The starting point for any sales forecast is a forecast of the patient population over the same time horizon. Long-term patient forecasts are not common in the published epidemiological literature, and the level of detail required for a commercial forecast far exceeds the typical detail in the secondary sources. Academically trained, but commercially inclined, our epidemiologists are driven by the market and the product, going well beyond the usual diagnostic boundaries that often overlook significant opportunity.

In order to provide our clients with a complete picture of their study population, the epidemiology analysis includes three phases. These are the same whether the need is for a simple prevalence estimate of a stable disease population, or a detailed patient flow analysis.

Phase I – Data Collection
This phase begins with an exhaustive review of the scientific literature, including:
  • Clinical trials
  • Published meta-analyses
  • Epidemiologic databases (NHANES, NCS, SEER, HCUP, etc.)
  • Published epidemiological studies

This identifies the extent to which the patient population is understood and studied, and allows us to build a sourced set of raw data to frame the subsequent analysis.

Phase II – Meta-Analysis

Once the universe of available data has been established, the next stage is to blend it into a cohesive picture of the patient population. Every patient population is derived from some combination of meta-analysis and modeling.

In the case of a straightforward prevalence analysis for a relatively stable disease state, the meta-analysis combines all available data, allowing the user to weigh individual studies against each other on an equivalent basis, yielding a set of age-specific prevalence rates that can then be applied to population projections for the final forecast.

In the case of a patient flow analysis, the meta-analysis provides a series of detailed outputs that describe the way patients move from one state of health to others. These are foundational to the patient flow model, which starts with incidence of new cases of illness, and tracks the incident cohorts over time, building a set of commercially relevant patient pools.

Often, a patient population is defined by a number of criteria, including the presence of specific signs and symptoms, risk factors, the absence of certain contraindications etc., any of which may overlap. This defines a set of comorbid segments, and care must be taken to neither neglect, nor double-count patients within this schema. As the prevalence of the individual components vary over time, and not always in lockstep, the comorbid subsegments that comprise them change as well. We have developed proprietary techniques for addressing this problem, allowing us to develop estimates that preserve the Odds Ratio between any particular pair of populations, for an arbitrary number of comorbidities, over time.

Phase III – Deliverables

The last component of the epidemiology analysis is the deliverable product to our client, typically a spreadsheet model of the population, with associated documentation of the sources and methodology. With experience, robust data collection practices, solid statistical analyses, and custom detailed deliverables, Humanumeric can help our clients to define and model the precise target population for their application.