Using Multiple Imputation to Address the Inconsistent Distribution of a Controlling Variable When Modeling an Infrequent Outcome

• Activity Number: 428
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Yujia Zhang* and Sara Crawford and Sheree Boulet and Michael Monsour and Bruce Cohen and Patricia McKane and Karen Freeman
• Companies: CDC and CDC and CDC and CDC and Massachusetts Department of Public Health and Michigan Department of Community Health and Florida Department of Health
• Keywords: : assisted reproductive technology ; inconsistent data distribution ; multiple imputation ; weighted sequential hot deck

Abstract

Temporal changes in methods to collect longitudinal data can generate inconsistent distributions of affected variables,but effects on parameter estimates have not been well described.We examined differences in Apgar scores of infants born in 2000-2006 to women with ovulatory dysfunction(risk)or tubal obstruction(reference)who underwent assisted reproductive technology(ART),using FL,MA,and MI birth certificate data linked to the Centers for Disease Control and Prevention's ART Surveillance database.FL had inconsistent information on induction of labor from a 2004 change in birth certificate format.To control for bias that may be introduced by the inconsistent distribution,we used multiple imputed data by Cox-Iannacchione weighted sequential hotdeck method for labor induction values in FL data collected before this change, and missing values in FL data collected after the change,as well as missing values for MA and MI data. The adjusted odds ratios for low Apgar score were 1.94(95%CI 1.32-2.85)using imputed and 1.83(95%CI 1.20-2.80)using non-imputed data. Compared with multiple imputation, the obtained estimate by original data was biased towards null with inflated standard errors.