Review of the facilitators and barriers to adoption of biofortified foods and food products

Abstract

Biofortification – the process of increasing the concentrations of essential nutrients in staple crops – is a means of addressing the burden of micronutrient deficiencies at a population level via existing food systems, such as smallholder farms. To realise its potential for global impact, we need to understand the factors that are associated with decisions to adopt biofortified crops and food products. We searched the literature to identify adoption determinants, i.e. barriers to (factors negatively associated) or facilitators of (factors positively associated) adoption, of biofortified crops and food products. We found 41 studies reporting facilitator(s) and/or barrier(s) of adoption. We categorised the factors using the Consolidated Framework of Implementation Research 2.0, resulting in a set of factors that enable or constrain adoption of biofortified foods across twenty-four constructs and five domains of this meta-theoretical determinant framework from implementation science. Facilitators of orange sweet potato adoption included knowledge about importance, relative advantage, efficient production and management practices; barriers included lacking timely access to quality vines and market remoteness (28 studies total). Facilitators of vitamin A cassava adoption included awareness of its benefits and access to information; barriers included poor road networks and scarcity of improved technology including inadequate processing/storage facilities (8). Facilitators of high-iron bean adoption included farmers’ networking and high farming experience; barriers included low knowledge of bean biofortification (8). Barriers to vitamin A maize adoption included low awareness and concerns regarding yield, texture and aflatoxin contamination (1). These barriers and facilitators may be a starting point for researchers to move towards testing implementation strategies and/or for policymakers to consider before planning scale-up and continuous optimisation of ongoing projects promoting adoption of biofortified crops and food products.

Background

Micronutrient (vitamins and minerals) malnutrition is an important public health problem affecting populations in all regions of the world. According to the most recent estimate, 372 million preschool-aged children (aged 6–59 months) and 1.2 billion non-pregnant women of reproductive age (aged 15–49 years) globally are suffering from at least one micronutrient deficiency⁽¹⁾. Additionally, 50–60% of the anaemia worldwide is due to iron deficiency⁽²⁾ affecting an estimated 27% of the world population in 2013, impacting developing nations at a disproportionately high rate⁽³⁾. Vitamin A and zinc deficiencies are also common and, coupled with undernutrition, have been attributed to approximately 45% of child deaths in 2011⁽⁴⁾. Micronutrient deficiencies have also been implicated in compromised immunity, retarded growth and lost developmental opportunity in children, and accelerated burden of chronic metabolic diseases with accumulated impact on health and productivity of humans. These deficiencies are due to lack of access to physiologically sufficient sources of food and are amplified by poor nutritional quality of crops. Therefore, increasing the consumption of essential micronutrients by the general population and especially of populations with high requirements remains a global priority.

Biofortification, the process of increasing the concentrations and bioavailability of essential nutrients in staple crops by conventional plant breeding and/or genetic engineering, is a promising approach to combatting micronutrient deficiencies⁽⁵⁾. Staple crops commonly targeted for biofortification are typically consumed in the daily diets of many populations and can include sweet potato, pearl millet, wheat, lentils, cassava, rice, beans and maize⁽⁶⁾. Micronutrients commonly enhanced by biofortification techniques are provitamin A, zinc and iron. Ultimately, biofortification serves as a cost-effective (because once biofortified varieties are developed, small costs accrue for maintenance but their seeds are self-replicating, multiplying the potential benefits) and sustainable (farmers may receive the seeds initially but then retain seeds for future crop cycles, a self-sustaining agricultural solution) means to address the burden of micronutrient deficiency at a population level via already existing food systems, such as growing biofortified crops on smallholder farms or through purchasing biofortified food products in markets⁽⁶⁾. To be fully realised as a sustainable solution to mitigate micronutrient deficiency, it is important to understand the factors that are associated with decisions to adopt and cultivate biofortified crops, including both facilitators and barriers to adoption.

Studies on adoption of agricultural innovation have often found that there is a large gap in time (ranging from 14 to 25 years) between availability of an innovation and its widespread adoption⁽⁷⁾. For example, the adoption level of vitamin A-biofortified cassava (VAC) variety was reported to be 35% in 2010 in sub-Saharan Africa⁽⁷⁾ and 39% in 2017 in Nigeria⁽⁸⁾. Vitamin A-biofortified maize (VAM) yields have only a modest gain in sub-Saharan Africa compared with that in Brazil, China and the USA⁽⁹⁾, and varietal replacement has been extremely slow, with farmers still growing 20-year-old varieties⁽¹⁰⁾. High-iron beans (HIB) have seen a 29% adoption level following intensive delivery efforts⁽¹¹⁾ despite a demonstrated significant positive effect of HIB on health outcomes⁽¹²⁾. Indeed, orange sweet potato (OSP) is often cited because of multiple examples of its scaling-up exercises. Since 2009 when the level of OSP adoption was 6.9%⁽⁷⁾, 6.2 million households were reached by July 2019 in fifteen sub-Saharan African countries (with considerable variation between countries). The scaling process of the OSP innovation involved several distinct phases with key factors occurring during each of these phases. Low and Thiele⁽¹³⁾ provided a detailed description of the initiatives aimed at scaling up the promotion and adoption of biofortified orange sweet potatoes. Experiences, findings, implementation challenges and lessons learned from these initiatives have been described previously⁽¹⁴⁾. Across these initiatives, households significantly increased OSP production⁽¹⁵⁾. Huey et al.⁽⁵⁾ in a systematic review described the impact pathway by which these initiatives increased OSP production leading to improved micronutrient status. Despite the many lessons learned, there remain unknowns in terms of implementation best practices and innovations needed to enhance delivery, engagement, uptake and impacts⁽¹⁵⁾. Therefore, there is a need to understand the factors driving crops’ slow adoption. In fact, there has been a call by the US government food security strategy for the identification of pathways for faster scaling in the early phase of research and collaboration among scale-up partners⁽¹⁶⁾. Understanding the factors that enable and inhibit the adoption of biofortified crops is crucial for designing future programmes focused on scaling.

Throughout the past decade, there has been a surge of studies reporting on biofortified crops. Several of them have explicitly examined the factors influencing adoption of OSP in Ghana⁽¹⁷⁾, Nigeria^(17,18) , Mozambique^(19–21), Tanzania⁽²²⁾ and Uganda⁽²⁰⁾. A few other studies evaluated the determinants of adoption of VAC among farmers^(8,23,24) . Vaiknoras et al.^(11,25) evaluated the impact of different delivery approaches on household time to adoption of iron-biofortified beans in Rwanda. While several reviews based on these primary studies have been conducted in the recent past restricting their foci on one country/region and/or one biofortified crop/product^(14,26–29) , these reviews did not use systematic review methods, which result in a synthesis of all available evidence across multiple databases and sources to limit bias in the overall conclusion. Systematic reviews are therefore more comprehensive and objective than reviews that are not systematic, as only some types of evidence may be included in the review. Two formal systematic reviews available with focus on implementation outcomes (e.g. acceptability and adoption) of a biofortified crop-based intervention were published in 2017⁽³⁰⁾ and 2023⁽³¹⁾. The 2023 review summarised the results of testing sensory acceptability across ten conventionally biofortified food products. The review published in 2017 summarised the determinants of both acceptance and adoption⁽³⁰⁾. However, in this 2017 review, studies included predicted the adoption of biofortified crops (i.e. acceptance and adoption remained mostly hypothetical or lab-in-the-field studies, based on choice experiments or experimental auctions) or participants were exposed to the biofortified crops for the first time (i.e. possibly biased findings from one-time exposure rather than actual experiences using of the biofortified crop), not based on real-world adoption or uptake data. In addition, large-scale effectiveness studies of biofortified crops are lacking, and reported adoption studies were mostly conducted under strict intervention conditions. Most of the studies in that review were not carried out among the groups most at risk of micronutrient deficiencies, such as young children and women of reproductive age. To harness the full potential of biofortified crops, it is imperative to understand the actual factors affecting all aspects of adoption and leverage on those that facilitate implementation and scale-up of biofortified crops. To the best of our knowledge, no review has been conducted that provides a comprehensive and consolidated overview of the barriers and enablers of adopting biofortified crops and their products at multiple levels including producers. Using the Consolidated Framework for Implementation Research (CFIR)^(32,33) , a determinant framework from the implementation science discipline, the objective of this review was to summarise the evidence around the factors that positively or negatively impact adoption decisions on biofortified crops and its products among smallholder farmers and consumers across several domains (e.g. individuals delivering/receiving the innovation, the settings/context, implementation processes and the characteristics of the innovation itself). Understanding these factors could support the decision making of multiple actors (e.g. practitioners, researchers, policy makers) in the design, implementation and scale-up of programmes for biofortified crops and their products.

Methods

Inclusion and exclusion criteria

Participants

Any human population was considered eligible for inclusion, and as such, in terms of sampling design, we considered growing farm households who planted, cultivated or marketed biofortified seeds, roots or vines, either provided to them by programmes or those which had been retained from previous years’ crops. We did not limit by region, and households at either local or national scales were eligible.

Interventions

We included studies utilising biofortified foods and food products, including those that have undergone processing post-harvest and that have been delivered as crops only or in the form of food products (as defined by study authors). Crops included those biofortified by conventional plant breeding approaches. We included any type of conventionally biofortified crop.

We did not include studies using agronomic biofortification methods, genetic engineering-based biofortification methods in order to focus the review on conventionally biofortified crops. We also excluded animal-based biofortified foods, such as dairy products or meat from animals that consumed biofortified feed, and plant-based protein-biofortified crops, such as quality protein maize, to focus on micronutrient biofortification.

Comparators

Comparators included either (A) a non-biofortified (i.e. conventional) version of the same crop or food product made using non-biofortified crops, or (B) food products industrially fortified with the same micronutrient.

Primary outcomes

The primary outcome was adoption determinants (i.e. any barrier or facilitator empirically shown or believed to are associated with the adoption of biofortified crops, foods and their products). We defined adoption as households’, farmers’ or other stakeholders’ engagement in planting, cultivating, retaining for future cultivation, or marketing seeds, roots or vines that produce biofortified crops and/or their consumption of such biofortified crops and products made thereof. A barrier was defined as any variable reported be associated with impeding or obstructing the adoption of biofortification interventions. In contrast, a facilitator was considered as any of those variables that were associated with promotion of the adoption of biofortification interventions. Studies that provided usable data for either of these two factors were included. Studies that only reported (with no unit of text relevant to adoption determinants) the effect of adopting biofortification intervention on health/nutrition were not included (e.g. Ref. ⁽³⁴⁾).

Study designs

As barriers and facilitators could be identified either from intervention studies assessing the effectiveness of biofortification interventions and/or associated implementation strategies or observational studies analysing factors of adoption of biofortification interventions, studies using any of the research designs ranging from qualitative to quantitative to mixed methods were eligible. Therefore, any relevant primary study conducted in the T2-2 phase (effectiveness research) and beyond along the translational research continuum (as described by Refs. ^(35–37)) were included. We included primary studies conducted in the T2-1 phase (efficacy research) only if they reported any determinants of adoption in addition to health outcomes.

Literature searches

We performed a search of relevant literature databases including MEDLINE (PubMed), AGRICOLA, AgEcon, CABI Abstracts (Web of Science) and organisational websites (e.g. Harvest Plus, CGIAR and partners) as described previously⁽³⁸⁾.

As a preliminary assessment of the literature on biofortification, we conducted a broad search in MEDLINE (PubMed) on 29 March 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, we conducted searches in additional databases, using broad or narrower searching depending on the topic focus of the database. These, including the original MEDLINE search, are summarised in Table 1. We also hand-searched websites of organisations that play roles in growing or distributing conventionally biofortified crops. The results are included in Table 2.

Table 1. Search strategy across included databases

Table 2. Results from hand-searching organisation websites

We also identified 1147 potential citations outside of the original search 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; or published recently in 2021, which we identified from journals’ table of contents alert feeds. Some of the latter included full-text versions of conference abstracts that were found and included in the original screening pool.

Data screening and extraction

We screened all records for eligibility, first at the title/abstract level and subsequently at the full-text screening level, and used a subset of articles to improve consistency among review authors. We then extracted data for each identified study, according to the following: study level details including authors or research group, study year and location; study design, population targeted, sample size and how crops were biofortified; and outcomes of the study.

We extracted core data relevant to implementation determinants of biofortified crops, foods and their products. Data extracted at this level included, in addition to barriers and facilitators of adoption, measures (definition) of adoption used, methods employed for identifying factors of adoption, and implementation strategies reported (if any) in each study. An inductive approach was taken to extract data on barriers and facilitators (which were later categorised using the CFIR codebook (see https://cfirguide.org/ for more information)⁽³⁹⁾ to allow capturing data segments on implementation determinants that did not fit the current CFIR constructs with potential to modify the CFIR matrix (i.e. testing applicability of CFIR in biofortification interventions). The decision to follow an inductive approach to data extraction was based on the anticipation that the CFIR might not fully reflect the implementation context of biofortification interventions.

Data synthesis and analysis

We conducted a directed (deductive) qualitative content analysis (DQCA)⁽⁴⁰⁾ to identify constructs related to implementation determinants extracted from the included studies. DQCA involves identifying a theoretical framework based on what the synthesis will be executed. This identification involved two steps and resulted in the selection of the CFIR as the theoretical framework⁽³⁹⁾. Initially, a list of candidate frameworks was developed by referring to three sources: (a) the website dissemination-implementation.org that lists theories, models and frameworks identified by Tabak and colleagues⁽⁴¹⁾, (b) Grol et al.’s ‘Planning and studying improvement in patient care: the use of theoretical perspectives’⁽⁴²⁾, and (c) Nilsen’s ‘Making sense of implementation theories, models and frameworks’⁽⁴³⁾. Then, candidate frameworks were rated against three criteria (familiarity, applicability and usability) online using the Theory Comparison and Selection Tool (T-CaST) (available at https://impsci.tracs.unc.edu/tcast/), and the CFIR scored the highest, indicating good fit to be used in this review. Although it is recommended to assess the CFIR constructs for their salience and adapt them for a particular context⁽³⁹⁾, this study assessed for each of the forty-eight constructs across five domains of the updated version of CFIR (i.e. CFIR 2.0)⁽³³⁾ as it was not possible to specify the most relevant constructs without having prior knowledge. Whether or not studies contributed data to the CFIR constructs was displayed through a matrix checklist, and the nature of the data contributed (i.e. barrier or facilitator or both) was indicated by differential colour-coding. The number of studies reporting on barriers and facilitators was aligned with each of the a priori constructs, and domains were counted and presented in tabular form. In addition to the frequency table, a summary table of barriers and facilitators was presented in tabular form for each construct.

Results

Study selection

For the four review topics, we found a total of 5142 records (Figure 1). Ultimately, we found overall 308 eligible records across the four review topics outlined previously.

Figure 1. PRISMA diagram⁽⁴⁴⁾.

Absence of data on implementation determinants was the major reason for excluding studies at the full-text screening phase: out of 308 articles assessed for eligibility, 267 studies were excluded as they did not report on implementation barriers and/or facilitators. As shown in Figure 1, forty-one studies met the eligibility criteria and were included in DQCA.

Characteristics of included studies

All the reports/articles included in this review were published within the past 15 years (2007 and later), with the majority (n = 37, 90%) being in or after 2016, the year the World Food Prize was awarded to four researchers who pioneered the development and dissemination of a biofortified crops⁽⁴⁵⁾. The included studies reported on four conventionally biofortified crops and/or their products: OSP, HIB, VAC and VAM. All studies specified the approach of biofortification used. Studies in this review predominantly represent OSP (n = 28, 68%), followed by VAC and HIB (n = 8, 20% each). While only one study reported VAM, three studies examined multiple biofortified crops. Geographically, East Africa (n = 29, 71%) and West Africa (n = 10, 24%) were most represented in this review, with a few studies from other regions including South America (n = 1, 2%) and South Asia (n = 1, 2%). This review presents evidence from nine different countries from Africa, with more studies coming from Kenya (n = 9, 22%), Nigeria (n = 9, 22%), Mozambique (n = 7, 17%), Rwanda (n = 6, 15%) and Uganda (n = 6, 15%) than from Tanzania (n = 3, 7%), Ghana (n = 2, 5%) and Malawi (n = 1, 2%) in Africa, India (n = 1, 2%) in South Asia, and Brazil (n = 1, 2%) in South America.

All but two studies included farming households in the rural setting: one study focused on purchase for consumption of OSP bread among urban consumers, while the other studied consumption and knowledge of OSPs and IBs in both rural and peri-urban areas. In terms of research phase, eight studies were a pre-implementation assessment to inform choice and development of implementation strategy and two other studies explicitly focused on external validity utilising effectiveness trial methodology (i.e. T2-2 phase). Five studies primarily had joint aims to test an implementation strategy (primary) and to assess health outcomes (secondary) (i.e. hybrid type 3 design). The majority of the studies (n = 19) only focused on implementation research, assessing the effect of implementation strategy(ies) on adoption status (n = 11) or post-implementation impact evaluation (n = 8). Only four studies assessed adoption and its drivers at scale (e.g. T4-1 phase in the translational research continuum). In terms of study type, this review included thirty-one (76%) quantitative studies compared with a few purely qualitative studies (n = 4, 10%). Six (15%) were mixed-methods studies.

Characteristics of adoption measures

Definitions and measurements used for the adoption outcome varied greatly across the studies. Eleven out of forty-one studies used both categorical and continuous measures of adoption, while thirty studies (73%) used only a categorical outcome measure. The definitions of adoption were not consistent across studies and ranged from one to five per study (Table 3). Twenty-five articles (61%) used a single definition, with the remaining using multiple definitions. The most frequently applied adoption measure was whether a household was growing a biofortified crop or not. Other measures included but were not limited to the decision to grow or consume in future, retention/preservation of planting materials, proportion of dedicated cultivation area, total cultivation area, and previous or ever consumption/production/purchase of biofortified crops and related products. The included studies differed in terms of their focus on the stages of adoption. Most studies (n = 36, 88%) conceptualised adoption in terms of agricultural and/or economic aspects (i.e. cultivation, multiplication, retention for future cultivation, and selling/purchasing of biofortified crops and/or their products). Nine studies (22%), however, had explicit focus on consumption (including three of these studies with sole focus on consumption). Studies also differed regarding the continuity aspect of the adoption measure. Thirty-one studies (76%) captured cross-sectional (first time) adoption, while ten studies (24%) focused on continuous adoption (e.g. voucher redemption for seeds/planting materials, storage or retention of seeds/planting materials, and cultivation of biofortified crops for multiple seasons/years).

Table 3. Mapping of barriers and facilitators to the Updated Consolidated Framework for Implementation Research (CFIR 2.0) constructs

Notes: CFIR 2.0 (33). HIB, high-iron beans; OSP, orange sweet potato; VAC, vitamin A cassava; VAM, vitamin A maize.

Blue, consistent facilitator of adoption; pink, consistent barrier; yellow, facilitator and barrier.

Barriers and facilitators to adoption of biofortified crops

Barriers to and facilitators of adoption of biofortified crops and products were spread across all the five domains of the CFIR 2.0: innovation, outer setting, inner setting, individuals and processes. The included studies reported factors of adoption that were captured by twenty-four out of the forty-eight constructs under these domains. Table 3 provides a mapping of each of the included studies based on which construct(s) and domain(s) it contributed data (i.e. factors of adoption) to align with. This matrix checklist also provides a visualisation of studies reporting only facilitators, only barriers or both facilitators and barriers. Some constructs of the CFIR 2.0 manifested as either facilitators (cells highlighted as blue) or barriers (cells highlighted as pink) to adoption. Other constructs captured factors that were reported to simultaneously be associated with adoption as barriers and facilitators (cells highlighted as yellow) in a single study.

Orange sweet potato (OSP)

The specific factors affecting adoption of OSP under the reported CFIR constructs are provided in Tables 4–8. Anecdotally, in one study, researchers found that, in 2011, conditions for maintaining vines were difficult and many households lost their vines; however, households who had received agricultural extension (the treatment group) were more likely to have been able to keep vines through the particularly dry year and continued to grow OSP in 2012⁽⁸⁵⁾.

Table 4. Factors affecting adoption of orange sweet potato (OSP) and its products

Table 5. Factors affecting adoption of orange sweet potato (OSP) and its products

Table 6. Factors affecting adoption of orange sweet potato (OSP) and its products

Table 7. Factors affecting adoption of orange sweet potato (OSP) and its products

Table 8. Factors affecting adoption of orange sweet potato (OSP) and its products

Innovation characteristics

Characteristics of OSP that offered relative advantage and facilitated its adoption included early maturity^{(17,21,51,54,57,64)} , superior yield^{(17,21,49,51,54,57,64)} and sensory attributes^(19,48,49) , and high sugar (dry matter) content^(17,21,64) . Popularity of OSP as a staple food crop enhances its local adaptability to be adopted by farmers⁽⁷¹⁾. In addition, resistance to disease (e.g. virus^(57,71) ) and drought^(21,57) and shortened time for cooking⁽⁵⁷⁾ were reported to favour OSP adoption. Ease of storage of OSP in the ground for longer durations may also favour adoption⁽⁵⁷⁾, but pest infestation remains a barrier⁽⁶⁷⁾. However, a few studies reported that OSP was sensitive to diverse weather conditions limiting its adoption; these include dying of planting materials in dry growing conditions^(51,70,71) , lower tolerance to drought in white-fleshed sweet potato⁽¹⁹⁾, and perishability of vines when it rains^(51,54) . Some studies also reported relative disadvantages of OSP including shorter shelf life and cracking of tubers⁽⁶⁷⁾, lower dry matter content^(57,67) and inferior taste^(57,64) . Furthermore, farmers perceive OSP as more demanding in terms of field preparation and weed maintenance⁽¹⁹⁾, requiring intensive knowledge and labour⁽⁷⁰⁾. Therefore, they find recommended OSP production practices difficult to integrate into their existing production systems⁽⁶⁴⁾. All factors are summarised in Table 4.

Characteristics of individuals

Reported capability-related factors in the literature can be grouped as physical (age^{(17,21,58,64,65)} , skill^(54,65,71) and experience⁽⁷¹⁾) and psychological (education^(21,64,67) and knowledge^{(21,47,48,51,55,56,58,62,71)} ). Older age was reported as both barrier^(58,65) and facilitator^(21,64) among farmers. One study reported mixed views (i.e. both enabler and barrier) on the association of getting older and OSP adoption⁽¹⁷⁾. Education level of farmers or household head was directly related with OSP adoption^(21,64,67) – with low education serving as a barrier and more education serving as a facilitator – as was their knowledge on vitamin A^{(21,55,56,71)} , source for OSP^{(47,51,58,62)} and nutritional benefits of OSP^{(47,51,58,62)} or its product (e.g. OSP bread)⁽⁴⁸⁾. Implementers’ skill to serve clients well⁽⁵⁴⁾ and farmers’ ability to practice intercropping⁽⁶⁵⁾ and maintain OSP vines between seasons⁽⁷¹⁾ were facilitators. Having previous experience of growing OSP was positively associated with adoption in contrast to being new to OSP⁽⁷¹⁾.

Opportunity factors influencing adoption of OSP were related mainly to the physical opportunity (environment) reflected in access to agricultural inputs including vines^(19,21,51) , land^{(21,51,55,56,59,63,70,71)} , labour^(19,71) and water⁽⁷⁰⁾. Having access to or ability to purchase vines was reported as a facilitating factor^(21,51) , while lack of timely access to sufficient, good-quality vines impeded adoption^(19,51) . The size of land area to which farmers had access has been found to have conflicting associations with different measures of adoption. We found mixed associations of land both within^(55,56,71) and across studies^{(51,55,56,63–65,67,71)} . However, having larger land area cultivated with sweet potato was consistently reported as a facilitator^(55,56,71) . Labour constraints^(19,71) including heavy domestic responsibilities of women⁽⁷⁰⁾ constrained adoption. Consistent to the relationship of labour constraint and adoption, greater household size⁽⁷¹⁾ and higher dependency ratio⁽⁶⁶⁾ were found as facilitators of adoption. Lack of water sources or access to wetland⁽⁷⁰⁾ and lack of purchasing power⁽⁴⁹⁾ or capital^(19,70) to get access to inputs such as land and irrigation limited adoption. Participation in salaried employment, however, remained as a barrier to adoption. Crop diversity – reflected in growing and selling of higher number of crop varieties – was associated with higher adoption^(21,71) . Similarly, two studies reported positive association between being wealthier and adoption status. However, owning livestock assets did not have unidirectional associations with adoption across studies^(17,59) . Such lack of unidirectionality was also observed for the sex of household head controlling resources including land^{(21,51,55,56,64)} . Regarding social opportunity, we almost exclusively found positive associations of better social network and support and adoption. Social networking including participation in community farmer organisations or community groups^(57,70) and being the leader of farmer groups^(55,56,71) facilitated adoption. Likewise, adoption was positively associated with a support system reflected in better access to extension services⁽⁵⁷⁾ and higher frequency of interaction between extension and farmers⁽¹⁷⁾. One study, however, reported unlikeliness of cooperative society members to cultivate a large proportion of their farmland to OSP.

Motivation factors affecting adoption were both reflective and automatic in nature. Farmers’ aversion of risk⁽⁵¹⁾, unwillingness to try innovative crops and/or recommended practices^(51,54) , and having poor yield in the first growing season⁽⁵¹⁾ acted as barriers that fall under the reflective motivation category. We also found four facilitators that align with this sub-component of motivation; these include presence of child or pregnant women in the household^(17,50) ; being cosmopolite (having had a cosmopolitan outlook in the last year) to seek information on OSP⁽⁶⁴⁾; the satisfaction of serving one’s own communities⁽⁵⁴⁾; and women’s goal setting concerning when and how to provide children with proper complementary foods⁽⁵⁰⁾. All the reported factors that fall under the automatic motivation category were cited as facilitators. These include positive attitude towards OSP^(22,58) , having belief that children enjoy eating OSP tubers⁽²²⁾; having belief to make adequate returns from OSP/vine sales^(22,51,71) ; feeling in control of using OSP-based complementary foods⁽⁵⁰⁾; perception that vine multiplication provides employment opportunity⁽⁶⁰⁾ and brings good health^(22,51,60) ; and desire to earn ‘more income’⁽⁶⁰⁾. All factors are summarised in Table 5.

Outer setting

Factors within the outer setting domain concerned critical incidents, local attitudes and conditions, partnerships and connections, existing policies and laws, and external pressure. Critical incidents acted as barriers and included extended hot dry season⁽⁷¹⁾, high virus pressure⁽⁷¹⁾, flooding⁽⁶³⁾, pest damage⁽⁶³⁾, theft⁽⁵¹⁾ and grazing animals⁽⁵¹⁾. Community recognition of implementers’ work⁽⁵⁴⁾ and approval of OSP use in child foods⁽⁵⁰⁾ helped facilitate adoption. In contrast, prevailing gender ideology limiting women’s mobility and bias of extension services towards male heads of households constrained adoption⁽⁷⁰⁾. Also, perceptions of OSP as ‘a women’s crop’ and ‘a poor person’s food’⁽⁵⁴⁾ were cited as barriers. Local conditions that favoured adoption include proximity and good access to markets for roots (vines) and means of transport⁽⁷¹⁾. Areas with two growing seasons a year also facilitated adoption⁽⁷¹⁾. In contrast, drought-prone areas characterised by high temperatures and little moisture posed hindrance to adoption^(49,63,70) as did areas with soil of low fertility and instances of soil erosion⁽⁶³⁾. High costs associated with transport, delivery of OSP vines, or agricultural inputs (labour, fertilisers, herbicides or vines) were cited by four studies^{(63,64,67,71)} . Greater distance to key places including agricultural field office, main road, vine multipliers’ locations and health facility were inversely associated with adoption^{(17,54,57,66,71)} . Other local infrastructural barriers included lack of access to markets (clients) for selling vines⁽⁷⁰⁾, inefficient transport system⁽⁶⁴⁾ and poor storage conditions⁽⁶⁷⁾.

Market pressure (i.e. competition of OSP (or vines) with peer crops) was associated with farmers’ adoption of OSP. Availability of markets and traders for roots and vines⁽⁷¹⁾, profit from selling roots/vines⁽⁶⁴⁾, and presence of market price incentives for OSP and its products⁽⁴⁹⁾ helped facilitate adoption. In contrast, inadequate market for the sale of OSP^{(51,64,65,67)} , unstable markets⁽¹⁹⁾, lack of demand⁽⁶²⁾ and lower price premium⁽⁶²⁾ hindered adoption.

Partnerships and connections with entities external to the implementing organisation were associated with implementation outcome. Adoption was reported to be facilitated by the greater numbers of nongovernmental organization (NGO) engaged in agricultural activities with smallholder farmers and the presence of strong public sector extension services or public institutions. However, protraction and difficulty in contracting suitable local partners constrained the adoption pathway⁽⁷¹⁾. In addition to NGO and extension services, availability of socially cohesive pre-existing groups and the existence of strong farmer organisations/associations at the local level were cited as facilitators⁽⁷¹⁾, but limited sharing of information and planting material across farmer networks limited adoption⁽¹⁹⁾. Weak linkages among producers, processors, manufacturers and marketers were also mentioned as barriers to the pathway to adoption⁽⁶⁷⁾. Although the existence of government policies, programmes and strategies for promotion and consumption of OSP helps facilitate adoption⁽⁷¹⁾, lack of collaboration at the national level about the promotion of OSP hindered adoption^(67,71) . Similar hindrance effects were also reported for under-financed, under-staffed and under-recognised nutrition programmes and policies⁽⁵⁴⁾. All factors are summarised in Table 6.

Inner setting (implementing organisation)

The characteristics of inner setting (i.e. implementing organisations) including institutional capacity, available resources, relational connections, and access to knowledge and information were associated with implementation of programmes aimed at increasing adoption. Failure to supply healthy planting materials of the desired varieties timely in sufficient quantities deterred adoption^(49,63,71) . Human resource inadequacy as barriers to programme delivery was cited by two studies that characterised extension staff as insufficient (resulting in high client loads), overcommitted (multitasking), underpaid and inadequately qualified^(54,71) . While high-quality training of extension personnel and farmer multipliers was reported to be critical to successful adoption⁽⁷¹⁾, lack of training and training materials was the most frequently mentioned constraint by implementers⁽⁵⁴⁾. Lack of adequate office space and transportation for implementers was also reported as barriers⁽⁵⁴⁾. However, a working environment with effective supervision and cooperative teamwork helped implementers stay motivated to effectively carry out their work to promote adoption⁽⁵⁴⁾. All factors are summarised in Table 7.

Implementation processes

The aspects of planning, teaming and assessing context that facilitated implementation of programmes aimed at increasing adoption were cited by two studies^(54,71) . Specific activities within these three sub-domains were having sufficient preparatory period for improvements in intervention design (i.e. planning), diagnostics analyses to identify implementation challenges (i.e. assessing context) and having local implementation partners with established systems and procedures to support coordination, networking, organisation of implementers, and ongoing monitoring and evaluation (i.e. teaming). These two studies also identified the positive associations between reflecting and evaluating as in timely detection of problems in implementation and their resolution using mixed-methods implementation evaluation⁽⁵⁴⁾ and operations research⁽⁷¹⁾. For example, following a dramatic drop of voucher redemptions for vines in the dry season, vine multipliers were provided with pumps for irrigation to ensure a steady flow of healthy, mature vines while farmers were allowed to redeem vouchers for an extended period as they could not keep vines alive during the dry season⁽⁵⁴⁾. Two separate studies identified the positive association between adoption and modifying the innovation (i.e. biofortified OSP crop) without altering its any core feature (i.e. adapting). These included altering the length of exposure to the OSP intervention⁽²¹⁾ and frequency of exposure to the OSP intervention⁽⁶⁹⁾. Tailoring of strategies to increase adoption was cited by twelve studies and included use of vouchers redeemable for vines^(46,54,71) , agronomic training^{(17,46,52,53,57,59,70)} , market development⁽⁷¹⁾, nutrition education through multiple channels⁽⁶¹⁾ and articulation of agriculture, nutrition and market sectors⁽⁶⁸⁾. Engaging innovation recipients to attract and encourage them to adopt OSP was mainly through nutrition education and awareness campaigns emphasising the benefits of growing and consuming OSP^{(17,46,51–54,57,59,69)} . However, de-incentivising implementers (i.e. community health workers) was associated with reduced probability of growing and consuming OSP⁽⁶⁶⁾. All factors are summarised in Table 8.

High-iron beans (HIB)

The specific factors affecting adoption of HIB under the reported CFIR constructs are provided in Tables 9–13.

Table 9. Factors affecting adoption of high-iron bean (HIB) and its products

Table 10. Factors affecting adoption of high-iron bean (HIB) and its products

Table 11. Factors affecting adoption of high-iron bean (HIB) and its products

Table 12. Factors affecting adoption of high-iron bean (HIB) and its products

Table 13. Factors affecting adoption of high-iron bean (HIB) and its products

Innovation characteristics

Characteristics of HIB that offered relative advantage and facilitated its adoption included better taste, better market return potential and shortened time for cooking⁽⁸⁴⁾. However, its poor yield was cited as a barrier⁽⁸⁴⁾. Nevertheless, tolerance to drought and resistance to pests and disease were reported to favour HIB adoption⁽⁸⁴⁾. All factors are summarised in Table 9.

Characteristics of individuals

Reported capability-related factors in the literature can be grouped as physical age^(80,81) and experience^{(11,25,77,78)} and psychological (education^{(11,25,77,78,80,81,84)} ) and knowledge^(47,80,81) . Education was found to have mixed effect on HIB adoption^{(11,25,77,78,80,81,84)} . Lack of awareness of or knowledge about the source and benefits of HIB were cited as barriers to HIB adoption^(47,80,81) . Older age of farmers was reported as barrier^(80,81) . However, more years of experience of growing HIB was associated with lower dis-adoption of HIB^(11,25) . Growers of traditional climbing beans were found less likely to adopt the improved bush variety^(77,78) .

Two physical opportunity factors – high price of improved seeds and lack of land – were barriers⁽⁶³⁾. Other physical opportunity factors including wealth^(79,84) , land area^(79–81,84) , crop diversity^(80,81) , household size^(77–79) and number of agricultural equipment owned^(77,78) were positively related with HIB adoption. Households with women as the main decision maker for bean production was associated with better adoption^(11,25) . This was also true for households with better access to information, markets and financial resources⁽⁸⁴⁾. Regarding social opportunity, we exclusively found that a better social network and support system were positively associated with adoption. Social networking reflected in membership to highly trusted social groups⁽⁸⁴⁾ and a higher number of fellow farmers with whom to discuss agriculture and nutrition^{(11,25,80,81)} facilitated adoption. Likewise, adoption was positively associated with support systems such as having access to agricultural extension agents^{(11,25,77,78)} . All factors are summarised in Table 10.

Outer setting

Factors within the outer setting domain concerned critical incidents, local conditions and external pressure. All factors cited under these three constructs posed barriers to HIB adoption and included drought⁽⁶³⁾, remoteness (distance to produce markets)^(80,81) and low market price for HIB⁽⁶³⁾. No facilitators were reported. All factors are summarised in Table 11.

Inner setting (implementing organisation)

The characteristics of inner setting (i.e. implementing organisations) that were associated with adoption of HIB were related to availability of resource materials. Failure to supply HIB planting material (seed) timely in sufficient quantities deterred adoption and was reported by two studies^(63,84) . No facilitators were reported. All factors are summarised in Table 12.

Implementation process

Two aspects of implementation process (adapting and tailoring strategies) were found to facilitate the implementation of programmes aimed at increasing adoption. Adapting included modifying the size of HIB seed packets. This helped farmers get access to small packets without travelling long distances to formal delivery channels while providing small-scale producers with large packets^(11,25) . Tailoring of strategies to increase adoption was cited by three studies and included use of seed delivery payback scheme^(11,25) , increasing visits from extension officers^(80,81) , and provision of nutrition training in addition to intensive agricultural training by extension officers^(82,83) . No barriers were reported. All factors are summarised in Table 13.

Vitamin A cassava (VAC)

The specific factors affecting adoption of VAC under the reported CFIR constructs are provided in Tables 19–21.

Innovation characteristics

Characteristics of VAC that offered relative advantage and facilitated its adoption included early maturity and superior yield⁽⁷⁶⁾. Several studies, however, reported disadvantages of VAC limiting its adoption including bitter taste of fresh roots⁽⁷⁶⁾, poor yield⁽²⁴⁾ and lower dry matter content (high moisture)^(24,67,76) . Resistance to drought⁽⁷⁶⁾ was reported to favour VAC adoption, but time-consuming root peeling of VAC and frying of gari⁽⁷²⁾, difficulty in sprouting⁽²⁴⁾ and perishability^(24,67) are complexities associated with VAC constraining its adoption. All factors are summarised in Table 14.

Table 14. Factors affecting adoption of vitamin A cassava (VAC) and its products

Characteristics of individuals

Among the psychological capability factors affecting VAC adoption, level of education was positively associated with adoption^(24,73,76) . Similarly, having increased awareness of the benefits of VAC facilitated adoption^{(23,67,73–75)} . Physical capability factors including older age and higher experience in farming activities were negatively associated with VAC adoption^(8,23,74,75) .

Lack of physical opportunity factors including capital^(74,75) , planting materials^(67,74,75) and labour⁽⁷²⁾ constrained adoption. Still, larger household size acted as a barrier to VAC adoption⁽²⁴⁾. Being a female youth was associated with more engagement in cassava production activities⁽⁷⁶⁾, and two studies reported a positive association between getting married and VAC adoption^(23,76) . Further physical opportunity factors supporting higher adoption included access to credit^(24,73) , VAC stem⁽⁸⁾ and media including phone and internet^(8,24,73) . Social opportunity facilitating adoption came in the forms of contact with extension agents to receive training on improved agronomic practices^(8,24,72,73) and membership of cooperative society⁽²⁴⁾.

Motivation factors affecting VAC adoption were both reflective and automatic in nature. Within the reflective motivation category, income from non-farm activities^(74,75) and higher number of under five children⁽²⁴⁾ acted as barriers. However, increased income from VAC further motivated farmers to adopt VAC⁽²³⁾. The only automatic motivation factor supporting adoption of VAC was having a positive perception about it⁽²³⁾. All factors are summarised in Table 15.

Table 15. Factors affecting adoption of vitamin A cassava (VAC) and its products

Outer setting

Factors within the outer setting domain concerned critical incidents, local attitudes and conditions, partnerships and connections, and external pressure. Critical incidents acted as barriers and included activities of Fulani herders destroying first planting of VAC⁽⁶⁷⁾, pest damage and disease infestation^(74,75) , theft^(74,75) and farm invasion of grazing animals^(74,75) . A prevailing general perception that the yellow colour of VAC indicates it is used for sick people also constrained adoption⁽⁶⁷⁾.

Local conditions that favoured VAC adoption include ease of transporting products to designated markets⁽⁷²⁾. In contrast, inadequate local capacity to supply freshly harvested VAC to large-scale buyers and processors posed hindrance to adoption⁽⁶⁷⁾. Expensive processing and storage facilities^(72,74,75) coupled with poor road networks^(74,75) particularly discouraged farmers’ adoption of VAC. These problems were further compounded by poor market infrastructure and middlemen exploitation⁽⁷²⁾ with associated fluctuations in VAC price^(74,75) . We also found one study citing poor soil fertility as a barrier to adoption⁽²⁴⁾.

Competition of cassava with peer crops was associated with farmers’ adoption of VAC. Non-availability of ready-made markets for VAC^(74,75) , low price⁽⁶⁴⁾ and oversupply of cassava in local markets⁽⁷²⁾ deterred VAC adoption. In contrast, higher consumer demand facilitated adoption⁽⁷²⁾.

Partnerships and connections with entities external to the implementing organisation were associated with implementation outcome. Absence of collaboration among agricultural stakeholders was cited by one study as the main issue behind poor adoption of VAC⁽⁶⁷⁾. Lack of extension services and poor extension supervision were also found to limit adoption^(67,72) . All factors are summarised in Table 16.

Table 16. Factors affecting adoption of vitamin A cassava (VAC) and its products

Inner setting (implementing organisation)

The characteristics of inner setting (i.e. implementing organisations) that were associated with the adoption of VAC were related to availability of resource materials. Failure to supply VAC planting material (seed) even when farmers were willing to pay for them constrained adoption and was reported by two studies^(67,73) . No facilitators were reported. All factors are summarised in Table 17.

Table 17. Factors affecting adoption of vitamin A cassava and its products

Implementation process

An aspect of implementation process found to be associated with VAC adoption was engaging innovation recipients. Low-awareness campaigns on VAC and its products were cited as a limiting factor⁽⁶⁷⁾, while involving farmers in participatory research process facilitated adoption⁽⁷³⁾. All factors are summarised in Table 18.

Table 18. Factors affecting adoption of vitamin A cassava and its products

Table 19. Barriers affecting adoption of vitamin A maize (VAM) and its products

Table 20. Barriers affecting adoption of vitamin A maize (VAM) and its products

Table 21. Barriers affecting adoption of vitamin A maize (VAM) and its products

Vitamin A maize (VAM):

The specific factors affecting adoption of VAM under the reported CFIR constructs are provided in Table 19. As a note, we found only one study, and this study only reported barriers of VAM adoption.

Innovation characteristics

Characteristics of VAM that offered relative disadvantage and hindered its adoption included lower yield and harder texture⁽⁶⁷⁾. Aflatoxin contamination and damage by storage pests were complexities associated with maize constraining its adoption⁽⁶⁷⁾. All barriers are summarised in Tables 19–21; no facilitators were reported in our included studies.

Characteristics of individuals

Among the psychological capability factors, we found low level of education and lack of awareness of VAM to be associated with lower adoption⁽⁶⁷⁾. Similarly, limited agronomic insights was negatively associated with VAM adoption⁽⁶⁷⁾. All barriers are summarised in Table 20.

Outer setting

Factors within the outer setting domain concerned local conditions and external pressure, extracted from one study⁽⁶⁷⁾. All factors cited under these two constructs posed barriers to VAM adoption and included inefficiencies in the credit market, consumers’ high preference for conventional maize with associated low demand of VAM, and lack of markets for selling VAM⁽⁶⁷⁾. All barriers are summarised in Table 21.

Summary: OSP, HIB, VAC, VAM

The frequency of studies reporting barriers and facilitators aligned with each of the constructs and specific to each of the four crops and/or their products (OSP, HIB, VAC, VAM) is presented in Table S1.

Among the characteristics of a biofortified crop, relative advantage and complexity were common across all the crops as positively associated with adoption. More studies (n = 11, 26.8%) identified relative advantage as facilitators rather than barriers (n = 7, 17.1%), while complexity was reported more commonly as barriers (n = 8, 19.5%) than facilitators (n = 5, 12.2%). Within the outer setting domain, local conditions and external pressure were common factors behind adoption. External pressure, particularly those related to market, emerged as hindrance to adoption in nine studies (22.0%). Aspects of local conditions were more common as barriers being cited by approximately one-third studies (n = 14, 34.1%). Within the context of implementing organisation, the only construct common to more than one crop was available resources being cited as barriers by several studies (n = 6, 14.6%). Constructs under the domain of implementation process were virtually absent in the literature reporting studies on VAC, HIB and MAC. While at least two studies on OSP reported each of the six constructs (teaming, engaging, assessing context, planning, adapting, tailoring strategies, and reflecting and evaluating) included in Table 22, they were almost exclusively identified as facilitators of adoption. Engaging innovation recipients (n = 8, 19.5%) and tailoring strategies to improve implementation (n = 13, 31.7%) were most frequently reported implementation processes facilitating adoption. In contrast to implementing organisation and implementation process domains, “characteristics of individuals” domains were heavily coded. The opportunity construct was most frequently cited – both as barriers (n = 15, 36.6%) and as facilitators (n = 23, 56.1%) – among all twenty-four constructs reported in this review. We also found capability features of individuals to have mixed associations with adoption with thirteen (31.7%) and sixteen (39.0%) studies reporting negative and positive associations, respectively, with adoption. Unlike capability and opportunity, motivation as a factor of adoption was not common across all the crops. Adoption of only OSP and VAC was reported in the literature to be affected by aspects of motivation among individuals.

Table 22. Geographies of include crops and studies

Regions based on World Health Organization regions⁽¹⁰¹⁾. HIB, high-iron beans; OSP, orange sweet potato; VAC, vitamin A cassava; VAM, vitamin A maize; AFR, African Region; AMR, Region of the Americas; SEAR, South-East Asian Region; EUR, European Region; EMR, Eastern Mediterranean Region; WPR, Western Pacific Region.

Discussion and conclusions

In this review, we synthesised forty-one studies to elucidate the barriers to and facilitators of adoption of conventionally biofortified crops. OSP represented nearly 70% of studies, followed by HIB and VAC (20% each), and one study in VAM, which is expected given the earlier roll-out of OSP in target regions. This is the first review to apply a comprehensive determinants framework (i.e. CFIR) from the implementation science field to identify factors associated with adoption of a biofortification intervention. We identified a wide array of barriers and enablers/facilitators to the adoption of biofortified crops and foods. For provitamin A-biofortified OSP, the most reported factors that promoted adoption were farmers having knowledge about importance, relative advantage, and efficient production and management practices of OSP, and the main barriers were lack of timely access or inability to maintain good-quality vines especially in drought-prone areas and remoteness to key places including markets. For HIB, the main factors that promoted adoption were farmers’ networking and high farming experience and the main barriers were lack of knowledge in the form of education and experience about HIB. For VAC, the main factors that promoted adoption were having awareness of the benefits of VAC and access to information, while the main barriers were poor road networks and inadequate processing and storage facilities. Finally, for VAM, factors identified were only barriers mainly related to low awareness about VAM among farmers and consumers in addition to concerns regarding yield, texture and aflatoxin contamination. Overall, agronomic and nutritional benefits of the biofortified crops, especially when farmers are aware of these relative advantages, were found to facilitate adoption. Conversely, production, storage and processing-related constraints arising from local environmental and market conditions coupled with inadequate agronomic insights of farmers limit their adoption.

A comprehensive understanding of adoption determinants is required and should act as an evidence base to inform the development and/or selection of breeding and scaling strategies. Arguably, the Consolidated Framework for Implementation Research (CFIR) that has been developed by Damschroder et al. is the most well developed and comprehensive among the determinant frameworks⁽⁸⁶⁾. By using a meta-theoretical determinant framework (i.e. CFIR) to capture adoption factors acting at multiple levels, we found adoption determinants of biofortified crops and their products to occur within the existing domains of the CFIR. This highlights that studies contributing to the evidence of this review explored a diverse array of aspects influencing adoption of biofortified crops. Kirk et al. studied how researchers utilise the CFIR in implementation studies and found that it is mostly employed to analyse what facilitates or hinders implementation (aligned with our goal)⁽⁸⁷⁾. Commonly cited determinant frameworks other than the CFIR are limiting in one way or other (e.g. do not explicitly address the relevance of end-users (farmers)) to be used as the synthesising architecture. As biofortified crops are an agricultural innovation with the goal of scaling up nationally rather than replicating a clinical innovation across healthcare facilities (i.e. the basis of developing the CFIR), it was difficult to delineate the CFIR’s implementing organisation and outer setting domains within the agriculture context. CFIR, being an organisation-centric framework, is often criticised for downplaying the central role of end users. However, CFIR has been revised on the basis of user feedback, and the version (i.e. CFIR 2.0)⁽³³⁾ we have employed in this review incorporates constructs based on Michie et al.’s⁽⁸⁸⁾ COM-B system that shape behaviour.

This comprehensive review, being the first of its kind to assess drivers of adoption among producers, to our knowledge, limited us in comparing the adoption factors with previous literature. However, the constructs and domains identified agree with factors previously reported to affect the implementation of agriculture-nutrition interventions in general^(89–91). Although factors of adoption identified in this review were summarised by using only 50% (twenty-four out of the forty-eight) of the CFIR constructs, this is not a perceived limitation of CFIR. Rather, this is possibly a reflection of the nascent state of implementation research and authors not reporting or incorporating many of the implementation aspects in their research. Other studies utilising the CFIR framework have also not utilised all the domains/constructs^(92,93) .

We found a relatively uneven distribution of factors across domains with relatively low representation from the inner setting (i.e. characteristics of the implementing organisation) and implementation processes domains. That we included in our review only those factors drawn directly from empirical evidence explains this discrepancy in the frequency of factors across domains. Another related reason behind the relatively infrequent reporting of inner setting factors and implementation processes is the low number of studies reporting process evaluations or implementation science research (i.e. those testing different implementation strategies with exclusive focus on improving implementation outcomes (e.g. adoption)). Most of the studies included in this review covered the initial stages of implementation that included support from research teams. Once this support is withdrawn and implementation progresses to advanced stages, e.g. real-world conditions, determinants from the inner setting (e.g. resources available to the implementing organisation) and implementation processes (i.e. activities and strategies used to implement the intervention) domains may start acting upon the adoption process⁽⁹⁴⁾.

For example, successful implementation of the biofortified OSP in Kenya has been largely attributed by Cole et al.⁽⁵⁴⁾ to the additional time, funding and other core resources available to the project team from project partners and donors during the implementation. Although the majority of studies lacked exclusive focus on these two domains, authors frequently mentioned potential association of these domains as recommendations to be considered before future development of biofortified crop project. However, our sole reliance on data-derived factors instead of authors’ subjective project experiences adds strength to the evidence presented in this review. In contrast to inner setting and implementation processes domains, we found a relatively common association between adoption and characteristics of individuals as reflected in the frequency with which studies reported factors that fell under capability, opportunity and motivation. This finding reflects the importance of not overlooking the individual characteristics when developing implementation strategies to promote adoption of biofortified crops and supports Powell et al.⁽⁹⁵⁾, who emphasised assessing the characteristics and preferences of receivers and deliverers of interventions before their delivery.

Although biofortified crops have been shown to have superior characteristics to the many non-biofortified varieties currently cultivated by farmers, factors within the intervention characteristics domain were not exclusively identified as facilitators. Empirical findings from the studies in this review confirm that adoption is contingent upon a comprehensive list of agronomic and sensory attributes of the biofortified crops and their products. For VAM, HIB and VAC, economic demands of higher yield should also be addressed in addition to palatability standards. For OSP, more attention should be paid especially to sensory attributes including dry matter content (sugar and starch composition), skin colour and root shape. There were far more barriers than facilitators identified in the inner setting and outer setting domains. For example, barriers including failure to timely supply adequate planting material and human resource inadequacy were reported across all the fortified crops in the inner setting domain. This is not surprising given the underfinanced public sector extension services in countries where the studies included in this review took place. Similarly, commonly reported barriers in the outer setting including physical remoteness, seasonality (especially dry season) and lack of markets and agro-products storage and processing facilities mostly fell beyond the control of implementing organisations. However, implementation processes that recognised local conditions during planning^(52–54), made choices that fit with existing conditions^(11,25,73) , allowed flexibility in project implementation from planned interventions in response to obstacles identified through ongoing monitoring and evaluation^{(46,54,71,80,81)} , and provided combined nutrition-agricultural training to generate demand and compensate for limited agronomic know-how^{(17,46,51–54,57,59,69,82,83)} allowed to overcome these barriers. This explains why the implementation process domain (especially the engaging construct) was predominantly coded as facilitator in sharp contrast to the inner setting and outer setting domains.

The determinants identified in this review should be interpreted not as the ones with the strongest association with adoption of biofortified crops, but the ones which were studied and reported by the authors of the included studies. This review only included determinants that were directly supported by the quantitative data or qualitative statement (self-report) describing the real experience of the study participants (i.e. no ‘hypothetical’ barriers, such as authors’ extrapolation of probable subsequent events acting as a barrier or facilitator, were included). Moreover ‘adoption’ was measured differently across studies. It is likely that adoption happens in stages, rather than being a discrete, one-off event. For example, Rogers⁽⁹⁶⁾ describes at least four stages of adoption (i.e. intention, initial decision, action to try or use, continued use). A major limitation of this review is that we are unable to explain how the identified factors are causally linked to adoption (or the stage of adoption) and their degree of influence.

Future studies on adoption of biofortified crops should consider the stages of adoption and identify the determinants across these stages⁽⁹⁷⁾. Researchers should also synthesise and recommend various implementation strategies based on identified barriers using the CFIR-Expert Recommendations for Implementing Change (ERIC) implementation strategy matching tool⁽⁹⁸⁾. It is imperative that the future studies adequately report the different characteristics of the implementation strategies (e.g. action, dose and duration) employed to address adoption following published recommendations (e.g. Proctor et al.⁽⁹⁹⁾). Future research on adoption determinants should also focus on the understudied crops, such as VAM, for which we found only one study, and understudied geographies (most included studies were done in the African Region as defined by the World Health Organization (Table 22)), mainly focused on OSP, due to its being rolled out there first as discussed previously⁽⁵⁾. As biofortified crops have been rolled out in thirty out of over 120 countries tested as of 2020⁽¹⁰⁰⁾, across all the WHO regions (Table 22), excluding the European Region, research is needed to inform adoption determinants in these varied settings, accounting for not only differences in adoption but also in delivery and diffusion, considering the diversity in crop type: vegetatively propagated crops like cassava and sweet potato, hybrid grain crops like maize and open pollinated crops like beans and pearl millet. Additionally, as non-biofortified wheat and rice are commonly consumed, it will be useful in a future study to compare the adoption determinants between these and biofortified crops. Finally, understanding the facilitators and barriers to adoption of crops biofortified through other means, including either agronomic or genetic engineering – which was beyond the scope of the current review – remains a research gap necessary to assess in a future review, given the varying appropriateness and applicability of different biofortification methods across different settings and geographies⁽³⁰⁾.

Conclusion

This review analysed forty-one studies reporting on aspects of adopting biofortified crops and their products and provides a set of factors that enables or constrains their across twenty-four constructs and five domains of a meta-theoretical determinant framework from implementation science. The barriers identified in this review are a useful starting point for implementation researchers to move beyond the currently dominant effectiveness research towards testing implementation strategies for scaling up biofortified crops and continuous optimisation of ongoing projects promoting adoption of biofortified crops. This review has revealed that scaling up research would be well served by studies that compare different delivery models (e.g. free versus subsidised distribution of planting materials), delivery platforms (e.g. commercial versus public to address the underfinanced inner setting) and delivery channels (e.g. facility-based versus community-based to address physical remoteness) and implementation processes to adoption of biofortified crops. Also, in-depth studies should be conducted in different contexts to delineate the interactive influences of different CFIR constructs including biological and cultural characteristics of individuals (e.g. age and sex) and economic factors (e.g. wealth and income) on adoption. Finally, as the current literature is heavily biased towards OSP, one of the first crops to be developed and rolled out, future works should involve examining adoption of other types of biofortified crops as their dissemination increases, such as VAM, iron and zinc pearl millet, zinc wheat and iron lentils, as well as VAC and HIB.