By Gary Johnson, expert-trainer of The Pharma Forecasting Course and the How to Price Better than Your Competitors for Successful Market Access course
The sales of a drug, in a given country, is the result of a cascade. Starting ‘at the top’ of the cascade, there are a certain number of people in the country. Of these, a certain proportion are afflicted by the disease in question. Of these, a certain proportion are diagnosed, … prescribed, … treated with our therapy class, ... treated with our brand. This approach to forecasting is called epidemiology-based forecasting, as opposed to forecasts based on historical sales data. Sometimes, to adopt this approach to forecasting is helpful, but often it causes unnecessary work and actually decreases the accuracy of your forecast.
It is a cardinal sin in forecasting to use models to tell the market it is wrong! If you are forecasting an antihypertensive, the diagnosis and prescription rates for this disease have been stable for many years. The market has spoken. Unless there is a major technological change (like a cure for hypertension), it is unlikely that these rates will change very much. Hence the number of treated hypertension patients can simply be extrapolated and all your forecasting effort can be concentrated where it belongs: where the action is – in this case, what share of antihypertensive is your class and your brand going to capture. In such case, a sales-based forecasting approach will thus suffice.
Of course there are times when the prescription and diagnosis rates of a disease can be expected to change markedly (as with the introduction of Viagra for erectile dysfunction) and then an Epi-based Forecast can be highly appropriate.
The point is that the best approach to forecasting is situational and should not involve the application of one rigid approach.