Researchers from the University of South Carolina’s Arnold School of Public Health, the United Nation’s Food and Agriculture Organization in Rome, Italy, and the World Health Organization in Geneva, Switzerland have completed a study on prediction intervals for penalized longitudinal models with multi-sourced summary measures to estimate childhood malnutrition prevalence. The study was led by Assistant Professor Epidemiology and Biostatistics Alexander McLain and Health Promotion, Education, and Behavior Professor Edward Frongillo and published in Statistics in Medicine.
The authors conducted this research in order to examine countries’ progress using malnutrition indicators of socio-economic conditions based on national surveys from varying sources. Global health analyses generally result in longitudinal data where summary measures from surveys (e.g., the estimated prevalence of stunting) rather than individual level data are available. Using summary measures from surveys in statistical analyses are challenging since each has an associated standard error, which is impacted by the survey sample size, study design and overall prevalence.
Administration of national surveys can be sporadic, resulting in sparse data measurements for some countries. Furthermore, the trend of the indicators over time is usually nonlinear and varies by country.
Drs. McLain and Frongillo and their team felt it was important to track the current level of malnutrition to determine if countries are meeting certain thresholds, such as those indicated in the United Nations Sustainable Development Goals. In addition, estimation of confidence and prediction intervals are vital to determine true changes in prevalence and where data is low in quantity and/or quality.
The authors use heteroscedastic penalized longitudinal models with survey summary data to estimate yearly prevalence of malnutrition quantities. They also develop and compare methods to estimate confidence and prediction intervals using asymptotic and parametric bootstrap techniques.
The intervals can incorporate data from multiple sources or other general data-smoothing steps, and the methods are applied to African countries in the UNICEF-WHO-The World Bank joint child malnutrition data set. The properties of the intervals are demonstrated through simulation studies and cross-validation of real data.