QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Leroy, J. L. Articles by Habicht, J.-P. Search for Related Content PubMed PubMed Citation Articles by Leroy, J. L. Articles by Habicht, J.-P.

---

Community and International Nutrition

Only Children of the Head of Household Benefit from Increased Household Food Diversity in Northern Ghana^1,2

³ Center for Evaluation Research and Surveys, National Institute of Public Health, Avenida Universidad 655, Col. Santa María Ahuacatitlán, CP 62508, Cuernavaca, Morelos, Mexico; ⁴ School of Medicine and Health Science, University for Development Studies, Tamale, Ghana; and ⁵ Division of Nutritional Sciences, Cornell University, Ithaca, NY 14853

^* To whom correspondence should be addressed. E-mail: jleroy{at}correo.insp.mx.

	ABSTRACT

TOP ABSTRACT Introduction Materials and Methods Results Discussion LITERATURE CITED

 
In many societies, foods are preferentially channeled to certain members of the household. We studied whether being the child of a powerful household member (head of household or first wife in a polygynous family) was associated with greater child stature in Northern Ghana and how this association varied with differences in household food availability. We used a sample of 464 children between 9 and 36 mo of age in extended households in rural Northern Ghana. Child stature was regressed on household food availability, the status of the child's father (head of household or other male), the status of the child's mother (marital order in a polygynous marriage), and the interaction terms between household food availability and parental status. The models were controlled for child age, sex, maternal height, parity, household size, and potential intra-community clustering. Household dietary diversity was associated with child stature (P < 0.05), but this association was limited to children of the head of household. For children of other males, there was no quantifiable association between household dietary diversity and child stature. Children of monogamous mothers were taller than children of second wives (P < 0.05). Our findings show that studies of intra-household allocation need to investigate beyond gender differences. Other structural household factors need to be considered in designing interventions, because they affect impact and even lead to increased intra-household inequality. Our results are relevant for Northern Ghana and as well as for similar settings elsewhere in the world.

	Introduction

TOP ABSTRACT Introduction Materials and Methods Results Discussion LITERATURE CITED

Various motivations for intra-household allocation patterns have been identified. A misperception of dietary needs of boys and girls in Guatemala led to differential infant feeding practices (2). Pitt et al. (3) found evidence that food allocation favoring men in Bangladesh was rational from an economic efficiency perspective, because investments in men had a greater perceived economic return than in women. The propensity to foster one's own genes has been suggested as a mechanism to explain discrimination against stepchildren (4). Status within the household has been identified as the source of power necessary to allocate resources (5).

In many cultures, the highest status in the household is held by the male "head." This individual has decision-making authority, economic responsibility for the household, and is often the most respected person (6). In Northern Ghana, as in many African societies, extended households are comprised of different families. Each family is composed of an adult male, his wives, and their children. One household member, normally an older male, is the head of household.

The head distributes access to land to the families within the household and allocates food from the household stores for the household's main meal. This meal is prepared for the whole household. All members within a family eat from the same pot. In contrast to the main meal, other food is prepared and eaten within the family in a less coordinated fashion. In the families making up the household, the first wife wields more decision-making power than the second and later wives and thus might be in a position to provide more or better quality foods to her children.

We studied children's attained height in extended (i.e. multi-family) households in Northern Ghana. In situations of inadequate linear growth, child height is primarily determined by dietary intake and health status (7). The dietary intake of the child is the result of the food available to the child's caregiver and the caregiver's feeding behaviors. The food available to the caregiver depends on how food is allocated within the household. Given the intra-household status relationships in this population, we postulated that children of powerful members of the household, i.e. the children of the head of household and the children of first wives, would be better fed and therefore taller than children of other members. We also postulated that improved household dietary diversity (a proxy for household food availability) would be associated with the stature of all children, albeit more so for children of the head of household and children of first wives.

	Materials and Methods

TOP ABSTRACT Introduction Materials and Methods Results Discussion LITERATURE CITED

 
    Data collection and sample. A survey of households with children <3 y old was conducted in 2001 by the International Food Policy Research Institute (USA) and the University for Development Studies (Ghana) in Savelugu-Nanton, a rural district in Northern Ghana. The survey was part of the baseline data collection for the evaluation of a UNICEF development program. All data were collected before any of the intervention activities took place.

Data were collected in a total of 64 villages (8). In every village, all households with a child <3 y old were selected. A household was defined as a group of people who consume food from the same stocks and share at least some meals together. The households with a child <3 y were identified in a census conducted before the survey. When there was more than 1 child <3 y in the household, an index child was selected based on the alphabetic order of the first name. Thus, only 1 child per household was included in this data set. In all but 2 (large) villages, all eligible households were included in the survey. In the 2 largest villages, a random subsample of 95 households was drawn. Data were collected in a total of 1684 households. Ninety-six households (5.7%) had incomplete data because of the households' refusal to participate and fieldworker errors.

Data were collected by locally hired fieldworkers who were trained for 6 wk. The training included extensive classroom and field exercises. Household data were collected by means of a structured questionnaire on household composition, household family structure and composition, sociodemographic characteristics, housing characteristics, household assets, household food consumption, maternal parity, and child health and nutrition. The questionnaire was extensively field tested by the research staff, field supervisors, and fieldworkers and is available from the International Food Policy Research Institute (9). Weight and recumbent length (in children <24 mo) or standing height (24- to 36-mo-old children and mothers) were measured using standard anthropometric methods (10) by highly trained and standardized anthropometrists (11). Weight was measured to the nearest 100 g with standardized electronic UNICEF scales (Uniscale, SECA) and length or height to the nearest mm with locally produced portable infantometers/stadiometers.

Verbal consent was obtained after reading the objectives and survey methods to the household members. The study protocol and the manner in which verbal consent was obtained from study subjects were approved by the Cornell University Committee on Human Subjects.

The analyses presented here were restricted to children who met the following criteria: 1) their households included at least 2 families comprised of an adult male, his wife (or wives), and their children, of which 1 family was headed by the individual identified as the "household head" (an example of a multi-family household structure is shown in Fig. 1); 2) the child's biological mother lived in the same compound as the father; and 3) the children were between 9 and 36 mo of age 2 (the age range of introduction of diverse diets to younger children in this population). We excluded 21 children of third and fourth wives, because their main and interactive coefficients were insignificant and they complicated the model without adding information. Of the 526 eligible multi-family households, height-for-age data could not be calculated for 55 children, the height of the primary caregiver was not available in 6 households, and household food availability was missing for 1 household. The sample size for the analyses consisted of 464 or 88% of the eligible households.

View larger version (15K): [in this window] [in a new window]   FIGURE 1  Example of a multi-family household in Northern Ghana, 2001.

The head of household was identified by the household members as the person with primary responsibility for providing members with food and care. We classified a child's mother as a first wife if she was the first wife in a polygynous marriage or the only wife in a monogamous marriage. Mothers who were the second wife in a polygynous marriage were classified as such.

Household dietary diversity was defined as the number of 9 all-inclusive food groups used in preparing the communal household meals. Household food consumption was obtained by asking all household cooks about the foods used in the preparation of communal household meals for the 7 d preceding the interview. The food groups used were dairy, cereals, roots and tubers, legumes, fruits, vegetables, meat, eggs, and fats.

Maternal parity was used as a proxy for birth order. It may be considered a proxy, because the index child was one of the mother's youngest children. Birth order affects HAZ at these ages. Four groups were created: mothers with a parity of one, 2–3, 4–6, and 7 children.

    Analytical methodology. To test the hypotheses, we used a regression model that included main effects for the status of the child's father (head of household vs. other male) and mother [first (or only) wife vs. second wife] and the interaction between status and household dietary diversity.

We first estimated the main effects model (i.e. the model without interaction terms) and then estimated the following interaction model:

where HAZ is the child's HAZ, H is a dummy variable for the status of the child's father (head of household or not); W captures the status of the child's mother (first wife or not); HDD is household dietary diversity adjusted for data collector fixed effects (8); W x HDD and H x HDD are the interaction terms between the previous variables; CA and CS are child age and sex, respectively; MH is maternal height, P is maternal parity; and SIZE is household size. The models did not control for parents' schooling, because <8% of the parents in the sample had any education.

For the head of household and the marital order variable, the hypothesized most powerful members were chosen as the reference group (i.e. the head of household and the first wife). With this coding system, (in the interaction model) is the slope of household dietary diversity for children of the high status household members, and and were the slopes for children of other males and second wives, respectively.

The models were adjusted for the potential lack of independence of children within villages by using random effects models. These models have an individual level error term (e_ij in the model) and a cluster or village level error term (u_j). There was no household clustering, because we only sampled 1 child per household. We tested for the significance of village level random effect (u_j) using the Breusch and Pagan Lagrangian multiplier test (16). Because these random effects were insignificant for all models, we used the ordinary least squares procedure.

One-sided significance tests were used for the covariates with a priori assumptions about the direction of the effect (head of household, marital order, household dietary diversity, and their interaction terms). For all other variables, we used 2-sided significance tests. P-value < 0.05 was considered significant for a main effect. A cut-off of 0.10 was used for interactions, because statistical power is lower for interactions than for main effects. All statistical analyses were carried out with STATA (StataCorp, version 8.2).

	Results

TOP ABSTRACT Introduction Materials and Methods Results Discussion LITERATURE CITED

 
The mean age of the children was 22 mo and slightly more than one-half of them were boys (Table 1). The mean HAZ was –1.95 and the prevalence of stunting was 47%. One-half of the children were children of the head of household. Almost all of the other children were either grandchildren of the head of the household or his nieces or nephews. Four of 5 mothers were the first wife of their husband. The mean annual per capita expenditure was equivalent to $78 U.S. dollars. Households spent over one-half of their income on food. The mean household size was between 16 and 17.

Model 2 tested the interaction between the parents' status and household dietary diversity. The interaction between household dietary diversity and the status of the child's father in the household (P = 0.061) was significant as predicted from our second hypothesis. However, in contrast to our prediction, the magnitude of the interaction eliminated any effect of increased household dietary diversity on child height in the children of other household males. The magnitude of interaction between household dietary diversity and the status of the child's mother (first vs. second wife) was substantial and negative as predicted but was not significant (P = 0.188).

Model 3 presents the parsimonious model, omitting the nonsignificant interaction from Model 2. At high levels of household dietary diversity, the children of the head of the household grew significantly better (0.2 ± SD) than other children, but they were equally stunted at low levels (Fig. 2). The results were more pronounced when using the 2006 WHO growth reference (15).

View larger version (8K): [in this window] [in a new window]   FIGURE 2  The predicted HAZ of children 9–36 mo of age in Northern Ghana, 2001, by household dietary diversity and the status of the child's father in the household. The HAZ were estimated between the first and 99th percentiles of household dietary diversity, adjusted for marital order of the child's mother, child age and sex, and maternal height (all covariates set to the mean and marital order to first wife).

	Discussion

TOP ABSTRACT Introduction Materials and Methods Results Discussion LITERATURE CITED

The literature on child growth in disadvantaged populations indicates that growth faltering is ultimately produced by the synergism between illness and poor diet (19). In our data, children of high-status members were not more likely to have a growth monitoring and immunization card and did not have less diarrhea than the other children. Healthcare seeking for diarrheal disease also did not differ. These findings thus suggest differences in food allocation, rather than health, as the explanatory pathway. Children of high-status members may be receiving more food and more macro- and micronutrient-dense foods [as has been described for animal source foods (5)]. From our knowledge of these households, we can think of 2 possible pathways. One is that the common meal is distributed preferentially to children of high status parents either in amount, in diversity, or both. Alternatively, the common meal is supplemented with more and better foods at other times by higher status parents. These parents have better access to household food stores or acquisitions, which are proxied by the household diversity score. However, these are hypothetical considerations, because we could not examine the dietary pathways satisfactorily in this data set. Future work on this subject requires the qualitative and quantitative data to understand the pathways.

Interestingly, the regression slope of the association between household dietary diversity and child HAZ in the multi-family households was similar to the association in single-family households. Controlling for the same covariates as in the analysis presented in Table 2, the slope in the single-family households was β = 0.123 (P = 0.036; n = 336), which was similar (P-value for difference, 0.56; analysis not shown) to the size of the association for children of the head of household in the multi-family households (β = 0.153; P = 0.010). Thus, household dietary diversity was associated with the height of the household head's children in a very similar way, irrespective of whether other males with children were present.

A key question raised by our findings is what do lower status members get from remaining in multi-family household arrangements? To stay in these households, it must provide them benefits that they cannot get from alternative living arrangements. Below some minimum level of food allocation, these other household members would find other arrangements. This would explain why there were no households in which child height was lower than the minimum level found for children of the head of household, i.e. the level at which the head of household's children were no better off than the other children (Fig. 2). The need to keep the household together thus dictates a minimum allocation of food. What is striking is that, in this population, the minimum allocation to low-status members is hardly increased even when household food availability is associated with large improvements in the height of the children of high-status parents. In other words, increasing food resources, as proxied by dietary diversity, are not distributed equally.

A limitation of this study is its cross-sectional nature, because household structures change over time. Married males separate from their household to form new households when they can live as well (or better) on their own as they could in the larger household. In this sense, household size is not only an indication of the number of people who need to share limited resources but also of the age of the households and, hence, the possible resource base that has been built up over time. Both variables, household size and age, could have opposite effects on child height. However, restricting the sample to households with at least 2 males with children reduced the possibility of bias due to household age, which is confirmed by the fact that household size was not associated with child stature.

The pattern of differences in height between children of first and second wives in Northern Ghana was similar to that found in the Datogo in Tanzania (13). Before adjusting for wealth (defined as the number of tropical livestock units, a measure of productive assets), Sellen (13) found children of second wives to be 0.5 SD shorter than children of first wives. The interaction between marital status and wealth was not significant. Gibson and Mace (14) found that children of second wives in polygynous marriages in rural Ethiopia had a weight-for-height Z-score that was 0.8 SD lower than children of first wives. This difference was not significant, which might have been due to the small number of women in both groups (75 and 46, respectively). Their model did not include a wealth x status interaction term.

Various possibilities have been offered to explain intrahousehold allocation patterns. It seems unlikely that families in this study perceived the dietary needs of children of the same age as different. The economic efficiency argument used by Pitt et al. (3) is also unlikely to be an underlying determinant, because the children were too young to be engaged in any economic activity. On the other hand, it is possible that those with greater power to allocate resources did expect that the children they were favoring would be more supportive of their needs and bring them economic resources when these children became adults. Those who were most likely to leave and less likely to return social and economic resources, the children of siblings and grandchildren, therefore received less.

Another explanation for our results is that the inequity is due to evolved behavior that fosters the reproduction of one's own genes (20). This behavior has been used to explain the "Cinderella effect," in which parents discriminate against stepchildren (4). In our study, the head may thus preferentially invest in his offspring rather than in the children of other household members. The majority of "other males" consisted of brothers or sons of the head, whose children share 25% of the genes with the head as opposed to 50% for the head's own children.

Our results show that intra-household inequality in child growth is related to status within a household. The analyses were based on the assumption that status either derives from or reflects power to allocate resources. This dictated the specification of the statistical model including the interaction terms between status and resources. Strong quantitative relationships such as those revealed in our analyses require examination of the pathways by which the power over resources results in better growth. Future studies should also explore constructs related to the sharing of resources and to child rearing and the development of the corresponding indicators.

In view of the conventional thinking that resources in the hands of women benefit children more than resources managed by men (21), a crucial question is whether the power of the head of the household operates through the first wife or not. Alternatively, the status of both the father and the mother have additive effects on child growth. Unfortunately, we were not able to examine this question for lack of statistical power.

The implications of our findings need to be taken into account in program planning both in developing and developed countries, not only for nutrition but for other social interventions. Extended households and polygyny are common in many developing countries. The specific household power structure and associated allocation patterns need to be investigated to determine how best to deliver program benefits. Our results are of particular importance considering the AIDS epidemic. As a consequence of the epidemic, an increasing number of children are being raised by nonbiological parents. South Africa, for instance, has a staggering 1.2 million AIDS orphans (22). Programs aimed at vulnerable children and orphans who are raised by 1 parent or in foster families need to evaluate whether food and other resources are allocated differentially to these children. Even though household structure in developed countries is very different, the Cinderella effect suggests that our findings may be relevant there as well. In the US, for instance, 10% of all children live in families with at least 1 nonbiological parent (23). When the intrahousehold power structure is not carefully considered, interventions could lead to increased intrahousehold inequality and may thus have limited or no impact on those household members in greatest need.

	ACKNOWLEDGMENTS

 
We thank Gretel Pelto for excellent comments on earlier drafts of this article and to Joseph Mensah-Homiah and Angelina Mensah for valuable information on household structure in Savelugu-Nanton.

	FOOTNOTES

 
¹ Supported by Howard Hughes Medical Institute through a fellowship to J. L. L., UNICEF Ghana, and the International Food Policy Research Institute (Washington, DC).

² Author disclosures: J. Leroy, A. Razak, and J. Habicht, no conflicts of interest.

Manuscript received 5 May 2008. Initial review completed 12 June 2008. Revision accepted 28 August 2008.

	LITERATURE CITED

TOP ABSTRACT Introduction Materials and Methods Results Discussion LITERATURE CITED

 

1. Haddad L, Pena C, Nishida C, Quisumbing A, Slack A. Food security and nutrition implications of intrahousehold bias: a review of literature. FCND Discussion paper 19. Washington, DC: International Food Policy Research Institute; 1996.

2. Tumilowicz-Torres AT, Pelletier D, Habicht J-P, Pelto G. Timing of growth stunting differs between indigenous Guatemalan boys and girls and maternal education attenuates stunting in girls. FASEB J. 2006;20:A1052.[Free Full Text]

3. Pitt MM, Rosenzweig MR, Hassan N. Productivity, health, and inequality in the intrahousehold distribution of food in low-income countries. Am Econ Rev. 1990;80:1139–56.

4. Daly M, Wilson M. An assessment of some proposed exceptions to the phenomenon of nepotistic discrimination against stepchildren. Ann Zool Fennici. 2001;38:287–96.

5. Gittelsohn J, Vastine AE. Sociocultural and household factors impacting on the selection, allocation and consumption of animal source foods: current knowledge and application. J Nutr. 2003;133:S4036–41.[Abstract/Free Full Text]

6. Bruce J, Lloyd CB. Finding the ties that bind: beyond headship and household. In: Haddad L, Hoddinott J, Alderman H, editors. Intrahousehold resource allocation in developing countries: models, methods, and policy. Baltimore and London: Johns Hopkins University Press for the International Food Policy Research Institute; 1997. p. 213–28.

7. Bhan MK, Bahl R, Bhandari N. Infection: how important are its effects on child nutrition and growth? In: Martorell R, Haschke F, editors. Nutrition and growth. Philadelphia and Vevey (Switzerland): Lippincott Williams & Wilkins; Nestec; 2001. p. 197–221.

8. Leroy JLJP. Childcare in developing societies: the importance of distinguishing between behaviors, resources and agents of disease in health and nutrition [thesis]. Ithaca (NY): Cornell University; 2005.

9. Ghana: Savelugu-Nanton Household Survey Dataset, 2001. Washington (DC): International Food Policy Research Institute. 2005 [cited 2008 Aug 27]. Available from: http://www.ifpri.org/data/ghana02.asp.

10. Lohman TG, Roche AF, Martorell R. Anthropometric standardization reference manual. Champaign (IL): Human Kinetics Books; 1988.

11. Habicht JP. Estandarización de métodos epidemiológicos cuantitativos sobre el terreno [Standardization of quantitative epidemiological methods in the field]. Bol Oficina Sanit Panam. 1974;76:375–84.[Medline]

12. Dibley MJ, Goldsby JB, Staehling NW, Trowbridge FL. Development of normalized curves for the international growth reference: historical and technical considerations. Am J Clin Nutr. 1987;46:736–48.[Abstract/Free Full Text]

13. Sellen DW. Polygyny and child growth in a traditional pastoral society: the case of the Datoga of Tanzania. Hum Nat. 1999;10:329–71.

14. Gibson MA, Mace R. Polygyny, reproductive success and child health in rural Ethiopia: why marry a married man? J Biosoc Sci. 2007;39:287–300.[Medline]

15. WHO Multicentre Growth Reference Study Group. WHO child growth standards. Acta Paediatr Suppl. 2006;95:1–104.

16. Breusch TS, Pagan AR. The Lagrange Multiplier Test and its applications to model specification in econometrics. Rev Econ Stud. 1980;47:239–53.

17. Reed BA, Habicht JP, Niameogo C. The effects of maternal education on child nutritional status depend on socio-environmental conditions. Int J Epidemiol. 1996;25:585–92.[Abstract/Free Full Text]

18. Ruel MT, Levin CE, Armar-Klemesu M, Maxwell D, Morris S. Good care practices can mitigate the negative effects of poverty and low maternal schooling on children's nutritional status: evidence from Accra. World Dev. 1999;27:1993–2009.

19. Lutter CK, Habicht JP, Rivera JA, Martorell R. The relationship between energy intake and diarrhoeal disease in their effects on child growth: biological model, evidence, and implications for public health policy. Food Nutr Bull. 1992;14:36–42.

20. Dawkins R. The selfish gene. Oxford: Oxford University Press; 1976.

21. Quisumbing AR, Brown LR, Feldstein HS, Haddad L, Pena C. Women: the key to food security. Food Policy Report. Washington, DC: The International Food Policy Research Institute; 1995.

22. Parikh A, DeSilva MB, Cakwe M, Quinlan T, Simon JL, Skalicky A, Zhuwau T. Exploring the Cinderella myth: intrahousehold differences in child wellbeing between orphans and non-orphans in Amajuba District, South Africa. AIDS. 2007;21:S95–103.

23. Kreider RM. Living arrangements of children: 2004. Current population reports, P70–114. Washington, DC: U.S. Census Bureau; 2007.

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Leroy, J. L. Articles by Habicht, J.-P. Search for Related Content PubMed PubMed Citation Articles by Leroy, J. L. Articles by Habicht, J.-P.