I think people who never drink, may tend to do so for religious reasons. If someone has never had a drink, they could have a tendency for alcoholism that never has a chance to manifest itself. So maybe that justifies excluding teetotalers from the data?