Smoking remains associated with education after controlling for social background and genetic factors in a study of 18 twin cohorts
Karri Silventoinen, Maarit Piirtola, Aline Jelenkovic, Reijo Sund, Ádám Domonkos Tárnoki, Dávid László Tárnoki, Emanuela Medda, Lorenza Nisticò, Virgilia Toccaceli, Chika Honda, Fujio Inui, Rie Tomizawa, Mikio Watanabe, Norio Sakai, Margaret Gatz, David A. Butler, Jooyeon Lee, Soo Ji Lee, Joohon Sung, Carol E. Franz, William S. Kremen, Michael J. Lyons, Cathérine Derom, Robert Vlietinck, Ruth J. F. Loos, Per Tynelius, Finn Rasmussen, Nicholas G. Martin, Sarah E. Medland, Grant W. Montgomery, Ingunn Brandt, Thomas Sevenius Nilsen, Jennifer R. Harris, Jessica Tyler, John L. Hopper, Patrik K. E. Magnusson, Nancy L. Pedersen, Anna K. Dahl Aslan, Juan R. Ordoñana, Juan F. Sánchez-Romera, Lucía Colodro‐Conde, Esther Rebato, Dongfeng Zhang, Zengchang Pang, Qihua Tan, Judy L. Silberg, Hermine H. Maes, Dorret I. Boomsma, Thorkild I. A. Sørensen, Tellervo Korhonen, Jaakko Kaprio
Abstract
We tested the causality between education and smoking using the natural experiment of discordant twin pairs allowing to optimally control for background genetic and childhood social factors. Data from 18 cohorts including 10,527 monozygotic (MZ) and same-sex dizygotic (DZ) twin pairs discordant for education and smoking were analyzed by linear fixed effects regression models. Within twin pairs, education levels were lower among the currently smoking than among the never smoking co-twins and this education difference was larger within DZ than MZ pairs. Similarly, education levels were higher among former smoking than among currently smoking co-twins, and this difference was larger within DZ pairs. Our results support the hypothesis of a causal effect of education on both current smoking status and smoking cessation. However, the even greater intra-pair differences within DZ pairs, who share only 50% of their segregating genes, provide evidence that shared genetic factors also contribute to these associations.