132 Annual Meeting Logo - Go to APHA Meeting Page  
APHA Logo - Go to APHA Home Page

Multiple outcome analysis of the effect of smoking on correlated pregnancy outcomes

Monina G. Bartoces, MS1, Robert McKeown, PhD1, Cheryl Addy, PhD1, Angela Liese, PhD1, and Kathryn J. Luchok, PhD2. (1) Department of Epidemiology and Biostatistics, University of South Carolina, Norman J. Arnold School of Public Health, Columbia, SC 29208, 803-777-7361, mgbarto@mailbox.sc.edu, (2) Dept. of Health Promotion & Education, School of Public Health, U of South Carolina, 800 Sumter Street, Rm. 216, Columbia, SC 29208

Background. Researchers have recognized the importance of investigating birthweight and gestational age simultaneously as outcome measures in reproductive epidemiology. A major issue arising from simultaneous investigation of LBW, preterm and small-for-gestational age (SGA) is appropriate modeling. The common approach is to conduct separate regression analyses for each of the outcomes. However, these outcomes are correlated, so the correlation must be taken into account to satisfy the independence assumption when investigating these outcomes simultaneously. Methods: We presented multiple outcome analysis using generalized estimating equations (GEE) as a novel approach to investigate simultaneously the associations between LBW, preterm, and SGA and smoking in pregnancy. Data were from the South Carolina Pregnancy Risk Assessment Monitoring Survey (PRAMS) from 1995 to 1999. Results: Smoking had different effects on LBW, preterm, and SGA; hence, the associations between smoking and these outcomes were modeled using outcome-specific effect model. Adjusted odds ratios indicate significant positive associations between smoking during the third trimester and LBW and SGA but not preterm. Conclusion: Multiple outcome analysis using GEE minimized Type I error by accounting for correlation among outcomes; allowed outcome-specific (different effects across outcomes) or common (similar effect) effect models depending on results of heterogeneity test. These models may provide insights on mechanisms underlying different associations between a factor and these outcomes; a common effect may be explained by a common biological mechanism whereas different effects may suggest otherwise. Identifying a factor that similarly affects these outcomes may translate to better prevention strategies.

Learning Objectives:

Keywords: Epidemiology, Methodology

Presenting author's disclosure statement:
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.

Maternal and Child Health Epidemiology

The 132nd Annual Meeting (November 6-10, 2004) of APHA