**Background:** Malnutrition in under-five children remains a critical global public health problem, with a worldwide stunting prevalence of 33.7% in 2022 and approximately 47 million children affected by wasting in 2020. Malnutrition contributes to increased morbidity and mortality—stunted children have a 5.48 times higher risk of death, and severely wasted children have an 11.63 times higher risk compared to normally nourished children. The positive deviance approach identifies uncommon but beneficial behaviors practiced by well-nourished children within malnourished communities and leverages these local solutions through community mobilization and education (Hearth sessions). This systematic review and meta-analysis aimed to evaluate the effectiveness of positive deviance-based interventions in reducing malnutrition among under-five children.
**Methods:** A systematic search was conducted across six databases (Science Direct, PubMed, ProQuest, SAGE Journals, Web of Science, and Scopus) from August 26 to September 4, 2022, supplemented by manual searches. The review followed PRISMA guidelines and was registered in PROSPERO (ID CRD42023413091). Inclusion criteria were original English-language interventional studies with control groups reporting LAZ, WAZ, or WHZ outcomes. Risk of bias was assessed using the Cochrane RoB 2 tool. Meta-analysis was performed using Review Manager 5.4 with a random-effects model, reporting mean differences (MD) and 95% confidence intervals. Heterogeneity was assessed using I² statistics.
**Key Results:** From 253 database records and 8 manual additions, 15 duplicates were removed, 33 articles underwent full-text screening, and 8 were included in the systematic review, of which 4 were eligible for meta-analysis (total n=2467 children). The four included studies were conducted in Ethiopia (Kang et al., n=1125, aged 6–24 months), Vietnam (Schroeder et al., n=240, aged 5–36 months), Cambodia (Young et al., n=330, aged 6–23 months), and Kenya (Anino et al., n=107, aged 6–59 months). Intervention durations ranged from 2 weeks to 12 months with follow-up of 6–12 months. For LAZ (3 studies, n=1574): MD 0.33 [95% CI −0.36 to 1.03], p=0.35, I²=98%. For WAZ (4 studies, n=1681): MD 0.28 [95% CI −0.04 to 0.61], p=0.09, I²=93%. For WHZ (3 studies, n=1574): MD 0.25 [95% CI −0.16 to 0.66], p=0.24, I²=93%. None of the pooled effects reached statistical significance. All three funnel plots were asymmetrical, though Egger's test could not be applied due to the small number of studies. Risk of bias was generally low overall, though one study (Anino et al.) had high risk due to non-randomized design, and two studies had high risk in the deviation-from-intended-intervention domain due to lack of blinding.
**Clinical Implications:** Although the pooled results did not reach statistical significance, the positive deviance approach consistently showed directionally favorable effects across all three anthropometric indicators. The approach is considered affordable, acceptable, and sustainable because it leverages local resources and culturally congruent practices. The high heterogeneity (I²=93–98%) likely reflects differences in sample age, intervention duration (10–12 days of Hearth sessions with variable follow-up), and follow-up periods across studies. The authors note that stunting improvement requires sustained behavior change—habit formation typically requires 21 days of consistent practice—and that younger children (<1 year) may benefit more from these interventions due to greater caregiver attention. The limited number of available interventional studies and incomplete data reporting (requiring exclusion of 4 of 8 eligible studies from meta-analysis) highlight the need for more rigorous, adequately powered trials with standardized outcome reporting and longer follow-up to determine optimal intervention duration and confirm effectiveness.