baseline (T0). Changes in flexible control within the first year of intervention (T1 v. T0) predicted changes in central obesity (P < 0·0001) and fasting plasma glucose (P = 0·025).

DELIGHT shows that flexible control characterizes individuals with a higher dietary quality, a lower waist circumference and a lower glucose level. Enhancing flexible control more than rigid control, and decreasing disinhibition, seems beneficial in terms of central adiposity and glucose levels.

To assess the association between dietary energy density and the prevalence of cardiometabolic risk factors among Iranian adult women.

Cross-sectional study.

Tehran, Islamic Republic of Iran.

We assessed habitual dietary intakes of 486 Iranian adult women by the use of a validated FFQ. Dietary energy density (DED) was calculated as each individual's reported daily energy intake (kJ/d, kcal/d) divided by the total weight of foods (excluding beverages) consumed (g/d). Fasting plasma glucose (FPG), lipid profiles and blood pressure were measured. Diabetes (FPG ≥ 126 mg/dl), dyslipidaemia (based on the National Cholesterol Education Program Adult Treatment Panel III) and hypertension (based on the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure) were determined. The presence of ‘at least one risk factor’ and ‘at least two risk factors’ of the three major risk factors for CVD (hypertension, dyslipidaemia and diabetes) was also evaluated. To explore the associations between DED and cardiometabolic risk factors, we obtained prevalence ratios in different models accounting for confounders.

Mean DED was 7·41 (sd 1·46) kJ/g (1·77 (sd 0·35) kcal/g). Consumption of energy-dense diets was associated with higher intakes of energy, dietary fat, cholesterol, vegetable oils, refined grains and high-fat dairy products, along with lower intakes of dietary carbohydrates, fruits, vegetables, and meat and fish. Adherence to an energy-dense diet was associated with elevated levels of serum TAG, total cholesterol and LDL cholesterol and lower levels of serum HDL cholesterol. Women in the top quartile of DED were more likely to have dyslipidaemia (61 % v. 31 %, P < 0·05), at least one (68 % v. 35 %, P < 0·05) and at least two (29 % v. 10 %, P < 0·05) cardiometabolic risk factors compared with those in the bottom quartile. Consumption of energy-dense diets was significantly associated with a greater chance of having dyslipidaemia (prevalence ratio in top quartile v. bottom quartile: 1·78; 95 % CI 1·33, 2·58), at least one (1·81; 1·44, 2·49) and at least two cardiometabolic risk factors (2·81; 1·51, 5·24). Additional control for BMI and total energy intake slightly attenuated the associations. No overall significant associations were found between consumption of energy-dense diets and risk of having diabetes or hypertension either before or after adjustment for confounders.

Consumption of energy-dense diets was significantly related to the high prevalence of dyslipidaemia, at least one and at least two cardiometabolic risk factors among Iranian adult women. Prospective studies are required to confirm our findings.

The present study aimed to characterize the dietary patterns of newlyweds married within one year and to examine the factors associated with weight changes.

A cross-sectional study was carried out in 2009.

Baoshan District, Shanghai, China.

A total of 342 pairs of newlyweds were recruited at their first visit. Weight and height were measured and food intake was measured based on an FFQ with forty-nine items. Dietary patterns were identified through factor analysis, and the factors associated with weight changes were determined by multiple regression models.

Average weight gain of the participants during marital transition was 1·9 kg (2·2 kg for men and 1·6 kg for women). Four major dietary patterns were identified: vegetable pattern, sweets and fats pattern, legume pattern and poultry, beef and mutton pattern. In multivariate analyses, the highest quintile of factor 2 (sweets and fats pattern) was significantly associated with weight gain after marriage in a comparison with the lowest quintile (β = 2·94; 95 % CI 0·75, 5·15; P = 0·01) in men. The highest quintile of factor 4 (poultry, beef and mutton pattern) was inversely associated with weight gain after marriage in a comparison with the lowest quintile (β = −1·21; 95 % CI −2·32, −0·11; P = 0·03).

Our results suggest that a pattern rich in sugar, oil and condiments may lead to greater marital weight gain in men; another pattern rich in poultry, beef and mutton may lead to less marital weight gain in both men and women.

The present study aimed to evaluate the quality of Internet information on the Mediterranean diet and to determine the relationship between the quality of information and the website source.

Website sources were categorized as institutional, pharmaceutical, non-pharmaceutical commercial, charitable, support and alternative medicine. Content quality was evaluated using the DISCERN rating instrument, the Health On the Net Foundation's (HON) code principles, and Journal of the American Medical Association (JAMA) benchmarks. Readability was graded by the Flesch Reading Ease score and Flesch–Kincaid Grade Level score.

The phrase ‘Mediterranean diet’ was entered as a search term into the six most commonly used English-language search engines.

The first thirty websites forthcoming by each engine were examined.

Of the 180 websites identified, thirty-two met our inclusion criteria. Distribution of the website sources was: institutional, n 8 (25 %); non-pharmaceutical commercial, n 12 (38 %); and support, n 12 (38 %). As evaluated by the DISCERN, thirty-one of the thirty-two websites were rated as fair to very poor. Non-pharmaceutical commercial sites scored significantly lower than institutional and support sites (P = 0·002). The mean Flesch Reading Ease score and mean Flesch–Kincaid Grade Level were 55·9 (fairly difficult) and 7·2, respectively. The Flesch–Kincaid Grade Level score determines the difficulty of material by measuring the length of words and sentences and converting the results into a grade level ranging from 0 to 12 (US grade level).

Due to the poor quality of website information on the Mediterranean diet, patients or consumers who are interested in the Mediterranean diet should get advice from physicians or dietitians.

The WHO/UNICEF Baby-Friendly Hospital Initiative (BFHI) is an effective strategy to increase breast-feeding exclusivity and duration but many countries have been slow to implement it. The present paper describes the development of a computer-based instrument that measures policies and practices outlined in the BFHI.

The tool uses clinical staff/managers’ and pregnant women/mothers’ opinions as well as maternity unit observations to assess compliance with the BFHI's Ten Steps to Successful Breastfeeding (Ten Steps) and the International Code of Marketing of Breastmilk Substitutes (Code) by measuring the extent of implementation of two to fourteen indicators for each step and the Code. Composite scores are used to summarize results.

Examples of results from a 2007 assessment performed in nine hospitals in the province of Québec are presented to illustrate the type of information returned to individual hospitals and health authorities.

Participants included nine to fifteen staff/managers per hospital randomly selected among those present during the interviewer-observer's 12 h hospital visit and nine to forty-five breast-feeding mothers per hospital telephoned at home after being randomly selected from birth certificates.

The Ten Steps Global Compliance Score for the nine hospitals varied between 2·87 and 6·51 (range 0–10, mean 5·06) whereas the Code Global Compliance Score varied between 0·58 and 1 (range 0–1, mean 0·83). Instrument development, examples of assessment results and potential applications are discussed.

A methodology to measure BFHI compliance may help support the implementation of this effective intervention and contribute to improved maternal and child health.

To signal key issues for harmonising approaches for establishing micronutrient recommendations by explaining observed variation in recommended intakes of folate, vitamin B12, Fe and Zn for adults and elderly people.

We explored differences in recommended intakes of folate, vitamin B12, Fe and Zn for adults between nine reports on micronutrient recommendations. Approaches used for setting recommendations were compared as well as eminence-based decisions regarding the selection of health indicators indicating adequacy of intakes and the consulted evidence base.

In nearly all reports, recommendations were based on the average nutrient requirement. Variation in recommended folate intakes (200–400 μg/d) was related to differences in the consulted evidence base, whereas variation in vitamin B12 recommendations (1·4–3·0 μg/d) was due to the selection of different CV (10–20 %) and health indicators (maintenance of haematological status or basal losses). Variation in recommended Fe intakes (men 8–10 mg/d, premenopausal women 14·8–19·6 mg/d, postmenopausal women 7·5–10·0 mg/d) was explained by different assumed reference weights and bioavailability factors (10–18 %). Variation in Zn recommendations (men 7–14 mg/d, women 4·9–9·0 mg/d) was also explained by different bioavailability factors (24–48 %) as well as differences in the consulted evidence base.

For the harmonisation of approaches for setting recommended intakes of folate, vitamin B12, Fe and Zn across European countries, standardised methods are needed to (i) select health indicators and define adequate biomarker concentrations, (ii) make assumptions about inter-individual variation in requirements, (iii) derive bioavailability factors and (iv) collate, select, interpret and integrate evidence on requirements.

To review the prevalence, severity and determinants of anaemia among women in West and Central Africa (WCA) and raise awareness among policy makers and programme planners in the region.

Systematic descriptive review of data in the public domain of the ORC Macro MEASURE Demographic and Health Surveys, national nutrition surveys, oral and technical communications at regional meetings, studies published in scientific journals, and WHO and UNICEF databases.

West and Central Africa region.

Women of childbearing age.

The prevalence of anaemia among pregnant and non-pregnant women is higher than 50 % and 40 %, respectively, in all countries. Within countries, this prevalence varies by living setting (rural v. urban), women's age and education. Across countries, socio-economic and climatic differences have no apparent association with the prevalence of anaemia among women. Several factors contribute either alone or jointly to the high rates of maternal anaemia in this region. These include widespread nutritional deficiencies; high incidence of infectious diseases; low access to and poor quality of health services; low literacy rates; ineffective design, implementation and evaluation of anaemia control programmes; and poverty.

Addressing the multiple causes and minimizing the consequences of anaemia on maternal and child health and development in WCA require integrated multifactorial and multisectoral strategies. This also calls for unprecedented, historical and stronger political will and commitment that put adolescent girls and maternal health at the centre of the development agenda.

Anaemia is common during pregnancy, and prenatal Fe supplementation is the standard of care. However, the persistence of anaemia despite Fe supplementation, particularly in HIV infection, suggests that its aetiology may be more complex and warrants further investigation. The present study was conducted to examine predictors of incident haematological outcomes in HIV-infected pregnant women in Tanzania.

Prospective cohort study. Cox proportional hazards and binomial regression models were used to identify predictors of incident haematological outcomes: anaemia (Hb < 110 g/l), severe anaemia (Hb < 85 g/l) and hypochromic microcytosis, during the follow-up period.

Antenatal clinics in Dar es Salaam, Tanzania.

Participants were 904 HIV-infected pregnant women enrolled in a randomized trial of vitamins (1995–1997).

Malaria, pathogenic protozoan and hookworm infections at baseline were associated with a two-fold increase in the risk of anaemia and hypochromic microcytosis during follow-up. Higher baseline erythrocyte sedimentation rate and CD8 T-cell concentrations, and lower Hb concentrations and CD4 T-cell counts, were independent predictors of incident anaemia and Fe deficiency. Low baseline vitamin D (<32 ng/ml) concentrations predicted a 1·4 and 2·3 times greater risk of severe anaemia and hypochromic microcytosis, respectively, during the follow-up period.

Parasitic infections, vitamin D insufficiency, low CD4 T-cell count and high erythrocyte sedimentation rate were the main predictors of anaemia and Fe deficiency in pregnancy and the postpartum period in this population. A comprehensive approach to prevent and manage anaemia, including micronutrient supplementation and infectious disease control, is warranted in HIV-infected women in resource-limited settings – particularly during the pre- and postpartum periods.