- Email: [email protected]
The influence of maternal intergenerational education on health behaviors of women in peri-urban Bolivia
Social Science & Medicine 50 (2000) 1189±1196
www.elsevier.com/locate/socscimed
The in¯uence of maternal intergenerational education on health behaviors of women in peri-urban Bolivia Deborah E. Bender a,*, Margaret F. McCann b a
Carolina Population Center, Department of Health Policy and Administration, School of Public Health, University of North Carolina, Chapel Hill, NC 27599-7400, USA b Chapel Hill, NC, USA
Abstract The in¯uence of maternal education on infant mortality has been demonstrated repeatedly in health and social science literature. Less explored is the in¯uence of the education level of the mother's mother. In the present paper the authors examine the possible eect of grandmother's education on maternal behaviors. The relationship between intergenerational education and selected health behaviors, including utilization of health services for prenatal care, breast-feeding and family planning, are reported. The data were collected in peri-urban Santa Cruz, Bolivia among mothers of infants between 0 and 18 months of age. It appears that grandmother's education does exert an eect on health behaviors above and beyond the eect of maternal education. This eect is more pronounced for health services which fall clearly in the domain of the formal health care system. While the results are exploratory, the results suggest the worth of further study and consideration of the in¯uence of mothers' mothers in the design of culturally sensitive quality health services. # 2000 Elsevier Science Ltd. All rights reserved. Keywords: Maternal education; Intergenerational in¯uences on health behaviors; Peri-urban Bolivia
Introduction The in¯uence of maternal education on infant mortality has been demonstrated repeatedly in health and social science literature (Caldwell, 1979; Ware, 1984; Cleland and VanGinneken, 1989; Bicego and Boerma, 1991). Higher levels of maternal education have also been found to be positively associated with protective health behaviors, including early and timely use of prenatal care and infant health, in research done by the
* Corresponding author. Tel.: +1-919-966-7383; fax: +1919-966-6961. E-mail addresses: [email protected] (D.E. Bender), [email protected] (M.F. McCann).
authors of this paper among women in peri-urban Bolivia (Bender et al., 1993, 1994; Bender and McCann, 1993). Caldwell has suggested that higher levels of maternal education result in (a) a reduced sense of fatalism in the face of children's ill health; (b) an increased sense of control over distribution of household resources; and (c) an increased ability to take positions dierent than those advocated by mothers-in-law or other authority ®gures in the household (Caldwell, 1979). Caldwell's explanation does not take into account the level of education of other authority ®gures in the household and the resultant quality of their in¯uence. Within the family decision-making model, there has been some exploration of the relationship between husband's education and health outcomes (Martin, 1995).
0277-9536/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 9 9 ) 0 0 3 6 9 - X
1190
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
However, this relationship is generally weak; stronger correlations appear to exist between husband's employment and income level and related health outcomes. Less explored is the in¯uence of the education level of the mother's mother (the person we call `grandmother', for ease of reference, in the remainder of this paper). The in¯uence of the grandmother on economic and demographic strategies, including family planning and child rearing, under conditions of rapid change and drought, has been studied by Ingstad in Botswana (1994). The study collected data on 94 households, using semi-structured interviews, observations and conversational interviews, during a 12-month period in 1984±1985. There, the study found, the maternal grandmother is the individual most often responsible for ensuring that the new mother and baby rest during the postpartum con®nement period. ``This time period'', the author states, ``is obviously an important time for the passing-on of knowledge about breastfeeding and child-care, in general . . . Grandmothers play an important part in ensuring that breast-feeding is given a successful start, thus in¯uencing child nutrition, and also Ð in an indirect way Ð child spacing'' (p. 222). In a theoretical article exploring the eects of maternal education on child mortality, Ware (1984, p. 194) suggests testing the relative power balance between younger and older generations. She suggests that opportunities for the ®rst generation to experience formal education broadly as compared to the previous generation (i.e. their mothers) would be greater than for other generations. This thesis may be particularly applicable to Bolivia, where the 1952 Agrarian Reform gave the majority Indian population legal access to formal education for the ®rst time. Illsley (1986) raised a similar question in a study of intergenerational social mobility in relation to low birth weight. Analyzing data from Scotland, Illsley found low birth weight to be aected by both current occupational class (de®ned by the husband's occupation) and mother's childhood occupational class (de®ned by her father's occupation). Low birth weight
was least common in the highest husband occupational group, intermediate in the intermediate occupational group and most frequent in the lowest occupational group. Furthermore, within each of those three husband occupational groups, the same pattern was found for mother's childhood occupational group. Thus, the highest rate of low birth weight was found for women with both husbands and fathers in the low occupational group. The question of interest in the present paper is whether there is an eect of grandmother's education on maternal behaviors with respect to the use of formal health care services, such as prenatal care and family planning. The authors explore patterns of association between mother's and grandmother's education, called intergenerational education, and selected health behaviors, including utilization of health services for prenatal care, breast-feeding and family planning.
Methods For this analysis, the variable intergenerational education (IGE) was created by combining years of completed education for the child's mother (called mother) and the maternal grandmother's (called grandmother). In delimiting the variable, the following groups were created (Table 1). Low IGE was de®ned as low mother's education (0± 4 yr of schooling) with low grandmother's education (0±4 yr) n 51); Medium IGE was de®ned as (1) medium or high mother's education (5±8 and 9 or more yr) with low grandmother's education n 123); High IGE was de®ned as (1) medium or high maternal education (5 or more yr) with medium or high grandmother's education (5 or more yr) n 110). An anomalous category, low maternal education
Table 1 Levels of mother's and grandmother's education used in creation of variable ``intergenerational education'' Grades of school completed by grandmother Grades of school completed by mother
0±4 (% (n ))
5±8 (% (n ))
9 or more (% (n ))
0±4 5±8 9 or more
29.3 (51)a 37.4 (65)b 33.3 (58)b
10.6 (5) 40.5 (34)c 53.6 (45)c
10.3 (3) 40.5 (8)c 53.6 (23)c
a
Low n 51). Medium n 123). c High n 110).
b
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
and medium or high grandmother's education (a category in which grandmother's education was greater than mother's education) was dropped due to low cell numbers (5 and 3, respectively). The data used in this analysis were collected in 1994 through a systematic household survey conducted in peri-urban Santa Cruz. The focus of the study was on infant feeding patterns and child spacing. In the study, 416 women with infants between 0 and 18 months of age were interviewed (Bender et al., 1992, 1997). Each woman was asked (a) if she had attended school; (b) how many grades she had completed; (c) if her mother attended school; and (d) how many grades her mother had completed. Data are available for 413 of the 416 respondents for maternal education level; about threefourths reported data on the grandmother's education. Although women were not asked about current residence of their mothers, they were asked to name the person who had had the greatest in¯uence on their decision to breastfeed.
1191
those less than 4 months, only 16% were being exclusively breastfed using the WHO breastfeeding indicators (WHO, 1981). Another 14% were being predominantly breastfed (breast milk plus non-milk liquids). Over half of the youngest age infants were receiving breastmilk plus other milks and 9% were receiving some solid foods as well. Among infants who were 6±9 months old, two-thirds were being fed according to WHO's ``timely complimentary feeding'' standard (breast milk plus solids) although one-third of infants in this age group were no longer breastfeeding. One-third of infants between the ages of 12 and 15 months continued to receive breast milk as well as other liquids and solids, meeting WHO's criteria for ``continued breastfeeding rate''. While the majority of women stated that they had made their own decision to breastfeed, about 10% reported that their mother was the person who had most in¯uenced the decision. The percent ranged from 5.7% in the lowest maternal education group to 13.4% among women with 9 or more years of schooling.
Results: description of the study The households of the majority of women appear to be marginal. Less than one-quarter of women in the sample had indoor plumbing (Table 2). About 90% of all households had electrical service, but signi®cantly fewer of the women with low education (0±4 grades) had these connections. More than two-thirds of the women responding were between 20 and 29 years of age and about 18% of women self-reported being single.
Results: breastfeeding practices In the Santa Cruz study, all but 19 of the infants were breastfed initially (96.4%). However, among
Results: intergenerational education In the analysis that considered IGE, women in the low IGE group are somewhat less likely to have had prenatal care than women in either the medium or high IGE group, but these dierences are not signi®cant (Table 3). The medium and high IGE groups report ever-use of prenatal care with similar frequency. These two groups had higher maternal education levels; the dierences in grandmother's education between these two groups thus had no eect. Rather the dierences were between the low level educated mothers and the two groups in which mothers had higher education. Dierences according to IGE group are signi®cant with respect to the number of prenatal care visits
Table 2 Selected socio-demographic characteristics of the study populationa Socio-demographic characteristic
Level of maternal education
Electricity Television Refrigerator Indoor plumbing Age: 20 and 29 yr Single or separated Speak Spanish plus native language at home a
(N 413). (pE0:05,
pE0:02,
pE0:01,
0±4 grades (% (n ))
5±8 grades (% (n ))
9+ grades (% (n ))
Total (% (n ))
16.5 13.6 8.0 3.1 10.9 3.9 3.4
35.1 26.9 17.2 8.5 34.1 7.8 1.5
37.0 32.4 23.5 12.6 24.7 7.8 1.7
88.6 72.9 48.7 24.2 69.7 19.5 6.6
(68) (58) (33) (13) (45) (16) (14)
pE0:001).
(145) (111) (71) (35) (96) (32) (6)
(153) (134) (87) (52) (102) (32) (7)
(366) (303) (191) (100) (243) (80) (27)
1192
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
Table 3 Use of health services during last pregnancy by level of intergenerational education Level of intergenerational education
Use of prenatal care n 284 w2 3:3, p 0:192 NSS Yes No Number of prenatal care visits n 240 w2 25:7, p 0:001 1 2±4 5 or more Birth attended by doctor or nurse n 284 w2 10:3, p 0:006 Yes No
pE0:0001). Those in the low IGE group are seven times more likely to have had only one visit than women in the high IGE group. Conversely, women in the high IGE group were almost twice as likely as women in the two lower groups to have had ®ve or more prenatal care visits. At the time this study was conducted, the Government of Bolivia's suggested number of visits for prenatal care was 5 (Normas, 1992). Particularly noteworthy is the fact that women in the medium IGE category (high maternal education, but low grandmother's education) were much less likely to have had ®ve or more visits than women in the high IGE category when the grandmother also had higher education. These results suggest that, in contrast to the previous variable, grandmothers with more education may be encouraging their daughters to go for more prenatal care. Finally, a physician or nurse attended the majority of the births themselves. However, women in the low IGE were three times more likely than women in the high IGE group to have been attended by someone other than a physician or nurse, usually a traditional birth attendant or relative pE0:006). The two groups
Low (%)
Medium (%)
High (%)
Total (%)
74.5 25.5
84.6 15.5
85.5 14.6
83.1 16.9
22.5 37.5 40.0
10.6 50.0 39.4
3.1 29.2 67.7
9.6 39.6 50.8
74.5 25.5
89.4 10.6
91.8 8.2
87.7 12.3
of higher educated mothers were similar to each other. Thus, as with prenatal care, maternal education appears to be an important factor but grandmother's education does not. Six variables related to breast-feeding practice were examined. For several of these variables, practices were almost universal and, thus, there was little dierence among the IGE groups. Virtually all women in the study breastfed, and most began breastfeeding within the ®rst 24 hours after birth (but seldom within the ®rst hour) (not tabled). Most infants were given mate or tea as the ®rst liquid after birth. Furthermore, about one-third of women in all three groups reported that they did not always have sucient breast milk (see below). However, women were split in rough thirds in responding to the question ``Who taught you to breast-feed''? Although the relationship does not reach statistical signi®cance, the trend is interesting. That is, women in the low IGE group were more likely to say no one taught them while those in the high IGE group are more likely to say that grandmother (i.e., their mother) taught them (Table 4). Comparing the two
Table 4 Breastfeeding practices by levels of intergenerational education Level of intergenerational education
Mother reports sucient breastmilk n 281 w2 1:8, p 0:40 NSS Yes No Who taught you to breastfeed n 260 w2 8:5, p 0:08 Mother's mother (i.e. grandmother) Doctor/nurse sta No one taught me
Low (%)
Medium (%)
High (%)
Total (%)
66.7 33.3
72.1 27.9
63.9 36.1
68.0 32.0
26.7 33.3 40.0
20.4 44.3 35.4
35.3 40.2 24.5
27.3 40.8 31.9
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
1193
Table 5 Knowledge of protective eect of breastfeeding on child spacing by level of intergenerational education Level of intergenerational education
Have heard of protective eects of breastfeeding n 282 w2 3:9, p 0:21 Yes No Duration of breastfeeding protection n 174 w2 6:1, p 0:05 All the time Part of the time
groups of higher-educated mothers, the group with the higher-educated grandmother was much more likely to say that their mother (i.e., the grandmother) taught them, whereas they were less likely to say ``no one'' had taught them. When outcomes related to knowledge of the protective eect of breast-feeding on child spacing were examined, two-thirds of women in all IGE groups stated that they had heard of the protective eects of breastfeeding (Table 5). However, women in the low and medium IGE groups were more than twice as likely as their high IGE counterparts to have stated, incorrectly, that protection against pregnancy through breast-feeding endured for the entire breast-feeding period. Women in the medium IGE group (high
Low (%)
Medium (%)
High (%)
Total (%)
60.8 39.2
57.0 43.0
68.2 31.8
62.1 37.9
33.3 66.7
31.5 68.5
15.5 84.5
25.3 74.7
mother's education/low grandmother's education) were more similar to those in low IGE group (low mother's education/low grandmother's education) than to those in the high IGE group (high mother's education/high grandmother's education). This again suggests a potential in¯uence of the more educated grandmothers in the high IGE group on the quality of information held by their daughters. With respect to family planning, women's responses to several of the questions did not vary signi®cantly by IGE strata (Table 6) but there were some interesting suggestive trends. Both groups of mothers with higher education (medium and high IGE) were somewhat more likely to be using contraceptives than the mothers with low education (low IGE). Among contra-
Table 6 Family planning knowledge and use by level of intergenerational education Level of intergenerational education
Current contraceptive use n 277 w2 1:3, p 0:51 NSS Yes No Method currently in use n 116 w2 4:1, p 0:40 NSS IUD Other modern Traditional Desire for additional information n 284 w2 8:4, p 0:08 NSS Yes Maybe No Main reason for non-use of contraceptive n 153 w2 10:5, p 0:10 NSS Lack of knowledge, fear of disapproval Breastfeeding Absence of husband Other reasons Have you talked with other women about contraceptive use n 277 w2 9:6, p 0:008 Never 1±2 times
Low (%)
Medium (%)
High (%)
Total (%)
36.0 64.0
44.6 55.4
45.3 54.7
43.3 56.7
5.3 66.7 27.8
13.5 61.5 25.0
23.9 50.0 26.1
16.4 57.8 25.9
86.3 0.0 13.7
94.3 1.6 4.1
94.6 1.8 3.6
93.0 1.4 5.6
37.5 31.3 9.4 21.9
33.3 27.0 19.1 20.6
25.9 44.8 22.4 6.9
31.4 34.6 18.3 15.7
64.0 36.0
51.7 48.3
38.5 61.5
48.7 51.3
1194
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
ceptive users, IUD use tended to increase from low to medium to high IGE groups, while use of other modern methods (primarily oral contraceptives) decreased. The percentages of contraceptive users who were using traditional methods were similar for all three IGE groups. In explaining their reasons for not using contraception, the percentage stating ``lack of knowledge'' or ``fear of disapproval'' declined steadily from low to medium to high IGE. Women in the high IGE group were also more likely to have reported that breast-feeding was the main reason for their non-use of a contraceptive than the two other groups of women. Interestingly, and supportive of Caldwell's explanation of a gradient of increasing control over one's world as maternal education increases, is the ®nding that the likelihood of having talked with other women about contraceptive use increased steadily, and signi®cantly, pE0:0008 from low to medium to high IGE. Furthermore, within the two higher maternal education groups, those with more educated grandmothers (high IGE) were more likely to have talked with others than counterparts in the group in which the grandmothers had less education (medium IGE). Continuing the exploration of the combined in¯uence of grandmother's and mother's education on selected health outcomes, the authors conducted both two-way chi-square and regression analyses. When dierences between low and medium IGE and between medium and high education were evaluated, with respect to their relationship to number of prenatal care visits (Table 3), dierences between medium and high IGE were found to be signi®cant pE0:001). Likewise, dierences between medium and high IGE were signi®cant in the case of breastfeeding instruction pE0:05 (Table 4); knowledge of the duration of breastfeeding protection pE0:05 (Table 5); and having ever talked with other women about contraceptive use pE0:05 (Table 6). Only in the case of birth attendant were dierences between low and medium IGE signi®cant pE0:001 (Table 3). Thus, for four of the ®ve analyses
with signi®cant dierences, the results indicate variation by grandmother's educational attainment (among mothers with the same educational levels). Only one of the comparisons indicates variation by mother's education (among grandmothers with the same education). In the regression analyses, maternal education and grandmother's education were entered into each model as independent variables. The two variables were highly correlated pE0:0001), according to Pearson correlation coecient. Four of the outcomes that were signi®cant in the univariate analysis were entered into separate models as dependent variables. The variables were: . number of visits for prenatal care; . birth attended by doctor or nurse; . having talked with other women about contraceptive use; and, . knowledge of the duration of pregnancy protection while breastfeeding. Dierences by grandmother's education were signi®cant with respect to number of visits sought for prenatal care pE0:002). That is, there is an eect of grandmother's education on frequency of prenatal care, which is in addition to the in¯uence of maternal education alone. The choice of birth attendant in this sample is in¯uenced more by maternal education (low versus medium or high) than by grandmother's education. Dierences are signi®cant. In the model predicting that women will have talked with other women about contraceptive use, grandmother's education was also signi®cant pE0:001). Access to family planning information in the Bolivian context is recent, and it is likely that women and their mothers are interested in acquiring additional information. Finally, there were no signi®cant dierences by grandmother's education in the model that considered knowledge of duration of pregnancy protection while breastfeeding, suggesting that the level of grandmother's education is less important in in¯uencing knowledge or practices that traditionally are familybased.
Table 7 Alternate categorization of levels of mother's and grandmother's education with mixed educational levels Grades of school completed by grandmother Grades of school completed by mother
0±4 (% (n ))
5±8 (% (n ))
9 or more (% (n ))
0±4 5±8 9 or more
29.3 (51)a 37.4 (65)a 33.3 (58)b
6.0 (5) 40.5 (34)b 53.6 (45)c
6.0 (3) 40.5 (8)c 53.6 (23)c
a
L/M(m)+L(g) n 116). M/H(m)+L/M(g) n 92). c H(m)+M/H(g) n 76).
b
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
1195
Table 8 Maternal and grandmother's education using dierent levels of `high' education for mothers and grandmothers Grades of school completed by grandmother Grades of school completed by mother
0±4 (% (n ))
5±8 (% (n ))
9 or more (% (n ))
0±4 5±8 9 or more
29.3 (51)a 37.4 (65)a 33.3 (58)c
6.0 (5)b 40.5 (34)b 53.6 (45)d
6.0 (3)b 40.5 (8)b 53.6 (23)d
a
L(m)+L(g) n 116). H(m)+L(g) n 50). c L/M(m)+M/H(g) n 58). d H(m)+H(g) n 68). b
This analysis is clearly exploratory. We have considered various ways of analyzing the data and reported those ®ndings. The ®ndings are consistent across methods. One alternative considered included some diagonal groupings (Table 7). Results for low, medium and high groupings were similar to these presented here. Another alternative would be to use dierent groupings of grades for mother and for grandmother, such as depicted in Table 8. The advantage of the latter is that it allows dierent de®nitions of `high' educational level to be used for mothers than for grandmothers. This application may be particularly revealing given that education was prohibited to people of Indian descent, the majority of the Bolivian population, until after the Agrarian Revolution in 1952. Even for women born in the years immediately following 1952, many may not have had access to schooling beyond some primary grades. This categorization produces four, rather than three, groupings, thus permitting comparisons of the eects of grandmother's education within both high and low maternal education groups (as well as comparisons of the eects of mother's education within both grandmother's education groups). It is acknowledged that the dierences among education groups may be attributable to factors other than education, per se. Some of these other factors were measured, and thus could be included in a multivariate analysis, while others were not measured (and may indeed be unmeasurable). Rather than suggesting that education itself is the reason for dierences among education groups, we propose that education is a marker for a constellation of factors. Our results indicate that grandmothers of varying educational levels may have varying eects on maternal behavior, which is important regardless of whether education itself is the reason for these dierences. Similarly, education can be used as a marker to assist in determining educational messages, both for mother and grandmother.
Discussion and recommendations A relationship between intergenerational education, beyond that of maternal education alone and selected formal health care system behaviors appears to exist. The strength of the relationship seems to vary according to the degree or amount of eort involved in the behavior. That is, going for ®ve prenatal care visits requires a more sustained eort than going for any prenatal care. Behaviors typically associated with the informal care sector or the family unit are less aected by IGE than are behaviors related to the formal health care system. This is shown in this example by breastfeeding and breastfeeding related knowledge. In Bolivia, population-wide access to family planning information and devices is recent. Because the reproductive health campaign has used multiple channels of communication, including formal health services, radio, television, billboards and outreach campaigns, it is dicult to know whether to classify modern contraceptive practice in the domain of the formal or informal health care system. In summary, it is apparent that grandmother's education exerts an eect on health behaviors above and beyond the eect of maternal education. This eect is more pronounced for health services which fall clearly in the domain of the formal health care system. Results from the few available studies suggest that a woman continues to rely on her mother, that is, the grandmother, for advice related to important decisions. Use of health services is clearly among these. Further emphasizing the potential strength of the mothergrandmother relationship is the fact that one-tenth of the respondents reported not having a spouse or that their spouse lived in another part of the country (9.4 and 9.8%, respectively). The question of interest in the present paper goes beyond the theoretical. In an article titled, ``Data Needs for the Future'', Vaessen (1991) lists ``edu-
1196
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
cational level of family members'' as one among several program topics for which additional data are needed. To continue this exploration, data which addresses grandmother's as well as spouse's school attendance ought to be routinely collected. In addition, information on grandmother's recent and past use of health services, grandmother's residence in relation to mother's residence, and explicit questioning of patterns of advice-seeking behavior would be useful additions in further explorations of the strength of the maternalgrandmother in¯uence on health behaviors and health outcomes. Following Illsley (1986), data on occupation and/or education of mother's father, as well as that of the husband, are necessary to explore the dierential eects of various family members on selected health outcomes. This additional information regarding the in¯uence of grandmothers on maternal health decision-making has potential for application in the design and implementation of quality health care services. Programs focused on prenatal care, safe delivery and well-child care, including extended breastfeeding and timely weaning, are examples where the inclusion of the grandmother in a health care visit has the potential to improve health outcomes by taking advantage of the grandmother's experience and the strength of her in¯uence. The usefulness of this team approach is likely to be highest in areas where there are high rates of female-headed households, either due to unmarried childbearing or prolonged absence of husbands.
Acknowledgements The authors want to thank collaborators at Nur University in Santa Cruz, Bolivia, the Earthwatch volunteers and the Bolivian students for their assistance during the data collection phase of the study. The authors also want to acknowledge the ®nancial support received from Earthwatch for the current study conducted in 1994. This study replicated research done in Cochabamba, Bolivia in 1991. The Cochabamba study was conducted with support from the Special Programme of Research, Development and Research Training in Human Reproduction of the World Health Organization.
References Bender, D.E., Baker, R., Dusch, E., McCann, M.F., 1997.
Integrated use of qualitative and quantitative methods to elicit women's dierential knowledge of breastfeeding and lactational amenorrhea in peri-urban Bolivia. Journal of Health and Population in Developing Countries. 1 (1), 68± 84. Bender D.E., Baker,R., McCann, M.F., Body, P., Dusch, E., 1995. Infant feeding practices and child spacing in lowland Bolivia. Replication of WHO's 1992 study; report submitted for publication to EarthWatch and the Center for Field Research, May. Bender D.E., Santander, A., Morales, L., Salinas,W., 1994. Proyecto Madre: a community-based reproductive health intervention to improve maternal knowledge, decisionmaking and service utilization in Peri-urban Bolivia. A ®nal report. UNC-SPH, Chapel Hill, NC. Bender, D.E., McCann, M.F., 1993. Validation of self-report of literacy among peri-urban women in Bolivia. International Journal of Epidemiology. 22 (4), 747±748. Bender, D.E., Rivera, T., Madonna, D., 1993. Rural origin as a risk factor for maternal and child health in peri-urban Bolivia. Social Science & Medicine. 37 (11), 1345±1349. Bicego, G.T., Boerma, J.T., 1991. Maternal education and child survival: a comparative analysis of DHS data. In: Demographic and Health Surveys World Conference Proceedings, vol. 1. IRD/Macro International, Columbia, MD, pp. 177±204. Caldwell, J., 1979. Education as a factor in mortality decline. Population Studies 33, 395±413. Cleland, J., VanGinneken, J., 1989. Maternal schooling and childhood mortality. Journal of Biosocial Science 10 (Suppl.), 123±124. Illsley, R., 1986. Occupational class, selection and the reproduction of inequalities in health. Quarterly Journal of Social Aairs 2 (3), 151±165. Ingstad, B., 1994. The grandmother and household viability in Botswana. In: Adepoju, A., Oppong, C. (Eds.), Work and Population in Sub-Saharan Africa. Heinemann, Portsmouth, NH (Published on behalf of the International Labour Oce). Martin, T.C., 1995. Women's education and fertility: results from 26 demographic and health surveys. Studies in Family Planning. 26 (4), 187±202. Normas nacionales para atencioÂn integral al ninÄo, al escolar, al adolescente y a la mujer, 1992. Manual de Normas y Procedimientos. Ministerio de PrevisioÂn Social y Salud Publica, La Paz. Vaessen, M., 1991. Data Needs for the Future. In: Demographic and Health Surveys World Conference Proceedings, vol. 3. IRD/Macro International, Columbia, MD, p. 2179. Ware, H., 1984. Eects of maternal education, women's roles and child care on child mortality. Population and Development Review 10 (Suppl.), 191±214. World Health Organization (WHO), 1981. Contemporary Patterns of Breastfeeding: Report on the WHO Collaborative Study on Breastfeeding. WHO, Geneva.
www.elsevier.com/locate/socscimed
The in¯uence of maternal intergenerational education on health behaviors of women in peri-urban Bolivia Deborah E. Bender a,*, Margaret F. McCann b a
Carolina Population Center, Department of Health Policy and Administration, School of Public Health, University of North Carolina, Chapel Hill, NC 27599-7400, USA b Chapel Hill, NC, USA
Abstract The in¯uence of maternal education on infant mortality has been demonstrated repeatedly in health and social science literature. Less explored is the in¯uence of the education level of the mother's mother. In the present paper the authors examine the possible eect of grandmother's education on maternal behaviors. The relationship between intergenerational education and selected health behaviors, including utilization of health services for prenatal care, breast-feeding and family planning, are reported. The data were collected in peri-urban Santa Cruz, Bolivia among mothers of infants between 0 and 18 months of age. It appears that grandmother's education does exert an eect on health behaviors above and beyond the eect of maternal education. This eect is more pronounced for health services which fall clearly in the domain of the formal health care system. While the results are exploratory, the results suggest the worth of further study and consideration of the in¯uence of mothers' mothers in the design of culturally sensitive quality health services. # 2000 Elsevier Science Ltd. All rights reserved. Keywords: Maternal education; Intergenerational in¯uences on health behaviors; Peri-urban Bolivia
Introduction The in¯uence of maternal education on infant mortality has been demonstrated repeatedly in health and social science literature (Caldwell, 1979; Ware, 1984; Cleland and VanGinneken, 1989; Bicego and Boerma, 1991). Higher levels of maternal education have also been found to be positively associated with protective health behaviors, including early and timely use of prenatal care and infant health, in research done by the
* Corresponding author. Tel.: +1-919-966-7383; fax: +1919-966-6961. E-mail addresses: [email protected] (D.E. Bender), [email protected] (M.F. McCann).
authors of this paper among women in peri-urban Bolivia (Bender et al., 1993, 1994; Bender and McCann, 1993). Caldwell has suggested that higher levels of maternal education result in (a) a reduced sense of fatalism in the face of children's ill health; (b) an increased sense of control over distribution of household resources; and (c) an increased ability to take positions dierent than those advocated by mothers-in-law or other authority ®gures in the household (Caldwell, 1979). Caldwell's explanation does not take into account the level of education of other authority ®gures in the household and the resultant quality of their in¯uence. Within the family decision-making model, there has been some exploration of the relationship between husband's education and health outcomes (Martin, 1995).
0277-9536/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 9 9 ) 0 0 3 6 9 - X
1190
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
However, this relationship is generally weak; stronger correlations appear to exist between husband's employment and income level and related health outcomes. Less explored is the in¯uence of the education level of the mother's mother (the person we call `grandmother', for ease of reference, in the remainder of this paper). The in¯uence of the grandmother on economic and demographic strategies, including family planning and child rearing, under conditions of rapid change and drought, has been studied by Ingstad in Botswana (1994). The study collected data on 94 households, using semi-structured interviews, observations and conversational interviews, during a 12-month period in 1984±1985. There, the study found, the maternal grandmother is the individual most often responsible for ensuring that the new mother and baby rest during the postpartum con®nement period. ``This time period'', the author states, ``is obviously an important time for the passing-on of knowledge about breastfeeding and child-care, in general . . . Grandmothers play an important part in ensuring that breast-feeding is given a successful start, thus in¯uencing child nutrition, and also Ð in an indirect way Ð child spacing'' (p. 222). In a theoretical article exploring the eects of maternal education on child mortality, Ware (1984, p. 194) suggests testing the relative power balance between younger and older generations. She suggests that opportunities for the ®rst generation to experience formal education broadly as compared to the previous generation (i.e. their mothers) would be greater than for other generations. This thesis may be particularly applicable to Bolivia, where the 1952 Agrarian Reform gave the majority Indian population legal access to formal education for the ®rst time. Illsley (1986) raised a similar question in a study of intergenerational social mobility in relation to low birth weight. Analyzing data from Scotland, Illsley found low birth weight to be aected by both current occupational class (de®ned by the husband's occupation) and mother's childhood occupational class (de®ned by her father's occupation). Low birth weight
was least common in the highest husband occupational group, intermediate in the intermediate occupational group and most frequent in the lowest occupational group. Furthermore, within each of those three husband occupational groups, the same pattern was found for mother's childhood occupational group. Thus, the highest rate of low birth weight was found for women with both husbands and fathers in the low occupational group. The question of interest in the present paper is whether there is an eect of grandmother's education on maternal behaviors with respect to the use of formal health care services, such as prenatal care and family planning. The authors explore patterns of association between mother's and grandmother's education, called intergenerational education, and selected health behaviors, including utilization of health services for prenatal care, breast-feeding and family planning.
Methods For this analysis, the variable intergenerational education (IGE) was created by combining years of completed education for the child's mother (called mother) and the maternal grandmother's (called grandmother). In delimiting the variable, the following groups were created (Table 1). Low IGE was de®ned as low mother's education (0± 4 yr of schooling) with low grandmother's education (0±4 yr) n 51); Medium IGE was de®ned as (1) medium or high mother's education (5±8 and 9 or more yr) with low grandmother's education n 123); High IGE was de®ned as (1) medium or high maternal education (5 or more yr) with medium or high grandmother's education (5 or more yr) n 110). An anomalous category, low maternal education
Table 1 Levels of mother's and grandmother's education used in creation of variable ``intergenerational education'' Grades of school completed by grandmother Grades of school completed by mother
0±4 (% (n ))
5±8 (% (n ))
9 or more (% (n ))
0±4 5±8 9 or more
29.3 (51)a 37.4 (65)b 33.3 (58)b
10.6 (5) 40.5 (34)c 53.6 (45)c
10.3 (3) 40.5 (8)c 53.6 (23)c
a
Low n 51). Medium n 123). c High n 110).
b
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
and medium or high grandmother's education (a category in which grandmother's education was greater than mother's education) was dropped due to low cell numbers (5 and 3, respectively). The data used in this analysis were collected in 1994 through a systematic household survey conducted in peri-urban Santa Cruz. The focus of the study was on infant feeding patterns and child spacing. In the study, 416 women with infants between 0 and 18 months of age were interviewed (Bender et al., 1992, 1997). Each woman was asked (a) if she had attended school; (b) how many grades she had completed; (c) if her mother attended school; and (d) how many grades her mother had completed. Data are available for 413 of the 416 respondents for maternal education level; about threefourths reported data on the grandmother's education. Although women were not asked about current residence of their mothers, they were asked to name the person who had had the greatest in¯uence on their decision to breastfeed.
1191
those less than 4 months, only 16% were being exclusively breastfed using the WHO breastfeeding indicators (WHO, 1981). Another 14% were being predominantly breastfed (breast milk plus non-milk liquids). Over half of the youngest age infants were receiving breastmilk plus other milks and 9% were receiving some solid foods as well. Among infants who were 6±9 months old, two-thirds were being fed according to WHO's ``timely complimentary feeding'' standard (breast milk plus solids) although one-third of infants in this age group were no longer breastfeeding. One-third of infants between the ages of 12 and 15 months continued to receive breast milk as well as other liquids and solids, meeting WHO's criteria for ``continued breastfeeding rate''. While the majority of women stated that they had made their own decision to breastfeed, about 10% reported that their mother was the person who had most in¯uenced the decision. The percent ranged from 5.7% in the lowest maternal education group to 13.4% among women with 9 or more years of schooling.
Results: description of the study The households of the majority of women appear to be marginal. Less than one-quarter of women in the sample had indoor plumbing (Table 2). About 90% of all households had electrical service, but signi®cantly fewer of the women with low education (0±4 grades) had these connections. More than two-thirds of the women responding were between 20 and 29 years of age and about 18% of women self-reported being single.
Results: breastfeeding practices In the Santa Cruz study, all but 19 of the infants were breastfed initially (96.4%). However, among
Results: intergenerational education In the analysis that considered IGE, women in the low IGE group are somewhat less likely to have had prenatal care than women in either the medium or high IGE group, but these dierences are not signi®cant (Table 3). The medium and high IGE groups report ever-use of prenatal care with similar frequency. These two groups had higher maternal education levels; the dierences in grandmother's education between these two groups thus had no eect. Rather the dierences were between the low level educated mothers and the two groups in which mothers had higher education. Dierences according to IGE group are signi®cant with respect to the number of prenatal care visits
Table 2 Selected socio-demographic characteristics of the study populationa Socio-demographic characteristic
Level of maternal education
Electricity Television Refrigerator Indoor plumbing Age: 20 and 29 yr Single or separated Speak Spanish plus native language at home a
(N 413). (pE0:05,
pE0:02,
pE0:01,
0±4 grades (% (n ))
5±8 grades (% (n ))
9+ grades (% (n ))
Total (% (n ))
16.5 13.6 8.0 3.1 10.9 3.9 3.4
35.1 26.9 17.2 8.5 34.1 7.8 1.5
37.0 32.4 23.5 12.6 24.7 7.8 1.7
88.6 72.9 48.7 24.2 69.7 19.5 6.6
(68) (58) (33) (13) (45) (16) (14)
pE0:001).
(145) (111) (71) (35) (96) (32) (6)
(153) (134) (87) (52) (102) (32) (7)
(366) (303) (191) (100) (243) (80) (27)
1192
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
Table 3 Use of health services during last pregnancy by level of intergenerational education Level of intergenerational education
Use of prenatal care n 284 w2 3:3, p 0:192 NSS Yes No Number of prenatal care visits n 240 w2 25:7, p 0:001 1 2±4 5 or more Birth attended by doctor or nurse n 284 w2 10:3, p 0:006 Yes No
pE0:0001). Those in the low IGE group are seven times more likely to have had only one visit than women in the high IGE group. Conversely, women in the high IGE group were almost twice as likely as women in the two lower groups to have had ®ve or more prenatal care visits. At the time this study was conducted, the Government of Bolivia's suggested number of visits for prenatal care was 5 (Normas, 1992). Particularly noteworthy is the fact that women in the medium IGE category (high maternal education, but low grandmother's education) were much less likely to have had ®ve or more visits than women in the high IGE category when the grandmother also had higher education. These results suggest that, in contrast to the previous variable, grandmothers with more education may be encouraging their daughters to go for more prenatal care. Finally, a physician or nurse attended the majority of the births themselves. However, women in the low IGE were three times more likely than women in the high IGE group to have been attended by someone other than a physician or nurse, usually a traditional birth attendant or relative pE0:006). The two groups
Low (%)
Medium (%)
High (%)
Total (%)
74.5 25.5
84.6 15.5
85.5 14.6
83.1 16.9
22.5 37.5 40.0
10.6 50.0 39.4
3.1 29.2 67.7
9.6 39.6 50.8
74.5 25.5
89.4 10.6
91.8 8.2
87.7 12.3
of higher educated mothers were similar to each other. Thus, as with prenatal care, maternal education appears to be an important factor but grandmother's education does not. Six variables related to breast-feeding practice were examined. For several of these variables, practices were almost universal and, thus, there was little dierence among the IGE groups. Virtually all women in the study breastfed, and most began breastfeeding within the ®rst 24 hours after birth (but seldom within the ®rst hour) (not tabled). Most infants were given mate or tea as the ®rst liquid after birth. Furthermore, about one-third of women in all three groups reported that they did not always have sucient breast milk (see below). However, women were split in rough thirds in responding to the question ``Who taught you to breast-feed''? Although the relationship does not reach statistical signi®cance, the trend is interesting. That is, women in the low IGE group were more likely to say no one taught them while those in the high IGE group are more likely to say that grandmother (i.e., their mother) taught them (Table 4). Comparing the two
Table 4 Breastfeeding practices by levels of intergenerational education Level of intergenerational education
Mother reports sucient breastmilk n 281 w2 1:8, p 0:40 NSS Yes No Who taught you to breastfeed n 260 w2 8:5, p 0:08 Mother's mother (i.e. grandmother) Doctor/nurse sta No one taught me
Low (%)
Medium (%)
High (%)
Total (%)
66.7 33.3
72.1 27.9
63.9 36.1
68.0 32.0
26.7 33.3 40.0
20.4 44.3 35.4
35.3 40.2 24.5
27.3 40.8 31.9
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
1193
Table 5 Knowledge of protective eect of breastfeeding on child spacing by level of intergenerational education Level of intergenerational education
Have heard of protective eects of breastfeeding n 282 w2 3:9, p 0:21 Yes No Duration of breastfeeding protection n 174 w2 6:1, p 0:05 All the time Part of the time
groups of higher-educated mothers, the group with the higher-educated grandmother was much more likely to say that their mother (i.e., the grandmother) taught them, whereas they were less likely to say ``no one'' had taught them. When outcomes related to knowledge of the protective eect of breast-feeding on child spacing were examined, two-thirds of women in all IGE groups stated that they had heard of the protective eects of breastfeeding (Table 5). However, women in the low and medium IGE groups were more than twice as likely as their high IGE counterparts to have stated, incorrectly, that protection against pregnancy through breast-feeding endured for the entire breast-feeding period. Women in the medium IGE group (high
Low (%)
Medium (%)
High (%)
Total (%)
60.8 39.2
57.0 43.0
68.2 31.8
62.1 37.9
33.3 66.7
31.5 68.5
15.5 84.5
25.3 74.7
mother's education/low grandmother's education) were more similar to those in low IGE group (low mother's education/low grandmother's education) than to those in the high IGE group (high mother's education/high grandmother's education). This again suggests a potential in¯uence of the more educated grandmothers in the high IGE group on the quality of information held by their daughters. With respect to family planning, women's responses to several of the questions did not vary signi®cantly by IGE strata (Table 6) but there were some interesting suggestive trends. Both groups of mothers with higher education (medium and high IGE) were somewhat more likely to be using contraceptives than the mothers with low education (low IGE). Among contra-
Table 6 Family planning knowledge and use by level of intergenerational education Level of intergenerational education
Current contraceptive use n 277 w2 1:3, p 0:51 NSS Yes No Method currently in use n 116 w2 4:1, p 0:40 NSS IUD Other modern Traditional Desire for additional information n 284 w2 8:4, p 0:08 NSS Yes Maybe No Main reason for non-use of contraceptive n 153 w2 10:5, p 0:10 NSS Lack of knowledge, fear of disapproval Breastfeeding Absence of husband Other reasons Have you talked with other women about contraceptive use n 277 w2 9:6, p 0:008 Never 1±2 times
Low (%)
Medium (%)
High (%)
Total (%)
36.0 64.0
44.6 55.4
45.3 54.7
43.3 56.7
5.3 66.7 27.8
13.5 61.5 25.0
23.9 50.0 26.1
16.4 57.8 25.9
86.3 0.0 13.7
94.3 1.6 4.1
94.6 1.8 3.6
93.0 1.4 5.6
37.5 31.3 9.4 21.9
33.3 27.0 19.1 20.6
25.9 44.8 22.4 6.9
31.4 34.6 18.3 15.7
64.0 36.0
51.7 48.3
38.5 61.5
48.7 51.3
1194
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
ceptive users, IUD use tended to increase from low to medium to high IGE groups, while use of other modern methods (primarily oral contraceptives) decreased. The percentages of contraceptive users who were using traditional methods were similar for all three IGE groups. In explaining their reasons for not using contraception, the percentage stating ``lack of knowledge'' or ``fear of disapproval'' declined steadily from low to medium to high IGE. Women in the high IGE group were also more likely to have reported that breast-feeding was the main reason for their non-use of a contraceptive than the two other groups of women. Interestingly, and supportive of Caldwell's explanation of a gradient of increasing control over one's world as maternal education increases, is the ®nding that the likelihood of having talked with other women about contraceptive use increased steadily, and signi®cantly, pE0:0008 from low to medium to high IGE. Furthermore, within the two higher maternal education groups, those with more educated grandmothers (high IGE) were more likely to have talked with others than counterparts in the group in which the grandmothers had less education (medium IGE). Continuing the exploration of the combined in¯uence of grandmother's and mother's education on selected health outcomes, the authors conducted both two-way chi-square and regression analyses. When dierences between low and medium IGE and between medium and high education were evaluated, with respect to their relationship to number of prenatal care visits (Table 3), dierences between medium and high IGE were found to be signi®cant pE0:001). Likewise, dierences between medium and high IGE were signi®cant in the case of breastfeeding instruction pE0:05 (Table 4); knowledge of the duration of breastfeeding protection pE0:05 (Table 5); and having ever talked with other women about contraceptive use pE0:05 (Table 6). Only in the case of birth attendant were dierences between low and medium IGE signi®cant pE0:001 (Table 3). Thus, for four of the ®ve analyses
with signi®cant dierences, the results indicate variation by grandmother's educational attainment (among mothers with the same educational levels). Only one of the comparisons indicates variation by mother's education (among grandmothers with the same education). In the regression analyses, maternal education and grandmother's education were entered into each model as independent variables. The two variables were highly correlated pE0:0001), according to Pearson correlation coecient. Four of the outcomes that were signi®cant in the univariate analysis were entered into separate models as dependent variables. The variables were: . number of visits for prenatal care; . birth attended by doctor or nurse; . having talked with other women about contraceptive use; and, . knowledge of the duration of pregnancy protection while breastfeeding. Dierences by grandmother's education were signi®cant with respect to number of visits sought for prenatal care pE0:002). That is, there is an eect of grandmother's education on frequency of prenatal care, which is in addition to the in¯uence of maternal education alone. The choice of birth attendant in this sample is in¯uenced more by maternal education (low versus medium or high) than by grandmother's education. Dierences are signi®cant. In the model predicting that women will have talked with other women about contraceptive use, grandmother's education was also signi®cant pE0:001). Access to family planning information in the Bolivian context is recent, and it is likely that women and their mothers are interested in acquiring additional information. Finally, there were no signi®cant dierences by grandmother's education in the model that considered knowledge of duration of pregnancy protection while breastfeeding, suggesting that the level of grandmother's education is less important in in¯uencing knowledge or practices that traditionally are familybased.
Table 7 Alternate categorization of levels of mother's and grandmother's education with mixed educational levels Grades of school completed by grandmother Grades of school completed by mother
0±4 (% (n ))
5±8 (% (n ))
9 or more (% (n ))
0±4 5±8 9 or more
29.3 (51)a 37.4 (65)a 33.3 (58)b
6.0 (5) 40.5 (34)b 53.6 (45)c
6.0 (3) 40.5 (8)c 53.6 (23)c
a
L/M(m)+L(g) n 116). M/H(m)+L/M(g) n 92). c H(m)+M/H(g) n 76).
b
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
1195
Table 8 Maternal and grandmother's education using dierent levels of `high' education for mothers and grandmothers Grades of school completed by grandmother Grades of school completed by mother
0±4 (% (n ))
5±8 (% (n ))
9 or more (% (n ))
0±4 5±8 9 or more
29.3 (51)a 37.4 (65)a 33.3 (58)c
6.0 (5)b 40.5 (34)b 53.6 (45)d
6.0 (3)b 40.5 (8)b 53.6 (23)d
a
L(m)+L(g) n 116). H(m)+L(g) n 50). c L/M(m)+M/H(g) n 58). d H(m)+H(g) n 68). b
This analysis is clearly exploratory. We have considered various ways of analyzing the data and reported those ®ndings. The ®ndings are consistent across methods. One alternative considered included some diagonal groupings (Table 7). Results for low, medium and high groupings were similar to these presented here. Another alternative would be to use dierent groupings of grades for mother and for grandmother, such as depicted in Table 8. The advantage of the latter is that it allows dierent de®nitions of `high' educational level to be used for mothers than for grandmothers. This application may be particularly revealing given that education was prohibited to people of Indian descent, the majority of the Bolivian population, until after the Agrarian Revolution in 1952. Even for women born in the years immediately following 1952, many may not have had access to schooling beyond some primary grades. This categorization produces four, rather than three, groupings, thus permitting comparisons of the eects of grandmother's education within both high and low maternal education groups (as well as comparisons of the eects of mother's education within both grandmother's education groups). It is acknowledged that the dierences among education groups may be attributable to factors other than education, per se. Some of these other factors were measured, and thus could be included in a multivariate analysis, while others were not measured (and may indeed be unmeasurable). Rather than suggesting that education itself is the reason for dierences among education groups, we propose that education is a marker for a constellation of factors. Our results indicate that grandmothers of varying educational levels may have varying eects on maternal behavior, which is important regardless of whether education itself is the reason for these dierences. Similarly, education can be used as a marker to assist in determining educational messages, both for mother and grandmother.
Discussion and recommendations A relationship between intergenerational education, beyond that of maternal education alone and selected formal health care system behaviors appears to exist. The strength of the relationship seems to vary according to the degree or amount of eort involved in the behavior. That is, going for ®ve prenatal care visits requires a more sustained eort than going for any prenatal care. Behaviors typically associated with the informal care sector or the family unit are less aected by IGE than are behaviors related to the formal health care system. This is shown in this example by breastfeeding and breastfeeding related knowledge. In Bolivia, population-wide access to family planning information and devices is recent. Because the reproductive health campaign has used multiple channels of communication, including formal health services, radio, television, billboards and outreach campaigns, it is dicult to know whether to classify modern contraceptive practice in the domain of the formal or informal health care system. In summary, it is apparent that grandmother's education exerts an eect on health behaviors above and beyond the eect of maternal education. This eect is more pronounced for health services which fall clearly in the domain of the formal health care system. Results from the few available studies suggest that a woman continues to rely on her mother, that is, the grandmother, for advice related to important decisions. Use of health services is clearly among these. Further emphasizing the potential strength of the mothergrandmother relationship is the fact that one-tenth of the respondents reported not having a spouse or that their spouse lived in another part of the country (9.4 and 9.8%, respectively). The question of interest in the present paper goes beyond the theoretical. In an article titled, ``Data Needs for the Future'', Vaessen (1991) lists ``edu-
1196
D.E. Bender, M.F. McCann / Social Science & Medicine 50 (2000) 1189±1196
cational level of family members'' as one among several program topics for which additional data are needed. To continue this exploration, data which addresses grandmother's as well as spouse's school attendance ought to be routinely collected. In addition, information on grandmother's recent and past use of health services, grandmother's residence in relation to mother's residence, and explicit questioning of patterns of advice-seeking behavior would be useful additions in further explorations of the strength of the maternalgrandmother in¯uence on health behaviors and health outcomes. Following Illsley (1986), data on occupation and/or education of mother's father, as well as that of the husband, are necessary to explore the dierential eects of various family members on selected health outcomes. This additional information regarding the in¯uence of grandmothers on maternal health decision-making has potential for application in the design and implementation of quality health care services. Programs focused on prenatal care, safe delivery and well-child care, including extended breastfeeding and timely weaning, are examples where the inclusion of the grandmother in a health care visit has the potential to improve health outcomes by taking advantage of the grandmother's experience and the strength of her in¯uence. The usefulness of this team approach is likely to be highest in areas where there are high rates of female-headed households, either due to unmarried childbearing or prolonged absence of husbands.
Acknowledgements The authors want to thank collaborators at Nur University in Santa Cruz, Bolivia, the Earthwatch volunteers and the Bolivian students for their assistance during the data collection phase of the study. The authors also want to acknowledge the ®nancial support received from Earthwatch for the current study conducted in 1994. This study replicated research done in Cochabamba, Bolivia in 1991. The Cochabamba study was conducted with support from the Special Programme of Research, Development and Research Training in Human Reproduction of the World Health Organization.
References Bender, D.E., Baker, R., Dusch, E., McCann, M.F., 1997.
Integrated use of qualitative and quantitative methods to elicit women's dierential knowledge of breastfeeding and lactational amenorrhea in peri-urban Bolivia. Journal of Health and Population in Developing Countries. 1 (1), 68± 84. Bender D.E., Baker,R., McCann, M.F., Body, P., Dusch, E., 1995. Infant feeding practices and child spacing in lowland Bolivia. Replication of WHO's 1992 study; report submitted for publication to EarthWatch and the Center for Field Research, May. Bender D.E., Santander, A., Morales, L., Salinas,W., 1994. Proyecto Madre: a community-based reproductive health intervention to improve maternal knowledge, decisionmaking and service utilization in Peri-urban Bolivia. A ®nal report. UNC-SPH, Chapel Hill, NC. Bender, D.E., McCann, M.F., 1993. Validation of self-report of literacy among peri-urban women in Bolivia. International Journal of Epidemiology. 22 (4), 747±748. Bender, D.E., Rivera, T., Madonna, D., 1993. Rural origin as a risk factor for maternal and child health in peri-urban Bolivia. Social Science & Medicine. 37 (11), 1345±1349. Bicego, G.T., Boerma, J.T., 1991. Maternal education and child survival: a comparative analysis of DHS data. In: Demographic and Health Surveys World Conference Proceedings, vol. 1. IRD/Macro International, Columbia, MD, pp. 177±204. Caldwell, J., 1979. Education as a factor in mortality decline. Population Studies 33, 395±413. Cleland, J., VanGinneken, J., 1989. Maternal schooling and childhood mortality. Journal of Biosocial Science 10 (Suppl.), 123±124. Illsley, R., 1986. Occupational class, selection and the reproduction of inequalities in health. Quarterly Journal of Social Aairs 2 (3), 151±165. Ingstad, B., 1994. The grandmother and household viability in Botswana. In: Adepoju, A., Oppong, C. (Eds.), Work and Population in Sub-Saharan Africa. Heinemann, Portsmouth, NH (Published on behalf of the International Labour Oce). Martin, T.C., 1995. Women's education and fertility: results from 26 demographic and health surveys. Studies in Family Planning. 26 (4), 187±202. Normas nacionales para atencioÂn integral al ninÄo, al escolar, al adolescente y a la mujer, 1992. Manual de Normas y Procedimientos. Ministerio de PrevisioÂn Social y Salud Publica, La Paz. Vaessen, M., 1991. Data Needs for the Future. In: Demographic and Health Surveys World Conference Proceedings, vol. 3. IRD/Macro International, Columbia, MD, p. 2179. Ware, H., 1984. Eects of maternal education, women's roles and child care on child mortality. Population and Development Review 10 (Suppl.), 191±214. World Health Organization (WHO), 1981. Contemporary Patterns of Breastfeeding: Report on the WHO Collaborative Study on Breastfeeding. WHO, Geneva.