Individual and community-level factors associated with animal source food consumption among children aged 6-23 months in Ethiopia: Multilevel mixed effects logistic regression model.

BACKGROUND: Diversified diet in childhood has irreplaceable role for optimal growth. However, multi-level factors related to low animal source food consumption among children were poorly understood in Ethiopia, where such evidences are needed for decision making.
OBJECTIVES: To investigate the magnitude and individual- and community-level predictors of animal source food (ASF) consumption among children aged 6-23 months in Ethiopia.
METHODS: We utilized a cross-sectional pooled data from 2016/19 Ethiopia Demographic and Health Surveys. A stratified two-stage cluster design was employed to select households with survey weights were applied to account for complex sample design. We fitted mixed-effects logit regression models on 4,423 children nested within 645 clusters. The fixed effect models were fitted and expressed as adjusted odds ratio with their 95% confidence intervals and measures of variation were explained by intra-class correlation coefficients, median odds ratio and proportional change in variance. The deviance information criterion and Akaike information Criterion were used as model fitness criteria. RESULT: in Ethiopia, only 22.7% (20.5%-23.9%) of children aged 6-23 months consumed ASF. Younger children aged 6-8 months (AOR = 3.1; 95%CI: 2.4-4.1), home delivered children (AOR = 1.8; 1.4-2.3), from low socioeconomic class (AOR = 2.43; 1.7-3.5); low educational level of mothers (AOR = 1.9; 95%CI: 1.48-2.45) and children from multiple risk pregnancy were significant predictors of low animal source consumption at individual level. While children from high community poverty level (AOR = 1.53; 1.2-1.95); rural residence (AOR = 2.2; 95%CI: 1.7-2.8) and pastoralist areas (AOR = 5.4; 3.4-8.5) significantly predict animal source food consumption at community level. About 38% of the variation of ASF consumption is explained by the combined predictors at the individual and community-level while 17.8% of the variation is attributed to differences between clusters.
CONCLUSIONS: This study illustrates that the current ASF consumption among children is poor and a multiple interacting individual- and community level factors determine ASF consumption. In designing and implementing nutritional interventions addressing diversified diet consumption shall give a due consideration and account for these potential predictors of ASF consumption.

Background

Under-nutrition remains a significant public health concern in Ethiopia among children [1], with 36.8%, 21.1%, and 7.2% of children are victims of stunting, underweight and wasting by 2019 and its adverse consequences respectively [2]. Major burden of malnutrition starts during pregnancy and early childhood where diversified and nutritious dietary habit could promote optimal growth, cognitive development and economic productivity [3].

The first 1000 days of early life is a critical period where the impacts of malnutrition will be devastating and proceeds to adulthood [4, 5]. Among these age groups, infant should be fed exclusively for the first six months and complementary feeding with a diversified diet is recommended starting from six months. Also, the infancy and the early childhood period is a risky period for widespread stunting and wasting due to problems in child feeding [6, 7]. Undiversified food consumption is one of the major predisposing factors for children aged 6 to 23 months of age [8].

Diversified diet in early childhood has irreplaceable role for optimal growth. Animal source foods (ASFs) have a high amount of critical nutrients that are more bio-available than nutrients from plants source foods [9]. ASFs are better sources for high quality proteins and essential amino acids [10], better and optimal child growth. In addition, ASFs can significantly improve nutritional status especially in children [11], due to the fact that it provides a high concentration of indispensable macro and micronutrients, to meet their daily requirements [12, 13].

ASFs (now onwards refers to consumption of egg and/or other flesh foods) have been linked to improved nutritional status in children, including reduced stunting [9, 14]. Prevalence of stunting is significantly lower in children who had consumed ASFs in the last 24 hours as compared to those who had not [10]. A review study indicated that better ASF consumption is linked to a positive physically observable anthropometric advantages and improved cognitive function among undernourished children from low-income settings [13]. A review on intervention trial of ASF supplement for children showed to improvement in the length, weight and Height for age Z scores [15].

Despite a better ASF production in the country, the rapidly rising food prices and poor economical access to diversified ASF consumption makes ASF consumption low, where the diet of many rural households were limited to maize and legumes [16]. Taking the burden of under-nutrition in early childhood and the limited diversified dietary habit and consumption into consideration, consumption of ASF could significantly contribute to the overall quality of diet and the long-term strategies to control malnutrition in the country.

However, the practice of ASF consumption is a function of various factors which might have a potential to hinder consumptions. So far, limited attempts were tried to understand the problem with the use of small-scale surveys [17-19], despite the multifaceted nature of the issue, where individual or household level factors only may not be sufficient. Hence, analysis approach considering individual and community level factors and design could allow to better understand the drivers of ASF consumption. In low-income settings like Ethiopia cultural factors like decision making power of women’s, religious factors like fasting and socio-economic factors like wealth status, ownership of livestock and other factors like low level of nutrition knowledge were among the reported factors known to affect ASF consumption [20, 21].

Hence, these factors could easily be understood if we could analyze using multi-layered approach like community/cluster level, household level and individual level factors [12, 22] at national scale where this gear towards informed decisions. Besides, the result may be used to design strategies and formulate programs that might improve the level of ASF consumption in young children. Previous study by Pott et.al indicated that limited individual level factors (religion, residence, child age) were associated with a better ASF consumption [20]. In addition, Workicho et.al identified residence, livestock ownership, socioeconomic class and literacy positively predicted better ASF consumption [16]. However, these studies were limited in some regions and does not involve a more robust assessment of community level factors, which may determine the ASF consumption due to the culture, livestock availability or other potential factors [23]. Furthermore, the effect of factors like high-risk fertility behaviors and multi-level factors on ASF consumption of children was poorly investigated, where studies showed that 57% to 72% of women had high risky fertility behavior [24, 25], which may predispose to a widespread food insecurity and malnutrition among children [24] Thus, this study aims to determine the level of ASF consumption and factors affecting it among children aged 6–23 months in Ethiopia.

Methods

Data source, study design and sample points

This study utilized a cross-sectional pooled data from Ethiopia Demographic and Health Surveys (EDHS) conducted in 2016 and 2019. The data were extracted from www.measuredhs.com.The data were nationally-representative population-based household surveys. The standard DHS have large sample size (usually between 5,000 and 30,000 households) carried out about every 5 years [26]. A community-based cross-sectional study design was used. The DHS surveys are based on a stratified two-stage cluster sampling design, where independent multi-stage samples are selected per strata. Within each stratum implicit stratification is applied to make sure that the selected primary sampling units are representative of different geographic levels and areas. Stratified primary sampling units (clusters) were sampled in the first stage and households in the second stage [27]. This two-stage sampling points allows to have a representative sample with a reduced sampling errors and appropriate coverage for target population. Sample size for this complex survey with clustering estimated with design effect (Deft). To prevent bias, no replacements or changes of the preselected households were allocated in the implementing stages [27]. The EDHS surveys used sample weights to account for complex survey design, survey non-response, and post-stratification for representativeness of the samples. The study population for this study were youngest living child age 6–23 month who is living with the mother (KR file) 24 hours preceding the interview. After data cleaning and exploration, a total weighted sample of 4,423 children aged 6–23 months were included in the survey and in our analysis.

Measures of variables

Dependent variable

According to WHO and UNICEF, ASF consumption among children age 6–23 months is defined as the percentage of children 6–23 months of age who consumed egg and/or flesh food on the previous day. This indicator is based on consumption of food groups 5 (flesh foods) and 6 (eggs) described in indicator 8 on minimum dietary diversity [28]. Children are counted as “consumed ASF” if either food group has been consumed, otherwise children are counted as or “not consumed ASF” [29, 30]. Hence, the dependent variable (outcome variable) is dichotomized as (“0”—consumed ASFs, “1”—do not consume ASFs).

Independent variables

Based on reviewed literature, both individual and community-level predictor variables were considered in our analysis. From the individual-level variables, child’s factor (age of child, sex of child, previous birth interval, and birth order), maternal factors (high-risk fertility behaviors, maternal age at birth) [24, 25, 31], socioeconomic factors (wealth index, maternal education, maternal occupation, exposure to media), and health service factors (place of delivery, and antenatal visit). We adopted the concept of high-risk fertility behaviors from DHS surveys, which considers three parameters, mother’s age at birth, birth order, and birth interval, to define high-risk fertility behaviors. The high-risk fertility behaviors were categorized as: no extra risk, unavoidable risk, single high-risk and multiple high-risk. The presence of any of the following 4 parameters was considered as a single high-risk fertility behavior: mother’s age <18 only, mother’s age >34 only, birth interval <24 months only and birth order above three. The combinations of two or more risk parameters are referred to as multiple high-risk fertility behaviors [16, 31].

Community-level variables (community poverty level, community-level education) were created from individual-level variables by aggregating them at the cluster (community) level by using the bysort command. We obtained the proportion of each community-level characteristic and the values were ranked into tertiles as low, medium, and high. Community-level education, which was the proportion of women with secondary or higher education in the community and categorized into tertiles as low, medium, or high. Similarly, the community poverty level was categorized into tertiles and classified as low, medium, or high poverty level [32].

Data management and analysis. The unit of analysis for this study was children aged 6–23 months in pooled DHS data and the data was exported and analyzed using Stata/SE version 14.0. Sample weights were applied for descriptive statistics adjusting for non-proportional allocation of the sample and non-response rate in all analyses. This makes sample data representative of the entire population. Categorization was done for continuous variables and further re-categorization was done for categorical variables. Descriptive analysis was carried out to present the data in frequencies and percentages.

Since EDHS data is nested data (4,423 children nested within 645 clusters) and a two-level clustered dataset with a multistage sampling design, we applied multilevel modelling, which acknowledges the nesting with in the survey. In nested data (hierarchical data), analyzing variables from different levels at one single common level with ‘standard’ analysis method is inadequate, and leads to loss of statistical power and conceptual problems (ecological fallacy and atomistic fallacy) [33, 34]. Thus, we fitted a two-level multilevel mixed-effects logistic regression (random-intercept model), with the log of the probability of inadequate ASF consumption was modeled using a two-level multilevel model as follows [34]:

Yij = βo + β1x1ij + μoj + eoij where, Yij is our outcome variable (animal source food consumption): the animal source food consumption for a child aged 6–23 months living in cluster j, βo is the intercept, β1 is the coefficient of explanatory variable x1, the part of the equation involving the β-coefficients, βo + β1x1ij, is called the fixed part of the model because the coefficients are the same for everybody; the residuals at the different levels, μoj + eoij, are collectively termed the random part of the model. We fitted four models for a mixed effects modeling for nested data to determine the model that best fits the data. Null model (M0) or the intercept-only model: a model with no explanatory variables, model-I: a model with only individual-level factors, Model-II: a model with only community-level factors, and model-III: a combined model that control the effects of both individual and community-level predictor variables on ASF consumption among children aged 6–23 months. The Stata command–“meqrlogit” was used to fit these models. The results of fixed effects were expressed as adjusted odds ratio (AOR) with their 95% confidence intervals (CIs).

The measures of variation were expressed as Intra-class Correlation Coefficients (ICC) or Variance Partition Coefficients (VPC), Median Odds Ratio (MOR), and Proportional Change in Variance (PCV). The ICC and VPC can be computed for random intercept models. In the case of a random intercept model, in a logistic regression model with no predictors, the ICC or VPC equals: , for logit model, the level-1 residual variance is 3.29. Therefore, [33, 34]. The MOR is a measure of heterogeneity while the VPC (ICC) is a measure of components of variance (clustering) that considers both between- and within-cluster variance. The MOR depends directly on the area level variance (the variance of the highest-level errors) and can be computed with the following equation: MOR = exp (√(2x Vc) x 0.6745] ≈exp (0.95√(Vc)) where, Vc is the between cluster variance. The proportional change in variance (PCV) is the percentage of proportional change in variance of subsequent models with respect to the empty model. The PCV can be computed by the equation: , where VA = variance of the initial model (empty model), and VB = variance of the model with more terms (consecutive models) [35]. The deviance Information Criterion (DIC), log likelihood and Akaike Information Criterion (AIC) were used to select the best model that explained the variation in ASF consumption well. The model with the smallest AIC is chosen as the one which fits the best. Models with a lower deviance fit better than models with a higher deviance [36, 37]. Variance Inflation Factors (VIF), Standard Error (SE), and Variance Correlation Estimator (VCE) were estimated to assess risk of multicollinearity among predictor variables.

Ethical considerations

Permission to access and download the EDHS datasets was obtained from DHS program. The accessed data were used for the purpose of registered research paper only. Confidentiality of the data were kept and no effort was made to identify any household or individual respondent interviewed in the survey. The data were not passed on to other researchers without the written consent of DHS. The data were fully accessed from www.dhsprogram.com with the respect to the data sharing policy.

Results

Socio-demographic characteristics of participants (children with their mothers)

A total of 4,423 children aged 6–23 months nested within 645 clusters (primary sampling units) from 2016 and 2019 Ethiopia DHS pooled data were included in our analysis. Out of the total participants, 2,154(48.7%) were male children. About 816(18.4%) children were in the age group 6–8 months,720 (16.3%) in the age group 9–11 months, 1,636 (37%) in the age group 12–17 months, and 1,251 (28.3%) were aged 18–23 months. The majority, 3,972 (89.8%) of mothers attended up to primary education while 3,649 (82.5%) were from rural areas.

Health related characteristics

From total participants, only 1,682(38%) of mothers had attended at least four ANC visits during their recent pregnancy. Concerning the place of delivery, nearly more than half of the mothers, 2,453 (55%) gave birth at home, whereas, 43% of households were in the lowest wealth quintile. A total of 1,608 (36.4%) falling into birth with any single high-risk category and 910 (20.6%) falling into births with any multiple risk category. The highest proportion, about 1,814 (41%) and 1,229 (28%) of children were from a community where the poverty level is high and low respectively (Table 1).

Table 1

Individual-and community-level characteristics of participants, pooled data from Ethiopia DHS 2016 & 2019 (n = 4,423).

Variables	Weighted frequency (%)
Sex of child
Male	2,154 (48.7%)
Female	2,268 (51.3%)
Age of child in months
6–8 months	816 (18.4%)
9–11 months	720 (16.3%)
12–17 months	1,636 (37%)
18–23 months	1,251 (28.3%)
Mother’s age at birth
> = 18 years	4,243 (96%)
<18 years	179 (4%)
Maternal educational level
Primary education or less	3,972 (89.8%)
Secondary education and above	450 (10.2%)
Place of residence
Urban	774 (17.5%)
Rural	3,649 (82.5%)
Attended 4+ ANC visit
No	2,740 (62%)
Yes	1,682 (38%)
Place of delivery
Health facility	1,855 (42%)
Home	2,453 (55%)
Other	115 (3%)
Household wealth index
Poorest	979 (22%)
Poorer	937 (21%)
Middle	925 (21%)
Rich	797 (18%)
Richest	783 (18%)
High-risk fertility behaviors risk category
No extra risk	1,135 (26%)
Unavoidable first birth risk	770 (17.4%)
Any single high-risk category	1,608 (36.4%)
Any multiple risk category	910 (20.6%)
Accesses to all three media at least once a week
No	4,395 (99.4%)
Yes	27 (0.6%)
Community poverty level
High	1,814 (41%)
Moderate	1,380 (31%)
Low	1,229 (28%)
Community-level education
Low	4,180 (94.5%)
High	242 (5.5%)

National animal source food consumption pattern among children

In this study, a total of 22.7% (95% CI: 20.5%-23.9%) of infant and young children age 6–23 months consumed ASF (either food groups: food groups 5 (flesh foods) or 6 (eggs) has been consumed) based on the standards in Ethiopia. An estimated 8.7% and 17.5% of children consumed flesh foods and egg products during the previous day of the interview, respectively (Table 2). The percentage of children with adequately ASF consumption is highest among children from the richest household family (37.2%) and lowest from the poorest household family (14.5%). The result also showed that ASF consumption improves with increasing household wealth status (Fig 1). A large-scale regional variation (41.7% highest in Addis Ababa and 5.9% lowest in Somali) was observed in ASF consumption among children aged 6–23 months in Ethiopia (Fig 2).

Table 2

ASF consumption among children aged 6–23 months in Ethiopia: Based on pooled data from 2016 and 2019 Ethiopia Demographic and Health Survey (n = 4,423).

Diets	Frequency (n)	Percentage (%)
Animal Source Food Consumption
No	3,421	77.3%
Yes	1,002	22.7%
Food Groups 5 (flesh foods) Consumption
No	4,036	91.3%
Yes	386	8.7%
Food Group 6 (eggs) Consumption
No	3,647	82.5%
Yes	775	17.5%

Fig 1

ASF consumption by household wealth status among children aged 6–23 months in Ethiopia.

Fig 2

Interregional variation in ASF consumption among children aged 6–23 months in Ethiopia.

Individual-and community-level predictors of ASF consumption

Measures of association (fixed effects) from multilevel logistic models

Bivariable multilevel mixed-effect logistic regression was computed and factors with a lower p-value (p-value below 0.25) and/or variables with strong theoretical relation with ASF consumption [38], and previously identified predictor variables were used as a cutoff to fit multivariable multilevel mixed-effects logistic regression to control confounding effects. Consequently, age of the child (in months), having four ANC visit, place of delivery, household wealth index, maternal education level, access to all three media at least once a week, place of residence, community poverty level, community-level education, double risk (birth order 4+ & age 34+), double risk (spacing <24 & order 4+), births with any multiple risk category, any avoidable risk category, and region were eligible for multivariable analysis. We fitted four consecutive multilevel mixed-effects logistic regression models. The intercept-only model (null model: _M0) (without predictor variables), model-I (a model with only individual-level factors), model-II (a model with only community level factors) and model-III (a combined model that controls the effects of both individual- and community-level predictor variables) on ASF consumption were fitted. Ultimately, associations with a p-value below 0.05 were used to identify a statistically significant predictors of ASF consumption among children aged 6–23 months in the full model (parsimonious model).

In our analysis, age of the child (in months), at least four ANC visits, place of delivery, the household wealth index, maternal education level, community poverty level, births with any multiple risk category, access to all three media at least once a week, place of residence, high-risk fertility behaviors, and the region were statistically significant predictors of ASF consumption among children aged 6–23 months in the final model (full model). Our finding revealed that, age of the child had significantly associated with ASF consumption among children aged 6–23 months. Younger children aged 6–8 months (AOR 95% CI: 3.1(2.4, 4.1) and 9–11 months (AOR = 1.54; 95% CI: 1.20–1.97) had a 3.1- and 1.5-times increased risk of inadequate ASF consumption as compared to older children aged (18–23 months).

This study also showed that, children of mothers who attended primary educational had 1.9 times the risk of inadequate ASF (AOR = 1.90; 95% CI: 1.48–2.45) as compared to children whose mothers attended at least secondary education level. Those children whose mothers gave birth at home had 80% more risk of having inadequate ASF consumption (AOR = 1.80; 95% CI: 1.4–2.3). Additionally, the household wealth index was significantly associated with ASF consumption. Accordingly, the odds of having inadequate ASF consumption among children living in the poorest households were 2.4 times (AOR = 2.43; 95% CI: 1.70–3.50) more likely than children from the richest households. Moreover, the odds of having inadequate ASF consumption among children aged 6–23 months living in the poorer households had 70% (AOR = 1.70; 95% CI: 1.22–2.37) more risk as compared to children living in the richest household. Furthermore, those children whose mothers had less than four ANC visits during pregnancy were 1.4 times more likely to have inadequate ASF consumption (AOR = 1.40; 95% CI: 1.02–1.83) as compared to those whose mothers had at least four ANC visits during the last pregnancy. Likewise, the region was significantly associated with ASF consumption. Also, the odds of being inadequate for ASF consumption among children aged 6–23 months whose mothers had accesses to all three media at least once a week was 72% less likely to have inadequate ASF consumption than (AOR = 0.28; 95% CI: 0.12–0.65) compared to whose mothers had access none of the three media. Likewise, community poverty level was significantly associated with ASF consumption among children aged 6–23 months in Ethiopia. Accordingly, the odds of having inadequate ASF consumption among children aged 6–23 months living in high community poverty level (cluster) were 53% (AOR = 1.53; 95% CI: 1.20–1.95) more likely than children from low community poverty level. This result also revealed that, children who lived in rural areas were 2.2 times (AOR = 2.20; 95% CI: 1.70–2.80) more likely to have inadequate ASF consumption compared with children from urban areas. Similarly, children born from multiple risk category, double risk with birth order 4+ & age above 34, and double risk with spacing <24 months, order 4+ were significantly associated with ASF consumption among children 6–23 months in Ethiopia (Table 3).

Table 3

Multilevel mixed-effects logistic regression modeling of individual- and community-level factors associated with no ASF consumption among children aged 6–23 months in Ethiopia, based on pooled data from Ethiopia DHS 2016 & 2019 (n = 4,423).

Individual-and community-level characteristics	COR [95% CI)	Full model: AOR [95% CI)
Age of child (in months)
6–8	3.1 (2.36,4)	3.1 (2.4, 4.1)
9–11	1.46 (1.15,1.87)	1.54 (1.2,1.97)
12–17		1.23 (1.02,1.5)
18–23	1.25 (1.03,1.51) 1	1
4+ ANC visit
No	1.98 (1.67, 2.33)	1.22 (1.01,1.5)
Yes	1	1
Place of delivery
Health facility	1	1
Home	2.5 (2.1, 2.96)	1.8 (1.4,2.3)
Household wealth index
Poorest	5.04 (3.93,6.46)	2.43 (1.7, 3.5)
Poorer	2.43 (1.88,3.14)	1.7 (1.22, 2.37)
Middle	2.1 (1.61,2.71)	1.44 (1.03,2.0)
Richer	1.79 (1.37,2.33)	1.37 (1.01, 1.87)
Richest	1	1
Maternal education level
Primary & below	3.3 (2.67, 4.06)	1.9 (1.48, 2.45)
Secondary+	1	1
Accesses to all three media at least once a week
No	1	1
Yes	0.19 (0.08,0.43)	0.28 (0.12, 0.65)
Place of residence
Urban	1	1
Rural	2.75 (2.24,3.37)	2.2 (1.7,2.8)
Community poverty level
Low	1	1
Moderate	1.6 (1.2, 2.03)	1.3 (1.01,1.69)
High	2.27 (1.8, 2.9)	1.53 (1.2, 1.95)
Community-level education
Low	1	1
High	2.4 (1.8, 3.4)	1.1(0.8,1.5)
Birth with any multiple high-risk fertility behaviors category
Double risk, birth order 4+ & age 34+
No	1	1
Yes	1.45(1.08,1.94)	1.4 (1.02,1.83)
Double risk, spacing <24, order 4+
No	1	1
Yes	2.5 (1.7,3.7)	1.93 (1.3, 2.9)
Births with any multiple risk category
No	1	1
Yes	1.8 (1.4, 2.2)	1.4 (1.02,1.83)
Region
Tigray	1	1
Afar	8.5 (4.98,14.38)	8.2 (4.8,13.9)
Amhara	3.7 (2.4,5.6)	3.3 (2.2,5.0)
Oromia	1.23 (0.87,1.74)	1.1 (0.8,1.5)
Somali	5.74 (3.66,9.0)	5.4 (3.4,8.5)
Benishangul	1.6 (1.05,2.32)	1.4 (0.9,2.0)
SNNPR	1.79 (1.24,2.56)	1.6 (1.1,2.3)
Gambela	1.11 (0.74,1.66)	1.2 (0.8,1.8)
Harari	0.74 (0.5,1.1)	0.9 (0.6,1.3)
Addis Ababa	0.59 (0.4, 0.9)	1.2 (0.8,1.8)
Dire Dawa	1.34 (0.87,2.05)	1.8 (1.14,2.70)

Random effects (measures of variations) and model fitness

We fitted the mixed-effects models for binary responses with- meqrlogit- Stata command. Mixed-effects logistic regression is logistic regression containing both fixed effects and random effects. We found the likelihood-ratio test (LR test) statistically significant (p-value < 0.05), with better fitted regression model better than the traditional. Based on the latent variable estimation method, the constant quantity π2/3~3.29 for the lower-level variance. Thus, for a two-level multilevel logistic regression model with a random intercept, ICC = VPC = Vc/ [vc+3.29). The ICC (VPC) quantifies the degree of homogeneity of the outcome within clusters. From the intercept-only model (null model), we found the variance partition coefficient (VPC): 17.8%. This implied that 17.8% of the variation in animal source food consumption among children aged 6–23 months to be attributable to differences between clusters (primary sampling units). In our data, the estimate of the between clusters variance was 0.71(0.51, 0.98: at p-value <0.001) for empty model (M0). The corresponding MOR was 2.23, indicating the odds of inadequate ASF consumption was 2.23 times higher in the children with the higher propensity the outcome of interest compared to those children with lower propensity. The MOR is a measure of heterogeneity in both between- and within-cluster variance. In this study, the final model (M3) revealed that a higher proportional change in variance (PCV) with 38%, showing about 38% of the variation of ASF consumption explained by the combined predictors at the individual-and community-level. For the reason that the models are nested, we used the –2 log likelihood (DIC) based statistic to compare models with regard to goodness of fit. Since the DIC in final model (M3) was the lowest in our analysis, the final model (M3) better fit to the data (Table 4).

Table 4

Random effects (measures of variations) and model fitness for no ASF consumption among children at primary sampling units (clusters) level by a multilevel mixed-effects logistic regression modeling based on pooled data from Ethiopia DHS 2016 & 2019.

Parameters	Null model (empty model)	Model-III
	model-0	(Final model)
Random-effects (measures of variations)
Cluster level variance (SE)	0.71 (0.12)	0.44 (0.1)
PCV (%)	Reference	38%
ICC or VPC (%)	17.75%	11.8%
MOR	2.23	1.88
Fit criteria (model diagnostics)
Loglikelihood (LL)	-2202.076	-1980.142
DIC (-2LL)	4,404.152	3,960.284
AIC	4408.153	4018.284

AIC: Akaike’s information criterion, DIC: Deviance information criterion, ICC: Intra-class correlation coefficient, MOR: Median odds ratio, PCV: Proportional change in variance, SE: Standard error, VPC: Variance partition coefficient

Discussion

The current analysis indicated that the ASF consumption of children in Ethiopia is 22.7%. This showed that nearly one in five children had ASF consumption. Diversified diet consumption is a proxy indicator for micronutrient adequacy, including iron, zinc and others which promotes optimal linear growth and development [18]. Low ASF consumption is more prevalent in more food insecure regions of the country (Afar, Somali, and other regions). In fact, ASFs are a significant contributor for the daily intake of bioavailable nutrients, with the potential to address the existing burden of stunting (37%) and protein-energy malnutrition (7.2%) in the country [1]. ASF consumption has the potential to decrease the stunting as compared to those with low ASF consumption [18], which emanates from eating monotonous plant source diets which in turn usually threatens nutrient bioavailability. Other study also showed that ASF consumption is low while the ASF expenditure is raising. Dairy products (14.7%), beef (3.2 kg), and egg (0.3%) were the most commonly consumed foods [39]. A better ASF consumption of 64% and 18% among Urban and rural was reported among school age children [40].

Despite a wide variation in ASF consumption by wealth status and residence, low ASF consumption is highly prevalent in different parts of the country. In this study, we have observed a very low ASF consumption in pastoralist communities like Somali region as compared to other parts of the country. Apart from the resource scarcity related factors, one of the reasons for this disparity could be measurement of ASF, as we do not considered milk and milk products, but it is widely used in pastoralist communities of Ethiopia [41]. Another study also indicated that children from the highest wealth quintile were more likely to have a better expenditure for ASF (16.5% and 4.9% of household expenditure) [39]. It also indicated that food support schemes like the productive safety net improve the dietary diversity and consumption of ASF, targeting parts of the society having low income [16].

In Ethiopia, the price index of ASF is rising at alarming rate with an average raise in price of ASF by 36% [17]. Strategies to increase access to ASF targeting the poor are crucial. Long-term strategies like employment schemes, and food for work approaches can aid in increasing diversified diet consumption in the short-term. However, long-term strategies shall be in place along with the short-term approaches in order to expect a lasting result [42, 43]. Furthermore, improving quality livestock production can increase access to ASF and consumption can alleviate malnutrition [44]. Such approaches allows sufficient intake at affordable price for better nutritional outcome and health status of children [41, 45].

Studies have shown that livestock and asset ownership can potentially improves ASF consumption [18]. Sustainable strategies through diversified agricultural production coupled with enhanced behavioral change communication approaches can improve ASF availability and consumption [46]. In addition, the concept of nutrition sensitive agriculture will play a mounting role if implemented with sectoral collaboration. However, the affordability of high quality, nutritious foods is threatened by the worsening market prices of food items in the country. The policy directions should encompass a multipronged approach to stabilize and increases the affordability of animal source foods.

The results of our analysis in the final model (model III) showed that both individual-level factors (age of the child, ANC visits, place of delivery, household wealth index, maternal education level, access to all three media at least once a week, and high-risk fertility behaviors), and community-level factors (community poverty level, place of residence, and region) were significant predictors of ASF consumption. This study showed that the proportional change in variance (PCV) of the full model was responsible for about 38% of the log odds of low ASF consumption in the communities. The outcome of MOR, a measure of unexplained cluster heterogeneity, is 2.23 and 1.88 in the null model and final model respectively. Thus, the results of MOR in this analysis indicate that there is unexplained variation between the clusters of the community. The VPC results also showed a significant cluster-level variance, which is a minimum precondition for a multilevel modeling.

In this study, children aged 6–8 months and 9–11 months had a 3.1- and 1.5-times increased risk of low ASF consumption as compared older children aged (18–23 months). This showed that consumption of ASF increases with age. Similar findings were found in studies conducted in Ethiopia. The possible justification could be lack of nutritional knowledge [19, 20, 22, 47, 48], late initiation of nutrient-dense complementary foods [49] and social norms and beliefs [20, 21]. The other possible reason for this association may be due to food taboos around ASF consumption in children [48]. Thus, Behavioral Change Communication (BCC) interventions are needed to improve the caregivers’ nutritional knowledge as effective strategy for infant and young children feeding practices. The finding of this study revealed that children whose mothers had at least four ANC visit and gave birth at health facility had a protective effect against low ASF consumption. Nutritional counseling and better exposure to nutrition information may have a crucial role for better ASF consumption among children. In addition, those families with access to ANC service are more likely educated, and from high socioeconomic class which may have a better ASF access than others [50]. The possible pathways (mechanisms) of this association may be due to ease of access to health information that may improve the likelihood of mothers practice of proper child feeding [49].

Similarly, this study revealed that children from poorest and poorer households were more likely to have low ASF consumption compared with children from richest households. This finding was similar with studies done in Cambodia, Ethiopia, and Indonesia [9, 20, 21, 49]. As the household wealth status increases, consumption of ASF increases. The possible mechanisms might be due to the agricultural policy and ASF prices and attitude related issues like use of ASFs as a source of income rather than feeding their own children [20, 21, 51]. Because the imbalance of the price and demands of ASFs in the community are high, almost every product of livestock is commonly intended for market purposes. As the result, the consumption of ASFs in poor households is hampered among children [20, 21]. The other possible justifications may be cultural factors like low women’s empowerment, leading to decreased prioritization of household nutrition[48, 51]. In contrary, when, women have greater decision-making power over household expenditures and control over resources, it is likely that the households would allocate larger proportions of their income and resources to improve their children’s diets [51]. This study found that, children of mothers who attended at least secondary educational level and have accesses to all three media at least once a week had less likely to have low ASF consumption as compared with their counter parts respectively. These findings were similar to that of studies conducted in Nepal [52]. These might be due to the fact that educated mothers might have better nutritional knowledge about the nutritional advantages of ASF consumption [20, 21, 51]. In addition, the role of media in raising awareness and shaping the practice of mothers towards better child feeding is substantial.

In Ethiopia, 35% [53], 57% and 36.8% [2] are a victims of Zinc deficiency, anemia and stunting, respectively where the positive role of ASF consumption in alleviating these problem is undisputable. Also, with this currently very low ASF consumption and less diversified foods, the 2030 targets for Zero hunger and stunting will be ideal without sustainable approaches to improve the dietary intake of children. Promoting nutrition sensitive agriculture, rearing livestock for own consumption; can improve consumption and income for better livelihoods. Above all, improving the livestock productivity, especially targeting the pastoralists through improved technologies and approaches allows widespread availability of ASF in the country [51]. Such strategies could sustainably improve the livestock productivity, and empowers women to address malnutrition in the developing countries [54].

Despite, the availability of livestock products, the household level understanding on the necessity for ASF, and household heads preference for market instead of feeding their child is the major factor in the majority of the rural area [52]. However, it is believed to be related to the deep routed problems of food insecurity, lack of knowledge and poverty which plays a great role for low level of diversified food consumption in developing country. Strengthened poverty reduction schemes and behavioral change communication (BCC) to improve the practice of proper child feeding should be conjoined [55]. In addition, consumption of ASF among children is low and have a widespread spatial variation. There should be a targeted strategies to increase access to ASF, and increase ASF production. Some cultural perspectives might also hinder access to ASF in some areas of the country. There should be a strengthened nutritional counseling and nutrition education strategies for an improved access and consumption of ASF for children. Furthermore, strong regulatory frameworks shall be in place to stabilize the increasing food prices in the country.

22 Nov 2021

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Sukumar Vellakkal

Academic Editor

PLOS ONE

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

2. Thank you for stating the following in your Competing Interests section:

“I have read the journal's policy and the authors of this manuscript have the following competing interests. The authors declare that they have no competing interest. All authors agreed on the submission of the manuscript.”

Please complete your Competing Interests on the online submission form to state any Competing Interests. If you have no competing interests, please state ""The authors have declared that no competing interests exist."", as detailed online in our guide for authors at http://journals.plos.org/plosone/s/submit-now

This information should be included in your cover letter; we will change the online submission form on your behalf.

3. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

4. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: It is great to see more research in nutrition using MLM methods. The discussion is rich in detail and the overall paper contributes importantly to growing interesting in ASF.

Above all, I would suggest a thorough copy editing from someone knowledgeable in the field. It would be made much stronger with some language revisions.

I have two, relatively large, suggestions for revision:

1. I would be interested to see the analysis re-done including dairy products. Dairy is also an ASF and an important one at that, and its exclusion as part of ASF is confusing.

2. It would be very helpful to see your iterations on model fit. Many of the variables seem to overlap or don't have a strong theoretical grounding, and some (e.g. livestock ownership) seem to be missing.

I've provided detailed feedback in the attached comment.

Reviewer #2: This is a very good piece of work. I have enjoyed reading it. The topic has important policy implications, the method suits the nature of the data structure, and the effort put into the manuscript to make a robust explanation is admirable. My general feedback is very good with mainly one editorial/additional estimation correction required apart from the routine editorial details I found. For this main comment, please see 2.1. Other than this, the followings are some of the comments I put forward in each subheading:

1. General comments

1.1. “Animal source food consumption is very low” (line 47) – reference? Compared to?

1.2. “despite a better livestock population, diversified diet consumption is poor..” should it be diversified ASF? ( line 76)

1.3. Which time period are the data points referring? (line 78)

1.4. “So far, limited attempts were tried to understand the problem with the use of small-scale surveys, despite the multifaceted nature of the issue, where individual or household level factors only may not be sufficient” (line 86 – 88). What are these studies? A few citation of these studies will be ideal. How is your study unique from these studies?

1.5. “Nationwide study addressing the factors that affect the level of ASF consumption has not been conducted in Ethiopia” (line 97 – 98) I am afraid I can’t agree to this claim. For example, Workicho et.al (2016) Household dietary diversity and Animal Source Food consumption in Ethiopia: evidence from the 2011 Welfare Monitoring Survey, Potts et.al (2019) Animal Source Food Consumption in Young Children from Four Regions of Ethiopia: Association with Religion, Livelihood, and Participation in the Productive Safety Net Program are a few examples. You can cite these researchers and point out what makes your study unique.

1.6. “Currently, the price of ASF is rising alarmingly in the country above 36%”. Lin 381. This should be re-written. There is no single ASF price as such unless it is a price index, I think.

1.7. “In Ethiopia, 35% [37], 57% and 36.8% [9] are a victims of Zinc deficiency, anemia and stunting…”line 438, who are these figures referring to? Of the Ethiopian population, Ethiopian children or ?

2. Methodology

Measurement of the dependent variable – while a dichotomized variable is a good starting point, it is also crude as it aggregates measurement and somehow undermines heterogeneity, for example in the type of ASF consumed here. In this regard, as a robustness extension,

2.1. In addition, “A large scale regional variation (..) in Addis Ababa and … lowest in Somali was observed in ASF consumption aged 6-23 months.” (line 262 – 264). And “Low ASF consumption is more prevalent in more food insecure regions of the country (Afar, Somali, and other regions).” (Line 361 – 362). “Apart from the resource scarcity related factors, one of the reasons for this disparity could be measurement of ASF, as we do not considered milk and milk products widely used in pastoralist communities of Ethiopia”.

This provides a wrong information and introduces systematic bias. Afar and Somali constitute one of the largest livestock population making children more likely to consume milk and milk products. These regions do not have similar wealth of chicken and are less likely to consume egg.

As a result, either

• Bring the definition of ASF used in this manuscript forward explicitly at the start (Mainly in the dependent variable section) and give a concerted direction to your reader what you mean by an ASF or

• Include milk and milk products and do additional analysis,

• Leave Afarand Somali from your analysis. My advice for you is to include milk and milk products and re do the analysis while keeping the current results as they are and see how things change. This will give you percentage values of ASF consumption and regression estimates to which you can compare results.

2.2. Will a separate regression (as a dichotomous variable) be an extension of this study or be used as a robustness check? For example, estimating the same model for the consumption of Egg and other flesh foods alone.

2.3. Counting the number of ASF and generating an ordered variable to run multi-level ordered logistic regression provide similar results?

2.4. “Only 27 (0.6%) of mothers had accesses to all three media at least once a week” which three media? I think this needs re-phrasing the statement. Looking at Table 1, this variable does not show variability worth to have it in a description. 99% said No to media access. This requires re categorization or will not make a good predictor. This is also true for “community level education” where only 5% are in the high education group. (Table 1)

2.5. Table 2 could give more information if the second row is subdivided into each type of flesh foods. The first row gives the aggregate measure already.

3. Results and discussion

3.1. “The finding of this study revealed that children whose mothers had at least four ANC visit and gave birth at health facility had a protective effects against low ASF consumption. The possible pathways (mechanisms) of this association may be due to ease of access to health information that may improve the likelihood of mothers’ practice of proper child feeding” (line 415 – 418). Can you provide evidence to this in your research, please?

3.2. “The current analysis indicated that the ASF consumption of children in Ethiopia is 22.7%” – better if written as “Currently, only 23% of children in Ethiopia consume ASF” of something similar along this line otherwise, the conclusion drawn from this will be over riding.

3.3. Policy implication scattered here and there. They should be brought together under one heading or in one or two paragraphs. For example, line 381 to 388 against lines 442 to 445

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: Review Abdu et al..docx

Click here for additional data file.

Submitted filename: TA review comments- PONE-D-21-25620.docx

Click here for additional data file.

14 Jan 2022

From authors

Manuscript title: Individual and community-level factors associated with animal source food consumption among children aged 6-23 months in Ethiopia: Multilevel mixed effects logistic regression model

Dear editor and reviewers

I am very grateful for all valuable comments and suggestions you gave us for the improvement of the paper. Hereunder, I responded to each points raised and these are corrected and amended in the revised manuscript when applicable (indicated in track changes) and clean copy of the final manuscript.

Editor’s comments

1. The funding statements

Response: since the authors did not receive any funding the funding statement was updated as “Funding: The authors received no specific funding for this work. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

2. Thank you for stating the following in your Competing Interests section

Response: We included an updated conflict of interest form and we filled the online system.

3. In your Data Availability statement,

Response: regarding the data availability, we updated the statement in accordance with the DHS data sharing policy (https://dhsprogram.com/data/terms-of-use.cfm) which stated that the data can only be used for the registered study only and will not be shared for other. Since it is a third party data which is obtained from the Demographic and Health survey repository and researcher can obtain the data through research project registration and can utilize the data up on agreeing for the terms and conditions for the data use.

Thus, we included the following statement as: ” The data used in this study are from the Ethiopian Demographic, and Health Survey (2016 and 2019) and can be requested from the DHS office at https://dhsprogram.com/Data/ using the details stated in the Materials and Methods section of this paper.”

4. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly.

Response: since the data is from third party (DHS program), the terms and conditions do not allow to share the data to others (only for the registered data only). Please refer to the above updated data availability statement.

Reviewer 1 comments

1. General comments

1.1. “Animal source food consumption is very low” (line 47) – reference? Compared to?

Response: diversified diet is recommended in ideal scenario and studies reported ASF consumption above the one we obtain in this study (22%). This indicates that only one in 5 children had consume ASF (refers to egg and flesh products) which is by far below the ideal case, where we said ASF consumption is poor among children in the study area.

1.2. “despite a better livestock population, diversified diet consumption is poor.” should it be diversified ASF? (line 76)

Response: It is just to refer to diversified diet containing diversified ASF consumption. It is corrected in the revised version.

1.3. Which time period are the data points referring? (Line 78)

Response: corrected (for the year 2019)

1.4. “So far, limited attempts were tried to understand the problem with the use of small-scale surveys, despite the multifaceted nature of the issue, where individual or household level factors only may not be sufficient” (line 86 – 88). What are these studies? A few citations of these studies will be ideal. How is your study unique from these studies?

Response: done, relevant citations are included to show the gap. This part is extensively revised and a more recent publication in the area are included and cited. In addition, the gap and the need for evidence is clearly shown.

1.5. “Nationwide study addressing the factors that affect the level of ASF consumption has not been conducted in Ethiopia” (line 97 – 98) I am afraid I can’t agree to this claim. For example, Workicho et.al (2016) Household dietary diversity and Animal Source Food consumption in Ethiopia: evidence from the 2011 Welfare Monitoring Survey, Potts et.al (2019) Animal Source Food Consumption in Young Children from Four Regions of Ethiopia: Association with Religion, Livelihood, and Participation in the Productive Safety Net Program are a few examples. You can cite these researchers and point out what makes your study unique.

Response: the comment was helpful to consider the studies done so far and justify our study. The revised justifications are included in the final paragraphs of the introduction. I am really grateful for that. As indicated in the revised manuscript, there are studies: both local studies and national wide studies, but the current data (updated and covers almost all regions of the country) and more robust statistical approaches allows to clearly identify the hierarchical predictors of ASF consumption among children.

1.6. “Currently, the price of ASF is rising alarmingly in the country above 36%”. Lin 381. This should be re-written. There is no single ASF price as such unless it is a price index, I think.

Response: corrected, it refers to the percentage increase in the price of food items (price index).

1.7. “In Ethiopia, 35% [37], 57% and 36.8% [9] are a victims of Zinc deficiency, anemia and stunting…”line 438, who are these figures referring to? Of the Ethiopian population, Ethiopian children or ?

Response: refers to children in Ethiopia based on the national survey done in 2019. It is corrected accordingly in the revised version.

2. Methodology

Measurement of the dependent variable – while a dichotomized variable is a good starting point, it is also crude as it aggregates measurement and somehow undermines heterogeneity, for example in the type of ASF consumed here. In this regard, as a robustness extension,

2.1. In addition, “A large scale regional variation (..) in Addis Ababa and … lowest in Somali was observed in ASF consumption aged 6-23 months.” (line 262 – 264). And “Low ASF consumption is more prevalent in more food insecure regions of the country (Afar, Somali, and other regions).” (Line 361 – 362). “Apart from the resource scarcity related factors, one of the reasons for this disparity could be measurement of ASF, as we do not consider milk and milk products widely used in pastoralist communities of Ethiopia”.

This provides a wrong information and introduces systematic bias. Afar and Somali constitute one of the largest livestock populations making children more likely to consume milk and milk products. These regions do not have similar wealth of chicken and are less likely to consume egg.

• Bring the definition of ASF used in this manuscript forward explicitly at the start (Mainly in the dependent variable section) and give a concerted direction to your reader what you mean by an ASF

Response: due to the nature of the data and the WHO/FAO definition for ASF consumption, we tried thoroughly for the first analysis and we finally decided to consider the mentioned food groups (also stated in the FAO definition) to our case. Then, considering the limitation of the analysis in the discussion part.

It shows us your critical review the paper and thank you very much for this. This question was also the main question for us at the start of this work. However, the FAO and WHO IYCF indicator states ASF consumption as the seventh important indicator which is operationalized there. The definition of ASF there states that” the percentage of children who consumed egg, poultry, animal fleshes during the past 24 hours). Dear reviewer, please kindly note that our target population are infants and children, where we expect almost all consume either breast milk and/or other milks during this time. Due to this, the indicator is defined not to consider milk and milk product in the indicator definition.

More specifically, in our country the rate of continued breast feeding till two years and median duration of breast feeding (reaches up to 36 months) is high and long, where children are being fed at least with breast fed. That is why the ASF consumption is operationalized like this.

As per the reviewer’s recommendations we defined the ASF consumption clearly in the introduction and variable section.

2.2. Will a separate regression (as a dichotomous variable) be an extension of this study or be used as a robustness check? For example, estimating the same model for the consumption of Egg and other flesh foods alone.

Response: The outcome is defined based egg and other flesh foods consumptions. So, for the description purpose, we presented the overall ASF consumption and each item separately. However, since our aim is to assess the predictors of ASF consumption, the outcome variable is ASF consumption (“1” for yes and “0” for No). So, there is no need to have a separate regression model.

2.3. Counting the number of ASF and generating an ordered variable to run multi-level ordered logistic regression provide similar results?

Response: No, please note that the nature of the dependent variable is Yes and No type (dichotomous, not ranked), where this variable could not be ranked to OLS. Thus, we applied the multilevel binary logistic regression for dichotomous outcome variable instead of OLS, which is not appropriate.

2.4. “Only 27 (0.6%) of mothers had accesses to all three media at least once a week” which three media? I think this needs re-phrasing the statement. Looking at Table 1, this variable does not show variability worth to have it in a description. 99% said No to media access. This requires re categorization or will not make a good predictor. This is also true for “community level education” where only 5% are in the high education group. (Table 1)

Response: these statements are revised and some confusing statements are deleted. The community level education is a community level aggregated indicator for education status (which is based on the actual measurements).

2.5. Table 2 could give more information if the second row is subdivided into each type of flesh foods. The first row gives the aggregate measure already.

Response: Here, the overall ASF consumption (first row) and the individual food consumptions (each food are presented in subsequent rows). The data did not specifically indicate specific flesh foods as these are many.

3. Results and discussion

3.1. “The finding of this study revealed that children whose mothers had at least four ANC visit and gave birth at health facility had a protective effect against low ASF consumption. The possible pathways (mechanisms) of this association may be due to ease of access to health information that may improve the likelihood of mothers’ practice of proper child feeding” (line 415 – 418). Can you provide evidence to this in your research, please?

Response: Evidence is provided in the revised manuscript.

3.2. “The current analysis indicated that the ASF consumption of children in Ethiopia is 22.7%” – better if written as “Currently, only 23% of children in Ethiopia consume ASF” of something similar along this line otherwise, the conclusion drawn from this will be over riding.

Response: it is corrected and revised accordingly.

3.3. Policy implication scattered here and there. They should be brought together under one heading or in one or two paragraphs. For example, line 381 to 388 against lines 442 to 445

Response: the policy implications were expressed and indicated in the discussion within each section. However, that may not be good to present like that. So that we prefer to present the policy implications and recommended strategies to be under each study finding, where it allows to clearly indicate the implications there than putting together.

Reviewer 2 comments and responses

Line 23: The background section could be strengthened by a) emphasizing the importance of ASF (rather than diversified diet), b) emphasizing the age group.

Response: ok, it is corrected

Line 27: Suggest changing ‘multi-layered’ to more specific – perhaps individual and community level.

Response: it is corrected as individual and community level …..

Line 30: Sentence beginning with “a stratified two-stage” is unclear. What is complex sample design referring to? How were weights used?

Response: corrected and the survey weight calculated by the DHS survey weighting method was applied. The survey weight for complex sample design is used ad applied for descriptive statistics.

Line 38: I’m curious about the inclusion of home delivered children as a variable, as it may have substantial overlap with some of the other predictors, particularly low SES. In addition, what overlap is there between low SES and high community poverty level, outside using it as both an individual and community level factor?

Response: yes, we are aware of the potential overlap and an interaction analysis was done for possible effect modification and we did not find a statistically significant effect modification in the model. The analysis was done for SES, Home delivery and other potential variables (P-value above 0.05). We are grateful for the importance notice and suggestions.

Line 47: I would love to see the conclusions be a bit more specific in terms of what targeting and policy decisions could be incorporated based on the findings.

Response: revised

Line 62: Suggest re-arranging the background to begin with animal-source foods role in optimal development, rather than the indicator, which belongs later or in the methods section.

Response: revised as per the reviewer request.

Line 71: There are also several more recent publications and meta-analyses which discuss the importance of ASF – inclusion of those would strengthen the background section, which relies on several single-sited analyses or position papers. Please see https://academic.oup.com/advances/article/10/5/827/5513046?login=true and https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD012818.pub2/full.

Response: we included this recent publication in the revised versions.

Line 76: Suggest “widespread” livestock production, or something similar

Response: it is revised and corrected.

Line 78: Suggest revising the line.

Response: Ambiguous sentence was deleted and restructured.

Line 85: Curious if ASF should be considered a practice? Practice to me would suggest a cultural component, but I believe what this analysis is getting at is its availability, affordability, accessibility, etc. Does not necessitate a revision if practice is what is intended – just some elaboration.

Response: here, we refer to ASF consumption to indicate the actual ASF consumption by children obtained from mother or caregivers’ response. So, it is actual consumption history and is not mainly concentrated on the availability and accessibility (which are secondary).

Line 93: Language is a bit informal

Response: corrected and revised.

Line 98: It would be helpful to discuss why high-risk fertility behaviors is specifically highlighted here.

Response: High-risky fertility behaviors are mainly related to a larger family, low socioeconomic class and other collinear variables. Thus, having high fertility-risky behaviors is related to lower availability and accessibility to ASFs, which limit its consumptions. Descriptions and evidences on it is included there.

Line 105: No need to include the names of variables in this section.

Response: deleted.

Line 120: It would be helpful to explain why the study used children ages 6-23 months in particular. This may be better included in the background where the importance of ASF in this age range is emphasized – otherwise there is an equally strong argument to use a wider age range.

Response: As we know failure to get adequate nutrition during the first one thousand days (1000 days) of life very critical period, where majority of stunting and micronutrition with all its adverse short and long-term problems can be prevented. This period includes pregnancy through the first two years of childhood. However, the period before six months infants and fetus are primarily dependent on maternal nutrient supply through breast feeding and placenta, where the role of extra meals starts after six months. This is clearly stated in the background part (it is stated in page 4 paragraph 2).

Line 127-130: I’m curious why Group 4 dairy products isn’t considered here, as this is also an animal source food. This would necessitate re-running all analyses, however unless there is a strong theoretical reason for excluding it, it should also be included.

Response: Response: due to the nature of the data and the WHO/FAO definition for ASF consumption, we tried thoroughly for the first analysis and we finally decided to consider the mentioned food groups (also stated in the FAO definition) to our case. Then, considering the limitation of the analysis in the discussion part.

It shows us your critical review the paper and thank you very much for this. This question was also the main question for us at the start of this work. However, the FAO and WHO IYCF indicator states ASF consumption as the seventh important indicator which is operationalized there. The definition of ASF there states that” the percentage of children who consumed egg, poultry, animal fleshes during the past 24 hours). Dear reviewer, please kindly note that our target population are infants and children, where we expect almost all consume either breast milk and/or other milks during this time. Due to this, the indicator is defined not to consider milk and milk product in the indicator definition.

More specifically, in our country the rate of continued breast feeding till two years and median duration of breast feeding (reaches up to 36 months) is high and long, where children are being fed at least with breast fed. That is why the ASF consumption is operationalized like this.

As per the reviewer’s recommendations we defined the ASF consumption clearly in the introduction and variable section.

Line 130: It’s important to note that this indicator does not exactly refer to adequate ASF consumption, but that ASF consumption is necessary for minimum dietary diversity. It’s therefore inaccurate to say “adequate animal source food consumption,” as this is only a criterion for MDD. I would simply rephrase to ASF consumption or no ASF consumption, but avoid labeling the variable as adequate, since there is technically no indicator for this.

Response: yes, I noticed the problems with the terms used in this analysis. Thus, we reframed as “consumed ASF” or “not consumed ASF” respectively. Thanks for that.

Line 135: It would be helpful to include citations that support the use of ‘high risk fertility behaviors’ from a theoretical perspective. I can understand why mother’s age <18 might influence ASF consumption as a proxy for poverty or decision-making, but not mother’s age >34 or birth order above three. These are more adequately capturing risks to the mother and child during birth and early life but I don’t see the connection to ASF consumption.

Response: As per the comment there are evidences that show high fertility risk behaviors is prevalent (72%) and linked to anemia and child nutritional status [1]. The link between high fertility risk behavior with ASF consumption is mainly due to the household size, poor socioeconomic status and low access to ASF and hence ultimately low consumption.

Line 148: Don’t need to include Stata commands or variable names throughout this or later sections. I’m curious here about the reasons for using community poverty level and community education level, as I don’t think they’re well explained. You might also note here that

Response: The stata commands are removed from the result, it is acceptable. The community level factors are very critical and important factors associated with the ASF consumption. It is known that the individual level factors only will not predict the ASF consumption, which may also be affected by the community level general context (geographical related factors), which needs to be evaluated. That is why we considered the community level factors, despite the issue of some correlations. Also, these community level indicators are a statistically robust indicator.

Line 250: I’ve discussed inclusion of different variables in several other sections but also want to highlight a few more that seem missing. In particular, I believe DHS asks whether households own any livestock, which would be important to include. I apologize if I am wrong on that.

Response: Here, your concern is good and previous studies showed that livestock ownership is associated with a better ASF consumption. However, on the other way round this may not hold true for the majority of the poor where they consider such foods sources as cash crop instead of own consumption. Due to such scenarios, and the DHS did not explicitly report that, we did not consider livestock ownership.

Line 273: I’m curious why variables with a p-value of <0.25 were included – you may want to include a citation for this. Otherwise, I would suggest revising many of these analyses so that there is a strong theoretical basis for their inclusion. Since the paper is interested in exploring variables that are associated with ASF and designing policy around that, it should be made clear why, for example, ANC visits or place of delivery are useful predictors of ASF consumption. The same holds true of fertility risk.

Response: lower p-value (p-value below 0.25) and/or variables with strong theoretical relation with ASF consumption and previously identified predictor variables were considered for the multivariable model, which is supported by many statistical approaches. Thus, we did not only rely on the p-value. Thus, we use a combination variable selection procedure including stepwise backward regression model for model building in combination with p-value and theoretical backgrounds.

Line 280-284: I would like to see in an appendix any additional model fitting that you carried out with different variables. As stated above, many of these variables don’t seem to have a strong rationale for inclusion. Model fit generally increases with more variables, but it would be important to demonstrate by how much, precisely, to justify their inclusion. You note that the very end that associations with a p-value below 5% were included, but in what stage of the process did you fit the variables?

Response: Here using the detail statistical modeling approaches stated above, we fitted a bivariable and multivariable model, where the bivariable is more likely to be confounded by multiple variables and that is why we prefer to present the multivariable fitted logistic regression model. We did not search for p-value 0.05 only rather we statistically fit the model and present the findings. In addition, the model fitting information for different iterations (AIC and DIC) are presented in the result part in detail.

Line 284: Correct to p-value below 0.05.

Response: It is corrected as p-value below 0.05.

Line 328: It would be helpful to re-organize this table to separate individual and community level factors more clearly.

Response: This table is already organized in to individual level factors first followed by community level factor thereafter. We prefer to present like this to avoid redundancy.

Line 365: This statistic comes from one survey in Indonesia, which should either be noted, or the statistic removed.

Response: done

Line 358-370: This discussion would be better used to present more of the overall findings from your analysis – rather than other analyses. This discussion only mentions that ASF consumption is low and more prevalent in food insecure regions. It should be connected back to the main themes from your findings.

Response: it is revised as per the comments

Line 416: This is a great explanation of possible pathways for ANC visits, though I would be curious whether these variables significantly improve model fit individual when included with education and wealth. If so, this is a great finding, but I would also like to see it explained more – e.g. what messages do mothers receive? Finally, is there a reason for including both ANC visits and home birth?

Response: The inclusion of these variables significantly improved the model fitness and are included in the final multivariable model. It is known that the ANC period is a potential contact point where maternal and child nutrition information are being delivered under the ENA program. And studies are showing that mothers are aware of the information and tends to change their behaviors.

Reference

1. Tamirat, K.S., G.A. Tesema and Z.T. Tessema, Determinants of maternal high-risk fertility behaviors and its correlation with child stunting and anemia in the East Africa region: A pooled analysis of nine East African countries. PLOS ONE, 2021. 16(6): p. e0253736.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

1 Mar 2022

7 Mar 2022

From authors

Manuscript title: Individual and community-level factors associated with animal source food consumption among children aged 6-23 months in Ethiopia: Multilevel mixed effects logistic regression model

Manuscript ID: PONE-D-21-25620R1

Dear editor and reviewers

I am very grateful for all valuable comments and suggestions you gave us for the improvement of the paper. Hereunder, I responded to each point raised in the second-round review and these amendments are indicated in track change in the revised manuscript when applicable and clean copy of the final manuscript is submitted.

Editor’s comments

#1 The issue regarding depositing the laboratory protocol in a relevant repository, since all methodological approaches are stated and indicated in the manuscript, it is not applicable for this case.

It is not applicable

#2 We did not make any change to financial disclosure

# 3 the issue of retracted references

Here we used the scite tool to check for retracted references and w updated the reference section accordingly. We also manually checked for retracted references and citation with retracted referencing were managed accordingly.

Reviewers’ comments

Fortunately, almost all the comments have been addressed in the first revision and editor and the reviewers duly acknowledged our effort, thanks for that. But there are some comments that may need clarification from the authors. First of all, we are glad to have your valuable comments.

Reviewer 1 comments

#1 on the introduction

Response; we accept the comments and we included other robust and more comprehensive evidence on the issue. The references are updated in the revised version.

#2 Reword table 3

Response: It is revised as requested “no ASF consumption”.

#3 the issue of Milk is different from ASF

Response: in the previous revision, we clearly indicated the definition of the indicators and ASF is defined in WHO in a way we define it in the current paper. But I accept, the issue to include correct references to the indictor definition. In order to be in line with the WHO updated new indicators, we still stand to define ASF consumption (as indicator)-as egg and/or flesh consumption [1] as such food groups are different from milk consumption among children due to

#1 majority of children consume breast milk up to two years

# 2 proportion of older children who consumer milk (cow, goat, camel) is high but,

#3 the typical consumption of egg or flesh meats has a paramount importance where access to such a food is low in developing countries due to rising price of animal source foods. Moreover, such foods are rich source of iron, zinc, vitamins and other mineral necessary for optimal growth in addition to protein, carbohydrates and fats which are common in milk. But the first three nutrients are found in scant amount from milk and milk products.

It is due to the above evidences that WHO excludes milk and milk product from the definition. I hope you will understand that.

#4 high risky fertility behaviors

Response: above all the point raised by the reviewer here is strong and logical to us to reconsider the decision regarding variable selection. The reason we included this variable in the model is due to the following.

A. The inclusion of this variable greatly improved the model fitness and it is found in many different literatures that High fertility behavior is strongly linked to food security, access to nutritious food and nutritional status of children (stunting).

We can clearly understand from the pooled analysis of DHS surveys from 2006 to 2012 that fertility behavior is important factor that determine access to food and nutritional status of children [2].

B. as this indicator shows the overall fertility risk of mother of children, this may indirectly indicate or correlate with birth interval and number of children, which in turn affects the ASF consumption negatively, we included it in the description part.

#5 copy edition and proofreads

The revised manuscript undergoes a thorough copy edition and proof read by professional editor as indicated in the revised manuscript (when applicable) and cleaned manuscript.

Reviewer 2 comments

-the reviewer is convinced by the revisions made and the compelling rational arguments raised by the author. Thanks for understanding the approaches and the methods used for this paper.

Additional comments; we uploaded the figures for PACE tool and the output was ok.

References

[1] Organization, W.H., Indicators for assessing infant and young child feeding practices: definitions and measurement methods. 2021.

[2] Rutstein, S.O. and R. Winter, The effects of fertility behavior on child survival and child nutritional status: evidence from the demographic and health surveys, 2006 to 20122014: ICF International.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

10 Mar 2022

Individual and community-level factors associated with animal source food consumption among children aged 6-23 months in Ethiopia: Multilevel mixed effects logistic regression model

PONE-D-21-25620R2

Dear Dr. Abdu,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Sukumar Vellakkal

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

23 Mar 2022

PONE-D-21-25620R2

Individual and community-level factors associated with animal source food consumption among children aged 6-23 months in Ethiopia: Multilevel mixed effects logistic regression model

Dear Dr. Abdu:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Sukumar Vellakkal

Academic Editor

PLOS ONE