Simultaneous interval estimates of the odds ratio in studies with two or more comparisons.