Food and Nutrition Bulletin

This paper details the steps to design and implement a positive deviance-informed, “Hearth” approach for the nutritional rehabilitation of malnourished children in the district of Leogane, Haiti. Groups of four to five children met daily for two weeks at the home of a local volunteer mother for nutritional and health messages and a well-balanced meal. Health messages and meal components were determined using information gathered from interviews with the mothers of positive deviant children in the community who are well nourished despite their family's limited economic resources. Hearth participants were then followed for six months in their own home by the program “monitrices,” women hired from each village and intensively trained to supervise the Hearth program, periodically weigh the children to evaluate their progress, and liaise between the hospital and the community. Monitoring from the first cycle indicated that 100% of children in eight villages and 66% of children in the remaining five villages continued to gain weight as fast or faster than the international standard median six months after participating in a Hearth program. At the conclusion of this cycle, programmers interviewed participant and non-participant families and made six modifications to the model, including the addition of a microcredit option for participating mothers.

Save the Children's (SC) successful integrated nutrition program in Viet Nam, the poverty alleviation and nutrition program (PANP), uses the positive deviance (PD) approach to identify key growth promoting behaviors and provides participatory adult education allowing mothers to develop skills related to these behaviors. We investigated whether improvements seen during a PANP intervention (1993–1995) were sustained three and four years after SC's departure. Cross-sectional surveys were administered to 46 randomly selected households in four communes that had previously participated in the PANP and 25 households in a neighboring comparison community in 1998 and 1999. Two children per household, an older child who had participated in the PANP and a younger sibling who had not, were measured (total n = 142 children), and their mothers were interviewed. Older SC children tended to be better nourished than their counterparts. Their younger siblings were significantly better nourished than those in the comparison group, with adjusted mean weight-for-age Z scores of −1.82 versus −2.45 (p = .007), weight-for-height Z scores of −0.71 versus −1.45 (p < .001), and height-for-age Z scores of −2.11 and −2.37 (ns, p = .4), respectively. SC mothers reporting feeding the younger siblings more than their counterparts did (2.9 versus 2.2 main meals per day, p < .001, and 96.2% versus 52% offering snacks, p < .01]. SC mothers reported washing their hands “often” more than comparison mothers (100% vs. 76%, p < .001). Growth-promoting behaviors identified through PD studies and practiced through neighborhood-based rehabilitation sessions persisted years after program completion. These sustained behaviors contributed to better growth of younger siblings never exposed to the program.

We compared the positive deviance (PD) approach in Save the Children's field guide with a case-control study (CCS) to identify behaviors associated with good nutritional status in Afghan refugee children 6 to 24 months of age in the Northwest Frontier Province (NWFP), Pakistan. The positive deviance inquiry (PDI), utilizing observations and interviews with mothers, fathers, and secondary caregivers in eight households, identified 12 feeding, caring, and health-seeking behaviors that were not widely practiced. The CCS, using the same selection criteria and content as the PDI with 50 mother-child pairs not in the PDI, yielded six significant associations with good nutritional status. Both the PDI and CCS detected feeding behaviors. The PDI alone identified complex phenomena (active feeding and maternal affect). The CCS alone confirmed the beneficial use of health services. The PD approach was an affordable, participatory, and valid method to identify feeding behaviors and other factors associated with good nutrition in this context.

Few prospective studies of child growth and its determinants take place in programmatic contexts. We evaluated the effect of Save the Children's (SC) community empowerment and nutrition program (CENP) on child growth, care, morbidity, empowerment, and behavioral determinants. This paper describes the research methods of this community-based study. We used a longitudinal, prospective, randomized design. We selected 12 impoverished communes with documented child malnutrition, three comparison, and three intervention communes in each of two districts in Phu Tho Province, west of Hanoi. SC taught district trainers in November 1999 to train local health volunteers to implement the 10-month CENP, including situation analysis, positive deviance (PD) inquiry, growth monitoring and promotion, nutrition education and rehabilitation program (NERP), deworming, and monitoring. PD inquiries aim to discover successful care practices in poor households that likely promote well-nourished children. NERPs are neighborhood-based, facilitated group learning sessions where caregivers of malnourished children learn and practice PD and other healthy behaviors. We dewormed all intervention and comparison children. We weighed all children less than 24 months of age living in the intervention and comparison communes and randomly selected 240 children (120 intervention and 120 comparison). We gathered information on nutritional status, diet, illness, care, behavioral determinants, empowerment, and program quality, monthly for six months with a re-survey at 12 months. We collected most information through maternal interview but also observed hygiene and program quality, and videotaped feedings at home. Some implementation and research limitations will attenuate CENP impact and measurement of its effectiveness.

Children who are weighed for growth monitoring are frequently clothed, especially in the cold weather. Health workers commonly estimate and subtract the weight of these clothes, but the accuracy of these estimates is unknown. We assessed the accuracy of child weights adjusted for estimated clothing typical of hot, cold, and extremely cold ambient temperatures. Trained field workers weighed a sample of 212 children 6 to 42 months old from the ViSION project, adjusted the weights using a job aid describing the weights of common clothing by season and age, and then weighed the clothing to calculate the actual clothing and child weights. Fieldworker estimates of the weight of the clothing that children wore during weighing were remarkably good. In nearly all cases (207 of 212; 97.7%), the difference between the estimated and actual clothing weight was less than the precision of the child scales (± 50 g), and most (181 of 212; 84.5%) were within 25 g. Thus, the calculated child weights were, in fact, equivalent to the actual child weights. Using simulations, we found that improperly accounting for clothing weight can overestimate weight-for-age by 0.1 to 0.4 Z score. Accurate weights are possible, even under adverse conditions. Our training methods, clothing album, and job aid might benefit nutrition research and programming in Viet Nam as well as settings with colder climates.

Integrated nutrition programs are widely used to prevent and/or reverse childhood malnutrition, but rarely rigorously evaluated. The impact of such a program on the physical growth of young rural Vietnamese children was measured. We randomized six communes to receive an integrated nutrition program implemented by Save the Children. We matched six communes to serve as controls. Our sample consisted of 238 children (n = 119 per group) who were 5 to 30 months old on entry. Between December 1999 and December 2000, we measured weight and height monthly for six months and again at month 12. Principle outcomes were weight-for-age Z score (WAZ), height-for-age Z score (HAZ), and weight-for-height Z score (WHZ), and the changes among these measures. As expected, anthropometric indicators relative to international references worsened as the children aged. Overall, children in the intervention communes who were exposed to the integrated nutrition program did not show statistically significant better growth than comparison children. Intervention children who were younger (15 months or less) and more malnourished (less than −2 Z) at baseline, however, deteriorated significantly less than their comparable counterparts. Between baseline and month four, for example, intervention children who were malnourished and less than 15 months old at entry lost on average 0.05 WAZ while similar comparison children lost 0.25 WAZ (p = .02). Lack of overall impact on growth may be due to a lower than expected prevalence of malnutrition at baseline and/or deworming of comparison children. Targeting nutrition interventions at very young children will have the maximum impact on growth.

Forty-two percent of Vietnamese children are stunted by two years of age. Since 1990, Save the Children Federation/US (SC) has implemented integrated nutrition programs targeting young children. We evaluated the effect of SC's nutrition program on the complementary food intake of young rural Vietnamese children. Using a longitudinal, prospective, randomized design, we followed 238 children (119 each from intervention and comparison communes) age 5 to 25 months old for six months with a re-survey at 12 months. We gathered 24-hour recall data at baseline and at months 2, 4, 6, and 12. Dietary energy intake was calculated using the 1972 Vietnamese food composition table. Key outcomes were daily frequency of consuming intervention-promoted food and non-breastmilk liquids and food, daily quantity of non-breastmilk liquids and food consumed, daily energy intake, and proportion of children meeting daily median energy requirements. Young rural children exposed to SC's program consumed intervention-promoted and any foods more frequently, ate a greater quantity of any food, consumed more energy, and were more likely to meet their daily energy requirements than comparison children. Some effects were only observed during the intensive intervention period; others persisted into or were evident only at the 12-month follow-up, approximately four months after program completion. Based on the mothers’ reports, the intervention did not apparently compromise breastfeeding prevalence or frequency. The intervention improved children's food and energy intake and protected them from declining as rapidly as comparison children in meeting their energy requirements.

Infectious disease and poor diet are the two proximal causes of malnutrition in children. During the 1990s, integrated nutrition programs implemented by Save the Children (SC) in Vietnam reduced severe child malnutrition, but it has not been clear if this impact was due primarily to improved diet or reduced disease. The aim of this study was to determine whether a community-based, integrated nutrition program in Vietnam reduced child morbidity due to diarrhea or acute respiratory infections. Children 5 to 25 months old were randomly selected from randomly assigned intervention and comparison communes. Caregivers of children from the intervention and comparison groups (n = 119 per group) were interviewed about their child's morbidity at program baseline and at study months 2, 4, 6, and 12. Multiple logistic regression and general estimating equations (GEE) were used to evaluate the effect of the intervention on the occurrence of any diarrhea and respiratory illness in the preceding two weeks. Respiratory illness, mainly upper respiratory illness, was more common than diarrheal disease at baseline (54% vs. 6%, respectively). During follow-up, children in the intervention communes had approximately half the respiratory illness experienced by those in comparison communes (AOR = 0.5; p = .001). Diarrheal disease was also lower in the intervention group, although differences were not statistically significant. We conclude that SC's integrated nutrition program was associated with reduced upper respiratory illness, perhaps due to improved hygiene practices and/or improved micronutrient intakes.

Rigorous assessments of program quality are uncommon in developing countries. We evaluated the quality of the two-week, volunteer-facilitated, caregiver-child rehabilitation “hearth,” or nutrition education and rehabilitation program (NERP), sessions in Save the Children's integrated nutrition program in Viet Nam. Field workers observed attendance, food contribution, food preparation, meal consumption, health message delivery, hygiene, and weighing at 240 NERP days at 59 NERP centers during seven months of implementation. Participation in cooking NERP meals (75.8%), washing mothers’ and children's hands (75.7% and 81.6%, respectively), and weighing (74.5% on days 1 and 12) were high, but attendance rates (50.3%), food contributions (20.3%), and health message delivery (20.0%) were lower than expected, all with wide variation among communes, ecology, NERP day, and NERP round. Contrary to protocol, food was often delivered to malnourished children who stayed at home to accommodate caregivers’ workloads and other constraints. While home-delivery of food prevented caregivers from learning from each other in a group, it did bring “tangible messages” (i.e., a large serving size of a new food) to the doorstep. Health volunteers were wise to stress active learning over repetitive message delivery. In summary, these NERPs were probably typical of previous NERPs although comparable quality measures are lacking. Despite imperfect implementation, the NERP's active-learning and local problem-solving helped achieve measurable impact on growth, diet, morbidity, and empowerment despite uncommon program challenges, such as uncharacteristically low baseline levels of malnutrition and high population dispersion. Regular quality monitoring may enhance impact even further.

Empowerment is often cited as a fundamental component of health promotion strategies. Anecdotes suggest that Save the Children's integrated nutrition project empowers local women and health volunteers. The aim of this research was to document the degree to which this is being accomplished. Using qualitative methodologies, we conducted a cross-sectional assessment to compare self-reported changes in identified empowerment domains among 17 program health volunteers and 20 mothers involved in a child nutrition intervention and among five Women's Union leaders and five mothers in a non-intervention comparison commune. Intervention mothers reported increased knowledge, confidence, and information sharing about child-care and feeding, while non-intervention mothers reported minimal changes in these domains. Both intervention health volunteers and non-intervention Women's Union leaders expressed improvements in knowledge, confidence, and relationships with community members. In this study we found that the relative increases in empowerment were greater for mothers than for health volunteers. Intervention mothers reported more sharing of information on child relationships with community members than Women's Union leaders. The increased information sharing has positive implications for spread of key messages to families that did not directly participate in intensive feeding and the sustainability of the intervention's impact. Future research should focus on developing culturally specific concepts of empowerment to better understand the effects of empowerment efforts. This study's identification of empowerment domains will inform future empowerment studies in Vietnam.

Style of child feeding may be an important determinant of child nutrition and health outcomes. Responsive feeding refers to the level and kind of interaction between caregiver and child that lead to a positive feeding experience, adequate dietary intake, and enhanced developmental opportunities. Responsive feeding behaviors may include active physical help and verbalization during feeding, role-playing, persistence, and positive feeding strategies. The aim of this study was to investigate styles of feeding among Vietnamese children 12 or 18 months of age from a rural province in northern Viet Nam. Forty child/mother pairs were videotaped during two, two-hour feeding episodes. Caregiver and child behaviors were coded at the level of the “intended bite” as observed through the videotape analysis of feeding episodes to assess caregiver behavior and the child's interest and acceptance of food. We found it feasible to use videotape and the modified coding and analysis scheme, originally developed for work in Peru, in Viet Nam. In Viet Nam, caregivers provided physical help to eat nearly all of the time in the younger children, and about 70% of the time among 18 month olds. Caregivers verbalized during only 30% of intended bites, and only half of these verbalizations were responsive in tone or words. Positive caregiver behaviors were significantly associated with higher child's acceptance of food, while non-responsive feeding behaviors were associated with child rejection of food. Future analyses of this data set will evaluate the degree to which an integrated nutrition program positively modified caretaker behaviors. More research is needed to demonstrate the relationships among the promotion of responsive feeding behaviors, acceptance of food, and improved nutrition and health status of children.

This study assessed barriers to exclusive breastfeeding in rural Viet Nam and identified how a few mothers were able to exclusively breastfeed despite barriers. A cross-sectional quantitative and qualitative assessment was carried out among 120 mothers of infants less than six months old in northern Viet Nam. Only 24% of the mothers exclusively breastfed. Adjusting for infant's age and who attended delivery, the risk of not exclusively breastfeeding was 14.0 times greater for women who had returned to work than for women who had not. Exclusively breastfeeding mothers (n = 4) who worked differed from other mothers in important ways. They all felt they had enough milk, all knew the appropriate time to introduce foods and liquids, and most were supported in their breastfeeding decisions by commune health workers and family members. This research suggests strategies that can be implemented now to increase exclusive breastfeeding in rural work environments. These include improving knowledge about the introduction of water and semi-solids, addressing perceptions of milk insufficiency, securing support from others, and presenting mothers with options for exclusively breastfeeding, even when they work outside the home.

A positive deviance (PD) inquiry identifies uncommon, model practices that a follow-on program can spread. PD has been used to rehabilitate malnourished children, but not for improving newborn health. Save the Children Federation/US (SC) conducted newborn PD cycles in communities (total population about 5,000 each) in two project areas in Haripur District, Pakistan among Afghan refugees and among local Pakistanis. Each PD cycle included planning, community orientation, situation analysis, PD inquiries, and community feedback with action planning. PD inquiries were in-depth interviews to identify uncommon behaviors among surviving asphyxiated newborns, thriving low birthweight babies, surviving newborns who had danger signs, and normal newborns. The Afghan caregivers showed better use of services and some household practices than their Pakistani counterparts, consistent with duration of SC presence (15 years vs. 18 months, respectively). The practices of both groups for clean delivery, thermal control, immediate and exclusive breastfeeding, and fathers’ involvement were weak. But PD individuals, families, and/or birth attendants modeled good maternal care and immediate, routine and special newborn care. Communities enthusiastically committed to change behavior and form neighborhood support groups for better newborn care, including a demand for hygienic delivery. The PD approach for the newborn is more complex than for child nutrition. Yet this pilot-test proposed a conceptual framework for household newborn care, suggested tools and methods for information gathering, identified PDs in two settings of different risk, galvanized SC staff to the potential of the approach, mobilized communities for better newborn health, and drafted a newborn PD training curricula.

Globally, the caregiving behaviors that contribute to good nutritional status are well understood; but it is not clear why some caregivers perform these behaviors while others do not. This formative qualitative research was designed to improve understanding about what distinguishes caregivers who practice optimal behaviors from those who do not. This study is a one-time, cross-sectional baseline assessment of factors that affect nutrition-related behavior change. It took place in a rural northern province in Viet Nam. One hundred caregivers of children 6 to 17.9 months of age from five communes were interviewed. None of the five communes were included in the larger prospective study designed to test the impact of the community empowerment and nutrition program (CENP). Four behaviors were examined: feeding the child “positive deviant” foods, feeding the child during diarrheal episodes, washing the child's hands, and taking the child to the health center when ill. Results indicate that for all four behaviors, favorable social norms distinguished those who practiced each behavior from those who did not. Positive, reinforcing beliefs and attitudes were important determinants of every behavior except handwashing. Likewise, self-efficacy differentiated doers from non-doers for all behaviors except feeding during diarrheal episodes. Findings from this research suggest that fathers and in-laws of non-doers are more likely to fail to advise mothers about infant feeding and health than they are to provide negative advice. By discovering what distinguishes those who practice optimal behaviors from those who do not, researchers, program planners, and others are better equipped to develop targeted interventions that lead to positive behavior change.

The positive deviance (PD) approach offers an alternative to needs-based approaches for development. The “traditional” application of the PD approach for childhood malnutrition involves studying children who grow well despite adversity, identifying uncommon, model practices among PD families, and designing an intervention to transfer these behaviors to the mothers of malnourished children. A common intervention for child malnutrition, the so-called “hearth,” brings mothers together to practice new feeding and caring behaviors under the encouragement of a village volunteer. Hearths probably work because they modify unmeasured behavioral determinants and unmonitored behaviors, which, in turn, result in better child growth. Some health outcomes require a better understanding of behavioral determinants and are not best served by hearth-like facilitated group skills-building. We propose testing “booster PD inquiries” during implementation to confirm behavioral determinants and efficiently focus interventions. We share early experience with the PD approach for HIV/AIDS and food security. The attributable benefit of the PD approach within a program has not been quantified, but we suspect that it is a catalyst that accelerates change through the processes of community attention getting, awareness raising, problem-solving, motivating for behavior change, advocacy, and actual adopting new behaviors. Program-learners should consider identifying and explicitly attempting to modify the determinants of critical behavior(s), even if the desired outcome is a change in health status that depends on multiple behaviors; measure and maintain program quality, especially at scale; and creatively expand and test additional roles for PD within a given program.