The potential contributions of bouillon fortification to meeting micronutrient requirements among women and preschool children in Senegal: A modeling study using household consumption and expenditure survey data

Abstract

To reduce micronutrient deficiencies, Senegal mandates the fortification of refined oil with vitamin A and wheat flour with iron and folic acid. Expanding Senegal's large-scale food fortification programs to include fortified bouillon could help fill the remaining gaps in dietary micronutrient requirements. Using 7-day household food consumption data collected between 2018 and 2019, we assessed the potential contributions of bouillon fortified with vitamin A (40–250 μg/g bouillon), folic acid (20–120 μg/g), vitamin B12 (0.2–2 μg/g), iron (0.6–5 mg/g), and zinc (0.6–5 mg/g) for meeting micronutrient requirements of women of reproductive age (WRA; 15–49 years old) and children (6–59 months old). Most households (90%) reported consuming bouillon, including poor and rural households. At modeled fortification levels, bouillon fortification reduced the national prevalence of inadequacy by up to ∼20 percentage points (pp) for vitamin A, 34 pp (WRA) and 20 pp (children) for folate, 20 pp for vitamin B12, 38 pp (WRA) and 30 pp (children) for zinc, and ∼8 pp for iron. Predicted reductions in inadequacy were generally larger among poor and rural populations, especially for vitamins A and B12. Our modeling suggests that bouillon fortification has the potential to substantially reduce dietary inadequacy of multiple micronutrients and could also help address inequities in dietary micronutrient inadequacies in Senegal.

INTRODUCTION

Although the adverse impacts of micronutrient deficiencies on population health, child survival, and economic productivity are well known,^1-3 deficiencies of one or more micronutrients remain remarkably common among women of reproductive age (WRA) and children under 5 years of age in low- and middle-income countries.⁴ In Senegal, the most recent nationally representative micronutrient survey data showed that 12.8% of children 12–59 months of age were deficient in vitamin A⁵ and approximately half were deficient in zinc.⁶ Over half of WRA were found to be zinc deficient,⁷ while ∼55% of WRA were deficient in folate.⁸

Micronutrient deficiencies are often driven by diets with low nutrient density^{3, 9, 10}; studies of dietary intake in Senegal are consistent with a high risk of micronutrient inadequacy.^11-13 Food fortification, which is the addition of one or more micronutrients to foods or condiments during processing, can be an effective strategy to improve the micronutrient adequacy of diets.^{9, 14, 15} In 2009, following the establishment of regional large-scale food fortification (LSFF) standards by the West African Economic and Monetary Union in 2007,¹⁶ Senegal implemented national standards mandating the fortification of refined oil with vitamin A and wheat flour with iron and folic acid.¹⁷ Bouillon, a condiment added to foods during cooking as a seasoning, is voluntarily fortified with iron or vitamin A in several countries in West Africa.^{18, 19} Because bouillon, like salt, is expected to be widely consumed across income groups and among urban and rural consumers alike,²⁰ bouillon may be an ideal candidate for fortification to complement existing fortification programs in Senegal.

In this paper, we present modeled estimates of the additional potential contribution (i.e., accounting for existing LSFF in Senegal) of bouillon fortified with multiple micronutrients (vitamin A, folic acid, vitamin B12, iron, and zinc) for meeting the micronutrient requirements of WRA (aged 15–49 years) and children aged 6–59 months in Senegal. In addition to assessing the potential contribution of fortified bouillon at the national level, we also assessed the extent to which bouillon fortification could help fill micronutrient gaps among poor and rural populations that are often harder to reach with other food vehicles either because they do not tend to consume the food vehicle or their consumption of the food vehicle comes from small-scale processing that may not be feasible or cost-effective.^{14, 21} This research is part of an initiative to generate evidence related to bouillon fortification including technical aspects of bouillon fortification, its potential role in contributing to intake of both micronutrients and sodium, and other issues that arise in conversations about bouillon fortification standards. A randomized controlled trial assessing the efficacy of bouillon fortified with vitamin A, folic acid, vitamin B12, iron, and zinc is currently underway in Northern Ghana.^{22, 23} Alongside the results of the ongoing efficacy trial and other evidence (Vosti et al., in preparation), the modeling results presented in this paper provide policymakers in Senegal with evidence to help inform discussions around the need for expanding the country's portfolio of intervention programs aimed at improving the micronutrient adequacy of diets and the potential for a national bouillon fortification program to help achieve that aim.

METHODS

Estimating apparent food consumption, micronutrient intake, and the adequacy of household diets

The primary source of data used in the analysis was the 2018–2019 Enquête Harmonisée sur les Conditions de Vie des Ménages (EHCVM),²⁴ a household consumption and expenditure survey implemented by the National Agency of Statistics and Demography in Senegal with support from the World Bank and the West Africa Economic Monetary Union Commission. The 2018–2019 EHCVM was administered to 7156 households in two waves (October–December of 2018 and April–July of 2019; each household was surveyed in one of the two waves) and is representative of the national level, regional level, and of urban and rural residence.²⁵ Primary sampling units were drawn from the most recent census data and then a two-stage sampling strategy was used in which 589 enumeration areas were first selected from the sampling frame and then 12 households per enumeration area were randomly selected for inclusion in the survey. We requested and received approval from Agence Nationale de Statistique et de la Démographie (ANSD) to use the 2018–2019 EHCVM data. All data used in the analyses were deidentified and were used in compliance with ANSD's data access policy.

Estimates of household food consumption were generated from data in the EHCVM household food consumption module, which asked the household respondent, who is typically the household member primarily responsible for making food purchase decisions and preparing household meals, to recall the quantities of 140 prespecified foods consumed by household members during the 7 days preceding the survey, as well as the source of consumption (purchases, own production, gifts, trade, etc.). Respondents were able to report quantities of food consumed in whatever units they preferred, including nonstandard units (e.g., piles of tomatoes, sachet of oil). Enumerators, who received theoretical and practical training on all survey modules, including household food consumption, used a reference album containing photos of food-specific nonstandard units to aid respondents in the identification of units of food consumed.²⁶ We managed extreme values in quantities of food apparently consumed by calculating the food- and region-specific 95th percentile of apparent household consumption per adult male equivalent (AME) and replacing food consumption quantities above the 95th percentile with the value at the 95th percentile.

We then matched each food in the food list with food composition estimates, primarily from the West African Food Composition Table (FCT),²⁷ supplemented with entries from the Nutrition Coordinating Center Nutrient Database for Standard Reference²⁸ and the Malawian FCT²⁹ for several food items for which a suitable match was not available in the West African FCT. Aggregate or generic food items, such as the single food item “Eggplant, Squash/Zucchini,” were matched with several food items to generate a weighted average of nutrient composition. Where relevant, we adjusted the total quantities consumed for the edible portion. For foods typically consumed cooked, we also adjusted for the yield factor from cooking and then matched those foods with cooked entries from the FCTs to account for changes in micronutrient content during cooking. We did not adjust for food waste or loss.

We calculated the average apparent household intake of energy, vitamin A, folate, vitamin B12, iron, and zinc by dividing the total household apparent intake by the 7 days of recall. For each of these micronutrients, we then calculated the apparent nutrient density of the household diet, expressed per 1000 kcal, by dividing the average daily household apparent intake of the micronutrient by daily household energy intake, multiplied by 1000.³⁰ Nutrient density is a metric of dietary quality, and classifications of adequacy and high intakes rest on the assumption that individuals are meeting their age- and sex-specific energy requirements.³⁰ Moreover, the use of household-level dietary data to assess the prevalence of inadequate and high micronutrient intakes among specific household members requires imposing the assumption that food is distributed within the household according to each household member's age- and sex-specific energy requirements. We qualify estimates of food consumption and nutrient intake as apparent to emphasize these and other assumptions inherent in using household-level data to assess dietary micronutrient adequacy and risk of high intakes.

Using information from the household roster in the ECHVM data, we identified target members (children 6–59 months of age and WRA) of each household. If a household had more than one child 6–59 months of age or more than one WRA, we randomly selected one household member from each target group to include in the analysis. We used the estimated average requirement (EAR) cut-point method to assess the adequacy of the household diet for meeting the vitamin A (μg retinol activity equivalents), folate (μg dietary folate equivalents), vitamin B12, and zinc requirements of target household members by comparing the nutrient density of the household diet to the critical nutrient density of target household members. We used the full probability approach, based on the iron density of household diets, to assess the adequacy of the household diet for meeting the iron requirements of children and nonpregnant WRA.¹⁴

Critical nutrient densities were calculated for each target household member as his/her EAR (or values along the distribution of requirements for iron) divided by his/her assumed energy requirements (Table S1). Nutrient requirements for vitamin A, folate, vitamin B12, and iron were from the US Institute of Medicine.^{31, 32} Based on dietary patterns in Senegal, we assumed 10% bioavailability of iron from all sources except fortified bouillon (detailed below).¹⁴ For each target household member, we calculated fractional zinc absorption as the ratio of absorbed zinc to total zinc by using published algorithms for children³³ and adults³⁴ to estimate absorbable zinc. Then, we estimated dietary zinc requirements (and critical nutrient densities) by adjusting physiological zinc requirements for children³⁵ and WRA³⁶ by the estimated person-specific fractional zinc absorption. Note that because pregnancy status was not collected in the EHCVM survey, to estimate the adequacy of the household diet for meeting the micronutrient requirements of both pregnant and nonpregnant WRA, we (1) estimated the prevalence of micronutrient inadequacy using critical nutrient densities for nonpregnant WRA, (2) estimated the prevalence of micronutrient inadequacy using critical nutrient densities for pregnant WRA, and (3) combined the two prevalence estimates as an average weighted by the proportion of pregnant and nonpregnant WRA in the population according to the most recent Demographic and Health Survey data.³⁷

For vitamin A (in the form of preformed retinol), folic acid, iron, and zinc, we also estimated the risks of high apparent intakes by comparing the nutrient density of the household diet to the critical upper density of the target household member, defined as the age- and sex-specific tolerable upper intake level (Table S2) divided by the age- and sex-specific energy requirement, expressed per 1000 kilocalories.

Modeling the contribution of large-scale food fortification

The Micronutrient Intervention Modeling-Secondary Data (MINIMOD-SD) tool was previously developed to estimate the nutrition benefits of micronutrient interventions using secondary data.³⁸ We used the MINIMOD-SD approach, coded in Stata 18, to model the contribution of LSFF, including bouillon fortification, in meeting dietary micronutrient requirements in Senegal. We first estimated the apparent consumption of refined oil, wheat flour, and bouillon among target household members using the AME method.³⁹ Specifically, we calculated an AME ratio for each target household member as his/her AME weight (his/her age- and sex-specific energy requirement to the requirement of an adult male aged 18–30 years) divided by the total number of AMEs in the household. We then multiplied the total daily household apparent consumption of the food vehicle by the target household member's AME weight to arrive at his/her apparent consumption of fortifiable foods. Estimates of apparent consumption of refined oil and wheat flour included both reported consumption of specific items on the food list as well as proportions of processed food items containing these food vehicles. More specifically, refined oil consumption included soybean oil, cottonseed oil, and “other” oils (which we assumed was primarily refined vegetable oil) plus the estimated quantity of refined oil in food items (e.g., cakes), while wheat flour consumption included wheat flour plus the estimated quantity of wheat flour in food items (e.g., bread).

We then modeled the contribution of existing LSFF by adjusting the nutrient density of the household diet to account for additional micronutrients provided via LSFF. Specifically, Senegal mandates the fortification of refined oil with vitamin A (17.5 mg/kg) and wheat flour with iron (60 mg/kg) and folic acid (2.5 mg/kg)¹⁷ (Table S3). We adjusted these fortification levels down to reflect the most recent information on the percent of each food vehicle that is fortifiable, the percent that is fortified to any extent, and the average fortification level at markets/households (Table 1, primary analysis). We multiplied the adjusted fortification levels by daily apparent household consumption of the food vehicle, recalculated nutrient densities of household diets with fortification, and compared nutrient densities with fortification to the critical nutrient densities of target household members.

TABLE 1. Fortification program modeling assumptions.

	Vehicle	Food vehicle that is industrially processed/fortifiable		Fortifiable food vehicle fortified to any extent		Average fortification level among fortified food as a percent of the standard after adjusting for expected losses from point of fortification to households
		Value	Source	Value	Source	Value	Source
Primary analysis	Refined oil	80%	Expert opinion^a	34%	Aaron et al. (2017)^54	30%	Expert opinion^a and adjusted for expected losses^b
	Wheat flour	100%	Expert opinion^a	96%	Republic of Senegal, Ministry of Industrial Development and Small and Medium Industries, & Institute of Food Technology (2020)^55	100%	Expert opinion^a and adjusted for expected losses^b
	Bouillon	100%	Assumption	75%	Modeling assumption	100%	Modeling assumption
Sensitivity analysis	Refined oil	80%	Expert opinion^a	75%	Modeling assumption	100%	Modeling assumption
	Bouillon	100%	Assumption	75%	Modeling assumption	100%	Modeling assumption

• ^a Expert opinion from USAID AFFORD assessment (interviews between March and May 2023).
• ^b Adjusted for expected micronutrient losses/degradation of 30% of vitamin A added to oil and 20% of folic acid added to wheat flour, based on expert opinion.

We assessed the potential additional contribution of bouillon fortification (i.e., in addition to the existing refined oil and wheat flour fortification program) over a range of possible fortification levels that were selected to allow for consideration of alternative micronutrient concentrations in fortified bouillon that might be chosen to balance reductions in micronutrient inadequacy with risk of high intakes. Specifically, we modeled 40–250 μg vitamin A per g bouillon, 20–120 μg folic acid per g bouillon, 0.2–2 μg vitamin B12 per g bouillon, 0.6–5 mg iron per g bouillon, and 0.6–5 mg zinc per g bouillon. To account for the possibility of less-than-perfect compliance, that is, if Senegal mandated the fortification of bouillon but not all bouillon was fortified in practice, we assumed 75% of bouillon would be fortified to the specified level in our modeling (i.e., we multiplied each fortification level by 0.75 to arrive at an average fortification level if 75% of bouillon were fortified). Then, we multiplied daily apparent bouillon consumption by each adjusted fortification level and recalculated the nutrient density of the household diet, prevalence of inadequacy, and where relevant, prevalence of high intake with both existing LSFF and bouillon fortification. We assumed that 2% of iron added to bouillon via fortification would be absorbed in the absence of additional compounds added to enhance iron absorption.^{40, 41}

Apparent micronutrient adequacy and consumption of fortifiable food were estimated at the national level, by wealth quintiles based on annual per capita household expenditures, and by urban and rural residence. We did not model the contributions of micronutrient supplements because consumption of supplements was not assessed in the EHCVM survey and because LSFF programs are often designed to reduce (or to eventually eliminate) the need for supplementation programs, so assessing the effects of fortification programs in the absence of supplementation programs can be useful for policy discussions. Similarly, we did not account for the contributions of voluntary fortification, as there is insufficient data on the extent of voluntary fortification and levels of micronutrients voluntarily added to some food products in Senegal. Finally, we did not account for the contribution of breastmilk to meeting the micronutrient requirements of young children because breastfeeding status was not collected in the EHCVM survey. As such, the modeling results reflect the adequacy of diets without breastmilk. We calculated per capita annual household expenditures based on household expenditure data reported in the EHCVM survey and used per capita annual household expenditures to construct quintiles of household socioeconomic status (SES). All comparisons of the prevalence of inadequacy at different fortification levels and across population subgroups, as well as apparent consumption of potentially fortifiable foods across food vehicles and population subgroups, were qualitative.

Regional estimates and sensitivity analysis

In addition to the modeling scenarios outlined above, we also modeled an additional scenario presented at the regional level, in which bouillon fortification levels were selected in order to meet 30% of the Codex nutrient reference values (NRVs). Calculations assumed 2.5 g of bouillon consumption per day among adults (per g bouillon: 96 μg vitamin A, 28.8 μg folic acid, 0.288 μg vitamin B12, 2.64 mg iron, and 1.68 mg zinc) (Table S3). These levels were chosen because they may be commercially feasible for voluntary fortification and because, according to Codex, meeting 30% of Codex NRVs allows for including a claim of “high in” on labels. As before, we adjusted these fortification levels to account for the assumption that 75% of the bouillon would be fortified to these levels.

Finally, as a sensitivity analysis, we estimated the potential additional contribution of fortified bouillon to meeting vitamin A requirements (and risk of high intakes) if compliance with Senegal's refined oil fortification program were improved from current levels (where current refers to the most recent evidence on compliance; Table 1, primary analysis) to a scenario in which 75% of fortified refined oil was fortified to the national standard (Table 1, sensitivity analysis). Because reported compliance with the wheat flour standard is currently high, we did not include wheat flour in the sensitivity analysis.

RESULTS

Apparent consumption of fortifiable foods and condiments

Most households in Senegal reported consuming refined oil or processed food products containing refined oil (90% of households nationally), wheat flour or processed food products containing wheat flour (89% of households), and bouillon (90% of households) in the 7 days preceding the EHCVM survey (Table 2). Reported consumption of refined oil and wheat flour was more common among wealthier households and urban households than households in lower quintiles of SES and rural households. Conversely, bouillon was consumed by a higher proportion of households in lower SES quintiles and by rural households. Among consumers, the median apparent consumption of refined oil was 44.6 and 21 g/day nationally among WRA and children, respectively, with higher average apparent consumption among households in higher SES quintiles and among urban households. The average apparent consumption of wheat flour among consumers, which was 54.5 g/day among WRA and 25.6 g/day among children at the national level, followed similar subnational patterns. Nationally, the average apparent consumption of bouillon was 1.9 g/day among WRA and 0.9 g/day among children, ranging from 1.3 to 2.7 g/day among WRA and 0.7 to 1.3 g/day among children by household SES, with higher median bouillon consumption among wealthier households. In contrast to the proportion of households consuming bouillon, median apparent consumption was modestly higher in urban compared with rural households.

TABLE 2. Apparent consumption of refined oil, wheat flour, and bouillon in the past 7 days based on the 2018−2019 Enquête Harmonisée sur les Conditions de Vie des Ménages (EHCVM) survey.

			National	SES 1^c	SES 2	SES 3	SES 4	SES 5	Urban	Rural
Refined oil	Households consuming^a (%)		90%	82%	88%	93%	93%	94%	95%	85%
	Median apparent consumption among consumers, g/day	WRA	44.6	26.6	38.8	46.2	52.0	59.0	48.0	39.8
		Children^b	21.0	12.8	19.1	23.1	26.3	28.8	23.3	18.6
Wheat flour	Households consuming^a (%)		89%	71%	89%	93%	95%	96%	94%	84%
	Median apparent consumption among consumers, g/day	WRA	54.5	30.2	46.1	54.7	60.5	70.1	58.2	48.8
		Children	25.6	13.1	22.5	28.6	29.9	31.8	28.1	22.5
Bouillon	Households consuming^a (%)		90%	92%	92%	93%	91%	79%	87%	92%
	Median apparent consumption among consumers, g/day	WRA	1.9	1.3	1.6	1.9	2.2	2.7	2.0	1.8
		Children	0.9	0.7	0.8	1.0	1.1	1.3	1.0	0.9

• Abbreviations: SES, socioeconomic status; WRA, women of reproductive age.
• ^a Households reporting any consumption of the food vehicle (or processed food containing the food vehicle), from household purchases during the recall period, are categorized as consuming the food vehicle.
• ^b Children 6−59 months of age.
• ^c SES wealth quintiles based on annual per capita household expenditures. SES 1 is poorest/lowest SES; SES 5 is wealthiest/highest SES.

Prevalence of apparent inadequacy and modeled contribution of bouillon fortification

With refined oil fortification at current levels of compliance (Table 1, primary analysis), the national prevalence of apparent vitamin A inadequacy among both WRA and children was 66% (bouillon fortification level 0 in Figure 1 and Table S4). Note that from here forward, if not explicitly stated, inadequacy refers to apparent inadequacy. The prevalence of inadequacy was more than 30 percentage points higher among households in the lowest SES quintile compared to the highest, and over 20 percentage points higher among WRA and children in rural households compared to urban households. Nationally, bouillon fortification was predicted to reduce the prevalence of vitamin A inadequacy to 25%–57% among WRA and to 25%–56% among children, depending on the modeled fortification level (Figure 1). Bouillon fortification was predicted to lead to substantial reductions in the prevalence of vitamin A inadequacy across SES quintiles, with vitamin A inadequacy dropping at least 40 percentage points among WRA and children in all but the highest SES at 250 μg vitamin A per g bouillon, though given the high prevalence of inadequacy without bouillon fortification, the diets of a substantial proportion (∼40%) of WRA and children in the lowest SES quintile would remain inadequate. Across modeled fortification levels, the prevalence of vitamin A inadequacy among WRA in urban households was predicted to drop to 16%–44% and to 35%–79% among WRA in rural households, with similar predicted reductions for children (Figure S1). The risk of high intakes of preformed retinol did not exceed 1% among WRA for any fortification level, but at higher bouillon fortification levels (i.e., ≥160 mg/kg), the risk of high intakes prevalence reached 3%–4% among children nationally and up to 6% among children in the highest SES quintile and in urban areas (Figure 1, Table S5, and Figure S1).

FIGURE 1

Prevalence of apparent vitamin A inadequacy (top) or risk of high intakes (bottom), nationally and by socioeconomic status (SES), with bouillon fortification added to existing fortification (34% of fortifiable refined oil fortified to 30% of the national standard of 17.5 mg/kg) at different bouillon fortification levels. Note that the modeled bouillon fortification levels (x-axis) are not always evenly spaced. SES quintiles based on annual per capita household expenditures. SES 1 is poorest/lowest quintile of SES; SES 5 is wealthiest/highest quintile of SES.

Accounting for the existing wheat flour fortification program, the prevalence of folate inadequacy among WRA was 42% nationally, ranging from 54% to 32% from the lowest to highest SES quintile, respectively, and was 5 percentage points higher among WRA in rural compared to urban households (Figure 2, Table S6, and Figure S2). Patterns were similar among children, though at lower levels of inadequacy. Fortification of bouillon with folic acid was predicted to reduce the national prevalence of folate inadequacy to 7%–27% among WRA and to 4%–17% among children, depending on the modeled fortification level. At the highest modeled fortification level, the prevalence of folate inadequacy was predicted to drop below 15% among WRA and below 10% among children across all SES and in urban and rural areas. At this fortification level, the risk of high folic acid intakes was predicted to impact 1% of WRA nationally (up to 2% across SES) and 3% of children nationally, across most SES quintiles, and in urban and rural areas (Figure 2, Table S7, and Figure S2).

FIGURE 2

Prevalence of apparent folate inadequacy (top) or risk of high intakes (bottom), nationally and by socioeconomic status (SES), with bouillon fortification added to existing fortification (96% of fortifiable wheat flour fortified to the national standard of 2.5 mg/kg) at different bouillon fortification levels. SES quintiles based on annual per capita household expenditures. SES 1 is poorest/lowest quintile of SES; SES 5 is wealthiest/highest quintile of SES.

Senegal does not currently mandate the fortification of a food or condiment with vitamin B12. Without bouillon fortification, the national prevalence of vitamin B12 inadequacy was 29% among WRA and 25% among children nationally, though with a substantially higher prevalence of inadequacy among households in lower SES compared to higher SES and among rural compared to urban households (Figure 3, Table S8, and Figure S3). With bouillon fortified to between 0.2 and 2 μg vitamin B12 per g bouillon, the predicted prevalence of inadequacy dropped to 7%–24% among WRA and to 6%–20% among children. At 2 μg vitamin B12 per g bouillon, the contribution of fortified bouillon to reducing the prevalence of inadequacy would be largest among WRA and children in the lowest SES quintile (41 and 36 percentage point reductions, respectively) and among WRA and children in rural households (20 and 25 percentage point reductions, respectively). Correspondingly, the difference in the prevalence of inadequacy between the highest and lowest SES quintiles decreased from 38 to 28 percentage points among WRA and from 37 to 29 percentage points among children.

FIGURE 3

Prevalence of apparent vitamin B12 inadequacy, nationally and by socioeconomic status (SES), with bouillon fortification at different bouillon fortification levels. Note that the modeled bouillon fortification levels (x-axis) are not always evenly spaced. SES quintiles based on annual per capita household expenditures. SES 1 is poorest/lowest quintile of SES; SES 5 is wealthiest/highest quintile of SES.

With iron-fortified wheat flour, the national prevalence of iron inadequacy was 43% among WRA and 31% among children, with limited variation across SES and urban and rural residence (Figure 4, Table S9, and Figure S4 at bouillon fortification level 0). Even at the highest modeled fortification level (5 mg iron per g bouillon), reductions in the prevalence of iron inadequacy would not exceed 8 percentage points among WRA and 6 percentage points among children, with very similar percentage point reductions across SES and urban and rural residence. The risk of high iron intakes was predicted to reach 2%–3% among WRA at 5 mg iron per g of bouillon (no risk of high intakes among children) (Table S10).

FIGURE 4

Prevalence of apparent iron inadequacy (top) or risk of high intakes (bottom), nationally and by socioeconomic status (SES), with bouillon fortification added to existing fortification (96% of fortifiable wheat flour fortified to the national standard of 60 mg/kg) at different bouillon fortification levels. Note that the modeled bouillon fortification levels (x-axis) are not always evenly spaced. SES quintiles based on annual per capita household expenditures. SES 1 is poorest/lowest quintile of SES; SES 5 is wealthiest/highest quintile of SES.

Finally, the prevalence of zinc inadequacy without fortification was 66% nationally among WRA and 74% among children. Across household SES quintiles, the prevalence ranged from 55% to 67% among WRA and 57% to 81% among children, and it ranged from 59% to 75% among WRA and 74% to 75% among children in urban and rural areas (Figure 5, Table S11, and Figure S5). Depending on the modeled zinc fortification level, the national prevalence of inadequacy was predicted to decline to 29%–56% among WRA and to 44%–68% among children, with fairly similar predicted percentage point reductions across SES and in urban and rural areas. However, at 1.8 mg zinc per g of bouillon, 8% of children nationally were predicted to be at risk of high zinc intakes, reaching 45% of children at 5 mg zinc per g bouillon (Table S12).

FIGURE 5

Prevalence of apparent zinc inadequacy (top) or risk of high intakes (bottom), nationally and by socioeconomic status (SES), with bouillon fortification at different bouillon fortification levels. Note that the modeled bouillon fortification levels (x-axis) are not always evenly spaced. SES quintiles based on annual per capita household expenditures. SES 1 is poorest/lowest quintile of SES; SES 5 is wealthiest/highest quintile of SES.

Modeled contribution of bouillon fortification at 30% of Codex by region

Estimates of the prevalence of inadequacy by region without bouillon fortification and with bouillon fortified to 30% of the Codex NRVs are presented in Figure 6 (WRA), Figure 7 (children), and Tables S13–S17. Across regions and accounting for the contribution of the existing refined oil fortification program at current levels of compliance, the prevalence of vitamin A inadequacy was between 48% in the Dakar and Ziguinchor regions and 86% in the Louga Region. With bouillon fortified at 96 μg vitamin A per g of bouillon, the predicted prevalence of inadequacy dropped to between 27% (Dakar Region) and 71% (Matma Region). Among children, the regional prevalence of vitamin A inadequacy without and with bouillon fortification were similar to those among WRA.

FIGURE 6

Prevalence of vitamin A, folate, vitamin B12, iron, and zinc inadequacy among women of reproductive age without (left panels) and with (right panels) bouillon fortification modeled meet 30% of the Codex nutrient reference values, calculated assuming 2.5 g of bouillon consumption per day among adults (per g bouillon: 96 μg vitamin A, 28.8 μg folic acid, 0.288 μg vitamin B12, 2.64 mg iron, and 1.68 mg zinc). See Tables S13−S17 for point estimates by region.

FIGURE 7

Prevalence of vitamin A, folate, vitamin B12, iron, and zinc inadequacy among children 6−59 months of age without (left panels) and with (right panels) bouillon fortification modeled meet 30% of the Codex nutrient reference values, calculated assuming 2.5 g of bouillon consumption per day among adults (per g bouillon: 96 μg vitamin A, 28.8 μg folic acid, 0.288 μg vitamin B12, 2.64 mg iron, and 1.68 mg zinc). See Tables S13−S17 for point estimates by region.

Accounting for current wheat flour fortification, folate inadequacy among WRA ranged from 29% in the Diourbel Region to 65% in the Kédougou Region. Bouillon fortified with 28.8 μg folic acid per g was predicted to reduce folate inadequacy to between 16% (Fatick Region) and 46% (Sédhiou Region). Among children, folate inadequacy was predicted to drop from between 15% (Kaffrine Region) and 48% (Kédougou Region) to between 6% (Fatick Region) and 26% (Sédhiou Region) with bouillon fortification.

Without bouillon fortification, the prevalence of vitamin B12 inadequacy among WRA varied widely across regions, from 16% in the Dakar Region to 80% in the Kédougou Region. With bouillon fortified with 0.288 μg vitamin B12 per g of bouillon, inadequacy was predicted to drop to between 9% (Dakar and Diourbel regions) and 69% (Kédougou Region). Regional patterns of vitamin B12 inadequacy were similar among children.

Accounting for the current wheat flour fortification program, iron inadequacy among WRA ranged from 37% in the Fatick Region to 56% in the Saint-Louis Region and among children from 24% in the Kaffrine Region to 41% in the Saint-Louis Region. Adding bouillon fortified with 2.64 mg iron per g of bouillon was predicted to decrease the prevalence of iron inadequacy by 2–7 percentage points among WRA and by 1–5 percentage points among children.

Finally, zinc inadequacy among WRA without bouillon fortification was 52% in the Dakar Region and ranged up to 82% in the Saint-Louis Region. Adding bouillon fortified with 1.68 mg zinc per g of bouillon was predicted to reduce the prevalence of zinc inadequacy among WRA to between 19% (Dakar Region) and 67% (Kaffrine Region). Among children, zinc inadequacy dropped from between 59% (Dakar Region) and 87% (Saint-Louis Region) without bouillon fortification to between 38% in the Dakar region and 72% in the Saint-Louis Region.

Sensitivity analysis

If compliance with the current refined oil fortification standard in Senegal were improved such that 75% of refined oil was fortified to the standard (Table 1, sensitivity analysis), the national predicted prevalence of vitamin A inadequacy would be 28% among WRA and 30% among children, although it would be above 50% among WRA and children in the lowest SES quintile and above 40% in rural areas (Figures S6 and S7). Across modeled fortification levels, the addition of vitamin A–fortified bouillon was predicted to reduce the prevalence of inadequacy, especially among WRA and children in poor households and in urban households. However, the risk of high preformed retinol intakes among children exceeded 5% nationally at 200 μg/g and reached 10% at 250 μg/g, with the risk of high intakes among children in households of the highest SES quintile reaching 18%. At 250 μg/g, the national risk of high preformed retinol intakes among WRA was 1%, reaching 2% among WRA in the highest SES quintile.

DISCUSSION

The design of new LSFF programs or redesign of existing LSFF programs with the potential to successfully reduce micronutrient inadequacies requires, among other things, evidence on the need for and potential impacts of LSFF programs.¹⁴ The preferred dietary data source to assess needs and predict program impacts is generally individual-level, 24-h dietary recall data,⁴² complemented with nutrient biomarker data to understand the extent of deficiency.¹⁴ However, given the scarcity of nationally representative 24-h dietary recall data, household food consumption data have increasingly been used to estimate the prevalence of micronutrient inadequacies in populations and to model the impacts of LSFF and other micronutrient interventions.^{43, 44}

Because Senegal does not have nationally representative 24-h dietary recall data, we used household food consumption data from the most recent household survey to model the potential contributions of implementing a nationally mandated bouillon fortification program. Our analyses showed that, even with the existing refined oil and wheat flour fortification programs, household diets in Senegal are inadequate to meet the vitamin A and zinc requirements of over 65% of WRA and children 6–59 months of age and the folate, vitamin B12, and iron requirements of more than 25% of these populations. Our analyses also showed significant subnational variation in the prevalence of micronutrient inadequacies, with the burden of inadequacy often substantially higher among WRA and children in poorer households and in rural households compared to wealthier households and those in urban areas. Reported consumption of bouillon was high among all households, including poor and rural households. As a result, our modeling predicted that the implementation of a bouillon fortification program could bring about sizeable reductions in the prevalence of inadequacy among all households, including poor and rural households that are typically harder to reach via fortification of refined oil and wheat flour. Specifically, across a range of potential fortification levels, we found that fortified bouillon had the potential to reduce the prevalence of vitamin A inadequacy among WRA and children in all but the highest SES quintile by up to 40 percentage points or more (predicted reductions were ∼32 percentage points in the highest SES quintile). Bouillon fortification could bring down the prevalence of folate inadequacy among WRA by 34 percentage points nationally and by 21 percentage points among children. The prevalence of vitamin B12 inadequacy could be reduced by up to ∼20 percentage points among WRA and children with bouillon fortification, with potential impacts among the poorest households almost twice that large. Zinc-fortified bouillon was predicted to reduce the high prevalence of zinc inadequacy among WRA (66% nationally) and children (74% children) by up to 38 percentage points among WRA and up to 30 percentage points among children. Given limited iron absorption from fortified bouillon, the predicted reductions in iron inadequacy were more modest (∼8 percentage points) and did not vary much across SES groups or urban and rural residence.

At the same time, our modeling showed that (without accounting for high-dose vitamin A supplementation) at bouillon fortification levels of 200 μg vitamin A per g and higher, more than 5% of children in households in the highest SES quintile were at risk of high vitamin A intakes. If compliance with the refined oil fortification standards improved, the predicted risk of high intakes among children reached 10% nationally with bouillon fortified at 250 μg per gram. As such, fortifying bouillon or other food vehicles, such as rice, with vitamin A would need to be carefully planned and coordinated with other micronutrient intervention programs. A high proportion of children were also predicted to be at risk of high zinc intakes at modeled bouillon fortification levels above 1.8 mg zinc per g of bouillon. However, given the high prevalence of zinc inadequacy of diets combined with recent criticism that tolerable upper zinc intake levels for children have been set below levels of observed usual dietary zinc intake that show no adverse effects,⁴⁵ the potential nutrition benefits of bouillon fortification would need to be weighed against the potential risk of high intakes when setting zinc fortification levels.

The results presented in this paper should be interpreted in the context of study limitations. First, we used household-level food consumption data. Therefore, our estimates of apparent consumption of fortifiable foods, the prevalence of dietary micronutrient inadequacy, and the modeled contribution of bouillon fortification to meeting micronutrient requirements of WRA and children are based on the assumption that food is distributed within the household in proportion to age- and sex-specific energy requirements. It is not possible to test this assumption using available data, nor is it possible to predict the direction of the error if this assumption does not hold in practice. Another limitation stemming from the use of household-level data is the likelihood of error in reported food consumption as a result of recall error and inadequate accounting for foods consumed away from home.⁴³ In an effort to account for some of this error, we assessed the adequacy of diets using the energy-adjusted nutrient-density metric. However, while nutrient density can help correct for measurement error in food consumption that is similar in nutrient composition to the reported household diet, it relies on the assumption that household members are meeting their energy requirements, and it cannot help to account for error in the reporting of foods that vary considerably in nutrient contents from the typical household diet (i.e., if specific micronutrient-rich foods were disproportionally over or underreported). Related to this, given inadequate accounting for foods consumed away from home, which may contain fortifiable food vehicles (e.g., refined oil or bouillon), the predicted impacts of LSFF on dietary adequacy may be underestimated, particularly for wealthier and urban households that, based on household expenditure patterns, tend to spend more on foods outside the home. Several studies have compared estimates of fortifiable food consumption and micronutrient intake and adequacy based on household consumption and expenditures surveys (HCESs) versus 24-h dietary recall data (e.g., Refs. 38, 46-49), but more research in a wider range of country contexts is needed. Guidelines for collecting high-quality food consumption data in HCESs have been developed,⁵⁰ but additional research designed to assess the accuracy of household-level food consumption data collected via HCESs and to establish best practice methods for analysis would also be valuable.⁴³ Finally, these analyses focused on WRA and children aged 6–59 months. Other groups, including adolescent girls and boys, men, and older adults, could also potentially benefit from a bouillon fortification program in Senegal. Similarly, the risk of high intakes may differ for some nutrients and some groups (e.g., adult men) than those assessed here.

There are also several important factors when considering the introduction of a new fortified food vehicle that are not considered here. First, a successful fortification program depends on technical feasibility, commercial viability, and consumer acceptance of the fortified product. Although consumer acceptance of bouillon fortified with vitamin A, folic acid, vitamin B12, iron, and zinc at 45%–125% of Codex NRVs in 2.5 g of bouillon has been established in Northern Ghana,⁵¹ and the bouillon cube currently being used in the randomized trial in Northern Ghana contains 200 μg vitamin A, 80 μg folic acid, 1.2 μg vitamin B12, 4 mg iron, and 3 mg zinc per g of bouillon,^{22, 23} the full range of fortification levels modeled here has not been tested for technical feasibility, commercial viability, or consumer acceptability. Cost, affordability among all stakeholder groups, and cost-effectiveness are also important considerations; these issues are addressed in Vosti et al. (in preparation). Finally, discussions around the fortification of bouillon, a condiment that is ∼half salt by weight, inevitably raise concerns about sodium intake and noncommunicable diseases. WHO has supported global commitments to population sodium intake reduction to reduce cardiovascular deaths.⁵² Recent studies have estimated that bouillon contributes a small share of dietary sodium (e.g., in Senegal, bouillon only contributes 15.4% of total dietary salt).⁵³ However, the most effective interventions (and dietary targets) to decrease total dietary salt intake remain a subject of discussion. Sodium reduction efforts are compatible with the fortification of bouillon if sufficient data are available to adjust the fortification program design (e.g., the amount of micronutrient added) to account for changes in consumption of the food vehicle.²⁰ Stakeholder concerns about the health impacts of ingredients in bouillon, including salt, monosodium glutamate, and other preservatives, should be addressed during discussions around bouillon fortification policy.

Even after accounting for mandatory oil and wheat flour fortification, our modeling suggests that diets in Senegal are inadequate to meet the micronutrient requirements of a large proportion of WRA and children, and for some micronutrients, the burden of inadequacy is higher among poor and rural households. Bouillon fortified with multiple micronutrients has the potential to bring about substantial reductions in the prevalence of vitamin A, folate, vitamin B12, and zinc inadequacy and could help address inequities in dietary micronutrient inadequacies. This evidence, alongside considerations of cost and cost-effectiveness, can help inform evidence-based decision-making around the design of a bouillon fortification program in Senegal.

AUTHOR CONTRIBUTIONS

K.P.A., S.A.V., and R.E.-S. designed the study and developed the methods. K.P.A., A.T., M.B., H.P., and S.K. made the food composition table matches. K.P.A. analyzed and modeled the data and wrote the first draft of the manuscript. All authors contributed to the data interpretation and revisions of the manuscript, and read and approved the final manuscript.

ACKNOWLEDGMENTS

This work was supported, in part, by the Bill & Melinda Gates Foundation via a grant to Helen Keller International [INV-007916]. Under the grant conditions of the Foundation, a Creative Commons Attribution 4.0 Generic License has already been assigned to the Author Accepted Manuscript version that might arise from this submission. The authors gratefully acknowledge the Agence Nationale de Statistique et de la Démographie (ANSD) in Senegal and the World Bank for the EHCVM data.

COMPETING INTERESTS

H.P. is employed by the Food Fortification Initiative and Emory University. These organizations help country leaders promote, plan, implement, monitor, or evaluate food fortification. All other authors have no competing interests to declare.

Open Research

PEER REVIEW

The peer review history for this article is available at: https://publons.com/publon/10.1111/nyas.15156