Original Articles

Community-based Nutrition Education for Improving Infant Growth in

Rural Karnataka
Kilaru A., Griffiths P.L.*, Ganapathy S. and Ghosh Shanti**
From Belaku Trust, 697 15th Cross Road, J.P. Nagar Phase II, Bangalore 560 078.
*Department of Human Sciences, Loughborough University, UK;
**Pediatrician and MCH Consultant, New Delhi, India.
Correspondence to: Kilaru A., Belaku Trust, 697 15th Cross Road, J. P. Nagar II Phase,
Bangalore 560 078, Karnataka, India.
E-mail: [email protected]

Objective: To evaluate a nutrition education intervention designed to improve infant growth and
feeding practices. Design: An intervention study using monthly nutrition education delivered by
locally trained counsellors targeted at caregivers of infants aged 5-11 months. Comparison of
outcomes for 2 groups - one non-intervention group of infants enrolled in 1997 that did not receive
the intervention in the first year of life, and an intervention group of infants enrolled 1998-1999 that
received the nutrition education. Setting: 11 randomly selected and 2 purposively selected villages
of south Karnataka. Subjects: 138 Infants (n = 69 intervention) aged 5 -11 months. Methods:
Families were administered a monthly questionnaire on feeding and child care behavior, and study
infants were weighed at this time, using the SECA solar scales, developed for UNICEF. Logistic
regression was used to examine differences between intervention and non- intervention infants in
infant feeding behavior outcomes. Results: Statistically significant improvement was found in
weight velocity for female infants in the intervention group. These infants were also more likely to
exhibit at least four positive feeding behaviors – intervention infants had a higher mean daily
feeding frequency (more likely to be fed solids at least four times a day (OR = 4.35, 95% CI =1.96,
10.00), higher dietary diversity (more likely to receive a more diverse diet OR = 3.23, 95% CI =
1.28, 7.69), and were more likely to be fed foods suggested by the counsellors such as bananas (OR
= 10.00, 95% = 2.78, 33.3) compared to non-intervention infants. Conclusion: Nutrition education
and counselling was significantly associated with increased weight velocity among girls and
improved feeding behavior among both boys and girls. These results provide further evidence that
community-based nutrition programs that emphasise appropriate feeding and care behavior can be
used to prevent and address early childhood malnutrition in poor households.
Key words: India, Infants, Intervention, Nutrition education.

C HILDHOOD malnutrition is a signifi-

cant health problem in developing
countries and one of the main causes of infant
faltering normally begins around six months
of age, the time when a diet based
predominantly on breast milk begins to
and child morbidity and mortality(1-4). Over include complementary foods, which when
half of the underweight children in the delivered inappropriately results in growth
world live in India, where approximately faltering(6). Infants in Karnataka display this
47% children under the age of three are pattern of malnutrition, with those aged less
underweight and 46% are stunted(5). Growth than six months having a low prevalence of

INDIAN PEDIATRICS 425 VOLUME 42__MAY 17, 2005

KILARU, ET AL. COMMUNITY BASED NUTRITION EDUCATION

underweight (6%), compared to a much higher identified in each of the 13 villages by

prevalence of underweight (39%) aged 6-11 fieldworkers through a combination of house
months(5). visits, the snowball technique, and by the
A lack of food is not the sole cause of Auxilary Nurse Midwife (ANM)) during the
malnutrition. Lack of awareness and know- study period were enrolled. As many of the
ledge about feeding amount, frequency, type respondents were illiterate, verbal consent was
of food, etc., contributes significantly to poor obtained. Information collection started with a
nutritional status among children even in baseline questionnaire including socio-
families where adults meet their daily demographic characteristics of the household.
requirements(7). Interventions have shown Monthly questionnaires were used to
that it is possible to improve infant growth and document feeding practices, including a 24-
feeding practices through action-oriented hour recall of foods and fluids consumed,
messages(8). Probably the most compre- weight of the infants and morbidity
hensive Indian studies of infant-focused experiences in the previous month; height was
nutrition education interventions have measured every three months. The study team
been conducted by Bhandari and visited infants until 24 months of age. Over the
colleagues(9,10). The findings from these three years in which the study was conducted,
studies suggest that nutrition education 468 infants were recruited.
messages can result in a longer duration of A year into the study, data revealed that
exclusive breastfeeding, decreased diarrheal levels of malnutrition were high and
morbidity, and increased energy intake. knowledge of appropriate infant feeding
practices was very low. The Belaku Trust’s
The objective of the current study was to ethical advisory group decided not to conduct
gain an understanding of infant feeding a randomised control study of the effect of
practices in the period when infants are nutrition education and instead start the
vulnerable to growth faltering and evaluate the intervention nutrition education for all
effect of a nutrition education intervention on families in the study. However, infants
these practices between 7-11 months of age in enrolled in the study in early 1997 were
rural Karnataka. This paper builds on an already a year old by then and had experienced
earlier report where it was shown that the their first year of life without the nutrition
nutrition education intervention had a positive education. This group was considered a non-
impact on the growth of female infants but not intervention comparison group (NI) (n = 69)
males(11). for the first year of life for the analysis and
Subjects and Methods other infants who were not older than 5 months
of age when the counselling began (n = 173)
Sample were considered an intervention group (I).
Infants aged less than six months were Hence, the sub-sample of infants for this
enrolled on an on-going basis between 1997- analysis is 242 (173 + 69). Although we are
1999 from 13 villages, 11 randomly selected comparing data from different calendar years
and two purposively selected to cover a range (because infants were recruited at different
of settings; one, a small village with a tribal times), there were no significant events that
community and the other, a large village with a influenced food supply in the area during
mix of public, private and traditional health 1997-1999 and there were no significant
care providers. All infants and recent births changes in public health services offered,

INDIAN PEDIATRICS 426 VOLUME 42__MAY 17, 2005

KILARU, ET AL. COMMUNITY BASED NUTRITION EDUCATION

therefore we consider the data to be The weight trajectory (using a growth chart),
comparable. The remaining infants in the episodes of illness, and developmental
study were exposed to the intervention some- milestones were also discussed. The tone of
time between 5 and 11 months of age and are the counselling was empathy and concern for
excluded for this analysis. Dropout from the the child and family’s well-being. Senior staff
study prior to 12 months of age was very low reviewed all questionnaires and were often
and thus we do not expect it to have a bearing present for the sessions. The counsellors were
on the results. trained to be mindful of household constraints
especially in terms of available household
The sixty-nine non-intervention infants
food, limited financial resources, decision-
(NI) (of whom 34 are females) did not receive
making capacity and privilege within the
the intervention in 1997 but were exposed to
family structure. The key counselling
the normal standard of care available in that
messages included in the study for infants
time period in the villages, provided by the
aged 5-11 months focussed on:
anganwadi workers (AWW) and auxiliary
nurse midwives (ANM). The 173 intervention • use of developmentally appropriate local
infants (I), of which 83 are females, were foods and preparation of these foods;
recruited into the study during 1998 and 1999 • appropriate feeding frequency;
and were exposed to the standard care
• gradually increasing food diversity;
available plus monthly nutrition counselling
from 5 months of age or earlier. • complementary feeding followed by
breastfeeding;
Intervention
• avoidance of feeding bottles.
The field research staff (high school or
Statistical Analysis
college educated fieldworkers who live in
close proximity to the study villages) also Analysis was done using SPSS(13).
provided the nutrition education. They were Logistic regression (to control for background
trained by Belaku Trust and external differences such as economic status) was used
MCH consultants (a pediatrician and a to examine differences between intervention
nutritionist)(12). and non-intervention infants between 7-11
months of age in outcomes related to the key
After administering the monthly question-
counselling messages. The analysis for the
naire, the field staff would discuss reported
effect of the intervention is examined from 7
feeding practices with the primary caregiver.
months of age to allow time for changes in
The total time for the questionnaire and
feeding practices and changes in the child’s
counselling was about 1½ hours. Field staff
weight to be captured.
would review dietary information from the
questionnaires and talk about ways of In addition to feeding behaviors, a linear
improving the quality and quantity of the diet regression was performed with frequency of
through questions and probes asked in a breastfeeding in 24 hours at 11 months of age
friendly manner. The field staff would often as the outcome variable. This was to assess
provide personal examples of how to whether any increase in the frequency or
overcome challenges in feeding young variety of complementary foods was offset by
children, including the kind of communication a decrease in the number of times an infant was
skills needed to coax young children to eat. breastfed.

INDIAN PEDIATRICS 427 VOLUME 42__MAY 17, 2005

KILARU, ET AL. COMMUNITY BASED NUTRITION EDUCATION

Since there were a larger number of the intervention and non-intervention groups,
intervention infants (173), than non- potentially confounding factors were
intervention (69) we wished to allow for any controlled for in the regression models. These
statistical effect caused by these different included sex, village of residence, caste,
numbers of infants in each group. A random mother’s education, economic level, and the
sample of 69 intervention infants was selected mother’s age at the child’s birth. Since higher
to test the analysis. A sample size of 69 in each weight velocity was observed only in
experimental group will detect differences in intervention girls in bivariate results, a sex-
the outcome variables studied of between 18- intervention interaction term was included in
24% between the intervention and non- the regression models. To account for baseline
intervention group (with P = 0.05 and 80% weight, weight at 6 months of age was also
power) in a bivariate analysis depending on controlled for in the models.
the outcome being examined. Results
No differences were observed in the Most background characteristics were not
outcomes, suggesting that the disparity in statistically different between the intervention
numbers does not affect our findings - hence and non-intervention groups of infants (Table
our final results presented are for the entire I), except for variable called village group.
group of 173 (I) group infants and 69 (NI) This was constructed by classifying the
group infants. In addition, because the villages in which the families live according to
children had not been randomly allocated to the level of health services available (facility

TABLE I–Characteristics of the Intervention and Non-Intervention Infants.

Intervention infants Non-intervention

(n = 173) (%) (n = 69)(%)

Female 54 46
Scheduled caste/ tribe 14 15
Mother no education 48 42
Village group 1* 61 51
Village group 2 13 30
Village group 3 26 19
Low standard of living+ 36 33
Mother aged < 20 at birth 33 43
Mean weight for age Z-score 6 months (SD) –1.08 –0.81
(1.00) (1.04)
* Villages were grouped according to geographic region and the type of health care available. Group 1 has poor
access generally to health facilities and no active health workers. Group 2 has moderate access to a primary
health sub-centre, and has an active Auxiliary Nurse Midwife (ANM) who is involved in outreach. Group 3
contains one large village which has access to a mix of services, private providers, a missionary hospital and a
government clinic, but no active outreach worker.
+ Composite index of ownership of household consumer goods, based on the local value of goods. Low standard

of living refers to the bottom tertile value of the owned household goods.

INDIAN PEDIATRICS 428 VOLUME 42__MAY 17, 2005

KILARU, ET AL. COMMUNITY BASED NUTRITION EDUCATION

type and extent of outreach services). Because the literature to be predictors of nutritional
of this, village group was controlled for in the status. Including a sex-intervention interaction
regression model. term in the regression model revealed that
female infants in the intervention group had
Females enrolled in the intervention had a
greater weight velocities relative to the other
weight velocity that was 77 g per month
groups (NI and I boys, and NI girls).
greater than non- intervention girls between 6-
10 months of age(11) (Fig. 1, bivariate Descriptive statistics for the feeding
results). This was also tested using linear behavior outcomes are displayed in Table II.
regression controlling for socio-demographic Significant differences were observed in the
factors that either varied between the I and NI feeding of bananas, with intervention infants
groups, or because they have been shown in being more likely to eat these (33%) compared

Fig. 1. Average weight velocity, stratified by intervention group.

TABLE II–Percentage of Infants Displaying Positive Feeding Behavior by Intervention Group.

Intervention Non-intervention P-value

(%) (%)
Not feeding animal milk ages 7-11 months 41 29 0.15
Not using a bottle ages 7-11 months 85 72 0.06
Feeding appropriate semi-solid foods ages 7-11 months 32 26 0.45
Feeding bananas ages 7-11 months 33 4 < 0.001
Feeding at least 4 times in 24 hours in addition to breast milk
ages 7-11 months 78 51 < 0.001
Feeding at least 5 different food groups* age 11 months 42 19 0.01
Displays at least 4 positive feeding behaviors† 65 49 0.06
* Food groups examined were dairy, cereal, protein, fruit, vegetables, oil and fat, sugar and savory snacks.
† Out of the six positive behaviors shown in the table above.

INDIAN PEDIATRICS 429 VOLUME 42__MAY 17, 2005

KILARU, ET AL. COMMUNITY BASED NUTRITION EDUCATION

to non-intervention infants (4%). Intervention in Zaire found that girls showed catch-up
infants were also significantly more likely to growth more quickly than boys(15) and what
be fed at least four times in 24 hours in we see here may be a demonstration of this. It
addition to breast milk (78% versus 51%) and is also possible that the fieldworkers paid
to receive foods from at least five different greater attention to families with girls,
food groups (42% versus 19%) in comparison although they were not instructed to do so,
to non intervention infants. because of an expectation of higher risk of
poor growth for girls due to a culture of female
Odds ratios with 95% confidence intervals
discrimination.
were examined from the logistic regression
models for feeding behavior. Non- Girls in the intervention group were more
intervention infants were observed to be likely to have at least 4 positive feeding
significantly less likely to be fed bananas behaviors compared to NI girls. This suggests
(adjusted OR = 10.00, CI 2.78, 33.33), to be that changes in a combined set of feeding
fed solids at least four times in 24 hours behaviors may account for the improved
(adjusted OR = 4.35, CI 1.96, 10.00); and to be weight gain observed in intervention girls
fed from at least five different food groups even though no single feeding practice
(adjusted OR = 3.23, CI 1.28, 7.69) compared appears significantly different for the girls.
to intervention infants. Further analyses will use observed feeding
In addition, we constructed a composite behavior in weight velocity regressions to see
score of 6 positive feeding behaviors and if the increase can be explained by these
compared intervention and non-intervention positive behavior changes.
groups. Intervention females were more likely The study limitations include the non-
to report at least 4 of the 6 positive behaviors randomised allocation of the intervention and
than non-intervention females (adjusted OR = non-intervention groups, which, as explained
2.78, CI 1.02, 7.69). No significant differences above, was done for ethical reasons. However,
were observed for male intervention or non- there is increasing recognition of the
intervention infants. importance and value of evaluation research to
Discussion public health policy and practice, especially
when it is unethical to conduct randomised
One of the most significant findings is the control trials because the intervention being
increase in weight velocity among the withheld has a strong potential for benefit(16).
intervention girls compared to NI girls. In addition, information on exact portion sizes
Although the increase is small, it suggests that would have been useful in assessing the
nutrition education can improve growth in impact of the intervention to see if there was a
poor households, and may be especially direct effect on growth; these data were
effective in regions where females are socially collected, but were unreliable, due to the well-
discriminated against. It is difficult to say why known difficulties of collecting such data in
this effect was seen only in girls; boys tend to the field.
have faster rates of gain relative to girls and as
such are considered more sensitive to the These findings support other evidence that
environment(14); we therefore may have there is significant scope for improving
expected more effect in boys in response to feeding behavior and growth through
improved feeding practices. However, a study counselling and education (9,10,17-19). We

INDIAN PEDIATRICS 430 VOLUME 42__MAY 17, 2005

KILARU, ET AL. COMMUNITY BASED NUTRITION EDUCATION

Key Messages
• Nutrition education and counselling lead to significant increase in weight velocity among girls.
• Counseling women and their families in appropriate feeding and core behavior could reduce
the incidence of malnutrition.

have observed significant positive differences Dr. Jayashree Ramakrishna, Dr. Usha
in weight velocity, feeding frequency, dietary Abraham, Shanti Mahendra, Srilakshmi
diversity, and in the use of specific Divakar and Dr. Thilaka Baskaran; we also
recommended and locally available foods(20). thank Jahnavi Vasudevan for critical
Ideally, it would be better to counsel women comments.
and their families much earlier in the infant’s Contributors: Ganapathy and Ghosh designed the
life, or even to begin during antenatal care. study; Kilaru and Ganapathy implemented the study.
Addressing groups of families through the The analysis and draft manuscript was done by
anganwadi, rather than individuals, would be a Griffiths and Kilaru. All four authors reviewed the
way of using existing programs and reducing manuscript.
the cost of nutrition education interventions. Funding: UNICEF, New Delhi.
Current recruitment requires AWWs to have at Competing interests: None stated.
least a 10th standard education, so training
REFERENCES
them to fulfil their role as community nutrition
advisors is indeed possible. Although macro- 1. Mosley WH, Cowley P. The challenge of
world health. Population Bulletin 1991; 46:
level strategies to reduce food insecurity are
2-33.
needed, in many poor households there is
usually some food available that can be 2. Gage TB, Zansky SM. Anthropometric
indicators of nutritional status and level of
modified and fed to young children with
mortality. Am J of Human Biol 1995; 7: 679-
appropriate frequency. Teaching families to 691.
increase feeding frequency, increase dietary
diversity, modify household food as well as 3. Murray CJ L, Lopez AD, editors. The global
burden of disease: A comprehensive
raising awareness on existing food taboos for assessment of mortality and disability from
young children (e.g., bananas and eggs), could diseases, injuries and risk factors in 1990 and
reduce the incidence of growth faltering and projected to 2020 (Vol 1. Global burden of
subsequent malnutrition. disease and injury series). Boston: Harvard
School of Public Health; 1996.
Acknowledgements
4. Murray CJ, Lopez AD. Mortality by cause for
We wish to thank UNICEF, New Delhi, eight regions of the world: Global burden of
for supporting this study. The authors would disease study. Lancet 1997; 349: 1269-1276.
like to thank the field staff Shobha Prakash, 5. IIPS (International Institute of Population
Jyothi, Mallika, Mani, Tejaswini, C.K. Shoba Sciences) and ORC Macro. National Family
and Anila, who delivered the intervention, and Health Survey (NFHS 2), India 1998-99.
Gopal; ANM Jayanti for her dedication to her Mumbai: IIPS; 2000.
work and assistance to the study. We also 6. World Health Organization, Department of
thank the other members of the research team: child and adolescent health and development.

INDIAN PEDIATRICS 431 VOLUME 42__MAY 17, 2005

KILARU, ET AL. COMMUNITY BASED NUTRITION EDUCATION

Complementary feeding family foods for 14. Bielicki T. Physical growth as a measure of the
breastfed children. WHO/NHD/00.1; Geneva; economic well-being of populations: the
2000. twentieth century. In: F. Falkner, J.M. Tanner
editors. Human Growth, vol 3, New York:
7. National Nutrition Monitoring Bureau. Diet
Plenum Press; 1986.
and nutritional status of rural population.
Technical report no. 21. National Institute of 15. Hauspie RC, Pagezy H. Longitudinal study of
Nutrition (ICMR), Hyderabad, India; 2001. growth of African babies: An analysis of
seasonal variations in the average growth rate
8. Caulfield LE, Huffman SL, Piwoz EG.
and the effects of infectious diseases on
Interventions to improve the complementary
individual and average growth patterns. Acta
food intakes of 6-12 month old infants in
Pediatr Scand 1989; Suppl. 350: 37-43.
developing countries: Impact on growth,
prevalence of malnutrition and potential 16. Kirkwood B. Making public health
contribution to child survival. Food and interventions more evidence based. BMJ
Nutrition Bulletin 1999; 20: 183-200. [online WWW]. 2004 April;328:966-967.
9. Bhandari N, Bahl R, Nayyar B, Khokhar P, http://bmj.bmjjournals.com/cgi/content/full/
Rohde JE, Bhan MK. Food supplementation 328/7446/966?etoc&eaf [23 April 2004].
with encouragement to feed it to infants from 4 17. Cerquieria MT, Olson CM. Nutrition education
to 12 months of age has a small impact on in developing countries: An examination of
weight gain. J Nutr 2001; 131: 1879-1880. recent successful projects. In: P. Pinstrup-
10. Bhandari N, Bahl R, Mazumdar S, Martines J, Andersen, D .Pelletier, H. Alderman editors.
Black RE, Bhan MK. Effect of community- Child Growth and Nutrition in Developing
based promotion of exclusive breastfeeding on Countries. Delhi: Oxford University Press;
diarrheal illness and growth: A cluster 1995.
randomised controlled trial. Lancet 2003; 361: 18. Zeitlin MF, Formacion C. Nutrition education,
1418-1423. study II. In: J. Austin, M.F. Zeitlin editors.
11. Ghosh S, Kilaru A, Ganapathy S. Nutrition Nutrition Intervention in Developing
Education and infant growth in rural Indian Countries: An Overview. Cambridge, MA:
infants: narrowing the gender gap. J Indian Oelgeschlager, Gunn and Hain; 1981.
Med Assoc 2002; 100: 483-490.
19. Hornik R. Nutrition Education: A State of the
12. Ghosh S. The feeding and care of infants and Art Review. ACC/SCN Nutrition Policy
young children. New Delhi: Voluntary Health Discussion Paper No 1. Rome: United Nations
Association of India; 1985. Food and Agricultural Organization; 1985.
13. SPSS. Statistical Package for the Social 20. Ghosh, S. Nutrition and Child Care; A
Sciences. Version 11. SPSS Inc. Chicago, Practical Guide. 2nd edition. New Delhi:
Illinois; 2001. Jaypee Bros; 2004.

INDIAN PEDIATRICS 432 VOLUME 42__MAY 17, 2005