Assessment of gestational age using birth certificate data compared with medical record data.

We assessed the extent to which use of medical record data might improve gestational age estimates compared with reliance on the birth certificate alone. Using population-based data from four northern Californian counties, we constructed an algorithm to select the best gestational age estimate from antenatal assessments recorded in medical records. A total of 172 singletons with moderate or severe congenital cerebral palsy from the California Cerebral Palsy Project were compared with 472 randomly selected controls with regard to discrepancies between the algorithm-derived estimated gestational age (bestgest) and an estimate based solely on the last menstrual period as recorded on birth certificates. Agreement between bestgest and birth certificate estimated gestational age was exact or within one week for at least 60% of both cases and controls in each of the three birthweight strata. In general, the greater the birthweight of the babies, the better the agreement. The mean number of weeks of overestimation by the birth certificate was 0.7 weeks for cases and 1.1 weeks for controls in the lowest birthweight group (< 1500 g). When compared with bestgest, clinical examination of the infant also tended to overestimate gestational age. In the < 1500 g birthweight group, cases were twice as likely as controls to have a bestgest of 'low certainty,' but antenatal estimates of 'high certainty' were obtained for at least a third of very low birthweight babies born during the mid-1980s. More widespread use of early ultrasound in more recent birth cohorts may result in a greater proportion of accurate antenatal estimates. When a distinction between immaturity and intrauterine growth retardation is important to the understanding of the aetiology of the outcome under investigation, the use of antenatal estimates from medical records may substantially improve the certainty of the data.