Coexistence of Mother Overweight and Stunting Children Within the Same Household in West Africa : Associated Factors

The double burden of malnutrition is the new reality in many low- and middle-income countries. Updating and meeting the challenge of double burdens on vulnerable people is essential to ensure healthy growth. This study is to explore the individual or household factors that may be associated with coexistence of mother overweight and stunting children within the same household (MOSC) in the West African countries. Besides, we sought to examine its prevalence using data from national surveys. Methods: We used nationally representative data from the Demographic and Health Survey of 11 West African countries. Data years range from 2012 to 2014. We used logistic regression to explore the association between the occurrence of coexistence of mother overweight and stunting children within the same household and individual or household factors. Logistic regression was performed on the pooled data to allow comparison between countries ranked by gross national income. kg/m ; anthropometric


Background
The double burden of malnutrition is the new reality in many low-and middle-income countries. Updating and meeting the challenge of double burdens on vulnerable people is essential to ensure healthy growth. This study is to explore the individual or household factors that may be associated with coexistence of mother overweight and stunting children within the same household (MOSC) in the West African countries. Besides, we sought to examine its prevalence using data from national surveys.

Methods:
We used nationally representative data from the Demographic and Health Survey of 11 West African countries. Data years range from 2012 to 2014. We used logistic regression to explore the association between the occurrence of coexistence of mother overweight and stunting children within the same household and individual or household factors. Logistic regression was performed on the pooled data to allow comparison between countries ranked by gross national income. Results: The Prevalence of MOSC was higher in Benin 11.52% (95%CI, 10.61-12.43), the lowest proportion was in Togo and Guinea [4.83% (95%CI, 3.88-5.77), 4.48% (95%CI, 3.57-5.40), respectively]. Women who did not receive antenatal cares were associated with higher odds of MOSC occurrence compared to those who received more than 4 antenatal cares in all countries except Gambia, this association was signi cant in 7 countries among which the MOSC in Ghana was four times more likely to occur in women who did not receive antenatal care compared to more than 4 antenatal cares [odds ratio (95%CI): 4.936 (2.701-9.019)]. Children over 24 months of age were signi cantly associated with higher odds of MOSC occurrence compared to under 24 months in all countries except Sierra Leone [odds ratio (95%CI): 1.215 (0.867-1.703)].

Conclusions:
In this study, none of the individual and household factors were strongly associated in the same way with MOSC occurrence in West African countries, they varied considerably between national databases and independent of national income per capita. However, Children's age and antenatal cares were the two covariates strongly associated with MOSC occurrence in most countries.

Background
The double burden of malnutrition is the new reality in many low-and middle-income countries (LMICs), and serious forms are in Sub-Saharan Africa and some parts of Asia [1,2]. The Global Nutrition Report reported multiple forms of malnutrition's burden among Children under 5 years of age and mother of reproductive age in several countries especially in West Africa [2,3]. Understanding in a speci c way who and where are most affected by the prevalence of these forms of malnutrition is a need to contribute to the implementation of political decisions and the direction of new research.
For many years, many countries have struggled to reduce the prevalence of undernutrition with slight progress, now they must cope with the recent increase in overweight due to the phenomenon of food transition among the population [4,5]. In 2018, more than 2.28 billion people around the world are estimated to be overweight and 150.8 million children under the age of ve are stunted [2,6].
In Africa, the prevalence of stunting children decreased from 38% in 2000 to 30% in 2018, this is still above 25% which is the average of stunting in LMICs. And the mother overweight has increased from 31.7% in 2000 to 41.1% in 2016 [7]. This rise in overweight is mainly due to the rapid change in eating habits, but also the socio-cultural in uence in which obesity is considered a beauty for mother and a ful lling life for men [8]. The double burden of malnutrition is the coexistence of undernutrition and overweight in the same population, household or individual [1,2], And studies have indicated that it is most prevalent in LMIC where undernutrition is still a public health problem [4].

Methods
Analyses were performed with available nationally representative data from the West African countries' Demographic and Health Survey (DHS). Out of the 17 countries of West Africa, DHS data of 11 countries were available and contained mother and their children body measurements. The years of the data range from 2012 to 2014. The DHS Program is designed to provide monitoring data of the population particularly children and women in developing countries and evaluate health and nutritional situation. Data for the population sample are collected with the collaboration of the governments of the countries. It is a strati ed two-stage cluster design and based on the selection of households to provide an adequate representation of urban and rural areas and geographical region. Data sets are available from the DHS website (https://dhsprogram.com/) [16]. The ICF provided technical assistance through the program, whose objective is to provide support to countries for carrying out demographic and health surveys [17]. Our study evaluated the association of interest in a representative sample of women of childbearing 15-49 years of age and their children under 5 years of age. We have chosen the 'Kids Recode' les which contain all the information relating to children and mother. We excluded all mother with non-plausible anthropometric data, non-pregnant, recently postpartum, having a height under 135m and BMI less than 12 kg/m 2 or higher than 60 kg/m 2 ; and children with non-plausible anthropometric data and height/age less than 6 or higher than 6 z-scores. If a mother had more than one child, we selected one of the children who was stunted or the last born if none of them was stunted.
Nutritional status was assessed following the World Health Organization reference standards (maternal body mass index (BMI) and height for age z-scores (HAZ) for children). Stunting children was de ned as a HAZ ≤ -2 SD of the WHO/NCHS/CDC new reference standards (WHO, 2006). Mother overweight was de ned as BMI ≥25.0kg/m 2 . The MOSC (de ned as the coexistence of mother overweight and stunting children in the same household) was determined by the presence in the household of overweight in a mother and stunting in at least one of her children. All outcomes variables were coded as a binary variable.
We included several Independent variables. We selected these covariates based on previous evidence of the determinants of nutritional status [13,18]. The mother-level covariates included place of residence (categorized as Urban and Rural), Wealth index (Poorest, Poorer, Middle, Richer, Richest), age group (categorized as 15-19, 20-29, 30-39, 40-49), Education level (categorized as None, Primary, ≥Secondary), Access to information (categorized as No and Yes), Source of water (categorized as Improved and Unimproved), Sanitation facility (categorized as Improved and Unimproved), the number of members in households (categorized as ≤5 members, >5 members), Antenatal Care (categorized as None, 3 times and less, more than 4 times). The child-level covariates included child age ( categorized as ≤24 months and >24 months), child gender ( categorized as Male and Female), Child small birth Weight ( categorized as Small, normal or Large and Missing), Child diarrhea during the last 2 weeks (categorized as No and Yes) and Month of birth (categorized as January, February, March, April, May, June, July, August, September, October, November and December).

Statistical analysis:
This study aims to examine country-speci c relationships between the independent variable and the MOSC occurrence and to conduct cross-country comparisons of the MOSC prevalence. Therefore, we present results as summary tables and make qualitative comparisons across studies. We calculated proportions for the outcome and independent variables, as well as for key characteristics. Multivariate logistic regressions were employed to estimate the association between prevalence of MOSC and the individual and household factors. Logistic regression was performed on the pooled data to allow comparison between countries ranked by gross national income. All analyses were adjusted for the weights and complex survey design of the DHS as recommended by the Guide to DHS Statistics DHS-7 [17]. The analyses were conducted using the Statistical Package for the Social Sciences (SPSS) software (Version 24.0) and Statistical Analysis Systems statistical (SAS) Studio University Edition. A 2-sided P value of .05 was used to assess statistical signi cance.

Descriptive data
Out of the 17 countries of West African countries, DHS data of 11 countries that were available and contain body measurements of mother and their children were analyzed. The years of survey were from 2012 to 2014. Datasets from demographic and health surveys by country used in the analysis, classi ed from lowest to highest in their Gross National Income (GNI) per capita (equivalent in US dollars) are shown in the table1. Table2 presents the descriptive characteristics of the analytical sample of 11 countries based on Demographic and Health Surveys, the sample population in each country except Liberia was predominantly rural. Liberia, Gambia and Ghana samples had the lowest proportion of rural households (46%, 52.8% and 52.7%, respectively) and the highest proportion was in Niger and Mali (80.8% and 79.3%, respectively). The proportion of household wealth index was predominantly that of the poor group in all countries regardless of the GNI per capita of countries (World Bank in 2019) except Niger and Benin in which the proportion of the rich group was higher (40.9%, 41.6%, respectively). The predominant mother's age group in each country was those between 20 to 39-year. The proportion of the mother who had secondary education or more was extremely low in Niger and Mali (4.9%, 8.3% respectively), while in Ghana, this proportion was much larger (53%). Over 60% of mother in each country had access to at least one source of information, the highest proportion was reported in Ghana and Gambia (90.2% and 90.1% respectively). More than 60% of the household in each country except Sierra Leone (56.%) had access on improved water sources, while the access to improved sanitary facilities was low in each country, Sierra Leone and Niger had the lowest proportion of 8.2% and 9.4%, respectively, and the highest of 37.5% was in the Gambia. More than half of households in all countries except Ghana (42.4%) were predominantly made up of 5 or more members. The proportion of mother who received 4 or more antenatal cares was lower in Niger and Mali (27.9% and 35.7%, respectively) and the higher proportion was reported in Ghana and Liberia (83.2% and 71.2%, respectively). The proportion of small birth weight was higher in Gambia and Ghana (50.9% and 51.3%, respectively), while the lowest proportion was in Benin and Niger (18.7% and 19.5%, respectively). The proportion of children who had Diarrhea in the last two weeks preceding the interview was higher in Liberia, Gambia and Ivory Coast (25.5% 21.1%, 21.5, respectively), while the lowest proportion was reported in Benin and Mali (7.9% and 9.3%, respectively).       OR: Odds Ratio from logistic regression models; CI: Con dence Interval. *P < 0.05; **P < 0.01. NA: Not Applicable.

Discussion
The patterns of associations between the individual or household factors and MOSC occurrence varied widely across the country data sets examined and was independently linked to their gross national income per capita.
In our study, the proportion of MOSC was more likely to occur in the urban area than the rural area in most West African countries. One explanation may be that the nutritional transition and sedentary lifestyles may more favor MOSC occurrence in urban areas. An analysis on pooled data of sub-Saharan Africa from a recent study showed a strong association between living in urban and MOSC occurrence [13], another study in which African countries are included reported that the association of mother overweight and stunting children with urbanization do not differ between rural and urban areas [11]. In Bangladesh, DBM was more likely to be more prevalent in urban areas [19], and authors state that the double burden is more likely to occur in rural areas in many LMICs [1,12].
Rich or middle households' wealth index were more associated with higher odds of MOSC occurrence compared to poor households' wealth index in all west African countries except Ivory Coast. Even if a signi cant association was found in 5 countries, MOSC was more likely to appear in the wealthier household but this occurrence was non-linked to the level of a country's GNI. One study reported that MOSC is associated with higher levels of a country's GDP per capita but that it is perhaps the poorest households in those countries that tend to have the MOSC [11]. A study in Guatemala showed that the coexistence of female obesity and stunting children was signi cantly associated with the middle economic quintile [20]. Other studies report that the prevalence of DBM seems to increase with an increase in the wealth index of households [1,12,18,19].
Mother age group was signi cantly associated with the MOSC occurrence in 5 countries Liberia, Niger, Benin, Mali and Nigeria. MOSC was more likely to occur in the 30-39-and 40-49-year age groups than in 20-29-year, this can be explained by the fact that the increase in a woman's age is associated with the overweight occurrence.
In Bangladesh, the maternal age group of 35 to 49 years was associated with the risk of the coexistence overweight/obese mother and child under-nutrition [19] Children over 24 months of age were associated with higher odds of MOSC occurrence compared to under 24 months in all countries except Sierra Leone.
This may be explained by the fact of the gradual accumulation of episodes of malnutrition during the rst years of childhood thus resulting in chronic conditions in advanced age [21]. The same result has been founded in pooled analysis on 14 African countries [13].In Bangladesh, the 24 to 49-month age group was associated with a high risk of the coexistence overweight/obese mother and child under-nutrition [19].
Mother with primary and/or secondary education were associated with higher odds of MOSC occurrence compared to those with no education in all countries except Togo and Guinea, however, MOSC was more likely to appear in mother with primary education with a signi cant association of more than three times in Mali and Gambia, and more than two times in Ivory Coast, while secondary education was signi cantly associated Ivory Coast (with more than three times), Gambia (with more than four times) and Nigeria. Other study found that Higher education in mother was associated with lower odds of MOSC, but was not among the factors strongly associated with it [13].
An explanation can be the cultural aspect that could interact in the sense that obesity considers as a mark of respect and admiration by society [8,22], educated people who move or are already in urban areas are more attached to tradition and growth fat voluntarily. Studies highlighted the pressure of the tradition on mother lead them to growth fat [23].
Mother's lack of access to information was signi cantly associated with lower odds of MOSC occurrence in the Gambia, while the reverse is observed in Ivory Coast and Nigeria in which it was signi cantly associated with higher odds of MOSC occurrence. This can be explained by the fact that most of the population are living in rural areas, have low education and access to information is limited [21].
Household access to the unimproved water source was only signi cantly associated with lower odds of occurrence of MOSC in Ghana, other countries do not show a signi cant association. Household access to unimproved sanitation was signi cantly associated with MOSC occurrence in two countries which are Liberia and Nigeria, while diarrhea in children was marginal or not associated with the occurrence of MOSC in all countries.
Diarrheal disease and infection are the most common cause of unimproved water supply and sanitation thus leading to stunting children. A recent study found that Improved sanitation was signi cantly associated with MOSC in sub-Saharan Africa [13]. However, in this study, it was households with an improved water source that were more likely to have a high proportion of MOSC in Ghana. This can be explained by inadequate conservation or storage of drinking water in households. Cultural practice, the availability of safe water and accessibility to soap can be an obstacle to good hygiene practice for the control of infection prevention in households [24]. Households made up of more than 5 members was signi cantly associated with the MOSC occurrence in the 3 countries with the highest gross income per capita which are Ivory Coast, Ghana and Nigeria. A study in Mali in 2006 reported that household size was not associated with the prevalence of stunting or overweight but the high number of household members is a protective factor against the prevalence of stunting [8].
Antenatal cares were signi cantly associated with MOSC occurrence in 7 countries: Liberia, Niger, Benin, Mali, Ivory Coast, Ghana and Nigeria, women who did not have antenatal cares were more likely to be exposed to the MOSC occurrence than those who had more than 4 antenatal care. That same association was observed in Guinea although antenatal cares didn't show stronger signi cance. The MOSC occurrence was not linked to the GNI of the country, for example, the MOSC was 4 times more likely to occur among women who did not have antenatal cares in Ghana while it was less than 2 times in Nigeria. This can be explained by the easy access to health care and the quality of the consultations in each country, this can be explained by the easy access to health care and the quality of the consultations in each country. The authors of a study from three Latin American countries found that the association of the prenatal consultation with stunting differs between countries and that is due to the quality of each country's prenatal consultation [25]. Another study in Nigeria concluded that a poor prenatal visit is associated with a risk of poor nutritional status in young children [26].
Child sex was not signi cantly associated with MOSC in the majority of countries except Sierra Leone and Mali, where MOSC was more likely to appear signi cantly in female than in the male. Studies in Nigeria and Bangladesh reported that malnutrition was more associated with female gender, while that association was strong among male gender in South Africa. Others studies in Brazil shown than male gender was associated with MOSC occurrence compared to female gender [26][27][28][29].
Children small birth weight was only signi cantly associated with the MOSC occurrence that in two countries, in Niger, it was those who had small birth weight who seemed to be more affected by the MOSC while the opposite was observed in Ivory Coast where small birth weight appeared the most affected. Studies in Bangladesh and Indonesia have reported a positive association between low birth weight and stunting children [30] [31]. An explanation may be that countries, in which normal or higher birth weight was more associated with the MOSC occurrence may have uncontrolled access of individuals or households to adequate, su cient and necessary food for an active and healthy life [32]. Another explanation can be the change in individual and family eating habits, a welleducated people living in urban areas may not have acquired food knowledge to feed properly their children, and for those living in rural areas will always face food insecurity from the area whatever the wealth index levels. Studies reported that the available food resources especially exported food may not be used adequately by many mother [8,33,34], leading to malnutrition and even after controlling by socio-economic factors, food insecurity in an area has a huge in uence on nutritional status [8,19].
Children month of birth were signi cantly associated with the MOSC occurrence in the four highest GNI per capita in 2015 countries: Benin, Ivory Coast, Ghana and Nigeria. In Ivory Coast, September was signi cantly associated with higher odds of MOSC occurrence compared to December, in Ghana no month compared to December showed signi cant association but the month of birth covariate was signi cantly associated with the MOSC occurrence, whereas in Nigeria all months were signi cantly associated with higher odds of MOSC occurrence compared to December except April and August. In Liberia, April, August and September were both signi cantly associated with higher odds of MOSC occurrence compared to December, but the covariate did not show statistical signi cance on analysis.
geographic position of countries. The authors have reported that this association decreases after the child's second year of life [35]. In India, the month of birth was signi cantly associated with stunting [36].
A comparison between countries showed that the MOSC occurrence was not necessarily related to countries' GNI and it was di cult to say that it was related to the proportion of stunting or that of overweight. Taking the country with the lowest GNI as a reference, the prevalence of stunting was signi cantly higher in all countries except Togo in which it was not signi cantly high and Ghana in which stunting was rather signi cantly low compared to that of Gambia.
Whereas the overweight prevalence was two times as high in Ghana but the MOSC did not show a signi cant difference compared to the Gambia. The prevalence of stunting was two times higher in Mali, Niger and Benin than other countries, while overweight in Niger was signi cantly low and that of Benin was signi cantly high but these two countries showed a signi cantly high prevalence of MOSC compared to The Gambia. We can believe that the prevalence of MOSC does not seem to be linked to the high prevalence of stunting or overweight but more to the high prevalence of both in the same country. The point we can make is that the MOSC in West African countries does not show such a large difference in proportion, not in the same way as that of overweight or stunting which shows large differences between countries.
Taking as a reference the previous studies on African countries the prevalence of MOSC has increased, for example, it increased from 2.2% in 1993 to 5.3% in 2014 in Ghana, from 2.0% in 1996 to 11.52% in 2013 in Benin. Apart from Benin all the other countries have their prevalence of MOSC below 10%.

Conclusion
In this study, none of the individual and household factors were strongly associated in the same way with MOSC occurrence in West African countries, they varied considerably between national databases and independent of national income per capita. However, Children's age and antenatal cares were the two covariates strongly associated with MOSC occurrence in most countries (10 and 7 out of 11 countries, respectively). The prevalence of MOSC does not seem to be linked to the high prevalence of stunting or overweight but more to the high prevalence of both in the same country. The message for decision-makers would be to establish speci c measures according to the factors associated with the MOSC of each country and not measures common to all countries.

List Of Abbreviations
LMICs: low-and middle-income countries MOSC: coexistence of mother overweight and stunting children in the same household Authors' contributions: the study was designed by SST and QL; SST performed the data collection, extraction and analysis, under the guidance of ANA, HZ and YH. All authors contributed to the veri cation and interpretation of the results and QL revised the nal draft manuscript. All authors have read and approved the nal version of the manuscript.