Children aged 6–24 months with WAZ < −3SD who had recovered from acute infection at the hospital of ICDDR,B were the study participants. We assessed a total of 553 children for eligibility and enrolled 507 children in the randomised trial. Children with weight-for-length Z-score (WLZ) < −3SD, oedema, fever, congenital disorders, diseases affecting growth, no fixed residence or primary caregiver not capable to provide stimulation due to illness were excluded (n = 46). Four community nutrition follow-up units (CNFU) were established in the the residential areas from where most malnourished children attended Dhaka hospital of ICDDR,B.
On discharge and after obtaining parental informed consent, the eligible children (n = 507) were randomly assigned to five groups: PS; FS; PS along with FS (PS + FS); clinic-control (CC) and hospital-control (CH). The CH group received fortnightly follow-up care at the hospital nutrition follow-up unit (HNFU) of ICDDR,B. The other 4 groups received fortnightly follow-up care at the CNFUs. All groups received similar basic care comprising growth monitoring, health education and micronutrient supplementation.
Randomisation was done by a researcher, who was not involved in the study. Four sets of separate randomisations were prepared for each CNFU, using a computer-generated, block-randomisation scheme, with permuted block lengths of 5 and 10.
Mothers/caregivers and their children were followed-up fortnightly at the assigned CNFU or HNFU for the first three months then monthly for the final three months. At each follow-up visit all groups received the following as a part of routine clinical management practiced at Dhaka Hospital of ICDDR,B ;
Children’s weight, length/height, and mid-upper arm and head circumferences were measured according to standard procedures .
Lessons on primary health care were provided by the play leaders (PLs).
At recruitment and during fortnightly follow up visits the children were supplied with multivitamin drops for a month (containing vitamin A, vitamin D, thiamin, riboflavin, pyridoxine, nicotinamide, calcium and ascorbic acid). They were also supplemented with zinc dispersible tablets and iron and folic acid tablets from 2–12 weeks after recruitment.
All study children were immunised according to the Expanded Programme of Immunisation guidelines , and children older than one year were de-wormed with a single dose of albendazole, provided that they had not been treated in the previous six months.
Trained female health workers (PLs) conducted sessions with play activities and parenting education for one hour with every child and mother of the PS and PS + FS groups. This was done at CNFU for six months; fortnightly for the first three months and then monthly for next three months i.e. 9 visits over 6 months.
The PL demonstrated play techniques with low-cost and culturally appropriate homemade toys and focused on helping the mothers to become more effective in teaching their children and enhancing maternal-child interactions. Mothers were encouraged to continue the activities at home. They were lent toys and picture books to take home and changed them at every visit for different toys.
The PL discussed early child development and the importance of mothers being responsive to their infants and playing, chatting, singing, showing love to and praising their child. They demonstrated how to chat and incorporate play into daily activities, such as feeding and bathing, in order to promote child development. The mothers were discouraged from using physical punishment. We also aimed to improve mothers’ self-esteem by praising them, listening to them and providing new skills. The sessions were participatory and mothers/caregivers were encouraged to share their views and suggestions.
Children in the FS and PS + FS groups were supplemented with food packets for the first three months of follow-up, according to the guidelines of the national food supplementation programme of Bangladesh. The packets were distributed at discharge from hospital and at each follow-up visit at the CNFU. One packet per day was provided for children below 12 months of age and two packets per day for older children. Food packets were also provided to siblings younger than five years in order to minimise food sharing. Each packet provided about 150 kcal (~ 630 k joules) of energy, with 11% of the energy derived from protein. It contained 20 g of roasted rice powder, 10 g of roasted lentil powder, 5 g of molasses and 3 g of soybean oil .
At the end of the third and the sixth month of follow-up children had anthropometric and developmental assessments (BSID-II) at the HNFU .
Stimulation at home
The quality of stimulation at home was measured by a modified version of the Infant/Toddler HOME inventory [27, 28]. The modification was made for use in Bangladesh [29–33]. The phrasing of some of the questions was modified to improve relevance considering differences in living conditions in Bangladesh. The modified version of inventory contained 60 items and six sub-scales: i) Organisation of the physical and temporal environment ii) Stimulation (opportunities for variations in daily stimulation at home) iii) Maternal involvement with the child iv) Play materials v) Avoidance of restriction and punishment and vi) Emotional and verbal responsivity of the mother. The total score was calculated by summing up all positive responses.
A research assistant, unaware of the group assignments, visited the children’s homes and administered the HOME at baseline and after 6 months. Before and during the study, 20 interviews were observed by the trainer and the inter-observer reliability between interviewer and trainer was high (r ranged from 0.91 to 0.98).
Research assistants unaware of the group assignment assessed the mother’s child rearing practices at baseline and after 6 months using a modified version of a questionnaire, previously used in Bangladesh . The questions concerned chatting, praising, showing love and affection, how the mothers teached children using different objects or play materials and how they interacted during activities such as feeding, bathing and dressing. We added few items to the parenting questionnaire on setting limits. Total scores were calculated by summing up all positive responses. The test-retest reliability on mothers of 20 children was good (r = 0.92).
Maternal depressive symptoms
At baseline and after 6 months the frequency of maternal depressive symptoms was assessed with a questionnaire based on the modified version of the Centre for Epidemiologic Studies-Depression Scale (CES-D) . This instrument was previously used in Bangladesh [35, 36] and Jamaica [37, 38]. The questionnaire addressed six aspects of depressive symptoms: depressed mood, worthlessness, helplessness/hopelessness, lethargy/fatigue, loss of appetite, and sleep disturbance. The scale was designed to assess the frequency of depressive symptoms expressed in number of days with these symptoms. After piloting, the wording of the questions was adapted to be more culturally appropriate. Mothers were asked to recall how many days in the past week they experienced depressive symptoms and the number of days (0–7) was recorded, and then summed to make a total depression score. Higher scores indicated the presence of more depressive symptoms. Inter-observer reliability between trainer and tester with 20 mothers was high (r = 0.98).
Data were collected on the families’ wealth, standard of housing, family structure and parental characteristics.
The outcomes of this analysis are the quality of home environment (HOME scores) and mothers’ child-rearing practices with the severely malnourished children six months after the start of intervention.
A sample size of 60 in each group was estimated to demonstrate a difference of 5 points in the MDI, based on a previous study in Bangladesh  where the MDI had a SD of 12, a significance level of 5% and power of 90% for five treatment groups. There were 59–77 mother-child pairs available for analysis providing 90% power to demonstrate a difference of 0.4 SD in HOME score between any two groups.
Data were analysed with SPSS version 18 (SPSS Inc, Chicago). Baseline characteristics were examined of the children and their families among the treatment groups, and of those analysed and lost to follow-up, by analysis of variance (ANOVA) for continuous variables and χ2 test for dichotomous variables.
In the analysis of treatment effects across the randomised groups, analysis of covariance (ANCOVA) was used to control for baseline HOME and child rearing practices scores. Adjustment was also made for age at final test and the factors differing among the groups on enrolment and between lost and analysed children, i.e. maternal education and depressive symptoms and father’s occupation.
The proposal was approved by the Institutional Review Board of ICDDR, B. Written informed consent was obtained from mothers on enrolment. Children who did not respond to treatment or had any major illness during the study were treated or referred to health facilities for examination and treatment.