You are here:
INTERIM REPORT, DEVELOP A SAMPLING STRATEGY: PREPARE TO IMPLEMENT A COHORT STUDY OF CHILDREN'S ENVIRONMENTAL HEALTH
Citation:
MENDOLA, P. AND S. MYERS. INTERIM REPORT, DEVELOP A SAMPLING STRATEGY: PREPARE TO IMPLEMENT A COHORT STUDY OF CHILDREN'S ENVIRONMENTAL HEALTH. U.S. Environmental Protection Agency, Washington, DC, EPA/600/R-05/017, 2005.
Impact/Purpose:
We propose sample design options to be considered for implementation in a North Carolina-based longitudinal cohort study
Description:
Overview
The goal of the National Children's Study (NCS) is to collect information on environmental and
social factors in the lives of children, starting before birth and continuing until age 21, to inform
researchers about their effects on the health and development of children. In this document, we propose
sample design options to be considered for implementation in a North Carolina-based longitudinal cohort
study (LCS). The purpose of the report is to present 'within-primary sampling unit (PSU)' procedures
for sampling, recruiting and collecting data for that 5,000-children cohort. EPA proposed in the work
assignment that initiated this work that RTI consider the sampling and design approaches currently under
consideration for the NCS, and develop sampling scheme options for the North Carolina-based pilot
study. Considerable work has already been done for the NCS by Battelle (Strauss et al, 2004) and Westat
(Westat, 2002) in developing sampling scheme options. This report builds on the sample design options
already proposed and focuses on the possible implementation of the sample design schemes in North
Carolina counties.
The first-stage sample design for the NCS involves the selection of PSUs in the U.S. in which the
study will be conducted. The goal of the within-PSU sample selection is to obtain approximately 25% of
the children for the study from women who are at a pre-conception stage and the remaining 75% of the
children from women who are pregnant at recruitment, ideally in their first trimester. Two sample
designs are evaluated as possible options for the within-PSU sample selection using a probability-based
sampling scheme. The first option is based entirely on an area household sampling frame, as requested by
the WAM. The second option is a dual frame design that has the potential to provide equivalent
inferential support at reduced cost by incorporating area household sampling as well as sampling in the
offices of prenatal care providers. The dual frame design was initially proposed in the Work Plan as
RTI's main sampling strategy, and this approach had already been developed when we received feedback
from the WAM. Thus, it is included here for consideration.