The article touches a little bit about how responses to a 0-27 point depression questionnaire are coded as 0 or 1, 1 being "has moderate to severe depression" and 0 being "does not have moderate to severe depression" at a threshold of 10 points or higher. Similar situation for suicidality. Instead of tracking the actual point changes across the 12 months, they just tracked the pooled percentages of 0 vs. 1.