The assumptions behind the forecast
We analyzed three anonymized proposals from large global CROs for the same study. Identical protocol, identical patient population.
| Measure | Range | Variance |
|---|---|---|
| Patients per site per month | 0.18 to 2.51 | 14x |
| Screen-failure assumption | 15% to 60% | 4x |
| Sites proposed | 2 to 38 | 19x |
| Enrollment period | 30 to 60 months | 2x |
None of the three is necessarily wrong. But the sponsor chose one without seeing the reasoning, then built the timeline, the budget and the site list on it. A rate that can vary fourteen-fold enters at the top of the funnel and is multiplied through every stage below it. And nobody questions the stages, because the rate was produced to win the bid, not to be true, and no one was asked to check it.