Early childhood caries in Romania: non-linear and spatial analysis of geographic sugar intake using LOESS, GAMs, and...

Excessive intake of free sugars in early childhood is a major modifiable risk factor for early childhood caries (ECC). Evidence on geographic variation in sugar consumption among preschool children remains limited, particularly in Eastern Europe.

To assess patterns of sugar intake among Romanian preschool children and to explore non-linear and spatial associations between sugar consumption and geographic areas using advanced statistical and mapping techniques.

Baseline data from the Smilebright Romania oral health programme were analyzed for 391 children aged 4-5 years. Parental questionnaires assessed the frequency of consumption of natural and processed sugars using a Likert scale. Sugar intake indices (total, processed, and natural) were derived as mean frequency scores. Non-linear associations between sugar intake and geographic variables were explored using locally weighted scatterplot smoothing (LOESS) and tested using generalized additive models (GAMs). County-level mean intake values were visualized using choropleth maps.

At the macro-regional level, significant variation was observed only for natural sugar intake (edf = 1.69, F = 7.32, p < 0.01), while processed and total sugar intake showed no significant differences. At the county level, significant geographic variation was identified for total, processed, and natural sugar intake (all p < 0.01), although adjusted R² values were modest (0.04-0.07). Spatial mapping revealed higher levels of total and processed sugar intake predominantly in north-eastern counties, whereas natural sugar intake was higher in south-western regions.

Sugar intake among Romanian preschool children demonstrates non-linear and spatially structured geographic patterns, particularly at finer geographic resolution. Although geographic factors explain a limited proportion of variability, their consistent influence supports the value of spatially informed analyses. Integrating flexible statistical modeling with geographic visualization can enhance understanding of dietary risk factors relevant to ECC prevention and inform targeted public health strategies details.