Associations of regular glucosamine use with all-cause and cause-specific mortality: causality assumptions need to be checked.