Client and birth record linkage: A method, biases, and lessons

Due to rising data collection costs and the availability of extant data files, record linkage is becoming increasingly important in evaluation research. We begin by discussing record linkage as a data-generating technique and its applicability to evaluation studies. This paper—part of a larger study assessing the perinatal outcomes of a comprehensive prenatal program for poor, urban women—describes a method for linking client records to live and stillbirth records. Biases which can enter into the linkage process and general issues which need to be addressed early in an investigation are also discussed.