August 4, 1993

Infant Birth Weight and Smoking Reduction-Reply

Author Affiliations

University of Alabama Birmingham

JAMA. 1993;270(5):579. doi:10.1001/jama.1993.03510050045023

In Reply.  —Dr Goldin raised a good point in asking whether "self-selection bias" could have influenced the results of our study. This is a potential problem whenever an intervention requires an active personal decision such as to stop smoking, to lose weight, or to quit drinking. To try to reduce this possibility, our study had controlled for many confounding variables including race, mother's age, height, weight, gestational age at birth, and the intervention status (experimental or control). To address Goldin's concern further, we reran our regression analyses with the following additional variables: parity, number of preterm deliveries, number of abortions, history of diabetes, history of hypertension, gender of the infant, and baseline hemotocrit level. Differences in the adjusted infant birth weight between quitters (Q) and no changers (NC), and between reducers (R) and no changers were slightly increased: 269 g (3285 g [Q] vs 3016 g [NC]) and 104 g