Examining the Long-Term Impact of Malaria Chemoprophylaxis in Everyday Life in Rural Gambian Adolescents

Abstract

Children in sub-Saharan Africa remain vulnerable to contracting malaria. While chemoprophylaxis is effective in preventing disease, its impact on social well-being is less understood. In this cross-sectional follow-up study conducted in 2001, we examined the role of chemoprophylaxis on social well-being in rural Gambia. Participants were 141 adolescents (age: 14.8–19.5; 40% male) from five villages, drawn from a longitudinal cohort in which one group received chemoprophylaxis in infancy, while the other received a placebo and began prophylaxis 1–2 years post-trial. Using a walking interview methodology, participants guided researchers through five village locations. Their narrations were coded for emotional content and social themes, followed by network analyses examining the relationships between these themes across treatment groups and gender. Emotional response analysis revealed significant gender differences in valence, with females exhibiting higher positive emotional tone than males (X² = 5.85, p = 0.016). Treatment effects showed gender-specific patterns: among males, the placebo group exhibited higher positive valence compared to the chemoprophylaxis group (X² = 8.34, p = 0.004), while females showed no treatment differences (X² = 1.11, p = 0.291). Affect analysis revealed high positive responses across all groups with no significant differences. Network analysis revealed significant gender differences in how adolescents organize social themes within their communities. Females demonstrated higher centrality in eigenvector (global influence; p = 0.039) and degree centrality (number of direct edges; p < 0.001), indicating greater interconnectedness and influence within the networks. No significant treatment group differences were observed in the network structure across any centrality indicators (all p > 0.05). This study provides an example of how the social–ecological framework and ecologically valid assessment methods, such as the walking interview, can be applied to investigate the interplay between early childhood health interventions, social dynamics, and individual development in a rural African context. The findings revealed that gender exerts a stronger influence than early treatment on adolescent social–emotional development.

1. Introduction

Nearly half of the world’s population, particularly those living in sub-Saharan Africa, Southeast Asia, and parts of Latin America, is at risk of contracting malaria. Approximately 200 million people get infected with malaria annually, with an average mortality rate of 0.24% [1]. This translates to roughly 696,000 deaths each year, emphasizing the critical need for prevention and treatment.

Preventing malaria involves a combination of vector control, personal protective measures, and preventative medications [1,2,3]. Among the strategies, chemoprophylaxis—the use of anti-malarial drugs to prevent infection—plays a major role, particularly for individuals in high-risk areas [4,5,6,7].

While chemoprophylaxis is effective in reducing the incidence of malaria, its impact extends beyond prevention. A meta-analysis by Angrist et al. (2023) [8], which included studies such as Clarke et al. (2008) [9]—who demonstrated significant effects of intermittent preventive treatment on educational outcomes in Kenyan schoolchildren—analyzed data from 4075 individuals across nine treatment groups. The meta-analysis registered significant positive effects of prophylaxis on cognitive skills (Cohen’s d = 0.12, 95% CI [0.08, 0.16]), particularly sustained attention. These findings suggest that malaria prevention through prophylaxis may have broader developmental benefits beyond physical health and well-being.

While the medical and developmental efficacy of chemoprophylaxis is well documented [10], its broader social and community impacts remain less understood, particularly in regions where traditional social structures play vital roles in daily life. This gap is particularly significant given that health interventions in community-centered societies may influence not only individual outcomes but also social relationships, community dynamics, and cultural practices that shape culturally specific adolescent development.

1.1. Gambian Culture and Norms

The Gambia offers an especially suitable setting for examining the broader social impacts of malaria prevention interventions. As Africa’s smallest mainland country with a population of 2.76 million people, it presents rich ethnic diversity with five major ethnic groups [11]. Malaria remains a significant public health problem in The Gambia, with an estimated 200,000 cases in 2021. This corresponds to an incident rate is 80.8 cases per 1000 people [12]. While the malaria incidence rate is below targets set by the WHO, the incidence rate of malaria in The Gambia has been increasing, highlighting the ongoing need for understanding the effects of these preventative strategies.

The majority (60%) of the population living in sub-Saharan Africa are rural farmers [13], operating within a collectivist social structure where family and village take precedence over individual pursuits, including formal education [14]. Even as educational opportunities expand for younger generations, societal values continue to prioritize family needs, primarily focused on household and agricultural tasks. While most Gambians traditionally lived in villages, about three-fifths of the population now resides in urban areas—though strong cultural ties and norms persist. This interconnection, combined with strong community-based culture and the continued malaria burden, provides a suitable setting for examining how health interventions interact with established social systems.

1.2. Gender Dynamics

Gender dynamics fundamentally shape social structures and daily interactions within rural Gambian communities, particularly influencing youth social development and community engagement [13]. Traditional Gambian society maintains distinct gender-based social structures that begin molding individual roles and relationships from childhood [15]. Women and adolescent girls typically maintain more extensive social networks centered around domestic duties, agricultural work, and childcare responsibilities [14]. These networks are reinforced through collective activities such as food preparation, water collection, and labor groups, fostering strong bonds across generations [16]. The formation of these gender-specific social networks begins early in childhood and intensifies during adolescence, with young girls progressively integrated into female social circles through participation in communal activities and sharing the workload [17].

In contrast, men’s and boys’ social networks often revolve around different spheres of community life, including religious activities, community decision-making, and income-generating ventures [18]. Young males typically experience more freedom in their social movements but have less structured social support systems compared to young females [19]. This gender distinction becomes particularly pronounced during adolescence, when boys begin to engage more in peer-based network activities, while girls become integrated into adult female social structures [20]. These gender differences in social organization and daily activities significantly influence how young people form and maintain social relationships, potentially affecting their response to health interventions [21].

1.3. Previous Research in Rural Gambia

Given this unique cultural and epidemiological context, The Gambia has served as an important site for malaria intervention research. In the 1980s, the World Health Organization (WHO) and the United Nations International Children’s Emergency Fund (UNICEF) recommended treatment alone due to concerns about the effects of chemoprophylaxis in children; yet, this recommendation was not supported by empirical evidence. Greenwood and colleagues (1988) [22] conducted a clinical trial comparing two malaria prevention strategies on 979 children aged 2 to 36 months. The first group received only treatment for presumptive malaria episodes, while the second group received both treatment and chemoprophylaxis. The combined intervention of treatment and chemoprophylaxis significantly reduced overall mortality in children (X² = 5.9, df = 1, p < 0.01), while treatment alone showed no significant effects on malaria-related mortality rates. This study demonstrated that combined treatment and chemoprophylaxis prevention offered a promising strategy for reducing child mortality from malaria.

Building on this foundational research, a follow-up study by Jukes and colleagues (2006) [23] investigated the long-term cognitive and educational impacts among a subset of recontacted participants (n = 579) from the original trial. The study was conducted in 10 villages near the bustling town of Farafenni—a strategic transit and commercial hub in the region. It employed a battery of five cognitive assessments adapted from Western psychological tools, validated for use with African populations, and administered in participants’ mother tongue to address methodological challenges in cross-cultural cognitive assessments. These included a visual search task, Raven’s Colored Progressive Matrices, digit span, categorical fluency, and vocabulary and proverb understanding.

The results showed no significant differences in cognitive function between the chemoprophylaxis and placebo groups (difference = 0.128, 95% CI [−0.052, 0.308], p = 0.162). However, a significant interaction was observed with the duration of post-trial prophylaxis (p = 0.034), where cognitive performance was higher in the prophylaxis group among those receiving no post-trial prophylaxis (effect = 0.24, 95% CI [0.02, 0.46]). This finding suggests complex timing effects of malaria prevention on developmental outcomes and highlights the importance of considering both the duration and timing of intervention when evaluating long-term impacts. While the 2006 follow-up provided valuable insights into the cognitive and educational outcomes of infancy chemoprophylaxis, it did not explore how such treatment might affect other domains of functioning in adolescence and adulthood.

1.4. Study Rationale

While the clinical efficacy of chemoprophylaxis is well-documented, its potential impacts on social constructs—particularly among adolescents in rural settings—remain understudied. This gap is particularly important to address, given the central role of social relationships in Gambian culture and the developmental significance of social connections during adolescence [21]. Understanding these broader social impacts requires methodological approaches capable of capturing the intersectionality between individual health interventions and community social dynamics within appropriate cultural contexts. This study employs walking interviews, an ecologically valid approach where participants guide researchers through meaningful locations in their village. This method, previously used in aging studies [24,25], is well-suited to rural populations in Africa, as it captures how these interventions influence daily life in community settings and aligns with Gambian culture traditions of meaningful conversations during walks or while sharing attaya (tea).

Drawing from our understanding of Gambian social structures and the differential roles of males and females in rural communities [14,26], we put forward several hypotheses. In terms of emotional responsiveness, we hypothesize that female participants will demonstrate higher overall positive valence and affect compared to males, reflecting gender differences in emotional expressivity. Second, we hypothesize that participants in the chemoprophylaxis group will exhibit more positive affect and valence than those in the placebo group. Additionally, given the distinct gender-based social structures and differential community roles described above, we expect that the effects of chemoprophylaxis on emotional responses will vary by gender, with males and females showing different patterns of treatment response.

In network analyses of social themes, we hypothesize that traditional gender roles in Gambian society will be reflected in centrality indicators. Given that females often maintain stronger ties to family and community networks through domestic responsibilities and agricultural labor [16,27], we expect them to demonstrate higher centrality in familial and sustenance-related themes. Conversely, males are expected to show higher centrality in social and cultural themes, reflecting established gender role divisions in rural Gambian society. This hypothesis is further supported by research on West African social structures, which highlights women’s role in maintaining community and family cohesion [19]. Second, we hypothesize that the impact of prophylaxis on social behavior patterns will vary by gender, with both genders exhibiting different centrality measures based on treatment groups. These predictions were evaluated using permutation-based inferential tests on network centrality metrics.

2. Methods

2.1. Theoretical Framework

Social–Ecological Model

The Social–Ecological Model (SEM) is an appropriate framework for understanding these complex dynamics. By viewing health as a complex web of physical, social, and mental factors, this framework addresses the task of health promotion while acknowledging how changes in one part of a social system can affect other aspects of an individual’s life. Recent applications of the SEM have demonstrated its effectiveness in understanding health-related community dynamics. Zhang and colleagues (2023) [28] conducted an extensive systematic review showing how the SEM effectively revealed factors affecting adults’ involvement in public open spaces (POS), considering various elements including economic resources, physical access, education levels, and social roles. Elder and colleagues (2007) explored this framework’s application to adolescent girls’ physical activity levels, demonstrating how individual health management can impact the surrounding social environment. Adolescent girls were reported to have declining rates of physical activity compared to their male counterparts. Because of this, they were at higher risk of physical and behavioral health problems, namely cardiovascular and mental health conditions. By implementing strategies to improve their activity levels, such as participating in non-competitive physical activity programs (e.g., walking club) and skill-building activities (e.g., problem-solving and goal-setting classes), the girls can foster a supportive community around them that encourages active pursuit of overall well-being by having confidence in their health and abilities. These activities, combined with environmental support through school-community partnership, enable females who were in this program to act as role models for the community and foster positive connections with friends and family.

The SEM is particularly relevant in the Gambian context due to the country’s emphasis on community and interdependence. Traditional Gambian society is characterized by a collective approach to life [14]. This aligns well with SEM’s multilevel perspective, which recognizes the interplay between individual, interpersonal, community, and societal factors. In Gambian culture, health and well-being are often viewed holistically, including not only physical health but also spiritual and social harmony [17]. By framing our research within the SEM, we aim to capture the ways in which malaria prophylaxis might influence not only individual health outcomes but also broader social dynamics in rural Gambia. To effectively implement this framework within the Gambian context, we employed the touring method as our primary data collection approach.

2.2. Touring Method

The touring method emerges as a particularly effective qualitative research technique when attempting to understand social–ecological dynamics. Often referred to as walking interviews, the touring method provides additional benefits compared to sedentary interviews, such as allowing for more participant comfort. Drawing from Carpiano’s (2009) [29] seminal work, the touring method provides a sophisticated means of investigation that closely aligns with the SEM, revealing the layered complexities of social-ecological interactions in real time. Specifically, as researchers walk alongside participants through their communities, they can peek into the physical, social, and cultural environments that shape daily experiences. This immersive approach enables the observation of subtle environmental influences that might be overlooked in traditional interview settings. Conducting research within participants’ natural settings may elicit richer responses about their personal connections to specific locations, as participants can directly reference and interact with meaningful places during the interview [30].

Multiple studies have demonstrated the effectiveness of the touring method [31], showing that it generates more spatially focused data and richer environmental insight compared to other assessment approaches. To illustrate, in a study exploring the social experiences of an elderly population, researchers found that this form of interviewing yielded much richer responses than expected, as it allowed them to glimpse the participants’ day-to-day routines [32]. Similarly, another study applied this method with another group of older adults to examine whether the neighborhoods in which participants lived influenced their social activity and mobility [33]. The researchers here also found the walking interviews to be highly efficient in eliciting in-depth qualitative responses.

Yet, there are also limitations to the touring method. Context and location are major factors in how effectively participants provide their responses [34]. The environment and physical landmarks must be relevant to the memories and experiences the researcher aims to evoke. Additionally, because these interviews are conducted in environments with less control over external factors, researchers may need to navigate unexpected obstacles during the interview—such as a passerby interrupting the conversation to speak with the participant.

While this method requires careful planning to address concerns such as weather conditions and physical accessibility, its ability to gather rich, contextual data makes it a valuable approach. The touring method facilitates a more equitable researcher–participant relationship by allowing participants to act as experts in their own environment. This is particularly important, as walking interviews shift traditional interview dynamics by enabling participants to lead and guide researchers through their space, rather than serving as passive respondents in a formal interview setting. This approach gives vulnerable, marginalized populations a larger voice in the research process [35]. Such methodological alignment strengthens social–ecological investigations by revealing hidden environmental barriers and facilitators that impact behavior, documenting the interplay between individual agency and structural constraints, and identifying specific locations where social determinants manifest in daily life.

In the Gambian context, the touring method holds particular promise due to its alignment with cultural traditions of oral communications and community-based interactions. The method may resonate deeply with Gambian culture, where meaningful conversations traditionally unfold during walks or while sharing attaya [13,14]. This approach naturally accommodates the communal nature of Gambian society, where social interactions often occur in shared outdoor spaces such as community wells and gathering trees [14,19]. Furthermore, by conducting interviews while walking through these culturally significant spaces, researchers can better understand how environmental and social factors specifically influence behaviors and decision-making processes within Gambian communities.

To investigate these hypotheses and apply the SEM theoretical framework, we designed a study that captured both the social-ecological context and the long-term impacts of malaria prevention among Gambian adolescents. By leveraging an existing longitudinal cohort of participants who had previously taken part in a malaria prophylaxis clinical trial, we examined the lasting effects of early childhood prophylaxis on social development and community engagement.

2.3. Participants

Participants were recruited from a longitudinal study that examined the long-term educational and cognitive impacts of early childhood malaria prophylaxis [23]. Of the original 579 participants, 141 (57 males and 87 females; mean age of 16.9, range 14.8–19.5 years) were able to complete the study. The walking interviews were conducted in 2001, approximately 13 years after the original clinical trial [22]. The final sample of 141 participants represents a convenience sample of reachable participants rather than a random subset of the original cohort. Multiple factors contributed to the substantial attrition from the original 579 participants, including migration of families, loss of contact, and logistical challenges during data collection. Participants were from 5 villages: Bambali (n = 48), Jajari (n = 40), Kani Kunda Suba (n = 18), Kani Kunda Wharf town (n = 14), and Kumbija (n = 21). Participants were divided into two groups based on the original infancy study. One group received malaria prophylaxis during the original trial (“Chemoprophylaxis”, n = 72), while the other group received a placebo during the trial period and then received active chemoprophylaxis one to two years after trial completion (“Placebo”, n = 69).

2.4. Village Walking Interview

Due to the rural location of our sample, the lack of access to educational resources, and minimal emphasis on educational attainment, the walking tour method was selected as the most effective approach to assess culturally relevant themes within these villages. All walking interviews were conducted by trained research assistants and lasted approximately 45 min. All interviews were conducted in participants’ preferred language, primarily Mandinka and Wolof. Verbatim field notes were translated into English prior to analysis. Given that some of the participants were minors, parental/caregivers’ consent was obtained prior to participation. To ensure authentic responses during these interviews, parents/caregivers did not actively participate. Approval from the Gambian Government was also obtained, and ethical approval was granted by the Medical Research Council Ethics Committee (MRC). We conducted 141 individual walking tours total—one tour per participant (72 tours with the chemoprophylaxis group (early treatment); 69 tours with the placebo group (delayed treatment)). Each participant completed their walking tour individually. At the beginning of each session, participants were given the following standardized instruction: “Now we’re going to go on a walk. I want you to show me your four favorite places around the village. Let’s start walking, and you bring me to any place that you like best”. Upon arriving at each location, the researcher would confirm with the participant using the prompt: “So this is one of your favorite places in the village?” Following confirmation, participants were given detailed instructions about describing activities: “Now, I want you to tell me all of the action words that are associated with this place/thing. In other words, I want you to name all of the things people do here/with this thing”. To ensure understanding, a standardized example was provided: “For example: If you showed me a school, could you name all the things that people do in the school, like teaching, learning, and writing”. There was no time limit for responses, and when a participant paused, they were encouraged with the prompt: “I know there are other things that people do here. Try to think of as many different things as you can”, The questioning for each location concluded only when the participants indicated that they could not think of any additional activities.

All interviews were conducted in the participants’ preferred language. All responses were documented verbatim by hand in a standardized field notebook. To maintain methodological consistency, each location had its own response sheet where researchers documented the specific questions asked, the participant’s complete response, and additional observations, including the use of present or past tense, verbal fluency, affect, valence, and the level of detail or knowledge shared. All participant responses were recorded verbatim, including any repetitions. This process was repeated for five different locations in the village, and all questions were administered orally. An example of the village tour worksheet can be found in Table S1. The analytical process involved systematic coding of participant responses across multiple dimensions. Initial coding categories were developed through a preliminary analysis of response patterns and consideration of cultural norms in rural Gambian communities. Two primary dimensions were coded: emotional content (affect and valence) and thematic elements.

2.5. Coding Process and Reliability

Affect, valence, and thematic coding were performed on the English translations using the adapted SPAFF5 rubric. The analysis involved systematic coding of participant responses across multiple time points by two trained coders who independently analyzed all transcripts. Initial coding categories were developed through a preliminary analysis of response patterns and consideration of cultural norms in rural Gambian communities. Two primary dimensions were coded: emotional content (affect and valence) and thematic elements.

2.6. Coder Training Protocol

Both coders underwent standardized training consisting of three phases. First, coders completed theoretical training on the coding framework, including detailed instructions on the distinction between affect and valence, the adapted Specific Affect Coding System (SPAFF5), and cultural considerations for interpreting responses in the Gambian context. Second, coders practiced a training set of 20 interview transcripts (not used in the analysis), discussing discrepancies and refining their understanding of the coding criteria through guided feedback. Finally, coders completed independent coding of a reliability subset before proceeding to the full dataset. The coders were trained to prioritize responses to the “Why” question when assessing affect, as these responses typically revealed participants’ emotional states and motivations for choosing specific locations. When “Why” responses were ambiguous or insufficient for affect coding, coders were instructed to refer to responses from the “What” category to provide additional context for determining emotional content.

2.7. Emotional Content Coding

Affect Assessment: Affect was operationalized using an adapted version of the Specific Affect Coding System (SPAFF5), modified for cross-cultural application in rural Gambian contexts. This system assesses expressed emotions or internal emotional states influencing behavior, specifically identified through emotional language, motivational statements, and expressions of personal feelings about locations (e.g., “I feel happy when I come here” or “This place makes me feel calm”). Affect was coded on a three-point scale (−1 = negative, 0 = neutral, 1 = positive) based primarily on participants’ responses to why they chose each location as meaningful, with reference to activity descriptions when motivational or affective content was unclear in the why answer.

Valence Assessment: Valence captured the emotional tone used to describe events, places, or activities themselves, coded using the same three-point scale based on evaluative adjectives and descriptive language in participants’ statements. Valence coding was primarily based on responses to the “Event” question, which captured participants’ descriptions of what happens at each location, supplemented by evaluative language about the locations and activities (e.g., describing events as “wonderful”, “peaceful”, or “difficult”).

2.8. Thematic Coding

Responses were analyzed for the presence of ten distinct themes: friendship, cultural, romance, sustenance, educational, social, commerce, familial, well-being, and undetermined. Each theme was coded using a binary system (0 = absent; 1 = present). Friendship encompassed peer relationships and non-community activities, while cultural themes included religious and traditional practices. Romantic references captured relationship-oriented content, and sustenance involved basic needs activities. Educational themes were restricted to formal learning contexts, excluding religious education. Social themes encompassed community-wide activities, distinct from friendship-based interactions. Commerce included all economic activities, while familial themes captured family-based interactions. Well-being referenced personal health and recreational activities. Responses not fitting these categories were coded as undetermined.

Table 1. Definition of Social Theme Categories.

Theme	Definition
Friendship	If the participants’ statements had any mention of a friend or any activities that involved friends and not the whole community
Cultural	If the participant mentioned religion or traditional ceremonies and rituals, or when spiritual views are expressed.
Romantic	If the participant mentions anything about any romantic interest, such as going to a location to meet a partner and starting a family.
Sustenance	If the participant mentioned activities important to sustenance, such as fetching water or food to cook.
Education	If they mentioned anything about education, such as schooling or gaining expertise training.
Social	This category is different from the friendship category as this encompasses more of the community instead of only friends. Cultural and community events such as drinking attaya (i.e., tea-sharing community).
Commerce	If the participant mentions anything about professions, trade, or any monetary exchange
Familial	Anything that involves a family member (regardless of the distance in relation) or even if a friend is mentioned along with a family member.
Well-Being	If a participant mentioned anything about personal health or well-being or wanted to do an activity just to relax and have fun.
Neutral	If no statement was given or they did not mention any of the previous categories.

provides detailed definitions for each thematic category to illustrate the coding framework.

2.9. Disagreement Resolution Protocol

When disagreements arose between coders, a structured resolution process was implemented. First coders independently reviewed the disputed response and provided justification for their coding decision. Second, coders met to discuss the disagreement, referencing the coding manual and training materials. If a consensus was reached through discussion, the agreed-upon code would be applied. For cases where coders could not reach an agreement (fewer than 2% of all responses), the response was coded conservatively.

2.10. Reliability

Reliability analysis demonstrated strong consistency across all measures. Cohen’s kappa for interrater agreement between the two trained coders showed an excellent reliability for affect coding using the adapted SPAFF5 (K = 0.85), strong reliability for valence coding (K = 0.78), and strong reliability for thematic coding across all categories (K = 0.88).

2.11. Analytic Approach

Building upon this coding framework, our analysis aimed to identify potential differences in well-being-related behaviors between individuals who received malaria prophylaxis and those who received a placebo in the original infancy trial [29]. Specifically, this analysis focuses on understanding how prophylaxis might influence participants’ social engagement, community involvement, and emotional responses to activities in their village.

The analysis followed a three-stage approach. First, descriptive statistics were calculated to summarize the frequency and distribution of social themes and emotional responses across all visits. Following this, frequency analyses were conducted to examine patterns of affect and valence for each location. Sign tests were then used to assess the significance of differences between positive and negative responses within each group (treatment vs. placebo) and gender. Additionally, two-sample proportion tests were conducted to compare these frequencies between groups, which allowed us to examine whether treatment status or gender influenced the distribution of emotional responses.

Finally, network analysis was implemented using the igraph package in R [30] to explore the relationships between social themes. In this analysis, themes were represented as nodes, with edges weighted by the frequency of co-occurrence between themes in participant responses. We selected three centrality measures to understand different aspects of theme importance in these networks. Eigenvector centrality was chosen to identify which themes were most influential in the overall social structure, so as to measure not just direct connections but also account for the influence through the social network. Centrality was selected to identify themes that served as bridges between different aspects of village life, and degree centrality was chosen to measure the number of a node’s direct edges connected to the node.

The network analysis was conducted separately for the prophylaxis and placebo groups, as well as gender, to identify potential differences in social theme clustering and importance. To determine if any observed differences between groups were meaningful rather than due to chance, we employed an edge randomization permutation test with 1000 iterations. For each iteration, group labels were randomly reassigned, and centrality measures recalculated, creating a null distribution that represented what we would expect to see if treatment or gender had no effect on social networks. A p-value was calculated as the proportion of permuted differences that exceeded the observed difference, allowing us to identify the true impacts of prophylaxis and gender on social structure. All statistical analyses were conducted using R version 4.1.0 [31], with a significance level of α = 0.05.

3. Results

3.1. Emotional Response

Longitudinal categorical plots were created to visualize participant movement across village locations by treatment groups and gender. Two visualizations were developed, one examining emotional affect (Figure 1) and another examining valence (Figure 2). Both plots revealed the distribution of participants’ emotional experiences during village tours, with color-coding representing emotional states (negative, neutral, or positive).

3.2. Emotional Analysis

Gender analysis revealed significant differences in emotional valence responses. Females demonstrated significantly higher positive valence compared to males (72.5% vs. 62.1%, X² = 5.85, p = 0.016, 95% CI [−0.189, −0.018], Cramér’s V = 0.10). Treatment group analysis revealed important gender-specific patterns.

Among males, significant differences emerged between treatment groups, with placebo males showing higher positive valence than chemoprophylaxis males (72.4% vs. 53.3%, X² = 8.34, p = 0.004, 95% CI [−0.337, −0.065], Cramér’s V = 0.20). Correspondingly, chemoprophylaxis males showed significantly higher negative valence compared to placebo males (47.7% vs. 27.6%, X² = 8.34, df = 1, p = 0.004, 95% CI [−0.337, −0.065], Cramér’s V = 0.20). In contrast, females showed no significant differences between treatment groups in valence responses (X² = 1.11, p = 0.291, 95% CI [−0.047, 0.167], Cramér’s V = 0.04).

Analysis of emotional affect (Figure 1) revealed high levels of positive responses across all groups when compared to negative responses. Gender comparison showed no significant differences in positive affect between males and females (96.9% vs. 94.4%, X² < 0.001, df = 1, p = 1.00, 95% CI [−0.112, 0.170], Cramér’s V = 0.00). Treatment group comparisons for affect showed no significant differences in any comparison (all p = 1.00, all Cramér’s V = 0.00).

3.3. Network Analysis

Network analyses revealed distinct patterns in social theme relationships across four gender x treatment combinations: female chemoprophylaxis, female placebo, male chemoprophylaxis, and male placebo. To contextualize the centrality results,

Table 2. Structural properties of treatment groups and gender-based social theme networks.

Network	Nodes	Edges	Density	Avg. Total Degree	Avg. Weighted Degree	Reciprocity	Transitivity (Global)
Male Chemoprophylaxis	9	41	0.57	9.11	25.78	0.79	0.66
Male Placebo	9	48	0.67	10.67	24.89	0.65	0.83
Female Chemoprophylaxis	10	56	0.62	11.20	36.80	0.78	0.81
Female Placebo	9	47	0.65	10.44	33.78	0.75	0.81

Note. Density = proportion of possible connections present; Reciprocity = proportion of mutual connections; Transitivity (Global) = overall tendency for nodes to form closed triangles (clustering).

summarizes basic network structural characteristics, including node count, edge count, density, and average degree. Supplementary Figures S3–S6 present weighted adjacency heatmaps for each gender x treatment network, illustrating connection strengths between domains, while Supplementary Figures S7–S9 display bar plots of betweenness, degree, and eigenvector centrality for all nodes. Each network comprises a set of nodes representing key domains—social, familial, sustenance, culture, friendship, commerce, well-being, romantic, neutral, and education—connected by directed edges indicating influence. Three centrality measures were computed to assess nodes’ structural importance: betweenness centrality (intermediary role), eigenvector centrality (global influence), and degree centrality (number of direct edges to the node).

Comparison of chemoprophylaxis versus placebo networks (Figures S1 and S2) revealed no significant differences between groups. The female chemoprophylaxis network (Figure 3a) revealed three distinct clusters, each representing interrelated domains with a varying degree of connectivity. The first cluster, highlighted in green, includes neutral, cultural, well-being, friendship, and commerce. Within this cluster, sustenance and neutrality emerge as central nodes, each with multiple bidirectional connections to cultural and friendship. Well-being and Commerce are more peripheral, with limited connectivity primarily directed toward neutral or sustenance. The second cluster, shaded in blue, comprises social, education, and romance, centered around the social node. The third cluster, highlighted in red, includes Familial and Friendship. Centrality analysis revealed familial as the most globally influential node in this network (eigenvector = 1.00), followed by Friendship (0.97) and Social (0.80). Cultural themes served as an intermediary across nodes (betweenness = 25.93), followed by Social (20.27). Degree centrality analysis revealed social and neutrality as the most efficient nodes (degree = 17), followed by familial (15) and tied scores for friendship and sustenance (14).

The female placebo network (Figure 3b) showed three clusters. The first cluster, highlighted in green, contains commerce, social, and cultural nodes. The second cluster, highlighted in red, consists of romantic, familial, and friendship nodes. And our last cluster, highlighted in blue, consists of neutral, well-being, and sustenance nodes. Sustenance emerged as one of the most influential nodes (eigenvector = 1.00), with friendship (0.98) and familial (0.73) also showing high influence. Commerce functioned as the key intermediary (betweenness = 13.67), followed by neutral (9.17). Sustenance was also the most efficient node (degree = 15), followed by well-being (14), and tied scores for friendship and education (13).

The male chemoprophylaxis network (Figure 3c) displayed the most fragmented structure, with three distinct clusters. The first cluster, highlighted in red, comprises well-being, culture, friendship, sustenance, neutral, and commerce nodes. The second cluster, highlighted in green, comprises Education and Familial nodes. The third cluster comprises only the social node. Despite its isolation, social maintained the highest global influence (eigenvector = 1.00), followed by friendship (0.39). Familial served as the primary intermediary (betweenness = 15.50), with neutral (12.67) and sustenance (12.33) also playing important bridging roles. Well-being had the highest degree centrality in the network (degree = 14), followed by friendship (13), and tied scores for education and culture (10).

The male placebo network (Figure 3d) also displayed three distinct clusters in the network. The first cluster, highlighted in green, contains cultural, education, social, commerce, sustenance, and neutral nodes. The second cluster, highlighted in blue, contains familial and well-being nodes. The third cluster comprises only the friendship node. Social demonstrated the highest global influence (eigenvector = 1.00), followed by friendship (0.59) and neutral (0.45). Education functioned as the primary intermediary (betweenness = 20.74), with familial (10.79) and social (10.33) also serving important bridging roles. Education has the highest degree centrality (degree = 17), followed by tied scores for friendship and well-being (14).

Permutation testing with 1000 iterations assessed the significance of network differences between groups. Treatment group comparison revealed no significant differences in network structure between chemoprophylaxis and placebo groups across all centrality measures: eigenvector centrality (p = 0.184, observed difference = 0.091), betweenness centrality (p = 0.071, observed difference = 2.088), and degree centrality (p = 1.00, observed difference = 0.6). Gender comparisons revealed significant differences in network organization. Males and females showed significant differences in eigenvector centrality (p = 0.039, observed difference = 0.158) and degree centrality (p < 0.001, observed difference = 2.437. But no significant differences were observed for betweenness centrality (p = 0.067, observed difference = 2.437). Gender x treatment interaction analysis showed no significant interactions across centrality measures: eigenvector centrality (p = 0.585), betweenness centrality (p = 0.279), and degree centrality (p = 0.655).

4. Discussion

In this study, we observed significant gender-specific differences in emotional responsiveness and social network patterns among rural Gambian adolescents who participated in an early childhood chemoprophylaxis clinical trial [22]. Using a longitudinal cohort and the walking interview methodology grounded in the SEM framework [36], we hypothesized that (1) females would demonstrate higher positive affect and valence than placebo participants; (2) chemoprophylaxis participants would exhibit more positive affect and valence than placebo participants; (3) chemoprophylaxis effects on emotional support would differ by gender; (4) females would exhibit higher centrality in familial and sustenance themes while males would exhibit higher centrality in social and cultural themes; and (5) chemoprophylaxis effects on social network centrality would differ by gender.

Our analysis of emotional responsiveness revealed a gender x treatment interaction that was not anticipated in our original hypothesis. While both genders exhibited predominantly positive emotional responses to village locations, significant treatment effects emerged only among males. Males who received chemoprophylaxis demonstrated significantly more negative emotional valence compared to placebo males (47.7% vs. 27.6%, p = 0.004), while females showed no treatment-related differences regardless of intervention status (p = 0.291). These findings align with emerging evidence suggesting that males demonstrate greater sensitivity to early environmental influences during critical developmental periods [37,38]. The differential treatment effects observed between genders may reflect sex-specific developmental trajectories in emotional regulation systems, with males showing prolonged vulnerability to early intervention effects that persist into adolescence [39].

The absence of treatment effects in females may reflect several protective mechanisms. Studies have shown that females typically show earlier maturation of emotional regulation systems and greater social connectedness, which may buffer against any potential negative effects of delayed interventions [40,41]. Additionally, cultural factors in rural Gambian society may provide differential social support systems for males and females that interact with biological predisposition [16]. Specifically, in rural Gambian contexts, these protective social support systems emerge through gendered socialization that shapes both the structure of social networks and norms around social expression [14,42]. From early adolescence, girls’ integration into domestic and agricultural labor groups reinforces sustained engagement in cooperative and supportive environments. This may encourage a positive affective tone [43]. In contrast, boys’ participation in more episodic, task-oriented community roles, often tied to religious or income-generating activities, may provide fewer opportunities for ongoing emotional reinforcement within their social networks [14]. These socio-cultural dynamics likely interact with biological differences to produce the observed behavioral gender patterns.

The walking interview methodology proved effective in capturing ecologically valid emotional responses, as evidenced by the high interrater reliability achieved (K = 0.85 for affect, K = 0.78 and valence) and the detection of meaningful between-group differences that would not have emerged if responses were simple due to social desirability bias. The ability to detect this gender x treatment interaction suggests that participants provided authentic emotional responses rather than culturally expected answers. This methodological approach aligns with previous research demonstrating how mobile interviewing techniques can facilitate more natural and content-rich responses [29,30], particularly in settings where traditional interview methods might create cultural barriers.

4.1. Social Network Structures

Regarding our hypotheses about social network structures, the results provided partial support for our predictions while revealing some other interesting patterns. The network analysis showed that gender significantly influenced social theme organization, with female adolescents demonstrating different centrality across multiple measures compared to their male counterparts (eigenvector centrality p = 0.039, degree centrality p < 0.001). This finding is consistent with traditional gender roles in Gambian society, where females maintain stronger familial and sustenance themes [14,26].

Contrary to our expectations, the network analysis revealed no significant treatment effects on social theme organization across any centrality measures (all p > 0.05). While descriptive differences existed between the treatment groups in terms of cluster patterns and node isolation, permutation testing confirmed that these differences were not statistically significant. This finding suggests that early childhood malaria interventions do not fundamentally alter how adolescents organize social themes within their community environments, despite affecting emotional responses. But our findings do suggest that variation in network centrality is more likely driven by gendered socio-cultural norms in rural Gambia than by early-life prophylaxis timing.

The preservation of social network organization despite emotional differences may reflect the dissociation between emotional processing systems and social cognitive frameworks [42,44]. This pattern suggests that while early intervention may influence emotional regulation systems, it may not disrupt core social developmental processes that are used for community integration and cultural transmission [45]. The gender differences in network centrality align with existing research on gender roles in rural African societies. The finding that females demonstrate higher centrality for the familial and sustenance themes suggests that girls play a more central role in community and family cohesion, likely due to their traditional responsibilities in familial and cultural spheres [14].

When considered through the lens of the Social–Ecological Model [36], our findings can be understood as operating across multiple ecological levels. At the individual level, early-life malaria chemoprophylaxis appears to influence aspects of emotional regulation—particularly in male adolescents—as evidenced by the higher proportion of positive affect observed in the early chemoprophylaxis compared to the placebo group. At the interpersonal level, gender differences in social network centrality, such as females’ higher eigenvector and degree centrality for familial and sustenance themes, reflect the influence of culturally structured roles and expectations that shape day-to-day interactions. At the community level, the persistence of traditional gender roles in rural Gambia, regardless of treatment status, highlights the role of broader socio-cultural systems in maintaining stable patterns of social organization. This multilevel framing underscores that health interventions may interact with, but not necessarily alter, entrenched social structures, which has important implications for intervention design and evaluation.

4.2. Limitations

While this study provides novel insights into the long-term social–emotional effects of early childhood malaria prevention, several limitations should be noted. First, the cross-sectional design limits causal inferences about the relationship between early childhood treatment and adolescent outcomes. Although participants were drawn from a well-characterized longitudinal cohort, the temporal gap between intervention and assessment (approximately 12–14 years) introduces potential confounding factors that could influence the observed relationship. Second, attrition represented a major limitation in this follow-up study. The final sample of 141 participants represents only 25% of the original 579 children contacted in the follow-up [23]. Multiple factors contributed to this substantial participant loss, including migration of families from study villages, changes in contact information, mortality, and the loss of our survey during data collection. Third, participants who remained available for follow-up may differ from those lost to attrition in ways that could significantly influence social–emotional outcomes. Families who remained in the same villages for over a decade may have different socioeconomic stability, health outcomes, or social support systems compared to those who migrated or were lost to follow-up. These differences could bias findings toward individuals with stronger community integration, potentially underestimating the variability present in the original cohort. Fourth, an important limitation is that all participants eventually received chemoprophylaxis treatment, although the timing differed between groups. The chemoprophylaxis group received treatment at the beginning of the original clinical trial [22], whereas the placebo group received it after the trial concluded. This timing difference complicates clear definitions of treatment groups but also highlights that delay in treatment may impact the development of social–emotional responses. Fifth, village-level factors may have influenced our findings, as the study design did not control for differences in community-specific cultural practices across the five villages. Sixth, the use of translated field notes may not capture all of the nuances of idiomatic speech. Seventh, the focus on a predominantly rural region limits the generalizability of findings to other cultural contexts or urban African populations. Cultural factors influencing gender roles, social organization, and emotional expression may vary significantly across different African societies, potentially affecting the magnitude or direction of observed effects. Additionally, other potential confounding variables, such as education level and socioeconomic status, were not statistically controlled for in the present analyses. These factors may influence emotional expression and social network structures and should be studied in future research. Lastly, because emotional responses and social themes were derived from self-reported narratives, results may be influenced by recall bias or social desirability effects. The use of thematic coding and network centrality metrics also introduces interpretive subjectivity, as coding decisions and analytical choices can shape the resulting network structure. While coder training and permutation testing were employed to mitigate these concerns, some residual bias is possible.

5. Conclusions

In conclusion, this study provides additional evidence that early childhood malaria chemoprophylaxis may have differential long-term effects on social–emotional development that persist into adolescence and are moderated by gender. The findings demonstrate domain-specific effects: emotional processing showed gender-by-treatment interactions, while social network organization exhibited the main effects of gender but no treatment or interaction effects. The preservation of social network organization despite differences in emotional processing suggests that early childhood malaria prevention does not disrupt core social developmental processes, while highlighting the importance of considering gender-specific vulnerability in intervention design and follow-up. To translate these insights into actionable strategies for public health in malaria-endemic regions like rural Gambia, we recommend the integration of gender-sensitive social–emotional screening into routine chemoprophylaxis programs. For example, community health workers using adapted walking interview tools may identify males at risk of negative emotional outcomes from delayed intervention. For policymakers, funding for longitudinal monitoring of non-clinical outcomes is essential to ensure equitable gender-conscious interventions.

The walking interview methodology proved effective in detecting meaningful differences between groups. By examining these relationships through a social-ecological lens, we identified gender as a critical moderating factor influencing vulnerability to early intervention effects in specific developmental domains, particularly emotional responsiveness. This integrated approach not only advances understanding of intervention impacts beyond immediate health outcomes but also provides a methodological framework for future research examining the long-term social impacts of public health initiatives. Future studies should continue to explore these methodological tools while considering how local cultural contexts and traditional social structures interact with health interventions to shape developmental trajectories.