Use of designing for behaviour change framework in identifying and addressing barriers to and enablers of animal source feeding to children ages 8–23 months in Bandarban Hill District in Bangladesh: Implications for a nutrition‐sensitive agriculture programme

Abstract Inadequate diet quality is a cause of undernutrition among children 6–23 months of age in Bangladesh, particularly in remote and isolated areas such as Bandarban District. Feeding animal source foods can help to combat stunting and wasting problems among children, but it may not be accessible or acceptable. A barrier analysis using the Designing for Behavior Change Framework was conducted in Bandarban district with participants from 4 ethnic groups, to explore potential barriers and key motivators by examining 12 behavioural determinants of consumption of animal‐source food in complementary feeding for children 8–23 months. Data were collected from 45 mothers of children 8–23 months, who provided animal‐source foods to their children (doers), and from 45 mothers who did not (non‐doers), for a total of 90 interviews. Nine determinants were statistically significantly different between doers and non‐doers as follows: self‐efficacy, positive consequences, negative consequences, social norms, access, reminders, perceived risk, perceived severity and perceived action efficacy. Nearby access to purchase animal‐source foods, rearing poultry or livestock at home and the support of household and community members are enablers to feeding animal‐source food. In contrast, these same factors are barriers for non‐doers. The lack of money to spend on animal‐source foods is also a barrier. An integrated nutrition‐sensitive and gender‐transformative animal‐based food production, and inclusive market programme could increase access to meat and eggs at the household level, increase opportunities to earn income and support gender‐equitable household workloads and decision‐making for optimal child feeding.


| INTRODUCTION
Globally, approximately one in five children under 5 years of age (149.2 million) are stunted and 45.4 million children are wasted (Development Initiative, 2021). Asia has the highest rate of stunting (53%) and wasting (70%) globally, and the majority of stunted and wasted children reside in South Asia (UNICEF, WHO & The World Bank, 2021). In Bangladesh, child undernutrition is a public health concern, with 28% of children under 5 years stunted, 25% underweight, and 10% wasted, with children in rural areas suffering from higher rates of undernutrition than in urban areas (NIPORT & ICF, 2019). Inadequate dietary consumption, especially among children 6-23 months, is a primary cause of poor nutrition outcomes for children (Tette et al., 2015). Increasing macro-and micronutrient intake, including carbohydrates, protein and fats, as well as vitamins A, B-12 and riboflavin, calcium, iron and zinc, can improve child growth Balehegn et al., 2019;Hirvonen et al., 2020;Pasha et al., 2013).
Beyond the age of 6 months, global guidance advises breastmilk alone is not sufficient to provide the required nutrients for children 6-23 months to achieve optimal physical and cognitive development, although continued breastfeeding is recommended in addition to other foods. To ensure and meet nutritional needs, complementary foods should be timely, adequate, safe and responsively fed . Animal-source foods (ASF) have high concentrations of essential nutrients Krebs et al., 2006;Zhang et al., 2016) and consumption of meat (including organ meat), fish, eggs and dairy products (milk, cheese and yoghurt) can support infants and young children to achieve adequate nutrient intake (Robert et al., 2021) and substantially reduce the risk of stunting and wasting (Dewey, 2013).
Children 6-23 months in Bangladesh consume low amounts of animal-source foods, despite increased meat, milk and egg production in recent years (JPGSPH BRACU & NNS, 2020). Thirty-nine per cent of breastfeeding children aged 6-23 months were found to have consumed meat, fish or poultry, 36% consumed milk products and another 59% consumed eggs (JPGSPH BRACU & NNS, 2020).
Multiple factors, including the mother or caregiver's knowledge of appropriate infant and young child-feeding practices , availability and access to animal-source foods (Haileselassie et al., 2020), sociocultural norms about feeding animal-source foods , poverty , place of residence (urban or rural) and food security status (Sarma et al., 2017) influence consumption of animal source foods by children 6-23 months.
The remote, hard-to-reach Chittagong Hill Tracts (CHTs) region of southwestern Bangladesh is home to 48% of the country's total ethnic minority population with 11 distinct ethnic groups, each with their own language and cultural traditions (Ministry of Health and Family Welfare, 2016). Access to health and nutrition services, education, markets and other services and resources is limited. Most recent data show Bandarban District of the CHT has the highest stunting rate (47.7%) in the country, with underweight and wasting rates of 27.6% and 10.4%, respectively (WFP & IFAD, 2015).
In 2015, a 6-year multi-sectoral programme, the United States Agency for International Development (USAID) Resilience and Food Security Activity Sustainable Agriculture and Production Linked to Improved Nutrition Status, Resilience and Gender Equity (SAPLING), was awarded to improve gender-equitable food security, nutrition and resilience of vulnerable households living in five subdistricts of Bandarban. The programme baseline in 2016 found 34% of children aged 6-23 months had a minimum acceptable diet, with inadequate consumption of animal-source foods: just 46% of children aged 6-23 months, who were breastfed, consumed flesh foods, 38% consumed eggs and 23% consumed dairy (ICF, 2017). Children's minimum acceptable diet was higher in households above the poverty line and where the head of household completed primary education or higher.
The baseline also found varying beliefs regarding feeding young children protein-rich animal-source foods, such as fish or other meat.
Little additional information is available on children's nutritional

Key messages
• The Designing for Behavior Change Framework is an effective barrier analysis tool to assess and identify food consumption determinants.
• Consumption of animal-source food for children aged 8-23 months is constrained by the unavailability of ASFs, lack of financial means to purchase, lack of support from household members and perceived negative consequences.
• Consumption of animal-source food by children aged 8-23 months is enabled by access through householdlevel animal rearing, proximity to points of sale, and household and community support.
• An integrated nutrition-sensitive and gendertransformative animal-based food production approach could increase equitable access to and utilization of animal-source foods.
trends in this remote, multi-ethnic area, including the consumption of animal-source foods. To inform programme implementation for improving child nutrition, the primary objective of this study was to explore the multi-dimensional behavioural determinants of animalsource foods in complementary feeding for children aged 8-23 months in the programme area.
2 | MATERIALS AND METHODS

| Conceptual framework
The study used the Designing for Behavior Change (DBC) framework (The TOPS Program, 2016) to identify and address the barriers to ASF feeding practices. The DBC is a comprehensive tool that helps to achieve better and sustainable behaviour change results by guiding programme designers through five key components as follows: (i) behaviour, (ii) priority group or influencing group, (iii) determinants, (iv) bridges to activities and (v) activities. In the DBC approach, based on the health belief model and theory of planned behaviour (Kalam, Davis, Shano, et al., 2021), barrier analysis (BA) is used to identify determinants, influencing groups and inform activities for behaviour change. The BA is a rapid assessment tool using doer/non-doer methodology that identifies the barriers and enablers, and the most important behavioural determinants associated with a particular behaviour . Comparative data are collected from doers (individuals who perform the behaviour) and doers and 3 non-doers were interviewed for a total of 9 doers and 9 non-doers in each subdistrict.

| Ethics approval and data collection
The study was conducted according to the guidelines laid down in the

| Data management and analysis
Data analysis was completed in three steps. First, data were deductively coded according to each preselected determinant.
Responses to open-ended questions under each determinant were inductively coded into themes or categories by a three-person team.

An analysis workshop was conducted to code open-ended data and
inter-rater reliability was undertaken through discussion and agreement in the workshop by the three coders. Responses to closedended questions were automatically coded to the response codes. times more likely to say rearing livestock and poultry in the household makes ASF easy, because it provides eggs, meat and milk (p = 0.014). They were further three times more likely to say feeding is easy, because animal-source foods are available at home (p = 0.005). Doers were also 2.4 times more likely to say support from family members (by purchasing, collecting, cooking and feeding these foods to the children) (p = 0.032) made it easy. Conversely, non-doers were 2.4 times more likely to say having money in hand would make ASF easy (p = 0.047).
The above-mentioned enablers are found to be barriers for nondoer mothers, because they do not have those facilities. Non-doers voiced significant challenges: they were 21.9, 11.2 and 12.6 times more likely to say not having money to purchase ASF (p = 0.000); a market is not available nearby (p = 0.004); and unavailability at home (p = 0.002) made it difficult to provide ASF to children. For doers, however, feeding animal-source food becomes difficult when children are sick (doers are 3.9 times more likely to say this; p = 0.001).
3.2 | Access: How easy it is to get animal-source food to feed to children daily Perceiving ease or difficulty to obtain or access animal-source foods influences the behaviour. Non-doers were 13.5 times more likely to say it is difficult to obtain animal-source foods to feed their children (p = 0.000), whereas doers were 7.4 times more likely to say it is not difficult (p = 0.000). The responses to the perceived self-efficacy questions presented above elaborate that non-doers said it would be easier to feed animal-source food if they reared livestock at home and if they had a market or vendor near their home. Likewise, doers said these two things made it easier for them to feed animal-source food to their children.
3.3 | Positive and negative consequences: Advantages and disadvantages of feeding animal-source foods regularly Regarding perceived negative consequences of giving ASF, doers were 3.6 times more likely to say there were no disadvantages to feeding it to children (p = 0.003). Non-doers were 3.6 times more likely than doers to say children get diarrhoea and stomach upset due to giving ASF (p = 0.010), whereas doers (18%) were more likely to say children develop fever and cough as a result of giving ASF (p = 0.045).

| Cues for action: How difficult it is to remember to feed animal-source food daily
Cues for action appeared to be a major barrier for the non-doers.
They were 3.8 times more likely to say it would be difficult to remember to feed ASF daily (p < 0.001), whereas doers are 3.8 times more likely to say it is not difficult to remember (p < 0.001).

| Susceptibility of risk: Likelihood that children will become malnourished
Perceived risk of becoming malnourished and the level of concern about getting sick also appeared to be a predictive determinant of feeding with animal-source foods. Non-doers were 2.5 times more likely to believe their children will be malnourished in the coming year (p = 0.023) Conversely,doers were 4.1 times more likely to say it is not likely at all their children will be malnourished in the coming year (p = 0.001).

| Perceived severity: Extent of severity if children become malnourished
The perceived severity of malnourishment was also predictive of feeding animal-source foods to children. Non-doers were 2.1 times more likely to believe the occurrence of being sick or malnourished in the coming year will be very or somewhat serious (p = 0.007), whereas doers were 5.1 times more likely to say the occurrence would not be serious at all (p = 0.004).
3.7 | Action efficacy: Likelihood that feeding animal-source goods daily to children will cause children's sickness or malnourishment Action efficacy differs from the susceptibility to risk in that it assesses the perception that the specific behaviour could cause malnourishment or sickness rather than assessing the perceived risk of becoming malnourished or sick from any cause. Doers were 8.7 times more likely to say feeding ASFs daily would not cause children to be sick or malnourished in the coming year (p = 0.000). Conversely, non-doers were 5.1 times more likely to say it was somewhat or very likely that feeding ASFs would cause children to become sick or malnourished (p = 0.000).

| Perceived social norms: Who approves or disapproves of feeding ASF daily
There were no significant differences between doers and non-doers for perceived social norms being an enabler or barrier, as they both said no one disapproves of feeding ASFs. Regarding the approval, household poverty, geographical remoteness and isolation, access to animal-source foods emerged as the most critical determinant in the study population. Three factors were found to directly impact access: (1) rearing poultry and livestock at home, (2) availability of a nearby place to purchase animal-source foods, and (3) not having the financial resources to purchase animal-source foods. The study highlighted that limited availability to purchase and not rearing animals at home was perceived as a barrier by non-doers, which aligns with findings from other studies in other rural areas of Bangladesh (Rasheed et al., 2011;Thorne-Lyman et al., 2017). In addition, lower purchasing power (i.e., not having money on hand) was also seen as a barrier to accessing animal-source foods, confirming findings from previous studies in Bangladesh (Affleck & Pelto, 2012;Mukta et al., 2015;Rasheed et al., 2011) and other countries (Burns et al., 2016;Colecraft et al., 2006;Wong et al., 2020).
Although not having a place to purchase animal-source foods nearby, not having money to purchase them, and not rearing them at home are barriers for the non-doers, these same factors were the enablers for doers who said having market access and rearing at home made it easy to feed animal-source foods to their children The global analysis of the EAT-Lancet reference diet showed that consuming the recommended diet, including animal-source foods, is not affordable for most people in low-and middle-income countries (Hirvonen et al., 2020). Other studies conducted in Bangladesh (Affleck & Pelto, 2012;Mukta et al., 2015;Rasheed et al., 2011;Thorne-Lyman et al., 2017), Sub-Saharan Africa (Hetherington et al., 2017), and Ethiopia (Haileselassie et al., 2020) have similarly found lack of affordability to be a barrier. Conversely, even those with access to poultry and livestock may have issues of availability as poultry, and livestock may be viewed as a source of income or asset and not a food source; households that rear poultry and other livestock may choose to sell their animals rather than consume them.
In Timor-Leste, results indicated that individuals chose to sell their animal products to earn income, because the products are not considered food, but rather an asset (they found a perception that 'eating meat' means 'eating money') (Wong et al., 2020).
In addition to access, support from household members, including husbands and mothers-in-law, and other community members emerged as a potential key enabler to feeding animalsource foods to children 8-23 months (it should be noted that not having support was not a significant barrier). Having a household member who supports appropriate child feeding could mitigate the potential negative effects on child diet that mothers' heavy workloads may have. In Bangladesh and other low-and middle-income countries, women suffer from severe time constraints and bear the burden of responsibility for household chores and primary care of children and other household members (Sinharoy et al., 2018;Sraboni & Quisumbing, 2018). In the CHTs, engagement in traditional cultivation, often in locations away from the house, and other income activities, which require labour and time, exacerbates already heavy workloads for women (Kabir et al., 2019;Uddin, 2014).
Mothers' heavy workloads and lack of support from husbands and other family members have been found to negatively impact appropriate complementary feeding in Bangladesh (Affleck & Pelto, 2012;Rasheed et al., 2011).
Other factors from the sociocultural context also influence child feeding. Some mothers in this BA said elders advise against feeding animal-source foods to children, which may be due to beliefs about negative consequences. Previous studies in Bangladesh found that mothers-in-law and elderly community members, neighbours and community health workers influence decisions regarding the child and maternal nutrition behaviours (Avula et al., 2013;Hoddinott et al., 2017;Kim et al., 2018). The sociocultural context influences child feeding practices (Boak et al., 2016;Monterrosa et al., 2012) and related norms, which inhibit feeding animal-source foods to children due to perceived negative consequences can also be a barrier, whereas perceptions of positive consequences can be an enabler. For mothers who do not feed animal-source foods to their children under 2 years, diarrhoea, upset stomach, cough and fever were perceived as negative consequences.
The idea that animal-source foods can cause digestive issues is found across the world. The Alive and Thrive programme in Bangladesh and Ethiopia observed similar perceptions in which mothers thought children cannot digest meat and other animal products, which can cause sickness, including vomiting and stomach upset (Alive & Thrive, 2010; Mukta et al., 2015). Similarly, Mexican female caregivers reported beef is not given to young children, because they are unable to digest it (Pachón et al., 2007). In this BA, mothers who fed animalsource foods to their children did not think it likely that the foods would cause their child to become sick or malnourished and reported it boosts children's physical growth, immune function, child health and motor skills. In contrast, mothers who do not feed animal-source foods not only were more likely to think these foods could cause their child to become sick or malnourished but also had a perception that if a child becomes sick, it is divine will (see Supporting Information: File 2) and out of their hands. If a child does become sick, all mothers find it difficult to feed animal-source foods to their children due to illness or lack of appetite.

| IMPLICATIONS FOR NUTRITION-SENSITIVE PROGRAMMES
Based on the findings of the BA, the SAPLING programme designed behaviour change activities following the DBC framework.  (Table 2).
While looking at the children's ASF consumption data, overall, different types of ASFs consumption increased gradually over the programme duration (Table 2). While looking at the results by type of ASFs, it was revealed that organ meat and small fish consumption increased by 18 percentage points, followed by eggs and other meats (16 and 15 percentage points, respectively) from 2017 to 2020. This strategy also aligned with other studies that found ownership of poultry and livestock is associated with young child consumption of animal-source foods in Bangladesh (Choudhury & Headey, 2018;Talukder et al., 2010), Ethiopia (Kim et al., 2019a), Tanzania (Bundala et al., 2020;Kim et al., 2019)  increasing access through income generation and availability in nearby points of sale, can form a holistic approach, which enables feeding animal-source foods to children 8-23 months.

CONFLICT OF INTEREST
The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available in