Video Abstract

Video Abstract

Close modal
OBJECTIVES:

Poor early childhood development in low- and middle-income countries is a major public health problem. Efficacy trials have shown the potential of early childhood development interventions but scaling up is costly and challenging. Guidance on effective interventions’ delivery is needed. In an open-label cluster-randomized control trial, we compared the effectiveness of weekly home visits and weekly mother-child group sessions. Both included nutritional education, whose effectiveness was tested separately.

METHODS:

In Odisha, India, 192 villages were randomly assigned to control, nutritional education, nutritional education and home visiting, or nutritional education and group sessions. Mothers with children aged 7 to 16 months were enrolled (n = 1449). Trained local women ran the two-year interventions, which comprised demonstrations and interactions and targeted improved play and nutrition. Primary outcomes, measured at baseline, midline (12 months), and endline (24 months), were child cognition, language, motor development, growth and morbidity.

RESULTS:

Home visiting and group sessions had similar positive average (intention-to-treat) impacts on cognition (home visiting: 0.324 SD, 95% confidence interval [CI]: 0.152 to 0.496, P = .001; group sessions: 0.281 SD, 95% CI: 0.100 to 0.463, P = .007) and language (home visiting: 0.239 SD, 95% CI: 0.072 to 0.407, P = .009; group sessions: 0.302 SD, 95% CI: 0.136 to 0.468, P = .001). Most benefits occurred in the first year. Nutrition-education had no benefit. There were no consistent effects on any other primary outcomes.

CONCLUSIONS:

Group sessions cost $38 per child per year and were as effective on average as home visiting, which cost $135, implying an increase by a factor of 3.5 in the returns to investment with group sessions, offering a more scalable model. Impacts materialize in the first year, having important design implications.

What’s Known on This Subject:

In low- and middle-income countries millions of young children have poor development. Efficacy trials show that stimulation and nutritional interventions benefit children’s development but evidence on cost-effective methods of going to scale is urgently needed.

What This Study Adds:

Mother-child group sessions were as effective on average as home visits in improving child cognition and language in Odisha, India. Groups required 28% of the cost of home visits, substantially improving scalability of child psychosocial interventions.

Poverty, malnutrition, and poor stimulation are preventing millions of young children in low- and middle-income countries from reaching their developmental potential.1  These disadvantages affect brain architecture and function with lifelong consequences.2,3 

Efficacy trials of psychosocial stimulation in early childhood improve disadvantaged children’s development,4,5  with limited evidence of sustained benefits into adulthood6,7  International agencies and professionals have called for these interventions to be scaled up.4,8,9  However, existing evidence says little about the costs and challenges of scaling up these interventions. Furthermore, even the most effective programs can fade out or not replicate.10,11  Trials of home visits,12,13  mother-child groups,14,15  clinic visits,16,17  and mixtures thereof18  all show some success in improving development but have used different outcomes, curricula, duration, and intervention intensities, making it difficult to evaluate the relative cost-effectiveness of alternative deliveries.

Previously,19  we adapted the Reach-Up and Learn13  home visiting intervention to the Odisha, India context. This had moderate impacts on children’s development but was labor-intensive and costly. Groups may reduce the cost because several mothers and children can participate in a session simultaneously. It is unclear, however, whether they can deliver the same impacts. Implementation is more challenging because facilitators must engage with children of different developmental levels and relate to several mothers at the same time, and mothers may have difficulty attending. However, mothers and children may learn more because of interactions and synergies within the peer group. We compare the effectiveness of the same early childhood development (ECD) intervention delivered to mothers and children in groups or through individual home visits. Given the poor nutritional status of children in this context, we included a nutritional education component. We collected midline data to evaluate progress over time. The combination of using local women with group delivery offers a potentially low-cost and genuinely scalable model. Hence, testing its effectiveness is critical.

We conducted an open-label, cluster randomized control trial in 3 districts: Cuttack, Salepur, and Bolangir. A total of 192 villages (clusters) were stratified by district and randomized into 4 groups: control, nutritional education, and combined nutritional education with ECD interventions delivered either through individual home visits or through mother–child group sessions.

The study was implemented by Pratham Education Foundation (Pratham). The Reach-Up and Learn13  curriculum was adapted to the local context by the Ambedkar University Centre for Early Childhood Education and Development (CECED). Data were collected by 2 independent organizations, Abdul Latif Jameel Poverty Action Laboratory and Morsel Research and Development, with training and support by study investigators. Measurements new to India were piloted and adapted to the context. Data were collected at baseline and 12 and 24 months later.

The study protocol was approved by the Institute for Financial Management and Research, India (IRB00007107, FWA00014616, IORG0005894), Yale University Human Subjects Committee (1112009492), University College London Ethics Committee (2168/002), Indian Council of Medical Research (5/7/822/2012/RCH), University of Pennsylvania Office of Regulatory Affairs (815027 IRB no. 8), and the Pratham Education Foundation FWA for the Protection of Human Subjects (FWA00019832). Trial registration numbers are ISRTN:18811205 and AEA RCT Registry: 0000958.

The sampling frame consisted of 300 villages in which Pratham had recently worked. On the basis of official sources and primary data, villages with <6 expected children of target ages, where facilities were shared, or scheduled for relocation were dropped, leaving 192 study villages. Children were identified through prebaseline household censuses and were deemed eligible if they were singletons, aged 7 to 16 months by the beginning of the intervention, and had no obvious disability. This age range was chosen because pregnant women often go away for their deliveries and most return home by 7 months.

In villages with ≤8 eligible children, all were approached to participate. In villages with >8 eligible children, clusters of children who lived within 0.7 km of each other were identified, and 1 child per cluster was randomly selected by using a random number generator in Stata-13 (Stata Corp, College Station, TX). This child and the 7 nearest neighbors were selected; remaining children were placed on reserve lists. Whenever possible, selected children who were unavailable for baseline were replaced by reserve-list children.

Participants were recruited between August 31 and December 19, 2015, and written or oral consent (depending on literacy) was obtained from both household heads and primary caregivers by survey staff. Households were asked again for consent before each survey round. Informed consent to participate in the intervention was collected from all households approached at baseline.

All interventions were conducted weekly for 24 months, beginning December 2015. Home visits lasted ∼60 minutes, group meetings lasted 90 minutes, and nutritional education visits 40 minutes. The nutritional education arm was delivered individually and was focused on improving the quality of children’s diets and basic hygienic practices in households through games, stories, and cooking demonstrations. Home visiting and group sessions were both focused on psychosocial stimulation and included some nutritional education content. Stimulation was based on the Reach-Up and Learn home visiting program,13  which has a structured home visiting curriculum, previously adapted to the local context.19 Facilitators showed mothers how to play and interact with and respond to their children in ways likely to promote development. They demonstrated play activities and encouraged mothers to participate by using toys made from locally available materials and purpose-designed books. Mothers were given the play materials to use at home and then exchanged weekly. The same toys were used ∼3 times but not repeated within <5 weeks, so they remained reasonably novel. Exchanging toys limits the materials required, reducing costs.

The home visiting curriculum was adapted for groups of 7 to 8 children. All mothers and children performed the same activities at the start of sessions, such as free play, singing, review of previous week’s activities, and child-rearing discussions. The groups were then divided into 2 by children’s ages for specific age-related play activities.

Pratham recruited 141 female facilitators from local communities. Average age was 25 years, 40% had bachelor’s degrees, 55% had completed secondary schooling, and 5% had not completed secondary school. Initial training lasted 3 weeks, followed by 3 short refresher trainings spread over the intervention. Facilitators were trained and supervised weekly by 28 mentors with social science degrees and experience working with children (see Supplemental Information for intervention details).

For ethical considerations, all study participants were given a Health and Nutrition Service Link (HNSL) service (Supplemental Information), informing them of the Integrated Child Development Services relating to health and nutrition.

The primary outcomes were children’s development, growth, and morbidity.

At baseline, because of time constraints, children were assessed by using an adapted version of the Ages and Stages Questionnaire, third edition (ASQ-3).20  At 12 and 24 months after baseline, children’s cognitive, language, and motor development were measured with the Bayley Scales of Infant and Toddler Development, third edition (Bayley-III)21  at centers accompanied by their caregivers. Testers had tertiary education or equivalent experience working with children and were trained for 8 weeks. Interobserver reliabilities were assessed before and during the study, and intraclass correlations ranged from 0.82 to 0.99 (n = 205).

Children were measured at home in all rounds using World Health Organization (WHO) guidelines22 : lengths or heights were measured with an infantometer or a stadiometer, respectively, and weights were measured with the Seca 876. Morbidity was assessed at midline and endline through mothers’ reports on occurrences of diarrhea, fever, and cough in the previous 2 weeks by using WHO definitions.23 

Children’s socioemotional development was assessed at endline by maternal reports by using the Strengths and Difficulties Questionnaire.24  Test items were combined into externalizing, internalizing, and prosocial scores.

Secondary outcomes were measured at all rounds in homes. Home-environment quality was assessed by using the UNICEF Family Care Indicators (FCI),25  including a play-materials scale, recording presence of certain types of toys and books, and a play-activities scale, including caregiver involvement with children in various play activities in the last 3 days. Selected items from the Responsivity and Involvement subscales of the Infant-Toddler Home Observation for the Measurement of the Environment (IT-HOME)26  were included at endline. The Knowledge of Infant Development27  was used to assess knowledge of child-rearing practices. All instruments measuring outcomes were translated into Odiya and extensively piloted.

Assuming 10% attrition and intracluster correlations up to 0.3, a target sample size of 1440 children was chosen (7.5 per village or cluster). This yielded power of 80%, with minimum detectable effects of 0.19 to 0.32 SDs on primary outcomes, and a minimum detectable difference in effects of 0.32 SD between treatment arms.

A total of 1449 children were initially selected from all 192 villages, out of which 1243 were successfully interviewed at baseline. Another 158 were successfully replaced (from reserve lists), leaving a final baseline sample of 1401 children.

Randomization of clusters to each arm was completed before baseline by a researcher not otherwise involved with the study using a random number generator in Stata-13.

Participants and intervention staff could not be masked to treatment status. Testers and enumerators were blind to treatment status, although enumerators could have potentially inferred treatment status from toys in intervened households.

We calculated composite scores for the Bayley-III scales. We present intention-to-treat (ITT) differences between each intervention group and the control, measured in control group SD. SEs were clustered at the village level, and two-sided P values were calculated by using t tests adjusting for multiple hypotheses testing by the Romano–Wolf stepdown procedure.28  Outcomes in the same panel of each table were tested jointly as a group. For all analyses, we controlled for covariates (baseline ASQ-3 scores, maternal education, first-born, child’s age and sex) to improve precision. Missing covariates were replaced by sample means of the same variables. For robustness, see Supplemental Table 4.

No clusters were lost during the study. A total of 1331 (92%) children were reinterviewed at midline and 1298 (90%) at endline. Analyses were conducted on children with completed endline testing with a total of 1298 for the Bayley-III and 1275 for anthropometric outcomes (Fig 1). Losses were balanced across groups.

FIGURE 1

Participant flow diagram showing clusters and children from initial screening to endline analysis. Boxes display number of children included in the analysis in each wave, along with the reasons for attrition between waves. Baseline loss numbers refer to those children not replaced.

FIGURE 1

Participant flow diagram showing clusters and children from initial screening to endline analysis. Boxes display number of children included in the analysis in each wave, along with the reasons for attrition between waves. Baseline loss numbers refer to those children not replaced.

Close modal

Table 1 shows baseline characteristics across all 4 groups: between 12% and 14% of children were stunted, and 58% to 64% of mothers had attained at least primary education. Access to clean water and good hygiene was limited, with 44% to 46% of households owning a toilet and 62% to 68% reporting purification of drinking water. Only the ASQ-3 communication score was significantly different across groups; we control for this in all analyses.

TABLE 1

Sample Characteristics at Baseline

ControlNutritional EducationHome VisitingGroup Sessions
Child characteristics     
 Male, n (%) 167 (52) 171 (54) 166 (52) 145 (44) 
 Age, mo, mean (SD) 12 (2.70) 12 (2.68) 12 (2.61) 12 (2.69) 
  HAZ, mean (SD) −0.69 (1.27) −0.67 (1.20) −0.71 (1.25) −0.59 (1.19) 
 Stunted at baseline (HAZ <−2), n (%) 39 (12) 36 (12) 45 (14) 48 (14) 
 No. siblings, mean (SD) 0.70 (0.82) 0.73 (0.84) 0.72 (0.85) 0.69 (1.04) 
 Mother is primary caregiver, n (%) 295 (92) 289 (90) 297 (92) 301 (90) 
 Primary caregiver has at least primary education, n (%) 204 (64) 207 (64) 202 (62) 195 (58) 
ASQ-3 scores, mean (SD)     
 Communication 82.33 (21.07) 86.23 (19.29) 84.78 (18.19) 88.34 (19.32) 
 Gross motor 85.84 (26.77) 88.33 (25.12) 88.34 (24.32) 88.28 (26.44) 
 Fine motor 93.14 (18.95) 95.09 (17.80) 95.29 (16.46) 95.50 (16.90) 
 Problem solving 92.88 (21.16) 96.16 (21.58) 96.33 (19.72) 95.21 (18.94) 
 Personal social 77.42 (21.67) 80.54 (20.23) 80.54 (20.39) 80.78 (19.58) 
Household characteristics     
 Primary caregiver Raven’s raw score, mean (SD) 11.88 (5.28) 12.48 (5.63) 12.12 (4.88) 11.71 (5.17) 
 Has access to electricity, n (%) 293 (92) 289 (90) 284 (88) 304 (92) 
 Owns a toilet, n (%) 146 (46) 146 (46) 145 (44) 154 (46) 
 Household purifies drinking water, n (%) 201 (64) 219 (68) 201 (62) 227 (68) 
Home environment and knowledge of child development scores, mean (SD)     
 FCI play materials (0–7) 2.70 (1.61) 2.84 (1.64) 2.9 (1.56) 2.58 (1.51) 
 FCI play activities (0–7) 2.53 (1.38) 2.78 (1.35) 2.74 (1.52) 2.56 (1.37) 
 Knowledge index (18–54) 38.38 (4.84) 38.70 (5.20) 38.67 (4.39) 39.05 (4.72) 
ControlNutritional EducationHome VisitingGroup Sessions
Child characteristics     
 Male, n (%) 167 (52) 171 (54) 166 (52) 145 (44) 
 Age, mo, mean (SD) 12 (2.70) 12 (2.68) 12 (2.61) 12 (2.69) 
  HAZ, mean (SD) −0.69 (1.27) −0.67 (1.20) −0.71 (1.25) −0.59 (1.19) 
 Stunted at baseline (HAZ <−2), n (%) 39 (12) 36 (12) 45 (14) 48 (14) 
 No. siblings, mean (SD) 0.70 (0.82) 0.73 (0.84) 0.72 (0.85) 0.69 (1.04) 
 Mother is primary caregiver, n (%) 295 (92) 289 (90) 297 (92) 301 (90) 
 Primary caregiver has at least primary education, n (%) 204 (64) 207 (64) 202 (62) 195 (58) 
ASQ-3 scores, mean (SD)     
 Communication 82.33 (21.07) 86.23 (19.29) 84.78 (18.19) 88.34 (19.32) 
 Gross motor 85.84 (26.77) 88.33 (25.12) 88.34 (24.32) 88.28 (26.44) 
 Fine motor 93.14 (18.95) 95.09 (17.80) 95.29 (16.46) 95.50 (16.90) 
 Problem solving 92.88 (21.16) 96.16 (21.58) 96.33 (19.72) 95.21 (18.94) 
 Personal social 77.42 (21.67) 80.54 (20.23) 80.54 (20.39) 80.78 (19.58) 
Household characteristics     
 Primary caregiver Raven’s raw score, mean (SD) 11.88 (5.28) 12.48 (5.63) 12.12 (4.88) 11.71 (5.17) 
 Has access to electricity, n (%) 293 (92) 289 (90) 284 (88) 304 (92) 
 Owns a toilet, n (%) 146 (46) 146 (46) 145 (44) 154 (46) 
 Household purifies drinking water, n (%) 201 (64) 219 (68) 201 (62) 227 (68) 
Home environment and knowledge of child development scores, mean (SD)     
 FCI play materials (0–7) 2.70 (1.61) 2.84 (1.64) 2.9 (1.56) 2.58 (1.51) 
 FCI play activities (0–7) 2.53 (1.38) 2.78 (1.35) 2.74 (1.52) 2.56 (1.37) 
 Knowledge index (18–54) 38.38 (4.84) 38.70 (5.20) 38.67 (4.39) 39.05 (4.72) 

HAZ, height-for-age z score.

During the trial, of caregivers interviewed at baseline, 11.2% in the nutritional education arm, 14.1% in the home visiting arm, and 25.5% in the group sessions arm (group sessions versus home visiting P < .05) stopped participating. However, data collection and measurement continued to include them. The average number of sessions attended varied by arm. Of children in the endline analysis sample, attendance rates were 84% in nutritional education, 75% in home visiting, and 51% in the group sessions arm (group sessions versus home visiting P < .05). The most common reason for missing sessions was temporary migration for the home visiting and nutritional education arms and mothers being too busy for the group sessions arm.

The Bayley-III scores were generally lower than the test reference population, with smaller SDs (Supplemental Table 4). Cronbach alphas were ≥0.8 for all subscales, and discriminant validity and stability over time were adequate (Supplemental Information).

Over the first year, the home visiting group improved significantly in cognition but not language, whereas the group sessions improved significantly in both cognition and language (Table 2). Over 2 years, home visiting had 0.324 SD impact on cognition (95% confidence interval [CI]: 0.152 to 0.496; stepdown P = .001) and a 0.239 SD on language (95% CI: 0.072 to 0.407; stepdown P = .009). Group sessions had a 0.281 SD impact on cognition (95% CI: 0.100 to 0.463; stepdown P = .007) and a 0.302 SD on language (95% CI: 0.136 to 0.468; stepdown P = .001). Differences in average impacts between home visiting and group sessions over 2 years were not significant. For both cognition and language, there was no significant difference between the impact at midline and at endline for either home visiting or group sessions. Neither treatment arm affected motor skills.

TABLE 2

Treatment Effect on Bayley-III Composite Score and the Strengths and Difficulties Questionnaire

Point EstimateCI: Lower BoundCI: Upper BoundP (Unadjusted)P (Adjusted)n
Panel 1: Bayley-III composite scores baseline to midline       
 Nutritional education       
  Cognitive −0.144 −0.320 0.033 .082 .359 1270 
  Language 0.037 −0.144 0.217 .613 .956 1273 
  Motor −0.036 −0.223 0.150 .683 .956 1269 
 Group sessions and nutritional education       
  Cognitive 0.298 0.104 0.493 .004 .018 1270 
  Language 0.313 0.115 0.510 .004 .006 1273 
  Motor 0.011 −0.210 0.232 .900 .956 1269 
 Home visiting and nutritional education       
  Cognitive 0.313 0.132 0.494 .002 .006 1270 
  Language 0.156 −0.053 0.365 .090 .359 1273 
  Motor −0.058 −0.262 0.146 .611 .956 1269 
Panel 2: Bayley-III composite scores baseline to endline       
 Nutritional education       
  Cognitive 0.037 −0.145 0.220 .673 .709 1298 
  Language 0.127 −0.038 0.292 .103 .342 1298 
  Motor 0.075 −0.087 0.238 .354 .657 1298 
 Group sessions and nutritional education       
  Cognitive 0.281 0.100 0.463 .002 .007 1298 
  Language 0.302 0.136 0.468 <.001 .001 1298 
  Motor 0.144 −0.036 0.324 .110 .342 1298 
 Home visiting and nutritional education       
  Cognitive 0.324 0.152 0.496 <.001 .001 1298 
  Language 0.239 0.072 0.407 .001 .009 1298 
  Motor 0.055 −0.108 0.218 .475 .709 1298 
Panel 3: Strengths and Difficulties Questionnaire       
 Nutritional education       
  Externalizing 0.010 −0.157 0.177 .900 .991 1297 
  Internalizing −0.058 −0.209 0.094 .438 .949 1297 
  Prosocial 0.073 −0.082 0.229 .328 .903 1297 
 Group sessions and nutritional education       
  Externalizing −0.039 −0.211 0.134 .651 .981 1297 
  Internalizing −0.008 −0.151 0.134 .908 .991 1297 
  Prosocial 0.179 0.021 0.336 .018 .126 1297 
 Home visiting and nutritional education       
  Externalizing 0.043 −0.117 0.204 .583 .981 1297 
  Internalizing −0.025 −0.172 0.122 .750 .981 1297 
  Prosocial 0.215 0.057 0.373 .008 .054 1297 
Point EstimateCI: Lower BoundCI: Upper BoundP (Unadjusted)P (Adjusted)n
Panel 1: Bayley-III composite scores baseline to midline       
 Nutritional education       
  Cognitive −0.144 −0.320 0.033 .082 .359 1270 
  Language 0.037 −0.144 0.217 .613 .956 1273 
  Motor −0.036 −0.223 0.150 .683 .956 1269 
 Group sessions and nutritional education       
  Cognitive 0.298 0.104 0.493 .004 .018 1270 
  Language 0.313 0.115 0.510 .004 .006 1273 
  Motor 0.011 −0.210 0.232 .900 .956 1269 
 Home visiting and nutritional education       
  Cognitive 0.313 0.132 0.494 .002 .006 1270 
  Language 0.156 −0.053 0.365 .090 .359 1273 
  Motor −0.058 −0.262 0.146 .611 .956 1269 
Panel 2: Bayley-III composite scores baseline to endline       
 Nutritional education       
  Cognitive 0.037 −0.145 0.220 .673 .709 1298 
  Language 0.127 −0.038 0.292 .103 .342 1298 
  Motor 0.075 −0.087 0.238 .354 .657 1298 
 Group sessions and nutritional education       
  Cognitive 0.281 0.100 0.463 .002 .007 1298 
  Language 0.302 0.136 0.468 <.001 .001 1298 
  Motor 0.144 −0.036 0.324 .110 .342 1298 
 Home visiting and nutritional education       
  Cognitive 0.324 0.152 0.496 <.001 .001 1298 
  Language 0.239 0.072 0.407 .001 .009 1298 
  Motor 0.055 −0.108 0.218 .475 .709 1298 
Panel 3: Strengths and Difficulties Questionnaire       
 Nutritional education       
  Externalizing 0.010 −0.157 0.177 .900 .991 1297 
  Internalizing −0.058 −0.209 0.094 .438 .949 1297 
  Prosocial 0.073 −0.082 0.229 .328 .903 1297 
 Group sessions and nutritional education       
  Externalizing −0.039 −0.211 0.134 .651 .981 1297 
  Internalizing −0.008 −0.151 0.134 .908 .991 1297 
  Prosocial 0.179 0.021 0.336 .018 .126 1297 
 Home visiting and nutritional education       
  Externalizing 0.043 −0.117 0.204 .583 .981 1297 
  Internalizing −0.025 −0.172 0.122 .750 .981 1297 
  Prosocial 0.215 0.057 0.373 .008 .054 1297 

Estimated coefficients are expressed in SDs of the control group. Sample size includes all target children who completed the relevant subscales of the Bayley-III at endline and midline for panel 1, and at endline for panel 2, and all mothers who completed the Strengths and Difficulties Questionnaire (at endline) in panel 3. Estimates in all panels controlled for baseline ASQ-3 problem solving, communication, fine motor, gross motor, and personal-social scales, as well as child sex, child age in days, maternal education, and parity with a dummy for first child. Estimated coefficients are expressed in SDs of the control group. The P values are two-tailed P values calculated using the T-bootstrap method (5000 replications) to account for the finite sample and for maximum comparability with the adjusted P values. Adjusted P values are two-tailed P values corrected for multiple hypotheses testing by using the Romano–Wolf stepdown procedure (5000 replications). CIs were constructed by using conventional critical values for individual hypotheses. The multiple hypotheses testing was applied within panels. All replications were done within district used in the randomization protocol and corrected for cluster effects at the village level.

The nutritional education arm had no significant impact on any of the Bayley-III outcomes. There were no significant differences among study arms in weight-for-age or height-for-age z scores (Supplemental Table 6). For morbidity (Supplemental Table 7), the nutritional education arm had lower fever prevalence at 12 months (95% CI: −0.190 to −0.045; stepdown P = .008), and group sessions had a significant reduction in fever at 24 months (95% CI: −0.168 to −0.027; stepdown P = .031). There were no other significant treatment effects. These results are relative to mean reported occurrences of 2.3% diarrhea, 46.7% coughing, and 31.4% fever over 2 weeks in the control group at endline.

In Table 2 panel 3, we report no impacts on Strengths and Difficulties Questionnaire internalizing and externalizing problems. The home visiting arm had a marginally significant 0.215 SD impact on prosocial skills (95% CI: 0.057 to 0.373; stepdown P = .054).

There was no significant effect on knowledge of child development in any of the treatments once adjustment for multiple testing was done (Table 3). The FCI play activities significantly increased in both home visiting and group sessions (home visiting: 0.383 SD; 95% CI: 0.242 to 0.524; stepdown P < .001, group sessions: 0.331 SD; 95% CI: 0.195 to 0.468; stepdown P < .001). Impacts were not significant on FCI play materials once intervention materials were excluded; the group sessions arm had a 0.221 SD effect on the IT-HOME Involvement scale (95% CI: 0.071 to 0.371; stepdown P = .034). Neither home visiting nor group sessions had impacts on the IT-HOME Responsivity scale once P values were adjusted for multiple testing.

TABLE 3

Treatment Effects on FCI, IT-HOME, and Maternal Knowledge on Child Development

Point EstimateCI: Lower BoundCI: Upper BoundP (Unadjusted)P (Adjusted)n
Nutritional education       
 FCI variety materials 0.130 −0.009 0.270 .045 .249 1298 
 FCI variety activities 0.301 0.162 0.440 <.001 <.001 1298 
 IT-HOME responsivity 0.129 0.002 0.256 .037 .228 1298 
 IT-HOME involvement 0.132 −0.029 0.293 .100 .437 1298 
 Maternal knowledge on child development −0.005 −0.082 0.072 .862 .862 1298 
Group sessions and nutritional education       
 FCI variety materials 0.066 −0.078 0.210 .324 .668 1298 
 FCI variety activities 0.331 0.195 0.468 <.001 <.001 1298 
 IT-HOME responsivity 0.144 0.028 0.260 .014 .109 1298 
 IT-HOME involvement 0.221 0.071 0.371 .003 .034 1298 
 Maternal knowledge on child development 0.035 −0.044 0.114 .254 .648 1298 
Home visiting and nutritional education       
 FCI variety materials 0.055 −0.088 0.198 .427 .668 1298 
 FCI variety activities 0.383 0.242 0.524 <.001 <.001 1298 
 IT-HOME responsivity 0.145 0.037 0.254 .006 .064 1298 
 IT-HOME involvement 0.099 −0.058 0.257 .199 .620 1298 
 Maternal knowledge on child development 0.075 −0.003 0.152 .013 .109 1298 
Point EstimateCI: Lower BoundCI: Upper BoundP (Unadjusted)P (Adjusted)n
Nutritional education       
 FCI variety materials 0.130 −0.009 0.270 .045 .249 1298 
 FCI variety activities 0.301 0.162 0.440 <.001 <.001 1298 
 IT-HOME responsivity 0.129 0.002 0.256 .037 .228 1298 
 IT-HOME involvement 0.132 −0.029 0.293 .100 .437 1298 
 Maternal knowledge on child development −0.005 −0.082 0.072 .862 .862 1298 
Group sessions and nutritional education       
 FCI variety materials 0.066 −0.078 0.210 .324 .668 1298 
 FCI variety activities 0.331 0.195 0.468 <.001 <.001 1298 
 IT-HOME responsivity 0.144 0.028 0.260 .014 .109 1298 
 IT-HOME involvement 0.221 0.071 0.371 .003 .034 1298 
 Maternal knowledge on child development 0.035 −0.044 0.114 .254 .648 1298 
Home visiting and nutritional education       
 FCI variety materials 0.055 −0.088 0.198 .427 .668 1298 
 FCI variety activities 0.383 0.242 0.524 <.001 <.001 1298 
 IT-HOME responsivity 0.145 0.037 0.254 .006 .064 1298 
 IT-HOME involvement 0.099 −0.058 0.257 .199 .620 1298 
 Maternal knowledge on child development 0.075 −0.003 0.152 .013 .109 1298 

Estimated coefficients are expressed in SDs of the control group. Estimates in all panels controlled for baseline ASQ-3 problem solving, communication, fine motor, gross motor, and personal-social scales, as well as child sex, age in days, maternal education, and parity with a dummy for first child. Estimated coefficients are expressed in SDs of the control group. The P values are two-tailed P values calculated by using the T-bootstrap method (5000 replications) to account for the finite sample and for maximum comparability with the adjusted P values. Adjusted P values are two-tailed P values corrected for multiple hypotheses testing by using the Romano–Wolf stepdown procedure (5000 replications) for all outcomes together. CIs were constructed by using conventional critical values for individual hypotheses. All replications are done within strata defined by the preprogram variables used in the randomization protocol and corrected for cluster effects at the village level.

Allowing for all overhead, training, materials, and personnel salaries, home visiting cost $135 per child per year assuming each home visitor performs 15 visits weekly. Group sessions cost $38 per child per year assuming each facilitator runs 8 groups per week with 8 children per group (see Supplemental Information for details).

Delivering the same ECD intervention via home visiting or group sessions had similar ITT impacts on children’s cognitive and language development. This is the first study to our knowledge comparing different ECD intervention delivery systems with a randomized design. Group sessions cost only 28% of home visiting in this context, with equivalent average effectiveness. At $38 per child per year, the group sessions model is genuinely scalable. Pratham is established nationally with capacity to select local women, train them, monitor fidelity, and run the program.

When designing Reach-Up and Learn, it was assumed that home visiting would facilitate teaching by targeting individual children’s proximal zone of development, help integrate play into everyday activities, and assist home visitors’ relationship with mothers.13  It is striking that, despite lower compliance for group sessions (attendance was 51% compared with 75% for home visiting), we observed similar ITT. We hypothesize that mothers and children in groups learned skills through observing others and isolated mothers received support by interacting with them. We also hypothesize that group sessions aided cultural acceptance of the promoted child stimulation and rearing practices. It remains possible that the type of intervention delivery may affect the sustainability of benefits, and longer-term follow-up is essential to answer this question.

The group model should be replicable in other poor rural Indian contexts and probably in other Asian countries with similar cultural contexts, where groups are acceptable, making it relevant for large populations. For example, in a recent Bangladeshi study, a similar ECD intervention16  was more effective when delivered to pairs of mothers and children than found in previous Bangladeshi studies when delivered individually; however, no direct comparison within the same experimental setup was attempted, in contrast to this study.

A disadvantage of group sessions compared with home visiting was that more mothers refused to enroll, dropped out, or attended poorly. There was concern that children who attended less were more disadvantaged, but their initial height-for-age, ASQ-3 scores, and home backgrounds were not worse. Implementation research10  is necessary at both community and program levels to determine how to increase attendance. The greatest dropout of group children (11.8%) occurred in the first 4 weeks; we need to improve outreach and understand the causes of lower attendance and adapt the program delivery appropriately.

Two years is longer than many interventions, and nearly all developmental improvements took place in the first year. In the second year, advantages were maintained, although group participation declined. The limited available data13,18,29  indicate that smaller improvements usually occur in the second year. This suggests a model with less intensive intervention in the second year or limiting it to one year. However, the effects of duration on sustainability of benefits need to be examined first.

The nutritional education intervention, worryingly, had no impact on growth. Stunting often increases over this age range in deprived contexts,30  and interventions may need to begin earlier to prevent it. It is estimated that if all known effective nutritional interventions were implemented together, only 20% of stunting would be reduced.31  Successful programs reducing stunting in early childhood have usually taken multisector approaches.32  Household food security was low in this population (14.6% of households reported at least 1 household member skipping meals in the past week), and nutritional supplementation may have been required. A further problem was poor sanitation, with less than half of households owning toilets, adding to risks of diarrhea or enteropathy.

The study strengths are its rigorous design, use of well-established developmental tests with midpoint evaluations, and use of a feasible delivery strategy, conducted by a national nongovernmental organization with readily available staff.

Study limitations are that morbidity occurrence was measured only on 2 occasions, so small effects may have been missed. The Bayley-III was not standardized for India; however, it appears valid in this population with acceptable levels of internal reliability, discriminant validity, and stability over time. Our minimum detectable effect between treatment arms (0.32 SD) was too large to detect differences between treatments; however, differences in impacts were minimal and all <0.06 SD. Despite the large size of our study, involving 1400 children, challenges not testable here may arise at scale.

In conclusion, nutritional education alone is unlikely to solve the problem of malnutrition in the context of food insecurity and poor hygiene, and more comprehensive programs may be required. The finding that a much cheaper ECD intervention delivered to groups of mother-child dyads had similar average impacts as individual home visiting has crucially important implications and should facilitate increasing coverage with limited resources in rural India and, if replicated elsewhere, in other countries.

Dr Grantham-McGregor conceptualized and designed the study, designed the initial curriculum and adapted it to this study’s setting, contributed to the design of the data collection instruments and to the interpretation of the data analysis, and drafted the initial manuscript; Ms Adya, Ms Day, Ms Kochar, Ms Makkar supported the adaptation of the curriculum and intervention to the setting, trained and supervised staff for implementation and data collection, and critically reviewed the manuscript for important intellectual content; Mr Attanasio, Ms Augsburg, Mr Behrman, Mr Meghir conceptualized and designed the study, contributed to the design and adaptation of the curriculum and the design of the data collection instruments, conducted analysis and interpretation of the data, and contributed to the drafting of the manuscript; Ms Caeyers contributed to the design of the data collection instruments and interpretation of the data analysis and critically reviewed the manuscript; Ms Jervis supported the adaptation of the curriculum and intervention to the setting, trained and mentored implementation staff, contributed to the design of the data collection instruments, conducted analysis and interpretation of data, and drafted the initial manuscript; Mr Phimister contributed to the design of the data collection instruments, supported training of data collection staff, conducted analysis and interpretation of data, and drafted the initial manuscript; Ms Rubio-Codina conceptualized and designed the study, supported the adaptation of the curriculum and intervention to the setting, trained and mentored implementation staff, contributed to the design of the data collection instruments, conducted analysis and interpretation of data, and critically reviewed the manuscript for important intellectual content; Ms Vats supported the adaptation of the intervention, trained and managed implementation staff, and critically reviewed the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

This trial has been registered with the ISRTN Register (http://isrtn.org) (identifier 18811205) and the AER RCT Registry (0000958).

Deidentified individual participant data (including data dictionaries) will be made available, in addition to study protocols and the informed consent form. The data will be made available through the World Bank Open Data Depository within a year after publication.

Dr Caeyers' current affiliation is the Norwegian School of Economics.

FUNDING: Funded and supported by the National Institutes of Health under grant R01 HD 72120; the Strategic Intervention Evaluation Fund, World Bank; the Economic and Social Research Council; the Cowles Foundation at Yale University; the Population Studies Center at the University of Pennsylvania; and the Center for Research in Inclusive Education, Chile under grant PIA ANID 160009. This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement No 695300 - HKADeC - ERC-2015-AdG/ERC-2015-AdG). The study funders reviewed and approved the preliminary study design, but had no further role in study design, data collection, data analysis, data interpretation or the writing of this article. Any opinions, findings, and recommendations are the authors and do not necessarily reflect the views of any of the institutions they represent, including those of the IDB, its Board of Directors or the countries they represent. All authors declare no competing interests. Funded by the National Institutes of Health (NIH).

     
  • ASQ-3

    Ages and Stages Questionnaire, third edition

  •  
  • Bayley-III

    Bayley Scales of Infant and Toddler Development, third edition

  •  
  • CECED

    Centre for Early Childhood Education and Development

  •  
  • CI

    confidence interval

  •  
  • ECD

    Early Child Development

  •  
  • FCI

    Family Care Indicators

  •  
  • HNSL

    Health and Nutrition Service Link

  •  
  • IT-HOME

    Infant-Toddler Home Observation for the Measurement of the Environment

  •  
  • ITT

    intention-to-treat

  •  
  • WHO

    World Health Organization

1
Lu
C
,
Black
MM
,
Richter
LM
.
Risk of poor development in young children in low-income and middle-income countries: an estimation and analysis at the global, regional, and country level
.
Lancet Glob Health
.
2016
;
4
(
12
):
e916
e922
2
Johnson
SB
,
Riis
JL
,
Noble
KG
.
State of the art review: poverty and the developing brain
.
Pediatrics
.
2016
;
137
(
4
):
e20153075
3
Black
MM
,
Walker
SP
,
Fernald
LCH
, et al.;
Lancet Early Childhood Development Series Steering Committee
.
Early childhood development coming of age: science through the life course
.
Lancet
.
2017
;
389
(
10064
):
77
90
4
Britto
PR
,
Lye
SJ
,
Proulx
K
, et al.;
Early Childhood Development Interventions Review Group, for the Lancet Early Childhood Development Series Steering Committee
.
Nurturing care: promoting early childhood development
.
Lancet
.
2017
;
389
(
10064
):
91
102
5
Aboud
FE
,
Yousafzai
AK
.
Global health and development in early childhood
.
Annu Rev Psychol
.
2015
;
66
:
433
457
6
Walker
SP
,
Chang
SM
,
Vera-Hernández
M
,
Grantham-McGregor
S
.
Early childhood stimulation benefits adult competence and reduces violent behavior
.
Pediatrics
.
2011
;
127
(
5
):
849
857
7
Tanner
JC
,
Candland
T
,
Odden
WS
.
Later Impacts of Early Childhood Interventions: A Systematic Review
.
Washington, DC
:
Independent Evaluation Group, World Bank Group
;
2015
. Available at: https://ieg.worldbankgroup.org/sites/default/files/Data/reports/chapters/ecd-later-outcomes_overview-introduction.pdf. Accessed June 10, 2020
8
World Health Organization
.
Nurturing care for early childhood development: a framework for helping children survive and thrive to transform health and human potential.
Available at: https://apps.who.int/iris/bitstream/handle/10665/272603/9789241514064-eng.pdf?ua=1. Accessed May 1, 2020
9
Richter
LM
,
Daelmans
B
,
Lombardi
J
, et al.;
Paper 3 Working Group and the Lancet Early Childhood Development Series Steering Committee
.
Investing in the foundation of sustainable development: pathways to scale up for early childhood development
.
Lancet
.
2017
;
389
(
10064
):
103
118
10
Tomlinson
M
,
Hunt
X
,
Rotheram-Borus
MJ
.
Diffusing and scaling evidence-based interventions: eight lessons for early child development from the implementation of perinatal home visiting in South Africa
.
Ann N Y Acad Sci
.
2018
;
1419
(
1
):
218
229
11
Duncan
GJ
,
Magnuson
K
.
Investing in preschool programs
.
J Econ Perspect
.
2013
;
27
(
2
):
109
132
12
Vazir
S
,
Engle
P
,
Balakrishna
N
, et al
.
Cluster-randomized trial on complementary and responsive feeding education to caregivers found improved dietary intake, growth and development among rural Indian toddlers
.
Matern Child Nutr
.
2013
;
9
(
1
):
99
117
13
Grantham-McGregor
S
,
Smith
JA
.
Extending the Jamaican early childhood development intervention
.
J Appl Res Child
.
2016
;
7
(
2
):
4
14
Fernald
LC
,
Kagawa
RM
,
Knauer
HA
,
Schnaas
L
,
Guerra
AG
,
Neufeld
LM
.
Promoting child development through group-based parent support within a cash transfer program: experimental effects on children’s outcomes
.
Dev Psychol
.
2017
;
53
(
2
):
222
236
15
Singla
DR
,
Kumbakumba
E
,
Aboud
FE
.
Effects of a parenting intervention to address maternal psychological wellbeing and child development and growth in rural Uganda: a community-based, cluster randomised trial
.
Lancet Glob Health
.
2015
;
3
(
8
):
e458
e469
16
Hamadani
JD
,
Mehrin
SF
,
Tofail
F
, et al
.
Integrating an early childhood development programme into Bangladeshi primary health-care services: an open-label, cluster-randomised controlled trial
.
Lancet Glob Health
.
2019
;
7
(
3
):
e366
e375
17
Chang
SM
,
Grantham-McGregor
SM
,
Powell
CA
, et al
.
Integrating a parenting intervention with routine primary health care: a cluster randomized trial
.
Pediatrics
.
2015
;
136
(
2
):
272
280
18
Yousafzai
AK
,
Rasheed
MA
,
Rizvi
A
,
Armstrong
R
,
Bhutta
ZA
.
Effect of integrated responsive stimulation and nutrition interventions in the Lady Health Worker programme in Pakistan on child development, growth, and health outcomes: a cluster-randomised factorial effectiveness trial
.
Lancet
.
2014
;
384
(
9950
):
1282
1293
19
Andrew
A
,
Attanasio
O
,
Augsburg
B
, et al
.
Effects of a scalable home-visiting intervention on child development in slums of urban India: evidence from a randomised controlled trial
.
J Child Psychol Psychiatry
.
2020
;
61
(
6
):
644
652
20
Bricker
D
,
Squires
J
,
Mounts
L
, et al
.
Ages and Stages Questionnaire
.
Baltimore
:
Paul H. Brookes
;
1999
21
Bayley
N
.
Bayley Scales of Infant and Toddler Development
, 3rd ed.
San Antonio
:
Psych Corp, Pearson
;
2006
22
World Health Organization
.
WHO child growth standards.
2006
. Available at: https://www.who.int/childgrowth/standards/en/. Accessed October 13, 2019
23
Levine
GA
,
Walson
JL
,
Atlas
HE
,
Lamberti
LM
,
Pavlinac
PB
.
Defining pediatric diarrhea in low-resource settings
.
J Pediatric Infect Dis Soc
.
2017
;
6
(
3
):
289
293
24
Goodman
R
.
The Strengths and Difficulties Questionnaire: a research note
.
J Child Psychol Psychiatry
.
1997
;
38
(
5
):
581
586
25
Kariger
P
,
Frongillo
EA
,
Engle
P
,
Britto
PM
,
Sywulka
SM
,
Menon
P
.
Indicators of family care for development for use in multicountry surveys
.
J Health Popul Nutr
.
2012
;
30
(
4
):
472
486
26
Caldwell
BM
,
Bradley
RH
.
Home Observation for Measurement of the Environment: Administration Manual
.
Tempe, AZ
:
Family & Human Dynamics Research Institute, Arizona State University
;
2016
27
MacPhee
D
.
Knowledge of Infant Development Inventory
.
Princeton, NJ
:
Educational Testing Service
;
1981
28
Romano
JP
,
Wolf
M
.
Exact and approximate stepdown methods for multiple hypothesis testing
.
J Am Stat Assoc
.
2005
;
100
:
94
108
29
McKay
H
,
Sinisterra
L
,
McKay
A
,
Gomez
H
,
Lloreda
P
.
Improving cognitive ability in chronically deprived children
.
Science
.
1978
;
200
(
4339
):
270
278
30
Victora
CG
,
de Onis
M
,
Hallal
PC
,
Blössner
M
,
Shrimpton
R
.
Worldwide timing of growth faltering: revisiting implications for interventions
.
Pediatrics
.
2010
;
125
(
3
). Available at: www.pediatrics.org/cgi/content/full/125/3/e473
31
Bhutta
ZA
,
Das
JK
,
Rizvi
A
, et al.;
Lancet Nutrition Interventions Review Group, the Maternal and Child Nutrition Study Group
.
Evidence-based interventions for improvement of maternal and child nutrition: what can be done and at what cost?
Lancet
.
2013
;
382
(
9890
):
452
477
32
Hossain
M
,
Choudhury
N
,
Adib Binte Abdullah
K
, et al
.
Evidence-based approaches to childhood stunting in low and middle income countries: a systematic review
.
Arch Dis Child
.
2017
;
102
(
10
):
903
909

Competing Interests

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

This is an open access article distributed under the terms of the (Creative Commons Attribution License), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Supplementary data