The key thing to note is that 13/11 to 15/11 is an "incident" and 01/12/2013 to 02/12/2013 is second incident.
The calc then used (No of instances) * (No of instances) * (No of days off)
so 2x2x5 to give 20
The problem is calculating no of incidents, since absence could span weekend, but weekend dates are not normally included in data.
In the attached I had added a couple of entries for a weekend, just to ensure I was trapping them (excluding). I have also added a column which links dates together as incidents (Values A,B etc). In the real data this does not exist.