1. A linear term for year the pregnancy attempt began was included in our statistical models to control for bias that could have been introduced by trends over calendar time in the opportunity for mercury exposure.56 Time trends may induce a spurious association between individual hygiene factors and time to pregnancy. Because we sampled on a woman's most recent pregnancy, women who took a long time to become pregnant (several years) had a greater opportunity to be exposed to poor mercury hygiene practices that were more common in the past than women who became pregnant more quickly and therefore, on average, more recently