Robust Approach to Optimize the Formulation of Iron-Rich Biscuits Using a Balanced Incomplete Block and Mixture Design

Abstract

In Senegal, 39.1% of children aged 5－17 years are affected by iron deficiency. Although fortification is a major mitigation strategy in Senegal, its full potential has not been reached in terms of public health impact due to factors linked to the characteristics of the food vehicle, availability, and acceptability by the target group from both an ethical and sensory standpoint. In this context, this study proposes a robust method for the rapid development of an iron-rich biscuit formulation to prevent iron deficiency among school-aged children. Nine biscuit formulations were developed based on local products using a mixture design, and the physicochemical and microbiological compositions were assessed. Hedonic tests were conducted using a balanced incomplete block design, followed by the “Just About Right” sensory evaluation test. The iron content of the biscuits was higher (3.4－5.8 mg/100 g) than that of commercial biscuits (0.4 mg/100 g). The biscuits were well appreciated, with scores averaging between 6.0 (like slightly) and 7.7 (like very much). Increasing the proportion of millet flour enhanced the iron content but reduced sensory acceptability. The mixture design was optimized with respect to overall appreciation and iron content. The best formula contained wheat, corn, and millet at 26.00, 18.17, and 7.82 g, respectively, per 100 g of dough, corresponding to an iron content of 4 mg/100 g and a hedonic score of 6.97. This research highlights the potential of local ingredients to produce affordable, nutritious, and well-accepted food vehicles, which are key to the prevention of iron deficiency anemia in school-aged children.

INTRODUCTION

Micronutrient deficiency (MND), or “hidden hunger,” is defined as a lack of vitamins and minerals essential for proper growth and development, which is estimated to affect more than two billion people worldwide, resulting in about one million premature deaths per year, particularly in sub-Saharan Africa (Hicks et al., 2019). MND has a direct impact on individuals and long-term effects on public health and economic development. Iron deficiency is the most common MND globally. In Africa, the prevalence of anemia is very high among children aged 6－59 months, amounting to 60% of this population (∼90 million individuals), and even higher among countries in West Africa, such as Senegal, where the rate is reportedly 71% (WHO, 2021). Among this age group of Senegalese children, 26%, 41%, and 4% have mild, moderate, and severe anemia, respectively (Diouf et al., 2015), while the prevalence of iron deficiency and iron deficiency anemia among children aged 5－17 years are 39.1% and 10.6%, respectively (Fiorentino et al., 2013). These figures are well above the threshold of 40% adopted by the World Health Organization to define anemia as a severe public health concern. Iron deficiency decreases physical capacity, reproduction, and mental capacity, and has been associated with stunted growth. Iron is an essential nutrient for cellular and physiological processes, especially oxygen fixation and transport in tissues. Iron requirements vary based on age, sex, and physiological state, but remain very high for children (Ohanenye et al., 2021).

Strategies to reduce iron deficiency and improve intake include nutritional supplementation, food fortification, and dietary diversification. Of these, nutritional supplementation is widely used to combat iron deficiency, although the success of this intervention is very limited due to low coverage rates, cultural beliefs and practices, insufficient supplies, and low compliance due to adverse effects, such as constipation, black stool, and metallic taste (Ndiaye, 2023). Fortification is a good alternative to nutritional supplementation. Although the incidence of anemia has decreased in countries that have adopted iron fortification, the impact on public health remains insufficient (Barkley et al., 2015). Fortification programs face several obstacles in Africa, particularly inadequate supplementary feeding practices. For instance, laws governing the fortification of refined oils with vitamin A and wheat flour with iron and folic acid have been adopted in Senegal, but the potential has not been realized due to low compliance. Moreover, quality control of enriched salt, oil, and wheat flour is lacking, while inadequate technological requirements limit production, and transport issues limit availability in rural areas (Ohanenye et al., 2021). Lastly, fortified products must meet the expectations of consumers from a sensory standpoint.

Biscuits, which are widely consumed in developing countries to improve the nutritional status of children, have an ideal fortification matrix as a functional food. Biscuits enriched with protein, fiber, minerals, and vitamins, with low fat and sugar contents are available in Senegalese markets. The substitution of wheat flour with local cereals and/or other ingredients is a new trend with great socioeconomic interest in Senegal. This substitution offers a double advantage, both economically and nutritionally. Sweet potatoes, tubers, other cereals, bananas, and more recently moringa and spirulina are widely used to enrich biscuits (Kaboré, 2012; Songre-Ouattara et al., 2017; Ashoush and Mahdy, 2019). Moringa, which is widely available in Senegal, has a high iron content and many therapeutic properties (Ndong et al., 2007). However, these plant-based ingredients can reduce iron bioavailability due to the presence of inhibitors, such as ascorbic acid and animal tissues, which are known to activate or catalyze the absorption of non-heme iron (Berger, 2004). A formulation with Moringa oleifera and other ingredients rich in vitamin C may increase the iron content of the product and potentially enhance non-heme iron bioavailability, but can affect organoleptic properties (color, texture, appearance, and taste) (Menasra, 2020). Many fortification programs have been hampered by the weak acceptance of the product by consumers, particularly children, who view organoleptic appeal as a premium determinant. Potato-enriched biscuits were found fair, pleasant, and very pleasant, respectively, by 59.37%, 31.25%, and 0% of tasters (Siedogo, 2009). Increased incorporation of moringa and spirulina leads to changes to the color and taste of biscuits that are not well accepted by consumers (Songre-Ouattara et al., 2017). Biscuits enriched with less than 1% moringa had “pleasant” ratings (Kaboré, 2012). To overcome these sensory limitations, while maintaining adequate iron levels, a filled-type biscuit was assessed in this study. The filling can serve as a vehicle for iron-rich ingredients, such as moringa, thereby limiting direct interaction with the dough and reducing the negative impact on organoleptic characteristics. In addition, the incorporation of jam or other fillings into baked biscuits has been shown to considerably enhance the organoleptic properties and provide a greater diversity of flavors and textures, thereby contributing to improved sensory appeal and consumer acceptance (Manley and Clark, 2011; Patrignani et al., 2022). Biscuits with acceptable sensory characteristics are often fortified at low levels due to the limited bioavailability of the fortificant (Berger, 2004). However, the formulation strategies used in prior studies of enriched biscuits were not specifically designed to simultaneously optimize nutritional value and sensory quality. Consequently, the identification of an optimal compromise between these two attributes remained limited. In this context, mixture design provides an efficient and rigorous scientific framework, as compared to conventional substitution approaches, to evaluate ingredient interactions, reduce the number of experimental trials, and identify optimal formulations (Goupy, 2006). However, this approach may generate a large number of formulations, thereby rendering complete sensory evaluation by each panelist particularly difficult. The balanced incomplete block (BIB) design enables reliable sensory evaluation of multiple formulations while reducing fatigue and minimizing confusion of the panelists (van der Linden et al., 2004).

The objective of this study was to develop an iron-rich biscuit based on local food ingredients to enhance the iron intake of school-aged children using a robust approach based on the BIB design. Nine different biscuit formulations were optimized to identify the best in terms of iron content and sensory score.

MATERIALS AND METHODS

Materials

Cereals (millet, wheat, and corn), M. oleifera powder, dried calyxes of Hibiscus sabdariffa, fruit pulp of Adansonia digitata and Detarium senegalense, peanut paste, sugar, water, and chemical yeast were purchased from local markets in Thiès and Dakar, Senegal.

Formulation

The selection of the biscuit ingredients was guided by the potential to improve the iron content of the formulation. Biscuits with a filling are popular among children. The filling can be cream, chocolate, or jam. The biscuit dough and filling were separately produced. As the primary ingredient in the dough, flour significantly influences the nutritional and organoleptic characteristics of biscuits. Thus, several biscuit doughs were produced based on varying the flour composition, but not the other ingredients. A mixture design was employed in the experimental framework to limit the number of trials.

Biscuit production

For Mixture 1, sugar and peanut paste were dissolved and mixed in water and oil to avoid lumps in the biscuit dough, while peanut paste was used as an oil substitute to improve the nutritional quality. For Mixture 2, cereal flour, moringa, and yeast were sifted and kneaded using a blender mixer into a homogeneous and machinable dough. The obtained paste was resistant to a certain degree of breakage and spread in thin layers. The dough was rested for 15 min before molding. The laminated and filled dough was baked at 190°C for 15 min.

Among the ingredients, moringa had the highest iron content and was incorporated at the desired levels to optimize the organoleptic quality of the biscuits. The filling ingredients consisted of moringa with H. sabdariffa, D. senegalense, and A. digitata. These ingredients are rich in ascorbic acid, which may enhance the bioavailability of non-heme iron in plant-based products. The filling formulation consisted of 8% moringa powder, 3% Hibiscus, 3% Adansonia, 2% Detarium, 11% sugar, and 3% cornstarch (Maizena^Ⓡ). The mixture was moistened to 70%, cooked to 60° Brix, and gelled with Maizena^Ⓡ diluted in tepid water. The filling accounted for 30% of the total biscuit weight in all formulations. All biscuit formulations included the same filling.

Mixture design

A mixture design was used to identify the optimal recipe while minimizing the number of assays (Goupy, 2006). The influence of the combination of three types of flour (wheat, corn, and millet) on the characteristics of the dough was investigated. The fundamental constraint of the mixture design required that the sum of all components must equal 100% of the mixture (52 g total flour content in the present study). In addition, the lower and upper amounts of each flour component were based on technological, nutritional, and sensory considerations. The combination of the mixture constraint and component amounts resulted in an extreme vertices design (Goupy, 2000). Based on a review of the literature, the minimum and maximum amounts of wheat, corn, and millet flour were 18/26, 8/26, and 0/26 g, respectively. The combination of these constraints defined the experimental domain. In addition to the four extreme vertices, four axial points and one center point were included to improve coverage of the constrained experimental region and enhance the estimation of mixture effects, resulting in a total of nine formulations. The different quantities of each type of flour used in the mixtures are described in Table 1. The total amount of flour in each mixture was 52 g/100 g of biscuit dough. The proportions of the remaining ingredients were constant across all formulations. Therefore, differences among biscuits were attributable only to the type of flour. The responses studied (iron content and sensory acceptability) were fitted using a special quadratic mixture model that included linear terms and estimable binary interactions among the flour components.

Table 1.

Proportions of ingredients among the different formulations

Ingredient (g/100 g)	Formula
	F1	F2	F3	F4	F5	F6	F7	F8	F9
Wheat1)	18	26	18	26	22	20	24	20	24
Corn1)	8	8	26	26	17	12.5	12.5	21.5	21.5
Millet1)	26	18	8	0	13	19.5	15.5	10.5	6.5
Sugar	10	10	10	10	10	10	10	10	10
Moringa	1	1	1	1	1	1	1	1	1
Peanut paste	6	6	6	6	6	6	6	6	6
Vegetable oil	15	15	15	15	15	15	15	15	15
Yeast	1	1	1	1	1	1	1	1	1
Water	15	15	15	15	15	15	15	15	15

¹⁾Correspond to the cereal flours, which are the only variable ingredients in the formulation (total=52 g for each formulation).

F1－F9, formulas 1 to 9 generated by the experimental design.

Microbiological analysis

The safety of the biscuit formulations was rigorously verified. Biscuit samples were tested for total aerobic mesophilic flora (TAMF), Escherichia coli, Salmonella, Clostridium, and yeasts/molds in accordance with the official standards for food microbiology (ISO, 2001; Afnor, 2002, 2009b, 2013a, 2017). These microorganisms are commonly monitored in biscuits and products intended for consumption by young children. Samples were considered compliant when the microbial counts met the acceptable limits of TAMF <10⁶ colony-forming units (CFU)/g, E. coli <10 CFU/g, Clostridium spp. <10² CFU/g, yeasts/molds <10³ CFU/g, and absence of Salmonella spp. in 25 g of product (de Luxembourg, 2011). Microbiological analyses of the biscuit samples were conducted one week after production. During storage, the biscuits were packaged in sealed polyethylene bags and stored at 25°C－30°C under normal laboratory conditions.

Sensory analysis

Sensory analysis was conducted to explore and elucidate the influence of specific factors of the formulations most appreciated by consumers, thereby providing a comprehensive understanding of consumer expectations regarding biscuits. These evaluations yielded objective data on taste preferences.

Consumer acceptability test

Sensory evaluation was conducted with 24 children aged 5－12 years, recruited from DIOUGOB Primary School in Sanar, a village located near the university. The participants represented the target population for the developed product. The panel consisted of equal numbers of boys and girls. The inclusion criteria were age of 5－12 years, regular school attendance, absence of known allergies to any of the ingredients used in the formulations, and the ability to understand the tasting instructions. Initially, a questionnaire was administered to the tasters (judges) to collect personal information (age, sex, ethnicity, educational level, allergies, etc.) as well as data regarding biscuit consumption habits (frequency, period, and mode of consumption). Then, the “Just About Right” sensory evaluation tool (JAR scale) and the hedonic test were employed to determine organoleptic preferences. The JAR scale included four descriptors (color, sweetness, softness, and mouthfeel) to identify the acceptable characteristics of biscuits and understand why consumers like or dislike a particular biscuit. The descriptors were scored by consumers using a three-level scale: “Too little,” “Just about right,” and “Too much.” For the hedonic test, each judge evaluated the coded samples by indicating their overall appreciation based on a 9-point scale ranging from 1 “extremely unpleasant” to 9 “extremely pleasant” (Mané et al., 2021).

Ethical statement

The study protocol was approved by the National Committee for Ethics in Health Research of the Ministry of Health and Social Action of Senegal (approval no. 000329/MSAS/CNERS/SP) and written informed consent was obtained from the parents or legal guardians of all participating children prior to inclusion in this study. In addition, verbal agreement was obtained from each child before the sensory evaluation sessions. Participation was voluntary, and all information collected was treated confidentially.

BIB design

A sensory test meeting several criteria was implemented based on the existing constraints. The number of judges or the number of products per judge is a common constraint. For each judge, testing of more than four products will result in mental weariness and fatigue, often referred to as “palate fatigue,” which could impact the opinion of the judge (van der Linden et al., 2004). Hence, a particular judge was not expected to rate all nine biscuit formulations during a single trial. To overcome the constraints of mental weariness and fatigue, a BIB design was used that met the two conditions given by Eq. 1 and 2 (Silva et al., 2014).

$$p\times r=s\times k$$ Eq. 1

$$\lambda =\frac{r\times (k-1)}{(p-1)}$$ Eq. 2

where p is the number of products studied, s is the number of subjects (judges) on the panel, k is the number of products tested by each judge, r is the number of repetitions per product (number of subjects testing a given product), and λ is the number of times a pair of products is scored (number of subjects simultaneously testing two given products). Each judge tested only three formulations to avoid fatigue, with the condition of λ=1. Given these findings, there was only one possible configuration (12, 9, 3) (Pagès and Périnel, 2007). In total, there were 12 judges and nine products, and each judge tested only three products (Fig. 1). To improve the reliability of the sensory evaluation, the same BIB design was replicated with a second independent group of 12 children. Consequently, the final sensory evaluation involved a total of 24 participants.

Fig. 1.

Repartition of the sensory test following the balanced incomplete block (12, 9, 3) design showing which product (F) was evaluated by each judge (J). Each shaded cell indicates that the corresponding biscuit formulation was evaluated by the respective judge.

Data analysis

The obtained data were assessed to determine the reliability of the BIB design. First, the coefficient of efficiency (E) was calculated to determine the quality of the plan. The E-value was defined as the inverse ratio between the variance of any product coefficient and the variance of the complete design coefficient (Eq. 3) (Fleiss, 1981; Pagès and Périnel, 2007):

$$E=\frac{p(k-1)}{k(p-1)}$$ Eq. 3

Second, panel effects were studied. The product rating is given by Eq. 4 (Pagès et al., 2006).

$$y_t=\mu +{\alpha }_i+{\beta }_j+{\gamma }_s+\alpha {\beta }_{ij}+\alpha {\gamma }_{is}+\beta {\gamma }_{js}+{\pounds }_{ijs}$$ Eq. 4

In the model, y_t represents the rating given to product (i) by judge (j). The constant term is denoted by µ. The effect of the product is represented by α_i, while the effect of the judge is indicated by β_j and the effect of rank (s) is denoted by γ_s. The interaction between the effect of the product (i) and that of the judge (j) is represented by αβ_ij. Additionally, αγ_is denotes the interaction between the effect of the product (i) and that of the session, while βγg_js illustrates the interaction between the effects of the judge and the session. Finally, £_ijs corresponding to the residual.

The various effects and performances of the panel were examined before providing a definitive response. The study of the effect of product αi required adjustment of the means (Tomassone, 1964a, 1964b). To this end, the total sum of the scores given by the judges who tested the product i (T_i) was checked and corrected. The corrected sum for product i (G_i) was calculated using Eq. 5.

$$G_i=S_i-\frac{T_i}{k}$$ Eq. 5

where S_i is the sum of the scores for product i. From the corrected sum, the corrected mean α_i is obtained by Eq. 6:

$${\alpha }_i=\frac{G_i}{r_E}$$ Eq. 6

where α_i is also considered to be the effect of product i (Fleiss, 1981).

The adjustment of the average takes into account the sum of the scores assigned to the product, the scores of the judges who evaluated the product, the partial averages, and the overall average.

Physicochemical analysis

Physicochemical analysis of the nine biscuit formulations was conducted based on defined standards and methods. Humidity was assessed by steaming in accordance with the standard NF V 03-903 (Afnor, 2013b). The protein content was quantified using the Kjeldahl Method, in accordance with standard NF V 18-100, the lipid content with the Soxhlet method as per the French standard NF V 03-905 (Afnor, 1997, 2009a), and the carbohydrate content via the Luff-Schoorl method. Energy values were calculated using the foundational energy conversion factors for macronutrients described by Atwater and Benedict. The levels of iron, zinc, and calcium were determined by atomic absorption spectrophotometry. Finally, the vitamin C content was assessed by high-performance liquid chromatography.

Data processing

The mixture design and iso-response curves were generated with Minitab 17 software (Minitab, LLC). Data collection and normalization, hedonic average adjustment, and figure construction for the JAR test were performed with Microsoft Excel 2019 (Microsoft Corp.). The BIB design was generated with R software (version 4.2.1; https://cran.r-project.org/bin/windows/base/old/). The R package “SensoMineR” (https://cran.r-project.org/web/packages/SensoMineR/index.html) was employed to process the sensory analysis data, panel evaluation, and box plot construction. The hedonic scores of the different biscuit formulations were compared by one-way analysis of variance with Fisher’s least significant difference test, followed by post hoc comparisons of the mean values. Differences in the physicochemical parameters of the formulations were compared with Fisher’s least significant difference test. A P-value<0.05 was considered statistically significant for all tests. The formulations were optimized using Minitab 17 software with desirability functions ranging from 0 to 1 to enable comparisons of different response variables on the same scale. Hedonic and iron content responses were used to determine the proportions of the ingredients and identify the biscuit with the highest hedonic score together with an iron content ≥4 mg/100 g. The calculations were conducted with normalized data given by Eq. 7.

$$y=\frac{x-\min \left(x\right)}{\max \left(x\right)-\min \left(x\right)}$$ Eq. 7

where x is the real value.

RESULTS AND DISCUSSION

Characteristics of biscuit formulations

Nine biscuit formulations were obtained from the BIB design. The biscuits were round with starry ends, a mean weight of 33±1.0 g, and yellow to beige to green in color owing to variations in the ingredients (Fig. 2). Yellow coloration was due to the addition of corn flour, while a slight green color was associated with the presence of moringa (Songre-Ouattara et al., 2017). Similarly, Ashoush and Mahdy (2019) reported that the use of moringa results in biscuits with greener coloration and diminished yellow coloration. Biscuits have a sweet taste because of the addition of sugar in the dough and filling. The general flavor varies depending on the formulations as well as the texture. The visual appearance of the biscuit is complemented by the results of the sensory analysis.

Fig. 2.

Representation of the color variation of nine biscuits produced with the balanced incomplete block design.

Food safety compliance

The results of the microbiological analyses demonstrated the conformity of all samples to the food safety criteria. The contents of TAMF, E. coli, Salmonella, Clostridium, and yeasts/molds did not exceed the threshold values (Table 2), confirming compliance with “good hygiene practices” and “good manufacturing practices” (Aderinola et al., 2020).

Table 2.

Results of microbiological analyses

Formula	TAMF (CFU/g)			Escherichia coli (CFU/g)			Clostridium (CFU/g)			Yeast (CFU/g)			Mold (CFU/g)			Salmonella (CFU/25 g)		Conformity
	N	Int		N	Int		N	Int		N	Int		N	Int		N	Int
F1	800	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C
F2	100	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C
F3	12,000	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C
F4	400	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C
F5	600	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C
F6	8,000	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C
F7	40,000	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C
F8	1,000	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C
F9	1,500	S		<10	S		<10	S		<10	S		<10	S		Abs	S	C

TAMF, total aerobic mesophilic flora; CFU, colony-forming unit; N, numeration; Int, interpretation; S, satisfying; Abs, absent; C, conform.

Sensory attributes

A BIB design was used to plan and evaluate the sensory tests. The efficiency coefficient was 75% among 12 subjects (Eq. 3), where values <67% generally indicate a significant loss of efficiency and that more evaluators should be added to the trial (Fleiss, 1981). A non-optimal design with an efficiency of 0.77 is considered a very satisfactory design (Pagès and Périnel, 2007). To increase the efficiency coefficient, two independent blocks of 12 subjects were formed, resulting in a total of 24 judges participating in the study. This increase resulted in doubling of the number of subjects testing a given product from 4 to 8. The efficiency coefficient remained at 75%, as replicating the same design does not change E. In a study of incomplete blocks, Pagès and Périnel (2007) concluded that when r=8, the efficiency of an optimal design created with an algorithm is always very satisfactory.

The second step was to evaluate the significance of the different effects of the parameters in Eq. 4. The results show that the only significant effect on the responses of the judges was based on the different formulations and that the other parameters did not influence their opinions when using the BIB design (Table 3). A study devoid of these effects guarantees the reliability of the results and facilitates reproduction. The reduction or elimination of these biases resulted in more accurate and meaningful evaluations, which is essential for sensory analysis. Hence, the average responses were corrected (Eq. 6) and reported as adjusted mean values (Table 4).

Table 3.

Assessment of panel performance using the AOV model (P<0.05)

Effect	Formula	Judge	Rank	Formula-judge	Formula-rank	Judge-rank
P-value	0.04	0.43	0.78	0.44	0.41	0.44

AOV, analysis of variance.

Table 4.

Calculated values of the adjusted means of the hedonic responses for each formulation

Formula	Si	Ti	xi	αi	yi
F1	50	154	6.3	—0.22	6.4
F2	52	160	6.5	—0.22	6.4
F3	48	151	6.0	—0.39	6.2
F4	60	171	7.5	0.50	7.1
F5	50	157	6.3	—0.39	6.2
F6	48	155	6.0	—0.61	6.0
F7	54	158	6.8	0.22	6.8
F8	52	156	6.5	0.00	6.6
F9	60	160	7.5	1.11	7.7

S_i, sum of the hedonic scores of the formulation (i); T_i, sum of hedonic scores given by the judges who evaluated the formulation (i); x_i, mean hedonic score of the formulation (i); α_i, corrected average of the formulation (i); y_i, adjusted mean hedonic score of the formulation (i).

Hedonic test results

The post hoc test on adjusted means showed two homogeneous groups (“a” and “b”) (Fig. 3A). F9 had the highest adjusted hedonic score and was significantly different from F1, F2, F3, F5 and F6, but not from F4, F7 and F8. According to the overall assessment, F9 was significantly more appreciated than F1, F2, F3, F5 and F6. The biscuits in group “a” were characterized by a low millet content and those in group “b” by a high millet content. The iso-response curve (Fig. 3B) confirmed that when the millet content was increased in the formulations, the overall appreciation decreased. Other studies have found similar results, with the sensory quality of biscuits decreasing in parallel with the substitution of millet flour (Kulthe et al., 2018; Kulkarni et al., 2021). Similarly, Ranasalva and Visvanathan (2014) reported that the acceptability of breads decreased as the level of millet flour was increased. Furthermore, McSweeney et al. (2016) showed that increasing the millet content in biscuits and extruded snacks resulted in lower flavor, texture, and overall acceptability scores. Likewise, Goswami et al. (2015) reported that biscuits made entirely from wheat flour achieved the highest overall acceptability scores, whereas those produced solely from millet flour received the lowest. The judges seemed to tend to penalize products enriched with millet flour, which may be explained by the fact that, as compared to wheat flour, millet flour is low in gluten, a protein that gives the dough elasticity, suppleness, and greater capacity to absorb moisture. These properties give the biscuits a texture and “melt-in-the-mouth” appearance, which is considered ideal by consumers (Fustier et al., 2009). Shimray et al. (2012) reported that replacing wheat flour with millet flour reduced the water absorption capacity, stability, and extensibility of the dough, mainly due to the absence of gluten. The lower gluten content associated with millet incorporation disrupts the formation of the viscoelastic network of the dough, resulting in decreased cohesiveness and extensibility. Consequently, biscuits produced with higher levels of millet flour often exhibit a more brittle texture, darker color, and reduced sensory acceptability (Reaz et al., 2023). However, biscuits and bread produced with local cereals tend to be unfamiliar to consumers, as these products are not widely available in local markets. Although Senegalese politicians promote the consumption of locally produced products, there has been no effect on the availability of biscuits made from local cereals.

Fig. 3.

Hedonic test results. (A) Boxplots of adjusted hedonic scores of the different biscuit formulations, with different letters indicating significant differences according to Fisher’s least significant difference test (P<0.05). Blue markers indicate the adjusted mean hedonic scores (yi) presented in Table 4. (B) Contour plot showing the predicted adjusted hedonic scores as a function of the proportions of wheat, corn, and millet.

Apart from millet flour, other ingredients used in the biscuit recipe, such as corn flour, peanut paste, and moringa, can reduce the organoleptic value of fortified products. Songre-Ouattara et al. (2017) found that the moringa content significantly affected the acceptability of enriched sorghum biscuits, as those produced with lowest levels of moringa received the highest scores. Moreover, they found a decrease in the sensory quality of biscuits when wheat flour is replaced with corn flour. The greater the number of ingredients, the more difficult it is to predict the sensory characteristics, which explains why cereal-based foods are often fortified by replacing wheat flour with a single ingredient.

JAR test results

The JAR test results are presented in “Too little,” “Just About Right,” “Too much” (Fig. 4).

Fig. 4.

Results of the Just About Right (JAR) test regarding color (A), sweetness (B), softness (C), and mouthfeel (D), expressed as percentages of the eight panelists who evaluated each formulation.

In regard to color, the formulations F2, F3, F8, and F9 had the highest appreciation rate, as 75% of the judges considered the color “just about right” (“Too little,” “Just About Right,” “Too much”). Biscuits with a high millet content (F1 and F6) were rated as too dark, those with moderate millet content were considered “just about right,” and those without millet (F4) were considered too light (Fig. 4A). The same trend was observed for biscuits enriched with millet flour, as the panelists described the biscuits as having an undesirable gray color, while the control samples (100% wheat flour) had lighter colors (Kulkarni et al., 2021). Arepally et al. (2023) observed significantly darker biscuits as the proportion of millet flour was increased. This darkening effect was attributed to the naturally darker color of millet flour, as well as its higher contents of phenolic compounds and minerals than wheat flour. In addition, moringa mixed with millet can enhance the dark color of biscuits. Ashoush and Mahdy (2019) reported that the appearance of biscuits became darker as the quantity of moringa was increased. Incorporating moringa gives the biscuits a darker color that is less appreciated by consumers (Songre-Ouattara et al., 2017). Notably, the F4 formulation, which did not contain millet, was considered too clear by more than 50% of consumers. Another study showed that formulations containing millet give biscuits a darker color as compared to wheat or corn flour, which yields a yellow color.

The sweetness was most appreciated in formulations F3, F4, F8, and F9, with at least 75% of the judges giving a “just about right” rating for each (“Too little,” “Just About Right,” “Too much,” Fig. 4B). These formulations were all characterized by a high corn content and low millet content, likely because of the higher carbohydrate content of corn kernels. Moreover, wheat and corn flour have higher levels of complex carbohydrates and simple sugars than millet flour. On the other hand, Kulkarni et al. (2021) found that incorporating millet flour gave biscuits a pleasant cooked-millet taste. However, an increase in the moringa content gives biscuits a bitter taste, which is not appreciated by consumers (Songre-Ouattara et al., 2017). The biscuits with the best taste were those with the lowest content of millet flour.

In regard to softness (Fig. 4C), F4 was the most appreciated, with 87.5% of judges considering the texture “just about right.” Unlike F4, only 37.5% of the judges liked F1, as many considered it “too hard” (“Too little,” “Just About Right,” “Too much”). F4 had no millet, while F1 had the most millet. Appreciation of softness varied inversely with the millet content of the biscuit. Kulthe et al. (2018) also found that non-millet enriched biscuits had a crispier texture, which may be explained by the lack of elasticity of millet flour, whereas wheat flour in the presence of gluten and water can form more friable walls.

For the descriptor “mouthfeel,” the judges rated F4 and F9 the highest, with 87.5% appreciation (“Too little,” “Just About Right,” “Too much,” Fig. 4D), confirming the previous results that the formulations with the least millet had the best texture. The presence of millet flour could explain this difference in perception. Increasing the millet flour content makes the biscuits grittier and less crisp (Kulkarni et al., 2021). Salama et al. (2023) found that incorporating small amounts of corn flour reduced the mouthfeel score, but at levels above 30%, the mouthfeel became more appreciated. This is in line with the results obtained in the present study.

Overall, as the millet flour content was increased, the appreciation for each of the four descriptors tended to decrease. It is important to note that regarding color, biscuits containing a low proportion of millet are more appreciated than those with a high content, as well as those that do not contain millet.

Physicochemical characteristics of the biscuits

The iron content of the biscuits varied from 3.4 to 5.8 mg/100 g (Fig. 5A). F1 and F6 had the highest iron contents, while F4 and F9 had the lowest. F1 and F6 had a large proportion of millet flour, unlike F4 and F9. Fig. 5B presents a contour plot of the predicted iron content (mg/100 g) as a function of the proportions of wheat, corn, and millet. The contour lines connect formulations with similar iron contents, while the color gradient represents variations in iron concentration within the experimental region. The plot shows that the predicted iron content increased with the proportion of millet flour. This may be because millet flour is richer in iron than corn and wheat flour. The iron quantity found in the different formulations was above that of the biscuits available in retail markets, averaging 0.4 mg/100 g (Prisca et al., 2022), and also higher than the iron content (1.74 mg/100 g) reported by Abd El-Hamid et al. (2019) for biscuits with 100% wheat flour. This difference is due to the presence of millet flour and moringa. Moringa is well known for its iron richness (18.39 mg/100 g) (Ndong et al., 2007). The addition of moringa increases the iron content of biscuits (Songre-Ouattara et al., 2017). Moreover, the biscuits produced in this study were richer in vitamin C, ranging from 1.5 to 2.1 mg/100 g. The vitamin C content was higher than reported by Mile et al. (2024) at 0.1－0.39 mg/100 g and Baqoyeva et al. (2023) at 0.58－1.47 mg/100 g, as were the biscuits available in markets at well below 1 mg/100 g or often negligible. These high vitamin C contents are probably due to the incorporation of baobab, Hibiscus, and Detarium, which are rich in vitamin C, a nutrient that enhances iron absorption. As one of the richest fruits in vitamin C, Detarium has a vitamin C content of 967 mg/100 g (Cisse et al., 2008). The consumption of baobab pulp as a rich source of vitamin C can increase iron absorption by a factor of two to three. The effects of baobab were similar to those of the direct addition of ascorbic and citric acids as enhancers of mineral absorption (Adetola et al., 2022). Ascorbic acid and animal tissues are known as activators or factors that catalyze the absorption of non-heme iron (Berger, 2004). The effect of ascorbic acid on iron absorption is more powerful than that of other organic acids, and is the most effective agent to improve the absorption of non-heme iron when stable in the food vehicle (Teucher et al., 2004). Note that millet contains phytates, which inhibit iron absorption, so the effect of vitamin C may be more pronounced in our biscuits. van der Merwe et al. (2019) reported that enriching pearl millet with moringa powder and baobab fruit pulp provided around three times more iron and zinc than enriched with pearl millet porridge only with a source of pro-vitamin A. Food fortification with ascorbic acid represents an important strategy to improve the iron status of individuals and populations (Teucher et al., 2004).

Fig. 5.

Compositions of the nine biscuit formulations. (A) Nutritional composition of the biscuit formulations expressed as the mean±standard deviation of triplicate determinations. Different letters indicate significant differences according to Fisher’s least significant difference test (P<0.05). (B) Contour plot showing the predicted iron content (mg/100 g) as a function of the proportions of wheat, corn, and millet within the experimental region.

The energy value of the produced biscuits largely met the Codex standard (Fig. 5A), which states that the energy value of processed cereal-based foods should not be less than 3.3 kJ/g (80 kcal/100 g). At the same time, the protein content of our biscuits complied with the Codex Alimentarius (2006), which sets the protein content of biscuits at ≤5 g/100 kcal. Similarly, the lipid content was in line with the recommendations of ≤3.3 g/100 kcal.

Taken together, these results indicate a trade-off between nutritional quality and sensory acceptability. While increasing the proportion of millet flour enhanced the iron content of the biscuits, it also reduced sensory acceptability, highlighting the importance of formulation optimization to achieve an appropriate balance between these attributes.

Biscuit recipe optimization

To find the best formula, optimization was performed based on the results of the hedonic test and the iron content of the different formulations. The cost of the biscuit was not significantly impacted by the formulation and thus not considered in the optimization process. The objective was to identify the best formulation to ensure a high iron content (≥4 mg/100 g) with high consumer acceptability (≥7).

The optimization results, as determined with Minitab 17 software, are shown in Fig. 6. The optimal proportion in each value is depicted in red font. Thus, the optimal formulation was found to contain wheat, flour, and millet at 26.00, 18.17, and 7.82 g, respectively. The wheat flour content was maximized, as wheat ensures the best texture and color. Corn was at a medium level, which enhanced sweetness, and millet was at a low level, which supplied the iron. For a complementary food, acceptability is a key factor for consumption and is essential to guarantee the desired nutritional effect.

Fig. 6.

Representation of the results of the optimization process as determined with Minitab software. D, desirability global; d, desirability of each parameter; targ, target value; y, optimal value.

Thus, sensory quality must be taken into great consideration during the formulation process. The results showed that the presence of a small amount of millet did not adversely affect acceptability. The color of biscuits containing a small proportion of millet was superior to that of biscuits without millet and those with a high millet content. In addition, the incorporation of millet helps to increase the iron content of the biscuits. However, although corn is not rich in iron, it seems to be preferred by consumers over millet.

The optimal formulation, therefore, has a normalized hedonic response y (0.58) and an iron content y (0.24). Using Eq. 7, the optimal formulation was an iron content of 4 mg/100 g and hedonic score of 6.97. The optimization was satisfactory because the composite desirability (0.98) was very close to 1. The optimization objective was achieved, as the resulting formulation was rich in iron, while offering a satisfactory organoleptic appreciation. An iron content of 4 mg/100 g corresponds to current recommendations for supplementation, while a hedonic score of 7 indicates that consumers find the formula pleasant.

The results of this study provide a rigorous approach to develop biscuit formulations based on local ingredients. The use of a mixture design for formulation and a BIB design for sensory analysis enables the study of a large set of formulations while minimizing bias. A trade-off was observed between nutritional quality and sensory acceptability, as increasing the proportion of millet flour enhanced the iron content of the biscuits but reduced consumer appreciation. Nevertheless, the optimization resulted in a formulation that combines high iron content with satisfactory sensory appreciation. The acceptability of biscuits to children is a key factor in encouraging regular consumption, which in turn guarantees the effectiveness and sustainability of supplementation. This research shows that it is possible to develop products that are both nutritionally rich and appreciated by consumers using local ingredients. This approach may contribute to the valorization of local agro-resources and support efforts to improve dietary quality. The optimal biscuit formulation identified in this study will be evaluated in future community-based studies involving school-aged children. Further investigations, including in vitro and in vivo assessments of iron bioavailability, are needed to better understand the nutritional potential. The findings of such studies will provide evidence regarding the potential contribution to improve iron nutrition in school settings.