Comparison of concordance and predictive validity of head injuries from parental reports and medical records.

Abstract

Objective: To examine agreement between parental reports of head injury and evidence of head injury in medical records and to compare these two measures in predicting early conduct disorder (CD).

Design and setting: Parent survey data was compared with records of child head injury from the National Health Services Register (Régie de l'assurance maladie du Québec, RAMQ) administrative database.

Participants: Children (N = 685) ages 6-9 with and without CD. There were 147 children with RAMQ recorded head injury and 39 children with parent-reported head injury.

Main measures: Indication of one or more head injury before 6 years of age as reported by parents and/or as noted in medical data. Early CD (present by age 9) according to parents and/or teachers.

Results: Results indicated poor agreement between the two forms of reporting κ = .161 (95% CI, .083 to .239), p < 0.001. Medical data significantly predicted the presence of CD in children, with a RAMQ coded head injury suggesting a child was 1.88 times more likely to have CD. Parent reports of head injuries did not significantly predict CD. Conclusion: Medical data should be prioritized in research addressing pediatric head injury, given that parent reports may fail to capture incidence of injury and therefore may be less predictive of other known correlates of head injury.