Should we adjust for pregnancy history when the exposure effect is transient?