UC San Diego

Background : Autoimmune (AI) diseases unlike most other chronic diseases in pregnancy have been understudied. The objectives of this dissertation were to evaluate the impact of AI disease severity on pregnancy outcomes, to examine how AI diseases affect the relationship between a common comorbid condition, nausea and vomiting in pregnancy (NVP), and spontaneous abortion (SAB), and lastly to explore missingness of HC in pregnant women with and without AI diseases. Methods : Secondary analyses were carried out on data collected prospectively in the OTIS Autoimmune Diseases in Pregnancy Project between 2004 and 2011. Analyses were carried out on different subsets of the main study to: (1) examine the relationship between rheumatoid arthritis (RA) disease severity measured during early pregnancy and selected pregnancy outcomes, (2) assess impact of maternal AI status on the occurrence of NVP and examine how NVP in these women is related to differential risk of SAB, and (3) examine a methodological issue on patterns and factors associated with missing exposure variables relevant to pregnancy cohort studies such as the OTIS Autoimmune Diseases in Pregnancy Project : whether "missingness" of head circumference (HC) measurement at birth in this dataset occurred completely at random or is systematically biased in ways that might compromise the validity of interpretation. Results: After adjusting for covariates, RA disease severity in early pregnancy was predictive of increased risk for preterm delivery in women with AI disease (aRR=1.55, 95% CI: 1.17, 2.07). Overall adjusted HR for NVP and SAB was 0.65 (95% CI: 0.29, 1.48). AI disease status did not affect the association between NVP and SAB. Adjusting for covariates, non-availability of medical records was strongly associated with missing HC. Conclusion : Disease severity in early pregnancy in women with RA is predictive of preterm delivery. There was no differential association between NVP and SAB in women with AI diseases. Finally, missing medical records was significantly associated with missing HC and we need to be cautious when interpreting results with missing covariates especially when the rate of missingness is not mentioned