3104.0: Monday, November 13, 2000 - 2:30 PM

Abstract #9636

Indiana birth certificate data quality study

Terrell W. Zollinger, DrPH, School of Public and Environmental Affairs, Indiana University, IUPUI, Bowen Research Center, 1110 West Michigan Street, Indianapolis, IN 46202, (317) 278-0300, tzolling@iupui.edu, Michael J. Przybylski, PhD, Bowen Research Center, Indiana University, 1110 W. Michigan Street, Indianapolis, IN 46202, and Roland E. Gamache, PhD, Epidemiology Resource Center, Indiana State Department of Health, 2 North Meridian Street, Indianapolis, IN 46204.

Natality statistics are commonly used to track trends in birth outcomes without regard to the accuracy of those data. The purpose of this study was to measure the reliability of 132 data elements recorded on the Indiana birth certificates as compared to the medical records. A two-stage stratified cluster design identified a sample of 1200 hospital births from 18 hospitals in Indiana in 1996. Medical records were abstracted and linked to the birth certificate data on 88% (n=1050) of the selected sample. The level of agreement between the birth certificates and the medical records in this sample varied widely by type of data element. The best agreement was in the areas of parents' demographics (.927) and overall birth outcome (.867), including low birth weight (.867) and preterm delivery (.791). Other variables, including whether the birth was by Cesarean delivery (.705) and overall aspects of pregnancy history (.608), had moderate agreement. Agreement was poor on the majority of other items, including most of the less frequently occurring items such as labor and delivery complications (.362), obstetric procedures (.276), concurrent illness (.117), and pregnancy complications (.049). There was no agreement on very rarely reported events such as congenital anomalies and abnormal conditions of the newborn. Although individual prenatal care variables did not have very good agreement, there was excellent agreement on the adequacy of prenatal care utilization index (.946) computed from them. Thus, significant caution must be used for most variables used to track trends in natality data.

Learning Objectives: At the end of this session, participants will be able to: 1. Distingish birth certificate data elements that are accurate and that can be used with confidence in community surveillance from those that have less value. 2. Apply specific measures of agreement that can be used in similar reliability studies. 3. Describe ways to improve birth certificate data collection

Keywords: Birth Outcomes, Surveillance

Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.

The 128th Annual Meeting of APHA