# Cannot replicate odds ratio

I am trying to replicate some odds ratio calculations which are reported in this paper. For example in the paper on page 3 in the Strome et al. data for tonsil: it states there are 52 cases and 48 controls. HPV % cases = 40% and controls = 6%.

Converting the percentages to raw figures, I get the following 2×2 table:

               Case  Control
HPV positive     21        3
HPV negative     31       45


Therefore the odds ratio should $=\frac{21\cdot45}{31\cdot3} = 10.16$.

However the paper reports this particular odds ratio as 18.2 on the following page. Does anyone know why this is? Are they using a different formula?

• Actually if you look at figure 1, the Strome 2002 data shows odds ratio of: 10.2 (2.8, 37.0), so that 18.2 looks like a typo to me. Also in figure 2 they report the OR as 10.2 from Strome data. – bdeonovic Mar 9 '14 at 0:09