Joint parental stimulation and early childhood development in 26 sub-Saharan African countries

Abstract

Background

Engagement in stimulation activities by both parents can have synergistic effects on early childhood development (ECD). While the role of maternal stimulation is well known, the association with paternal stimulation remains underexplored and inconsistent. The current study aimed to assess the distribution of different combinations of maternal and paternal stimulation and explore their associations with ECD in the context of sub-Saharan Africa (SSA).

Methods

Using the Multiple Indicator Cluster Surveys and Demographic and Health Surveys from 26 SSA countries, we included 69 637 children aged 3–4 years in the analysis. The standardised ECD z-score was constructed using the ECD index. We estimated three-level linear regressions to assess its association with parental engagement in stimulation activities which was classified into four categories: mother low, father low (fewer than four activities); mother high, father low; father high, mother low; mother high, father high. Stratification analyses were conducted to examine whether the association differs by socioeconomic, geographic and learning-related variables.

Results

Only 4% of parents provided high stimulation jointly (both high), which was associated with higher ECD z-score compared with both parents providing low stimulation (both low), after adjusting for covariates (b=0.04, 95% CI=0.01, 0.08). The association between joint parental stimulation and ECD was greater for children from wealthier households compared with their poorer counterparts (interaction p=0.012).

Conclusions

Children showed better developmental status when both parents jointly provided high stimulation. Policies and interventions should target both mothers and fathers, and transform traditional norms about parenting to achieve the most optimal development outcomes of children.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• In the context of sub-Saharan Africa, parental engagement in stimulation activities can improve their child’s early developmental status. However, while the positive role of maternal stimulation is well known, findings for paternal stimulation are inconsistent. The effect of joint stimulation by both parents, in particular, remains underexplored.

WHAT THIS STUDY ADDS

• To our knowledge, this study is one of the first to quantify the different combinations of maternal and paternal stimulation and their significant associations with early childhood development (ECD) in sub-Saharan African countries. Using the nationally representative data, the present study advances our understanding of the role of joint parental engagement in stimulation activities on ECD by suggesting that children show the most optimal developmental status when both parents engage in stimulation activities.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The present study suggests the need for future studies that identify the barriers discouraging parents, particularly fathers, from fully engaging in stimulation activities for their children. To achieve the most optimal developmental outcomes of children in sub-Saharan Africa, policies and interventions aimed at improving parenting practices should holistically target both parents and transform traditional norms about parenting that prevent parents from fully participating in overall childcare activities.

Introduction

Early childhood is a critical period for promoting healthy growth, school readiness and academic achievement in the short term, and for ensuring better health and well-being at later adulthood.1,3 At population level, early childhood development (ECD) is directly linked to human capital development which is needed to accelerate sustainable economic growth.^{2 4}

In the context of sub-Saharan Africa (SSA), with countries undergoing rapid population growth and demographic transition,⁴ investing in ECD is particularly important. However, in SSA, approximately 40% of children remain at risk of developmental delay^{5 6} with significant socioeconomic and geographic inequalities (ie, children from poorer households and rural areas are at greater risk of being developmentally off-track⁵). Hence, ECD is a global priority to ensure all children thrive to reach their full potential.

The Nurturing Care Framework by the WHO introduces five essential components for promoting ECD: good health, adequate nutrition, responsive caregiving, opportunities for early learning, and security and safety.⁶ In particular, early learning and responsive caregiving are important for building emotional bonds and linguistic development in children.^{1 2 7} Caregivers’ engagement in stimulation activities including reading, playing and naming/counting objects with the child can provide contexts for responsive caregiving and early child learning. Various studies have reported positive associations between caregivers’ stimulation and ECD outcomes in low-income and middle-income countries (LMICs).38,10 The conceptual framework of multigenerational life course of development by Black et al² also illustrates that such parenting activities enhance developmental potential in early childhood, which exchange diverse influences with the rest of the components (eg, good health, adequate nutrition) and are fundamentally sustained by broader contexts such as enabling environment for caregivers and families or social, economic and political conditions.

Existing literature has focused mostly on the role of mothers in optimising ECD^{3 11} since they are assumed to be the primary caregivers of young children globally. Nonetheless, assessing paternal stimulation and its effect on ECD is also important because studies have suggested that fathers may engage in childcare through unique ways that have additive effects on ECD. For example, qualitative studies found that fathers were more involved in physical play (ie, ‘rough and tumble’) while mothers tended to engage in overall play and caregiving behaviours during the interaction with their child.12,14 Despite its potential importance, paternal stimulation remains far inadequate in LMICs, with only 3.9% of fathers engaged in high stimulation in SSA countries (compared with 14.6% for mothers).¹⁵ In terms of its association with ECD outcomes, prior studies reported mixed findings for paternal stimulation. For example, positive associations between paternal stimulation and child’s cognitive, linguistic and socioemotional development were found in studies from 38 LMICs,¹¹ Rwanda¹⁶ and Pakistan.¹⁷ In contrast, other studies from Kenya,¹⁸ Bangladesh,¹⁹ rural Pakistan²⁰ and Honduras²¹ found null to negative associations.

What remains underexplored in this literature is the potential synergistic effects from joint stimulation by both parents. When estimating the effect of paternal stimulation, the majority of previous studies16,1820 failed to adjust for maternal stimulation in their analyses, potentially resulting in overestimation of the association between paternal stimulation and ECD. Only a few studies adjusted for maternal stimulation,^{11 19 21} and none considered maternal and paternal stimulation jointly. Exploring levels of maternal and paternal stimulation in combination is important as it can reveal further insights into coparenting and capture the interactive influence of the caregiving practices between the two caregivers in supporting the child’s development.^{12 22} To the best of our knowledge, there has been no attempt to assess the association between joint parental stimulation and ECD in SSA countries.

Finally, the association between joint parental stimulation and child development can also vary depending on other factors such as sociodemographic and early learning-related characteristics, given that they influence child development simultaneously and their effects are not mutually exclusive.² For example, among children in wealthier households, those who receive care from other extended family members or those who attend early education programmes, parents’ stimulation may be correlated with child development to a lesser degree since there are many other enriching sources of stimulation, compared with those without these additional early learning opportunities.^{11 23} By assessing stratified analyses by these various factors, one can estimate the association between parental stimulation and ECD outcomes among children at high risk (ie, those in poorer settings) and identify subgroups that would benefit the most from stimulation interventions.²⁴

Hence, based on Black et al’s conceptual framework,² the current study sought to assess the relationship between joint parental stimulation and child development as described in online supplemental figure 1. More specifically, using the nationally representative data of 26 SSA countries, the study aimed to assess the distribution of the different combinations of maternal and paternal engagement in daily stimulation activities for children (as defined by four mutually exclusive categories of both mother and father engaging in high stimulation; only mother engaging in high stimulation; only father engaging in high stimulation; neither engaging in high stimulation) and explore their associations with ECD. Furthermore, we aimed to investigate whether this association differs by various socioeconomic, geographic and learning-related characteristics.

Methods

Data and study sample

Data from the Multiple Indicator Cluster Surveys (MICS) and Demographic and Health Surveys (DHS) were used. The MICS and DHS are one of the largest sources of comprehensive and internationally comparable data on children and women worldwide.^{25 26} Among 48 SSA countries classified by the World Bank Group, we excluded two surveys that were not publicly available either from the MICS and DHS, four surveys that were conducted before 2010 when a questionnaire for ECD was first introduced²⁷ and 14 surveys that did not collect ECD data. Two additional surveys were excluded for incomplete data about attendance to early childhood education (ECE) programmes, which resulted in 26 country surveys (21 from MICS; 5 from DHS) (figure 1).

Figure 1. Flow chart of sample construction (N=69 637).

Our study sample was restricted to children aged 3 and 4 years who were the focal age group for the ECD questionnaire (N=114 568). Children who were not residing with both parents (N=37 198) were excluded from the analysis because they lacked information on characteristics for parents (eg, age, education level). We additionally excluded those missing data for parental stimulation (1.4%), ECD outcomes (5.2%) and other covariates (0.0%–7.2%), leaving a total of 69 637 children for the main analysis (table 1). Even after the exclusion, the final sample showed overall similar distribution in sociodemographic characteristics when compared with the raw sample of 114 568 children (data not shown).

Table 1. Countries, survey years, sample sizes and distribution of parental stimulation and ECD mean score (N=69 637).

Country	Survey year	N (%)	Parental stimulation, N (%)				ECD score (SD)	ECD z-score (SD)
			Both low	Mother high, father low	Father high, mother low	Both high
Total		69 637 (100.00)	55 065 (79.07)	10 288 (14.77)	1477 (2.12)	2807 (4.03)	3.34 (1.20)	0 (1)
Benin	2017–2018	3213 (4.61)	2618 (81.48)	446 (13.88)	71 (2.21)	78 (2.43)	3.39 (1.18)	0.05 (0.97)
Burundi	2016–2017	3555 (5.11)	2977 (83.74)	455 (12.80)	22 (0.62)	101 (2.84)	2.78 (1.25)	−0.45 (1.03)
Cameroon	2014	1812 (2.60)	1523 (84.05)	205 (11.31)	25 (1.38)	59 (3.26)	3.45 (1.08)	0.10 (0.89)
Central African Republic	2018–2019	2180 (3.13)	1697 (77.84)	302 (13.85)	88 (4.04)	93 (4.27)	2.72 (1.22)	−0.50 (1.00)
Chad	2019	4394 (6.31)	3568 (81.20)	561 (12.77)	171 (3.89)	94 (2.14)	2.97 (1.26)	−0.29 (1.04)
Congo	2014–2015	2213 (3.18)	1580 (71.40)	416 (18.80)	109 (4.93)	108 (4.88)	3.32 (1.94)	−0.01 (0.98)
Congo Democratic Republic	2017–2018	5287 (7.59)	4320 (81.71)	619 (11.71)	91 (1.72)	257 (4.86)	3.15 (1.25)	−0.14 (1.02)
Cote d'Ivoire	2016	2347 (3.37)	2063 (87.90)	156 (6.65)	39 (1.66)	89 (3.79)	3.58 (1.15)	0.20 (0.95)
Eswatini	2014	263 (0.38)	198 (75.29)	49 (18.63)	6 (2.28)	10 (3.80)	3.62 (0.95)	0.24 (0.78)
Gambia	2018	2571 (3.69)	2448 (95.22)	98 (3.81)	12 (0.47)	13 (0.51)	3.53 (0.92)	0.17 (0.76)
Ghana	2017–2018	2035 (2.92)	1741 (85.55)	193 (9.48)	44 (2.16)	57 (2.80)	3.23 (1.14)	−0.08 (0.94)
Guinea	2016	2249 (3.23)	1863 (82.84)	270 (12.01)	30 (1.33)	86 (3.82)	3.13 (1.11)	−0.16 (0.91)
Kenya	2013–2014	663 (0.95)	614 (92.61)	38 (5.73)	8 (1.21)	3 (0.45)	3.44 (1.17)	0.09 (0.96)
Lesotho	2018	281 (0.40)	237 (84.34)	33 (11.74)	4 (1.42)	7 (2.49)	4.10 (0.97)	0.64 (0.80)
Madagascar	2018	3489 (5.01)	3099 (88.82)	296 (8.48)	34 (0.97)	60 (1.72)	3.53 (1.14)	0.16 (0.94)
Malawi	2019–2020	3446 (4.95)	2947 (85.52)	383 (11.11)	56 (1.63)	60 (1.74)	3.45 (1.25)	0.10 (1.03)
Mali	2015	5255 (7.55)	4056 (77.18)	937 (17.83)	76 (1.45)	186 (3.54)	3.41 (1.01)	0.07 (0.83)
Mauritania	2015	2339 (3.36)	1761 (75.29)	393 (16.80)	47 (2.01)	138 (5.90)	3.41 (1.15)	0.07 (0.95)
Nigeria	2016–2017	9534 (13.69)	6242 (65.47)	2059 (21.60)	337 (3.53)	896 (9.40)	3.45 (1.17)	0.10 (0.97)
Rwanda	2019–2020	2221 (3.19)	1586 (71.41)	504 (22.69)	37 (1.67)	94 (4.23)	4.22 (1.00)	0.73 (0.82)
Sao Tome and Principe	2019	425 (0.61)	315 (74.12)	78 (18.35)	7 (1.65)	25 (5.88)	3.50 (1.05)	0.14 (0.86)
Senegal	2019	1246 (1.79)	1215 (97.51)	22 (1.77)	6 (0.48)	3 (0.24)	3.63 (1.18)	0.25 (0.97)
Somalia	2011	2650 (3.81)	1639 (61.85)	876 (33.06)	37 (1.40)	98 (3.70)	3.10 (1.31)	−0.19 (1.08)
Togo	2017	1418 (2.04)	1172 (82.65)	145 (10.23)	39 (2.75)	62 (4.37)	3.20 (1.26)	−0.11 (1.03)
Uganda	2016	3312 (4.76)	2636 (79.59)	545 (16.46)	43 (1.30)	88 (2.66)	3.43 (1.15)	0.09 (0.94)
Zimbabwe	2019	1239 (1.78)	950 (76.67)	209 (16.87)	38 (3.07)	42 (3.39)	3.71 (1.06)	0.37 (0.87)

Note. Parental stimulation was considered high if a mother or father engaged in four or more stimulation activities out of six in the past three3 days preceding the survey. (ECD) score is a continuous score ranging from 0 to 5. ECD z-score is a standardizedstandardised estimate having a mean of 0 and an standard deviationSD of 1 in the pooled dataset. For Benin, Burundi, Rwanda, Senegal and Uganda, the countries signified with an asterisk (*), data from the Demographic and Health Surveys waswere used; data from the Multiple Indicator Cluster Surveys waswere used otherwise.

ECD, early childhood development.

Exposure

Parents’ engagement in stimulation activities was reported by the primary caregiver (mostly mothers) in terms of the number of activities (0–6) that a mother or father of a child engaged in over the past 3 days. These activities include: (a) read books, (b) told stories, (c) sang songs, (d) took outside for a walk, (e) played with and (f) named or counted things with. For mothers and fathers, separately, high stimulation was defined as engagement in four or more activities and low stimulation as engagement in three or less activities per scoring guidelines developed by UNICEF²⁸ and consistently with prior literature.^{15 29} Based on this information, a joint variable was constructed to reflect the four different combinations of maternal and paternal stimulation: both mother and father had low stimulation (both low), mother had high stimulation but father had low stimulation (mother high, father low), father had high stimulation but mother had low stimulation (father high, mother low) and both mother and father had high stimulation (both high).

Outcome

ECD Index (ECDI) is an indicator of ECD that was developed and validated across LMICs by UNICEF.²⁷ It measures the developmental status of children aged 3 and 4 years based on 10 caregiver-reported items that encompass physical, literacy-numeracy, socioemotional and cognitive development.²⁷ However, physical and literacy-numeracy development items are considered less appropriate for children in this age range.^{30 31} For instance, physical development items represent a developmental skill that is generally achieved before the age of 1 year (ie, child can pick up a small object with two fingers) and child’s overall health rather than developmental status (ie, child is sometimes too sick to play).^{30 31} Literacy-numeracy development items (child can identify/name at least 10 letters of the alphabet; child can read at least four simple, popular words; child knows the name and recognises the symbol of all numbers from 1 to 10) were designed to capture preacademic knowledge that is more advanced than developmental capacity of children aged 3–4 years.^{30 31} Hence, the ECD score for this study was generated using three socioemotional development items (child gets along well with other children; child sometimes kicks, bites or hits other children or adults; child gets distracted easily or quickly) and two cognitive development items (child follows simple directions on how to do something correctly; child can work on a task, including play tasks, independently). The ECD score ranged from 0 to 5, with a higher score indicating more optimal developmental status, and was standardised to have a mean of 0 and an SD of 1 in the pooled dataset (ECD z-score).

Covariates

In regression analyses, we adjusted for the following variables based on the multigenerational life course framework of development² and relevant prior literature1118,21 32 33: child’s sex (girl/boy) and age in years, mother and father’s age in years and their education level (no education to primary/secondary or higher education), household wealth quintiles (poorest/poorer/middle/richer/richest), type of place of residence (rural/urban) and number of children aged under 5 years at home were included. Household wealth index score is a composite index of relative standard of living at household level, which was created in the MICS and DHS based on household characteristics and assets.^{25 26} Additionally, the following variables related to children’s learning and stimulation were used as covariates: number of child books at home (0–10+), variety of toys at home (0–3), stimulation by other caregivers (low/high) and attendance to ECE programmes (no/yes). Other caregiver’s stimulation was considered high if any caregiver besides the child’s parents and older than 15 years engaged in four or more activities in the past 3 days, following the definition used for parents.

Statistical analyses

We first described the general distribution of parental stimulation and child’s ECD scores in the pooled and country-specific sample. The distribution of ECD scores and covariates was then presented by the level of parental stimulation to assess heterogeneity.

The main analysis using the joint variable of parental stimulation (both low; mother high, father low; father high, mother low; both high) was performed as a three-level random effects linear model for 69 637 children (level-1) nested within 13 713 clusters (level-2) and 26 countries (level-3). This can be described as following: ${\pi }_{ijk}=\alpha +\beta x_{ijk}+t_{ijk}+u_{jk}+v_k$ where ${\pi }_{ijk}$ is the estimated ECD z-score of a child i in cluster j in country k. $x_{ijk}$ is a vector of all covariates while β is that of estimated coefficients. t_ijk, u_ik and $v_k$ each represent the unexplained residuals at the individual-level, cluster-level and country-level with the assumption of: ${\displaystyle t_{ijk}\sim N\left(0,{\sigma }_t^2\right)}$, ${\displaystyle u_{jk}\sim N\left(0,{\sigma }_u^2\right)}$, and ${\displaystyle v_k\sim N\left(0,{\sigma }_v^2\right)}$. Regressions included child’s demographics (sex, age) in model 1 and sequentially adjusted for socioeconomic and learning-related characteristics of parents and households (age and education level of parents, household wealth quintiles, type of place of residence, number of children, number of child books, variety of toys, attendance to ECE programmes) in model 2.

To assess the heterogeneity in the association between the different combinations of parental stimulation and child’s ECD z-score, stratified regressions were estimated by household wealth (poorest/poorer vs middle/richer/richest), maternal education level (no education/primary vs secondary/higher education), type of place of residence (rural vs urban), other caregiver’s stimulation (low vs high) and attendance to ECE programmes (no vs yes). For each of the stratification variables, their interactions with parental stimulation were estimated and relevant p values from likelihood ratio tests were provided.

As a secondary analysis, we conducted regressions separately for maternal and paternal stimulation to assess their independent associations with ECD z-score. In addition to the same set of covariates included in models 1 and 2 of the main analysis, spouse’s stimulation was additionally adjusted for in model 3. To ensure the robustness of our results, two sensitivity analyses were conducted. First, the 10-point ECD score was generated with the addition of two physical and three literacy-numeracy development items and standardised to have a mean of 0 and an SD of 1 for the subsample of 68 398 children. Second, a revised cut-off for parental stimulation was applied based on the median number of stimulation activities that mothers and fathers engaged in. In this sensitivity analysis, parental stimulation was thereby considered high if a mother engaged in two or more activities and if a father engaged in one or more activities. For these newly generated variables, two linear regression models were additionally estimated with the same set of covariates included in the main analysis.

Statistical outputs obtained from the regressions were reported in the form of beta coefficients (b) with 95% CIs and were considered statistically significant at p<0.05. For all regressions, survey years (2011–2020) were included. All analyses were performed via STATA/MP V.16.1, with the use of mixed STATA command for three-level regression analyses (for full STATA codes, see online supplemental text 1).

Patient and public involvement

It was not appropriate or possible to involve patients or the public in the design, or conduct, or reporting, or dissemination plans of our research.

Results

In the pooled sample, 4.0% of children had parents who jointly engaged in high stimulation (both high; ranging from 0.2% in Senegal to 9.4% in Nigeria) (table 1). 14.8% of children had high maternal stimulation and low paternal stimulation (mother high, father low), and 2.1% had the opposite (father high, mother low). For the remaining 79.1%, both the mother and father had low stimulation (both low). The mean ECD score in the pooled sample was 3.3 out of 5 (SD=1.2) and varied across countries (ranging from 2.7 in Central African Republic to 4.2 in Rwanda).

ECD score was highest among children whose mother and father jointly had high stimulation (both high; 3.5), followed by ‘mother high, father low’ (3.4), ‘father high, mother low’ (3.4) and ‘both low’ groups (3.3) (table 2). Parents with high stimulation were also likely to have a higher education level, belong to richer households and live in urban areas. Number of books and variety of toys were greater for parents with high stimulation, and these parents were more likely to have their children attend ECE programmes. All the above group differences were statistically significant (analysis of variance/χ² p<0.001).

Table 2. Descriptive statistics of the pooled and stratified sample by parental stimulation.

Variables	Pooled (N=69 637)	Parental stimulation				P value
		Both low (N=55 065)	Mother high, father low (N=10 288)	Father high, mother low (N=1477)	Both high (N=2807)
	N (%)/mean (SD)
ECD score	3.34 (1.20)	3.32 (1.21)	3.42 (1.18)	3.38 (1.12)	3.51 (1.13)	<0.001
ECD z-score	0 (1)	−0.01 (0.99)	0.07 (0.97)	0.05 (0.92)	0.15 (0.93)	<0.001
Child’s sex
Girl	34 362 (49.3)	27 035 (49.1)	5347 (52.0)	653 (44.2)	1327 (47.3)	0.37
Boy	35 275 (50.7)	28 030 (50.9)	4941 (48.0)	824 (55.8)	1480 (52.7)
Child’s age (year)	3.48 (0.50)	3.48 (0.50)	3.48 (0.50)	3.50 (0.50)	3.48 (0.50)	<0.001
Mother’s age (year)	31.38 (7.16)	31.49 (7.23)	30.96 (6.85)	30.44 (7.12)	31.25 (6.80)	<0.001
Father’s age (year)	39.98 (10.09)	40.16 (10.22)	39.36 (9.65)	38.88 (9.86)	39.42 (9.12)	<0.001
Mother’s education level
No education and primary	54 566 (78.4)	45 384 (82.4)	6681 (64.9)	960 (65.0)	1541 (54.9)	<0.001
Secondary and higher	15 071 (21.6)	9681 (17.6)	3607 (35.1)	517 (35.0)	1266 (45.1)
Father’s education level
No education and primary	47 494 (68.2)	39 475 (71.7)	6077 (59.1)	662 (44.8)	1280 (45.6)	<0.001
Secondary and higher	22 143 (31.8)	15 590 (28.3)	4211 (40.9)	815 (55.2)	1527 (54.4)
Household wealth
Poorest	18 055 (25.9)	15 488 (28.1)	1827 (17.8)	260 (17.6)	480 (17.1)	<0.001
Poorer	15 935 (22.9)	13 345 (24.2)	1885 (18.3)	257 (17.4)	448 (16.0)
Middle	13 539 (19.4)	10 862 (19.7)	1954 (19.0)	275 (18.6)	448 (16.0)
Richer	11 780 (16.9)	8849 (16.1)	2049 (19.9)	314 (21.3)	568 (20.2)
Richest	10 328 (14.8)	6521 (11.8)	2573 (25.0)	371 (25.1)	863 (30.7)
Type of place of residence
Rural	50 903 (73.1)	41 700 (75.7)	6543 (63.6)	951 (64.4)	1709 (60.9)	<0.001
Urban	18 734 (26.9)	13 365 (24.3)	3745 (36.4)	526 (35.6)	1098 (39.1)
Number of children	2.17 (1.27)	2.20 (1.30)	2.07 (1.19)	1.97 (0.92)	2.02 (1.04)	<0.001
Number of child books	0.25 (1.04)	0.14 (0.76)	0.56 (1.52)	0.68 (1.67)	1.03 (2.08)	<0.001
Variety of toys	1.59 (0.95)	1.54 (0.93)	1.76 (1.00)	1.88 (1.00)	1.93 (0.96)	<0.001
Other caregiver’s stimulation
Low	57 194 (82.1)	45 794 (83.2)	8229 (80.0)	1244 (84.2)	1927 (68.6)	<0.001
High	12 443 (17.9)	9271 (16.8)	2059 (20.0)	233 (15.8)	880 (31.4)
Attendance to early childhood education programmes
No	55 915 (80.3)	46 043 (83.6)	7170 (69.7)	979 (66.3)	1723 (61.4)	<0.001
Yes	13 722 (19.7)	9022 (16.4)	3118 (30.3)	498 (33.7)	1084 (38.6)

Note. Parental stimulation was considered high if a mother or father engaged in four or more stimulation activities out of six in the past three3 days preceding the survey. (ECD) score is a continuous score ranging from 0 to 5. ECD z-score is a standardizedstandardised estimate having a mean of 0 and an standard deviationSD of 1 in the pooled dataset. P- values were estimated from ANOVA/chi-squareχ2 tests.

ANOVA, analysis of variance; ECD, early childhood development.

When children’s demographics were adjusted for in model 1, high stimulation by both mother and father was associated with the highest ECD z-score (b=0.13, 95% CI=0.10, 0.17), followed by the other categories of parental stimulation (b=0.07, 95% CI=0.05, 0.09 for mother high, father low; b=0.07, 95% CI=0.02, 0.12 for father high, mother low) (table 3). The associations attenuated after additionally adjusting for sociodemographic and learning-related characteristics. More specifically, high stimulation by one parent alone showed a beta size approximating zero and was no longer associated with ECD z-score in model 2 (b=0.01, 95% CI=−0.01, 0.03 for mother high, father low; b=0.00, 95% CI=−0.05, 0.05 for father high, mother low). Nevertheless, high stimulation by both parents was associated with 0.04 SD higher ECD z-score at statistically significant level (b=0.04, 95% CI=0.01, 0.08), when compared with low stimulation by both parents.

Table 3. Associations between parental stimulation and early childhood development z-score (N=69 637).

Variables	Beta coefficients (95% CI)
	Model 1	Model 2
Parental stimulation
Both low	Reference	Reference
Mother high, father low	0.07 (0.05, 0.09)***	0.01 (−0.01, 0.03)
Father high, mother low	0.07 (0.02, 0.12)**	0.00 (−0.05, 0.05)
Both high	0.13 (0.10, 0.17)***	0.04 (0.01, 0.08)*
Child’s sex
Girl	Reference	Reference
Boy	−0.10 (−0.11, to 0.09)***	−0.10 (−0.11, to 0.09)***
Child’s age (year)	0.16 (0.15, 0.18)***	0.15 (0.14, 0.16)***
Mother’s age (year)		0.00 (0.00, 0.00)
Father’s age (year)		0.00 (0.00, 0.00)
Mother’s education level
No education and primary		Reference
Secondary and higher		0.05 (0.02, 0.07)***
Father’s education level
No education and primary		Reference
Secondary and higher		0.04 (0.02, 0.06)***
Household wealth
Poorest		Reference
Poorer		0.03 (0.01, 0.05)**
Middle		0.05 (0.03, 0.07)***
Richer		0.06 (0.03, 0.09)***
Richest		0.11 (0.08, 0.15)***
Type of place of residence
Rural		Reference
Urban		−0.01 (−0.03, 0.02)
Number of children		−0.02 (−0.02, −0.01)***
Number of child books		0.02 (0.02, 0.03)***
Variety of toys		0.04 (0.03, 0.05)***
Other caregiver’s stimulation
Low		Reference
High		0.01 (−0.01, 0.03)
Attendance to early childhood education programmes
No		Reference
Yes		0.14 (0.12, 0.16)***

Note. Parental stimulation was considered high if a mother or father engaged in four or more stimulation activities out of six in the past three3 days preceding the survey. Early childhood development z-score is a standardizedstandardised estimate having a mean of 0 and an standard deviationSD of 1 in the pooled dataset. Child’s demographics (age, sex) were adjusted for in Mmodel 1. Parents’ age and education level, household wealth quintiles, type of place of residence, number of children, number of child books, variety of toys, other caregiver’s stimulation, and attendance to early childhood education programsprogrammes were additionally included in Mmodel 2. For all analyses, survey years were included. -values are signified with * at p; ** at p; *** at p.

^*p<0.05, **p<0.01, ***p<0.001.

When we assessed interaction between parental stimulation and socioeconomic, geographic and learning-related variables (figure 2), a significant interaction was found for household wealth only (interaction p=0.012). Children from wealthier households were associated with a 0.09 SD higher ECD z-score if both parents had high stimulation (b=0.09, 95% CI=0.04, 0.13), whereas the association was null for poorer households (b=−0.03, 95% CI=−0.09, 0.04). The association between parental stimulation and child’s ECD z-score did not substantially differ by other variables including mother’s education level (interaction p=0.435), type of place of residence (interaction p=0.062), other caregivers’ stimulation (interaction p=0.838) and attendance to ECE programmes (interaction p=0.106).

Figure 2. Stratified associations between parental stimulation and early childhood development score (N=69 637). Parental stimulation was considered high if a mother or father engaged in four or more stimulation activities out of six in the past 3 days preceding the survey. Children whose mother and father both had low stimulation served as the reference category. Early childhood development z-score is a standardised estimate having a mean of 0 and an SD of 1 in the pooled dataset. Child’s demographics (age, sex), parents’ age and education level, household wealth quintiles, type of place of residence, number of children, number of child books, variety of toys, other caregiver’s stimulation, attendance to early childhood education programmes and survey years were adjusted for in each regression model except for its stratification variable (eg, for regression (A), household wealth was not included). For each stratification variable, its interaction with parental stimulation was estimated, and the relevant p values from likelihood ratio tests are provided at the top left corner of each figure. P values for beta coefficients are signified with *p<0.05; **p<0.01; ***p<0.001.

The secondary analysis estimated the association between maternal and paternal stimulation and ECD z-score separately and found that the association was statistically significant only in model 1 (online supplemental table 1). For example, after adjusting for children’s demographics, the ECD z-score was 0.08 SD higher for mothers with high stimulation (b=0.08, 95% CI=0.06, 0.10) and 0.10 SD higher for fathers with high stimulation (b=0.10, 95% CI=0.07, 0.13). After additionally adjusting for sociodemographic characteristics and spouse’s stimulation in model 3, the size of associations became substantially small and statistical significance disappeared for both maternal (b=0.01, 95% CI=0.00, 0.03) and paternal stimulation (b=0.02, 95% CI=−0.01, 0.05).

The sensitivity analyses reported consistent findings when the ECD score was estimated using the 10-point scale (online supplemental table 2). The estimated association between ECD z-score based on the 10-point scale and high stimulation of both parents was statistically significant (b=0.17, 95% CI=0.13, 0.20) and substantially greater than the association for other stimulation categories (ie, mother high, father low; father high, mother low). However, our finding was sensitive to the operationalisation of high stimulation; for example, using a more liberal cut-off for parents’ stimulation whereby engagement in two or more stimulation activities for mothers and one or more activities for fathers was considered as high stimulation, it was no longer associated with the increase of ECD z-score (b=0.01, 95% CI=−0.01, 0.03 for both high).

Discussion

Using nationally representative data from 26 SSA countries, we estimated the prevalence of different combinations of maternal and paternal engagement in stimulation activities and their associations with ECD. The present study has three main findings. First, parental engagement in stimulation remains far inadequate in SSA. For example, in 80% of the sample, both parents had low stimulation, while 15% had high stimulation from mothers only and 2% had high stimulation from fathers only. Only 4% reported a high level of parental stimulation jointly, and this ranged from less than 1% in Gambia, Kenya and Senegal to 6%–9% in Mauritania, Nigeria, and Sao Tome and Principe. Overall, this finding was in line with prior work^{11 15} where most mothers and fathers were found to provide low stimulation in SSA; for instance, 15% and 4% of mothers and fathers in 23 SSA countries, respectively, were reported to engage in high stimulation, whereas their average across 62 LMICs was approximately threefold higher.¹⁵ Though various interventions have been implemented to improve parents’ caregiving practices in SSA,^{8 34} the current prevalence of early stimulation for children remains far behind Target 4.2 of the Sustainable Development Goals to ensure access to quality care for all children,³⁵ and hence calls for more targeted parenting programmes and policies in this region.

Second, we found that joint parental stimulation was associated with the child’s development—that is, children showed better developmental status when both parents provided high stimulation. This association remained statistically significant even after adjusting for socioeconomic characteristics, which were found to be related to ECD both in our results and consistently in prior literature.1,3 However, our findings should not be overinterpreted as the effect sizes were small in magnitude despite their statistical significance driven by large sample size and adequate statistical power. Previous studies reported positive associations between maternal stimulation and ECD.^{21 32 33 36 37} Evidence on paternal stimulation was more mixed, such that its association with ECD varied across studies in terms of its direction and magnitude, regardless of the inclusion of maternal stimulation in study models.^{11 19 21} We, however, attempted to explore the combination of maternal and paternal stimulation and its association with the child’s development. In fact, the increase of ECD z-score was found to be significant only for children whose parents had jointly high stimulation, but not for those whose mother or father alone had high stimulation. Joint parental stimulation can be translated into an absolute increase in the total amount of stimuli exposed to a child, directly related to a better developmental outcome. At the same time, stimulation from both mother and father is expected to improve the diversity of stimuli. The kinds of plays and learning activities usually vary by the gender of caregivers such that fathers are known to engage more in physical plays than mothers do.12,14 Interactions of different characteristics can stimulate different aspects of development, and hence the caregiving of both parents can fill the gap in childcare that their spouse does not generally address.

Our finding on the increase of ECD score only when both parents jointly engaged in high stimulation can be further explained by the family systems theory.³⁸ In a family system, all elements (ie, mother, father and child) are interdependent and one dyadic relationship (eg, mother–child) is influenced and even strengthened by other dyads (eg, father–child).³⁸ Within a triadic relationship involving parents and a child, interaction between mother and child, for example, is affected by a father–child dyad.³⁹ This is because the father’s engagement in childcare can function as interpersonal support to the mother and help her cope with stress and tension arising in the child-involving dyad.²² In fact, father’s support to his spouse was associated with smaller risk for maternal depressive symptoms in Kenya¹⁸ and Pakistan,²⁰ which can potentially affect mother’s childcare practices and ECD outcomes in the long term. Though the statistical interaction between maternal and paternal stimulation was insignificant in this study, our findings suggest that the parents’ caregiving can jointly benefit ECD within this triadic relationship.

Finally, the association between parental stimulation and ECD did not substantially differ by socioeconomic, geographic and learning-related variables except for household wealth. When stratified by household wealth, we observed that the association between high stimulation of both parents and ECD score was more pronounced among children from richer households. This finding—contrary to our initial hypothesis—may be attributed to the higher quality of caregiving practices in affluent households where parents may have better knowledge of childcare.⁴⁰ These parents may also have engaged in stimulation activities with longer duration and at greater frequency or purchased educational materials with better quality even at the same level of reported stimulation. On the other hand, there are poverty-related risk factors for ECD (eg, disadvantaged home/neighbourhood environment, child’s malnutrition, maternal depression) that may reduce the extent to which a child can benefit from parental stimulation in poorer households.⁴¹ Due to such complexity, findings on moderation by socioeconomic status from prior literature remain partly mixed, such that some studies supported more pronounced association between responsive caregiving and child’s development in socioeconomically disadvantaged settings,^{23 34} while others did not.⁴¹ Our findings suggest that responsive caregiving of parents may be important for fulfilling a child’s developmental potential across all socioeconomic groups for different reasons, necessitating future studies to explore the heterogeneity in this mechanism.

Despite the importance of parental stimulation, to date, only a limited number of parenting interventions engage fathers.^{34 42} For instance, the systematic review of Jeong et al³⁴ suggested that only 3 out of 41 trials of parenting interventions conducted in LMICs invited fathers as a key agent while others maintained their focus solely on mothers. A good example of holistic parenting interventions is the First Steps programme implemented in Rwanda,⁴³ which provided weekly education to both mothers and fathers with respect to responsive caring, stimulation and promotion of the child’s literacy via radio sessions. The programme resulted in improved engagement in learning/play, nurturing and positive discipline behaviours of both parents as well as the developmental status of their children.⁴³ As such, policy interventions aimed at fulfilling a child’s developmental potential, particularly through better parenting, should engage both parents and consider father-friendly delivery (ie, timing and location of programmes) and approaches (eg, gender-sensitive recruitment and monitoring) that can encourage fathers’ participation in intervention programmes.⁴²

A few important methodological limitations should be noted. First, parental stimulation was operationalised based on self-reported responses from mothers or primary caregivers and thus can be biased. Father’s stimulation, in particular, could have been underreported if a respondent (ie, mother) was unaware of the interaction between her spouse and children, or conversely, has the possibility of overreporting due to the social desirability bias.^{11 15} Moreover, a set of six stimulation activities measured through the MICS and DHS may not fully represent parents’ caregiving because it does not include the quality, frequency and other characteristics of stimulation (eg, duration, whether activities were led by a child, materials used in activities).^{11 15} We, however, made the best use of available data in the MICS and DHS, and these six activities were previously found to have acceptable reliability and predictive validity.⁴⁴ In our primary analysis, high stimulation was defined based on the MICS classification³⁵ and previous studies,^{16 29} but this could have been a high standard for parenting practices in SSA. Indeed, our sensitivity analysis suggests that results can be sensitive to the cut-off used to define high stimulation.

Second, although the ECDI has been validated across many LMICs,²⁷ the set of five cognitive and socioemotional development items is still limited in its ability to fully capture a child’s overall developmental potential.^{5 11 30} In fact, the advanced version of ECDI (ECDI2030) has been developed to reflect a child’s developmental capacity more comprehensively and was introduced to the MICS and DHS since 2020.⁴⁵ Including items for health, learning and psychosocial well-being, the ECDI2030 is expected to provide better estimates of a child’s developmental status.⁴⁵ Future studies should explore its validity and how components of nurturing care are associated with this newly measured ECD.

Third, the temporal order between parental stimulation and a child’s developmental status cannot be established due to the cross-sectional nature of the surveys employed in this study. Additionally, there can be residual confounding due to unobserved characteristics related to the child’s health and learning (eg, preterm birth, nutritional status, interaction with peers, quality of ECE programmes), who the primary caregiver is, other caregiving practices (eg, diets, healthcare use), family structures, and neighbourhood environments, and therefore we do not claim causality from our results.^{5 22} Fourth, we restricted the sample to children who were residing with their biological mother and father at the time of the survey, and thus the findings cannot be directly generalised to those in different family structures (eg, single-parent families, grandfamilies).¹¹

Conclusions

Across 26 SSA countries, the majority of young children are living in households where both the mother and father are engaged in low stimulation. Our results highlight the importance of joint stimulation by both parents for a child to reach their full developmental potential. Future studies should identify the barriers that discourage parents, particularly fathers, from fully engaging in stimulation activities for their children to better understand how to involve more fathers in childcare. Moreover, further research is needed to enrich our understanding about how and to what extent financial and social resources alter the effects of parental stimulation. Moving forward, policies and interventions aimed at improving parenting practices should holistically target both parents and transform traditional norms about parenting that prevent parents from fully participating in overall childcare activities, to achieve the most optimal developmental outcomes of children in SSA.

The funders had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript; and decision to submit the manuscript for publication.

Data availability statement

Data are available in a public, open access repository.