Testing volumes are low, meaning we fail to capture a sizeable chunk of cases.

The testing strategy is bad. We’re testing people that are less likely to be positive.

And possibly, the model is just wrong.