- Susan P Walker, professor ()1,
- Susan M Chang, lecturer1,
- Christine A Powell, senior lecturer1,
- Emily Simonoff, professor2,
- Sally M Grantham-McGregor, professor3
- 1 Epidemiology Research Unit, Tropical Medicine Research Institute, University of the West Indies, Mona, Kingston 7, Jamaica
- 2 Institute of Psychiatry, London SE5 8AF
- 3 Institute of Child Health, London WC1N 1EH
- Correspondence to: S Walker
- Accepted 7 June 2006
Objective To determine whether dietary supplementation or psychosocial stimulation given to growth retarded (stunted) children age 9-24 months has long term benefits for their psychosocial functioning in late adolescence.
Design Sixteen year follow-up study of a randomised controlled trial.
Setting Poor neighbourhoods in Kingston, Jamaica.
Participants Of 129 stunted children identified at age 9-24 months, 103 adolescents aged 17-18 were followed up.
Intervention Supplementation with 1 kg milk based formula each week or psychosocial stimulation (weekly play sessions with mother and child), or both, for two years.
Main outcome measures Anxiety, depression, self esteem, and antisocial behaviour assessed by questionnaires administered by interviewers; attention deficit, hyperactivity, and oppositional behaviour assessed by interviews with parents.
Results Primary analysis indicated that participants who received stimulation had significantly different overall scores from those who did not (F = 2.047, P = 0.049). Supplementation had no significant effect (F = 1.505, P = 0.17). Participants who received stimulation reported less anxiety (mean difference - 2.81, 95% confidence interval - 5.02 to - 0.61), less depression (- 0.43, - 0.78 to - 0.07), and higher self esteem (1.55, 0.08 to 3.02) and parents reported fewer attention problems (- 3.34, - 6.48 to - 0.19). These differences are equivalent to effect sizes of 0.40-0.49 standard deviations.
Conclusions Stimulation in early childhood has sustained benefits to stunted children's emotional outcomes and attention.
Growth retardation or stunting affects 30% of children under 5 years globally and is associated with poor development.12 Limited evidence suggests that cognition and school achievement are affected up to late adolescence,3 but we found no studies of psychosocial functioning in people who were stunted in early childhood.
Stunted children show behavioural changes in early childhood—they are less happy, more apathetic, and more fussy.4 Children who were admitted to hospital for severe malnutrition in early childhood show aggressive behaviour, attention deficits, and poor social relationships when they reach school age.5–7 These findings suggest that stunted children have psychosocial problems.
In high risk children in developed countries, psychosocial stimulation in early childhood reduces antisocial behaviour and delinquency in adolescence.89 Other psychosocial outcomes have rarely been assessed.
We conducted a two year randomised trial of dietary supplementation and psychosocial stimulation in stunted Jamaican children aged 9-24 months. Each treatment benefited development.2 When assessed at 7 and 11 years, stimulation significantly benefited cognitive function.10 Small benefits from supplementation seen at 7 years were not detected at 11 years. Parents of stunted children reported more problems with conduct than parents of non-stunted children at 11 years, and in a subgroup observed at 9-10 years stunted children were more inhibited and less persistent than non-stunted children.1112 We conducted a follow-up study of the participants at 17-18 years to determine any benefits from the interventions in early childhood on their current psychosocial functioning.
In 1986-7 we identified children aged 9-24 months by house to house survey of poor neighbourhoods of Kingston, Jamaica. We assigned all 129 stunted children identified (length for age less than - 2 standard deviations of the National Center for Health Statistics references13) to one of four groups—control, supplementation, stimulation, or both interventions. The order of group assignment was determined randomly. Other enrolment criteria have been reported previously.2 We calculated the sample size to have 80% power at 5% significance to detect a 5 point difference (equivalent to 0.5 standard deviation, as in our population the standard deviation of the test is typically 10 points) in developmental levels from either intervention.
We visited all participants weekly for two years. Supplementation comprised 1 kg milk based formula each week.14 Stimulation comprised weekly one hour visits to the home that focused on enhancing interactions between mother and child. Trained community health workers demonstrated play techniques to the mother and involved her in a play session with her child. Mothers were encouraged to talk to their child and to use praise and positive reinforcement. Physical punishment was discouraged. Home made toys and simple picture books were left with the families, and mothers were encouraged to play with their children between visits and to incorporate play activities into daily routines. One hundred and twenty seven children completed the study. We re-examined the children at 7 and 11 years.
Follow-up at 17-18 years
We assessed the psychosocial functioning of the cohort in 2002-3 when participants were aged 17-18 years. The aim was to assess symptoms and we did not try to diagnose specific disorders. We interviewed 103 participants, 80% of those who started the trial (figure). Written informed consent was obtained from the participants and their parents.
We assessed self esteem (how I think about myself questionnaire), anxiety (what I think and feel questionnaire), depression (short mood and feelings questionnaire), and antisocial behaviour (behaviour and activities checklist).15–18 Questionnaires were administered in a private interview at our research unit. Table 1 shows the number of items in each scale, range of scores, and sample questions.
By age 17, young people are thought to be valid informants for many behaviours, but not attention deficit. Thus, to obtain information on this condition we questioned the mothers or primary carers by using the Conners' parent rating scale (short form).19 This scale also provides scores for cognitive problems and lack of attention, hyperactivity, and oppositional behaviour (such as anger and resentfulness).
We piloted and reworded the questionnaires as necessary so that they were easily understood. Table 1 shows internal consistency (Chronbach's α) and test-retest reliability (intraclass correlation coefficient) after a two week interval. Different interviewers (who were unaware of the participants' group) administered the questionnaires to the participants and parents. Inter-observer agreement with another trained interviewer in 23 interviews was ≥ 97% for all items. Higher scores indicate worse psychosocial functioning except for the self esteem scale, where higher scores indicate better self esteem.
We used a questionnaire to obtain information on education, sexual relationships, pregnancy, contact with the police, and exposure to violence.
We asked participants about the frequency of hunger due to lack of food at home during the previous year. Homes were visited and information obtained on water and toilet facilities (each rated on a 1-6 scale), crowding (number of people per room), and number of household possessions from a list of 11 items. A housing score was derived by factor analysis of these variables. We recorded the mother's or primary carer's education and occupation. We assessed the mother's verbal intelligence with the Peabody picture vocabulary test when the children were 11 years old.
We screened outcome variables for normality and used square root transformations to normalise the depression and antisocial behaviour scales. Effects of the interventions were determined by intention to treat. For the primary analysis, we used multivariate ANOVA to determine whether the interventions had significant overall effects on the outcomes; in secondary analyses, we used multiple regression to examine the effects of interventions on individual outcomes. The interventions were coded as: supplementation: supplemented and both interventions = 1, control and stimulation = 0; stimulation: stimulation and both interventions = 1, control and supplementation = 0.
Loss from study
Most children (75%) were lost from the study because of migration (figure), and the proportion lost was similar in all groups. Enrolment measures were similar in participants and children who were lost except that in the non-stimulated groups children lost from the study had lower weight for height (P = 0.014) and younger mothers (P < 0.001) than those who were assessed. Mother's age was not associated with any of the outcome variables and weight for height was associated with antisocial behaviour only. In analyses of antisocial behaviour we therefore controlled for initial weight for height.
Table 2 shows the characteristics of the participants at enrolment and follow-up. At follow-up, we found no significant differences between the groups. When we combined the groups in which the children received psychosocial stimulation and compared them with those who did not, the stimulated group reported less hunger (P = 0.03). We found no significant differences by intervention status (stimulation or supplementation) in age, proportion of males and females, or any other social background variables.
Effects of treatment
Table 3 shows the psychosocial outcomes by group. Multivariate ANOVA with all outcomes as the dependent variables and supplementation and stimulation as factors indicated a significant effect of stimulation (F = 2.047, P = 0.049) but not supplementation (F = 1.505, P = 0.17). In secondary analyses, we performed regressions for the individual outcomes to determine which outcomes differed between participants who had or had not received stimulation. Because of the factorial design of our study we included dummy variables for stimulation and supplementation, but the results for supplementation are not shown owing to the overall lack of significance in the multivariate analysis. Participants who received psychosocial stimulation in early childhood reported significantly less anxiety, fewer symptoms of depression, and better self esteem than non-stimulated stunted participants (table 4). They were also rated as having fewer attention problems by their parents. If, however, we use a more stringent level of significance (P < 0.01, because of the multiple comparisons) only the difference in reported anxiety is statistically significant. We found no significant interactions between stimulation and supplementation. We entered hunger (which differed by stimulation status) as a factor and re-analysed the outcomes that had been significantly affected by stimulation. Hunger was not a significant predictor of any of the variables.
The groups did not differ regarding contact with the police or law. Apart from being stopped for questioning by the police (reported by 47.6% of participants), few contacts were reported—five participants were detained and one was convicted of an offence. The groups did not differ in sexual behaviour (having had sex, number of partners, having been pregnant) or in use of alcohol, cigarettes, and marijuana. Participants who had received stimulation were less likely to have been suspended from school (stimulation 31.3%, no stimulation 47.3%; P = 0.10) or expelled (stimulation 2.1%, no stimulation 10.9%; P = 0.08) than participants who did not receive stimulation.
Psychosocial stimulation in early childhood had sustained benefits for the psychosocial functioning of stunted children. We found no sustained benefits from dietary supplementation.
In our study, supplementation had small benefits to height during the intervention, and supplementation only increased the net energy intake by 440 kJ/day.14 A higher level or longer duration of supplementation may have led to greater benefits.
Stunted children who participated in the stimulation programme reported fewer symptoms of depression and anxiety and better self esteem and attention than those who did not receive the intervention. The changes in scores for anxiety, depression, and self esteem were 0.4-0.49 of a standard deviation. At an individual level, these effects are not large (although in line with some mental health treatment outcomes). However, at a population level, these changes could have an important role in reducing emotional disorders.
Loss to follow-up was modest, and the interviewers were blind to the participants' group. The only enrolment characteristics that differed between stunted children lost to follow-up and those assessed were weight for height and mother's age. Mother's age was not related to any of the outcomes and weight for height was related only to antisocial behaviour, so the intervention was probably responsible for the benefits seen.
Few long term studies have evaluated interventions in early childhood in disadvantaged but adequately nourished children.20 The main benefit reported in psychosocial functioning is reduction in antisocial behaviour.89 In our study, participants in the stimulated group were suspended or expelled from school less often, but we found no benefits to self reported antisocial behaviour.
We found no previous reports of interventions in early childhood with long term benefits to internalising problems or attention, but these behaviours were rarely assessed. Intervention studies in the United States have shown that children who are more disadvantaged benefit most.8 The Jamaican children studied here were undernourished and came from very poor families, and this may have contributed to the wide ranging benefits of intervention, which include cognitive and educational benefits.21
The intervention was aimed at improving the mother-child relationship; it emphasised listening and chatting to the children, allowing them to experience success, praising their actions, and reducing punishment. Any change in these behaviours could help improve psychosocial functioning, particularly self esteem. In another recent trial with a similar intervention, mothers in the intervention arm had reduced symptoms of depression.22 Thus, improved psychosocial functioning in the mothers may also have been beneficial to the children. The benefits of stimulation to the educational achievement of the participants may also have contributed to better emotional outcomes.
Much attention has been paid to poor cognition and school achievement of undernourished children, but changes in their psychosocial functioning may be as important to their quality of life. Stimulation in early childhood produced sustained improvements to the psychosocial functioning of stunted children. The next challenge is to develop interventions that can meet the needs of the enormous number of stunted children.
What is already known on this topic
Linear growth retardation (stunting) in childhood is associated with cognitive deficits and poor educational achievement in late adolescence
Stunting is also associated with behavioural problems at 9-11 years of age
Psychosocial stimulation in early childhood has long term benefits for stunted children's cognitive outcomes
What this study adds
A programme of home based psychosocial stimulation has sustained benefits for stunted children's emotional outcomes and attention
Thanks to Amika Wright and Sydonnie Shakespeare for conducting the interviews and the participants and their parents for their continued cooperation.
Contributors All authors conceived and designed the study. SC and SW conducted the study. SW and SGM analysed the data and SW drafted the manuscript. All authors critically reviewed the paper and approved the final draft for publication. SW is guarantor.
Funding Supported by a grant from the Wellcome Trust (066088). The funders approved the study design but had no involvement in data collection, analysis, interpretation, or writing the report.
Competing interests None declared.
Ethical approval Ethics committees of the University of the West Indies and the Institute of Child Health, University College, London.