Integrated Early Childhood Development in Cambodia: Protocol of a Cluster Stepped-Wedge Trial

The IECD program is a 5-year intervention (2021–2025) that aims to improve children’s development through changes in caregivers’ practices, informed by nurturing care.^7,8  The intervention delivers community, group, and home-based activities that integrate responsive caregiving and early learning, nutrition, health and hygiene, and household stability and support into daily activities.

RTI International and Helen Keller International work in partnership with a local team in Cambodia, led by a program director who sets strategic directions with USAID and the RGC. Memoranda of understanding codify the shared objectives, respective roles, and responsibilities.

RGC representatives participate in material development and technical review, serve as master trainers, and supervise field activities. Ministries of Social Affairs, Veterans, and Youth Rehabilitation; Health; Education, Youth, and Sport; and the Council for Agricultural and Rural Development contribute to the content of IECD nurturing care modules. Training is conducted jointly with the relevant ministry, including national, provincial, and district representatives. Ministries serve as lead partners for technical activities related to their expertise: Ministry of Health for health, hygiene, and nutrition activities; Ministry of Social Affairs, Veterans, and Youth for household stability and support activities; and Ministry of Education, Youth, and Sport for responsive care and early learning activities (Table 1).

The IECD program aims to generate political support for sustained investment in early childhood interventions at the provincial and district levels, and to strengthen commune and village-level early childhood services, in alignment with Cambodia’s decentralized Social Accountability Framework and its National Action Plan for Early Child Care and Development.²¹  At the village level, the IECD program engages existing frontline community health workers (village health volunteers [VHVs]) to deliver IECD activities to families. IECD has established a new role of child development leaders (CDLs), drawn from existing commune councils for women and children, to provide oversight and support to VHVs operating across villages in their communes. The VHVs and CDLs receive training and a modest stipend from the IECD program.

Through the contributions of RGC representatives and program staff, the IECD intervention activities incorporate Cambodian knowledge and expertise related to child development, nutrition and health, and disability screening and support. Program activities align with national strategies (eg, the National Strategy for Food Security and Nutrition).

The IECD program is informed by principles and evidence from nurturing care, including equity (see Supplemental Information), and delivers program components through community-, group-, and home-based platforms. The theory of change targets caregiver behavior in 4 domains (responsive caregiving and early learning; nutrition; health and hygiene, and household stability and support) (Fig 1). Table 2 presents the targeted caregiving domains, including specific behaviors and community-, group-, and home-based activities.

Across communities, IECD implements social behavior change campaigns, including mass media, interpersonal communication, and community activities open to all community members (eg, cooking demonstrations, sanitation campaigns, and gender equality workshops). IECD implements disability and nutrition screening using the locally developed Community-Based Development Milestones Assessment Tool and mid-upper arm circumference, and refers those identified to indicated professional services. IECD supports selected households to serve as agricultural demonstration sites, modeling improved production practices for consumption and income generation. IECD provides village-level support for eligible families to apply for government financial support, including conditional cash transfers.

The IECD program provides 10 biweekly group sessions for caregivers of children aged <5 years and pregnant women, following 10 modules (detailed in Table 3) that integrate content from the program’s 4 targeted caregiving domains. VHVs and CDLs deliver group sessions using IECD-created session scripts, including an introductory activity, recaps from previous sessions, new content featuring 2 to 3 key messages, and opportunities for modeling, practice, and feedback. RGC personnel and IECD staff provide supervision and support to build local capacity.

Home visits are provided to households with pregnant or lactating women, in the highest poverty quintile (based on RGC’s “ID Poor” designation), a child with additional needs (eg, developmental delays, undernourished), and families experiencing domestic violence (identified by local government). VHVs and CDLs conduct home visits using IECD-provided session scripts that mirror the group session modules, with emphasis on practicing targeted behaviors.

Subnational capacity-building aims to increase the commitment of commune and district governments to expand services for families with young children. Consistent with evidence from nurturing care, households with multiple adversities receive the most intensive interventions, and their children are expected to experience the largest impacts.⁵ 

VHVs, CDLs, and IECD staff monitor program delivery using a tablet-based digital data collection tool. Representatives of commune and district government, and lead technical ministries, attend program activities to observe and provide feedback to the IECD team. Observation tools focus on fidelity, including adherence to the scripts, participant engagement, and session time devoted to participants practicing new skills. The IECD team monitors program reach (number of participants receiving each service) and session attendance. Data from observations are used to improve training and materials to strengthen intervention delivery.

This study evaluates the mediating effects of caregiving behaviors on children’s development using a modified randomized cluster stepped-wedge design,^22,23  with longitudinal tracking of study participants across enrollment assessment and 3 subsequent waves of data collection. The IECD program is delivered to 3 cohorts (clusters) of villages sequentially: Cohort 1 receives the intervention first, followed by Cohort 2 and then Cohort 3. Participants in all 3 village cohorts are recruited at baseline and complete the enrollment assessment. All are assessed again at the initiation and conclusion of each intervention cohort (Fig 2). Cohorts not receiving the intervention serve as comparison through a crossover procedure.

The IECD program is being implemented in Kampong Thom and Preah Vihear provinces located in the Tonlé Sap basin, in the freshwater floodplain of the lower Mekong River. These predominantly rural provinces were selected because they are located in USAID’s investment focus area, and have relatively high prevalence of stunting and poverty and low prevalence of infant and young child feeding practices, based on the 2014 Cambodian Demographic and Health Survey.

Village eligibility criteria (eg, geographic clustering [for feasibility of implementation], current agriculture production, and absence of other, similar donor-funded programs) informed the selection of 345 villages to receive the IECD intervention. The RGC selected 75 villages to assign to Cohort 1. We randomly assigned the remaining 270 villages to Cohorts 2 and 3, using a village-level randomization procedure. Within each village, we used systematic random sampling to select households and visited each selected household to assess eligibility. Eligibility criteria included at least 1 child aged <5 or a pregnant woman and primary caregiver (including during pregnancy) age 18 or older.

We calculated the power curves for detecting a statistically significant difference for a 2-way interaction term of intervention group (intervention versus comparison) and time (pre- versus postintervention) in a logistic regression model with significance of P value <.05. We generated 1000 replicates for each target proportion and sample size combination, using the proportion of cases in the intervention group/posttest group. The other 3 proportions, comparison/preintervention, comparison/postintervention, and treatment/preintervention, were set to 0.5. We accounted for the longitudinal design and primary caregiver clustering (ie, repeated measures), and we assumed that the cross-classification of comparison/treatment and preintervention/postintervention groups had an equal sample size. To detect a 10 percentage-point difference treatment/posttest group (ie, a proportion of 0.6, using an interaction term in a logistic regression model with 0.8 power) requires 1500 primary caregivers. We assumed a 90% retention rate across waves and intended to enroll 1825 caregivers from 345 villages. Six villages were inaccessible because of coronavirus disease 2019 mitigation measures during the enrollment assessment, resulting in a baseline sample of 1790 caregivers from 339 villages.

The local IECD team hires, trains, and supervises enumerators who have previous relevant data collection experience and local familiarity. Enumerators are independent from the intervention team and unaware of group assignment. The IECD team provides a 5-day training on instruments, tablets equipped with the data collection software, consent procedures, child safeguarding, and coronavirus disease 2019 safety protocols. Enumerators are trained to read the questions aloud without using verbal or nonverbal cues. Training includes piloting administration procedures, with supervision by IECD staff. Data collection supervisors and quality control monitors use quality control checklists and software error checks to monitor data collection teams.

Data are collected in private locations during household visits at enrollment and follow-up. Children who pass their fifth birthday “age out” and children are added as caregivers give birth. At enrollment, we gathered demographic data on all household members, including sex, age, marital status, and additional data on caregivers, including ability to read and write, employment status, and school attainment. We used the RGC’s household wealth classification system (ID Poor), to determine relative household poverty. Data for children include sex, age, birth weight, gestational age, and enrollment in preschool, day care, or other group care. At each data collection wave, we collect data on the primary outcome measures, children’s development, and on the secondary outcomes, targeted caregiving behaviors (Table 4).

To assess children’s development, we administer validated caregiver-report measures that can be administered by field personnel and incorporated into sustainable follow-up evaluations. For children aged 6 to 35 months, we use the Caregiver-Reported Early Development Instrument,²⁴  which has concordance with the Bayley Scales of Infant and Toddler development and has been used in Cambodia.²⁵  For children aged 36 to 59 months, we use the Early Childhood Development Index 2030 developed by the United Nations Children’s Fund.²⁶  In data collection waves 2, 3, and 4, we administer both the Early Childhood Development Index 2030 and the Caregiver-Reported Early Development Instrument to children aged 24 to 36 months to examine transition between measures.

To assess whether the impact of the intervention on children’s development is mediated by caregivers’ key behaviors, we administer measures of the secondary outcomes. Responsive caregiving and early learning are assessed using the Family Care Indicators^17,27  and a project-developed questionnaire. Nutrition is assessed using the World Health Organization infant and young child feeding measures and scoring for breastfeeding, minimum acceptable diet for children aged 6 to 23 months, and minimum dietary diversity for women of reproductive age (15–49 years).^28,29  We use project-developed measures to assess health and hygiene, and household stability and support. The health and hygiene questionnaire includes items on handwashing practices, latrine use, making water safe to drink, and caring for a sick child. The household stability and support questionnaire includes items on caregiver task-sharing, gender norms and attitudes, caregiving routines, and ability to access Cambodia’s conditional cash transfer program.

The study examines:

1. the direct effects of the IECD intervention on child development;

2. mediation through targeted caregiving behaviors;

3. moderation by household wealth, maternal education, and birth weight; and

4. sustainability of child development impacts.

We will apply analytic procedures for stepped-wedge designs that account for the confounding between-time and intervention effects inherent in this design by comparing exposed and unexposed time periods for the clusters. We will use guidance from a recent review of stepped-wedge cluster analysis strategies,³⁰  which extend the original Hussey and Hughes model.^31,32  We will fit generalized linear mixed-effect regression models to examine the effects of the IECD intervention activities on child development, controlling for propensity scores, if necessary, and accounting for repeated measurements.

We will conduct confirmatory factor analyses to develop scores for the 4 caregiving targets: (1) responsive caregiving and early learning, (2) nutrition, (3) health and hygiene, and (4) household stability and support. To assess mediation through the caregiving targets on child development, we will conduct path analyses and apply bootstrapping to estimate the confidence intervals for the indirect effects. We will test each caregiving target individually, and the cumulative caregiving targets, adjusting for clustering and propensity scores.

To examine whether the effects of the intervention are moderated by household wealth, maternal education, and birth weight, we will introduce interaction terms between intervention status and the 3 potential moderators into the model. We will use a likelihood ratio test to assess if the addition of the moderator significantly improves the model fit. If the model fit is significantly improved by the inclusion of the interaction term, we will conduct a simple slope analysis of the moderated mediation model and estimate the indirect effect. All models will be adjusted for clustering and propensity score weights, as necessary.

We will evaluate model fit using fit statistics, including the χ² statistic, recommended to be nonsignificant; the Comparative Fit Index and Tucker–Lewis Index recommended to be ≥0.9; root mean square error of approximation recommended to be ≤0.05; and the standardized root mean square residual recommended to be ≤0.08. Full information maximum likelihood will be used to account for missing data. We will interpret statistical significance as P < .05.

Finally, we will examine whether the effects of the intervention on caregiving targets and children’s development are sustained (over 1 year for Cohort 2 and 2 years for Cohort 1), using repeated measures.

We obtained ethical approval from the institutional review board of RTI International and from the Cambodian National Ethics Committee (Ministry of Health). All participants provide written informed consent (registration: www.clinicaltrials.gov, #NCT05197985).

This article outlines a longitudinal study to evaluate an integrated, multicomponent, multiplatform intervention developed in collaboration with government partners to enhance children’s development through caregiving activities, beginning prenatally through age 5 years. Building on evidence that caregiving practices are modifiable with well-structured support⁵  and implementation science principles, the IECD intervention is designed to help families adopt nurturing care practices throughout their daily activities. The project trains and engages village-level volunteers to implement multicomponent caregiving activities through community-, group-, and home-based sessions, with the goal of building capacity for sustainable interventions. Partnership with the RGC and involvement of RGC representatives increase the likelihood of ongoing governmental engagement and support for these activities.

The protocol is designed on the basis of principles of nurturing care and equity. The basic intervention is universal; intensity of services (eg, home visits and referrals for indicated professional services) increases with household or child vulnerability. This approach enables governments to consider policy objectives according to distinct population needs, such as universal coverage with a pro-poor, equity-enhancing investment in children and families most in need.³³  The partnership involves the US government with funding from USAID. The IECD program emphasizes the need for approaches supported by rigorous evidence, and the alignment of donor-funded programs with national governmental policies.

This study has several methodological limitations. First, the villages included in the sample may not be representative of the 2 provinces or Cambodia. Second, the study relies on caregiver-reported caregiving activities and child development, rather than observations. However, the measures are validated,²⁴  economical, and increasingly used to evaluate programs and policies related to childhood development. In alignment with implementation science principles, we selected population-based measures that can be incorporated into future government-led screening. Third, because of financial constraints, the nutrition assessment is limited to mid-upper arm circumference. Finally, the caregiving target measures meet psychometric criteria, but are sample-defined and may not apply in other contexts. Future programs should have a stronger co-creation process and well-developed process measures that involve all stakeholders including program participants.

The IECD program represents an advance over individual and group-oriented interventions by partnering with the RGC. Government representatives enhance their expertise in reducing disparities and promoting children’s development through participation in IECD implementation. In addition to evaluating sustainability once direct services are withdrawn, the design provides evidence to equip the RGC to develop policies that ensure equity by enacting the principles of nurturing care.

In conclusion, this multisectoral, multiplatform intervention will be evaluated over the coming years, generating useful learning about IECD interventions. The potential for all children to thrive is an integral component of the United Nations’ 2030 Sustainable Development Goals. Basing future interventions on governmental partnerships, with particular attention to the relevance of program targets, effectiveness, feasibility, local capacity, and sustainability, will advance children’s development and achievement of the Sustainable Development Goals.

We thank the contributions to the program design and implementation of IECD from Mr Hou Kroeun, Ms Rany Khoy, Mr Vannary Hun, Mr Channy Check, Ms Borany Aum, Ms Bunlang Sok, Mr Sopheap Phim, and Ms Sreignep Say. This protocol benefited from the contributions and leadership from RGC, particularly the Ministry of Social Affairs, Veterans, and Youth; the Ministry of Health; the Ministry of Education, Youth, and Sport; and the Council for Agriculture and Rural Development.

CDL

child development leader

IECD

Integrated Early Childhood Development

RGC

Royal Government of Cambodia

USAID

United States Agency for International Development

VHV

village health volunteer