Patient-based forecast models follow patient cohorts through a framework which is the forecasters best understanding of a markets dynamics at a given time. Each forecast tells a story through which we can see all expected events and their impacts over a particular time period of interest. The longer this period is, the greater the risk that unexpected events will impact the model.
The primary driver of a patient-based model is generally the treated population being fed into the model. This population interacts with the dynamics set up by the forecaster to produce outputs which hopefully are indicative of what is happening in the market. Usually this is a relatively easy population to derive if you understand the pathway of patients with a disease and how they are typically treated.
Issues may arise however when there is a substantial shift in the treatment algorithm of patients driven by new techniques or therapies occurring upstream of your market of interest, potentially leading to substantial unaccounted for trickle down effects which are not being replicated in the real world. One example of this could be the approval of sunitinib as an adjuvant therapy for Renal Cell Carcinoma prior to this event resection was the only possible treatment for locally advanced or high risk RCC. This approval could significantly improve patient outcomes and reduce the number of patients who enter the first-line metastatic treatment setting, Which in turn would drive different outputs for subsequent treatment populations.
Any downstream market forecasts need to be aware of upstream events if they are to provide a good picture of true market dynamics over time, a forecaster should always be looking for potential changes to a treatment algorithm and how this could impact their model.Henry Barling