In message <[log in to unmask]>,
[log in to unmask] writes
>
>If so, then 28/1255 in the active group would have suffered raised liver
>enzymes
> and 8/1255 in the placebo group would have suffered raised liver
>enzymes
>
>Intuitively this looks to be a real difference. Statistically I'm not so
>sure.
>On the basis that they may not be 'statistically different' then the
>company
>might be justified in saying they are 'comparable'.
>
They are:
Test and CI for Two Proportions
Sample X N Sample p
1 28 1255 0.022311
2 8 1255 0.006375
Estimate for p(1) - p(2): 0.0159363
95% CI for p(1) - p(2): (0.00665426, 0.0252182)
Test for p(1) - p(2) = 0 (vs not = 0): Z = 3.37 P-Value = 0.001
Looks like a statistically significant difference to me! The 95% CI
gives NNH from 150 to 40. The important question is not so much whether
the treatment is significantly more likely to cause raised liver enzymes
than placebo (it is!) but how harmful to patients these raised liver
enzymes are compared to the benefit they might get from the treatment.
Toby
--
Toby Lipman
General practitioner, Newcastle upon Tyne
Northern and Yorkshire research training fellow
Tel 0191-2811060 (home), 0191-2437000 (surgery)
Northern and Yorkshire Evidence-Based Practice Workshops
http://www.eb-practice.fsnet.co.uk/
|