Factors Associated with Stunting among Children Aged 0 to 59 Months in Harare City, Zimbabwe

: Background : Child malnutrition is a major public health problem in Zimbabwe and it has been one of the leading causes of morbidity and mortality in children under five years of age in the City of Harare. Methods : An analytic cross sectional survey was conducted on 342 mothers – child pairs of children aged 0 – 59 months in Harare’s suburbs between July and August 2014.The aim of the study was to identify factors associated with the high levels of stunting among children age 0 – 59 months. Bivariate and multivariate analyses were used to establish the risk factors for U5 years stunting. Findings : A total of 29.2% of the children were stunted with (19.3% being moderately stunted and 9.9% severely stunted). More females (32.9%) were stunted compared to (25.9%) males. The main contributing factors for stunting among the children 0 - 59 months were found to be lack of mother’s education adjusted pOR=0.49 (95%CI: 1.04-3.27); being unemployed pOR=1.22 (95%CI: 1.08-2.03); residing in high density suburbs pOR=2.14 (95%CI: 1.61-7.55); child ever being hospitalized pOR=1.04 (95%CI: 1.01-2.14); breastfed after > 1 hour after birth pOR=1.11 (95%CI: 1.09-1.80); complementary feeds < 6 months pOR=1.30 (95%CI: 1.17-2.21); low birth weight pOR=1.46 (95%CI: 1.29-3.51) and having a poor dietary diversity pOR=1.07 (95%CI: 1.01-1.84). Conclusions : Lack of mother’s education is the principal risk factor for stunting among the under five children in Harare exacerbated by unemployment of either the father or mother or relative, area of residence, feeding practices, birth weight, illness and infections and household dietary diversity. Stunting is on the increase in Harare. The findings also revealed a high level of in-equalities in social determinants of health since stunting is related to socio-economic factors. Recommendations : Any nutrition intervention has to use strategies that will comprehensively address the social determinants of health and improve the standard of living of households in the high density areas so as to have a ripple effect on the nutritional status of U5s


INTRODUCTION
Stunting is a nutritional problem that is deeply rooted in poverty and deprivation. Grantham [1] purports that stunting affects mainly developing countries. Globally, it is estimated that about 178 million children younger than 5 years of age are stunted, WHO [2]. It is estimated that 95% of malnourished people live in developing countries in Sub-Saharan Africa and South Central Asia, WHO [2].
According to WHO and UNICEF [2] an estimated 165 million below the age of five years were stunted translating to 26% of under-fives being stunted globally.The World Health Organization [2] classifies a child as stunted (chronically undernourished) if his or her height-for-age index is more than 2 standard deviations below the reference median. Children who Furthermore the Food and Nutrition Council [5] survey indicates that boys were more likely to be malnourished than girls across all indices. In Zimbabwe the prevalence of stunting among children aged 0 -59 months has increased over the years from 21% in the 1994 ZDHS [6] to 28% in the 2005/6 ZDHS [7] to the current 33.8% as indicated in the 2010/11 ZDHS [8].
The prevalence of children with a short stature in Harare as of 2010 was 28% (CI: 25.1 -32.4), Food and Nutrition Council [5]. The ZDHS [8] 2010-11 indicates that the overall prevalence of stunting in Harare stands at 29% and that for severe stunting is 9.3%. According to the Zimbabwe National Nutrition Survey [9] of 2010 the percentage of children between 6 and 59 months of age classified as stunted is 28.7% against a target of below 2.3%. As a consequence approximately one in every three children aged 0-59 months in Harare is stunted.
According to UNICEF [3] Reports on nutrition, causes of stunting are multifaceted and multidimensional and include poor complementary feeding as well as not exclusively breast feeding children.
In Harare the City Health Department's mortality patterns indicate that over 40% of children under 5 years who die have malnutrition as an underlying cause. Thus malnutrition is the 4 th cause of death among the < 24 months age group and 3 rd leading cause of death among the <5 age group. Stunting increased over the years from 2007 to 2011 in Harare, with a year on year increment in its prevalence (Harare City Health Department annual reports, 2007 -2011) [10].
Against this background we therefore set out to undertake a study so as to explore the possible risk factors that determine chronic malnutrition among children under 5 years of age in Harare City during the year 2014.
Social determinants of health are factors that can determine the health outcome of households and communities in both negative and positive terms. Social determinants are also responsible for exacerbating inequalities in society and the inequalities further decrease chances of individuals to attain health. Various studies for U5 nutrition in developing countries have identified that child malnutrition is related to these inequalities.
Mandrefo et al. [11] carried out a community based cross sectional survey to determine risk factors and the prevalence of malnutrition among 796 children 6-59 months paired with care-givers in Ethiopia. Their findings indicate that underweight was significantly associated with having more than 4 children born to the mother, poor education of the father, and having had diarrhoea in the past 2 weeks. Not practicing family planning and time of commencing complementary feeding were related to wasting.
Ngianga-Bakwin Kandala et al. [12] investigated the impact of geographical setting on the nutritional status of children U5 in the Democratic Republic of Congo using data from a national household representative sample. Their methods of analysis enabled them to examine geographical variation of the nutritional status of U5. They were also able to determine the extent to which the geographical pattern is influenced by other factors such as socio economic factors, and soft but volatile issues such as political, environmental, conflicts and cultural factors. In the end their analysis showed that malnutrition was present in all the provinces but it was more marked in rural areas Tishome et al. [13] conducted a cross sectional study to identify the determinants of stunting focusing on food surplus areas in Ethiopia so as to identify the differential impact of other determinants such as demographics and psychographics and health factors. Factors related to malnutrition included the sex of the child, diarrhoeal episode, age at introducing complementary feeding and duration of breastfeeding.
Zere and D. McIntyre [14] analyzed data from the Standards and Development Survey from a sample of 3755 under five children so as to assess the burden of socio economic inequalities against under-five (U5) malnutrition in South Africa. Household income was used as a proxy indicator for economic status. The different types of malnutrition were calculated using the illness concentration index. Stunting was found to be concentrated in the Eastern Cape and Northern provinces which are already known to be very poor provinces. The findings also revealed that malnutrition was related to socio economic status.
However there were no inequalities in nutritional status among children from the white households. Their findings supported the assumption that inequalities in socio economic status of households does affect the nutritional status of U5 children negatively or positively.

Statement of the Problem
Maternal and child undernutrition is directly attributable to 35% of all global child deaths [4]. In Harare the Health Department's mortality patterns indicate that over 40% of children under 5 years who die have malnutrition as an underlying cause. Thus malnutrition is the 4 th cause of death among the < 24 months age group and 3 rd leading cause of death among the <5 age group. The deterioration in nutritional status of the children under-five years is a public health concern. Stunting increased over the years from 2007 to 2011 in Harare [6][7][8] with a year on year increment in its prevalence Harare City Health Department annual reports, 2007 -2011 [10].

Justification
Stunting is preventable and measures can be put into place to prevent its occurrence. The study was conducted so that modifiable factors associated with the development of stunting can be identified and targeted for preventive programming. Factors associated with and the possible causes of stunting among the < 59 months age group had not yet been explored in Harare's suburbs. Evidence generated from the study is required to inform policy, practice as well as to generate interest in further studies.

1.
What is the prevalence of stunting among < 59 months age group in Harare city?

2.
What are the principal causes of stunting in Harare city among < 59 months age group?

Objectives
To identify and determine factors associated with stunting among children age 0-59 months in Harare.

Specific Objectives
To determine the prevalence of stunting among children aged < 59 months age group.
To assess the socio-economic factors associated with stunting in < 59 months age group.
To determine health service related factors associated with stunting.
To determine patient related factors associated with stunting.
To make appropriate recommendations for interventions.

Data Capturing and Analysis
Epi Info version 3.5.3 was used to enter, clean and analyze the data. Frequencies and proportions were be generated for independent variables. In the Univariate analysis, chi-square test/ univariate logistic regression analysis was performed to determine associations between independent variables and outcome of interest stunting. Stratified analysis was done to control for confounding factors and to identify effect modifying variables. Multivariate logistic regression analysis was performed to identify independent risk factors associated with stunting. The logistic regression model included all variables with a p-value of 0.25 or less. Logistic regression results were presented using odd ratios (OR) and their 95% confidence interval (CI).

METHODS
An analytical cross sectional study was conducted in 4 districts in Harare City on children aged 0 -59 months and their caregivers. A Harare resident with a child aged 0 -59 months and caregiver aged 18 years or above were included in the study. A minimum sample size of 337 respondents was determined. Multistage sampling technique was used.The first stage involved the random selection of four 4 districts out of 8 administrative districts using the table of random numbers. The second stage involved randomly selecting 2 suburbs from each selected district. At the third stage, households were randomly selected using the table of random numbers from the list of households available. At household level a motherand-child pair was identified and studied.

Ethical Procedures
The proposal was approved by the College of Health Sciences Joint Institutional Review Board (JREC) and the Medical Research Council of Zimbabwe. During data collection Guardians were provided with explanation about the objectives of the study and were given time to ask questions in any aspects of the study that they did not understand All respondents signed a consent before the data collection process was carried out.

Data Collection
Data was collected using a structured interviewer administered questionnaire to a mother/caregiverchild pair to establish the determinants and risk factors of stunting among the 0 -59 months old children. This questionnaire contained sections to assess information on demographic, socioeconomic, anthropometric, health related and dietary factors.

RESULTS
A total of 342 children aged 6 -59 months were sampled into the study, their median age was 16 months (with interquartile range (IQR); Q 1 =11, Q 3 =26) and 178 (52%) were males. A total of 342 questionnaires were administered to the mother/caregiver -child pairs and corresponding anthropometry was measured. Most (96%) of the children in the sample were born in health care institutions. Overall stunting in children 6 -59 months was 29.2%, with moderate stunting at 19.3% and severe stunting at 9.9% ( Table 1). Compared to males, more females were stunted (32.9% vs. 25.9%), while more males were wasted than females ( Table 2).
Most of the mothers/caregivers had attained secondary education 271 (79.2%) whereas 55 (16%) had completed primary schooling. Most respondents were unemployed 242 (71%) and these women further rely on their husbands as the main sources of family income 334 (97.7%).
The median weight of non-pregnant/breastfeeding mothers were 62 kgs (with IQR; Q 1 =54.9; Q 3 =70.2 kgs) with a corresponding median body mass index of 23  of the interview. The report of a child having ever been breastfed was universal with 335 (98%) having been given mother's milk.  Most of the children (96.2%) were delivered at a health institution, however only 201 (60%) were put to breast within the first hour, furthermore 139 (81.5%) were weaned from breast milk before the internationally agreed 24 months. The children' median weight was 10.3 kgs, median height was 10.kgs and the mid-upper arm circumference was15.1 centimeters.  Table 7 above shows the history of food intake in the last 24 hours and seven days. Most (73.3%) of the respondents reported a food score of more than 6 out of 10. The median birth weight of children was 3 015 grammes (with IQR; Q 1 = 2 700, Q 3 = 3 350) while the median age of introduction of complementary feeds was 6 months (with IQR; Q 1 = 3, Q 3 = 6). In addition the median number of meals children ate the previous day was 3 (with IQR; Q 1 = 2, Q 3 = 4). Most of the mothers 284 (83%) were aware of the consultation fees by type of service required.
Household treatment of drinking water was low with a proportion of 133 (38.9%). Most respondents reported sharing one toilet with other households and the median for families sharing a toilet was 2 families (with IQR; Q 1 = 1, Q 3 = 4). Toilet usage was however high with most mothers 308 (90.1% indicating they had used toilets to dispose of their children' faeces. Majority of the mothers 139 (40.6%) highlighted the borehole as their preferred choice of water for drinking. Municipal water was unpopular as indicated by the 123 (36%) who resort to drinking municipal water as boreholes in their neighbourhood are not working. Borehole water was the main alternative source of water for drinking 208 (60.8%), followed by municipal water 69 (20.2%) and protected wells coming third at 54 (15.8%).     Harare as a metropolitan city has sewer systems with flush toilets 296 (86.5%). However some respondents indicated they have been without water for periods spanning years, hence they constructed Blair toilets or pit latrines that is 36 (10.5%) and 9 (2.6%).
From the univariate analysis factors, Tables 5-7 residing in high density (OR = 2.1495% CI 1.61-7.55)), having an unemployed caregiver (OR = 1.2295% CI 1.08-2.03) and having a mother with primary education (OR = 1.8595% CI 1.04-3.27) were significantly associated with stunting. In particular residing in high density areas increased the risk of child stunting by 21%, while having an unemployed caregiver increased the risk by 12% and having a mother with primary education increased child stunting risk by 19%.  Ever being admitted to hospital (OR = 1.04,95% CI 1.01-2.14), being put to breast after 1 hour (OR = 1.11,95% CI 1.09-1.80), having complement feeds before 6 months of age (OR = 1.30, CI95%1.17-2.21), and a child birth weight of less than 2500 grammes (OR = 1.46,95% CI 1.29-3.51) were significantly associated with child stunting. In particular ever being admitted to a hospital increased risk of child stunting by 10%, having complementary feeds before 6 months increased child stunting risk by 13%, and low birth weight increased child stunting risk by 15%.
Having a household dietary diversity score of less than 5 out of 10 was a statistically significant risk factor to child stunting. Drinking treated water at household level was protective of stunting but this was not statistically significant.
Stratified analysis was carried out for a number of child variables and maternal elements to check for effect modification and control for possible confounding. After controlling for confounding, only mother's education remained statistically significant. Having a caregiver who completed secondary education reduced the child risk of stunting by 49% (Adjusted POR 0.49 95% CI (0.26 -0.94).

DISCUSSION
The study aimed to establish the prevalence and risk factors for stunting in Harare city. The study found out that (29.2%) of children aged 0 -59 months were stunted. Compared to the year 2012, this demonstrated a 1% increase in the prevalence of stunting. Our findings of 29% prevalence are consistent with the classification of Zimbabwe as a medium to high stunting country according to the WHO [5]. Furthermore prevalence established is almost similar to the ZDHS [8] 2010/11 stunting prevalence for Harare which stands at 28.3%. The main statistically significant risk factors for stunting found from the study were: the mother having had only primary education, being unemployed, a child being admitted to a hospital, a child who was put to breast after 1 hour of birth, a birth weight of 2 500 grammes and less and a household dietary diversity score of less than 5 out of a possible 10. Significant protective factors to stunting identified from the study were: being married, having the husband as the main source of income, a household monthly income of more than USD$300 and drinking treated water/clean water at household level.
The insignificant protective factors were: gender being male, home deliveries, having a full gestational term for the child and stopping breast feeding after 24 months of age. It must however be noted that all study children had up to date vaccination statuses hence vaccination was not analyzed in this current study.
In the current study, being from the less privileged groups was a risk factor initially but disappeared in subsequent multivariate analysis. The effects of income are known to be mediated through other underlying determinants. These findings are consistent with findings of a study on stunting in Libya by Taguri, A.E. et al. [15]. These factors determine the ability of the family to combine their knowledge, resources and patterns of behaviour, to promote, recover or maintain health status and to cope with a difficult environment. Such factors include parental education, psychosocial stimulation and household environment. People living in urban areas are supposed to be provided with better access to health services, education and other social support systems which are no longer readily available or easily accessible to residents in Harare in the last decade due to economic decline and over population.
Lack of education was shown to be a significant risk factor to stunting. This is consistent with a study conducted in Ethiopia by Christiansen [16] which concluded that a child with a caregiver with less than 12 years of schooling was at highest risk of stunting POR 3.2 at 95% CI (2.9 -3.89). In agreement with most studies low maternal education was a major determinant of stunting in the study. These findings are further supported by Mandrefo et al. [11] who identified that undernutrition was significantly associated with poor education of the father and having more than four children born to the mother.
Expectedly, as the duration of mother's education increases, her finances and consequent contribution to the family income also increase. This places the family at a higher social class and, therefore, better nutritional status. In addition, educated mothers are more likely to make decisions that improve child nutrition and health. An educated woman is more likely to send all her children to school, thereby breaking the chain of ignorance; making better use of the childhood survival strategies, such as sufficient breastfeeding, immunization, household oral rehydration therapy, and family planning.
The occupational status of the mother, specifically being unemployed was a significant risk factor for child stunting. This can be linked to lack of education which results in women having a low socioeconomic status. This affirms B. Teshome et al. [13] findings in their study on magnitude of stunting in children less than five years of age in food surplus region of Ethiopia. They found that unemployed women were almost two times likely to have a stunted child than their economically privileged counterparts OR 1.98 (95% CI 1.21 -2.86) [13].
Area of residence was shown to have an influence on stunting. Residing in high density was shown to be a risk factor for stunting in the study. This affirms what R.A. Ricci, et al. [17] concluded whereby place of residence can be used as a proxy for economic status. In a similar study the finding Kandala et al. [12] in the democratic Republic of Congo confirmed that living in a rural area within certain environmental cultural settings impacted upon nutritional status where those in rural poor areas were stunted compared to the urban group.
According to the study high density residence entailed overcrowding, inadequate water and sanitation facilities. This sustained exposure of children to adverse environments (poor water and sanitation) retards child linear growth. Findings from a study by Zere and McIntyre [14].
To identify the burden of socio economic inequalities on U5 nutritional status support the above findings since stunting was found to be related to poor socio economic status of households in the Eastern Cape and Northern Provinces which are known to have a poor socio economic status.
According to this current study being male was shown to be an insignificant protective factor to stunting. This is plausible as it is consistent with Teshome et al. [13] studies in Ethiopia which showed that female children are at a higher risk of stunting than male children.
Place of birth, specifically home delivery was shown to be a protective factor to stunting. This statistically insignificant factor is not credible as it disagrees with findings on studies conducted on chronic malnutrition. To authenticate this, A. Rayan et al. [18] in his study in Bangladesh proved that children delivered at home under traditional delivery systems without medical care had a higher risk of stunting. This was largely attributed to the fact that in medical set ups health education on complications; maternal and child nutrition and best feeding practices is given which may have positive outcomes on maternal and child health.
The father as the main source of income was shown to be a protective factor. Though insignificant, this is in harmony with findings by Christiansen [16] where father's education emerged as an important factor associated with stunting. Their analysis concluded that children whose fathers are educated were less likely to be stunted compared to children with uneducated fathers. This was because fathers are the main income earners and decision-makers, thus their education plays an important role in ensuring they provide better nutrition for their families. Fathers may have an advantage in accessing a better job and thus increasing family income. Adored and Abiola [19] Assessed factors related to nutritional status of children under five. The prevalence of stunting was found to be 12.5%. Factors such as household size, low educational status of the mother negatively correlated with nutritional status of the child. Higher educational status of the mother, hygienic practices in food preparation were positively correlated with the nutritional status of the child.
Marital status was shown to be protective from stunting. This insignificant characteristic is paradoxical solely due to the factors that studied marriage in relation to stunting focused on monogamous and polygamous marital unions. According to Senbanjo et al. [20], polygamy was shown to be a risk factor for stunting in such studies as there may be competition for nutritional resources within this largely predominant African custom.
Though insignificant, monthly income of more USD$300.00 was shown to be protective to stunting in the study. This is consistent with findings from a study by Orach [21] in Ghana on the effect of economic inequality on chronic childhood undernutrition in Ghana. He found that children from the poorest quintile of households are more than twice as likely to suffer from stunting compared to children from the richest quintile.
Amongst the child characteristics associated with stunting, time first put to breast was found to a statistically significant factor. In essence being put to breast after 1 hour is a risk factor. This concurs with studies in Ethiopia by B. Teshome [13] which indicated that babies who did not receive colostrum were at a higher risk for stunting. This was attributed to assertions that infants who do not receive colostrum may have high incidence, duration and severity of illnesses such as diarrhoea.
Giving children complementary feeds before they reach six months of age was also a significant risk factor in the study. This is in agreement with other studies where prominence of other food materials before 3 months was among feeding practices explaining risk of stunting. Ayaya [22] in Ethiopia showed that children who received complements before 6 months were found to be at a significantly higher risk of stunting than those who did not.
Low birth weight was established to be a significant risk factor for stunting. This affirms the finding by [23] where the prevalence of stunting, especially among children less than 2 years of age, reflected the prevalence of low birth weight in a population. This influence of birthweight on child linear growth confirms other findings that showed links between birth weight and growth through infancy and beyond.
Amongst the plausible yet insignificant child characteristics analyzed was history of illness in the last 2 weeks for the child. Not being ill is protective of stunting according to this study. This is in tandem with findings of other studies where incidence of diarrhoea is negatively associated with stunting. Guerrand et al. [24] reviewed the evidence relating to the association between stunting and episodes of diarrhoea. They noted that a large proportion of populations in the world live in absolute poverty. This situation is further exacerbated by lack of access to clean water and safe sanitation. These communities experience high episodes of diarrhoea among the under five children. Diarrhoea in turn contributes to an impoverished gut which reduces the absorption of nutrients which in turn results in malnutrition and stunting of the child under five. The cycle of diarrhoea malnutrition perpetually contributes to chronic malnutrition thus demonstrating the reciprocity of disease and the environment in form of social determinants of health.
According to this study age at which a child is stopped breastfeeding if more than 24 months was shown to be protective to stunting. This however is divergent with B. Teshome et al. [13] findings where their study revealed that prolonged breastfeeding is indeed a risk factor for stunting. It was paradoxical children breastfed for more than 2 years were at increased risk of stunting than those who did not receive breast milk beyond 2 years.
In as far as household and community characteristics, household dietary diversity that is a total food score of less than 5 out of a possible 10 was revealed to be a risk factor. This concurs with other studies in the developing regions where diets are mainly cereal based. Thus complements given to children are plant sources and conversely feeding with animal sources is low, so is the vegetable and fruit consumption thus consequently children fail to access enriched complementary diets.
Notable plausible insignificant household and community factors were drinking treated water at household level and use of proper sanitary facilities. These factors were shown to be protective form stunting. This is in accordance with findings by Judith A. Ricci [17] in Philippines who found out that continued exposure to adverse environmental conditions over an extended period of time retards child linear growth.

CONCLUSIONS
The study aimed to establish the prevalence of stunting and the factors associated with stunting In addition also aimed at establishing the specific determinants of stunting at household and community level so as to provide recommendations for interventions in the City of Harare.
The findings have provided a very worrying trend in terms of the prevalence of stunting among the 0-59 month age group which is clearly on the increase as shown by various studies. This is reflected by the fact that the findings indicate that (29.2%) of children aged 0 -59 months were stunted. This shows an increase of 1 % since 2012. The status of stunting is not getting any better since Zimbabwe is already classified by WHO as being in the medium range 30-39.9% which in global stunting. The ZDHS [6] carried out every five years also confirms prevalence of stunting to be at 29% for children U5 years (0-59) months in Harare. The findings show that malnutrition is getting worse. The findings however show that the factors that are associated with stunting ca be addressed through improving the environment by addressing social determinants of health such as poor access to clean water and access to employment. It is also worrying that the risk factors in the overall framework of social determinants of health are currently being exacerbated by the poor economic situation which has devastated Zimbabwe in the past decade. The study was carried out in the high density suburbs where inequalities in social determinants of health such as: housing, access to water, education and employment are normally found. It is our hope that the inclusion of the Nutrition cluster in the current Zimbabwe Economic blue print ZIMASSET [25] will facilitate the improvement of household and country food security through improved agricultural methods and economic reform and general improvement in institutional and economic governance at all levels.