Kristen M Rappazzo1, Danelle T Lobdell1, Lynne C Messer2, Charles Poole3, Julie L Daniels3. 1. National Health and Environmental Effects Research Laboratory, U.S. Environmental Protection Agency, Office of Research and Development, Chapel Hill, North Carolina, USA. 2. School of Community Health-College of Urban and Public Affairs, Portland State University Portland, Portland, Oregon, USA. 3. Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina, Chapel Hill, North Carolina, USA.
Abstract
OBJECTIVES: Estimating gestational age is usually based on date of last menstrual period (LMP) or clinical estimation (CE); both approaches introduce potential bias. Differences in methods of estimation may lead to misclassification and inconsistencies in risk estimates, particularly if exposure assignment is also gestation-dependent. This paper examines a 'what-if' scenario in which alternative methods are used and attempts to elucidate how method choice affects observed results. METHODS: We constructed two 20-week gestational age cohorts of pregnancies between 2000 and 2005 (New Jersey, Pennsylvania, Ohio, USA) using live birth certificates: one defined preterm birth (PTB) status using CE and one using LMP. Within these, we estimated risk for 4 categories of preterm birth (PTBs per 106 pregnancies) and risk differences (RD (95% CIs)) associated with exposure to particulate matter (PM2.5). RESULTS: More births were classified preterm using LMP (16%) compared with CE (8%). RD divergences increased between cohorts as exposure period approached delivery. Among births between 28 and 31 weeks, week 7 PM2.5 exposure conveyed RDs of 44 (21 to 67) for CE and 50 (18 to 82) for LMP populations, while week 24 exposure conveyed RDs of 33 (11 to 56) and -20 (-50 to 10), respectively. CONCLUSIONS: Different results from analyses restricted to births with both CE and LMP are most likely due to differences in dating methods rather than selection issues. Results are sensitive to choice of gestational age estimation, though degree of sensitivity can vary by exposure timing. When both outcome and exposure depend on estimate of gestational age, awareness of nuances in the method used for estimation is critical. Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://www.bmj.com/company/products-services/rights-and-licensing/.
OBJECTIVES: Estimating gestational age is usually based on date of last menstrual period (LMP) or clinical estimation (CE); both approaches introduce potential bias. Differences in methods of estimation may lead to misclassification and inconsistencies in risk estimates, particularly if exposure assignment is also gestation-dependent. This paper examines a 'what-if' scenario in which alternative methods are used and attempts to elucidate how method choice affects observed results. METHODS: We constructed two 20-week gestational age cohorts of pregnancies between 2000 and 2005 (New Jersey, Pennsylvania, Ohio, USA) using live birth certificates: one defined preterm birth (PTB) status using CE and one using LMP. Within these, we estimated risk for 4 categories of preterm birth (PTBs per 106 pregnancies) and risk differences (RD (95% CIs)) associated with exposure to particulate matter (PM2.5). RESULTS: More births were classified preterm using LMP (16%) compared with CE (8%). RD divergences increased between cohorts as exposure period approached delivery. Among births between 28 and 31 weeks, week 7 PM2.5 exposure conveyed RDs of 44 (21 to 67) for CE and 50 (18 to 82) for LMP populations, while week 24 exposure conveyed RDs of 33 (11 to 56) and -20 (-50 to 10), respectively. CONCLUSIONS: Different results from analyses restricted to births with both CE and LMP are most likely due to differences in dating methods rather than selection issues. Results are sensitive to choice of gestational age estimation, though degree of sensitivity can vary by exposure timing. When both outcome and exposure depend on estimate of gestational age, awareness of nuances in the method used for estimation is critical. Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://www.bmj.com/company/products-services/rights-and-licensing/.
Authors: David A Savitz; James W Terry; Nancy Dole; John M Thorp; Anna Maria Siega-Riz; Amy H Herring Journal: Am J Obstet Gynecol Date: 2002-12 Impact factor: 8.661
Authors: Amy E Kalkbrenner; Gayle C Windham; Marc L Serre; Yasuyuki Akita; Xuexia Wang; Kate Hoffman; Brian P Thayer; Julie L Daniels Journal: Epidemiology Date: 2015-01 Impact factor: 4.822
Authors: Kristen M Rappazzo; Julie L Daniels; Lynne C Messer; Charles Poole; Danelle T Lobdell Journal: Environ Health Perspect Date: 2014-05-30 Impact factor: 9.031