Testing the impact of orphan status, illness in the household, and a child's relationship to the household head on wasting in Nigeria: implications for Hamilton's rule

ABSTRACT

Existing studies on the impact of orphanhood on wasting in sub-Saharan Africa were historically conducted in East and South Africa. These studies have been interested in understanding how close kinship ties with household heads impact orphans' risk of wasting, considering economic resources in the household. The unavailability of secondary data in the past made exploring the idea difficult. Little is known on the subject in West Africa, particularly Nigeria. This research undertakes this topic by analyzing the 2013 Nigeria Demographic and Health Survey Data and employing a multi-level logistic regression model. The research examines the interaction of being a paternal orphan or vulnerable child and the child's relationship to the household head on early childhood malnutrition as measured by weight-for-height z-score, wasting. It also ascertains whether the interaction effect varies with household income. Results show that, regardless of residence in poor or non-poor households, being a paternal orphan or vulnerable child coupled with being a grandchild produces the highest probability of wasting, especially in poor households. Further, results show that extended families and relatives, particularly grandparents-headed households, are not meeting the nutritional needs of paternal orphans or vulnerable children who are grandchildren, especially in poor households.

KEYWORDS:

• Extended family
• wasting
• kinship
• orphans
• poverty
• sub-Saharan Africa

Disclosure statement

No potential conflict of interest was reported by the author(s).

Availability of data and materials

The datasets generated during and/or analyzed during the current study are available in the Demographic and Health Surveys repository; web address [https://dhsprogram.com/data/available-datasets.cfm].

Research involving human participants and/or animals

This article does not contain any studies with human participants or animals performed by any of the authors.

Authors’ contributions

First author made substantial contributions to the creation and conception of the ideas in the research. She also wrote the initial draft of the manuscript. The second author performed the data analyses and worked with the first author in the interpretation of data analyses results. Both authors have reviewed, edited, and approved the submitted version.

Notes

1 An orphan is defined from a biological standpoint of being a child under the age of eighteen years whose mother, father, or both parents have died due to any cause (UNAIDS, UNICEF and USAID, 2004). A maternal or paternal orphan is a child for whom one parent (mother or father, respectively) has died. A double orphan has no living parent (UNAIDS, UNICEF and USAID, 2004). Orphanhood can also be defined from a social perspective when a child is abandoned because of poverty by his or her biological parent or the child grows up with at least one of them absent (Abebe & Aase, 2007; Chirwa, 2002; Coneus et al., 2014).

2 The United Nation's Convention on the Rights of the Child (UNCRC) and the Organization of African Unity's African Charter on the Rights and Welfare of the Child (ACRWC) define childhood as the period in which a child is below the age of eighteen years old, so long as the law in the country of residence does not state that majority is attained earlier (OAU, 1990; UNCRC, 1989).

3 Extended families are large networks of people extending through varying degrees of relationship including multiple generations, over a wide geographic area, and involving reciprocal obligations (Ankrah, 1993; Biemba et al., 2010; Foster & Williamson, 2000).

4 Eastern and southern African countries were hit the hardest early by the prime age adult HIV/AIDS crisis in Africa, which led to a high rate of orphans in the excess of ten percent in some countries (Case et al., 2004). Initially, western African countries did not experience the high levels of adult HIV/AIDS that were observed in East and South Africa, though the percent in the former is now climbing (Case et al., 2004).

5 Multilevel logistic regression was used because the dependent variable is dichotomous, Yes/No, which cannot be analyzed any other way. Measures such as correlations are suitable for bivariate relations that involve mostly interval-ratio variables. The research questions called for a multivariate analysis because we controlled for several independent variables as we examined the dependent variable. Indeed, our research objectives were met as indicated from the analysis.

6 Theory provides that some variables act in tandem to affect the dependent variable. That is why we have incorporated variables that work in combination to influence the dependent variables in the model. When the interacting variables are modelled the main effects are not interpreted, rather the interaction is what is important and as such interpreted. For logistic regression, interactions are non-linear, so we cannot omit them from the model because they are not significant as per the p-value. Further analysis using predicted probabilities is undertaken on to measure how the interaction influences the dependent variable at various levels of the interacting variables. This not only makes it easier to interpret the interaction but also provides an intuitive tool to examine the relationship.

7 Details and confirmation on the validity and reliability of all measures can be found in the Survey Organization Manual/Demographic and Health Surveys Methodology that accompany the 2013 Nigeria Demographic and Health Survey Data at https://dhsprogram.com/Data/.

8 The children who have no relation to their household head were included in the category of relation because there were only seventy-seven observations.

9 Cross-tabulation is a bivariate measure of how two variables are associated or related. It does not control for other factors that are likely to influence the dependent variable. That can only be done in a multivariate context.

10 Clusters are enumeration areas that are equivalent to local administrative areas or neighborhoods.

11 The result for the null model is not shown here. It is available upon request.

12 Both odds ratio and probability measure how likely an outcome will occur. Odds ratio provides a unique measure that describes the effect of an independent variable on a dichotomous dependent outcome in a logistic regression framework. An odds ratio that is equal to 1 means that the event is equally possible. An odds ratio that is greater than 1.00 means an increase in the odds of the event occurring, and an odds ratio below 1.00 implies a decrease in the odds of the event happening. On the other hand, predicted probabilities in a logistic regression provides a sense of how the dependent variable behaves over a range of values of key independent variables, which is not possible with odds ratio. Predicted probabilities are especially useful and intuitive when evaluating interaction terms in a non-linear regression, which logistic regression is a part. Probabilities ranges from 0 to 1 with values close to 0 and 1 indicates low and high probability, respectively.

Additional information

Funding

This research did not receive any specific grant from agencies in the public, commercial, or not-for-profit sectors.