An Implementation Evaluation of A Group-Based Parenting Intervention to Promote Early Childhood Development in Rural Kenya

Abstract

Early childhood development (ECD) parenting interventions can improve child developmental outcomes in low-resource settings, but information about their implementation lags far behind evidence of their effectiveness, hindering their generalizability. This study presents results from an implementation evaluation of Msingi Bora ("Good Foundation" in Swahili), a group-based responsive stimulation and nutrition education intervention recently tested in a cluster randomized controlled trial across 60 villages in rural western Kenya. Msingi Bora successfully improved child cognitive, receptive language, and socioemotional outcomes, as well as parenting practices. We conducted a mixed methods implementation evaluation of the Msingi Bora trial between April 2018 and November 2019 following the Consolidated Advice for Reporting ECD implementation research (CARE) guidelines. We collected qualitative and quantitative data on program inputs, outputs, and outcomes, with a view to examining how aspects of the program's implementation, such as program acceptance and delivery fidelity, related to observed program impacts on parents and children. We found that study areas had initially very low levels of familiarity or knowledge of ECD among parents, community delivery agents, and even supervisory staff from our partner non-governmental organization (NGO). We increased training and supervision in response, and provided a structured manual to enable local delivery agents to successfully lead the sessions. There was a high level of parental compliance, with median attendance of 13 out of 16 fortnightly sessions over 8 months. For delivery agents, all measures of delivery performance and fidelity increased with program experience. Older, more knowledable delivery agents were associated with larger impacts on parental stimulation and child outcomes, and delivery agents with higher fidelity scores were also related to improved parenting practices. We conclude that a group-based parenting intervention delivered by local delivery agents can improve multiple child and parent outcomes. An upfront investment in training local trainers and delivery agents, and regular supervision of delivery of a manualized program, appear key to our documented success. Our results represent a promising avenue for scaling similar interventions in low-resource rural settings to serve families in need of ECD programming. This trial is registered at ClinicalTrials.gov, NCT03548558, June 7, 2018. https://ichgcp.net/clinical-trials-registry/NCT03548558.

Keywords: CARE guidelines; early childhood development; implementation evaluation; parenting intervention; rural Kenya.

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2021 Luoto, Lopez Garcia, Aboud, Singla, Zhu, Otieno and Alu.

Figures

Figure 1

Logic model. Child and parent outcome indicators are reported in Luoto et al. (14).

Figure 2

Mean CHV Delivery Scores by Session. Figure plots mean scores from supervisor monitoring forms across all CHVs by session for items measured on a 1–5 scale from “poor” to “excellent” for six of 13 delivery measures using solid lines. N = 686 supervisor rating forms across 16 sessions and 40 villages. Some earlier sessions had more than 1 supervisor present. For session “quality” and “fun,” figure plots mean scores on 1–10 scale on right-hand-side axis using dashed lines. All 13 delivery measures show similar upward trends and the full list is in Figure 3. Red vertical lines represent review sessions when sessions were home visits for mixed-delivery arm and group sessions for group-only arm.

Figure 3

Delivery Ratings: Supervisor and Self-Evaluations. Findings are mean ratings from supervisor or self-assessment monitoring forms across all 16 sessions and 40 intervention villages. N = 686 for supervisor monitoring forms; N = 2,043 for CHV self-assessment forms that include each home visit.

Figure 4

Mean maternal attendance by study arm and session. From attendance monitoring data. Dashed faded lines represent 95% confidence intervals; red vertical lines represent review sessions when sessions were home visits for mixed-delivery arm and group sessions for group-only arm. The only sessions with statistically significant differences in attendance at 95% level or higher are the four review sessions (4, 8, 12, and 16).