How does Prism compute the 95% confidence interval for the difference between two proportions?

As part of the contingency table analysis, Prism reports the 95% confidence interval for the difference between two proportions (which in some experimental designs is the attributable risk). Prism does this using a simple equation:

FractionInTopRow = A / (A + B)
FractionInBottomRow = C / (C + D)
Diff = ABS(FractionInTopRow - FractionInBottomRow)
P = (A + C) / (A + B + C + D)
SEofDiff = Sqrt[P * (1 - P) * (1/(A + B) + 1/(C + D))]
UpperCI = Diff + CriticalZ(alpha, tail) * SEofDiff
LowerCI = Diff - CriticalZ(alpha, tail) * SEofDiff

Newcombe (1) reviewed this problem, and discusses 11 different ways to compute the confidence interval for the difference between two proportions. Prism (up to version 6) uses his method 1, which is the only method discussed in most texts. In the future, we'll probably create an option  and allow his method 11 as an alternative. 

1.    Newcombe, R. G. R. Interval estimation for the difference between independent proportions: comparison of eleven methods. Statistics in medicine 17, 873–890 (1998).