Children from low- and middle-income countries (LMICs) face multiple challenges that significantly impede their cognitive and socioemotional development [1]. These challenges pose a substantial threat to their well-being in adulthood. Children’s early language and cognitive development, at risk in such contexts, are important determinants of subsequent literacy skills and later school progress [2,3].

Several socioeconomic conditions and poverty-related factors contribute to compromised child developmental progress [4-6]. These include nutritional deficits, which directly impact both physical and cognitive growth; limited access to educational resources, such as learning toys; elevated risk of chronic stress exposure; and the challenges posed by overcrowded and inadequate housing conditions. All of these disrupt the educational process [1,7] and children’s health status [8,9].

Compromised caregiver-child interaction and limited caregiver activities that support early learning are other key factors prevalent in deprived contexts that hinder children’s cognitive and socioemotional development [6,10,11]. Early difficulties in the caregiver-child relationship raise the risk of insecure infant attachment [12], itself predictive of a range of adverse child outcomes [13]. In addition, promoting warm and responsive interactions among caregivers, including vocalizations and smiles, has been shown to positively influence child health outcomes [14].

Providing education to parents in dialogic book-sharing (DBS) techniques has been proposed as a simple and deliverable means of obviating adversity risk to the caregiver-child relationship and child development [15,16]. There is considerable evidence for the benefits of book-sharing interventions provided to families living in high-income countries [16-19]. While a similarly beneficial impact has been reported from urban and periurban areas of LMICs [15,20-22], evidence is limited from poor rural communities, where the need for such early child development intervention may, arguably, be the greatest [10,23,24].

This paper presents the protocol for the Effects of a DBS Intervention for Female Caregivers in Rural Tanzania (EDBiT) study. EDBiT is a randomized controlled trial (RCT) designed to determine the effects of the DBS intervention delivered in rural Tanzania on cognitive development (specifically language skills and attention) as well as other parental and child outcomes.

During the preparation phase of the trial, we paid particular attention to the five recommendations proposed by Noble et al [18] to increase the probability that education in DBS will benefit the targeted child outcomes: (1) The DBS intervention delivered in this study involved a higher dosage compared to most previous studies [10,16,18,20-22,25], as the core 8-week educational program (intensive intervention) was followed by regular meetings of female caregivers swapping books and exchanging experiences with book-sharing and the upbringing of children (extensive intervention). (2) In addition to the waitlist control group, we included an active control group to test specific effects of book-sharing over the nonspecific effects of being subjected to some educational program focused on some parenting skills. (3) The design of this trial allows a potential follow-up that will assess the long-term effects of the index intervention. We expect the data collection for this follow-up study to be conducted in 3- to 5-year intervals, with the theory of change (ToC), presented in the Overview of the Study Design section, determining variables that will be the focus of our interest. (4) We collected data on a broad range of outcome variables to provide a rich assessment of the effects of book sharing on a variety of outcomes. (5) Children from diverse socioeconomic backgrounds living in the target region were included, allowing for the investigation of moderating social factors.

The aim of the EDBiT study is to assess the effects of DBS, a parenting program for female caregivers of children aged between 15 and 45 months, implemented in rural Tanzania. We aim to assess the impact of the intervention on the following domains: child cognitive and socioemotional skills, parenting and parental stress, and child health. We will evaluate the moderating influence of certain variables, including caregivers’ cognitive abilities.

This study is a 3-arm cluster RCT. Registered female caregivers were randomized at the individual level to an index intervention group attending DBS educational sessions, an active control group attending educational sessions focused on playful activities (PAs) designed to promote the development of a child’s motor skills, or a waitlist (passive) control (WLC) group. We collected data on all 3 groups at baseline, after intensive intervention, and after extensive intervention, approximately 1 year after the start of the intervention (refer to the Intervention section for the description of the intensive and extensive phases of the intervention).

On the basis of previous research on the efficacy of the version of DBS implemented in this study [15,20-22], significant differences between the index group and the WLC group are to be expected. The comparison between DBS and PAs is to control for the fact that any parenting interventions may yield positive results for children because of the caregivers’ increased attentiveness to their children rather than because of the content of the index intervention itself. While DBS primarily aims to develop child language, PAs were strategically designed to cultivate other skills that are not the primary focus of DBS (refer to the subsequent sections).

Several authors argue for adding qualitative methods to quantitative ones in RCTs for studies to paint a richer picture of the evaluated interventions and increase the validity of findings [26-28]. However, most existing RCTs of DBS programs apply solely quantitative methods [17,18,20-22]. In EDBiT, we adopted a mixed methods approach, with qualitative research capturing the subjective perspectives of caregivers in both DBS and PAs on the intervention’s effects. Qualitative data will provide an additional layer of information with the potential to further explain the results of the quantitative analysis, possibly shedding light on the factors that facilitated or hindered the realization of intervention outcomes. They will also examine how the intervention influenced caregivers’ parenting practices and attitudes toward their role in a child’s development. Caregivers’ book-sharing patterns were monitored through self-completed diaries and interviews. The level of parental engagement could moderate the index intervention’s impact on child outcomes [18].

Considering the importance of theories of change for impact evaluations [29], we constructed the causal chain (Figure 1). The model includes the main activities of the evaluated intervention, the intended immediate outcomes, and an indicative list of expected long-term impacts (beyond the scope of this study).

This study has been conducted in the rural areas of the Usangu Plain, South West Tanzania. Specifically, it is located along the southern and eastern edges of the Usangu area, historically labeled as the “cultivation zone” [30]. This populous area relies on small-scale farming, mainly on the production of rice and maize. Implementing the EDBiT study in Usangu provides an opportunity to test the feasibility and effectiveness of a DBS intervention in a context that is close to the socioeconomic median for Africa rather than at either extreme. The Multidimensional Poverty Index places both the South West Highlands region, where Usangu is situated, and Tanzania as a whole near the median of their respective national and continental distributions [31]. Findings from such a setting may offer valuable insights into the potential scalability of the DBS program elsewhere across the African continent.

This study covers 12 towns, villages, and geographically isolated neighborhoods (referred to as “villages” in Figure 2) from Utengule Usangu, Kongolo, and Igurusi wards (Figure 2) [32,33]. Assessment and education sessions were held in each of the 12 villages. Village executive officers were informed about the project, and leaders in all villages approved both the intervention and the study.

This study received ethical approval from the Faculty of Science Ethics Committee at Palacký University Olomouc (reference 23-03) and from the University of Dar es Salaam (reference AB3/12B). Caregivers provided informed consent for their own and their children's participation after the full consent form was read aloud to them during registration, and again verbally prior to each instance of data collection. All consent was audio-recorded. Participation was voluntary, and caregivers were informed that they could withdraw at any time without consequences. No personally identifying information was collected, and all data were anonymized prior to analysis. Participants did not receive financial compensation, but small refreshments and gifts were provided as a token of appreciation. All the data are securely stored on a cloud server and made available only to the core research team members to maintain confidentiality. The study has been registered with the International Standard Registered Clinical/Social Study Number (ISRCTN12613329).

Female caregivers caring for a child born between December 1, 2019, and June 1, 2022 (aged between 15 and 45 months at the time of baseline assessment), were eligible for participation in this study. Eligible women were the primary persons responsible for childcare and not necessarily biological mothers.

Recruitment of caregivers was conducted between July 29, 2023, and September 3, 2023. Members of the project team visited prospective households, accompanied by respected women from the local communities. Eligible caregivers were enrolled in the study if they (1) provided consent to participate in the study; (2) declared their availability to attend the 3 rounds of assessments; and (3) agreed that we could, if necessary, delay their participation in the project by 1 year.

In total, 542 eligible caregivers enrolled in this study. For each village, the location of each household was determined by GPS, and mutual distances between households were computed. To prevent cross contamination, we divided groups of households into clusters. To avoid large clusters of households, we selected a subset of households that could be split into smaller clusters of no more than 5 households. The distance between households from distinct clusters was set as >100 m. Some eligible households were excluded from the study to fulfill this condition. These caregivers were invited to join savings groups a year later (refer to the Intervention section). The number of excluded households (n=99) was kept to a minimum by using advanced graph theory methods. Therefore, 443 (81.7%) female caregivers were included in this study.

Clusters, each including 1 to 5 households, were randomly assigned to the DBS, PA, or WLC group. The randomization process mimicked a block design; it considered dwelling place (village) and child age (children aged between 15 and 30 months vs children aged >30 to 45 months) and ensured a balanced distribution of these characteristics across the study groups.

The study participants were not informed which program (DBS or PA) constituted the studied intervention and which served as a control condition.

Consistent delivery of both the DBS and PA programs was tracked with an 18-item implementation fidelity checklist. In total, 11 items assessed the substantive content of each session, for example, whether the facilitator explained key concepts clearly and gave accurate, supportive feedback during mothers’ practice, whereas 7 items captured organizational logistics and caregiver-oriented factors (eg, punctual arrival of facilitators and mothers, maternal attentiveness during the theoretical segment, distribution of porridge to children at the start of the session, and the babysitter’s behavior).

Each checklist item was transformed to a 0% to 100% scale; the mean of the 11 content items formed the content fidelity subscore, and the mean of the 7 organizational items formed the procedural fidelity subscore. A session was deemed adequate only when both subscores were ≥75%. Facilitators whose scores fell below this threshold received targeted supervisory feedback before the next block of sessions. Fidelity distributions will be summarized descriptively in trial reports.

An independent rater conducted live, in-person observations of 2 sessions per facilitator, with a minimum interval of 3 weeks between visits. Sessions were not audio or video recorded. Fidelity scores will be used solely to define the per-protocol analytic sample (dyads taught by facilitators whose average content and procedural fidelity across the 2 observed sessions are each ≥75%). Because the checklist serves primarily as a quality-assurance gate, fidelity will otherwise be summarized descriptively and will not be modeled as an explanatory variable.

Data were collected on tablets and mobile phones using a Kobo platform [51] and uploaded to a secure online server. Child assessments and caregiver-child book-sharing interactions were video recorded. A random sample of videos was regularly checked for fidelity of the data collection procedure. Interviews and focus groups were recorded on a voice recorder and checked for quality and adherence to the prescribed questions.

The EDBiT trial addresses a critical evidence gap—whether DBS can enhance early language, socioemotional competence and caregiver behavior in resource-constrained, rural settings in sub-Saharan Africa. By comparing DBS with both a WLC group and an active PA control group, the study will isolate benefits specific to DBS rather than the generic advantages of increased parent-child interaction. Its mixed methods design, underpinned by an articulated ToC, will also illuminate how and for whom DBS works, thereby advancing implementation science for scalable parenting interventions.

Most RCTs of book sharing have been conducted in high-income or urban settings and have rarely included an active control group, leaving open the question of whether the gains stem from DBS itself or simply from extra parent-child time. EDBiT also helps close this gap by testing the intervention in rural Tanzania, where material resources are limited. In addition, this study’s mixed methods approach brings quantitative impact estimates into dialogue with caregivers’ lived experiences, generating implementation insights that purely quantitative trials often miss.

Nevertheless, inevitable field-trial constraints remain. The findings may not be readily generalizable to other locations or time points, and reliance on self-report for some caregiver outcomes could introduce social-desirability bias despite triangulation with observational data.

If DBS intervention proves effective, future work could explore delivery to other household members, such as older siblings and fathers. Digital adaptations might be tested where printed materials are scarce. In addition, future work might experiment with tailoring program content or dosing for distinct subgroups, for example, mothers with lower cognitive ability or those living in extreme poverty, to determine whether customized approaches enhance the interest of mothers and intervention impact.

Findings will be shared with (1) local stakeholders in the Mbeya region of Tanzania through community forums and briefing sessions and (2) the global research community through presentations at international early-childhood conferences and open-access publications in peer-reviewed journals. Anonymized datasets will also be deposited in an open repository.