Maternal depression is associated with child undernutrition: A cross‐sectional study in Ethiopia

Abstract Child undernutrition is widespread in low‐ and middle‐income countries (LMIC) and is associated with health and economic losses. Undernutrition is estimated to contribute to 3.1 million deaths per year in children less than 5 years of age. A complex causal and contextual factors contributing to child undernutrition have been assessed, but maternal depression, which could contribute to child undernutrition by interfering with the mother's child caring practice and ability, has been received little attention. The objective of this study was to assess the association between maternal postpartum depression symptoms and infant (5–10 months of age) stunting in northern Ethiopia. A community‐based cross‐sectional study was conducted among mother–infant pairs (n = 232) between March and April 2018. Through interviewer‐administrated questionnaire, information on sociodemographic variables were collected, and maternal depression symptoms were assessed using the Edinburgh Postnatal Depression Scale (EPDS≥13). Infants' length and weight were measured and converted to length and weight for age Z scores using the WHO growth standards. Breastfeeding was a norm, but the adequacy of complementary feeding practice was sub‐optimal. Only 25% of the infants met the minimum meal frequency (MMF), less than 10% met the minimum dietary diversity (MMD; 9%) or minimum acceptable diet (7%). Maternal depression was prevalent (22.8%) and was significantly associated with inappropriate complementary feeding and stunting (P < .05). Improving complementary feeding practices is central to preventing stunting in this and other settings. However, such efforts should integrate interventions that address maternal depression to improve child feeding and caring practices to effectively prevent stunting.

were more likely to be stunted but not underweight. However, the few studies that investigated association between maternal depression symptoms and child undernutrition in Ethiopia produced mixed results. For example, Nguyen et al. (2013) reported that maternal common mental disorders were associated with undernutrition in Bangladesh and Vietnam but not in Ethiopia. However, this and other studies evaluated mental disorders but not specifically maternal depression symptoms. They also included a wider age group (6-59 months), which could underestimate the effect of maternal depression symptoms as the highest need for maternal care is in the first 2 years coinciding with the time that undernutrition reaches a peak. Particularly, postnatal depression could be highest in the child's first year of life. Therefore, the aim of this study was to assess the association between maternal depression, specifically postnatal depression, and undernutrition in infants aged 5-10 months old in northern Ethiopia.

| Study site
This community-based cross-sectional study was conducted between March and April 2018 in rural areas of northern Ethiopia. The region is characterized by high prevalence of child undernutrition, with a stunting prevalence of 46%, which is above the national average (38%) (CSA). The study was conducted in 10 kebeles (smallest administration unit) in the Meket district.

| Sample size and sampling procedures
A total of 238 households targeted for the Ethiopian government's Productive Safety Net Program (PSNP 4 ) were included in this study using a multi-stage sampling technique. Of the 27 kebeles in Meket district, 10 were randomly selected. A listing of PSNP 4 households with infants 5-10 months of age was compiled through house-tohouse visits, and a random sample in proportion to the size of the kebeles was selected. Being resident in the district and having an infant 5-10 months old were the eligibility criteria for participating in the study. The interviewees were mothers with infants 5-10 months.

Key messages
• Maternal depression especially postnatal is a strong predictor of infants' undernutrition measured as stunting and underweight.
• Maternal depression is strongly associated with inappropriate complementary feeding practices of infants.
• Increased postnatal care visits should be promoted, and maternal depression awareness should be integrated in maternal and infant health programmes like health extension packages and other preventive programmes aiming to improve the care and feeding practices of the mother for her infant.
• Maternal mental health care and treatment guidelines especially for postnatal depression are crucial, and attention to maternal mental health is recommended for all countries because of its public health significance.
Future longitudinal study is recommended to evaluate the potential cause-effect relationship between addressing maternal depression and infants' undernutrition in rural areas of low income countries.

| Sociodemographic characteristics
Data were collected by face-to-face interviews with the mother in a quiet room at the health post. Basic sociodemographic data such as age of the mother, marital status, estimated household's average monthly income, religion, educational status and occupation as well as other data concerning household economic status were collected.

| Household food insecurity access scale
The Household Food Insecurity Access Scale (HFIAS) (Coates, Swindale, & Bilinsky, 2007) was used to assess household food insecurity during the 4 weeks preceding the survey. The households were categorized into four groups: food secure, mildly food insecure, moderately food insecure and severely food insecure. Finally, these were merged in to two groups: food secure and food insecure (mildly, moderately and severely) households.

| Anthropometric measures
All anthropometric measurements were taken by the lead author.
Length was measured to the nearest 0.1 cm using a portable wooden recumbent length board with a fixed head and sliding foot piece.
Weight was measured to the nearest 0.01 kg, with an electronic scale (UNICEF Seca 770) with light clothing. Measurements were done in duplicate, and the mean value was calculated. Infant's age was recorded from the immunization card or birth certificate. However, in the absence of such documents, maternal recall with a calendar of events was used. Infant anthropometric status was calculated as length for age z-score (LAZ), weight for length z-score (WLZ) and weight for age z-score (WAZ) (WHO, 2006) reference data.

| Household socioeconomic status
Household wealth index (low, medium and high) was constructed using principal component analysis (PCA). Selected household assets, ownership and number of livestock, materials used to build house and agricultural land size were used to calculate the index. The factor scores used explained 68.4% of the total variance in the household's wealth index.

| Maternal workload
PCA was also used to calculate maternal workload. The 10 distinct laborious activities in the community in the preceding 1 week of the survey were included in the analysis. The variables included in the analysis were collecting firewood and fetching water (walking distance > 30 min), washing clothes, land preparation, weeding, manual mowing, threshing, grinding and pounding grain and terracing. The final components which explained 70.6% of the total variance were retained and women were classified into three categories (low, medium and heavy) of workload based on their scores.

| Health-seeking behaviour and practice
The place of delivery of the infant and immunization status was recalled by the mother, and whenever available was confirmed by checking the immunization cards. Infant illness (morbidity) and healthcare-seeking practices of the mother were assessed through maternal recall of signs and symptoms related to acute respiratory infections, diarrhoea, cough and/or cold and fever over the 2 weeks prior the survey.

| Maternal depression assessment
Maternal depression was assessed using the Edinburgh Postnatal Depression Scale (EPDS) (Cox, Holden, & Sagovsky, 1987) which included 10 items. Mothers rated the extent to which each item matched their feelings during the 7 days prior to the survey. Possible scores ranges from 0 to 30, with higher scores indicating greater severity of depression. A score of 13 and above indicated the postnatal depression. The EPDS has been used as a screening tool for detecting postnatal depression in different cultures (Kakyo et al., 2012) and has been used in several African countries (Madeghe, Kimani, Stoep, Nicodimos, & Kumar, 2016;Ukaegbe, Iteke, Bakare, & Agbata, 2012;Yator, Mathai, der Stoep, Rao, & Kumar, 2016). A validation study in Ethiopia also demonstrated the acceptability and utility of the Amharic version for use as a screening tool for postnatal depression (Tesfaye, Hanlon, Wondimagegn, & Alem, 2009). The internal consistency measured by Cronbach's alpha (0.80) was acceptable.

| Data processing and analysis
Data were checked, coded and entered into SPSS, v. 20, for analysis. Categorical data were analysed by descriptive statistics (frequency and percentage), whereas range, mean and standard deviations were used to present continuous variables. The normal distribution of the data was checked with the Kolmogorov-Smirnov test (Mishra et al., 2019).
Bivariate analyses were conducted, and all variables with P < .25 were included in multivariate logistic regression. The model fit was checked using the Hosmer-Lemeshow goodness of fit test. Statistical significance was considered at the 95% confidence interval, and both crude and adjusted odds ratios are reported.

| Sociodemographic and anthropometric characteristic
Among the 238 recruited mother-infant pairs, 232 completed the study (97.5% response rate). Almost half (49.1%) of the study mothers were in the 25-34 year range, and 95.3% were married (Table 1).

| Breast feeding and complementary feeding practices
Early breast feeding practices were recalled by mothers. Amongst the studied mothers, breast feeding initiation within an hour of birth was 77.2% and 95.7% of mothers currently breast feeding (Table 2). Complementary food intakes of infants were assessed and analysed for infants aged 6-10 months (n = 198). Although a number of infants were not yet being fed complementary foods, cereals/grains was the most commonly consumed food group (35.9%) and only a very few infants (5.1%) had consumed meat in the 24 h before the survey. Most infants (81.5%) had low dietary diversity (0-2 food groups from 7 groups) in addition to breast milk. Three-fourths (74.7%) of infants aged 6-10 months did not meet recommendations for minimum number of meals during the day before the survey.

| Factors associated with infants' stunting and complementary feeding
Lack of maternal education (P = .02) and maternal depression symptoms (P = .009) predicted stunting, whereas being in the upper wealth

| DISCUSSION
Our findings highlight the relatively good practice of continued breastfeeding but also illustrate the very poor complementary feeding practices characterized by low dietary diversity and meal frequency.
Poor complementary feeding practices, low socioeconomic status and lack of maternal education are common predictors of infant stunting, but maternal depression symptoms also was found to be associated with inappropriate complementary feeding and stunting.
The first 1000 days from conception to the child's second birthday is considered to be a critical window of opportunity for nutrition that, if missed, leads to grave consequences such as impaired cognitive function, overall poor health and loss of future productivity (Cursik & Georgieff, 2016;Hoddinott, Alderman, Behrman, Haddad, & Horton, 2013). Considering the importance of the first 1000 days, WHO/FAO developed child feeding guidelines and indicators to routinely monitor and support programmes that aim to improve nutrition in this critical period (WHO, 2010). National nutrition policies are also focusing on this critical window of opportunity, and, consequently, a number of prioritized interventions (e.g. nutrition education, fortification and vitamin A supplementation) are being implemented WHO, 2016).
In the present study that only focused in the first 1 year of life, one-third of the children were already stunted. Although continued breastfeeding was a norm, inappropriate complementary feeding practices were widespread. Consumption of nutrient-dense food groups such as animal source foods and vitamin A-rich fruits and vegetables was very low. This is unfortunate, as dietary diversity in general and animal source food consumption in particular are consistently associated with lower risk of stunting (Kraseve, An, Kumapley, Bégin, & Frongillo, 2017). Poor knowledge about appropriate complementary feeding practices, low production diversity and availability, poor market access and unaffordability of nutrient-dense foods are among possible constraints limiting optimal complementary feeding practices (Abay & Hirvonen, 2017;Hirvonen, Hoddinott, Minten, & Stifel, 2017;Muehlhoff et al., 2017;Ruel, 2019). Maternal depression symptoms affected a fifth of the study participants and was significantly associated with both poor complementary feeding practices and stunting, even after controlling for covariates like socioeconomic status and maternal education. Postpartum maternal depression is quite common in LMIC and prevalence of up to 46% has been reported in Bangladesh (Hagos, Shikur, Hanlon, & Lindtjørn, 2018;Khan & Flora, 2017). Such levels of maternal depression were previously found to compromise care and infant feeding practices, including breastfeeding and complementary feeding (Stewart et al., 2008), which can lead to stunting. Indeed, the association between maternal depression symptoms and child stunting highlighted by our study is supported by a recent multi-country study that evaluated psychological risk factors among women and child stunting in 137 LMIC (Fawzi et al., 2019). This study used effects sizes from a meta-analysis (Surkan et al., 2011) to estimate impact of maternal depression on stunting. However, given the observational nature of the evidence, more rigorous community-based trials evaluating the impact of addressing maternal depression are warranted.
The present study has a number of limitations that need to be considered when interpreting our findings. First, the cross-sectional T A B L E 5 Results of bivariate and multivariate logistic regression analyses of association between maternal depression and inappropriate complementary feeding practices of infants b