A study of maternal reports of childhood injuries that result in hospital attendance or admission: do they match National Health Index database records?
MetadataShow full metadata
Background: Childhood injury is one of the leading causes of morbidity and mortality worldwide and concern is increasingly expressed at the lack of adequate data to inform policies and injury prevention strategies. Objective: The aim of this thesis was to examine the reliability and validity of the use of maternal recall of childhood injuries in a birth cohort study of Pacific mothers residing in New Zealand. Maternal reports of child injuries and medical attendance events reported were matched to listings held within the National New Zealand Health Information Service (NZHIS), computer database of public hospital events, the National Medical Discharge Summary (NMDS) database. Methods: The study included 1,354 Pacific children born in Auckland in 2000 whose mothers completed a questionnaire at 6-weeks, 1-year, 2-years, 4-years, and at the 6-year measurement waves. Mothers identified injury events by proxy report, in face to face interviews with ethnic specific interviewers. Information was gained on the type of injuries sustained, their frequency and medical attendance events associated with these injuries. The Statistical Classification of Diseases and Related Health Problems Code 10th Revision (ICD-10-AM) was used to ascertain injury and non injury status for the NMDS medical attendance listings. Results: Kappa statistics demonstrated a modest level of agreement between the NHI database listings and the mothers reporting of childhood injuries between the 0-6-year old children. However McNemar's test of symmetry revealed no systemic under-reporting on behalf of the mothers, suggesting that the use of maternal proxy reporting of childhood injuries is indeed a valid measure. Conclusion: While maternal proxy reporting of childhood injuries was found to be a valid measure, some evidence of misinterpretation of questions was found; suggesting continued vigilance and development of maternal completed childhood injury questionnaires is, warranted. Further investigation exploring the reliability of maternal recall over time and development of an internationally recognised and standardized questionnaire for capturing parent reports of childhood injury is advocated. The responsible use of de-identified data in child health studies is believed to have a pivotal role to play in reliability and validity studies in the future.