Dietary diversity, place of residence and malnutrition among school-going adolescent girls in Western Maharashtra, India: a structural equation modelling analysis

Background Adolescent girls in India bear a disproportionate share of the country’s nutritional burden, with undernutrition and overnutrition increasingly coexisting within the same population. Dietary diversity is a modifiable proxy indicator for diet quality. Methods A community-based cross-sectional analytical study was conducted among 703 school-going adolescent girls (350 urban, 353 rural) in Western Maharashtra, India. Dietary diversity was assessed by 24-h recall. Associations were tested with Pearson’s chi-square and binary logistic regression. For the outcomes (i.e., undernutrition, overnutrition, and malnutrition), an exploratory path-analytic structural equation model was fitted using maximum likelihood estimation, with place of residence and Minimum Dietary Diversity (MDD) specified a priori as hypothesized serial intermediate variables between sociodemographic characteristics and nutritional status. Model fit was evaluated using CFI ≥ 0.90 and RMSEA ≤0.08, with TLI reported alongside. Cluster-robust generalized estimating equations (GEE) were used as a sensitivity analysis for within-school clustering. Results Malnutrition affected 70.13% of participants overall (undernutrition 60.31%, overnutrition 9.82%). MDD was inadequate in 90.18% (rural 95.18% vs. urban 85.14%, p < 0.001). After multivariable adjustment, rural residence (AOR = 4.55) and father’s elementary occupation (AOR = 2.36) were independently associated with higher odds of inadequate MDD, whereas outdoor physical activity of ≥2 h/day was associated with lower odds (AOR = 0.49). In the path models, dietary diversity showed a significant indirect association with undernutrition (standardized indirect association via MDD = 0.014, 95% CI 0.001–0.029) and with overall malnutrition (indirect association = 0.014, 95% CI 0.001–0.035), while place of residence was significantly associated with dietary diversity and lifestyle behaviors in all three models. Fit was acceptable on CFI (0.93–0.94) and RMSEA (0.058), but TLI (0.85–0.86) was below 0.90; in the GEE sensitivity analysis, which is limited by the small number of schools, the direction of the key associations was unchanged. Conclusion Dietary diversity and place of residence showed statistically significant direct and indirect associations with undernutrition and overall malnutrition among adolescent girls. These exploratory, cross-sectional findings describe hypothesized pathways and do not establish mediation or causation; confirmation requires longitudinal data and a design with more schools.