Modelling birthweights with different sampling rates: a comparison of three procedures in SAS.
by
Kristie Carter
University of Auckland
Coauthors: Chris Wild - Department of Statistics, John Thompson, Ed Mitchell - Department of Paediatrics

The Auckland Birthweight Collaborative (ABC) study is a case-control study that was conducted in Auckland from October 1995 to November 1997. This study collected information on the risk factors relating to small for gestational age (SGA) infants in New Zealand. SGA age babies were sampled at a much higher rate than control babies (appropriate for gestational age, AGA). There were also different sampling rates employed over two time periods due to changes in maternity care in the study regions.

The main analyses of the ABC study data were conducted using logistic regression analyses comparing SGA babies to AGA babies. This paper discusses linear regression analysis of the birthweights themselves on study risk factors. Most of the analysis was performed in SAS using the SURVEYREG procedure, a relative newcomer to SAS that can that can account for different sampling rates. This talk will describe how SURVEYREG takes into account the different sampling rates and compare the results of logistic regression, linear regression, and linear regression analyses with proper allowance for survey weights.

Date received: August 30, 2001