Six-Year Follow-up of Childhood Stimulation on Development of Children With and Without Anemia

The initial study was a cluster randomized controlled trial of psychosocial stimulation provided to children with IDA (serum transferrin receptor ≥5.0 mg/L and hemoglobin >80 and <110 g/L) and children who are NA and iron sufficient aged 6 to 24 months conducted in 30 villages. Eligible children were screened for anemia and iron status, and an average of 7 children with IDA were selected from each village. Children with obvious disabilities (8 children) and twins (2 pairs) were excluded from the evaluation. Children with severe anemia were referred for treatment. Within each village, every child with anemia was matched with a child without anemia for age (±6 months) and sex. Where more children were available, they were selected randomly. The villages were then randomized to stimulation or control. The resulting sample was 117 treated and 108 control children with anemia and 107 treated and 102 control children without anemia (Fig 1).

At enrollment, the mean (SD) age of the children was 15 (5) months. There were 219 (50.5%) girls. The NA group had higher mean (SD) height-for-age z score −1.75 (1.25) than the IDA group −2.15 (1.10). The NA group’s mothers were better educated and had a higher BMI and family care indicator scores than those in the IDA group.

All children with or without anemia in the treatment villages received weekly play sessions at home for 9 months, using a curriculum designed in Jamaica³⁶  and adapted for Bangladesh.²⁶  All children with anemia in both stimulation and no stimulation groups received 30 mg of iron syrup daily for 6 months delivered to their home weekly. The children were assessed on the Bayley Scales of Infant Development, Second Edition, before and after the intervention, and blood samples were taken for iron status analysis. All children in the IDA group, except 1, improved in hemoglobin concentration (at least 10 g/L). Psychosocial stimulation improved the Mental Development Index (MDI) of the NA group but not the IDA group, and the interaction between anemic group and stimulation approached significance. The children with anemia had lower motor development scores and were less comfortable meeting the tester than the children without anemia at end line. There were no other anemia group differences, once imbalances in socioeconomic backgrounds were controlled. Further study details are available.²⁹ 

The children were relocated by field organizers, recruited from the community. They invited mother–child dyads to a nearby place to assess the children’s performance and interview the mothers or caregivers.

Full-Scale IQ (FSIQ) was assessed with the Wechsler Abbreviated Scale of Intelligence (WASI)-II.³⁷  The WASI was developed and standardized in the United States, and was adapted for this population to be culturally appropriate, but it was not standardized for Bangladesh. Therefore, the scores should not be compared with children elsewhere and were only used to assess group differences. The adapted WASI-II has been used previously in Bangladesh.³⁸  The vocabulary and similarity subscales were used to measure the children’s Verbal Comprehension Index (VCI), and the block design and matrix reasoning subscales were used to measure the Performance Reasoning Index (PRI). The VCI and PRI together formed the FSIQ.

Motor development was assessed with the Movement Assessment Battery for Children-2 (age-band 2 for 7–10 years).³⁹  This test has also been used in Bangladesh previously.³⁸ 

School achievement was measured using a locally developed tool based on the Wide Range Achievement Test.⁴⁰  The test included reading, spelling, and mathematics, and was administered individually. The test was previously used in Bangladesh with primary school children.^38,41 

The quality of the home environment was assessed using the Middle Childhood inventory of Home Observation for Measurement of Environment.⁴²  The questionnaire was translated and adapted for use in Bangladeshi children.

Children’s anthropometry, height, weight, mid-upper arm circumference, and occipitofrontal circumference were measured by the assessors using World Health Organization standard procedure.⁴³ 

Hemoglobin status of the children was assessed from venous blood using cyanmethemoglobin.

Sociodemographic information was collected by the assessors through interviewing mothers or caregivers. The information included the number of household members; age of mother; family structure; parental education and occupation; the quality of roof, floor, and walls of the house; and availability of electricity in the residence.

Five assessors (4 females, 1 male) having at least a bachelor’s degree assessed the child development outcomes and interviewed the caregivers for demographic details and the home environment. The assessors were recruited from outside the study area and conducted assessments rotating through intervention and control clusters, and were blind to the children’s group assignment. A psychologist with 10 years’ experience in child development assessment trained the assessors for 1 month on the assessment tools and questionnaires. The assessors began testing when they achieved >90% agreement with the trainer in each test.

The protocol was approved by the Research and Ethical Review Committees of the institutional review board of icddr,b, and written consent was obtained from the parents of the children.

To examine differences of the children’s background characteristics by groups, we used 2-way analysis of variance. We conducted univariate analysis to determine the differences among the 4 groups (NA stimulation [NA-stim] versus NA no stimulation [NA-no stim], IDA stimulation [IDA-stim] versus IDA no stimulation [IDA-no stim]). Similarly, univariate analysis was also conducted to assess differences between IDA versus NA groups on outcome variables at follow-up. We also used t test to check if the lost and tested children at follow-up were different on any of the measured variables.

The enrollment MDI and follow-up IQ and its subscales were internally standardized using the SD of the population. Intention-to-treat analyses were used to detect all treatment effects. We performed multiple linear regression analysis to assess the effects of stimulation on all primary outcomes. Covariates for the models were selected if the background characteristics were different by treatment groups on enrollment. Inverse probability-weighted estimation was used to correct for possible bias because of loss to follow-up. These weights were used in all multiple regression models of the impact of intervention. We included enrollment MDI, the anemia group (NA/IDA), an interaction term between stimulation and anemia groups in the regression model, and adjusted for testers’ effect and cluster design. We also conducted multiple regression analysis for IDA and NA groups separately. Statistical significance was set at P < .05. All data were analyzed using Stata (version 13.0).

Out of 412 targeted children who completed the original study, 372 (90%) were assessed at this follow-up. Reasons for loss to follow-up were migration (n = 34) and deaths (n = 5) because of drowning, dog bite, and diarrhea. One child could not be tested because of a behavioral problem and was dropped. This resulted in a final sample of 102 (92.7%) children in the IDA-stim group, 94 (88.7%) in the IDA-no stim group, 92 (88.5%) in the NA-stim group, and 84 (91.3%) in the NA-no stim group (Fig 1).

There were no significant differences in the background characteristics of the tested and lost children at enrollment (Supplemental Table 5).

The mean (SD) ages of the children of NA and IDA groups were 106.5 (7.34) and 106.0 (6.6) months, respectively. There were more girls in the NA group compared with the IDA group. The mothers of the NA group had higher education and home stimulation scores than the mothers of the IDA group. The NA group still had higher hemoglobin than the IDA group, but their height-for-age z scores were no longer different. (Table 1).

Table 2 presents the mean scores of primary outcomes by 4 groups using univariate analysis. There were small differences between the NA and IDA groups in all IQ scales and school achievement outcomes except in motor development (data not shown).

The NA-stim group had higher FSIQ (P = .03) and PRI (P < .02) scores than the NA-no stim group, but there were no significant differences between IDA-stim and IDA-no stim groups. There were no differences in the mean scores of IQ, school achievement, and motor development in the anemia groups.

On assessing the treatment effects on IQ with multiple regression analysis with all 4 groups included, there was a significant stimulation effect on FSIQ (B, 0.29; 95% confidence interval [CI] 0.04 to 0.54) and PRI (B, 0.32 CI 0.07 to 0.57), but no difference between the anemia groups. However, there was a significant interaction between the stimulation and anemia groups in FSIQ and VCI (Table 3).

We therefore analyzed the anemia groups separately. In the NA group, there were significant stimulation effects on FSIQ (B, 0.43; CI, 0.08 to 0.79) and PRI (B, 0.48; CI, 0.08 to 0.89), but not on VCI. In the IDA group, there were no significant treatment effects (Table 4).

There were no significant treatment effects in either the NA or IDA group in reading, spelling, mathematics, or Movement Assessment Battery for Children (Supplemental Table 6).

Children who were either with or without anemia in early childhood and participated in a cluster randomized trial of psychosocial stimulation intervention²⁹  were followed up 6 years after intervention ended, at age 8 to 9 years. Benefits had been found to the NA group’s mental development immediately after intervention and benefits to the FSIQ and the PRI were present at follow-up but there was no effect on motor development or school achievement. It is encouraging that benefits from early intervention to IQ were sustained and the findings add to the scant literature on medium term sustainability of benefits in LMICs. Acquiring literacy and mathematics skills depends to some extent on the quality of teaching, which may not have been of a high standard in these children’s schools.

Conversely, in children with anemia, no benefits were found on development immediately after intervention and none on IQ, motor scores, or school achievement at follow-up. Therefore, the only difference at follow-up between the NA and IDA groups in any of the developmental outcomes was the lack of response to intervention in children with anemia compared with benefits found in children without anemia. Many studies of psychosocial stimulation in early childhood have shown child development benefits immediately after intervention; however, the medium and long-term benefits have rarely been examined, especially in LMICs.³² 

Follow-up of the Jamaican study, which showed comprehensive adult benefits, found no school achievement benefits at age 11 years,⁴⁴  but benefits were apparent at 17 years.⁴⁵  It is therefore possible that improvements in school achievement in the current study may appear at a later age in the NA group.

A few short-term follow-up studies up to 6 years of age in LMICs have found sustained child development benefits: in Pakistan at age 4 years,⁴⁶  in Uganda at age 3 years⁴⁷  and in Jamaica at age 6 years;⁴⁸  whereas, in Colombia at age 4 years, no benefits were found.⁴⁹  There is considerable debate as to whether IDA detrimentally affects young children’s development, although there is a well-established association.⁹  As expected, initially, the IDA group had several disadvantages compared with the NA group, including IDA children being more likely to be stunted and have poorer social backgrounds.²⁹  However, children with anemia had similar scores to the children without anemia on all development outcomes at enrollment once background differences were controlled. At the end of the initial trial, the children’s MDI was similar in both groups, but the IDA group had lower scores on the Psychomotor Development Index compared with the NA group. Most interestingly, they did not benefit from the intervention, unlike the NA group; the significant interaction between treatment and the anemia group at follow-up confirms this finding.

In the original study, children with anemia in both treatment and control groups received iron syrup for 6 months because of ethical concerns. Therefore, it is not possible to know the developmental response to iron. Nevertheless, another study conducted in Bangladeshi children found that iron supplementation alone had a beneficial effect on orientation–engagement and both iron and zinc supplementation promoted motor development.⁵⁰  A systematic review and meta-analysis documented that iron supplementation on children with anemia aged 6 to 24 months improved hemoglobin status, but the developmental benefits were uncertain.¹¹  There are very few studies that followed long-term effects of iron supplementation on cognitive development.

A limitation in the current study is that it cannot help to clarify the controversy over the effects of iron supplementation on child development,¹¹  because the control group was treated with iron because of ethical concerns. However, it was remarkable that there was no difference in developmental outcomes between the anemia groups at follow-up except in their response to stimulation. The findings suggest that children with anemia may be less responsive to improvements in the environment. However, the IDA group was more disadvantaged than the NA group and it is possible that the other unmeasured disadvantages, and not IDA per se, affected their response to intervention.

Another limitation is that, whereas both the groups with anemia received iron treatment, the NA groups did not get a placebo. It is possible that giving iron reduced the visit time spent on child development and may have reduced the intervention’s effectiveness. In addition, the IDA control group also received weekly visits when we delivered the iron syrups, whereas the NA control group did not receive visits. These visits may have increased the attention to the IDA-no stim group. It therefore remains uncertain why the IDA group failed to benefit from stimulation.

The strength of the study was that it comprised 2 parallel cluster randomized trials and both IDA and NA groups had the same home visitors and the same intervention. Furthermore, the background characteristics of the stimulated and control groups were balanced and the attrition was small.

Children who were not anemic and participated in a home visiting psychosocial stimulation intervention in early childhood had sustained benefits on their IQ and perceptual reasoning at 8 to 9 years of age, 6 years after the intervention. The persistence of benefits at 8 to 9 years provide further support for the promotion of child development interventions in rural children. Further follow-up to assess longer-term sustainability would be helpful. In contrast, matched children with IDA showed no benefit. This lack of response should be explored further.

We thank all the families who participated in the study. We also thank the assessors who measured the children’s development and the local field organizers who located the participants. We thank the commitment of Swedish International Development Assistance for its research efforts. We also thank the governments of Bangladesh, Canada, Sweden, and the United Kingdom for providing core/unrestricted support.

FSIQ

Full-Scale IQ

IDA

iron-deficiency anemia

IDA-no stim

iron-deficiency anemia no stimulation

IDA-stim

iron-deficiency anemia stimulation

LMIC

low- and middle-income country

MDI

Mental Development Index

NA

nonanemic

NA-no stim

nonanemic no stimulation

NA-stim

nonanemic stimulation

PRI

Performance Reasoning Index

VCI

Verbal Comprehensive Index

WASI

Wechsler Abbreviated Scale of Intelligence