The stress from prestressing at the bottom flange of the section is P*e/S + P/A. See figure 7.
First, Pe in the problem is misleading since the "e" subscript is used to mean effective. So the bending effect of prestress is Pe*e/S.
e=h/3
The area to use for this problem is the area of the beam (40in^2). The problem says to assume losses occur before the deck, so the correct area is the girder area (no creep into the deck section)
However, the S to use is the composite section since that is there to resist bending. This is given in Ex 3.18 as 135in^3
Just coincidentally for the girder S=bh^2/6=66.65in^3. In this problem ONLY (don't apply this as a rule) the S to use for the composite section is (nearly) equal to 2 times the S for the girder.
So take the prestress equation above and put all that in in terms of the girder.
P*e/S is equal to P*(h/3)/(2*b*h^2/6) which reduces to P/b*h = P/A
Then there is the P/A from that portion of the equation to get the 2P/A.
So I think the math works, but I totally get the confusion.
As an aside, I don't understand these kind of reductions. What is the point? Prestressing extreme fiber stress is always P*e/S+P/A with proper signs. There is no reason to combine those terms. A similar one I see all the time in geotech reports is pci for subgrade modulus. While this is mathematically correct, psi/i, the intent is the pressure to displace one inch (not to be confused with a density or unit weight).
Good luck!