As perinatal events have been linked with diseases of later onset, epidemiological studies on child development and adult health require information on the perinatal period. When national neonatal registers do not exist, review of medical records may be impractical. However, neonatal information could be obtained by asking mothers to complete a postal questionnaire using data from the Personal Child Health Record (PCHR). We assessed agreement between medical records and mothers' reports for information on delivery and the newborn's condition at birth, based on the PCHR, a short time after delivery. Of 711 women giving birth in 3 French hospitals and invited to participate in the study, 580 (82%) completed a postal questionnaire 6 weeks after delivery, copying the data recorded in the PCHR when available. Information on pregnancy, delivery and the newborn's health at birth was independently extracted from medical records by physicians of the maternity departments. Agreement between medical records and maternal reports for a range of perinatal factors was assessed in 580 newborn-mother dyads using kappa coefficients. Agreement was excellent for first and second stages of delivery, gestational age, birth weight, birth size and head circumference (kappa coefficients 0.80-1.00) and good for hospitalization during pregnancy, but poor for Apgar scores. With this exception, mothers' reports appeared reliable when compared with medical records. As PCHRs exist in most developed countries, this approach could be used in epidemiological studies on child development to increase the reliability of mothers' reports of their newborn's condition at birth.

Apgar scores reported in PCHRs may not yield reliable information for epidemiological studies. When the PCHR is the only source of information for the neonatal period in an epidemiological study, it would be preferable to use a composite neonatal indicator rather than the Apgar score.