Predictors of asthma in young children: does reporting source affect our conclusions?

Both the size and statistical significance of sociodemographic and early health risk factors on childhood asthma vary across studies, in part because some studies rely on parents' retrospective reports of health conditions while others are based on medical records. The authors compare predictors of asthma alternately using maternal reports and medical records for the same set of children. Data are from the 1988 National Maternal and Infant Health Survey and 1991 Longitudinal Follow-up, which collected information from birth certificates, medical records, and mothers of a nationally representative, population-based cohort, allowing comparison across data sources for a consistent sample of young children in the United States. Concordance between maternal reports and medical records on asthma is moderate (kappa = 0.48). The authors find considerable discrepancies in both the estimated prevalence of asthma and the distribution across children with different sociodemographic and health characteristics, depending on the source of asthma data. Black race, male gender, and preterm birth are found to be risk factors for asthma regardless of data source. Poverty, large family size, urban residence, maternal smoking, and breastfeeding are significantly associated with asthma based on maternal reports but not medical records. Lower health care utilization among poor, uninsured, and urban children may account for part of the discrepancy.