Sensory acceptability of biofortified foods and food products: a systematic review

Abstract Context It is important to understand the sensory acceptability of biofortified food products among target population groups if biofortification is to be realized as a sustainable strategy for mitigation of micronutrient deficiencies, able to be scaled up and applied through programs. Objective This systemic review aims to summarize and synthesize the sensory acceptability of conventionally bred iron-, zinc-, and provitamin A–biofortified food products. Data Sources MEDLINE (PubMed), AGRICOLA, AgEcon, CABI Abstracts (Web of Science), and organizational websites (eg, those of HarvestPlus and CGIAR and their partners) were searched for relevant articles. No access to any market research that may have been internally conducted for the commercial biofortified food products was available. Data Extraction This review identified articles measuring the sensory acceptability of conventionally bred biofortified food products. Extraction of the hedonic ratings of food products was performed. Data Analysis An “Acceptability Index %” was defined based on hedonic scoring to determine an overall rating, and used to categorize biofortified food products as “acceptable” (≥70%) or “not acceptable” (<70%). Additionally, this review narratively synthesized studies using methods other than hedonic scoring for assessing sensory acceptability. Conclusions Forty-nine studies assessed the acceptability of 10 biofortified crops among children and adults, in mostly rural, low-income settings across Africa, Latin America, and India; food products made from mineral and provitamin A–biofortified food products were generally acceptable. Compared with studies on provitamin–A biofortified food products, few studies (1 to 2 each) on mineral-enhanced crops such as rice, cowpeas, lentils, and wheat were found, limiting the generalizability of the findings. Similarly, few studies examined stored biofortified food products. Few commercial food products have so far been developed, although new varieties of crops are being continuously tested and released globally. Certain crop varieties were found to be acceptable while others were not, suggesting that particular varieties should be prioritized for scale-up. Determining sensory acceptability of biofortified food products is important for informing programmatic scale-up and implementation across diverse populations and settings.


INTRODUCTION
Biofortification, ie, the process of increasing the concentrations and bioavailability of essential nutrients in staple crops by conventional plant breeding, agronomic techniques, and genetic engineering, is a promising approach to combating micronutrient deficiencies in at-risk populations around the world, estimated to affect 1 in 3 women of reproductive age and 1 in 2 preschool-aged children globally. 1Ultimately, biofortification has the potential to serve as an economically and environmentally sustainable means of contributing to addressing the burden of micronutrient deficiencies at a population level via already existing food systems.
To be fully realized as a sustainable solution for mitigating micronutrient deficiencies, it is important to understand how well the sensory characteristics of biofortified foods and food products are accepted among target population groups, particularly in comparison to substitute foods made from nonbiofortified conventional crop varieties.Staple crops commonly targeted for biofortification are often traditionally consumed in the diets of many populations and can include sweet potato, pearl millet, wheat, lentils, cassava, rice, beans, and maize. 2 However, the process of biofortifying foods may lead to visible or sensory changes in the conventional crop varieties and the resulting foods (eg, crops with greater provitamin A content are likely to be yellow or orange in color compared with their conventionally white counterparts, and biofortified orange sweet potato (OSP) is likely to be mushy or soft compared with white sweet potato 3 ).Previous reviews have summarized the acceptability of biofortified foods in lowand middle-income countries. 3,4These reviews found that, broadly, biofortified foods were acceptable among consumers.However, the findings were limited by the lack of biofortified food dissemination and availability in the populations being studied and therefore partially relied on "hypothetical acceptability", ie, based on interviews to ascertain the customer's own perceived likelihood of accepting a biofortified food after hearing a description of that food, rather than based on directly comparing the acceptability of biofortified foods and nonbiofortified foods through empirical evaluation (hedonic testing, eg, actual tasting and consumption).Further, these reviews did not always record which biofortified food varieties were examined, though it was noted that the variety tested was sometimes an important factor in determining the acceptability of the sensory characteristics. 3Understanding the sensory acceptance of biofortified foods among target populations, ascertained through more direct methods, comparing both the sensory acceptance of nonbiofortified foods with that of biofortified foods, and comparing that of biofortified crop varieties, is important for informing food product development and potentially which varieties to emphasize for growing and commercialization.As a note, examining the acceptability of adopting biofortified crops by farmers, and their uptake, is beyond the scope of this review and will be covered separately. 5he objective of this review was to determine and summarize the sensory acceptability of food and food products made from conventionally bred iron-, zinc-, and provitamin A-biofortified staple crops, measured using hedonic scoring, or other methods, in which the biofortified food product was directly tested and compared against the same foods made with (a) other varieties of that biofortified crop or (b) nonbiofortified foods.

METHODS
The protocol for this review was registered on PROSPERO (ID: 254461, no.CRD42021254461), the international prospective register of systematic reviews of the University of York and the National Institute for Health Research, on June 11, 2021. 5

Inclusion and exclusion criteria
Our eligibility criteria are summarized in the PICOS (participants, interventions, comparisons, outcomes, and study design) format in Table 1.

Participants
Any human population was considered eligible for inclusion, including populations of infants, children, and adults.

Interventions
Included studies utilized biofortified crops-based foods and food products, including those that have undergone processing post-harvest, that have been delivered in the form of food products (as defined by trialists).Crops included those biofortified by conventional plant breeding approaches.Interventions utilizing agronomic biofortification methods, genetic engineering-based biofortification methods, or animal-based biofortified foods (such as dairy products or meat from animals that consumed biofortified feed) were excluded from this review.Additionally, protein-biofortified crops such as quality protein maize (QPM) were excluded, allowing a focus on micronutrient biofortification.

Comparators
Comparators included either (a) a nonbiofortified [ie,  control] version of the same food or a food product made using nonbiofortified crops; or (b) food products industrially fortified with the same micronutrient.

Primary outcomes
1. Sensory acceptability: sensory parameters (eg, taste/ flavor, smell/odor, appearance/color, texture/mouthfeel, overall, or as defined by trialists), as assessed using a hedonic scale including the facial hedonic scale.

Secondary outcomes
2. Other scales: a.Just About Right (JAR) b.Children's food intake (amount of food consumed) c.Quantitative Descriptive Analysis (QDA) method d.Paired preference test

Study designs
This review included studies wherein biofortified cropbased foods were tested empirically by consumers, panelists, or households, and did not include studies in which the results were derived from modeling or hypothesized judgments.For example, the analysis included only sensory acceptability studies in which the food or food product was directly tasted and/or consumed by participants to inform their hedonic scale ratings.

Methods for evaluating acceptability
Sensory acceptability.This section briefly describes the methods for analyzing the sensory acceptability of food products, including hedonic testing and variants thereof, ie, JAR analysis, grams eaten, and paired preference tests.The reader is referred to a previous review for a detailed description of sensory evaluation methods and hedonic testing methods. 4donic scale.Hedonic testing measures the degree to which a consumer likes, accepts, or prefers a given product. 4Scales can range between 1 and 5, 7, 9, or higher, with a 9-point scale being the most common.In a 9-point hedonic scale, there is a midpoint of 5 (considered neutral or neither like nor dislike), with 4 positive and 4 negative categories or verbal anchors per side, ranging from "dislike extremely" to "like extremely." 6After participants test the food product, they indicate which number represents their opinion.This simplicity means the testing is accessible to a large population and does not require extensive participant training before use.Additionally, this method can be used for testing overall acceptability in addition to more specific sensory characteristics (eg, smell, texture).In 1 study, among children over 5 years old, 9-point scales discriminated better than 7-point scales. 7However, due to the small number of sensory parameter categories (usually 4 or 5) and end-point avoidance, ceiling effects may occur. 8,9This method is most reliable with sample sizes of n 60 participants, so it is not appropriate for smaller-scale studies. 10Additionally, for children, it may be more useful to use a verbal liking scale and using the terminologies of Peryam & Kroll (P&K)instead of "like extremely" and "dislike extremely", the terms "super good" and "super bad" are used, respectively. 7Additional details, advantages, and limitations of hedonic scale testing are discussed in a previous review. 8cial hedonic scale.A facial hedonic scale, or modified category scale, may be used instead of the traditional hedonic scale when surveying populations in which illiteracy is prevalent or among children.The scale includes text and pictures of faces with various expressions (emoticons) to show the range in acceptability from "dislike very much" to "like very much".After tasting the food product, participants select the emoticon that represents their liking of the sample.The range of this scale varies from study to study, but the most common range is 5 points.In cases where specific sensory attributes are being measured (eg, smell, texture) in lower-literacy populations, it is important that researchers educate the participants on the attribute before testing.Additionally, it may be helpful to support results from the facial hedonic scale with the use of another method, such as a paired preference test, which is suitable for semiliterate and illiterate populations. 11st-About-Right sensory analysis.The JAR scale is a bipolar measurement using 2 semantically opposite anchors at either end of the scale. 12The center point is called "Just About Right" or "Just Right" and assumed to be a participant's ideal level, and the product may be evaluated as deviation from this ideal level, for example, "Too Little" or "Too Much". 12The JAR scale typically targets more specific sensory characteristics than a traditional hedonic scale (eg, fermented odor, crumbliness).A weakness of this method is that it requires 3 decisions on behalf of the consumer: (1) perception of the intensity of an attribute; (2) the location of the consumer's optimal point; and (3) comparing the difference between perceived intensity and this ideal point. 12ams eaten (child acceptability).4][15] In this method, children are given a weighted sample (additional servings are available ad libitum) and are fed until the child refuses food.The remaining food is weighted to calculate the amount consumed.A greater quantity of food consumed correlates to a higher acceptability and vice versa.Feeding studies have found that young children's acceptability by food intake may be more reliable than adult's measurements, which may be biased. 13With this method, it is important that the mothers and children are randomized separately, and that the mothers eat after their children to avoid influencing the child's acceptability.
Quantitative Descriptive Analysis.QDA is a method based on the understanding that humans are better at perceiving relative sensory differences than they are at perceiving absolute differences. 9Unlike other methods discussed, QDA requires a smaller panel of 10-12 trained individuals.A line scale of about 15.24 cm goes from left to right in increasing intensities (eg, weak to strong). 9Panelists score food products using the scale, ensuring a relative, rather than absolute, measurement. 11The results of this method are traditionally presented graphically in a "spider web." 9 Paired preference test.Young children's acceptability of food products cannot be measured using standard methods due to illiteracy and reduced comprehension.However, a paired preference test has been found to be reliable when testing children older than 2 years. 16In this test, children taste a sample of a control food and a sample of the modified food.They then indicate which sample was preferred.This method is useful for including very young children and semi-literate/illiterate populations in studies, but its simplicity also means that the degree of acceptability cannot be measured.Thus, comparisons between food products are less reliably made.

Literature search
A search of relevant literature databases was conducted to include: MEDLINE (PubMed), AGRICOLA, AgEcon, and CABI Abstracts (Web of Science), and organizational websites (eg, Harvest Plus, CGIAR, and partners).As a preliminary assessment of the literature on biofortification, a broad search was conducted in MEDLINE (PubMed) on March 29, 2021, using the following key terms: "Biofortification"[MeSH] OR "biofortif*"[tiab] OR "bio-fortif*"[tiab].This resulted in 1434 results.After screening these results and ascertaining key words to use for increasing the sensitivity of the search, the team conducted searches in additional databases, using broader or narrower searching depending on the topic focus of the database.These searches, including the original MEDLINE search, are summarized in Table 2. Organization websites were also handsearched (Table 3).Additionally, 1147 potential citations outside of the original search were identified during the screening process.These included studies that were: cited in review papers but did not include variations of the term "biofortification" in their abstracts; not indexed in any of the literature databases described above and were thus missed by the original search;

Data synthesis and analysis
For each study that used the hedonic scale or facial hedonic scale method, the mean (SD) or median (IQR) hedonic rating for each parameter (eg, taste, aroma, etc.) for each biofortified food was recorded.The mean rank score was converted to a percentage of the sum of the number of hedonic scale ranks to standardize this outcome across studies with different scale lengths; for example, a mean score of 4.00 out of a 5-point hedonic scale, or a 7.20 out of a 9-point hedonic scale, would both convert to 80% acceptable.This percentage is known as the "acceptance index," as described previously. 17 cutoff of 70% was used to classify each hedonic scale parameter as having good sensory acceptance, according to previous work. 17,18This review considered a product to have good sensory acceptance overall if at least half of the sensory parameters had an acceptance index of 70%.
If biofortified foods are acceptable compared with nonbiofortified foods, or if both types of foods result in 70% acceptability indices, then biofortified foods were considered to be adequately acceptable. 14,15or studies that used the hedonic scale but reported data as the number of participants who chose each level of the scale, instead of reporting a mean value, the mean was back-calculated manually by weighting the number of participants by the hedonic value and dividing by the total number of participants.
In general, hedonic scales show the minimum rank (eg, 1) as being the least acceptable, and the maximum rank (on a 5-point scale, a rank of 5) as being most acceptable.Any hedonic scales that used opposite scoring (ie, 1 as most acceptable, 5 as least acceptable) were converted to maintain consistency in our review.
Studies that used sensory acceptability assessment methods other than a hedonic scale were synthesized narratively.

RESULTS
For the 4 review topics, a total of 5141 records (Figure 1) were identified.Overall, 305 eligible records were found across the 4 review topics outlined previously.Excluded studies did not report on sensory acceptability.For this review topic, 49 studies (63 reports) on acceptability across 10 types of crops were included: 14 studies on OSP, 13 studies on maize, 6 studies on beans, 8 studies on cassava, 5 studies on pearl millet, 2 studies on rice, 1 study on cowpeas, and 1 Figure 1 Flowchart of search and selection process. 82trition Reviews V R Vol. 82 (7):892-912 study that examined both separately and in combination OSP, pearl millet, lentils, and wheat.Nine studies were done in South Africa, 9 in Nigeria, 6 each in India and Brazil, 4 in Colombia, 2 each in Rwanda, Uganda, and sub-Saharan Africa, and 1 each in Bolivia, Burkina Faso, Ghana, Guatemala, Malawi, Mozambique, Panama, Tanzania, Kenya, and Zambia.All controls used in these studies were nonbiofortified versions of the same crop; no studies examined industrially fortified crops as comparators.

Summary of overall results
The most frequently used hedonic scale in this review was the 9-point hedonic scale, although 4-point, 5point, and 7-point hedonic scales were also used (Tables 4-10, see Tables S1-S10 in the Supporting Information online).All scales are represented by "dislike extremely" at the extreme minimum (1) and "like extremely" at the maximum (either as originally constructed or as converted by us).Several studies used methods other than hedonic scale testing for assessing sensory acceptability, such as JAR analysis, weighed intakes, QDA, and a paired preference test.
Biofortified OSP vs nonbiofortified OSP.Chutney, rasam, thick shake, barfi, payasam, and sharbat made from either Kamalasundurai or WSP were acceptable among breastfeeding mothers in rural India. 24Several studies across Africa found boiled OSP (A45, Chipika, Kadyaubwerere, Zondeni, Karote DSM, Resisto) or the control (A40, Kenya, Polista, Sinia B) were acceptable. 19,25,27Chips or crisps made with Kemb 10 or Dutch-Robyjin -a regular potato with similar moisture content to that of sweet potato -using either corn oil or palm oil were all acceptable. 26One study examined acceptability of boiled sweet potato using a 9-point hedonic scale; however, quantified results were not reported. 20,21ofortified OSP: variety or preparation methods comparison.In consuming boiled OSP, particular varieties (187017-1, 420027, and an unreported OSP variety) were less acceptable than other varieties (see Table S1 in the Supporting Information online). 27,30Boiling without skin for 15 minutes was less acceptable than 10 minutes, or boiling for either 10 or 15 minutes with skin. 32Frying OSP was acceptable. 32Juiced OSP alone was more acceptable than juiced OSP with pineapple juice added. 23An unreported OSP variety was roasted or prepared as tamales, pesada, or soda; all were acceptable. 23Baked goods made with varying amounts of Beauregard flour, including cake (36 g to 64 g flour/ 100 g plus oil [1.3 g/100 g-12.7 g/100 g]), 34 bread (20%-60% flour), 31 and cookies (extruded or non-extruded flour) were all acceptable. 28ofortified OSP: other sensory acceptability methods.In western Kenya, n ¼ 501 caregivers of children under 5 years of age or pregnant women used a 5-item JAR scale (including anchors of "much too little," "too little," "just about right", "too much", and "much too much") to rate acceptability of the VITAA varietal of OSP, prepared by boiling. 35,36Several sensory attributes (including sweetness, smell, color, texture/softness, taste, and crumbliness) were compared between VITAA and nonbiofortified white or yellow sweet potato.Generally, the majority of participants rated each OSP attribute "just about right", with the exception of an even split of 68% of participants rating "smell" as being "too little" or "just about right" (34% each).
One study in a population of 15 trained panelists in Brazil used the QDA measure to assess the sensory acceptability of OSP (variety: Beauregard) processed into chips, then stored in various conditions. 37In brief, storage with nitrogen, in either polyester/aluminum foil/low-density polyethylene (PET/Al/LDPE), biaxially oriented polypropylene (BOPP)/metallized (met) BOPP  without an oxygen scavenger, and the latter with an oxygen scavenger showed less significant sensory alterations (in flavor, odor, color, crispness) during 207 days of storage, than chips packed without nitrogen in BOPP/metBOPP after 153 days or chips packaged with nitrogen in PETmet/LDPE for 184 days.The QDA method entailed using a 9-cm nonstructured scale ranging from 0 to 9, with a score of 4.5 defined as product rejection (therefore, scores closer to 0 were considered better sensory acceptance).The extremes of the scales for each attribute were: color (0 ¼ intense orange, 9 ¼ light yellow); odor (0 ¼ characteristic, 9 ¼ odd); oxidation odor (0 ¼ absent, 9 ¼ strong); flavor (0 ¼ characteristic, 9 ¼ not characteristic); oxidation flavor (0 ¼ absent, 9 ¼ strong); crispness (0 ¼ crunchy, 9 ¼ limp); overall quality (0 ¼ excellent, 9 ¼ dreadful).This study was able to determine the best storage conditions for OSP chips to minimize oxidation and optimize palatability.
Biofortified maize vs nonbiofortified maize.The fermented beverage, Amahewu, was not well accepted when made with either biofortified or control maize, 42 but adding roasted Bambara flour resulted in acceptability in both biofortified and nonbiofortified forms. 43hutu, a crumbly maize porridge, was acceptable when made from either PVA pool A, PVA pool H, or the control white maize, 41,51 but was not acceptable when prepared from KP-78, KP-79, or KP-77. 51Serving PVA Aor control WE-3172-based phutu alongside chicken curry or cabbage curry was acceptable, but PVA A-based phutu served with bambara groundnut curry was not acceptable, while control WE-3172-based phutu served with the same curry was acceptable. 25][46] Nshima porridge made from biofortified orange maize was less acceptable than nshima porridge made from an also non-acceptable isogenic white maize, or an acceptable local white maize, 38 but nshima made from prototype high provitamin A orange maize, or 2 controls, yellow and white maize, were all acceptable. 50Thin porridge and samp (coarse corn meal) made from KP-78, KP-79, or KP-77 was less accepted than the same dishes made with the white maize control SC-701. 51Finally, porridges made from various proportions of fermented A9895032 maize flour or common yellow maize were all acceptable. 39ofortified maize: variety or preparation methods comparison.Maturation time of the maize appeared to influence the acceptability of 8 boiled maize varieties (see Table S2 in the Supporting Information online), with younger maize (harvested 20 days after pollination) being accepted across all 8 varieties. 40Only 2 out of the 8 varieties were still acceptable at 27 days pollination, while only 1 was acceptable at 34 days after pollination.Maize snacks made with 5 varieties of extruded maize with or without amaranth leaf powder scored consistently low for sensory acceptability. 48ofortified maize: other sensory acceptability methods.
Finally, 1 study that reported hedonic scale results among secondary school children (n ¼ 54) and adults (n ¼ 50) also used a paired preference test among these groups as well as among preschool (n ¼ 52) and primary school children (n ¼ 56) 51 to measure sensory acceptance of biofortified maize made into phutu, thin porridge, and samp in KwaZulu-Natal.Among preschool and primary school children, the biofortified maize variety KP-79 and the control, a commercial white maize (SC-701), were tested.Preschool children preferred the biofortified versions of all 3 foods.Conversely, primary school children had no preference for phutu and samp, but preferred thin porridge made with control maize.In testing the additional biofortified varieties, KP-78 and KP-77, secondary school children and adults strongly preferred control white maize relative to the biofortified yellow varieties.
Iron-biofortified beans.ained panelists in rural subSaharan Africa, N ¼ 10 Biofortified beans vs nonbiofortified beans.Bean varieties BIO-101 and BIO-107 or a local control, cooked and served with vegetable sauce, were acceptable. 53The ROBA1 or control K131 varieties, made into porridge or ground and made into a bean sauce using different flour processing methods were acceptable. 616][57][58][59][60] Finally, rice with cooked beans, including SMR 4, SMN 18, SMC 14, or SMB 17, or control Testigo, were acceptable among children, but not among adults. 52ofortified beans: variety or preparation methods comparison.Variety SMN 18 made into cookies with either 15% or 20% bean flour was acceptable. 54o-vitamin A-biofortified cassava.A total of 8 studies reported on the sensory acceptability of provitamin Abiofortified cassava (Table 7, [65][66][67][68][69][70][71][72] see Table S4 in the Supporting Information online).Seven studies were conducted on populations in Nigeria, and 1 study was conducted on a population in Brazil.All studies were undertaken in rural settings.All the studies had populations consisting of adults, including university staff, graduate students, and semi-trained panelists.In this study the food product was referenced as "Xhima." Additionally, all studies except for 1 65  Biofortified cassava vs nonbiofortified cassava.From a non-peer-reviewed pre-print, bread made with 20% of either TMS 01/1368 cassava or control TME 419, and 80% wheat were accepted. 70Pasta made from TMS 07/ 0593 was more accepted than pasta made from control white cassava. 72Eba made from TMS 01/1368 or TMS 01/1371, or undisclosed very deep yellow or light yellow cassava, were all acceptable, as was eba made from control white cassava. 65,71Gari made from very deep yellow or light yellow cassava, and fufu made from TMS 01/ 1368 or TMS 01/1371, and their control cassava counterparts, were all acceptable. 65,71Fufu was also acceptable, made from either TMS 01/1368, TMS 01/1371, or control TME 419. 71ofortified cassava: variety or preparation methods comparison.Chips made from BRS Dourada, BRS Gema e ovo, BRS Jari, or Hybrid 2003 1411 cassava with the addition of onion and parsley flavoring showed greater acceptability than similar chips without the flavoring. 66Eba and gari paste made from TMS 1358 were acceptable when fermented for 0 days-4 days; as a note, at 4 days, hedonic scale parameters moldability and overall were rated as not acceptable. 67Gruels with varying proportions (60%-100%) cassava starch, and either 0%-40% partially defatted soybean flour 68 or 2%-12% whole egg, 69 were also all acceptable.
Biofortified cassava: other methods for determining sensory acceptability.Two studies in this review incorporated the JAR sensory analysis. 35,36,71In Nigeria, West African food products, ie, eba and fufu, were made from either biofortified crops or the control (ie, the nonbiofortified conventional crop or industrially fortified conventional crop-based products and/or other industrially fortified products eg, fortified oil) cassava and tested for sensory acceptability using a 3-level JAR scale among n ¼ 122 adult consumers. 71The attributes included color (being too white, JAR, or too yellow); fermented odor (too weak, JAR, too strong); texture (too soft, JAR, too hard).Biofortified varietals included TMS 01/1368 and TMS 01/1371, while the control conventional cassava was TME 419 processed with or without red palm oil to fortify it with vitamin A. In general, consumers were satisfied with the color, smell, and odor of the biofortified cassava-based eba and fufu, and the authors noted that the addition of red palm oil in the fortified foods may have caused an untoward odor and softened the texture, leading to greater liking for biofortified and nonfortified cassava food products.
To ˆand pancakes made with decorticated or whole pearl millet varied in acceptability: to ˆwas only accepted made with control (local Gampela) pearl millet, while pancakes were consistently acceptable when made with variety GB 8735, either whole or decorticated, but only acceptable when variety Tabi was decorticated. 76Bhakri made from Dhanashakti or a local variety 75,77 ; idli, mudde, and porridge, 24 and several dessert items (cookies, laddu, sheera, cake, nankhatai, sweet porridge, and puranpoli) and savory items (khichdi, upma, dhokla, idli, vegetable cutlet, kothimbir wadi, thempla, pav bhaji, pakoda, and vada), 14,15 made from Dhanashakti or control DG 9444 varieties were all acceptable.Finally, cookies made from either AHB or control MRB pearl millet were acceptable, either in a 70:30 or 60:40 ratio of whole wheat flour and germinated pearl millet flour. 73ll other biofortified pearl millet food products were acceptable.
Biofortified pearl millet: variety or preparation methods comparison.Khichdi made from Dhanashakti was found to be acceptable. 74iofortified pearl millet: other sensory acceptability methods.Two studies measured young (6-month-old-24month-old) children's acceptance of biofortified pearl millet through biofortified pearl millet intake amount in rural Southern India 24 and in the urban slums of Mumbai 14,15 (see Table 8 for varieties).Children were given a weighed sample (additional servings were available ad libitum) and were fed until the child refused food.The remaining food was weighed to calculate the net amount consumed, with a greater quantity of food consumed correlating to a higher acceptability of that food.In general, there were no differences in the amount of consumption of foods prepared with the biofortified crops or with control crops across both studies, indicating the acceptability of food products made with biofortified pearl millet among children.
Zinc-biofortified rice.In total, 2 studies reported on the sensory acceptability of biofortified rice (Table 9, 78,79 see Table S6 in the Supporting Information online).The countries surveyed included Bolivia, Columbia, and India.The biofortified varieties included PCT-25-  In the study Yareshimi 2013, mean sensory scores of cooked biofortified rice varieties are given in Figures 1 and 2, but it is not clear how these means were calculated.It appears that a "Figure 3" is missing from the paper, which may contain each parameter's individual score.We have included the mean score here for reference.Abbreviations: NR, not reported.
C2-329-4-2SR-5P, CT22154-9P-1SR-1P-3SR, BF14AR021, BF14AR035, Karidaddi, Makam, IVT (SHW) 91, Badshahbhog, and BI 43; the control nonbiofortified varieties included local varieties MAC-18, CICAA4, and BI 33.The only age group represented was adults, and the only processing method included was cooked, which was found to be acceptable in 1 study, 79 while acceptability results were not fully reported in the other study. 78on cowpeas, iron lentils, zinc wheat, and combinations of multiple crops.Cowpeas, 80 lentils, 24 and wheat 24 were investigated in 1 study each (Table 10, 24,80 see Tables S7-S10 in the Supporting Information online).The countries surveyed included Brazil (cowpeas) and India (lentils, wheat).The age groups represented were adults (in Brazil) and breastfeeding infants and their mothers (in India).The lentil and wheat data came from the same study.Biofortified lentils (biofortified: Pusa Vaibhav; control: Moitree) made into sambar were acceptable, as was biofortified wheat (biofortified: BHU-6; control: HD2967) made into porridge. 24owever, the study examining cowpeas (variety BRS Xiquexique) did not include a comparison with a control nonbiofortified crop in the methodology; instead, cheese bread made with either 5.6% or 8.0% cowpea flour were both found to be acceptable.One study compared multiple biofortified crops-based dishes with the same dishes made with nonbiofortified crops (see Table S10 in the Supporting Information online), including pulao, kesari, and poli; all were rated acceptable. 24

Main findings
This systematic review summarized the results of testing sensory acceptability directly in target populations across 10 conventionally biofortified food products, including those made from OSP, maize, beans, cassava, pearl millet, rice, cowpeas, lentils, wheat, and combinations of multiple biofortified crops into a single composite meal.From the current evidence base, the findings suggest that foods and food products made from biofortified crops generally had acceptable sensory characteristics, reaching hedonic score-based acceptance indices of at least 70% and showing sensory acceptability using other methods, across children, adolescents, and adults.
The general acceptance of most biofortified foods among consumers in this review is consistent with findings on the acceptance of biofortified OSP, maize, cassava, pearl millet, beans, and rice in previous reviews on biofortified crop acceptance in lower-to-middle income countries, 3,4 but these reviews differed in that they did not examine lentils, wheat, cowpeas, or composite meals, likely due to lack of available data.Furthermore, these reviews differed in that they included all types of biofortified crop (ie, those biofortified by not only conventional breeding, but also agronomic methods and genetic engineering); included the result of hypothetical acceptance from customers' own perceptions of the crop but not direct experience with the crop; and did not distinguish the biofortified varieties tested.

Reasons for non-acceptance of certain biofortified foods and recommendations
Only a few studies found that certain biofortified foods had an overall "not acceptable" rating, which may aid further development of that food product, such as changing the additional ingredients that are added to the food product.For example, in testing OSP juices, adding pineapple juice for sweetness resulted in lower acceptability than adding plain OSP juice; for maize amahewu, adding wheat bran or malted maize inoculum was not acceptable, but amahewu made with roasted Bambara flour was acceptable, suggesting multiple formulations of a given food product may need to be tested to determine acceptability.Chips made from cassava showed improved acceptability with the addition of onion and parsley flavoring, which may be a factor for manufacturers to consider in designing processed foods.
Other methods for aiding development of the food product may involve adjusting the methods of growing, preparation, or storage.For example, OSP boiled with or without skin for 10 minutes or with skin for 15 minutes was acceptable, but OSP boiled without skin for 15 minutes was not acceptable.Maize that was allowed to pollinate for 20 days and boiled was more acceptable than maize that was allowed to pollinate for 27 days-34 days and boiled.For pearl millet, decorticated grains were more accepted than whole-grain versions for some food products (gruel); other foods, such as pancakes, were found to be unacceptable using either decorticated or whole grains.For storing OSP chips, particular packaging is needed to maintain optimal sensory characteristics and acceptability after storage for several months and to minimize undesirable flavors and odors due to oxidation.These findings may also aid developers in finding the ideal growing, preparation, and storage techniques.
Particular varieties of OSP, maize, and beans were more acceptable than others, indicating that producers and processors should select only varieties shown to be more acceptable by consumers.For OSP and maize, this was apparent when the crop was boiled whole, while for beans this was shown only when beans were made into porridge or sauce.Conversely, the studies included did not indicate variety-informed sensory acceptability for whole boiled cassava.In preparing boiled OSP, 2 out of 15 varieties tested were not acceptable, suggesting that variety may be an important consideration for developing food products for particular processing techniques.
Bias against certain sensory attributes may also explain lower acceptance for certain foods.For example, 1 study noted that it is possible, from accompanying focus group discussions, that preconceptions about biofortified maize -for example, yellow maize being associated with animal feed -may bias older individuals towards preferring white maize. 51This study showed that biofortified maize may be acceptable in preschool programs but would require additional strategies (eg, intensive nutrition education programs) to incorporate into the diets of older individuals.

Gaps and limitations
In our review, several crops, including rice, cowpeas, wheat, and lentils, were represented by only 1 to 2 studies each, indicating a gap in the literature, possibly informed by limited dissemination.However, investigating the overall acceptability of mineral-biofortified crops from only a studies limits the overall generalizability; diverse populations and settings likely have variability in sensory acceptability.Other gaps include the populations tested (few studies included preschool children, schoolchildren, or adolescents) and testing the acceptability of composite meals including multiple kinds of biofortified crops, which was done in only 1 study.For example, rice and beans is a commonly consumed meal in Latin America, which is an area where biofortified crops have been introduced 81 ; therefore, this meal could be made using zinc-biofortified rice and iron-biofortified beans and tested for sensory acceptability.Finally, only 1 study appeared to examine the sensory acceptability of a food product that was stored for an extended time period -a major gap given that crops are not always going to be consumed just after harvesting and cooking or otherwise processing, considering seasonality.
While this review found data for 10 types of biofortified crops, the sensory acceptability of other crops currently being developed by HarvestPlus was not found in the evidence base developed from our literature search -for example, vitamin A-biofortified banana and plantains; iron-biofortified irish potato; zinc-biofortified sorghum. 81These are in dissemination, national performance trials for release, or being tested, or released across several countries (depending on crop) 81 ; it will be important to generate data on the sensory acceptability of these crops as an early step to incorporating such crops.

CONCLUSIONS
This review adds detailed data on sensory acceptance of biofortified food products and recommends potential paths for processors of biofortified crops to help improve acceptability.Generally, biofortified crops were well accepted.Lack of acceptance could be attributed to specific ingredients added to the food product, to particular preparation methods, or to biased perceptions regarding the food.Several research gaps remain, including sensory acceptability studies in diverse populations and settings on: biofortified rice, cowpeas, lentils, and wheat; combinations of multiple biofortified crops into composite meals; stored biofortified food products; and foods made from newer biofortified crops currently being introduced across the world.Studies evaluating new biofortified crops, crop varieties, and food products should include a formal sensory acceptability evaluation where feasible, to inform programmatic scale-up and implementation and ensure success across diverse populations and settings.Development (BMZ) and the Netherlands Ministry of Foreign Affairs for the Commercialisation of Biofortified Crops programme, as the funders, had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.Declaration of interest.S.M. holds equity in a diagnostic start-up focused on developing assays for low-cost and point-of-care measurement of certain nutrients from a drop of blood, using results from his research as a faculty member at Cornell University.GAIN is a not-forprofit organization supporting and promoting biofortification programs; V.M.F., M.N.N.M., E.M., and A.M.N. are employees of GAIN.All other authors have no relevant interests to declare.

Supporting Information
The following Supporting Information is available through the online version of this article at the publisher's website.
Table S1 Sensory acceptability for biofortified orange sweet potato Table S2 Sensory acceptability for biofortified maize Table S3 Sensory acceptability for biofortified beans Table S4 Sensory acceptability for biofortified cassava Table S5 Sensory acceptability for biofortified pearl millet Table S6 Sensory acceptability for biofortified rice Table S7 Sensory acceptability for biofortified cowpeas Table S8 Sensory acceptability for biofortified lentils Table S9 Sensory acceptability for biofortified wheat Table S10 Sensory acceptability for multiple biofortified crops made into a composite meal Table S11 PRISMA Checklist

Table 1
PICOS criteria for inclusion of studies

Table 2
Search strategy across included databases

Table 3
Results from hand-searching organization websites

Table 4
Summary of studies investigating sensory acceptability of provitamin-A biofortified orange sweet potato (OSP) via hedonic score testing
b Control refers to a nonbiofortified, non-industrially fortified, conventional crop.

Table 5
Summary of studies investigating sensory acceptability of provitamin A-biofortified maize via hedonic score testing

Table 5 Continued
a Foods were considered acceptable if they had an overall sensory acceptability score of 70%.b Control refers to a nonbiofortified, non-industrially fortified, conventional crop.c

Table 6
Summary of studies investigating sensory acceptability of iron-biofortified beans via hedonic score testing a Foods were considered acceptable if they had an overall sensory acceptability score of 70%.b Control refers to a nonbiofortified, non-industrially fortified, conventional crop.

Table 7
Summary of studies investigating sensory acceptability of provitamin A-cassava via hedonic score testing a Foods were considered acceptable if they had an overall sensory acceptability score of 70%.bControl refers to a nonbiofortified, non-industrially fortified, conventional crop.Abbreviations: NR, not reported.Nutrition Reviews V R Vol. 82(7):892-912

Table 8
Summary of studies investigating sensory acceptability of iron-6 zinc-biofortified pearl millet via hedonic score testing bControl refers to a nonbiofortified, non-industrially fortified, conventional crop.

Table 10
Summary of studies investigating sensory acceptability of iron-biofortified cowpeas, iron-biofortified lentils, zinc-biofortified wheat, and composite meals made from combinations of multiple provitamin A-, iron-, and zinc-biofortified crops, via hedonic score testing Foods were considered acceptable if they had an overall sensory acceptability score of 70%.b Control refers to a nonbiofortified, non-industrially fortified, conventional crop.

Table 9
Summary of studies investigating sensory acceptability of zinc-biofortified rice via hedonic score testing c