Levels in the follow-up of a long-term clinical process.

I developed a 5-stage, 7-compartment prognostic model for the long-term follow-up of 28 patients with sickle chronic lung disease. Estimated hazard functions for transition varied with Stage 1, 2, 3, or 4 of sickle chronic lung disease as well as with stage at diagnosis and duration of observation with hematologic disease before the recognition of sickle chronic lung disease. Individualized hazard matrices produced 28 prognostic probability functions of time that had heterogeneous levels and shapes. The probability of death from the time of diagnosis to the end of each patients period of follow-up determined that patients prognostic integrated hazard. Such integrated hazard values, developed from a prognostic model, provide a superior context for carrying out standard formal tests and inferences regarding the effects of alternative treatments in randomized clinical trials. (Epi? demiology 1994;5:35-41)