Abstract

The objective of the study was to estimate the quality of the data recorded in the Estonian Medical Birth Registry.
METHOD. A sample was drawn from all births (12,593 births) in 1997. Copies of the original hospital records were requested (666 births) and received for 641 births. The records were re-abstracted by trained gynecologists at clinics. The level of agreement between the hospital records and the records of the Estonian Medical Birth Registry was measured using agreement proportion and Cohen’s kappa (with its 95% confidence interval).
RESULTS. Most of the sociodemographic data, variables in the history of pregnancy,  number of antenatal visits, date of the last menstrual period, child’s gender, birth weight, birth length, and dates of hospitalization and discharge were reliable. The date of the first antenatal visit, maternal smoking, gestational age (in days) and duration of delivery were at an unsatisfactory level. Concerning gestational age, the agreement reached a satisfactory level when full weeks were compared. Data quality in the pregnancy- or delivery-related diagnoses and procedures was the poorest.
CONCLUSION. This study showed that the registry data were accurate for many variables, however, poorly recorded variables should be used with caution in research.