Agreement of injury reporting between primary care medical record and maternal interview for children aged 0-3 years: implications for research and clinical care.

Abstract

To assess agreement of injury reporting between primary care medical record and maternal interview.
Cross-sectional study of data from a randomized controlled trial of home visiting. The setting was Hawaii's Healthy Start Program (HSP). Subjects comprised a population-based sample of children in at-risk families with 3 years of primary care medical records and maternal interviews (n = 443). Outcome measures were percentage of children injured unintentionally and mean number of injuries per child in the first 3 years of life by primary care medical record and maternal interview.
We identified 490 injuries: 48% by primary care medical record, 22% by maternal interview, and 30% in both sources. More children were reported injured by primary care medical record than maternal interview (51% vs 39%, P< .001). The mean number of injuries per child was 0.87 by primary care medical record and 0.51 by maternal interview (difference 0.36, 95% confidence interval 0.27-0.45, P< .001). Agreement between data sources was fair (kappa = 0.47).
This study estimates that 25% of childhood injuries may not be reported in the medical record, highlighting the need for reconsideration of the use of medical records as the gold standard for unintentional injury data. Caution should be used when interpreting injury data from one source, especially from families with stressful life situations. Poor communication regarding injuries between social service, primary care and urgent care providers may contribute to decreased quality of primary care and missed opportunities for injury prevention.