Prevalence and predictors of infant and young child feeding practices in sub-Saharan Africa

Abstract Background This study assessed the prevalence and predictors of minimum dietary diversity (MDD), minimum meal frequency (MMF), and minimum acceptable diet (MAD) in sub-Saharan Africa (SSA). Methods A sample of 87 672 mother–child pairs from the 2010–2020 Demographic and Health Surveys of 32 countries in SSA was used. Multilevel binary logistic regression analysis was carried out to examine the predictors of MDD, MMF, and MAD. Percentages and adjusted odds ratios (aORs) with a 95% confidence interval (CI) were used to present the findings. Results The prevalence of MDD, MMF, and MAD in SSA were 25.3% (95% CI 21.7 to 28.9), 41.2% (95% CI 38.8 to 43.6), and 13.3% (95% CI 11.6 to 15.0), respectively. Children aged 18–23 months were more likely to have MDD and MAD but less likely to have MMF. Children of mothers with higher education levels were more likely to have MDD, MMF, and MAD. Children who were delivered in a health facility were more likely to have MDD and MAD but less likely to have MMF. Conclusions Following the poor state of complementary feeding practices for infants and young children, the study recommends that regional and national policies on food and nutrition security and maternal and child nutrition and health should follow the internationally recommended guidelines in promoting, protecting, and supporting age-appropriate complementary foods and feeding practices for infants and young children.


Introduction
The United Nations Convention on the Rights of the Child stated that every child has the right to good nutrition. 1This implies that good nutrition should not be a privilege, but is a necessity for every child.However, this right is yet to be met for many children, as the World Health Organization (WHO) reported that in 2020, about 149 million children < 5 y of age were estimated to be stunted (too short for their age), 45 million were wasted (too thin for their height), and 38.9 million were overweight or obese. 2 It has been reported that few children receive nutritionally adequate and safe complementary foods, emphasizing that in most countries, less than a quarter of infants aged 6-23 months met the criteria for dietary diversity and feeding frequency that is International Health appropriate for their age. 35][6] It is worth noting that a key determinant of the nutritional status of young children or infants in SSA is their feeding practices. 7 , 8WHO has several indicators for measuring infant and young child feeding practices.These include minimum dietary diversity (MDD), minimum meal frequency (MMF), and minimum acceptable diet (MAD). 9DD is the proportion of children 6-23 months of age who receive foods from a minimum of five of the eight groups of foods, namely breast milk, grains, roots and tubers; legumes and nuts; dairy products; flesh foods (meat, fish, poultry and organ meats); eggs; vitamin A-rich fruits and vegetables; and other fruits and vegetables. 10 , 11MMF is the proportion of children 6-23 months of age who receive solid, semisolid or soft foods (but also including milk feeds for non-breastfed children) the minimum number of times or more (two feedings of solid, semisolid or soft foods for breastfed infants ages 6-8 months; three feedings of solid, semisolid or soft foods for breastfed children ages 9-23 months; and four feedings of solid, semi-solid or soft foods or milk feeds for non-breastfed children ages 6-23 months, whereby at least one of the four feeds must be a solid, semisolid or soft food). 10 , 12Lastly, MAD refers to receiving at least the MDD and MMF for their age during the previous day for breastfed children and for non-breastfed children receiving at least the MDD and MMF for their age during the previous day as well as at least two milk feeds. 10 , 135][16] These inappropriate feeding practices account for more than two-thirds of child and infant mortality in SSA. 17 Children who do not receive sufficient dietary diversity, meal frequency or acceptable diet after 6 months of age are at a higher risk of becoming stunted, despite having been optimally breastfed. 11Moreover, the findings of Masuke et al. 17 indicated that inappropriate feeding practices are associated with a higher risk of stunting, wasting, and underweight among young children.Also, Yarnoff et al. 18 reported that inappropriate feeding practices are associated with diarrhoea, fever and cough among infants and young children in SSA.The quality of feeding practices in infants and young children is dependent on the frequency of the meal, the food groups contained in the diet and the acceptability of the diet.However, most infants and young children are introduced to normal household foods, which are predominantly cereals and starchy foods, and are poor in quality. 19 , 20he 2014 report of the International Food Policy and Research Institute on good global nutrition reported that to achieve nutritional requirements and prevent deficiencies, child feeding practices in the first 2 y of age need special attention, especially in SSA. 21There have been some studies conducted on infant and young child feeding practices in SSA, 22 , 23 but these studies have limited generalizability due to the small sample sizes and different methods of assessment.The present study sought to fill the gaps in the literature accordingly.This study aimed at assessing the prevalence of MDD, MMF, and MAD in SSA at the subregional level, as well their predictors.The findings of this study will provide evidence to improve the dietary practices of infants and young children in SSA in line with the Sustainable Development Goal 3.2. 24

Study design and data description
Our study was multicountry and involved a cross-sectional data analysis of the Demographic and Health Surveys (DHS) conducted from 2010 to 2020 in 32 sub-Saharan African countries.The data were extracted from the children's file (KR) of the 32 countries.The DHS is a nationally representative and comparative survey conducted every 5 y in > 90 low-and middle-income countries (LMICs) globally. 25Respondents for the survey were selected using a two-stage cluster sampling technique.The detailed sampling procedure has been published elsewhere. 26The DHS employed a standardized, structured questionnaire to collect data from respondents on health indicators such as child nutrition and feeding practices. 25In the present study, we included 87 672 mother-child pairs in the final analysis (see Table 1 ).The dataset for the 32 countries used is freely available from https://dhsprogram.com/data/available-datasets.cfm .

Outcome variable
The outcome variables for this study were the three core indicators for infant and young child feeding practices measured as MDD, MMF and MAD.These measures were defined per the WHO's requirements 2 , 27 and were categorized as 'yes' and 'no'.

Explanatory variables
A total of 18 explanatory variables were included in the study.These variables comprised of the characteristics of the child and the mother and were grouped into individuallevel, household-level, and contextual-level variables.The International Health variables were selected based on their association with MDD, MMF, and MAD from previous literature, 9-13 , 20-23 , 28-37 as well as their availability in the DHS dataset.For the individual-level variables, we maintained the existing coding for the sex of the child, mother's age (years), level of education, marital status, current working status and postnatal care attendance (PNC) as found in the DHS dataset.The other variables were recoded as age of child (6-8, 9-11, 12-17, 18-23 months), birth order (1, 2-4, 5 + ), size of the child at birth (large, average, smaller), antenatal care (ANC) attendance (none, 1-3, 4 + ) and place of delivery (home, health facility, other).The coding for the household-level variables included household size (small, medium, large), frequency of watching television (not at all, less than once a week, at least once a week), frequency of listening to radio (not at all, less than once a week, at least once a week), frequency of reading a newspaper/magazine (not at all, less than once a week, at least once a week), and wealth index (poorest, poorer, middle, richer, richest).The contextual-level variables consisted of place of residence (urban/rural) and geographic subregion (south, central, east, west).

Statistical analyses
We performed both descriptive and inferential analyses using Stata version 16.0 (StataCorp, College Station, TX, USA) in this study.Descriptively, percentages with their corresponding confidence intervals (CIs) were used to present the prevalence of MDD, MMF, and MAD using forest plots.Cross-tabulation was used to determine the distribution of the outcome variables across the explanatory variables.Pearson's χ 2 test of independence was employed to examine the relationship between the outcome variables and the explanatory variables.All the variables that had a p-value < 0.05 were regarded as statistically significant.To obtain the best combinations of variables as predictors, we employed the 'best subset variable selection method' to select the explanatory variables for the regression analysis.In doing this, we used the Stata command "gvselect".A detailed description of the best selection method is provided in the literature. 38 , 39Log-likelihood, Akaike information criterion (AIC), and the Bayesian information criterion (BIC) were used to present the results of the best selection method.The combination of variables with the lowest AIC was selected for the regression analysis.
A multilevel binary logistic regression was conducted to determine the factors associated with each of the outcome variables.Five models (Models O-IV) were built to examine the predictors of the outcome variables.Model O was built to contain the outcome variable with the results indicating the variance in each of the outcome variables attributed to the clustering of the primary sample units (PSUs).Models I, II and III were built to contain the individual-level, household-level and contextual-level variables, respectively.The final model (Model IV) was fitted to include all the explanatory variables.The results of the regression analyses were presented as adjusted odds ratios (aORs) with their corresponding 95% CIs.The model fitness and comparison were checked using the AIC, with the lowest AIC showing the best-fitted model.Statistical significance was set at p < 0.05.The writing of this manuscript was guided by the Strengthening Reporting of Observational Studies in Epidemiology reporting guidelines. 40We applied the sample weights to obtain unbiased estimates according to the DHS guidelines.Also, the Stata survey command svy was used to adjust for the complex sampling structure of the data in the χ 2 and regression analyses.

Results
Prevalence of infant and young child feeding practices (MDD, MMF, and MAD) in SSA

Discussion
This study examined the prevalence and predictors of MDD, MMF, and MAD in 32 countries in SSA.The prevalence of MDD, MMF, and MAD in SSA were 25.3%, 41.2%, and 13.3%, respectively.The low prevalence of MDD, MMF and MAD recorded in this study are similar to what was recorded in 80 LMICs, 41 49 LMICs 42 and 48 LMICs. 43A possible reason for the similarities in the findings could be attributed to the similarities in the socio-economic status of the LMICs and the employment of a large sample size. 42 , 43ariations were also detected for the prevalence of infant and child feeding practices among the studied countries.The prevalence of MDD ranged from 5.8% in Burkina Faso to 49.4% in South Africa.This could be attributed to the lower proportion of women reaching the MDD in Burkina Faso. 31The study also found that while Liberia had the lowest (25.4%) prevalence of MMF, Lesotho recorded the highest (59.3%).The level of poverty in Liberia could also account for this finding. 28For MAD, Burkina Faso recorded the lowest prevalence (3.9%) while Rwanda had the highest (22.3%), probably because of the mothers' low level of perceived selfefficacy to provide the daily required food groups for their children. 44he prevalence of MDD, MMF, and MAD vary across different countries in SSA.For example, a study conducted in Ghana found the prevalence of MDD to be 35.6%, 45whereas a survey in Ethiopia found a lower prevalence of MDD, which was 10.8%. 22n Malawi, a 27.7% prevalence of MDD was found.For MMF, an Ethiopian survey found the prevalence to be 44.7%, 22whereas a study in Nigeria reported a prevalence of 56%. 46The prevalence of MAD was found to be as low as 7.0% in a study conducted in Ethiopia, 47 while in Nigeria, a study reported the prevalence of MAD to be 8.0%. 46tudies conducted in different sub-Saharan African countries have reported the factors associated with MDD to include the mother's education, wealth quintile, urban residence, home gardening, media exposure, mother's knowledge of dietary diversity, employment status, household assets, and optimal household water access. 11 , 22 , 23 , 48 , 49On the other hand, factors such as child age, parity of the mother, wealth quintile, mother's education  12 , 28 , 41 , 46 The factors that are associated with MAD across different parts of SSA include ANC visits, mother's education level, household wealth quintile, age of the child, sex of the child, mother's media usage, mother's working status, and birth interval. 28 , 46he study found that infants and young children who were 18-23 months of age were more likely to have MDD and MAD.This finding is similar to the findings of previous studies. 22 , 44 , 45n explanation for these findings could be a result of an increase in the consumption of other food groups as children age. 22 , 44owever, those who were 18-23 months were less likely to have MMF.It is possible the reduction in exclusive breastfeeding practices as children grow could have influenced their likelihood of receiving the recommended MMF compared to their younger counterparts. 50imilar to the findings of previous studies, 23 , 51-53 this study found that infants and young children whose mothers had higher education levels were more likely to have MDD, MMF, and MAD.This finding could be that women who are more educated are better equipped with knowledge about providing their children with the appropriate complementary foods they require, increasing their likelihood of adequately feeding their children. 23 , 51 , 53his finding could be because women who are more educated are mostly employed, which suggests that they may have a greater ability to afford the variety of required food groups their children need for optimum growth and development. 23 , 51 , 53orroborating the findings of previous studies, 53 -55 the study found that infants and young children who were delivered in a health facility were more likely to have MDD and MAD.A potential explanation for this finding could be that women who deliver in a health facility receive some education about complementary feeding practices, increasing their likelihood of adequately feeding their children. 53 , 54However, those who were delivered in a health facility were less likely to have MMF.This finding was also confirmed by a previous study in Tanzania. 53A possible reason for this could be that mothers continuously practice exclusive breastfeeding as children grow instead of introducing complementary foods. 53oreover, infants and young children whose mothers attended PNC were more likely to have MDD, MMF and MAD.The finding of this study is similar to the findings of previous studies. 23 , 53 , 55Women who frequently visited a health facility could have benefited from the education given to mothers regarding choosing and providing complementary feeding for their children to aid their development while reducing the occurrence of preventable diseases. 53 , 55imilar to the findings of other studies, 22 , 51 , 56 our study found that infants and young children whose mothers were exposed to mass media were more likely to have MDD, MMF and MAD.Our finding could imply that women who are exposed to mass media are educated through this means regarding the appropriate feeding practices available to them, increasing their likelihood of adequately feeding their children. 22 , 51orroborating the findings of previous studies, 21 -23 , 51-53 this study found that infants and young children whose mothers are of the richest wealth index were more likely to have MDD, MMF and MAD.Our finding could be the fact that wealthy households are able to afford and provide a variety of the required food groups to facilitate proper growth and development of young children, increasing their likelihood of being adequately fed. 21 , 22 , 52imilar to the findings of previous studies, 51 , 56 this study found that infants and young children whose mothers reside in rural areas were less likely to have MDD and MAD.Our finding may be due to economic reasons, given that rural areas may be less developed and individuals living there may not be as educated and financially able, which may reduce their likelihood of adequately feeding their children. 51 , 56

Strengths and limitations
The use of a relatively large sample size of nationally representative samples from several countries makes the findings of the study more generalizable to the study populations used in this study.However, the study has some limitations.First, the study was cross sectional, thus causal interpretations of the findings are limited.Again, the variables used were self-reported, thus respondents might have under-or overreported the feeding practices of their children, which could influence the findings.

Conclusions
The study found that the prevalence of MDD, MMF, and MAD among infants and young children in 32 countries in SSA remains low.Variations in the prevalence of feeding practices among infants and young children in the studied countries were also identified.The study also identified the factors that are associated with MDD, MMF, and MAD.Public health interventions aimed at improving complementary feeding practices among infants and young children in SSA should focus on the factors identified in this study.Regarding the poor state of complementary feeding practices for infants and young children, the study recommends that regional and national policies on food and nutrition security and maternal and child nutrition and health follow the internationally recommended guidelines in promoting, protecting and supporting age-appropriate complementary foods and feeding practices for infants and young children.

Figure 1 .
Figure 1.Forest plot showing the prevalence of MDD in SSA.

Figure 2 .
Figure 2. Forest plot showing the prevalence of MMF in SSA.

Figure 3 .
Figure 3. Forest plot showing the prevalence of MAD in SSA.

Table 1 .
Description of the study

Table 2 .
Bivariate analysis of predictors of infant and young child feeding practices in SSA Table2shows the results of the bivariate analysis of predictors of MDD, MMF, and MAD in SSA.The study found that the age of the child, birth order, size of the child at birth, maternal age, maternal education level, current working status, ANC, place of delivery, PNC, marital status, household size, frequency of watching television, frequency of listening to radio, frequency of reading a newspaper/magazine, wealth index, place of residence and geographic subregion were significantly associated with MDD and

Table 3 .
Predictors of MDD among children in SSA

Table 3 .
Continued MAD.The age of the child, size of the child at birth, maternal education level, ANC, place of delivery, PNC, marital status, household size, frequency of watching television, frequency of listening to radio, frequency of reading a newspaper/magazine, wealth index, and place of residence were associated with MMF.

Table 4 .
Continued Confidence Interval; * p < 0.05, ** p < 0.01, *** p < 0.001; 1 = Reference category; PSU = Primary Sampling Unit; ICC = Intra-Class Correlation; AIC = Akaike's Information Criterion level, maternal age, mother watching television, size of the baby, mode of delivery, and health service contact have been reported to be associated with MMF in different countries in SSA.