We also used the discrete-time survival analysis method to estimate the survival function for remission and hazard function for relapse into problem use. The survival probability was based on the model including only the time indicators and illustrates the cumulative probability of remitting over time. Hazard function graphs were created for the full sample over time by remission status at 1 year, using the model with time indictors and remission status at 1 year (abstinent, non-problem user). This graph describes the probability of relapsing back into problem use over time.
