A generalized additive mixed model to estimate trends in stunting in a longitudinal study of children covered by social inclusion programs in northeast Argentina

PÉREZ, ADRIANA; CUETO, GERARDO RUBÉN; TURJANSKI, PABLO; FERNÁNDEZ, MARÍA SOLEDAD; NUÑEZ, PABLO

Barcelona

Conferencia; XXIXth International Biometric Conference; 2018

International Biometric Society

Abstract Body: Restricted growth (stunting), which impair children?s potential for growth and development, is the most prevalent form of undernutrition in the globe and the best proxy for child health inequalities. Argentina´s government implemented programs Plan Nacer (2004) and Programa Sumar (2013) with the goal of equitably increasing the access and quality of prioritized health services for pregnant women and children up to 5 years. We present here an evaluation of the impact and potential of these programs on the nutritional status of the target children population. Between 2005 and 2015; 941,233 set of anthropometric measures (including birth and visit dates, age, gender, weight and height) corresponding to 114,676 younger than 5 years children were collected at 291 health centers located throughout the Misiones province, Argentina. We defined the prevalence of stunting (HAZ, height-for-age Z score) as the proportion of children who fell below an SD of −2. We fitted a generalized additive mixed model (GAMM) to model non-linear trends and seasonal variation in prevalence of stunting. Age, gender of children and health center were added as fixed variables. A numeric variable ?visit month? was added with a cyclic cubic spline smooth, to forces the spline to connect the estimate at December with the estimate at January. A random effect ?individuals? was included to account for within-individual correlation. Over the study period, stunting prevalence showed significant declining inter annual trend, varying from 22.9% in 2005 to 9.5% in 2015. A significant seasonal variation of prevalence was registered, being higher in winter months. The stunting prevalence was also higher in boys than in girls, and for both genders, decrease in a non-linear way with age. These findings support substantial decreases in stunting prevalence in the population analyzed. Programs provide high-quality and big-data sources that could be coupled to robust monitoring and evaluating frameworks to generate updated population-based evidence. We consider this as a key focus of policy makers and planners aiming to improve child survival, health, and nutrition. The maintenance of these strategies over time becomes essential to achieve sustained improvements in health-related indicators in the post-2015 sustainable development agenda.