Early initiation of breastfeeding, bottle feeding, and experiencing feeding challenges are associated with malnutrition

Abstract Malnutrition remains a public health concern amidst low proportions of the core infant and young child feeding (IYCF) practices, yet, data on specific child feeding practices that are associated with undernutrition are rare. Hence, this study sought to assess child feeding practices and their association with undernutrition among young children. An analytical cross‐sectional design was used among mothers/caregivers with children aged 6–23 months, attending child welfare clinics in Techiman municipality, Ghana. Simple random sampling was used to select 8 health facilities, and 403 participants were selected from those facilities using proportional stratification. A 24‐h dietary recall based on seven food groups was used to collect data on children's dietary intake and used to derive WHO child feeding indicators. The length, weight, and age of children were taken and used to compute anthropometric z‐scores. The proportions of children who met their minimum dietary diversity (MDD), minimum meal frequency (MMF), and minimum acceptable diet (MAD) were 44%, 56%, and 36% respectively. Children 6–8 months [AOR=2.24, CI (1.037–4.841); p = .04] and 9‐11 months [AOR=2.47, CI (1.096–5.573); p = .029], those who were not breastfed within the first hour of delivery [AOR = 3.56, CI (1.833–6.912), p < .001], and those who were bottle fed [AOR = 2.87, CI (1.374–5.973); p = .005] were more likely to be wasted. Children 6–8 months [AOR = 0.29, CI (0.126–0.672); p = .004] and 9–11 months [AOR = 0.24, CI (0.104–0.544); p = .001] and those who experienced feeding challenges [AOR = 0.52, CI (0.301–0.905); p = 0.021] were protected against stunting. The percentages of children who met their MDD, MMF, and MAD were low and not associated with undernutrition. Early initiation of breastfeeding and bottle feeding were associated with acute malnutrition and experiencing feeding challenges was associated with chronic malnutrition. Promoting appropriate child feeding practices can reduce the risk of undernutrition.


| INTRODUC TI ON
Complementary feeding refers to the timely introduction of safe and nutritious foods in addition to breastfeeding typically from 6 to 23 months of age (Bhutta et al., 2013).A survey in low-income countries showed that eating diverse diets lowers the chances of being stunted and/or underweight, and minimum meal frequency reduces the chances of being underweight (Bhutta et al., 2013).In Ghana, the proportion of children aged 6-23 months who met the three core infant and young child feeding (IYCF) practices was only 13% (GSS et al., 2015).Another study found that the proportions of children aged 6-23 months who met the minimum meal frequency (MMF) and minimum dietary diversity (MDD) were 46.0% and 51.4% respectively (Issaka et al., 2015).
Only 29% of infants globally meet their MDD (Bégin & Aguayo, 2017).According to Lutter et al. (2011), on the global burden of malnutrition and the highlight of data on child feeding practices, less than 30% of children aged 6-23 months achieved the MDD.A study by Kassa et al. (2016) showed that only 18.8% of infants in Benin met their MDD.Similarly, a study by Mekonnen et al. (2017) reported that 27.3% of children met their MDD with most of them consuming grains (84.6%) while just a few consumed fish (9.7%) and 1.1% consumed iron-rich foods (liver).
Globally, only 52% of children between the ages of 6-23 months met their MMF (White et al., 2017).Kassa et al. (2016) found that 67.3% of children met their MMF.Vitta et al. (2016) also reported that 70.3% of children between the ages of 6-23 months met their MMF in Tanzania which is higher than most studies.However, some studies in Ghana have found lower proportions of children receiving the recommended frequency of foods compared to the global prevalence, that is, 29% for Ga West Municipality and 40% for Tema Municipality (Agbozo et al., 2016;Bentil et al., 2016).
Most studies have shown that few children are able to meet both MDD and MMF, and do not meet their MAD and optimal complementary feeding with a global prevalence of MAD representing only 15.9% (White et al., 2017).Lower prevalences have been recorded in some African countries 12.3%, 7.3%, and 21.1%, for Benin, Nigeria, and Ethiopia respectively (Kassa et al., 2016;Mekonnen et al., 2017;Udoh & Amodu, 2016).Since minimum acceptable diet is directly associated with optimal complementary feeding, the findings above show that complementary feeding is a challenge in developing countries and progress towards meeting the standard for optimal complementary feeding is slow.These underscore the relevance of examining current feeding practices which may be critical for achieving optimal nutritional status, in addition, malnutrition is a public health concern in Ghana (GSS et al., 2015).Hence, there is the need to identify and promote feeding practices that have the potential to improve child nutritional status, yet, there is little data on the relationship between child feeding practices and nutritional status in the middle belt of Ghana, specifically, Techiman municipality.Hence, this study sought to assess child feeding practices and their association with undernutrition among children 6 to 23 months old in the Techiman municipality of Ghana.

| Key message
• One-third of mothers experienced feeding challenges when feeding children, and one out of every six children was bottle fed.
Feeding challenges were categorized into three: (1) Food insecurity; (2) Vomiting and refusal to/after eat(ing); and (3) Inadequate knowledge on appropriate child feeding practices and lack of time.
• Early initiation of breastfeeding and bottle feeding were associated with acute malnutrition and experiencing feeding challenges was associated with stunting.Child feeding indicators such as minimum dietary diversity, minimum meal frequency, and minimum acceptable diet were not associated with malnutrition.
• The findings could be relevant for nutrition educators and nutrition programmers, as efforts could be tailored towards improving infant and young child feeding practices and strengthening existing programs such as growth monitoring, and promotion activities in child welfare clinics.

| Study area
The study was conducted in the Brong Ahafo region (which has been recently divided into Bono East and Ahafo regions) which has two farming seasons.It lies between the northern and southern part of Ghana, mostly regarded as the middle belt.The Brong Ahafo region is one of the major producers of cocoa and timber in the forest zone of Ghana.

| Design, target population, and sampling
The analytical cross-sectional design was used among mothers/ caregivers with children, aged 6-23 months, attending child welfare clinics in Techiman municipality, Ghana.Eligible study participants who declined consent and sick children were excluded from the study.Decision on consent was taken by mothers/caregivers; sickness was defined by observation, self-report of mothers/caregivers, consultation with healthcare personnel and child health records booklets.A sample size of 403 was obtained using the formula for sample size determination with single proportion (Hawkins, 1989); an assumed prevalence of 50% for undernutrition, a critical value of 1.96 at 95% confidence level, and a 0.05 margin of error.Simple random sampling was used to select 8 health facilities (Grace Hospital, Opoku Agyemang Hospital, Holy Family Hospital, Tanoso Health Centre, Bamire Health Centre, Mt. Olive Hospital, Atabuoso Health Centre, and Akrufi Health Centre) from a total of 14 health facilities in the municipality, and the 403 participants were selected from these facilities using proportional stratification.

| Assessment of child feeding indicators
A single 24-h recall was used to collect data on children's dietary intake.Mothers were asked to recall foods that were given to their children in the last 24 h, and the various foods mentioned were classified into seven food groups; Cereals; Legumes and Nuts; Dairy products; Flesh foods; Eggs; Vitamin A rich fruits and vegetables; Other fruits and vegetables.Children were given a score of 1 for consuming food from a food group and 0 for otherwise.The sum of the individual food groups gave dietary diversity score, the maximum score for a child was 7 and the minimum score was 0 when a child did not consume from any of the food groups.Children whose dietary diversity scores were ≥4 were considered to have met their MDD; children who were 6-8 months old or 9-12 months old, were still breastfed, and had eaten a solid or semi-solid food 2 and 3 times in a day respectively were considered to have met their MMF.Children aged 6-23 months who were not breastfed and had eaten a solid or semi-solid food 4 times in a day were also considered to have met their MMF.Children who met both MDD and MMF were considered to have met their MAD.All these computations were performed in line with WHO guidelines (WHO, 2010).Early initiation of breastfeeding, bottle feeding, feeding challenges, introduction of solid or semi-solid foods at 6 months, continued breastfeeding at 1 year, and giving colostrum to the child were also assessed by asking mothers to respond to yes or no questions on the aforementioned indicators.
For those who said they experienced feeding challenges, an openended question (what challenges did you experience?) was used to probe in order to understand the challenges they experienced.

| Assessment of nutritional status
Length of each child was taken with a height measuring board and weight was measured using an electronic weighing scale (SECA).All measurements were recorded to the nearest 0.1 cm or 0.1 kg for height and weight respectively.Ages of children were confirmed from records on date of births in child health record booklets.The height/length, weight, and age were used to compute anthropometric z-scores.Anthropometric z-scores of <−2 standard deviations (−2SD) from the median of the WHO Growth Standards were defined as stunting, wasting, and underweight (WHO, 2006).Data on the kind of feeding challenges participants experienced were analyzed using thematic analysis.

| Independent variables
The independent variables were child's age, place of delivery, and child feeding practices.The child feeding practices measured in this study were initiation of breastfeeding, colostrum use, continued breastfeeding at 1 year, foods introduced before breastmilk, introduction of solid and semi-solid foods at 6 months, feeding challenges experienced by mothers, bottle feeding, MDD, MMF, and MAD.

| Dependent variables
The dependent variable was undernutrition which was measured using stunting, wasting, and underweight.

| Data analysis
Data were entered, cleaned, and analyzed using Statistical Package for the Social Sciences (SPSS), version 21 (IBM Inc.).Microsoft Excel 2014 and WHO anthro software were used to calculate weight-forheight z-scores (WHZ), weight-for-age z-scores (WAZ), and lengthfor-age z-scores (LAZ).Categorical variables were presented as frequencies and percentages, and continuous variables were presented as means and standard deviations.First-line analysis was performed using Chi-square/Fisher's Exact test to identity child feeding indicators that affected malnutrition (See Table S1).Variables with p-values <.2 were fitted in the multiple logistic regression models (Hosmer Jr et al., 2013;Sitotaw et al., 2019) to determine their independent contributions to children's nutritional status.The goodness of fit of the models was assessed using Hosmer & Lemeshow test; stunting model (X 2 = 6.32, p = .61),wasting model (X 2 = 1.66, p = .80),and underweight model (X 2 = 2.18, p = .70),which indicated acceptable goodness of fit.Statistical significance was set at p < .05for the logistic regression analyses.

| Socio-demographic characteristics of respondents
A total of 403 participants were interviewed.About 46% of children were between 12 and 23 months old, 53% of them were males and 83% of them were born in health facilities.More than half of the mothers (52.6%) had low educational background and 8 out of every 10 mothers were gainfully employed (Table 1).

| Distribution of dietary practices and nutritional status by age
Children who met their MDD were 44%, out of this percentage, 65% were 12 to 23 months old (Table 2).Children who met their MMF were 56%, and six out of every ten of them were 12 to 23 months old.Thirty-six percent of them met their MAD and 25% of these children were 9 to 11 years old.Twenty-one percent (20.6%) of all children were stunted and 69% of the stunted children were 12 to 23 months old, four out of ten children who were wasted were 6 to 8 months old (11% of all children were wasted) and 15% of underweight children were 9 to 11 months old (10% of all children were underweight) (Table 2).

| Feeding characteristics of children
The percentage of children that were fed within the first hour of birth was 80.4%, less than one percent (0.7%) of them were not given colostrum and nine out of every ten (93.5%) were still breastfed at 1 year.Nearly all (98%) children were not given any food before breastmilk and more than three-quarter (79.7%) of them started eating solid and semi-solid foods at 6 months old.One-third of mothers/caregivers experienced feeding challenges when feeding their children and 84% of children were not bottle fed (Table 3).Feedback from probing on the feeding challenges experienced were grouped into three: (1) Food insecurity; (2) Vomiting and refusal to eat/ vomiting after eating; and (3) Inadequate knowledge on appropriate child feeding practices and lack of time.

| Multivariable analysis of child feeding factors and nutrition status of children
In Table 4, the logistic regression analysis revealed that children 6 to 8 months old were 71% less likely to be stunted compared to children

| DISCUSS ION
The study sought to investigate the association between child feeding practices and undernutrition, specifically, stunting, wasting, and underweight, and found that not giving breastmilk to children within first hour of delivery and practicing bottle feeding were positively associated with wasting.The study also revealed that younger child age and experiencing feeding challenges were protective against stunting.
In the current study, the percentage of children that was bottle fed was almost 18% and 43% higher than the number of children who were bottle fed in studies by Jemide et al. (2016) and Abeway et al. (2018) in Nigeria and Ethiopia respectively.Degefa et al. (2019) also reported bottle feeding prevalence of 35.1% in Burayu, Ethiopia.The differences in these findings may be because of tailored health messages on dangers of bottle feeding given to mothers in the current study during growth monitoring and promotion service at child welfare clinics.The percentage of children introduced to breastmilk within the first hour of birth was 80% in the current study which confirmed the findings of Abeway et al. (2018) but much higher than the average figure reported for urban and rural communities in Bangladesh (51%) (Sheikh et al., 2020).The current study also reported a high percentage of children being introduced to solid and semi-solid foods at 6 months, this was in line with the finding of Jemide et al. (2016).Sheikh et al. (2020) also reported that 65% of children were introduced to solid and semi-solid foods in Bangladesh.
The percentage of children that met their MAD was higher than other studies conducted in Ghana (Saaka et al., 2015) and Nigeria (Jemide et al., 2016).The variation in findings may be because health workers in the current study became more experienced and savvy with nutrition counseling in the course of providing growth monitoring and promotion services at child welfare clinics over time.
However, the findings of Saaka et al. (2015) on proportion of children meeting their MMF was similar to the finding of the current study.This implies that more children would meet their energy TA B L E 1 Socio-demographic characteristics of participants.requirements since MMF is a proxy for measuring energy intake (Mekonnen et al., 2017).

Characteristic
The current study did not show any association between the WHO main child feeding indicators; MDD, MMF, and MAD, and malnutrition in the multivariable logistic regression analyses.Our findings were similar to studies from Cambodia (Reinbott et al., 2015) and Ethiopia (Tessema et al., 2013).Additionally, another study in Ghana (Saaka et al., 2015) did not show any association between minimum dietary diversity, minimum meal frequency, and stunting.On the contrary, some studies indicated increased risk for stunting and underweight among children who did not meet their minimum dietary diversity (Fekadu et al., 2015;Khamis et al., 2019).Other studies by Kimwele and Ochola (2017) and Girma et al. (2019) showed association between minimum meal frequency and dietary diversity as well as all three forms of malnutrition.Similarly, in Nigeria, increased odds of underweight and stunting among children who did not meet their minimum dietary diversity were noted (Udoh & Amodu, 2016).Saaka et al. (2015) suggested that the lack of association between stunting and MDD, MMF, and MAD may be because the feeding indicators are not sensitive to chronic malnutrition.
Child feeding indicators such as early initiation of breastfeeding and bottle feeding remained significantly associated with wasting after multivariable logistic regression analysis.The current study established that bottle feeding increased the risk of wasting compared to children who were not bottle fed.This confirm findings from other studies that suggested increased odds of wasting among bottle fed children (Ejigu et al., 2018;Liben et al., 2016).Our study also showed an increased risk of wasting among children who were not breastfed within the first hour of birth.Marriott et al. (2012)reported significant association between early initiation of breastfeeding and protection against underweight; their study also showed that children face a greater risk of underweight if breast feeding does not continue to between 12 and 15 months.Underweight may reflect wasting or stunting (Prendergast, 2015), hence, it is important to adhere to WHO recommendations on child feeding to avoid acute and subsequent malnutrition.Other studies also showed increased risk of stunting among children who were not breastfed within the first hour of birth (Kismul et al., 2018;Muchina & Waithaka, 2010;Shine et al., 2017;Simanjuntak et al., 2018).
Interestingly, the findings of the current study indicate that children of mothers who experienced feeding challenges were protected against stunting.It would make sense to suggest that these mothers probably visited health facilities and nutrition experts more, and the present observation could be a reflection of knowledge and skills they acquired from health personnel and child feeding support groups on how to overcome such challenges.Considering the total number of children who were appropriately fed in this study, and the high prevalence of malnutrition, there is the need for more efforts in the area of social behavior change communication among mothers and pregnant women, social behavior change materials such as posters and calendars could be made more readily available than it is currently.In addition, health workers or community health volunteers could make regular follow-ups to ensure that mothers are implementing all that they learn at child welfare clinics.
This has the potential to enhance uptake of key nutrition messages and also improve access to child welfare clinic services which has been suggested to improve infant and young child feeding (Aborigo et al., 2012;Apanga, 2015;Otoo et al., 2009).As has been shown in the current study, inappropriate child feeding practices namely giving the child breastmilk after the first hour of delivery and bottle feeding increase the risk of undernutrition.Therefore, promoting appropriate child feeding practices particularly early initiation of breastfeeding can reduce the risk of undernutrition.
This study had some limitations.For instance, its cross-sectional nature made it impossible to determine the effect of child feeding practiceson malnutrition.Also, mothers/caregivers reliance on memory during 24-h recall could have introduced some recall bias but several probing techniques were used to minimise recall bias.That notwithstanding, we believe the current study establishes that child feeding practices such as early initiation of breastfeeding, bottle feeding, and mothers/caregivers experience of feeding challenges are associated with malnutrition among children.The findings could be relevant for nutrition educators and nutrition programmers, as efforts could be tailored towards improving infant and young child feeding practices and strengthening existing programs such as growth monitoring and promotion activities in child welfare clinics.
TA B L E 4 Multivariable analysis of child feeding factors and nutrition status of children.
Feeding characteristics of children.