Current and potential contributions of large-scale food fortification to meeting micronutrient requirements in Senegal: a modelling study using household food consumption data

Baseline prevalence of micronutrient inadequacy

We estimated the baseline prevalence (ie, without the contributions of LSFF or other micronutrient interventions) of apparent micronutrient inadequacies using household-level food consumption data from the 2018–2019 Enquête Harmonisée sur les Conditions de Vie des Ménages (EHCVM).17 The 2018–2019 EHCVM, which was implemented by the National Agency of Statistics and Demography in Senegal, is a household consumption and expenditure survey that collected data on household living conditions from 7156 households and is representative at the national level, by region, and by urban and rural residence. We requested and received approval from ANSD to use the 2018–2019 EHCVM data, which we used in compliance with ANSD’s data access policy. The dataset is now publicly available from the World Bank Microdata Library.18 All estimates account for survey weights. Note that we refer to ‘apparent’ food consumption and ‘apparent’ micronutrient adequacy as a reminder that these estimates are based on household-level data, which require imposing a number of assumptions (described below) to generate individual-level estimates.

The survey included a food consumption module in which the household respondent was asked to recall the quantity of each of 140 prespecified food items (see online supplemental table S1 for the list of food items) consumed by household members during the 7 days preceding the survey (food consumption module data were available for 7129 households). Using conversion factors provided with the survey data, we converted all reported quantities of food consumed to grams and calculated the average daily apparent quantities consumed. We adjusted for extreme outliers in reported quantities of food consumed by identifying the food-specific and region-specific 95th percentile of daily apparent household consumption per adult male equivalent (AME) (ie, we normalised the total quantity consumed across households by dividing that quantity by the number of AMEs in the households) and replacing quantities above the 95th percentile with the value at the 95th percentile.

Next, we estimated the energy and micronutrient contents of each food in the food list by matching it with a food composition table (FCT) entry. Matches were made based on input from in-country collaborators. Aggregate or generic food items were matched with several FCT entries to estimate the nutrient contents based on an average or weighted average (eg, the food item ‘sweet potato’ was matched with white-flesh sweet potato (assigned 80% wt) and orange-flesh sweet potato (assigned 20% wt)). Almost all matches were made with food items in the 2019 West African FCT,19 supplemented with entries from the Nutrition Coordinating Center Nutrient Database for Standard Reference20 and the Malawian FCT21 for several foods that did not have appropriate matches in the West African FCT. Where applicable, we adjusted the total quantity of food apparently consumed by the household for the edible portion and yield factor from cooking. We did not adjust for food waste.

We assessed the micronutrient adequacy of household diets for meeting the needs of WRA and children 6–59 months of age based on the nutrient density of the household diet. For each micronutrient (vitamin A, thiamin, riboflavin, niacin, folate, vitamin B₁₂, iron and zinc), we calculated the nutrient density of the household diet as the ratio of total daily apparent intake of the micronutrient to total daily energy intake, expressed per 1000 kcal by multiplying that ratio by 1000. Except for iron, we assessed the adequacy of the household diet for meeting the needs of WRA and children using the cut-point method. Specifically, we compared the nutrient density of the household diet to the critical nutrient density of a randomly selected WRA and/or child aged 6–59 months in the household, where critical nutrient densities were calculated as the age-specific and sex-specific estimated average requirement (EAR) divided by the age-specific and sex-specific energy requirements of the WRA or child, expressed per 1000 kcal. Because the iron requirements of children and non-pregnant WRA are not normally distributed, we used the full probability method, based on the iron density of diets, to asses iron adequacy.3 In short, estimating iron adequacy based on the full probability method accounts for skewed iron requirements by comparing iron intakes (or, in this case, densities) to a distribution of requirements to calculate the probability of inadequacy.

EARs for all micronutrients except zinc were from the US Institute of Medicine22 (online supplemental table S2). Based on dietary patterns in Senegal, we assumed a 10% bioavailability of iron.3 Zinc adequacy was assessed relative to physiological zinc requirements for WRA23 and children.24 Using published algorithms for children25 and adults26 to estimate absorbable zinc, for each target household member, we calculated fractional zinc absorption as the ratio of absorbed zinc to total zinc. Then, we estimated dietary zinc requirements (and critical nutrient densities) by adjusting physiological zinc requirements by the estimated person-specific fractional zinc absorption. Note that we were not able to assess iodine adequacy because the West African FCT does not include estimates of the iodine content of foods.

Pregnancy status was not collected in the EHCVM survey. Therefore, we assessed micronutrient adequacy among pregnant and non-pregnant WRA by first assuming all WRA in the sample were pregnant and estimating the prevalence of inadequacy and then assuming all WRA in the sample were not pregnant and estimating the prevalence of inadequacy. These separate estimates were then combined into an overall estimate of micronutrient adequacy among pregnant and non-pregnant WRA as a weighted average of the two estimates, where weights were based on 2019 Demographic and Health Survey (DHS) estimates of the proportion of WRA currently pregnant and not pregnant.27

Consumption of fortifiable foods

We estimated the apparent consumption of select fortifiable foods (refined oil, wheat flour and imported rice) for WRA and children using the AME method (as explained below, only imported rice, which was listed separately from local rice in the EHCVM food list, was considered fortifiable in this analysis). The AME method is based on the assumption that the intrahousehold distribution of food is in proportion to each member’s age-specific and sex-specific energy requirements.28 Based on age and sex information collected in the household roster, we assigned an AME weight to each household member, calculated as the ratio of each household member’s assumed energy requirement to the energy requirement of a male aged 18–30 years (online supplemental file 1).29 Then, each household member was assigned an AME ratio, calculated as his/her AME weight divided by the total number of AMEs in the household. Apparent consumption of each fortifiable food vehicle (including consumption of processed products containing the food vehicle using fortifiable food equivalents) was then calculated as total household apparent consumption of food vehicle multiplied by the AME ratio of the selected WRA and/or child in the household.

Modelled contribution of LSFF

To assess the contribution of LSFF to meeting the micronutrient requirements of WRA and children, we modelled a set of scenarios based on Senegal’s existing refined oil (vitamin A, 17.5 mg/kg) and wheat flour (iron, 60 mg/kg and folic acid, 2.5 mg/kg) fortification standards (online supplemental table S3). We also modelled hypothetical scenarios to assess the potential contribution of (1) revising the mandated folic acid fortification level from 2.5 mg/kg to 5 mg/kg to align with current WHO guidelines16 and (2) adding thiamin (3 mg/kg), riboflavin (2 mg/kg), vitamin B₁₂ (0.02 mg/kg) and zinc 55 (mg/kg) to the wheat flour fortification standards at fortification levels in alignment with current WHO guidance. Finally, we modelled the hypothetical establishment of a rice fortification standard based on the current World Food Programme’s technical specifications for rice fortification30 (online supplemental table S3). Note that we converted folic acid from fortification to dietary folate equivalents to account for the higher bioavailability of folic acid compared with dietary folate.31

Informed by recent qualitative and quantitative testing of market samples of refined oil and wheat flour in Senegal, we modelled refined oil fortification under a status quo current compliance scenario (59% of fortifiable refined oil fortified to an average of 82% of the national standard32) and an improved compliance scenario assuming that 75% of refined oil was fortified to 100% of the standard (online supplemental table S4). We also modelled a status quo current compliance scenario for wheat flour (76% of wheat flour fortified to an average of 43% of the national standard32) as well as an improved compliance scenario in which we assumed 76% of wheat flour was fortified to 100% of the standard. In line with the improved compliance scenarios, we modelled the potential contribution of rice fortification assuming 75% of fortifiable rice was fortified to the specified standard. Note that, currently, ~60% of Senegal’s rice supply is imported and could potentially be fortified prior to importation.13 In our modelling, we assumed all imported rice was fortifiable. A negligible share of the rice supply (~1.5%) is domestically refined at industrial scale (and hence may be feasible for fortification). We did not account for domestically milled rice in the modelling. Note that all modelling assumptions we applied at the national level.

For each scenario, we increased the nutrient content of the food vehicle to account for additions via LSFF, recalculated the nutrient density of the household diet and then compared the recalculated nutrient density of the household diet to critical nutrient densities for WRA and children. Where relevant, assumed fortification levels at the point of consumption were based on adjustments for expected micronutrient losses between the point of fortification and the point of consumption (online supplemental table S5).

For each scenario and micronutrient for which a tolerable upper intake level (UL) for safe consumption exists (vitamin A in the form of preformed retinol, folic acid (excluding dietary folate), iron and zinc), we also estimated the risk of high intake by comparing the nutrient density of the household diet to the critical upper density for the micronutrient. UL values were taken from the US Institute of Medicine22 33 (online supplemental table S6).

Accounting for VAS

Senegal implements a national VAS programme in which high-dose vitamin A supplements are provided every 6 months to children aged 6–59 months. Given the importance of balancing micronutrient contributions from fortification and supplementation where the risk of high intake may be a concern, we also modelled the contribution of VAS to meeting micronutrient requirements as well as the proportion of children who receive VAS (or do not receive VAS) and (1) are at risk of high retinol intake with LSFF, (2) have adequate (but not high) vitamin A intake with LSFF and (3) have inadequate vitamin A intake with LSFF. Receipt of VAS was based on 2019 DHS coverage estimates and linked to the ECHVM data (see online supplemental methods).27 Following Engle-Stone et al,34 receipt of VAS was converted to a daily equivalent intake of 167 µg RAE/day. This daily equivalent value was incorporated into the vitamin A density of the diet and compared with the vitamin A requirements of children aged 6–59 months.

Patient and public involvement

Because these were analyses of secondary data, patients and public were not involved in the design, conduct or reporting of this research. Results were disseminated at a validation workshop in Senegal.

Prevalence of baseline micronutrient inadequacies

Based on the nutrient density of household diets, the national baseline (ie, without accounting for the contributions of LSFF or other micronutrient interventions) prevalence of apparent micronutrient inadequacy among WRA was above 60% for all micronutrients analysed except for niacin and vitamin B₁₂ and above 70% for all micronutrients except niacin, vitamin B₁₂ and iron among children (table 1). There was notable subnational variation in the baseline prevalence of inadequacy by region. Among WRA, there were particularly large discrepancies in the prevalence of apparent vitamin A inadequacy (ranging from 52% in the Ziguinchor Region to 91% in the Saint-Louis and Louga regions), vitamin B₁₂ (14% in the Diourbel Region to 80% in the Kedougou Region) and zinc (44% in the Ziguinchor Region to 82% in the Saint-Louis Region). The largest regional discrepancies in the prevalence of apparent inadequacy among children were for vitamin A (ranging from 5% in the Ziguinchor Region to 90% in the Saint-Louis Region), folate (40% in the Kaffrine Region to 95% in the Saint-Louis Region) and vitamin B₁₂ (9% in the Diourbel Region to 76% in the Kedougou Region). Given the very low prevalence of apparent niacin inadequacy nationally and subnationally among both WRA and children, we did not model the potential addition of niacin to wheat flour.

Apparent consumption of refined oil, wheat flour and imported rice

Most households in Senegal reported consuming purchased refined oil or products containing refined oil (90% of households) and wheat flour or products containing wheat flour (89% of households) in any amount in the 7 days preceding the survey (table 2). Across regions and households residing in urban and rural areas, household consumption of these food vehicles was reported by at least 70% of households. Consumption of imported rice in the 7 days preceding the survey was reported by 52% of households nationally, with similar proportions of urban (55%) and rural (49%) of households reporting imported rice consumption. Across regions, imported rice consumption ranged from 12% in the Saint-Louis Region to 78% in the Fatick Region.

Based on the subsamples of households that reported consuming these food vehicles in any amount, median apparent consumption of refined oil was 45 g/day among WRA and 21 g/day among children. Median apparent consumption of wheat flour among consumers was 55 g/day among WRA and 26 g/day among children, while median apparent imported rice consumption among consumers was 200 and 99 g per day among WRA and children, respectively. Subnationally, while there was little variation in quantities of each food vehicle apparently consumed between urban and rural consumers, median daily apparent consumption of refined oil was over four times higher in some regions than others (with the highest and lowest median apparent consumption in the Saint-Louis and Kolda regions, respectively), and median daily apparent consumption of wheat flour was over twice as high in the Matam Region as the Kolda Region. Daily average apparent consumption of imported rice was more similar across regions than refined oil or wheat flour.

Modelled contribution of LSFF to meeting micronutrient requirements and risk of high intake

The predicted contribution of existing and hypothetical LSFF programmes to the adequacy of household diets for meeting the micronutrient requirements of WRA and children aged 6–59 months is shown in figure 1. All point estimates at the national level, by urban and rural residence, and by region are available in online supplemental table S7–S19).

Request permission

Vitamin A-fortified refined oil at current levels of estimated compliance with the national standard was predicted to reduce the prevalence of apparent inadequacy among both WRA and child by >30 percentage points (from 70% to 35% among WRA and from 70% to 37% among children), although inadequacy in rural areas remained above 50% among both population groups (figure 1; online supplemental table S7). If compliance with the refined oil fortification standard improved, the prevalence of vitamin A inadequacy was predicted to further decline to 24% among WRA and to 26% among children. Adding fortified rice to the refined oil fortification programme at current compliance was predicted to reduce the national prevalence of vitamin A inadequacy to 24% among both WRA and children. With the combination of fortified refined oil at improved compliance and rice fortification, the national prevalence of vitamin A inadequacy was predicted to decline to 16% and 17% among WRA and children, respectively, but even with this combination, the prevalence of inadequacy was >25% in rural areas. Across the regions of Senegal, the predicted prevalence of vitamin A inadequacy among WRA declined to between 14% in the Dakar Region and 64% in the Tambacounda Region with refined oil at current compliance, to between 8% (Dakar Region) and 52% (Tambacounda Region) with improved compliance with the refined oil standard (figure 2). Across regions, the addition of vitamin A-fortified rice to refined oil fortification with improved compliance was predicted to reduce vitamin A inadequacy among WRA to between 5% and 39%.

Request permission

Across all modelled scenarios, the predicted prevalence of high vitamin A intake was 1% or lower among WRA (online supplemental table S8). However, while the national predicted risk of high vitamin A intake among children was 2% or lower for all modelled scenarios, risk of high intake exceeded 5% in the Dakar Region in all scenarios with refined oil fortification and reached 9% with the combination of refined oil at improved compliance plus rice fortification (the risk of high intake among urban children was also 5% in this scenario). Across all modelled LSFF scenarios, 1% of children nationally were predicted to be both at risk of high vitamin A intake with LSFF and also received VAS, and a similar proportion was at risk of high vitamin A intake without receiving VAS (online supplemental table S10). A predicted 23%–35% of children with adequate (but not high) intake with LSFF also received VAS (22%–45% with adequate intake with LSFF did not receive VAS). Finally, 5%–18% of children were predicted to have inadequate intake with LSFF and also received VAS (12%–36% had inadequate intake and did not receive VAS). In the Dakar Region, between 2% and 4% of children across LSFF scenarios were predicted to be both at risk of high vitamin A intake and also received VAS.

If Senegal added thiamin to the national wheat flour fortification standard at the WHO recommended level (3 mg/kg), the national prevalence of apparent inadequacy was predicted to drop less than 5 percentage points among both WRA with thiamin-fortified wheat flour at current compliance. With improved compliance, the national prevalence of thiamin inadequacy was predicted to decline by 8–11 percentage points, although given high baseline prevalence of inadequacy, the prevalence of inadequacy, nationally and in both urban and rural areas, remained above 80% among WRA and above 70% among children (figure 1; online supplemental table S11). Thiamin-fortified rice at 9.7 mg/kg (the WFP specification) was predicted to reduce the national prevalence of thiamin inadequacy among WRA from 92% to 47% and among children from 86% to 44%, with similar reductions in urban and rural areas. The combinations of rice and wheat flour fortification at current and improved compliance were predicted to reduce the prevalence of inadequacy by several additional percentage points.

Adding riboflavin to Senegal’s wheat flour fortification standard at the WHO recommended level (2 mg/kg) reduced the predicted prevalence of apparent inadequacy by several percentage points among WRA and children; however, in both the current and improved wheat flour compliance scenarios, the diets of 92%–94% of WRA and 90%–92% of children remained inadequate in riboflavin (online supplemental table S12). Because riboflavin turns fortified rice kernels orange/yellow, which can negatively impact consumer acceptance, riboflavin is typically not recommended for rice fortification,35 and therefore, we did not model rice fortification with riboflavin.

Senegal’s wheat flour fortification programme at currently estimated compliance was predicted to reduce the national prevalence of folate inadequacy among WRA by 10 percentage points (from 92% to 82%) and improved compliance with the standard was predicted to bring inadequacy to 53%, although inadequacy was predicted to remain above 70% in some regions (figure 1, folate (a); online supplemental table S13). The existing programme at current and improved compliance was also predicted to reduce folate inadequacy among children by 20 percentage points (from 76% to 56%) and 45 percentage points (to 31%), respectively. If Senegal revised the wheat flour fortification standard from 2.5 mg/kg to 5 mg/kg folic acid to align with current WHO guidance, predicted folate inadequacy among WRA was predicted to decline to 61% at current compliance and 25% with improved compliance, although inadequacy among WRA in the Kedougou and Sedhiou regions remained above 50% even with improved compliance. At this higher fortification level, folate inadequacy among children was predicted to decline to 18%–35%, depending on assumed compliance with the standard. There was no predicted risk of high folic acid intake among either WRA or children in any of these scenarios (online supplemental table S14).

Compared with wheat flour fortification alone, the combination of wheat flour at the current standard plus folic acid-fortified rice resulted in substantial reductions in folate inadequacy among WRA, reducing by 40 percentage points (from 82% to 42%) under the current compliance scenario and 25 percentage point (from 53% to 28%) under improved compliance (figure 1, folate (b)). Among children, this combination was predicted to reduce folate inadequacy by 28 percentage points (from 56% to 29%) under current wheat flour compliance and by 15 percentage points (from 31% to 16%) under improved wheat flour compliance. If the wheat flour standard was increased to 5 mg folic acid per kg, the combination of fortified wheat flour and rice was predicted to reduce folate inadequacy among WRA to between 13% and 31% and among children to between 9% and 18%, depending on assumed compliance with the wheat flour fortification standard. In terms of risk of high folic acid intake, fortified rice, either alone or in combination with fortified wheat flour, would put <1% of WRA at risk of high intake nationally and up to 3% of WRA in several regions with the combination or fortified rice and wheat flour at the higher fortification level of 5 mg/kg. Between 2% and 7% of children, nationally, were at risk of high folic acid intake with the combination or fortified rice and wheat flour, with much higher proportions of children in some regions at risk of high intake (eg, between 8% and 24% in the Ziguinchor Region; see online supplemental table S14).

Introducing vitamin B₁₂ into the wheat flour fortification standard was predicted to reduce the national prevalence of apparent inadequacy among WRA from 29% to 24% (currently estimated compliance) or 19% (improved compliance) and from 25% to 21% (currently estimated compliance) or 17% (improved compliance) among children (figure 1; online supplemental table S15). However, with wheat flour fortification alone, the prevalence of vitamin B₁₂ inadequacy was still >30% among WRA in rural areas (>25% among children) and remained above 50% among WRA in some regions (figure 3). Vitamin B₁₂-fortified rice was predicted to reduce inadequacy to 15% nationally among WRA and to 12% nationally among children. The combination of fortified wheat flour and rice reduced inadequacy to 9%–12% nationally among WRA and to 8%–9% among children, although >30% of WRA would still have inadequate vitamin B₁₂ in the Tambacounda Region as well as ~40% of both WRA and children in the Kedougou region.

Request permission

The current wheat flour fortification programme was predicted to reduce apparent iron inadequacy among WRA by 7–15 percentage points (from 63% inadequate to 54% inadequate with current estimated compliance and to 46% inadequate with improved compliance) and by 5–10 percentage points among children (from 43% to 38% inadequate with current estimated compliance and to 33% with improved compliance), with reductions ranging from 8 to 21 percentage points among WRA and 7–15 percentage points among children across regions under the improved compliance scenario (figure 1; online supplemental table S16). The combination of iron-fortified rice and wheat flour at current or improved compliance was predicted to reduce inadequacy to 33%–39% and 23%–27% among WRA and children, respectively. There were no predicted risks of high iron intake among WRA or children as a result of wheat flour and/or rice fortification (online supplemental table S17).

Finally, adding zinc to Senegal’s wheat flour fortification standard (55 mg/kg) reduced the prevalence of apparent zinc inadequacy among WRA from 66% to between 47% and 56% nationally, depending on compliance with the standard, although the prevalence of inadequacy remained above 50% among WRA living in rural areas as well among women in many regions (figures 1 and 4; online supplemental table S18). Among children, wheat flour fortification reduced zinc inadequacy from 74% to between 61% and 67% nationally. Rice fortified with zinc was predicted to reduce the national prevalence of zinc inadequacy by several percentage points more than wheat flour fortification, and the combination of wheat flour and rice fortification reduced the predicted prevalence of inadequacy to between 32% and 38% among WRA and 48% and 52% among children. However, rice fortification alone and in combination with wheat flour fortification resulted in a substantial proportion (33%–41%) of children at risk of high zinc intake nationally. With wheat flour fortification alone, the national risk of high zinc intake among children was <5% but reached 5%–6% in the Dakar, Kolda and Matam regions with improved compliance.

Request permission