OBJECTIVES: We want to develop a parametric model to predict mortality for future patients with a disease that have specific demographic and clinical characteristics while considering patient trajectories over time for a set of biomarkers, which are critical predictors of disease progression.  This is an important tool for many health economic projections. METHODS: In time-to-event studies, longitudinal measures are collected for important disease progression biomarkers.  Using only the last available value of these measures in survival models discards important information from the longitudinal evolution.  We used data from a 3-year observational study to estimate the covariate coefficients in a Cox-proportional hazards model in the presence of time-dependent biomarkers via SAS® PHREG.  In addition, we applied a Weibull accelerated failure time model to estimate the scale/shape of a parametric survival distribution using SAS®LIFEREG, which, unlike PHREG, does not allow for direct incorporation of longitudinal measures.  In developing the final prediction model, we combined the coefficients from PHREG and the scale/shape from LIFEREG to compute the probability of survival. RESULTS: By applying the Weibull model without considering patient trajectories over time, we predicted a 3-year survival rate of 55.1%.  However, the hybrid combination approach of the Cox/Weibull model, predicted a more accurate 3-year survival rate of 46.7%, which fell within the confidence bounds of the original observational study.  CONCLUSIONS: Ignoring the additional variability of patient trajectories over time, when modeling survival, can lead to biased estimates.  We have implemented a hybrid approach by which we incorporated the impact of time-dependent biomarkers of the disease along with the scale/shape of a parametric survival distribution to more accurately project survival time in health economic modeling.