Calidad de los datos del Instituto Nacional de Estadística para la elaboración de los indicadores de salud perinatal: pequeño y grande para su edad gestacional

Abstract

BACKGROUND Relative measures of birthweight (small and large-for-gestational age, SGA-LGA) are increasingly preferred to absolute measures (low birthweight, macrosomia). In this study we assess whether the national vital statistics provided by the Spanish National Statistical Institute (INE) reliably estimate SGA and LGA. Also, we will assess whether missing data (selection) and misreported information (bias) are systematically associated with parental socioeconomic information. METHODS We linked the information on 6,339 births at the Hospital Clínico San Carlos of Madrid (2005-06) with the vital statistics records (successful for the 95% of the observation). Validity measures and concordance were estimated for low birthweight (LBW, <2500 gr), macrosomia (>4500 gr), SGA (<10th percentile) and LGA (>90 percentile). Logistic regressions were fitted. RESULTS The prevalence estimated with the hospital data were: LBW (6%), macrosomia (0.5%) SGA (1%) and LGA (15%) and, with the data from INE: 5% 0.5% 2% 12% respectively. Kappa statistics: LBW (83%), macrosomia (79%), PEG (24%) and LGA (82%). Missing and misreported data vary with parental nationality and their situation in the labor market (OR between 1.5 y 2.2). CONCLUSIONS Vital statistics overestimate the prevalence of SGA and underestimate the prevalence of LGA. The concordance between the sources is very good for low birthweight, macrosomia and LGA, and moderately good for SGA. Both missing and misreported birthweight and gestational age are associated with parental socioeconomic characteristics.