Cover Image
close this bookCare and Nutrition - Concepts and Measurement (IFPRI, 1997, 56 p.)
View the document(introduction...)
View the documentForeword
View the documentAcknowledgments
View the documentSummary
View the document1. Introduction
View the document2. Developments in Conceptualizing Care
View the document3. Resources for Care
View the document4. Care Practices
View the document5. Conclusions
View the documentBibliography
View the documentInternational Food Policy Research Institute

3. Resources for Care

As research in a particular domain increases, there is a natural progression from an initial definition of relevant constructs5 to the development of valid and appropriate measurement tools and finally to the determination of indicators of risk. For example, with respect to children's nutritional status, researchers now largely agree on the relevant constructs (such as growth), measurement tools (anthropometric measurements), and risk indicators (-2 standard deviation height-for-age on NCHS norms) (UN ACC/SCN 1989). After considerable debate, most investigators accept that these indicators apply across cultural and ecological contexts. A similar process must occur for care. Constructs have to be agreed upon, measurement tools identified and tested, and appropriate indicators of risk and benefit determined. Although there has been considerable progress in industrialized countries in defining constructs and measurement tools relating to cognitive and motor development in children, similar progress has not been made in developing countries, and there has been much less attention given to the effects of these caring practices on children's nutritional status in any setting.

5 A construct is "a concept used in a particular theoretical manner that ties together a number of observations." It cannot be observed directly but, like gravity or evolution, is indirectly inferred from data (Ray 1997, 22).

Cultural variation is likely to be a more important consideration in developing indicators for care practices than for nutritional status. First, cultural differences in caregiving practices and resources are often substantial. In one society, conversation-like interaction with infants is assumed to be essential for their cognitive development, whereas, in others, talking directly to an infant who cannot yet say any words is seen as of no value. For example, one study found that Kenyan Gusii mothers were as likely to hold, touch, or carry a nine-month-old infant as a three-month-old, whereas American mothers were more likely to verbally stimulate a nine-month-old than a three-month-old (Richman, Miller, and Le Vine 1992). Second, care behaviors are likely to be determined by the society's perception of their goals for children. In some societies, obedience, loyalty, and hard work are valued in children, whereas, in others, verbal assertiveness and independence are valued (Nsamenang 1992). Third, even though there may be good agreement on the constructs that are important, such as active feeding behaviors, and even the measurement tools, finding indicators that are not dependent on the context may be difficult. For example, active feeding of young children may be important where food is of poor quality and anorexia in children is common, whereas it may be irrelevant where there is high-quality food and little anorexia. Thus it will probably be necessary to examine the definition of constructs, measurement techniques, and indicators of risk and benefit as three separate processes. Developing measures of care that can be used in different settings is not an easy task. One possible solution may emerge from cross-cultural psychology.

Cross-cultural psychologists suggest that there are three theories about cultural differences: (1) total cultural relativity (every culture has such different functions that one cannot make any judgments across cultures); (2) absolutism (every culture has the same functions, and the same judgments can be made in every culture, thus culture should not be included in analysis); and (3) universalism (Berry et al. 1992). The latter approach suggests that all cultures share similar functions (for example, a greeting) but that the ways that these functions are expressed will differ by culture (wave, verbal, gaze, touching, and so forth). For example, Freedman (1979) asked Native American and Euro-American women to attract the attention of their infants. The Euro-American women tended to vocalize, some almost continuously, and the infants' response was to wave their arms and legs. The Native American women, on the other hand, used looking and gazing to attract their infants' attention, and the infants returned the gaze. Thus the function of getting children's attention is probably universal, but the appropriate way to do so may vary by culture. If one simply observed care-giver verbalization, the functional equivalence of the two practices (vocalization and gaze) would be missed. The universalist approach is generally accepted as the most adequate and is attempted here.

A second principle guiding cross-cultural psychologists is that individual differences within a culture may be as important as between-culture differences. In fact, several cross-cultural studies suggest that within-culture differences in many of these practices are larger than the between-culture differences, and that what appear initially to be unique and exotic differences between cultures are more similar when better understood (Berry et al. 1992). The existence of important differences both within and between cultures needs to be recognized in defining care practices, but the possible functional equivalence of practices should also be evaluated. As indicators and assessment strategies are developed, they must focus on the function that the practice is intended to achieve within the cultural context.

There are many different techniques that can be used to measure the constructs of care or constraints to care. Some of these constructs can be measured fairly easily, such as whether the caregiver washes hands with soap and water prior to food preparation, whereas others are harder to assess and analyze, such as the quality of responsiveness of the caregiver to the child. Similarly, some of the constraints to care have well-defined constructs, such as the nutritional status of the caregiver, whereas others are more difficult to define, such as the caregiver's self-esteem. Many of these care practices and constraints (for example, maternal responsiveness or child attachment) have proved amenable to quantitative analysis in the United States and Canada, but the validity of the measures has not been assessed in other cultural contexts. Given the universalist perspective, one should search for the functional equivalent of that practice in each cultural context. Qualitative methods will be useful for this phase of the process.

A second approach to defining the appropriate constructs, measurement tools, and indicators is to assume that there are a variety of possibly adaptive child-rearing methods and a much smaller range of inappropriate methods. Therefore, the research endeavor becomes one of identifying instances of inappropriate or possibly harmful practices or of debilitating constraints to care. In all cases, it will be essential to evaluate and respect the current cultural adaptations for care.

Various measurement tools have been developed and employed to assess care. These include questionnaires or interviews asking parents or caregivers to describe their care practices and their children's typical behaviors and daily activities; qualitative rating scales with behaviorally defined scale points to assess broadly stated characteristics of care; systematic observations of the caregiver and the child in their natural setting (not in a laboratory); experimental procedures followed by observation of the child's and caregiver's responses and coding of the frequency of particular practices to construct a summary score of different items; and detailed observational coding of ongoing sequences of caregiver-child interaction, aimed at characterizing relevant features of the relationship of the pair.

Some of these strategies for assessing the quality of care in the research studies require extensive observations and technical methodologies. Therefore, they are not feasible for practical use in the field. However, some of these approaches to the measurement of care may be adapted for practical field use in assessing breakdowns in or threats to care. The relevance of these constructs and measures in different cultures should always be of concern to investigators. It is important in all such approaches to develop indicators that have meaning and validity across a variety of cultures. A behavior that may appear to be maladaptive to an outside observer may be the norm within a particular culture. Therefore, it is essential to approach the development of measures with a healthy skepticism and to attempt to use a within-culture yardstick to compare a caregiver with others within her or his own cultural group.

It must be noted that the quality of a construct, measurement tool, or indicator will increase as more investigators and program planners use and refine them. For example, because of extensive work on measuring daily dietary intake, there are useful techniques for measurement, even though the construct of "daily intake" is complex. Similarly, the measurement of cognitive ability has received an enormous amount of attention, and numerous valid measurement tools exist. When more time is given to the assessment and analysis of care practices or behaviors, there is every reason to expect that valid measures can be developed and that indicators will eventually be defined.

The remainder of this chapter is a discussion of suggested constructs and measures of care: both quantitative, relatively well recognized, and objective measures and those that are either of a qualitative nature, more difficult to collect, less well understood, or more specific to the culture. In some cases, the construct is defined, but the measure is not yet available. The six categories of resources for care identified earlier are examined in detail.

Caregiver Education, Knowledge, and Beliefs

Effects of Maternal Education on Child Care

Maternal education is associated with the level of care provided. Three examples of caring practices are considered here: breast-feeding, health care seeking, and interacting with the child. The pathways through which maternal education affects caregiving practices are also discussed.

Feeding Practices. The relationship between maternal education and breast-feeding practices is complex. Education increases both the ability to earn income and the ability to appreciate the importance of caregiving. The former tends to mitigate against breast-feeding, particularly in urban areas, as the caregiver's opportunity cost of time increases. The latter tends to promote caregiving, particularly in supportive workplace environments. For example, in Israel, mothers with the lowest and highest levels of education engage in long-term breast-feeding (Ever-Hadani et al. 1994; Mansbach, Greenbaum, and Sulkes 1991). In developing countries, however, the negative effects on breast-feeding tend to predominate. In Brazil, for example, maternal educational levels have been strongly correlated with earlier termination of breast-feeding (Giugliani et al. 1992). DaVanzo and Starbird (1991) have reported a negative relationship between level of maternal education and breast-feeding duration in Malaysia. In the Philippines, an increase in maternal education by one year is associated with a 36 percent decrease in the probability of exclusive breast-feeding for a six-month-old infant (Cebu Study Team 1991).

Maternal education is associated not only with the quantity of breast-feeding (duration, frequency), but also the quality of feeding. Guldan et al. (1993) in a study conducted in rural Bangladesh found that maternal education is associated with variables that reflect more intensive care for their children (that is, less distraction while feeding, a cleaner feeding environment, and more frequent initiation of child feeding). In the same study, however, more education is also associated with less adequate feeding practices, such as termination of feeding by the mother more often than by the child, a larger number of bottle feeds per day, and fewer breast-feeds per day.

Home Health Practices. Another caring behavior associated with child health and nutrition is the family's home health practices, both preventive health care (immunization, antenatal care for the mother, and so forth) and seeking health care in the event of morbidity. The effects of education on health-care-seeking practices are well documented; it is becoming increasingly evident that maternal education affects a child's health and nutritional status through its effect on the mother's health-care-seeking practices. Better-educated women are more likely to use available health care and community service facilities than women with no education (Joshi 1994; Caldwell 1986; Barrera 1990; Cebu Study Team 1991; Thomas, Strauss, and Henriques 1991).

Child-Caregiver Interactions. Observational studies of mother-child interactions of educated and uneducated women have revealed patterns of behavior that reflect a more committed attitude toward child care among educated women (Le Vine et al. 1991; Richman, Miller, and Le Vine 1992). In Cuernavaca, Mexico, Le Vine et al. (1991) found that mothers who had attended school longer adopted a style of interaction with their infants that was stimulating to infant development, rather than the nurturing style adopted by mothers with fewer years of education. Better-educated women are more vocal with their infants even though their infants are likely to grow up to be more vocal themselves and therefore to require more attention as toddlers (Richman et al. 1988). Richman, Miller, and LeVine (1992) find that better-educated Mexican mothers are more likely to feed their children when they cry. The better-educated mothers are more likely to modify their responsive practices to the age of their infants than less-educated women do. For example, when their children are 10 months old, less-educated women continue to hold them frequently, whereas better-educated women are more likely to interact conversationally at this age.

The pathways through which maternal education affects caregiving practices are (1) the ability to process information, (2) the ability to acquire skills, and (3) the ability to model behavior (Figure 4).

Processing of Information. According to Thomas, Strauss, and Henriques (1990) and Barrera (1990), it is predominantly because an educated woman is more knowledgeable that she is better able to use health care facilities, keep her environment cleaner, and thereby benefit her children. Barrera (1990) proposes that maternal education affects child health by "affecting the productivity of inputs and lowering the costs of information." These sources suggest that one of the likely channels through which maternal education affects child height is by improving the woman's ability to acquire new knowledge and process it appropriately. Information processing, measured by reading newspapers, listening to the radio, watching television, and retaining the information, differs from traditional measures of schooling because it reflects the current capacities of the woman, rather than her history. Thus it may be a better indicator of her current abilities to care for her children. These measures explain almost all of the impact of maternal education on child health, but the exact type of "information" that brought about this effect is not identified (Thomas et al. 1990).

Figure 4 - Pathways of interaction of education with caregiving

Following similar reasoning, Tucker and Sanjur (1988) use "maternal differentiation" rather than maternal education in their analysis of correlates of child nutrition in Panama. Maternal differentiation is a composite variable that incorporates not only years of education, but also current nutrition knowledge, frequency of reading, and a measure of household productivity. Thus, this measure includes evidence of use and retention of information, which they feel is theoretically more coherent than merely using years of schooling. "Maternal differentiation" is positively associated with children's dietary intake and anthropometric status in their study.

The Acquisition of Skills. Skills acquisition, or learning material from school, is one way for women to build a knowledge base to guide behavior and learn patterns of practices that are useful for participating in modern bureaucracies (Joshi 1994; LeVine et al. 1991). With this knowledge, it is hypothesized that these women can make better use of health care services, interact effectively with doctors and nurses (Joshi 1994), and comply better with treatment recommendations (Ware 1984). The primary skill learned in school is literacy. Joshi's (1994) finding that the association between maternal schooling and health care utilization behavior (talking to a doctor) loses significance when controlled for literacy suggests that the effect of schooling on child health may primarily stem from becoming literate, rather than from the more often elaborated notion of becoming an "information processor."

Identity Acquisition. The third suggestion is that women change and take on a different sense of self or identity when they become schooled. The theory of "identity acquisition" proposes that it is not literacy alone that determines the practices of educated women. This hypothesis assumes that schooling leads to behavior change through imitation of people in the "modern sector" (Joshi 1994). Another aspect of this theory is that schooling helps women identify with the role of a teacher as well as that of a student (LeVine et al. 1991). This role change makes them more amenable to new information and also more interactive and stimulating in their child care practices.

Using data from Nepal, Joshi (1994) finds that the caregiving behavior of keeping a child clean is more affected by maternal identity as a schooled person than by the skill of literacy. Maternal identity is assessed using maternal appearance and posture as proxy measures; women who are more erect and assertive are judged to have a "schooled" identity. Making sure that a child is clean when taken to a doctor, the caring behavior, is significantly associated with years of schooling when controlled for literacy but not when controlled for identity. Therefore, Joshi suggests that the association of maternal years of schooling with child cleanliness is mediated through identity, not literacy.

Clearly, the use of education as a measure of care is complex, even though some studies reported here demonstrate that the quality of care given is enhanced by education of the mother. The measures of education used in the literature include literacy status (literate or illiterate), the level of education completed (primary, secondary, and so forth), and the number of years of education, as well as measures of the skills gained from schooling - the ability to read and comprehend written passages, listen and comprehend, and the ability to use decontextualized language. More testing of the various hypotheses of the pathways of interaction of education and child care may be needed before additional cultural-specific behavioral measures of education for care are developed. Identifying practices of educated women and how they differ could also help develop cultural-specific measures for the processes through which education affects care. (The measures developed by Joshi [1994] are good examples of this approach.) For suggested constructs and measures, see Table 1.

Caregiver Knowledge and Beliefs

Available data on cultural beliefs related to caregiving at different stages in the life of an infant are often qualitative, rather than quantitative, in their approach. They can be invaluable, however, from the point of view of assessing the sociocultural causes of and reactions to malnutrition, infant feeding practices that are unique to certain cultures, and beliefs and practices relating to lactation, before embarking on a large-scale survey. The purpose of this section, therefore, is to illustrate the nuances in beliefs among cultures that are gleaned primarily through qualitative and anthropological studies. The care resource measures that are expected to emerge from this discussion are beliefs about breast-feeding, infant feeding, and infant growth.

Table 1 - Education of caregiver


Measurement tools


Years of schooling

Self-report, school records, existing data

Number of years that makes a difference varies by context

Literate or illiterate

Self-report, simple test, or existing data

May be approximated by more than three years of schooling

Skills acquisition

Testing of functional use of language, information processing

May need to be adapted to the cultural setting

Identity acquisition

Observation of teaching role taken by mother with respect to child, professionals

Will depend on the setting

Breast-Feeding Initiation. Cultural beliefs appear to be important in determining both the initiation of breast-feeding and its termination. Both of these practices are closely associated with the growth and development of young infants. In a number of developing societies, breast-feeding is a universal practice, which is initiated soon after birth (Harrison et al. 1993; Cominsky, Mhloyi, and Ewbank 1993; Almedom 199 la, 1991b). In other cultures, particularly in the Indian subcontinent and parts of Southeast Asia, there is a strong belief that colostrum is highly undesirable, and prelacteal feeds of sweetened water, goat's milk, or diluted cow's milk are commonly given in the first two to three days postpartum (Reissland and Burghart 1988; Blanchet 1984; McDonald 1987; McGilvray 1982, cited in Reissland and Burghart 1988).

Cessation of Breast-Feeding and the Timing of Weaning. In a number of studies, the common reasons for cessation of breast-feeding and weaning onto an adult diet are another pregnancy (or the desire for another child), perceived breast-milk insufficiency, certain developmental milestones achieved by the child, or a combination of these reasons (Harrison et al. 1993; Almedom 199 la, 1991b; Cominsky, Mhloyi, and Ewbank 1993; Martines, Ashworth, and Kirkwood 1989; and others). Perceived breast-milk insufficiency is the most commonly reported reason and the perception of insufficiency is often based on the crying of the infant. One study reports that for some, the father made the decision to terminate breast-feeding, possibly because of the belief within that culture that breast-feeding mothers should not have intercourse (Harrison et al. 1993, in Egypt). This study also observed that the quality of breast milk is perceived to change with the age of the child, and this affects the choice of a wet nurse (a mother with an older infant is not allowed to nurse the younger infant of another woman).

In Egypt, the timing of the weaning is important to minimize the risk of exposure to the "evil eye" by ensuring no contact with a menstruating woman or a newly wed woman at the time of complete weaning (Harrison et al. 1993). In Ethiopia, the timing of weaning depends on the season, with the preferable season for weaning being the winter (related to the abundance of barley in the winter (Almedom 1991b).

Maternal Characteristics and Breast-Feeding. In Egypt, breast-feeding is associated with responsibility and maturity in a woman, and very young mothers are not expected to breast-feed (Harrison et al. 1993). The same study reports that the psychological state of the mother is considered important to successful breast-feeding. The breast milk of mothers who are sad or emotionally disturbed is believed to cause diarrhea in their children. In some cultures, the diet of a mother is believed to be important when she is breast-feeding; if the infant falls ill, the mother is given special foods or drinks so that the breast-fed infant will benefit from it (Harrison et al. 1993; Gryboski 1996). The belief that the quality of breast milk is a determinant of the child's health is seen in a number of cultures. Often, the mother is blamed for an infant's illness and may even be treated for it (Reissland and Burghart 1988).

Complementary Feeding. The fear of "spirits" that can disturb the child pervades a number of care-related practices in Indonesia; effort is taken to keep the child calm during the first few months of life, but once the child is able to sit (an important developmental milestone in Indonesia), he or she is considered to be less vulnerable to spirits (Gryboski 1996). The importance of the child's emotional state and contentedness and perception of "emotional maturity" is evident in Gryboski's study in Indonesia, where sibling care-givers are taught to yield to the infant's demands so that the infant is not upset until the child is in late infancy. For instance, infants are fed to induce sleep and calmness in early life, but as the child develops, the child is not pressed to eat if he or she appears upset; "consent feeding" is the norm rather than force feeding as in the preceding months (Launer and Habicht 1989). The child's ability to control feeding is believed to come into play after 7 months, because the child is believed to be helpless from birth until 7 months (Gryboski 1996; Launer and Habicht 1989).

Beliefs regarding complementary feeding have implications for child nutrition, since the age at which children are reported to be most vulnerable to growth-faltering is the period between 6 and 18 months, which is the period of transition between breast milk and an adult diet. The transition period varies by culture; in Bangladesh, Zeitlin and Ahmed (1995) report that the period between 13 and 18 months may be the most crucial.

Beliefs about appropriate time of initiation of complementary feeding also varies across cultures, with the earliest incidence of complementary feeding seen in Indonesia (Kardjati 1996; Launer and Habicht 1989), where rice and mashed bananas are introduced in the first week of life. The belief that supports this practice is that children who are fed a meal will be calmer and sleep more, which will help the mother carry on with her work. In Egypt, Harrison et al. (1993) report that mothers believe that supplementation (after 40 days of full breast-feeding) is necessary to promote growth and "fatness." Mixed feeding is also said to help the mother by reducing the time she needs to spend in breast-feeding the child. The belief that breast-feeding is time-consuming is widespread: in Honduras, mothers believe that exclusive breast-feeding takes longer than supplemental feeding, even though observational studies show that breast-feeding and food preparation and serving take the same amount of time as exclusive breast-feeding for 4- to 6-month-old children (Cohen, Haddix, et al. 1995).

Finally, many food taboos for young children may limit the types of foods that can be offered (Van Esterik 1989). In Iran, the introduction of a variety of foods is often delayed, based on the perception that young children cannot digest the foods that are available to the family (for example, beans) or that some foods cause stammering and delayed speech (eggs) and impair the intellect if introduced before 18 months of age (cheese) (Rabiee and Geissler 1992). This implies that children may not receive adequate amounts of protein- and micronutrient-rich foods until they are 18 months of age.

Characteristics of the Child that Affect Decisions about Feeding. Many beliefs about the termination of breast-feeding depend on characteristics of the child and the child's developmental level. Events related to the initiation and completion of the weaning process include the eruption of teeth (Almedom 1991a, 1991b; Harrison et al. 1993), the onset of walking, and a perception that the child is old enough to consume an adult diet (Harrison et al. 1993). A second factor is the child's appetite as perceived by the mother. Care-givers often see a child's good appetite as an indicator of health (Bentley et al. 1991; Bentley, Black, and Hurtado 1995).

Beliefs about caregiver control of feeding can influence child intake. Dettwyler, in a series of reports from Mali, finds that mothers tend to believe that children should control the amount of food they ingest and that the child's hunger or apparent interest in food should determine the amount of food provided to the child (Dettwyler 1986, 1987). Where levels of anorexia among children are high, this belief can lead to under-nutrition. Engle et al. (1995) find variations in these beliefs among mothers within a single culture; mothers who felt that a child who refuses food should be encouraged to eat more had better-nourished children than those who felt that a child's refusal should not trigger offers of additional food.

Perceptions about the child's state of health have also been reported to influence decisions about breast-feeding, particularly duration of breast-feeding. Adair and Popkin (1996) report that a mother's perception that her infant is small increases the likelihood of her not breast-feeding, even when she intended to do so before the birth of the infant. Conversely, the perception that the infant is doing well increases the likelihood that breast-feeding will continue (Adair, Popkin, and Guilkey 1993). A study in rural Senegal finds that small and thin infants are preferentially fed millet-gruel in addition to breast milk, because their mothers perceive the need to feed them something in addition to breast milk (Simondon and Simondon 1995). Among Peruvian women, Piwoz et al. (1994) report that the strongest indicator of change in feeding practices is low weight in gain infants.

Beliefs about Illness and Malnutrition. Beliefs about illness and practices adopted during illness can have implications for children's health and nutrition. Some information about existing beliefs, in addition to the more easily available information on the use of health care facilities, may prove invaluable in the planning process in these areas. For example, certain illnesses are seen to be a part of the normal development of the child (Gryboski 1996, in Indonesia; Cominsky, Mhloyi, and Ewbank 1993, in Zimbabwe). Some of these illnesses could precipitate malnutrition or aggravate existing situations (for example, some kinds of diarrhea and upper respiratory tract infections). In Pakistan, Mull (1991) reports that mothers seldom associate protein-energy malnutrition (called marasmus) with consumption of too little food; it is often linked to the influence of spirits or a "bad" person on a child. Similar findings are reported in East Africa, India, Nepal, and Tanzania, where protein-energy malnutrition is often seen as a result of adultery, breast-feeding while pregnant, or an evil spirit (Gerlach 1964; Morley, Rohde, and Williams 1983; Tanner 1959; Reissland and Burghart 1988). Scheper-Hughes (1992) describes a process called the "medicalization of hunger" in shantytowns in northeast Brazil; mothers believe that the symptoms of nutrient deficiencies should be treated with medicine, not food. Mull (1991) is especially concerned with the belief that having a marasmic child is a stigma, and, therefore, the illness may not even be reported in many cases. Relying on reported data about morbidity prevalence in settings such as these could well yield underestimated figures. For suggested measures, see Table 2.

Physical Health and Nutritional Status of the Caregiver

Figures on the current nutritional situation of women in the developing world indicate that iron-deficiency anemia is widespread among pregnant and nonpregnant women in developing countries, with the highest rates in South Asia (over 60 percent), using a cutoff of blood levels of hemoglobin less than 12 grams per deciliter. These rates have increased in South Asia and Sub-Saharan Africa over the past decade (UN ACC/SCN 1992). Stunting and low body mass index (BMI) are common in developing countries. Low BMI (less than 18.5), also known as chronic energy deficiency, has been found in 40 percent of women in samples from South and Southeast Asia and in 20 percent in Sub-Saharan Africa (UN ACC/SCN 1992).

Table 2 - Knowledge, attitudes, and beliefs of caregiver


Measurement tools


Beliefs and knowledge about initiation of breast-feeding, colostrum

Surveys, qualitative measures; varies by culture

May vary both between and within cultural groups; may need an individual and a normative measure

Beliefs about termination of breast-feeding

Surveys, qualitative measures; varies by culture

May vary both between and within cultural groups; may depend on developmental milestones, opinion of other family members, appearance of other children

Beliefs about complementary feeding: timing, types, control of intake

Surveys, qualitative measures; varies by culture

Child's physical and emotional state may play a major role in feeding decisions or cultural beliefs about personality or "personness" of the child

Beliefs about maternal characteristics in relation to breast-feeding

Surveys, qualitative measures; varies by culture

Psychological state, maturity, and diet and health of mother may affect beliefs about her ability to breast-feed

Beliefs about relationship between food and malnutrition

Surveys, qualitative measures; varies by culture

Some illnesses are perceived as part of the developmental process (types of diarrhea, upper respiratory tract infections). Caregivers may not associate protein-energy malnutrition with food but believe it to result from spiritual or evil influences, ill-effects of breast-feeding during pregnancy, or adultery; protein-energy malnutrition is often seen as a stigma on the child and is deeply feared.

The linkage between caregiver nutritional status and caregiving has rarely been studied; Winkvist (1995) was one of the first authors to address this issue in detail. Two pathways are possible for linking maternal health to caregiving (Figure 5): (1) a direct link between nutritional status and caregiving capacity and practices through maternal energy levels, and (2) an indirect link whereby the biological consequences of malnutrition for the pregnant and lactating woman could affect the characteristics other child, both physical and behavioral, which could, in turn, affect care-giving practices.

Direct Link

Research on the direct linkage between nutritional status and caregiving is limited. Most findings reported here are from the Nutrition CRSP projects conducted in Kenya (Neumann, Bwibo, and Sigman 1992) and Egypt (Kirksey et al. 1992). Using very small samples, data from the CRSP studies indicate that anemic women in Egypt are less active caregivers than non-anemic women (Rahmanifar et al. 1992). In Egypt, McCullough et al. (1990) found an association between low levels of vitamin B6 in mothers and failure to respond adequately to infant vocalizations, as well as less effective maternal responses in cases of infant distress.

Chronic energy deficiency, measured using BMI, may affect productivity negatively by modifying physical activity patterns (Shetty and James 1994). The nutritional status of the mother is also expected to affect her ability to care for her children. The Food and Agriculture Organization of the United Nations/World Health Organization/United Nations University (FAO/WHO/UNU 1985) Joint Consultation on Energy and Protein Requirements estimated the energy cost of child care activities as 2.2 x basal metabolic rate (BMR), which falls into the category of moderate physical activity. According to Torun et al. (1989), marginally malnourished individuals tend to become more sedentary at the expense of social interactions and discretionary activities. Child care could be classified as "discretionary activity," which is described as "additional activity outside working hours, the energy requirement to cover which should not be regarded as dispensable as it contributes to the physical and intellectual well-being of the individual, household, or group" (FAO/WHO/UNU 1985). Because women with low BMI are less economically productive6 (Ferro-Luzzi et al. 1992), one could expect that they would spend more time in their homes and therefore on child care. There is, however, very little literature assessing the quality of child care performed with low reserves of energy. Energy expenditure studies among men have shown that increased dietary intakes result in more efficient salaried work, less time spent napping, and more physical activity after work (Torun 1989). Data from the Kenyan Nutrition CRSP (McDonald et al. 1994) shows that during a temporary food shortage (a famine), mothers held and cared for their children significantly less than before the shortage. The increased need to procure food resulted in increased child care by siblings and other family members.

6 Shetty and James (1994) cite unpublished FAO data provided in a personal communication by P. J. Françoise in 1990.

Productivity is also influenced by iron status (Yip 1994); supplementation with iron increased women's productivity on tea plantations (Bothwell and Charleton 1981; Edgerton and Gardner 1979; Levin et al. 1990), on farms (Vijayalakshmi, Kupputhai, and Uma-Maheshwari 1987), and in cotton mills in China (Li et al. 1994). Although these patterns suggest that women's health status may affect caregiving, more research must be done before conclusive statements can be made about the effects of nutrient deficiencies on engagement in child care activities or responsive-ness to children.

Indirect Linkages

The effect of maternal nutrition on the pregnancy outcome, particularly birth weight, has been discussed extensively and is not dealt with here (Abrams 1991). However, infant behavior is a major component in establishing an affective relationship between mother and child; therefore, the effects of maternal nutrition on infant behavior will be examined.

Findings from the Egypt Nutrition CRSP (Kirksey et al. 1994) indicate that maternal intakes of energy and protein from animal sources, iron, and zinc are positively associated with neonatal "habituation" behaviors (a measure of early information processing). Rahmanifar et al. (1992) report that maternal diet during lactation, especially lower intakes of animal source foods and certain B vitamins, are associated with infant drowsiness, and infant drowsiness is negatively associated with caregiver vocalization. Similar findings from the Mexico Nutrition CRSP (Alien et al. 1992) indicate that maternal weight and dietary factors are more strongly associated with infant behavioral variables, especially habituation, than are sociocultural factors.

Figure 5 - Possible pathways of interaction of maternal health and caregiving

One must be wary of attributing causality to these relationships for two reasons. First, as Burger, Haas, and Habicht (1993) note, the existence of a statistically significant association between the nutrient deficiency and the behavioral outcome may not imply causality, since socioeconomic status and other factors can affect both behavioral and nutritional outcomes. Most of the data reported here on the relationship between nutrient deficiencies and caregiving behaviors have come from the Nutrition CRSP's, and intervention studies will be needed to demonstrate a causal relationship. Some of the CRSP studies of behavior were conducted using small samples and were limited to biochemical assessments of a few nutrients. The existence of relationships between deficiencies of these specific nutrients and caregiving practices does not preclude the existence of similar relationships with other nutrients.

Second, it is unclear what the direction of the relationship might be. Numerous reports suggest that child behavior changes with poorer nutritional status and that these behaviors may mean that children are less able to elicit caregiving, as discussed earlier. Malnutrition has been associated with apathy, lower energy, delays in verbal development, and delays in motor development, all of which can reduce the child's ability to solicit care.

Other maternal attributes that can be expected to affect caregiving activities are morbidity and reproductive health (for example, the number of children previously born [parity], interpregnancy interval, and reproductive status). The effects of maternal morbidity on caregiving in developing-country situations have not been studied in detail, but illness could exert its effects through the first pathway (direct effects) by influencing maternal nutrient levels and energy reserves. Data from the Kenya CRSP indicate that illness among women, particularly pregnant women, forces them to reallocate a number of tasks, including child care, to other family members (Neumann, Bwibo, and Sigman 1992, 17). The groups requiring the most assistance and task reallocation are found to be, in descending order, pregnant women, adult males, and nonpregnant women. A study on the effects of schistosomiasis infections on women's time allocation patterns shows that physical activity was considerably reduced among infected women, as was activity related to personal care (Parker 1992). The study shows no effects on child care time, but there are no data on the age groups of the children, nor was the definition of "child care" clear.

Women between the ages of 15 and 49 in developing countries spend a significant proportion of their lives in a state of pregnancy or lactation, or both, and the stresses of these periods may lead to a considerable depletion of maternal nutrient levels (Merchant et al. 1989; McGuire and Popkin 1989). Data are not available on the effects of these factors on caregiving per se, but children of mothers who have had many children in The Gambia have poorer early growth, as well as lower rates of catch-up growth in height, than children of mothers with few children (Prentice, Cole, and Whitehead 1987). Miller (1994) finds that the combination of a high birth order and short interval between pregnancies holds the greatest risk of bearing a child with low birth weight. Short intervals between pregnancies are also associated with increased risk of premature births and low birth weight (Mavalankar, Gray, and Trivedi 1992). The case for improving maternal health and nutrition is stronger than ever before, based on both existing evidence of the effects of maternal health on birth outcomes and emerging evidence of their effect on infant behavior and caregiving.

Finally, the extent of violence toward women in their homes is beginning to be recognized. Despite the underreporting of domestic violence, a summary of 35 studies from a variety of countries shows that "one-quarter to more than half of women report having been physically abused by a present or former partner. An even larger percentage have been subjected to ongoing emotional and psychological abuse, a form of violence that many battered women consider worse than physical abuse" (Heise, Pitanguy, and Germain 1994,4).

Violence toward children tends to occur in these same households, and data from the United States suggests that it is even more frequent than violence toward spouses (Finkelhor and Dziuba-Leatherman 1994). An atmosphere of psychological and physical violence may have devastating consequences for children's nutritional status, but few investigators have examined the issue. For suggested measures, see Table 3.

Mental Health, Self-Confidence, and Lack of Stress

In the United States, a large literature links maternal depression with poor caregiving and problematic outcomes for children (Rutter 1990). This issue is summarized in more detail in Engle and Ricciuti (1995). Depression probably also plays a major role in poor caregiving in many developing countries as well. However, despite reports of high levels of anxiety and depression among women in developing countries,7 studies linking these psychological factors with child care have not been done. A report on a slum improvement project in Bangladesh suggests that "the social isolation of women, coupled with the lack of extended family networks in urban areas, is thought to have negative effects on the mental health of women, which, in turn, is likely to reduce the quality of child care, even when the mother is physically present" (UNICEF 1994,9).

7 See, for example, Chakraborty 1990 on Calcutta.

Depression in the United States has been measured by a number of instruments, few of which have been adapted outside of the United States. However, since the most common instruments rely on self-reports (for example, the Beck Depression Inventory) and the questions are fairly straightforward, the possibility of adapting the instrument exists. An anxiety and depression scale was used, for example, in the Egyptian Nutrition CRSP project (Kirksey et al. 1992).

Stress refers to a person's discomfort when exposed to difficult and uncontrollable circumstances, and it is one of the characteristics of high-risk mothers in the United States. Stress may also be linked with poorer caregiving. Although there are many measures of stress, one that might be adaptable to different cultural settings is a symptom checklist, such as the Health Opinion Survey, a report of the woman's physical problems in a recent period (Weisner and Abbott 1977). Items selected reflect psychosomatic difficulties, such as headaches, feeling tired, inability to eat, or sweating palms. Scores on the Health Opinion Survey are associated with sources of stress, such as inadequate family support among Kenyan market women (Weisner and Abbott 1977) and problems with the spouse for Guatemalan periurban mothers (Engle 1989b).

Table 3 - Physical health and nutritional status of caregiver


Measurement tools


Body Mass Index (BMI) (weight/height2)

Requires careful anthropometry

Usually, BMI less than 18.5 is considered an indicator of risk.

Iron deficiency anemia

Various measures, such as hemoglobin, hematocrit

Usually, hemoglobin less than 12 grams per deciliter is considered a risk factor.

Other micronutrients

Biochemical assay

B12 in plasma vitamins, B6 in breast milk (proxy deficiencies for serum levels); not clear that these are key; should evaluate a variety of vitamins.

Reproductive health

Number of children borne, birth spacing, age at first birth, pregnancy complications

Data may be available from the Demographic and Health Survey or census surveys.


Type, frequency, severity of morbidity

Self-report, physician's records

Domestic violence

May require qualitative methods to obtain; low rates likely from surveys

Definitions may or may not include emotional abuse.

A commonly used measure of stress in the United States is the Life Events Change Scale, developed by Dohrenwend and Dohrenwend (1974) to measure the amount of change that an individual has experienced in a recent period. Changes, whether positive or negative, are assumed to increase stress, and they are ranked according to degree of intensity of the change. In the United States, the most dramatic change is death of a spouse, followed by separation from a spouse. Other events include moving to a new area and starting a new form of employment. The degree of intensity of the change can be established for any particular cultural setting, and items can be added or deleted. This measure of stress may be easier to adapt across cultures than the symptom checklist because the life events are by definition specific to a cultural setting.

The level of confidence of the caregiver is often cited as a critical factor for complementary feeding, particularly for anorexic children, but this relationship has not been tested systematically. Program experience suggests that it plays a major role (Gibbons and Griffiths 1984; Griffiths 1988). Even though increased self-confidence is frequently noted as an outcome of a project, systematic measurement of this concept has eluded investigators.

In industrialized countries, the measurement of self-esteem (how one values oneself compared with social norms) has a long history, and numerous measurements have been developed (see, for example, Coopersmith 1981). These measurements are associated with a number of outcomes, such as school performance, job success, and test scores, although relationships are not strong. However, their adaptation to a broader cultural context has been limited.

One of the difficulties in adapting the scale to other cultures is that the basis for self-esteem probably differs by culture (Berry et al. 1992). Another problem is that all of the instruments require the individual to judge whether a number of items are "like me," and this process of judging may be unfamiliar in a less self-conscious society than the United States. Even if overall self-esteem could be measured in another culture, the scale might not measure the kinds of changes in women that are often observed in "empowerment" programs. The apparent increase in confidence often noted may be due to factors other than an increase in overall "self-esteem." These changes may be due to increases in assertiveness, defined as the ability to ask for what one has the right to, which is often very low in traditional societies (Engle 1989b). Changes may also be a function of increases in perceived self-efficacy (Bandura 1984), conceptualized as one's ability to be successful in a particular situation (Engle and Davidson Hunt 1991). The possibilities of developing measures of these more specific abilities are much greater than finding a culture-free assessment of overall global self-esteem or self-rating. For suggested measures, see Table 4.

Caregiver Autonomy and Control of Resources

Autonomy and control of resources refers to the care-giver's ability to play a role in decisions made within the household and the community. A number of studies have addressed this issue. For the Côte d'Ivoire, Haddad and Hoddinott (1994), for example, suggest that mothers are more likely to allocate extra resources under their control to children than are fathers. And the higher the percent of income earned by women, the greater their control over resources (Blumberg 1988; Engle 1991, 1993, in Guatemala). However, working for income does not automatically mean that women control their incomes; in many societies, income is automatically assumed to be the property of the husband (Nsamenang 1992). Women generally enjoy greater autonomy in female-headed households, and some studies show that in spite of lower incomes, children in these living situations do better than might be expected, probably because intrahousehold distribution practices favor children more in female-headed households than in households headed by men (Haddad 1992; Johnson and Rogers 1993; Onyango, Tucker, and Eisemon 1994; Kennedy and Peters 1992; and others).

In many societies, mothers do not have the authority to make decisions regarding the care and feeding of their young children. These decisions may be made by the child's father or, in many cases, by a mother-in-law or older female in the husband's family. In Jordan, Doan and Bisharat (1990) found that the most significant factor associated with child nutritional status was the degree of autonomy of the mother within the household, even controlling for the woman's age, education, and household size. Castle (1995) found that some of the most malnourished children in her Malian sample belonged to low status women in high-income households. She suggests that it may not be the level of household wealth that determines a mother's resources for child health, but rather the mother's access to these resources (Engle, Castle, and Menon 1996). For suggested measures, see Table 5.

Caregiver Workload and Time Availability

Women's time commitments have been recognized as a zero-sum game; no new activities (including new caring practices) can be incorporated into their lives unless other activities are replaced or performed more efficiently in less time (McGuire and Popkin 1990a, 1990b). In addition to activities related to child rearing, women are typically engaged in other time-intensive domestic activities such as water carrying and fuel-wood gathering, and nondomestic production activities, such as agricultural work, informal labor, and formal labor market activities.8 Improved collection of data on time use indicates that women spend more time than men in all work activities, and that in three Asian countries, they spend significantly more time than men in domestic production activities (United Nations 1995; Brown and Haddad 1995).

8 Women's involvement in economic activities in the developing countries varies widely by region, from a high of 56 to 58 percent in eastern and central Asia, to 53 to 54 percent in Sub-Saharan Africa and southeastern Asia, 50 percent in the Caribbean and Oceania, about 30 percent in Latin America and western Asia, and 21 percent in northern Africa. Over the past two decades, men's economic activity rates have declined, whereas women's have increased substantially in all regions except Sub-Saharan Africa and eastern Asia, where they were already high (United Nations 1995). Work rates are higher for rural than for urban women in all parts of the world except for Latin America, where most of the rural work tends to be agricultural and seasonal, but possibly requiring fewer hours per day. Women in Guatemala who reported doing agricultural work were working, on average, only an hour a day (Engle 1989a).

Table 4 - Mental health, self-confidence, and lack of stress of caregiver


Measurement tools



Could adapt existing instruments, such as Beck Depression Inventory

Requires careful adaptation


Could adapt instruments such as Life Event Changes, Symptom Checklist

Requires careful adaptation


Could be adapted from other instruments; includes multiple definitions because it has not yet been clearly defined

Needs further development

Perceived self-efficacy

Adapted locally as a list of tasks relevant to caregiving. Woman is asked to rate her expectation that she is able to perform that task.

Shows promise of being adaptable across cultures

Table 5 - Autonomy and control of resources in the household by caregiver


Measurement tools


Status of woman with respect to others in household

Demographic survey of woman's relationship to head female; qualitative data; self-report

Requires culturally appropriate specification

Household headship (female or male)

Self-report of headship (may depend on economic contribution, age, or kinship patterns; reasons for absence of male partner)

Term generates much confusion; better to use several definitions

Income earner

Self-report (including informal labor); demographic data disaggregated by gender may exist

Income earner does not necessarily control income.

Decisionmaking within household

Caregiver's evaluation of who makes decisions; survey data

Validity of this measure may be low; should interview various family members

Access to resources

Caregiver's perception of access to family income, family land, inheritance laws and customs

May use societal measures as well

The literature on the effects of women's employment on child nutritional status and health outcomes reveals that there is not a simple association between the two (see, for example, Leslie 1989). A few recent studies have found significant negative associations of work for earnings with child nutritional status. In an evaluation of almost 2,000 rural mothers in India, Abbi et al. (1991) found that children of mothers who worked in agricultural labor on their own farms for 5 to 6 hours per day were likely to be significantly malnourished, regardless of who the alternate caregiver was. The women did not have control of their earnings. Rabiee and Geissler (1992) report significantly lower weight-for-age and higher incidence of diarrhea among Iranian children during the time of seasonal agricultural work, despite the relative wealth of the region. In this case, the caregivers were siblings from 8 to 13 years, and assessments of specific practices (for example, disposal of child wastes) suggest that their level of competence was much lower. The most malnourished children were those who were given sedatives in order to keep them quiet while the mother was working. Gryboski (1996) also found negative effects of maternal work in children under a year.

Other studies have found either no negative effects of work (Wandel and Holmboe-Ottesen 1992a, 1992b) or positive effects of work on children (de Groote et al. 1994, in Mali; Brown, Yohannes, and Webb 1994, in Niger; Blau, Guilkey, and Popkin 1996, in the Philippines; Engle 1991, 1993, in Guatemala; LaMontagne, Engle, and Zeitlin 1996, in Nicaragua). In the Philippines, Blau, Guilkey, and Popkin's (1996) analyses with the Cebu data set underline the importance of well-paid work; children whose mothers work in higher-paying occupations have equal or better growth rates, using a rigorous model. When the work was well paid, when the income was in the hands of the mother, or when the child was more than one year old, the effects on either child nutrient intake or nutritional status were positive.

Some studies examine flexibility of work, closeness of work to home, and time-based work as measures of compatibility of work with child care. Compatibility is very important but should be defined in the local context. Informal work may not be as flexible as is often assumed; it may not be compatible with child care if there is much pressure to finish work by a certain time (Doan and Popkin 1993; Joakes 1989).

Clearly, however, very young infants of women from poor households, who are engaged in time-intensive production activities, who have little control over income allocation, and who do not have good alternate caregivers are at risk of low growth.

Important variables for examining the effects of the mother's time availability and workload on children are the coverage and quality of the alternate caregiving system, the age of the child and its characteristics, the woman's control of earned income, the wage rate and flexibility of the work, and the poverty of the household. Some of these variables are included in Table 6, but the remainder, as indicated in the last row, are in other tables. The use of time allocation as a measure of child care, probably the only widely used measure of child care to date, is discussed later. In this list, observed and recalled time are listed separately, as they have differing benefits and costs, an issue that will receive attention farther on.

Social Support Received by the Caregiver

The support provided to the primary caregiver can include explicit child care assistance or information or emotional support provided to the caregiver. One of the most important types of social support is alternate child care. The abilities of the caregiver to provide care may be particularly important for complementary feeding. Engle (1992) distinguishes between levels of care needed at various stages of development of the child. Care by anyone but the mother or a competent adult in the first year of life is associated with higher infant mortality; care needs in the second year of life are still very demanding, although the shortcomings of the caregiver can perhaps be ameliorated by the availability of good quality food and a healthy and safe environment. By the third year of life, many children are capable of some degree of self-care. Leslie's (1988) summary of findings suggests the possibility of negative outcomes for children of mothers who worked during the first year of life but neutral or positive outcomes in later years of life.

The quality of alternate caregiving is rarely investigated. The only dimension of quality that has been examined so far is the age of the alternate. Some studies suggest that care by a preteen caregiver is associated with lower nutritional status of the child under two years, controlling for mother's education and socio-economic status (Engle 1991; LaMontagne, Engle, and Zeitlin 1996). These effects are not unidirectional. Although women's work for earnings normally increases after their children pass through the critical first year, in the Philippines, women in the lowest income groups with more than one preschool child are more likely to work than those with fewer preschool children (Doan and Popkin 1993). Presumably, they have a greater need to work regardless of the availability of alternate caregivers.

Table 6 - Workload and time availability of caregiver


Measurement tools


Observed time spent on work and child care

Observed in sample of time or continuously; define terms as including supervision or not

Observe all caregivers; age of child will be a critical factor

Recalled time spent on work and child care

24-hour recall most common; should interview caregiver

Validity of the measure may be quite limited; can improve measure with good interview techniques

Work characteristics: occupation, wage rate, security

Self-report, census data; often small-scale informal work is not counted without special effort

Important to examine control of income, alternate caregivers, and level of assets of household

Flexibility of employment, compatibility with child care

Work location, time-based (wage) or not, transportation issues, how absenteeism due to child illness is handled

Informal work and piecework at home may not be flexible; depends on need for income, pressures for production

Quality of care during work time (mother or other caregiver)

Surveys of characteristics of alternate caregivers (for example, age, gender) and their availability; observation of quality of care

Also should assess instructions to alternative caregivers, amount of food preparation, and so forth

Table 7 - Social support for caregiver


Measurement tools


Availability of alternate caregivers

Survey of caregiver when mother is working; ratio of children to adults in household (dependency ratio); quality of alternate caregivers (for example, age); survey of whether alternates are kept out of school for this purpose

Need more information on quality of alternate caregiver; dependency ratio should focus on children under 3 years of age

Father's provision of emotional support

Surveys of father's role in decisionmaking; surveys and observations of father's sharing of household and child care tasks; qualitative methods used to identify other providers of emotional or informational support

Observation is difficult; may use unobtrusive observations of father's role; important to interview both men and women

Community support

Assessment of community institutions for child feeding and care programs; qualitative focus groups

Depends on local situation

To date, the availability of institutional care for very young children is extremely limited in developing countries. However, a number of experimental attempts to provide this kind of care are under way, and some have been able to provide care for the youngest children (Leonard and Landers 1992). The quality of these programs depends enormously on the quality of the support provided to the personnel (Young 1995). There are examples of community organizations that have helped increase the amount of food consumed by children when mothers were unable to do so.9

9 One example is the Iringa, Tanzania, program sponsored by UNICEF (1989).

Although there is some evidence that female-headed households provide better care for children, the overall trend toward an increasing number of female-headed households, a higher percentage of women in the labor force, and more older family members to care for raises concerns about the burden placed on the primary caregivers - women (Bruce et al. 1995). Although men should be more involved with child care as women increase their time in the labor force, this change has been slow in coming, with men still providing far less than an equal share of time in household chores and child care (United Nations 1995).

Fathers are particularly important as a source of emotional and informational support (Engle and Breaux 1994). There is some evidence that when they contribute a higher percentage of their incomes to family budgets, children are better nourished (Engle 1993, 1995). Their roles are particularly important where females are traditionally secluded, as in Pakistan (Jahn and Aslam 1995). Their opinions about child care-giving can have significant effects on decisions about infant feeding, particularly breast-feeding (Scrimshaw et al. 1987). One program, the Nutrition Communication Project in Mali, was successful in actively promoting male involvement in nutrition decisions by encouraging men to purchase liver for their pregnant wives (personal communication, E. Piwoz, 1996). Understanding father's attitudes and targeting fathers for education offers promise for nutrition education programs. For suggested measures, see Table 7.