Supplemental Material can be found at: http://jn.nutrition.org/content/suppl/2009/08/20/jn.109.10488 5.DC1.html

The Journal of Nutrition Community and International Nutrition

A Responsive Feeding Intervention Increases Children’s Self-Feeding and Maternal Responsiveness but Not Weight Gain1,2 Frances E. Aboud,3* Sohana Shafique,4 and Sadika Akhter5

Abstract Responsive complementary feeding, whereby the mother feeds her child in response to child cues and psychomotor abilities, is low in some countries and likely contributes to malnutrition. Interventions are needed to evaluate whether promoting responsive feeding would add any benefit. Using a cluster-randomized field trial, we evaluated a 6-session educational program that emphasized the practice of child self-feeding and maternal responsiveness. A total of 108 mothers and their 8- to 20-mo-old children in 19 clusters were randomly assigned to the intervention group and 95 in 18 clusters were assigned to the informational control group. Outcomes were assessed at pretest, postintervention, and follow-up. Research assistants, who were unaware of group assignment, observed and coded mother and child midday meal behaviors. At follow-up, the percent of self-fed mouthfuls was 47.8 6 42.4 (mean 6 SD) in the responsive feeding group children compared with 32.2 6 41.0 in the controls (P = 0.01); likewise, the number of responsive verbalizations was 6.55 6 5.9 in the responsive feeding mothers and 4.62 6 4.5 in controls (P = 0.01). Intervention mothers recalled more messages. Mouthfuls of food eaten by children and weight were equivalent in the 2 groups. Lack of change in foods eaten and small quantities may explain the similarly low levels of weight gain. These results provide evidence that self-feeding and maternal verbal responsiveness, two developmentally important behaviors, can be increased by targeting specific behaviors with appropriate behavior change strategies of modeling and coached practice. Weight gain may require more nutritional input, especially in areas of high food insecurity. J. Nutr. 139: 1738–1743, 2009.

Introduction A recent series of publications has drawn attention to high levels of malnutrition in children under 5 y in developing countries, particularly South Asia where over 40% of this age group is underweight (1). In addition to inadequate quantities and diversity of food, nutritionists increasingly recognize an unresponsive feeding style as a crucial impediment to healthy nutrition. To this end, WHO and UNICEF advocate responsive complementary feeding in their guidelines (2). To date, despite several descriptive studies on the topic (3–5), only 1 published study has demonstrated the effectiveness of an intervention aimed solely at responsive feeding (6). This study pursues this inquiry by evaluating an intervention implemented in a foodinsecure region of Bangladesh to increase self-feeding and maternal responsiveness. Responsive caregiver behavior refers to the caregiver observing, interpreting, and then responding to a child’s cues in a 1

Supported by the Social Sciences and Humanities Research Council of Canada and the United Kingdom’s Department for International Development in Bangladesh. 2 Author disclosures: F. E. Aboud, S. Shafique, and S. Akhter, no conflicts of interest. * To whom correspondence should be addressed. E-mail: frances.aboud@mcgill. ca.

1738

contingent and appropriate manner and in accordance with the intended meaning of the signal (7). In Bangladesh, this style of feeding is challenged in 2 ways (5). One is that mothers tend to feed in a controlling manner and so encounter frequent refusals, which in turn lead to forceful tactics to get the food in. Second, mothers do not let children feed themselves during the second year despite children’s increasing motor capabilities and signals of interest to do so. Several attempts to promote responsive complementary feeding have been combined with messages about providing more nutritious foods in Peru and India, but feeding behaviors were not assessed (8–10). More recently, mothers who attended 6 behavior-change sessions in Bangladesh had children with greater self-feeding and greater weight gain than control mothers who received knowledge-based sessions (6). Maternal responsiveness, however, did not increase. A more careful analysis of the mothers’ responsive behavior indicated that it declined in both groups (6). One explanation was that as children fed themselves more, responsiveness may have become less necessary; mothers had less opportunity to feed in a responsive manner if they were not feeding. However, this interpretation misrepresents the purpose and benefits of responsiveness and children’s continuing need for it, albeit in more distal forms such as words, as they take on responsibility for

0022-3166/08 $8.00 ã 2009 American Society for Nutrition. Manuscript received January 22, 2009. Initial review completed March 16, 2009. Revision accepted June 16, 2009. First published online July 8, 2009; doi:10.3945/jn.109.104885.

Downloaded from jn.nutrition.org at LONDON SCHOOL OF HYGIENE & TROPICAL MEDICINE on February 8, 2012

3 Department of Psychology, McGill University, Montreal H3A 1B1, Canada; 4International Centre for Diarrhoeal Diseases Research, Dhaka 1212, Bangladesh; and 5Plan International, Dhaka 1212, Bangladesh

Participants and Methods The design was a cluster-randomized field trial in which the fixed variable was assignment to the responsive feeding or control group. Clusters were the villages where mothers lived and met in a group to receive either the responsive feeding or regular nutrition education. Ethical approval was granted by the International Centre for Diarrheal Diseases Research, Bangladesh and McGill University. Mothers were informed that they would receive nutrition education and signed consent forms to participate in data collection. The study took place between April and December 2007 and was registered with International Standard Randomized Controlled Trial number 150000469. Participants. The mothers and children resided in the rural subdistrict of Jaldhaka, in the district of Nilphamari, Bangladesh, 650 km north of the capital Dhaka, where a nongovernmental organization regularly delivered parenting programs to mothers of children under 3 y (hereafter called “the organization”). The population served by the organization is mostly Muslim with a sizeable minority of Hindus involved in farming, small business, or wage labor and classified as poor and very poor (15). Malnutrition is higher than the overall current prevalence in Bangladesh (15,16). Eligible clusters consisted of 114 village groups where mothers of children under 3 y had met monthly since September of 2006 to receive lessons on health and nutrition from the organization. From these clusters, children were eligible to participate in the study if they were between the ages of 8 and 20 mo at pretest. Sample size estimation. The sample size was based on the number of verbally responsive acts of the mother. From a previous study that yielded 3.8 6 3.0 (mean 6 SD) verbally responsive acts at the midday meal, we predicted a mean difference of 1.5 (or 0.5 SD). With an a of 0.05 and power of 0.80, the required sample size was calculated to be 80 per group. Intracluster correlation was set low at 0.03 (6). Cluster sizes were expected to be 6. Based on the formula 1+ intracluster correlation(cluster size – 1), the sample size was multiplied by 1.15 to accommodate clustering, thus requiring a sample of 92 per group. Randomization and recruitment. Randomization of village clusters occurred before mothers were recruited to the study.

Using a random numbers table and a numbered list of the villages with parenting groups, we first selected 37 village groups and then randomly assigned 19 as intervention clusters and 18 as control clusters. To avoid contamination, none of the intervention clusters was adjacent to a control cluster. Eight research assistants who were not aware of group assignment visited mothers at home and recruited them into the study during May. They recruited all eligible mothers from the organization’s ongoing health and nutrition program. The research team’s independence from the implementation of sessions was maintained; research assistants were not present in the area when the intervention was being implemented. After follow-up they were still unaware that there were 2 distinct programs. Peer educators implementing the responsive feeding intervention received extra training and knew that they were participating in an atypical program. Mothers’ awareness of different programs was not assessed. Intervention. Interventions were delivered to clusters. The intervention group received 6 sessions in addition to the regular program. As part of the regular program, both control and intervention groups received 12 sessions on child development earlier (September to November, 2006) that discussed how parents could help children learn, provide stimulation through toys and talk, and use gentle discipline (17). Likewise, both groups received 12 monthly information sessions on health and nutrition (18) during which the months of January to May concerned complementary feeding. During June, the intervention group additionally received weekly sessions on responsive feeding. More details of the control and responsive feeding messages follow. The regular program provided 5 information sessions relevant to nutrition (18); control mothers received this information only. Mothers were told about food requirements, nutrients and their functions, growth monitoring, meal frequencies, low appetite, and breast-feeding. Local community health workers gave the information verbally and with the help of a picture book and stories, using what is known as the adult education approach. Although a few mothers brought children, the latter were not actively involved except to be weighed. To help control mothers recall the food messages, in mid-June we gave each one a laminated picture from their program of foods to feed children. The intervention mothers and children received 5 additional sessions on responsive feeding during the month of June and a booster session in early October (after the post-test and 6 wk before the follow-up). Peer educators were young women from the village, normally with grade 8 to 10 education, who had previously delivered the child development sessions and conducted village preschools in the morning. They used our 20-page Responsive Feeding Manual (Supplemental Material), in Bangla, with the following messages that mothers discussed and practiced with children during the sessions: 1) Wash your child’s hands before he/she picks up food; 2) self-feed: let the child pick up food and eat; 3) be responsive: watch, listen, and respond in words to your child’s signals; 4) when your child refuses, pause and question why; don’t force feed or threaten (5 alternative strategies were discussed); and 5) offer a variety of foods, including fish, eggs, fruits, and vegetables. The behavior change strategy, based on components of social-cognitive learning theory, included practice, problemsolving, and peer support (19). To allow participants to practice self-feeding and responsive feeding, peer educators provided 5 or 6 small, soft pieces of fruit, vegetable, and/or boiled egg on a plate. The peer educator demonstrated with 1 child and then coached mothers as they practiced with their own child. Responsive feeding intervention in Bangladesh

1739

Downloaded from jn.nutrition.org at LONDON SCHOOL OF HYGIENE & TROPICAL MEDICINE on February 8, 2012

their own feeding (11–13). Consequently, the present intervention emphasized verbal responsiveness as a way of showing love, encouragement, and 2-way communication (14). Self-feeding was also promoted for this sample of 8- to 20-mo-old rural Bangladeshi children. To assess the benefits of this responsive feeding program, we trained peer educators to implement a 6-session module to groups of rural mothers and their children in Bangladesh and compared them with mothers and children who received only the regular nutrition education. A cluster-randomized field trial was used in which randomization occurred and interventions were delivered at the cluster level (village) and outcomes measured at the individual level. The hypotheses were that, compared with controls, mothers in the responsive feeding intervention would show more verbally responsive behaviors and their children would show more selffeeding and greater weight gain. Secondary objectives included more face-to-face feeding, more hand-washing, and greater recall of messages by intervention mothers.

Measurement of outcomes. Primary outcomes were weight, mouthfuls eaten, self-fed mouthfuls, and mother’s responsive verbal acts. Secondary outcomes were other potentially relevant feeding behaviors, such as child refusals and maternal nonresponsive encouragement, feeding position, hand-washing, foods fed to the child, and messages recalled by the mother. All but the last outcome were measured at 3 points: recruitment (pretest), 2 wk after the sessions ended (post-test), and at a follow-up 5 mo after the sessions ended and 6 wk after the booster. All research assistants collected data from intervention and control clusters. Child and mother behaviors were observed during a midday meal when mothers were most likely to feed the child individually. The time was arranged during an interview in the morning when the mother provided information on the child’s birth date, breast-feeding status, illness in the past week, and foods fed yesterday. Data on sociodemographic status, e.g. mother’s and father’s ages, education, and 11 family assets (e.g. bed, table, radio, bicycle, electricity) were collected. After the interview, the child was weighed twice on an electronic scale to the nearest 50 g (Seca model 881) and the mean value used in analyses. Weight rather than weight-for-age was analyzed, because we could not be sure of children’s exact ages, although results were similar. The behavioral outcomes were measured through observation of a midday meal as follows. The research assistant sat unobtrusively in a position to observe and record all feeding behaviors of the mother and child, along with foods fed. Subsequent coding of each behavioral act is described shortly. To assess its representativeness, when the meal was over, the mother was asked if the child was hungry before the meal, if this was the usual time for the child’s meal, and if it was a typical amount (yes, no for each). Messages recalled by the mothers were assessed at follow-up by asking if they remembered any messages given by the organization. After each, the assistant asked, “What else did they tell you?” Answers were organized according to 10 categories such as foods to feed, hygiene, nutritional disorders, responsive feeding, and self-feeding. Mothers received a score of 1 for each category for which they recalled a message. 1740

Aboud et al.

Behavior assessment training and reliability. Training ensured that research assistants wrote reliably full records of observed feeding behaviors and subsequently coded each behavior reliably. Research assistants were trained in a 1-wk period before conducting the pretest and given refresher training prior to post and follow-up testing. This included observation of 10 videos of mother-child interaction for which written records taken by assistants could be compared with ones done frame-byframe by experienced observers. All reached a minimum of 80% accuracy before the end of training. Observed feeding behaviors were later coded using a responsive feeding framework (5,6), particularly for mouthfuls swallowed, child self-feeding, and mother verbally responding to a preceding child signal in accordance with the intent of the child’s behavior (e.g. child takes the previous mouthful and the mother says “eat more rice”; child refuses and the mother asks, “Do you want water?”; child reaches for food and mother says, “you will eat by yourself”). Coded but not primary outcomes were child refusals, mother’s encouraging but nonresponsive behavior (e.g. after refusal, the mother offers again and says, “Eat, eat"), child’s feeding position (child seated facing or sideways so mother can see child’s face, not lying down or fed from behind), and washing child’s hands with at least water prior to eating. Each meaningful unit of behavior in the transcripts received a code. For example, the written statement, “Child picks up egg and puts in mouth” received a code of self-feeding; if the mother subsequently said, “you eat nicely,” a code of mother verbally responsive was given. Consequently, a frequency count of each behavior code was available for each mother-child pair, reflecting the number of times the behavior occurred. Frequencies are psychologically more meaningful than rates in motherchild research (20), because a child’s memory for an event is directly related to the frequency of its occurrence, not to the rate such as occurrences per minute. Inter-rater reliability on 10 transcripts coded independently by the principal investigator and coders exceeded 90% agreement. Method of analysis. All analyses used the MIXED procedure in SAS for continuous scores, except for foods observed and messages recalled, which required the GLIMMIX procedure for dichotomized scores (version 8.2). Individual child outcomes were analyzed based on intention to treat. The between-subjects fixed variable was group assignment; the random variables were clusters and mothers within clusters. A restricted maximum likelihood fitting method was used. Groups were compared at baseline adjusting for clusters; post-test and follow-up analyses additionally covaried the pretest value of the outcome variable, child’s age and sex, mother’s education, and family assets. Values in the text and tables are raw means 6 SD. Effect sizes for d, or the number of SD separating the 2 means, were calculated at follow-up. Although we expected child’s age to affect weight and behaviors, we did not expect (or find) age to interact with the intervention. Those who were lost to follow-up were compared with those who were analyzed. We used chi square and correlation to verify whether a typical meal was related to sickness and mouthfuls.

Results Ninety-five control mothers and 108 intervention mothers were recruited (Fig. 1). Approximately 5% of the sample was lost to follow-up, 7% of control mothers and 2% of intervention mothers. Observations of 8 intervention sessions during wk 3 and irregular visits by the organization’s supervisors confirmed high

Downloaded from jn.nutrition.org at LONDON SCHOOL OF HYGIENE & TROPICAL MEDICINE on February 8, 2012

Discussions focused on the answers to frequently asked questions and flexible solutions to common problems. Mothers repeated aloud the 5 messages with the help of a picture poster depicting a mother washing her child’s hand as she notices the child’s other hand signaling a desire to self-feed from a plate with diverse foods. In mid-June, to help recall the messages, intervention mothers received the laminated colored picture shown in the poster. In the booster session, 3 mo later, mothers discussed problems enacting the behaviors and practiced feeding their child responsively. The 19 intervention peer educators received group training over a 4-d period. Role plays with infants were particularly important in convincing peer educators that children would eagerly feed themselves while mothers praised. Special emphasis was placed on mothers’ verbal responses to her child’s signals, such as how to talk to children after they took or refused a mouthful. Multiple verbal responses to refusals were outlined, because initial interview data revealed that over onehalf of the mothers responded by simply directing the child to eat (“eat, eat”), diverting the child’s attention (e.g. “look at the bird”), and then slipping food in, forcing food in, and threatening to give food away. Findings from a formative qualitative study conducted in the area were incorporated (Purnima Menon, personal communication, Cornell University, March 2007).

TABLE 1

Baseline sociodemographic, child, and feeding characteristics1

Variable

Control 95 14.02 6 3.7 48 (50.5) 48 (50.5) 90 (94.7) 23.50 6 6.3 4.15 6 3.6 6.16 6 1.9 5.18 6 1.7 7.93 6 1.0 15.10 6 9.9 19.20 6 35.1 4.35 6 4.3

108 13.72 6 3.3 42 (38.9) 71 (65.7)* 103 (95.4) 23.63 6 5.3 3.00 6 3.5 6.19 6 2.3 4.96 6 1.7 8.01 6 1.2 13.77 6 9.7 18.67 6 32.9 3.77 6 3.2

1 Values are means 6 SD or n (%). *Different from control, P = 0.03 (cluster-adjusted ANOVA). 2 Mouthfuls refer to the total number of mouthfuls of food taken in 1 midday meal. 3 The number of mothers’ verbal acts that responded to the meaning or intent of the child’s preceding act.

FIGURE 1 Trial profile. Of those lost, 2 had migrated permanently and 5 mothers had gone to visit their family.

fidelity (.95%) to the Responsive Feeding Manual (Supplemental Material) on 10 critical elements such as demonstration, practice, coaching, feeding position, and mothers’ participation in discussion. Attendance was likewise high (.90%) for all except 2 who never attended. Only a few control sessions were observed; peer educators closely followed the Health and Nutrition Manual (18) except for irregular growth monitoring. Randomization appeared to be largely successful according to an analysis of the pretest variables (Table 1). Only one indicator was significantly different: more intervention children were sick at pretest. The sample was representative of Bangladesh in that 33.7% of control children and 39.8% of intervention children were .2 SD below the median of the WHO child growth standards. The 9 participants missing follow-up did not differ from those retained on any of the pretest variables except household size, for which lost families was smaller (lost: 4.00 6 1.4; retained: 5.11 6 1.7; P = 0.05).

Preliminary descriptive analyses indicated that intraclass correlations for the analyzed outcomes were low, varying between 0.00 and 0.05. Cluster sizes ranged from 3 to 8. The variables weight and self-fed mouthfuls were not skewed, the number of mouthfuls and mother’s verbally responsive behaviors were positively skewed but could be normalized with a square root transformation. Consequently, weight and self-fed mouthfuls were analyzed raw and number of mouthfuls and verbal responses were analyzed in the transformed state. Other behavior variables such as child refusals and nonresponsive encouragement were transformed. Follow-up results are described unless otherwise indicated (Table 2). Children in the 2 groups did not differ in weight, indicating that weight gain was similarly low for both groups with a mean of 0.97 kg gain over the 7 mo from start to finish. Similarly, the number of mouthfuls taken did not differ with an overall mean of 21.4. However, child self-feeding was significantly greater for intervention than control children (d = 0.37). Mothers’ verbal responsiveness was also higher in the intervention group (d = 0.36). Child refusals were significantly lower among intervention children at post-test (d = 0.27) but not at follow-up. Mothers’ nonresponsive encouragement declined similarly in both groups; however, the subcategory of nonresponsive verbal encouragement increased among intervention mothers and exceeded that of control mothers. More intervention mothers sat their child in a position to see his/her face and washed the child’s hands before eating. Mothers thought that their child was hungry (92–99% reported so), that this was the usual time for their child to eat (91–97%), and that it was a typical meal (67–90%) for the 2 groups, which did not differ over the 3 time periods, except at follow-up when intervention mothers reported less-than-typical intake (x2 = 5.93; P = 0.02; control 90% typical, intervention 76%). At 3 time periods, the mothers’ assessment of whether the intake was typical correlated positively with number of mouthfuls (r = 0.16–0.28; P , 0.02), although only at follow-up was typicality negatively related to reports of child illness (x2 = 8.02; P = 0.004). Thus, not surprisingly, sick children may have eaten less food. Only fruit was more often reportedly given to intervention children the previous day (Table 3). At follow-up, of the 7 Responsive feeding intervention in Bangladesh

1741

Downloaded from jn.nutrition.org at LONDON SCHOOL OF HYGIENE & TROPICAL MEDICINE on February 8, 2012

Children, n Child age, mo Girls, n (%) Sick, past week, n (%) Mother housewife, n (%) Mothers age, y Mothers education, y Family assets, n (of 11) Household size, n Childs weight, kg Mouthfuls2, n Self-fed mouthfuls, % of total Mother verbally responsive,3 n

Intervention

TABLE 2

Effect of responsive feeding intervention on nutritional and feeding behavior outcomes1

Outcome variable

Post-test2

Follow-up2 Foods reported by mother

95 108

93 107

88 106

7.93 6 1.0 8.01 6 1.2

8.16 6 0.9 8.21 6 1.1

8.94 6 1.0 8.95 6 1.1

21.73 6 0.9 21.76 6 1.0

21.93 6 0.8 21.93 6 0.9

21.86 6 0.9 21.87 6 0.9

15.10 6 9.9 13.77 6 9.7

20.11 6 16.0 20.78 6 14.0

21.62 6 15.5 21.23 6 14.3

19.20 6 35.1 18.67 6 32.9

36.21 6 41.5 48.53 6 45.2

32.21 6 41.0 47.79 6 42.4**

4.35 6 4.3 3.77 6 3.2

5.61 6 5.3 9.15 6 7.7***

4.62 6 4.5 6.55 6 5.9**

6.34 6 5.5 7.88 6 6.9

6.57 6 5.9 5.11 6 5.2*

4.12 6 4.9 3.36 6 4.2

6.40 6 3.8 7.71 6 6.4

6.11 6 4.0 5.88 6 4.1

4.76 6 3.8 4.69 6 3.4

1.65 6 1.6 1.74 6 1.6

1.54 6 1.9 2.06 6 1.8

1.32 6 1.6 1.90 6 2.0*

35 (36.5) 26 (24.3) 6 (6.4) 2 (1.8)

41 (44.1) 61 (57.5)**

39 (41.0) 63 (58.3)*

31 (33.7) 63 (58.9)

43 (48.9) 65 (61.3)*

Values are raw means 6 SD or n (%). *Different from control, P , 0.05; **P , 0.01; ***P , 0.0001 (cluster-adjusted ANCOVA, covarying pretest levels, child’s age and sex, mother’s education, and family assets). 2 Mouthfuls refer to the total number of mouthfuls of food taken in 1 midday meal. 3 Self-fed mouthfuls refer to child putting food in own mouth as a percent of total mouthfuls. 4 A verbal act of mother that responded to the meaning or intent of the child’s preceding act. 5 Child refusal is a food offering rejected by the child. 6 Nonresponsive encouragement is an act by mother that encouraged the child to eat but did not respond to the child’s preceding act. 7 Nonresponsive verbal encouragement is a verbal act by mother encouraging the child to eat but not responding to the child’s preceding act. 8 Child is seated facing the mother or sideways allowing her to view the child’s face. 9 Child’s hands were washed with water prior to eating (with or without soap).

critical food categories (21), intervention children ate a mean of 3.25 foods, and control children ate 2.93 compared with 2.62 and 2.82 at pretest, respectively. Only 28.4% received $4 food categories. Foods offered at the midday meal were not significantly different and diversity was low. More intervention mothers recalled messages at follow-up (Table 4). Aboud et al.

Control Children, n Food type Rice Dal Animal (fish) Egg Fruit Vegetables Cow milk2 Carbohydrate3 Biscuit/sugar Oil Dietary diversity

Foods observed at midday meal

Intervention

88 2.78 0.10 1.20 0.15 0.03 2.33 0.32 0.59 1.45

106 times eaten, n 6 0.7 2.67 6 0.8 6 0.4 0.19 6 0.6 6 1.3 1.27 6 1.1 6 0.4 0.21 6 0.5 6 0.2 0.25 6 0.6*** 6 1.7 1.91 6 1.4 6 0.7 0.41 6 0.9 6 0.8 0.38 6 0.8 6 1.1 1.55 6 1.3

2.93 6 0.92

3.25 6 0.99

Control

Intervention

88

106 n (%)

86 (97.7) 4 (5.7) 39 (44.3) 6 (6.8) 1 (1.1) 59 (67.0) 4 (4.5) 16 (18.2) 3 (3.4) 76 (86.4) 2.26 6 0.69

104 (98.1) 7 (6.6) 39 (36.8) 10 (9.4) 1 (0.9) 67 (63.2) 7 (6.6) 19 (17.9) 10 (9.4) 90 (84.9) 2.22 6 0.68

Values are means 6 SD or n (%). ***Different from control, P = 0.0001 (clusteradjusted ANCOVA). 2 Cow milk was often mixed with rice flakes and salt/sugar. 3 Carbohydrate foods were mainly potato. 1

Discussion

1

1742

Comparison of control and intervention children’s diets yesterday as reported by the mother and foods observed during the midday meal, at follow-up1

The intervention was successful in increasing self-feeding by children and verbal responsiveness by mothers; the responsive feeding position likewise became more prevalent. The behavioral component of the intervention was therefore successful. On closer inspection of the post-test data, in particular hand-washing and self-feeding, it was noted that 5 of the control villages increased from near zero to very high levels, suggestive of contamination from the first message (“Wash your child’s hands before he/she picks up the food”). Although this is unfortunate from a research perspective, it demonstrated that the behaviors being promoted were acceptable to mothers and compatible with cultural practices despite their initially low levels. It was expected that self-feeding and fewer refusals would lead to more mouthfuls in the intervention group. However, the number of mouthfuls eaten by children did not differ. Fortunately, there was no decrement despite mothers’ expectation that TABLE 4

Comparison of control and intervention mothers’ recall of messages at follow-up1

Message recalled

Control

Intervention

Mothers, n Ages to feed Foods to feed How often to feed Hygiene Growth monitoring Nutritional disorders Responding, not forcing Self-feeding Nonresponsive encouragement Nonresponsive stimulation2

88 0 30 (34.1) 6 (6.8) 12 (13.6) 0 2 (2.3) 6 (6.8) 0 (0) 4 (4.5) 2 (2.3)

106 0 83 (78.3)*** 17 (16.0) 84 (80.0)*** 0 1 (1.0) 64 (60.4)*** 44 (41.5)*** 24 (22.6)** 14 (13.2)*

1 Values are n (%). *Different from control, P , 0.05, **P , 0.01, ***P , 0.0001 (cluster-adjusted ANCOVA on mother’s recall of a message (no, yes) controlling for child‘s age and sex, mother’s education, and family assets). 2 Includes telling stories or telling the names of foods without encouraging eating and not in response to a child signal.

Downloaded from jn.nutrition.org at LONDON SCHOOL OF HYGIENE & TROPICAL MEDICINE on February 8, 2012

Children, n Control Intervention Child's attained weight, kg Control Intervention Weight-for-age, Z Control Intervention Child mouthfuls,2 n Control Intervention Self-fed mouthfuls,3 % of total Control Intervention Mother verbally responsive,4 n Control Intervention Child refusals,5 n Control Intervention Nonresponsive encouragement,6 n Control Intervention Nonresponsive verbal encouragement,7 n Control Intervention Responsive feeding position,8 n (%) Control Intervention Hand washing,9 n (%) Control Intervention

Pretest

TABLE 3

Acknowledgments We thank Plan Bangladesh’s Manager, Haider Yaqub, and the program unit manager in Jaldhaka, Syed Rashid, for facilitat-

ing our use of their parenting groups as implementing sites and Md. Mamun Rashed for skillfully training the peer educators. Xun Zhang conducted the statistical analyses.

Literature Cited 1.

2. 3. 4.

5. 6.

7.

8.

9.

10.

11. 12.

13.

14. 15. 16. 17. 18. 19.

20.

21.

22.

23.

Black RE, Allen LH, Bhutta ZA, Caulfield LE, de Onis M, Ezzati M, Mathers C, Rivera J, for the Maternal and Child Undernutrition Study Group. Maternal and child undernutrition: global and regional exposures and health consequences. Lancet. 2008;371:243–60. WHO. Guiding principles for complementary feeding of the breastfed child. Geneva: WHO; 2003. Engle PL, Bentley M, Pelto G. The role of care in nutrition programmes: current research and a research agenda. Proc Nutr Soc. 2000;59:25–35. Ha PB, Bentley ME, Pachon H, Sripaipan T, Caulfield LE, Marsh DR, Schroeder DG. Caregiver styles of feeding and child acceptance of food in rural Viet Nam. Food Nutr Bull. 2002;23 Suppl 4:95–100. Moore AC, Akhter S, Aboud FE. Responsive complementary feeding in rural Bangladesh. Soc Sci Med. 2006;62:1917–30. Aboud FE, Moore AC, Akhter S. Effectiveness of a community-based responsive feeding programme in rural Bangladesh: a cluster randomized field trial. Matern Child Nutr. 2008;4:275–86. Eshel N, Daelmans B, Cabral de Mello M, Martines J. Responsive parenting: interventions and outcomes. Bull World Health Organ. 2006;84:991–8. Penny ME, Creed-Kanashiro HM, Robert RC, Narro MR, Caulfield LE, Black RE. Effectiveness of an educational intervention delivered through the health services to improve nutrition in young children: a cluster-randomised controlled trial. Lancet. 2005;365:1863–72. Bhandari N, Bahl R, Nayyar B, Khokhar P, Rohde JE, Bhan MK. Food supplementation with encouragement to feed it to infants from 4 to 12 months of age has a small impact on weight gain. J Nutr. 2001;131: 1946–51. Bhandari N, Mazumder S, Bahl R, Martines J, Black RE, Bhan MK. An educational intervention to promote appropriate complementary feeding improves child feeding practices and linear growth in rural Haryana, India. J Nutr. 2004;134:2342–8. Isabella RA. Origins of attachment: maternal interactive behavior across the first year. Child Dev. 1993;64:605–21. Isabella RA, Belsky J, von Eye A. Origins of infant-mother attachment: an examination of interactional synchrony during the infant’s first year. Dev Psychol. 1989;25:12–21. Tamis-LeMonda CS, Bornstein MH, Baumwell L. Maternal responsiveness and children’s achievement of language milestones. Child Dev. 2001;72:748–67. Tomasello M. Cooperation and communication in the 2nd year of life. Infant Child Dev Persp. 2007;1:8–12. Aboud FE. Evaluation of an early childhood parenting program in rural Bangladesh. J Health Popul Nutr. 2007;25:3–13. UNICEF. Child and mother nutrition survey in Bangladesh, 2005. Dhaka (Bangladesh): UNICEF; 2007. Plan Bangladesh. Parenting manual and picture book. Dhaka (Bangladesh): Plan International; 2004. Plan Bangladesh. Health of mother and child manual. Dhaka (Bangladesh): Plan International; 2005. Baranowski T, Perry CL, Parcel GS. How individuals, environments and health behaviors interact: social cognitive theory. In: Glanz K, Rimer BK, Lewis FM, eds. Health behavior and health education: theory, research, and practice. San Francisco: Jossey-Bass; 2002. p. 165–84. Hubbs-Tait L, McDonald Culp A, Culp RE, Miller CE. Relation of maternal cognitive stimulation, emotional support, and intrusive behavior during Head Start to children’s kindergarten cognitive abilities. Child Dev. 2002;73:110–31. WHO. Indicators for assessing infant and young child feeding practices: conclusions of a consensus meeting held November 6–8 2007, Washington, DC: World Health Organization; 2008. Dewey KG, Adu-Afarwuah S. Systematic review of the efficacy and effectiveness of complementary feeding interventions in developing countries. Matern Child Nutr. 2008;4:24–85. Saha KK, Frongillo EA, Alam DS, Arifeen SE, Persson LA, Rasmussen KM. Household food security is associated with infant feeding practices in rural Bangladesh. J Nutr. 2008;138:1383–90.

Responsive feeding intervention in Bangladesh

1743

Downloaded from jn.nutrition.org at LONDON SCHOOL OF HYGIENE & TROPICAL MEDICINE on February 8, 2012

children who fed themselves would eat less. In fact, there was a strong positive correlation between number of mouthfuls and proportion self-fed (r = +0.37, adjusting for child’s age; P , 0.0001) and a corresponding inverse correlation with motherfeeding. However, compared with other nutrition education (22) and responsive feeding interventions (6), the nutritional benefit was less. Dietary diversity was low, with only 28.4% eating $4 of 7 critical food types the previous day. This is regarded as a food-insecure region of the country because of its low daily wage and seasonal shortages of foods. The low diversity and few mouthfuls indicate insufficient quality and quantity of intake. Changes to behavioral aspects of infant and child feeding may require more than the commonly offered nutrition education sessions. Mothers in the control group received 12 monthly health and nutrition sessions, yet they recalled few of these messages. The approach to nutrition education based on adult education theory may be ineffective when delivered to large groups of people who lack literacy skills, because they cannot retain the information, much less use it to guide their behavior. In contrast, the behavior change theory underlying our intervention emphasized the use of a respected model from the village, who demonstrated the behavior, and then coached mothers as they practiced with their child; the picture served as a reminder cue when home. The findings are important in that they are the second evaluation of a behavior change program explicitly designed to practice self- and responsive feeding. The results point to the importance of household resources (e.g. education, assets, food security) to improve the child’s dietary diversity, which must accompany self-feeding and responsive behaviors if nutritional status is to improve (6,22,23). Intervention mothers recalled messages about diet, hygiene, and responding but did not act on the diet messages, possibly because of limited resources. Limitations of the study include potential contamination of the control group, the delivery of the programs, and measurement. Although we attempted to ensure that intervention and control villages were not adjacent, visiting peer educators and mothers might have passed on messages or shown their picture to control village mothers. Despite potential contamination by the hand-washing and self-feeding message, differences were significant. The responsive feeding message was given after the regular nutrition education; this might explain better recall of the former. Village clusters receiving the 2 programs differed in size; the responsive feeding sessions were attended by 3 to 8 mothers and their children, whereas the control sessions were attended by 10–20 women, only some with children; this was due to our age restriction for responsive feeding. Our behavioral outcomes were measured through observation and records written by trained and reliable assistants. Video recordings might have provided more reliable data but were not used because they aroused too much self-consciousness. Also, although we observed only 1 midday meal, it was reported as typical. In conclusion, the responsive feeding intervention tested here provided significant improvements in children’s self-feeding and mothers’ verbal responsiveness. Behavior changes and message recall indicate that the implemented behavior change strategy may be a useful addition to existing education programs for malnourished children. However, in food-insecure areas, food supplementation may be necessary to translate these behavior changes into weight gain.