2.1. Search strategy

This systematic review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [10] framework and followed Cochrane standards for systematic reviews and meta-analyses. A comprehensive search was conducted in MEDLINE/PubMed, EMBASE, Cochrane Library and LILACS for publications up to December 12, 2025, focusing on the Health education in the prevention and management of infectious diarrhea in childhood. Search terms included combinations of “Diarrhea”, “Health Education”, “Child, Preschool” and “Diarrhea/prevention and control”. The full search strategies are presented in S1 Table. In addition to electronic database searches, a manual search of the reference lists of all included studies and relevant reviews was performed to identify additional eligible trials. The study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO), bearing the registration number CRD420250643885.

2.2. Inclusion and exclusion criteria

Original studies published in any year with full-text availability, regardless of the language of publication, were included. Eligible studies consisted of randomized clinical trials. The population of interest included mothers/caregivers of children under 5 years of age who had access to health education methods (e.g., face-to-face, audiovisual, digital interventions) for the prevention of acute infectious diarrhea. No minimum duration of intervention exposure was predefined, as educational strategies varied substantially across studies. Clinical outcomes of interest included knowledge regarding methods of childhood diarrhea prevention, the number of cases, and adoption of daily hygienic practices following the educational intervention, as well as other outcomes potentially referenced in the articles. Outcome definitions adhered to those provided in the included studies. Cohort studies, case reports, case-control studies, case series, literature reviews, and systematic reviews were excluded. Studies exclusively focused on pharmacological prophylaxis or non-educational preventive interventions were excluded.

2.3. Identification and selection of studies

Two authors independently conducted the search, selection, and application of eligibility criteria. The selection process occurred in three stages: (1) removal of duplicates; (2) exclusion of articles that did not meet eligibility criteria based on title and abstract review; (3) and subsequently, full-text review of selected articles with re-application of eligibility criteria to assess their quality and relevance to the proposed objective. Any disagreements between the authors were resolved through discussion and dialogue, with a third author present. The “Rayyan QCRI” tool was used for article selection, and the two researchers were blinded to each other’s decisions throughout the process.

2.4. Data extraction

Data extraction was conducted using a predefined data collection form. The following characteristics were collected: study title, reference, country, center, publication year, follow-up period, sample size, and the number of individuals lost to follow-up. Additionally, data on health education methods, topics addressed, implementation and monitoring processes, and participant demographics—including age, participants per group, caregivers’ marital status, education levels, and basic sanitation conditions—were gathered. All outcomes evaluated in the studies were extracted, with an emphasis on health education’s role in preventing and managing childhood infectious diarrhea. Caregiver-reported and other qualitative outcomes were extracted and analyzed according to the definitions, instruments, and criteria adopted in each original study. Two authors independently performed data extraction, and discrepancies were resolved through discussion with a third author.

2.5. Risk of bias assessment

The risk of bias assessment included the evaluation of methods for randomization, treatment allocation, blinding, selection and comparability of study groups, and outcome evaluation, conducted at the study level. The following tools were used: the Cochrane Risk of Bias for Randomized Trials (RoB-2) for randomized controlled trials. Judgments were classified as “low risk,” “some concerns,” or “high risk.” These assessments were independently performed by two authors, and discrepancies were resolved through discussion with a third author. No statistical analysis was performed due to the heterogeneity among the included studies.

2.6. Data synthesis

Due to heterogeneity in intervention types, follow-up periods, and outcome definitions, meta-analysis and pooled quantitative analyses were not performed. This heterogeneity was particularly related to differences in effect measures and follow-up periods, as well as variability in the content and delivery of educational interventions. Outcomes were reported using different analytical measures, including incidence, prevalence, incidence density, knowledge scores, self-efficacy scales, and behavioral indicators, which limited comparability across studies and precluded meaningful quantitative synthesis. Instead, results were summarized using a narrative synthesis, grouping studies by type of educational intervention and outcome measures.

3.1. Identification and selection of studies

A total of 1,917 records were initially identified across databases. After removing 543 duplicates, 1,374 studies were screened for title and abstract. 24 studies underwent full-text review, with nine excluded based on predefined eligibility criteria. Ultimately, 15 studies met all inclusion criteria and were included in the qualitative analysis, as illustrated in Fig 1.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA [10] flow diagram of study selection.

https://doi.org/10.1371/journal.pntd.0014442.g001

3.2. Study characteristics

The main characteristics and outcomes of the included studies are summarized in Tables 1 and 2. The final analysis included 15 controlled trials evaluating community-based interventions for the prevention of childhood diarrhea (Table 1). Collectively, these studies enrolled 44,794 caregivers of children between 1985 and 2018. Sample sizes ranged from 51 to 16,301 participants. The studies were geographically diverse, including four conducted in Asia (China, India, Pakistan, Nepal), four in Africa (Gambia, Ghana, Tanzania, Ethiopia), three in Latin America (Brazil, Bolivia, Honduras), and others in South Asia (Bangladesh).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics of included studies.

https://doi.org/10.1371/journal.pntd.0014442.t001

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of the study populations, objectives, and outcomes.

https://doi.org/10.1371/journal.pntd.0014442.t002

Follow-up duration exhibited considerable variability, ranging from 2 to 32 months. The majority (n = 9) implemented follow-up periods of 3–12 months, while others maintained 24-months follow-up (Guo et al., 2018 [11]; Oles et al., 2024 [12]), and Manaseki-Holland et al. (2021) [13] extended follow-up to 32 months. Most studies reported caregiver demographics, with ages typically ranging from early adulthood to middle age, although some studies did not specify age ranges (n = 10), and were therefore recorded as “NA” in Table 1.

The majority of studies (n = 11) employed a two-arm design comparing control versus intervention groups, including Luo et al. (2019) [14], Ma et al. (2019) [15], Hashi et al. (2017) [16], and Luby et al. (2004) [17]. Others adopted more complex multi-arm approaches (n = 4), such as Briceño et al. [18], who compared handwashing, sanitation, and combined interventions against a control, while Penha et al. (2022) [5] and Lindquist et al.(2014) [19], evaluated multiple intervention combinations.

The studies included in this review examined population characteristics, intervention objectives, and outcomes related to childhood diarrhea prevention (Table 2). Most interventions targeted caregivers of children under five years of age, with eligibility criteria including child’s age, household location, and caregivers consent to participate.

3.3. Interventions and outcomes

The interventions aimed to reduce diarrhea incidence or improve hygiene practices. Oles et al. (2024) [12] and Penha et al. (2022) [5] focused on maternal education, whereas Manaseki-Holland et al. (2021) [13] and Luo et al. (2019) [14] implemented food safety and parenting programs. Several studies, including Ma et al. (2019) [15], Hashi et al. (2017) [16], and Briceño et al. (2017) [18], assessed diarrhea prevalence as the primary outcome, while Guo et al. (2018) [11] and Kapoor et al. (2016) [20] evaluated hygiene-related behaviors.

Secondary outcomes varied across studies, including respiratory infections (Manaseki-Holland et al.,2021) [13], water quality (Hashi et al., 2017) [16], and caregiver knowledge or satisfaction (Ansari et al.,2012 [22]; Bloch et al., 2013 [23]). These caregiver-reported outcomes were based on structured questionnaires or interviews assessing knowledge acquisition, perceived caregiving practices, and satisfaction with the intervention. In contrast, Penha et al. (2022) [5], Lindquist et al. (2014) [19], and Luby et al. (2004) [17] did not report secondary outcomes.

3.4. Narrative synthesis of intervention strategies

The educational strategies and corresponding outcomes are synthesized in Table 3. The studies employed diverse health education strategies to prevent and manage childhood diarrhea (Table 3). Delivery methods varied, ranging from traditional face-to-face sessions and printed materials, audiovisual resources, community-based visits, and Behavior Change Communication (BCC) campaigns.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Characteristics of the health education method used.

https://doi.org/10.1371/journal.pntd.0014442.t003

Three studies implemented home visits as the primary strategy for delivering educational content. Oles et al. (2024) [12], Luo et al. (2019) [14], and Ma et al. (2019) [15] used structured home-based approaches led by trained community health workers or volunteers, supplemented with visual or digital tools. These interventions focused on hygiene practices, diarrhea prevention, breastfeeding, and child nutrition, often incorporating regular monitoring and caregiver feedback mechanisms.

Other studies emphasized group-based educational tools or community engagement strategies. Manaseki-Holland et al. (2021) [13] used performing arts and public meetings, while Penha et al. (2022) [5] applied video and booklet formats based on self-efficacy theory. Hashi et al. (2017) [16] and Guo et al. (2018) [11] conducted face-to-face educational sessions, reinforced by printed materials, community demonstrations, and public broadcasting (e.g., loudspeakers, posters, or radio).

BCC models were central to studies such as Lindquist et al. (2014) [19] and Briceño et al. (2017) [18], which trained local agents to promote hygiene practices and sanitation through structured community interventions. Similarly, Nicholson et al. (2014) [23] applied school- and household-based activities to reinforce child handwashing habits, observing measurable spillover effects on family health.

Bloch et al. (2013) [22] and Ansari et al. (2012) [21] interventions focused specifically on improving caregiver knowledge, using videos, questionnaires, and illustrated sessions to enhance understanding of diagnosis, home care, and signs dehydration signs. Luby et al. (2004) [17] and Stanton et al. (1988) [24] combined education with product distribution, providing printed materials and soap to reinforce hygiene practices.

Overall, 12 studies reported positive outcomes, including reduced diarrhea incidence, improved hygiene behavior, and increased caregiver knowledge. However, a few studies noted that while intermediate behavioral outcomes improved, the interventions alone did not yield significant health impacts (e.g., Briceño et al., 2017 [18]). Monitoring approaches included questionnaires, symptom tracking, unannounced visits, and biological indicators, such as hand swabs or water testing. These outcomes were assessed using caregiver-reported measures related to knowledge, practices, and satisfaction.

3.5. Analytical synthesis of the included studies

Demand-side interventions demonstrated robust efficacy in reducing diarrheal disease, with the highest magnitude of effect observed in structural hardware provision. In periurban Bolivia, the installation of hollow fiber water filters was associated with a diarrheal prevalencer ratio of 0.21 (95% CI 0.15 -0.30), representing a 79% reduction in prevalence. Notably, the addition of hygiene BCC did not statistically enhance the filter’s efficacy (PR 0.27, 95% CI 0.22–0.34), while BCC alone produced a non-significant reduction (PR 0.71, 95% CI 0.59–0.86) [18]. Intensive Handwashing with Soap promotion showed high strength of association across urban and rural cohorts. In Karachi, Pakistan, the intervention achieved a 53% lower incidence of diarrhea (95% CI -65% to -41%) [19]. In rural Ethiopia, Hashi et al. reported a longitudinal adjusted incidence rate ratio of 0.65 (95% CI 0.57–0.73), confirming a 35% reduction in diarrhea incidence. Furthermore, a hazard analysis critical control point food safety program in Gambia reduced reported diarrhea with an adjusted relative risk (aRR) of 0.39 (95% CI 0.32 -- 0.48) at 6 months [16]. The benefit of hygiene education extended to viral enteric pathogens. In China, Guo et al. reported that intensive education reduced the incidence of hand, foot, and mouth disease from 4.2% in control groups to 2.1% in intervention groups (x2 = 22.138, p < 0.001). This effect was supported by laboratory findings, with significantly lower coliform contamination detected on hand swabs in the intervention group compared with controls (2.0% vs. 9.45%) [11].

Synthesis of secondary outcomes reveals significant protective effects beyond enteric pathologies. Handwashing promotion in Mumbai, India, was associated witha 15% reduction in episodes of acute respiratory infection (ARI) episodes (95% CI -30% to -8%) and a 46% reduction in eye infections (95% CI -58% to -31%) among 5-year-olds children [23]. Similarly, the Gambia trial observed an aRR for ARI of 0.67 (95% CI 0.53 - 0.86). These data suggest that the mechanism of action for hand hygiene extends beyond the fecal-oral route to respiratory and ocular pathogen transfer pathways [13].

Integrated programs demonstrated the ability to alter cognitive trajectories without necessarily impacting distal physical growth outcomes. In rural China, an integrated home-visitation program improved cognitive development by 0.24 standard deviations (SD) (95% CI 0.04–0.44 SD, p = 0.01). However, no significant effects were found for motor or social-emotional development [14]. A critical finding across the literature is that reductions in morbidity do not necessarily translate into measurable gains in nutritional status. For example, in Bangladesh an intervention achieved a 22% protective efficacy against diarrhea (incidence density ratio of 0.78, 95% CI 0.74–0.83). Despite this reduction, children in both intervention and control groups exhibited identical patterns of weight gain and remained at 76% of the NCHS standard for weight-for-age. This underscore the complexity of growth outcomes and suggests that factors such as food security, dietary quality and metabolic stressors may limit the impact of hygiene-only interventions on nutritional recovery [24].

Educational technologies reliably improved knowledge scores and maternal self-efficacy (MSE). In the United States, Bloch et al. demonstrated that a 3-minute video intervention significantly outperformed written instructions, with higher knowledge scores both immediately (12.2 vs. 8.9) and at follow-up (11.1 vs. 7.8) [22]. In Brazil, the use of combined video and booklets significantly improved MSE scores, with the risk of childhood diarrhea in the combined group (AB) plummeting from 8.5 to 1.1 over the follow-up period [5]. In Honduras, counseling led to significant shifts in primary care practices. Targeted parents were 16.4% more likely to seek professional newborn checks within 3 days (95% CI 3.1%–29.8%, p = 0.016) and 19.6% more likely to avoid traditional umbilical cord wrapping (95% CI 4.2%–35.1%, p = 0.013) [12].

The strength of the reported associations is moderated by several identified confounders that were not fully controlled in the primary studies. In Ghana, Ma et al. (2019) noted a district-wide cholera outbreak triggered supply-side interventions (e.g., water tablets) that caused diarrhea prevalence in the control group to drop from 20.1% to 7.0%, potentially masking the true effect of volunteer visits (overall RR 0.73, 95% CI 0.37–1.45). Subgroup analysis in the Ghana trial identified that health benefits were only statistically significant when implementation met specific fidelity thresholds: ≥ 70% community coverage and ≥30 minutes per visit (Diarrhea RR 0.23, p = 0.003) [15]. In Bolivia, the extreme baseline contamination of tanker truck water (71.8% of households) acted as a structural confounder that rendered behavioral education ineffective without the physical barrier of a filter. In the end, most studies relied on maternal 7-day or 14-day recall, and several lacked blinding between the delivery and data collection teams, introducing potential social desirability bias [19].

4.1. Biases and limitations

The risk of bias analysis indicated that some studies had a high risk in domains such as inadequate randomization, lack of blinding, and substantial loss to follow-up (Fig 2). Studies that failed to clearly describe randomization methods or to implement strategies to minimize performance and detection bias tend to yield less reliable results, compromising internal validity and the strength of the synthesized evidence. Oles et al. (2024) and Penha et al. (2022) were particularly notable for high risk related to randomization and deviations from intended interventions, including the use of self-reported data and potential lack of blinding [5,12]. Penha et al. (2022) also showed high risk due to missing outcome data, with follow-up losses that may have influenced the results [5]. Similarly, Lindquist (2014) and Nicholson (2014) exhibited high risk in outcome measurement due to reliance on caregiver-reported morbidity without clinical verification, increasing the possibility of detection bias [19,23].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Risk of Bias Graph for the Different Clinical Trials Included in the Systematic Review.

https://doi.org/10.1371/journal.pntd.0014442.g002

Other studies, such as Luo et al. (2019), Ma et al. (2019), and Manaseki-Holland et al. (2021), were classified as having “some concerns,” especially regarding allocation concealment and outcome measurement [13– 15]. Conversely, the direction and strength of the observed associations were more consistent when supported by studies classified as having Low Risk of Bias. For example, Luby (2004) and Hashi (2017) provided higher-certainty evidence that intensive handwashing promotion and consistent WASH messaging were associated with reductions in diarrhea incidence ranging from 35% to 53% in high-risk environments. Studies categorized as having “some concerns,” such as Manaseki-Holland (2021) and Luo (2019), also demonstrated favorable short-term reductions—such as the 60% decrease in self-reported diarrhea observed in The Gambia following multiple reinforcement visits—although these findings require cautious interpretation due to methodological constraints.

Conversely, findings from studies with High Risk of Bias should be interpreted with greater caution, particularly when based on self-reported outcomes or in the presence of incomplete follow-up data, as these factors may overestimate intervention effects. These findings highlight the need for caution when interpreting the effectiveness of the interventions.

Despite consistent behavioral improvements, several interventions did not translate into clear clinical benefits, suggesting that structural determinants, such as inadequate sanitation and limited access to clean water, may attenuate the isolated effect of educational strategies (Freeman et al., 2014; Luby et al., 2006) [17,29]. This contextual dependency represents an important limitation when interpreting effectiveness estimates across heterogeneous settings. Environmental and district-level confounders also influenced observed impacts. In Ma (2019), for instance, a district-wide cholera outbreak triggered concurrent supply-side interventions in the control group, potentially masking the true effect of the educational strategy. In addition, incomplete reporting of participant characteristics, particularly caregiver age, in several included studies restricted the assessment of population comparability and may have contributed to observed clinical heterogeneity. Together, these factors limit causal inference and underscore the need for cautious interpretation of pooled qualitative findings.