Public Health Nutrition: 20(12), 2114–2123 doi:10.

1017/S1368980017000969

Grandmothers’ knowledge positively influences maternal

knowledge and infant and young child feeding practices
Chandni Karmacharya1, Kenda Cunningham2,*, Jowel Choufani3 and
Suneetha Kadiyala1
1
Department of Population Health, Faculty of Epidemiology and Population Health, London School of Hygiene and
Tropical Medicine, London, UK: 2Helen Keller International Nepal, Kathmandu, Nepal: 3Environment and Production
Technology Division, International Food Policy Research Institute, Washington, DC, USA

Submitted 21 July 2016: Final revision received 31 March 2017: Accepted 18 April 2017: First published online 5 June 2017

Abstract
Objective: To examine associations between grandmothers’ knowledge and infant
and young child feeding (IYCF) practices and to test whether the associations are
independent of or operate via maternal knowledge.
Design: Cross-sectional household survey data from households with a child
under 5 years (n 4080). We used multivariate regression analyses, adjusted for
child, maternal, grandmother and household characteristics, and district-level
clustering, to test associations between grandmothers’ knowledge and IYCF
practices for children aged 6–24 months living with a grandmother. We used
causal mediation to formally test the direct effect of grandmothers’ knowledge on
IYCF practices v. maternal knowledge mediating these associations.
Setting: Two hundred and forty rural communities, sixteen districts of Nepal.
Subjects: Children aged 6–24 months (n 1399), including those living with
grandmothers (n 748).
Results: We found that the odds of optimal breast-feeding practices were higher
(early breast-feeding initiation: 2·2 times, P = 0·002; colostrum feeding: 4·2 times,
P < 0·001) in households where grandmothers had correct knowledge v. those
with incorrect knowledge. The same pattern was found for correct timing of
introduction of water (2·6), milk (2·4), semi-solids (3·2), solids (2·9), eggs (2·6) and
meat (2·5 times; all P < 0·001). For the two pathways we were able to test, mothers’
correct knowledge mediated these associations between grandmothers’ knowl-
edge and IYCF practices: colostrum feeding (b = 10·91, P < 0·001) and the
introduction of complementary foods (b = 5·18, P < 0·001). Key words
Conclusions: Grandmothers’ correct knowledge translated into mothers’ correct Nepal
knowledge and, therefore, optimal IYCF practices. Given grandmothers’ influence Child nutrition
in childcare, engagement of grandmothers in health and nutrition interventions Grandmothers
could improve mothers’ knowledge and facilitate better child feeding. Infant and young child feeding

Nepal has experienced some of the fastest rates of reduc- IYCF practices are important for child survival, child
tion in child undernutrition globally: between 1996 growth and development(6,7). The WHO-recommended
and 2011, the prevalence of stunting (height-for-age IYCF practices cover both breast-feeding and com-
Z-score < –2) among children under 2 years of age fell from plementary feeding of children under 2 years of age(4).
48 to 27 %. Improvements in health-service access and According to the 2011 Demographic and Health Survey,
utilization, sanitation and education, coupled with reduc- about 80 % of mothers reported to have exclusively breast-
tions in poverty, drove this stunting reduction over time in fed their children under 6 months of age. Complementary
Nepal(1–3). However, child undernutrition remains a sig- feeding practices continue to be suboptimal: not even one
nificant public health burden, with more than one in four in three children aged 6–24 months consume foods
Nepali children under 2 years of age being stunted(1,2). from at least four of seven food groups, the standard
Infant and young child feeding (IYCF) practices, which for minimum dietary diversity. Also, contrary to the
did not improve between 1996 and 2011, might be a key WHO recommendation of introducing children to com-
contributor to persistent child undernutrition in Nepal(1–5). plementary foods at 6 months of age, nearly one in four

*Corresponding author: Email [email protected] © The Authors 2017

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 Grandmothers’ knowledge and IYCF practices 2115
children 4–5 months of age are fed solid or semi-solid Studies on the role of grandmothers in IYCF practices in
foods and at least 40 % of children 6–8 months of age still South Asia are limited; those conducted to date are
do not consume solid or semi-solid foods(8). This high- primarily qualitative and the quantitative studies were
lights that complementary foods are sometimes introduced based on small sample sizes. There are no prior quanti-
too late and other times, too early. tative studies on this topic in Nepal. In the present study,
Maternal and household demographic and socio- we investigate the associations between grandmothers’
economic factors likely drive poor IYCF practices. Prior correct knowledge and correct practices for the following
studies in Nepal have found maternal employment, edu- IYCF indicators: breast-feeding initiation, colostrum
cation, age and media exposure as determinants of IYCF feeding and timing of introduction of six types of
practices(3,6,7). Cultural beliefs, knowledge and perceptions complementary foods (water, milk (and milk products
have all been shown to influence complementary feeding, other than breast milk), semi-solid foods, solid foods, meat
including which types of foods to introduce first, the timing and eggs) for children 6–24 months of age residing in rural
of their introduction and who should feed young chil- Nepal. We also investigate whether any found association
dren(9). Behaviour change interventions often target is direct or whether maternal knowledge mediates the
mothers with the aim of improving their IYCF-related association.
beliefs, attitudes and knowledge and, ultimately, their IYCF
practices. However, in much of South Asia, including
Nepal, grandmothers are also childcare providers who Methods
influence maternal decisions regarding child feeding. Judi
Aubel highlights the central role grandmothers play in child Survey design and sampling
feeding in Bangladesh, Pakistan and Nepal: grandmothers We use data collected for a baseline survey of a quasi-
advise mothers on caregiving, including child feeding, and experimental impact evaluation of Suaahara, an integrated
they are often direct caregivers for young children(10,11). nutrition programme to improve maternal and child
In South Asia, grandmothers are considered storehouses nutrition in rural Nepal. Data were collected from mid-
of knowledge and wisdom on a wide array of household June to early October (rainy season) of 2012 across
topics. Given their revered status, grandmothers often Nepal’s three agro-ecological zones: mountains, hills and
serve as advisors and supervisors to the next generation, terai(8). Households were selected through a multistage
playing an influential role in child health and nutrition(10). cluster design. First, sixteen districts were purposively
In a qualitative study about the role of Tamang mothers-in- selected to include eight intervention districts and eight
law in Mwakanpur, Nepal, Masvie found that most comparison districts, matched based on various agro-
mothers were assisted by their mothers-in-law for young ecological and sociodemographic characteristics. Second,
child feeding(12). Another qualitative study in Nepal five village development committees per district were
reported that mothers were obliged to obey their mothers- selected and third, three wards per village development
in-law and follow local practices even when the mother committee were selected; both village development
wanted to follow a doctor’s recommendations: although committees and wards were selected using probability-
mothers were responsible for food preparation, grand- proportional-to-size techniques. Lastly, within each ward,
mothers had primary decision-making power over which seventeen households with a child under 5 years of age
foods the child would receive and when(13). Furthermore, were randomly selected from a census carried out by
in Nepal, about four out of five women residing in rural study enumerators. The total sample included 4080
areas are engaged in agricultural labour; many mothers households across 240 wards, including 1399 children
depend on grandmothers to provide care for their young between the ages of 6 and 24 months.
children while they work in the fields(14). Locks et al.
found that 55 % of mothers and 39 % of grandmothers are Ethics approval
primary caregivers of children in Baitadi, Nepal(14). In rural Ethics approval was obtained from the Nepal Health
Nepal, grandmothers provide childcare in part because of Research Council (NHRC), Nepal’s ethical review board,
the heavy workloads of mothers and the authoritative and the internal review board of the International Food
cultural space that grandmothers occupy in Nepali Policy Research Institute (IFPRI) in 2012. For the additional
households, but also because of the drastic increase in analysis in the present publication, ethics approval was
labour emigration of both men and women. The number also obtained from the London School of Hygiene and
of labour permits issued to men increased from 211 371 in Tropical Medicine (LSHTM) in 2015.
2008 to 492 724 in 2014, a 133 % increase, but the number
of labour permits issued to women grew even more Data collection
drastically from 8594 in 2008 to 29 154 in 2014, a steep The baseline household survey involved interviewing
incline of 239 %(15). If this trend continues, grandmothers three household members: (i) the mother of a randomly
may play an increasingly significant role in child health selected child under 5 years of age (the index child);
and nutrition. (ii) a major adult economic decision maker (the mother’s

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 2116 C Karmacharya et al.
husband or adult male, when available); and (iii) the introduction of complementary foods. We were unable to
child’s grandmother, if she resided in the same household. test pathways for early initiation of breast-feeding, as the
These household surveys included asking the primary data set did not include a maternal knowledge variable for
female and male respondents a diverse set of questions this. Given the binary nature of our child feeding outcome
regarding household sociodemographic characteristics variables, we used generalized structural equation mod-
and knowledge and practices relating to maternal and elling for our mediation analysis.
child health and nutrition; water, sanitation and hygiene; In all adjusted models, we controlled for various
agriculture; family planning; and empowerment. The potential confounding factors at the child, mother,
grandmothers’ questionnaire included questions regarding grandmother and household levels, based on our knowl-
their knowledge and beliefs on key maternal and child edge of the local setting and prior studies regarding IYCF
health and nutrition topics, including IYCF. in Nepal(13,14,16–19). Child variables included age in
months, sex (male or female) and decision maker
Data analysis regarding child feeding (mother, grandmother or other).
All statistical analyses were performed using the statistical Maternal characteristics included age in years, level of
software package STATA version 13.0 (2013). For the present education (years of formal schooling completed), whether
IYCF-focused study, we restricted our analysis to households participated in wage or salary employment in the year
with at least one child 6–24 months of age (n 1399) and for prior to the survey (yes or no), number of sources from
our regression analysis, we focused on index children in this which nutrition information was ever heard (newspaper/
age range residing with a grandmother (n 748). magazine, radio, television, brochure/banner/poster, bill-
The primary outcome variables were IYCF practices, as board, flipchart, counseling card, announcements in
reported by mothers: whether or not the child was fed loudspeakers, community/village gatherings, religious
colostrum (yes/no); whether or not there was early gathering/meetings, mother’s group, street drama, health
initiation of breast-feeding, defined as the child receiving facility, female community health volunteers) and resi-
breast milk within the first hour of birth (yes/no); and dency with her own mother (yes or no). Grandmother
whether or not the child was introduced to water, milk characteristics included age in years and level of education
(and milk products other than breast milk), semi-solid (years of education completed). Household characteristics
foods, solid foods, eggs and meat at the appropriate included the number of children <5 years of age residing
age, defined as mothers who reported introducing each of in the household, the agro-ecological zone of residency
the six types of complementary foods between 6 and (mountains, hills or terai), caste/ethnicity (Dalit, dis-
9 months of age (yes/no). These specific IYCF variables advantaged Janajatis, disadvantaged non-Dalit Terai, reli-
were selected based on both the research question gious minorities, relatively advantaged Janajatis, Brahmin/
and data availability from the grandmothers’ survey for Chhetri/Sanyasi), level of food security (food secure,
construction of the parallel primary explanatory variables. mildly food insecure, moderately food insecure or
The primary explanatory variables were grandmothers’ severely food insecure) measured by the Household Food
knowledge and beliefs on these three IYCF practices. Insecurity Access Scale (HFIAS)(20) and wealth quintile
Grandmothers were asked whether or not a child should be (poorest, second poorest, middle income, second
given colostrum; when breast-feeding should start; and wealthiest or wealthiest), calculated using principal com-
when (in months) they believed each of the six types ponent analysis. We also controlled for potential clustering
of complementary foods should be introduced. Binary at the district level and included a variable identifying
variables were created for both breast-feeding variables to whether the district was an intervention or comparison
denote correct knowledge: (i) child should be fed colostrum area, given the purposive selection of intervention areas
(yes/no); and (ii) child should receive breast milk within the by degree of need and how this may influence the results.
first hour of birth (yes/no). Binary variables were also
created for correct knowledge on the timing for introduction
of complementary foods, defined as grandmothers who Results
reported that each of the six types of complementary foods
should be introduced between 6 and 9 months (yes/no). Table 1 presents key descriptive statistics for the child,
We used ANOVA and χ2 tests to explore characteristics mother, grandmother and household; these statistics are
of households with (n 748) and without (n 651) a grand- presented for both types of household: those with (n 748)
mother in residence. Next, multivariate logistic regression and without a grandmother (n 651) in residence. Among
models were used to test the associations between those living with a grandmother in the household, half of
grandmothers’ IYCF knowledge and parallel IYCF prac- the children were male and the mean age was 14 months.
tices. Lastly, we empirically tested the direct v. indirect Mothers were on average 24 years old, with age ranging
effects for grandmothers’ correct knowledge and the from 15 to 42 years. Mean maternal education was 6 years
actual practices, via mothers’ correct knowledge as the of formal schooling. Among surveyed mothers, only 5 %
hypothesized mediator, for colostrum feeding and timely resided with their own mothers and only 13 % claimed to

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 Grandmothers’ knowledge and IYCF practices 2117
participate in wage or salary employment in the past year. disadvantaged caste group in Nepal. The average number
On average, women received nutrition information from of children <5 years of age per household was one. More
only one source of a possible ten sources. Mean age of the than eight out of ten surveyed households were food
grandmothers was 55 years, but their age ranged from 35 secure.
to 85 years. On average, these grandmothers had less than Table 1 also presents some statistically significant dif-
half a year of formal schooling. About half of the house- ferences we found between characteristics of households
holds resided in hill districts, with residency for the other with and without grandmothers. Households with grand-
half split fairly evenly between mountains and terai dis- mothers tend to have mothers that are younger (P < 0·001)
tricts. About half of the households were from most and more highly educated (P < 0·001); however, these
advantaged ethnic group while 17 % were Dalit, the most mothers are less likely to have participated in wage/salary

Table 1 Characteristics of the sample of households from 240 rural communities in sixteen districts of Nepal, 2012

Households with Households without

grandmothers grandmothers
(n 748) (n 651)

% or Mean SD % or Mean SD P value†

Child
Sex: male (0/1) 49·3 – 51·0 – 0·53
Age (months; range 6–24) 14·2 5·2 14·6 5·1 0·15
Age group
6–8·9 months 17·7 – 14·0 –
9–11·9 months 18·7 – 17·8 –
12–14·9 months 18·3 – 18·7 –
15–17·9 months 14·8 – 17·1 –
18–20·9 months 13·8 – 15·8 –
21–23·9 months 16·7 – 16·6 – 0·39
Decision maker: child feeding
Mother 72·1 – 92·2 –
Grandmother 10·3 – 0·3 –
Other 17·6 – 7·5 – <0·001
Mother
Age (years; range 15–42) 23·6 4·5 26·8 6·1 <0·001
Education (years of formal schooling completed; range 0–14) 5·9 4·1 4·0 4·1 <0·001
Participation in wage/salary employment in past 12 months, yes (0/1) 13·2 – 18·0 – 0·01
Number of sources from which nutrition information ever heard (range 0–10) 0·9 1·4 0·8 1·4 0·12
Currently residing in own mother’s home, yes (0/1) 4·7 – 1·8 – 0·003
Grandmother
Age (years; range 35–85) 55·1 9·3 NA – NA
Education (years of formal schooling; range 0–12)‡ 0·3 1·2 NA – NA
Household
Number of children <5 years of age 1·4 0·6 1·4 0·52 0·72
Agro-ecological zone of residency
Mountain 23·5 – 27·0 –
Hill 52·7 – 51·8 –
Terai 23·8 – 21·2 – 0·25
Caste/ethnicity
Dalit 17·4 – 21·8 –
Disadvantaged Janajatis 21·9 – 24·0 –
Disadvantaged non-Dalit Terai 5·2 – 3·2 –
Religious minorities (Muslim/Churoute) 0·8 – 1·5 –
Relatively advantaged Janajatis 5·2 – 4·8 –
Most advantaged (Brahmin/Chettri/Thakuri/Sanyasi) 49·5 – 44·7 – 0·06
Food security§
Severely food insecure 0·9 – 3·2 –
Moderately food insecure 5·8 – 13·1 –
Mildly food insecure 11·2 – 19·1 –
Food secure 82·1 – 64·7 – <0·001
Wealth quintile
Poorest 20·1 – 20·0 –
Second poorest 20·1 – 20·0 –
Middle income 19·9 – 20·0 –
Second wealthiest 20·1 – 20·0 –
Wealthiest 19·9 – 20·0 – <0·001

NA, not available.

†P values calculated using ANOVA for variables with mean values reported and χ2 test for variables with prevalence.
‡Data missing for one or two individuals (n 747, n 746).
§Measured with the Household Food Insecurity Access Scale (HFIAS)(20).

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 2118 C Karmacharya et al.
employment in the past 12 months (P = 0·01). Compared Mothers’ knowledge was higher than grandmothers’
with those households without a grandmother present, a knowledge for all IYCF variables analysed. Mothers
higher proportion of households with grandmothers are reported that six of ten children were breast-fed within the
from the Brahmin/Chettri/Thakuri/Sanyasi caste and first hour of birth and that about nine of ten children
ethnicity group and more have achieved food security, received colostrum. Thirty-seven per cent of the mothers,
as measured by the HFIAS (P < 0·001). in households with grandmothers, reported feeding water
Table 2 summarizes both grandmothers’ and mothers’ and milk and milk products at the correct time, while 64
IYCF knowledge and household IYCF practices included in and 70 % of the mothers reported feeding semi-solid foods
the analysis. Most grandmothers had correct breast-feeding and solid foods at the correct time. Half of the mothers
knowledge: breast-feeding should be initiated within an reported correct practice of introducing eggs and meat at
hour of birth (71 %) and colostrum should be given to the 6–9 months of age. None of the differences in maternal
baby (83 %). While slightly less than half of the grand- knowledge or IYCF practices between households with
mothers had correct knowledge that water and meat pro- and without grandmothers were statistically significant.
ducts should be introduced at 6–9 months of age, a majority Table 3 shows results from the regression analysis of
of the grandmothers had correct knowledge on when semi- grandmothers’ correct knowledge and optimal household
solid foods and solid foods should be introduced (71 and practices for breast-feeding initiation, colostrum feeding
75 %, respectively). Half of the grandmothers had correct and timing of introduction of each of the six types of
knowledge regarding timing of introduction of milk and complementary foods. When adjusting only for district-level
eggs. Only 17 % of the grandmothers reported correctly clustering, the odds of correct IYCF practices were
(6–9 months of age) for when all six of these com- about two to four times higher in households where
plementary foods should be introduced. grandmothers had correct knowledge compared with

Table 2 Infant and young child feeding (IYCF): grandmothers’ knowledge, maternal knowledge and household practices in the sample of
households from 240 rural communities in sixteen districts of Nepal, 2012

Households with Households without

grandmothers grandmothers
(n 748) (n 651)

n % n % P value
Grandmothers’ knowledge
Initial breast-feeding: should be within an hour of birth 744 71·2 NA – NA
Colostrum: should be given to child 740 83·1 NA – NA
Each complementary food should be introduced at 6–9 months of age:
Water 743 46·4 NA – NA
Milk and milk products 743 54·9 NA – NA
Semi-solid foods 743 70·7 NA – NA
Solid foods 743 74·2 NA – NA
Eggs 743 49·3 NA – NA
Meat 743 45·1 NA – NA
All six complementary foods 743 17·2 NA – NA
Mothers’ knowledge
Initial breast-feeding: should be within an hour of birth NA NA NA NA NA
Colostrum: should be given to child 744 92·9 647 90·4 0·17
Each complementary food should be introduced at 6–9 months of age:
Water 748 54·8 651 54·5 0·92
Milk and milk products 747 66·0 651 63·6 0·35
Semi-solid foods 747 78·2 651 78·6 0·83
Solid foods 747 76·3 651 78·3 0·36
Eggs 747 56·4 651 56·4 0·99
Meat 747 53·5 651 53·3 0·93
All six complementary foods 747 27·7 651 27·6 0·98
IYCF practices
Initial breast-feeding: within an hour of birth 748 64·2 651 67·4 0·20
Colostrum: given to child 740 90·1 649 91·1 0·54
Each complementary food introduced at 6–9 months of age:
Water 748 36·9 651 35·8 0·67
Milk and milk products 748 36·6 651 37·6 0·70
Semi-solid foods 748 64·3 651 63·1 0·65
Solid foods 748 70·4 651 67·7 0·27
Eggs 748 51·6 651 51·1 0·87
Meat 748 51·3 651 52·4 0·70
All six complementary foods 748 12·6 651 11·5 0·55

NA, not available.

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 Grandmothers’ knowledge and IYCF practices 2119
Table 3 Associations between grandmothers’ correct knowledge and correct infant and young child feeding practices among children
6–24 months of age in the sample of households from 240 rural communities in sixteen districts of Nepal, 2012

Household practices

Crude† Adjusted‡

n OR 95 % CI P value§ OR 95 % CI P value§
Breast-feeding initiation 740 1·89 1·17, 3·06 <0·001 2·16 1·33, 3·52 0·002
Colostrum feeding 740 3·84 2·94, 5·01 <0·001 4·24 2·81, 6·40 <0·001
Timing of introduction of complementary foods
Water 735 2·46 1·63, 3·72 <0·001 2·60 1·64, 4·11 <0·001
Milk (other than breast milk) 739 1·92 1·20, 3·06 0·007 2·38 1·49, 3·80 <0·001
Semi-solid foods 741 3·30 1·98, 5·50 <0·001 3·20 1·88, 5·44 <0·001
Solid foods 739 3·07 2·23, 4·26 <0·001 2·93 2·08, 4·13 <0·001
Eggs 739 2·05 1·34, 3·11 0·001 2·64 1·57, 4·43 <0·001
Meat 739 2·01 1·46, 2·77 <0·001 2·46 1·67, 3·62 <0·001
All six complementary foods 678 2·08 1·05, 4·12 0·035 2·55 1·42, 4·56 0·002

†Controlled for district-level clustering.

‡Adjusted for district-level clustering, all child factors (age, sex and decision making on child feeding), maternal factors (age, education, participation in wage/
salary employment, number of sources from which nutrition information ever heard and currently residing in own mother’s home), grandmother factors (age and
education) and household factors (number of children <5 years of age, agro-ecological zone of residency, ethnicity, food security, wealth quintile) and whether
the household belonged to the intervention area or not.
§From Wald test.

households where grandmothers had incorrect knowledge. Discussion

In the adjusted models, the odds of appropriate breast-
feeding initiation and colostrum feeding were 2·2 and 4·2 The present study explored the association between
times higher (P = 0·002 and P < 0·001), respectively, in grandmothers’ knowledge and selected IYCF practices of
households where grandmothers had correct knowledge children under 2 years of age: initial breast-feeding,
compared with households where grandmothers had feeding colostrum, and the appropriate timing of intro-
incorrect knowledge. Similarly, in the adjusted models, duction of water, milk, semi-solid foods, solid foods, meat
correct household timing of introducing water, milk, semi- and eggs as complementary foods. The results from the
solid foods, solid foods, eggs and meat were all two to three cross-sectional data set that included children 6–24 months
times higher in households where grandmothers had of age (n 1399), of whom 748 lived with grandmothers
appropriate knowledge compared with households where and 651 did not live with grandmothers, showed that more
grandmothers did not have appropriate knowledge (all than half initiated breast-feeding within the first hour of
P < 0·001). Results were also robust when checked for birth, nearly all were given colostrum, and the percentage
clustering at the ward level (results available upon request). of households with appropriate timing of introduction of
Tables 4 and 5 present results from our generalized complementary foods varied among the six types of food
structural equation modelling, based on our hypothesized (water, milk, semi-solid foods, solid foods, meat and eggs).
models (Figs 1 and 2) of how grandmothers’ knowledge Using logistic regression models we found that, among
may influence IYCF practices directly as well as indirectly households with grandmothers in residence, the odds of
via maternal knowledge. Our results show that grand- appropriate IYCF practice were at least two times or more
mother’s correct knowledge on feeding colostrum mostly likely in households where grandmothers had correct
influences the actual practice via influencing maternal knowledge compared with households were grand-
knowledge. The direct effect is almost zero (Fig. 3), mothers had incorrect knowledge and these results were
whereas there is a large, significant indirect effect: grand- statistically significant. Our mediation analyses showed
mothers’ correct knowledge improves the odds of that almost all of the association between grandmothers’
mothers’ having correct knowledge by 2·1 times practice-specific IYCF knowledge and actual practice is
(P < 0·001) and mothers’ correct knowledge improves the mediated by maternal knowledge: grandmothers’ IYCF-
odds of colostrum being fed by 5·3 times (P < 0·001). related knowledge influences maternal IYCF knowledge
Similarly, grandmother’s correct knowledge on timely and in turn household IYCF practices.
introduction of complementary foods has almost no direct In the present study, the variation across IYCF practices
effect (Fig. 4), but the indirect effect is positive and sig- even among the same population was substantial.
nificant: the odds of maternal correct knowledge are 2·1 Although more than 90 % of children in rural Nepal were
times higher in households where grandmothers have given colostrum, initial breast-feeding in the first hour after
correct knowledge (P < 0·001) and correct maternal birth happened only in about two-thirds of households.
knowledge improves the odds of complementary foods This prevalence was higher than in Nepal’s 2011
being introduced at the right time by 2·5 times (P < 0·001). Demographic and Health Survey, which showed that less

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 2120 C Karmacharya et al.
Table 4 Path analysis: exponentiated effects for grandmothers’ Table 5 Path analysis: exponentiated effects for grandmothers’
knowledge and colostrum feeding (n 737) in the sample of house- knowledge and introduction of complementary foods (n 741) in the
holds from 240 rural communities in sixteen districts of Nepal, 2012 sample of households from 240 rural communities in sixteen dis-
tricts of Nepal, 2012
Colostrum Maternal knowledge:
feeding† colostrum feeding† Timely introduction
of all Maternal knowledge:
Grandmothers’ correct 0·12 2·06*** complementary introduction of
knowledge foods† complementary foods†
Mothers’ correct knowledge 5·30*** NA
Child age Grandmothers’ correct 0·15 2·07***
knowledge
9–11·9 months −0·10 0·66 Mothers’ correct 2·50*** NA
12–14·9 months 0·30 0·45 knowledge
15–17·9 months 0·61 −0·09 Child age
18–20·9 months 0·77 −0·07 9–11·9 months −1·93*** 0·62(*)
21–23·9 months 1·26 0·54 12–14·9 months −1·14*** 0·33
Child sex 0·67 −0·57 15–17·9 months −1·52*** 0·67(*)
Decision maker: child feeding 18–20·9 months −1·72** 0·34
Grandmother 0·50 −0·14 21–23·9 months −1·22*** 0·24
Other −0·68 −0·09 Child sex −0·63(*) 0·01
Maternal age 0·01 0·10* Decision maker: child feeding
Grandmother −0·76 −0·13
Maternal education Other −0·54 0·42
Some primary −0·33 0·39 Maternal age 0·03 −0·02
Completed primary −0·46 0·34 Maternal education
Some secondary −0·73(*) 1·45** Some primary −0·20 0·01
Completed secondary 0·85 0·98* Completed primary 0·55 0·05
Higher level −0·63 2·53* Some secondary −0·37 0·52
Maternal employment 0·24 −0·68 Completed −0·32 0·34
Maternal nutrition information: 0·04 −0·14 secondary
number of sources Higher level −1·40(*) 0·28
Maternal residency with own −1·03 1·27 Maternal employment 0·18 0·32
Maternal nutrition 0·01 −0·04
mother
information: number
Grandmother’s age 0·01 0·01 of sources
Grandmother’s education Maternal residency 0·84 −0·81
Some primary −1·57* −1·19 with own mother
Completed primary 14·1*** 14·4*** Grandmother’s age 0·01 0·01
Some secondary 13·8*** 13·7*** Grandmother’s education
Completed secondary 13·6*** 11·9*** Some primary 0·88 −1·10*
Higher level 13·3*** 14·5*** Completed primary −13·8*** −0·69
Household agro-ecological zone of residency Some secondary 3·01*** 0·37
Hill 0·19 −1·18(*) Completed −10·4*** −14·4***
secondary
Terai 0·51 −1·09
Higher level −13·0*** 0·94
Household caste/ethnicity Household agro-ecological zone of residency
Disadvantaged Janajatis −0·03 0·62 Hill −0·39 −0·07
Disadvantaged non-Dalit 0·64 0·00 Terai −1·52(*) −0·41
Terai Household caste/ethnicity
Religious minorities 17·1*** −1·13 Disadvantaged −0·53 0·02
(Muslim/Churoute) Janajatis
Relatively advantaged 0·05 −0·43 Disadvantaged −13·4*** 0·63
Janajatis non-Dalit Terai
Most advantaged (Brahmin/ 0·25 0·31 Religious minorities −12·0*** −14·9***
Chettri/Thakuri/Sanyasi) (Muslim/Churoute)
Relatively 0·03 −0·28
Household food security advantaged
Mildly food insecure −0·44 0·09 Janajatis
Moderately food insecure −1·13 −0·43 Most advantaged −0·17 −0·51
Severely food insecure 1·61 −1·07 (Brahmin/Chettri/
Household wealth quintile Thakuri/Sanyasi)
Second poorest 0·63 −0·20 Household food security
Middle income 0·01 0·17 Mildly food 0·15 0·09
Second wealthiest −0·25 0·86 insecure
Wealthiest 0·21 0·54 Moderately food −0·22 0·15
Household number of children <5 years of age insecure
Severely food −15·7*** 0·20
2 0·36 0·19 insecure
3 or more 0·27 0·98 Household wealth quintile
Second poorest −0·69(*) 0·91**
NA, not available. Middle income −0·37 0·18
(*)P < 0·10, *P < 0·05, **P < 0·01, ***P < 0·001. Second wealthiest −0·72* 0·44(*)
†Adjusted for clustering at district level. Wealthiest −2·30** 0·92*
Household number of children <5 years of age
2 0·39 0·27
3 or more −0·22 0·14
than 45 % of children in rural areas born in the previous
NA, not available.
2 years were breast-fed within an hour of birth. Similarly, (*)P < 0·10, *P < 0·05, **P < 0·01, ***P <0·001.
the timing of introduction of complementary foods varied †Adjusted for clustering at district level.

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 Grandmothers’ knowledge and IYCF practices 2121
age and called for future child nutrition interventions in
Grandmother’s Colostrum feeding
knowledge on South Asia to include all family members who are childcare
colostrum feeding
providers(21). In a mixed-methods study in rural Haryana,
Mother’s knowledge on India, Kausal et al. found that grandmothers believed that
colostrum feeding complementary foods should only be introduced when the
Fig. 1 Empirical model for direct and indirect effects of infant starts walking or asking for food, while the mothers
grandmothers’ knowledge on colostrum feeding and actual believed complementary feeding should be introduced
practice
much earlier(22). This is consistent with our findings,
underscores that IYCF-related knowledge may differ
Grandmother’s Timely introduction of CF among adult childcare providers in the same household
knowledge on
timely introduction and highlights the need for ensuring that grandmothers are
of CF
also aware of optimal IYCF practices, given their influential
Mother’s knowledge on
timely introduction of CF role in child feeding in South Asia(23). Analysing a
cross-sectional data set from the Young Lives study in
Fig. 2 Empirical model for direct and indirect effects of
grandmothers’ knowledge on timely introduction of Andhra Pradesh, India, Moestue and Huttly found inde-
complementary foods (CF) and actual practice pendent associations between maternal, paternal and
grandmother characteristics and child nutritional status,
further highlighting the need for programmes to target
b = 0.12 other family and community members, rather than just
Grandmother’s Colostrum feeding mothers(24). Studies from outside South Asia have also
knowledge on
colostrum b = 2.06*** b = 5.30***
emerged showing the important role of grandmothers for
feeding
child health and nutrition and the need for targeting them
Mothers’s knowledge
on colostrum feeding with behaviour change communication interventions(25–27).
As mentioned, commonly in South Asia and in Nepal,
Fig. 3 Exponentiated coefficients (b) for direct and indirect
effects of grandmothers’ knowledge on colostrum feeding and senior women are highly revered and hold the top position
actual practice in Nepal, 2012; ***P < 0·001 in the family hierarchy, often having power and authority
over younger women, including relating to infant feed-
b = 0.15
ing(27,28). A study in Maharashtra, India showed that
Grandmother’s Timely introduction
grandmothers take care of infants and teach first-time,
knowledge on of CF inexperienced mothers about breast-feeding, prelacteal
timely b = 2.07*** b = 2.50***
introduction feeding, complementary feeding and feeding during child
of CF Mothers’s knowledge
on timely introduction of CF
illness. That study found that mothers often follow practices
that their mothers and mothers-in-law suggest, perhaps due
Fig. 4 Exponentiated coefficients (b) for direct and indirect
to their own inexperience or to avoid conflict and maintain
effects of grandmothers’ knowledge on timely introduction of
complementary foods (CF) and actual practice in Nepal, 2012; tradition(23). Mothers generally consult grandmothers
***P < 0·001 regarding feeding of infants and young children and some
grandmothers feel that it is their responsibility to teach
substantially by type of food, with water and milk given mothers and pass on their knowledge, perceiving them-
too early and eggs and meat given too late, both in the selves as ‘providers of perinatal care’(12), a perception
present study and other studies in Nepal. A previous study strengthened by first-time mothers relying on them for
noted that this is in part because of a belief that com- expertise and knowledge regarding IYCF practices.
plementary foods should be soft to avoid diarrhoea and The present study suffers from a few limitations. First, the
stomach aches and that animal-source foods should only study is based on a cross-sectional data set and thus our
be introduced once the child develops teeth and the ability regression results cannot be interpreted as causal. Second,
to digest appropriately(13,14). Attention to each of these this was not a nationally representative sample; the sixteen
IYCF practices is important given their importance for districts of Nepal included in the survey were purposively
ensuring optimal child nutrition. selected for an impact evaluation baseline. Third, there may
The present study highlights the potential role of other be residual unaccounted for confounders not available in
adult household members, specifically grandmothers, in the data set or biases which are inevitable in self-reported
child feeding and nutrition. The positive associations practices (e.g. IYCF). We asked mothers of children up to
between grandmothers’ knowledge and IYCF practices are 2 years of age to remember child feeding practices since
consistent with earlier, albeit scant, literature in South Asia. birth and acknowledge potential recall bias as a study
In a study in Vadodara, India, Sharma and Kanani found limitation(29–34). Finally, in this data set, the grandmother
that households in which a grandmother resided, in com- data are limited and certain interesting data points, such as
parison to households without a grandmother, delayed whether the grandmothers had been exposed to nutrition-
initiation of complementary feeding to beyond 6 months of related counselling or other inputs, are unavailable.

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 2122 C Karmacharya et al.
Despite these limitations, the present study is the first deficiencies, are all important steps in the right direction.
empirical study in Nepal, and one of few globally, to inves- Our findings suggest these efforts may be more effective in
tigate associations between grandmothers’ knowledge and improving IYCF practices and reducing undernutrition if
IYCF practices. Our study adds to prior studies, which were other adult household members, such as grandmothers,
mostly qualitative or had very small sample sizes, failed to are also explicitly targeted(24,36,37). In rural Nepal, this is
adjust for potential confounding and clustering, and looked now particularly important in the light of increasing exit
only at either breast-feeding or complementary feeding. migration of women from Nepal and young children being
Furthermore, our study disaggregates the complementary left behind to be cared for by grandparents. Nepal’s pre-
foods rather than lumping the various food types, which is sent climate of commitment to nutrition and to addressing
important given variation in use of these foods for child the burden of undernutrition, particularly among children
feeding. The mediation analysis also allowed us to explore in the first ‘1000 days’ of life, must be pushed forward. This
how grandmothers’ knowledge may influence nutritional is an ideal time to adopt innovative approaches to nutri-
practices compared with traditional regression analysis, which tional obstacles so that further reductions in child nutrition
answers only whether associations exist or not. Therefore, can be seen and to engage in rigorous research to
our study suggests additional hypotheses around how assess the effectiveness of these approaches and build a
household factors influence IYCF practices in South Asia. stronger evidence base for how to address persistent
undernutrition.
Future research and programme and
policy implications Acknowledgements
Globally, most child nutrition policies and programmes
target mothers of young children, aiming to increase their Acknowledgements: The authors would like to acknowledge
awareness and motivation to engage in optimal IYCF various organizations within the Suaahara project, including
practices. In South Asia, grandmothers’ roles as decision the US Agency for International Development (USAID),
makers and advisors on child health and nutrition suggest Save the Children and Helen Keller International (HKI).
that they should be integrated into existing programmes or They also thank Akriti Singh and Swetha Manohar for pro-
that interventions designed specifically for grandmothers viding valuable feedback to this paper. They are particularly
and other adult household members could prove bene- grateful to the thousands of households who participated in
ficial. One way of addressing knowledge gaps may be to the Suaahara baseline study for their time and energy. C.K.
improve health workers’ knowledge and their counselling would also like to thank Laura Brye, Pooja Pandey and
skills for communicating accurate ICYF knowledge and Dale Davis for facilitating the internship at HKI Nepal so that
ensure that health workers are reaching childcare provi- this study could be carried out and Bishnu Dulal for assisting
ders at large with this information, rather than just with data analysis. Financial support: Data collection and
mothers. Different child nutrition approaches and pro- work on this manuscript was supported by Suaahara,
grammes may be needed to target households with and funded by USAID. USAID had no role in the design, analysis
without grandmothers, since we observed differences in or writing of this article. Conflict of interest: HKI,
demographics between these two types of household. a Suaahara-implementing organization, provided an unpaid
Additional studies could analyse factors that may shape internship opportunity for C.K. to carry out this analysis in
grandmothers’ knowledge, pathways and mechanisms for Nepal and funded a separate research consultancy for K.C.
how this influences child feeding and the nutritional status However, none of this analysis assesses the impact of
of the child, and how these differ by various demographic Suaahara. Authorship: C.K., K.C. and S.K. designed the
(e.g. caste) and socio-economic (e.g. wealth) character- research; C.K., J.C. and K.C. analysed the data; C.K., K.C.
istics. Detailed attention should be given to which house- and S.K. wrote the paper; C.K. had primary responsibility for
hold member undertakes which specific childcare roles and final content; all authors read and approved the final
how often to inform the design of future interventions. manuscript. Ethics of human subject participation: Ethics
Rigorous experimental and quasi-experimental studies are approval was obtained from the NHRC and the internal
needed to investigate how grandmothers, and other adult review board of IFPRI (2012); and also from the LSHTM
household members, effect change in IYCF practices. While (2015) for the additional analysis presented herein.
the focus in the present study was on feeding infants and
young children, a similar analysis for other nutrition-related
practices, including diet for women during pregnancy References
and lactation, food hygiene, and water, sanitation and
hygiene behaviours, would also be informative. 1. Cunningham K, Headey D, Singh A et al. (2017) Maternal
The Government of Nepal’s National Multi-sector and child nutrition in Nepal: examining drivers of progress
from the mid-1990s to 2010s. Glob Food Sec (Epublication
Nutrition Plan and the various nutrition-sensitive and ahead of print version).
nutrition-specific interventions(35), aiming to address the 2. Cunningham K, Singh A, Headey DD et al. (2016) Reaching
high prevalence of stunting, wasting and micronutrient new heights: 20 years of nutrition progress in Nepal.

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press

 Grandmothers’ knowledge and IYCF practices 2123
In Nourishing Millions: Stories of Change in Nutrition, and Health Survey 2006–2007. Matern Child Nutr 8,
pp. 115–123 [S Gillespie, J Hodge, S Yosef et al., editors]. Suppl. 1, 60–77.
Washington, DC: International Food Policy Research 20. Coates J, Bilinsky P & Coates J (2007) Household Food
Institute. Insecurity Access Scale (HFIAS) for Measurement of Food
3. Headey D & Hoddinott J (2015) Understanding the rapid Access: Indicator Guide. Washington, DC: Food and Nutri-
reduction of undernutrition in Nepal, 2001–2011. PLoS One tion Technical Assistance, FHI 360.
10, e0145738. 21. Sharma M & Kanani S (2006) Grandmothers’ influence on
4. Devkota MD, Adikari RK & Upreti S (2016) Stunting in child care. Indian J Pediatr 73, 295–298.
Nepal: looking back, looking ahead. Matern Child Nutr 12, 22. Kaushal M, Aggarwal R, Singal A et al. (2005) Breastfeeding
257–259. practices and health-seeking behavior for neonatal sickness
5. Headey D, Hoddinott J & Park S (2016) Drivers of in a rural community. J Trop Pediatr 51, 366–376.
nutritional change in four South Asian countries: a dynamic 23. Aubel J (2012) The role and influence of grandmothers on
observational analysis. Matern Child Nutr 12, Suppl. 1, child nutrition: culturally designated advisors and care-
210–218. givers. Matern Child Nutr 8, 19–35.
6. Pan American Health Organization & World Health Orga- 24. Moestue H & Huttly S (2008) Adult education and child
nization (2003) Guiding Principles for Complementary nutrition: the role of family and community. J Epidemiol
Feeding of the Breastfed Child. Washington, DC and Community Health 62, 153–159.
Geneva: PAHO and WHO. 25. Ahrari M, Houser RF, Yassin S et al. (2006) A positive
7. Black RE, Allen LH, Bhutta ZA et al. (2008) Maternal and deviance-based antenatal nutrition project improves birth-
child undernutrition: global and regional exposures and weight in Upper Egypt. J Health Popul Nutr 24, 498–507.
health consequences. Lancet 371, 243–260. 26. Aubel J, Touré I & Diagne M (2004) Senegalese grand-
8. Ministry of Health and Population (Nepal), New Era & ICF mothers promote improved maternal and child nutrition
International (2011) Nepal Demographic and Health practices: the guardians of tradition are not averse
Survey. Kathmandu and Calverton, MD: Ministry of Health to change. Soc Sci Med 59, 945–959.
and Population, New Era and ICF International. 27. Aubel J & Alvarez M (2011) The Roles and Influence of
9. Stewart CP, Iannotti L, Dewey KG et al. (2013) Con- Grandmothers and Men: Evidence Supporting a Family-
textualising complementary feeding in a broader framework Focused Approach to Optimal Infant and Young Child
for stunting prevention. Matern Child Nutr 9, 27–45. Nutrition: Literature Review. Washington, DC: US Agency
10. Aubel J & Alvarez M (2011) The Roles and Influence of for International Development.
Grandmothers and Men: Evidence Supporting a Family- 28. Simkhada B, Porter MA & van Teijlingen ER (2010) The role
Focused Approach to Optimal Infant and Young Child of mothers-in-law in antenatal care decision-making
Nutrition. Washington, DC: US Agency for International in Nepal: a qualitative study. BMC Pregnancy Childbirth
Development. 10, 34.
11. Aubel J & Project TG (2005) Grandmothers: A Learning 29. Natland ST, Andersen LF, Nilsen TIL et al. (2012) Maternal
Institution. Washington, DC: US Agency for International recall of breastfeeding duration twenty years after delivery.
Development. BMC Med Res Methodol 12, 179.
12. Masvie H (2006) The role of Tamang mothers-in-law in 30. Li R, Scanlon KS & Serdula MK (2005) The validity and
promoting breast feeding in Makwanpur District, Nepal. reliability of maternal recall of breastfeeding practice. Nutr
Midwifery 22, 23–31. Rev 63, 103–110.
13. Seckel L (2011) Factors that Constrain or Prevent Optimal 31. Karkee R, Lee AH, Khanal V et al. (2014) Infant feeding
Infant and Young Child Feeding Practices in Rural Nepal: information, attitudes and practices: a longitudinal survey in
Findings from a Formative Research Study in Three central Nepal. Int Breastfeed J 9, 14.
Districts. Kathmandu: Nepal Family Health Program II, 32. Bailey S & Hedlund K (2012) The Impact of Cash Transfers
Johns Hopkins Bloomberg School of Public Health. on Food Consumption in Humanitarian Settings: A Review
14. Locks LM, Pandey PR, Osei AK et al. (2013) Using formative of Evidence. London: Humanitarian Policy Group, Overseas
research to design a context-specific behaviour change Development Institute.
strategy to improve infant and young child feeding practices 33. Bai DL, Fong DYT, Lok KYW et al. (2017) Practices, pre-
and nutrition in Nepal. Matern Child Nutr 11, 882–896. dictors and consequences of expressed breast-milk feeding
15. Government of Nepal (2014) Labour Migration for in healthy full-term infants. Public Health Nutr 20, 492–503.
Employment A Status Report for Nepal: 2013/2014. 34. Gibson RS (2005) Principles of Nutritional Assessment.
Kathmandu: Government of Nepal. Oxford: Oxford University Press.
16. Pandey S, Tiwari K, Senarath U et al. (2010) Determinants of 35. Government of Nepal, National Planning Commission
infant and young child feeding practices in Nepal: second- (2012) Multi-Sectoral Nutrition Plan for Accelerating the
ary data analysis of Demographic and Health Survey 2000. Reduction of Maternal and Child Undernutrition in Nepal.
Food Nutr Bull 31, 352–365. Kathmandu: Government of Nepal, National Planning
17. Pokhrel S & Sauerborn R (2004) Household decision- Commission.
making on child health care in developing countries: the 36. Moestue H, Huttly S, Sarella L et al. (2007) ‘The bigger the
case of Nepal. Health Policy Plan 19, 218–233. better’ – mothers’ social networks and child nutrition in
18. Cunningham K, Ploubidis GB, Menon P et al. (2015) Andhra Pradesh. Public Health Nutr 10, 1274–1282.
Women’s empowerment in agriculture and child nutritional 37. Duffield A, Abdullah-Al-Harun, Hossain M et al. (2003) Thin
status in rural Nepal. Public Health Nutr 18, 3134–3145. on the Ground. Questioning the Evidence behind
19. Joshi N, Kingsley EA, Dibley MJ et al. (2012) Determinants World Bank-Funded Community Nutrition Projects in
of inappropriate complementary feeding practices in young Bangladesh, Ethiopia and Uganda. London: Save the
children in Nepal: secondary data analysis of Demographic Children.

https://doi.org/10.1017/S1368980017000969 Published online by Cambridge University Press