The relationships between predictors (e.g., LR) and the outcomes (e.g., initial remission) were evaluated with a series of DTSAs constructed using a latent hazard function representing the distribution of the time of AUD remission (see Figure 1). Discrete-time hazard is the conditional probability that an individual will experience the event (e.g., remission of an AUD) at an age, given he did not have the event at an earlier time point (Singer and Willett, 1991; Willett and Singer, 1993). The resulting pattern of remissions (hazard function) was then tested for significant changes over time using the Likelihood Ratio Test (LRT), comparing a model constrained for no time change vs. an unconditional model.
