Adaptive analyses are presented of a trichotomous respiratory status data set with an outcome having three possible values over a baseline and four subsequent clinic visits using discrete regression with multinomial, ordinal, or censored Poisson probabilities. Results are compared for partially modified generalized estimating equations (GEE), fully modified GEE, and extended linear mixed modeling (ELMM). Linearity for the probabilities in visit with constant dispersions is addressed as well as whether unit dispersions are appropriate for these data, a comparison to standard GEE, and the dependence of means and dispersions on visit. Adaptive additive and adaptive moderation models are generated for visit and being on active treatment. A comparison to the standard linear moderation model is provided as well as direct variance modeling of trichotomous respiratory status levels. Models based on directly specified correlation structures are compared to models based on random effects/coefficients. A summary of the analysis results is also provided. SAS code for generating these analyses is described along with output generated by that code.