The statistical analysis in the article by Lindén and Alm,1 which was published in the July 1997 issue of the ARCHIVES, is incorrect. The data were analyzed as if they came from a 2×2 factorial experiment, but such an analysis would be appropriate only if each subject received only 1 treatment (placebo, latanoprost alone, physostigmine alone, or both medications). In fact, every subject received every treatment combination, the outcomes at each point are correlated, and the estimates of SE are wrong. Also, no allowance is made for multiple significance tests over time. One approach to these data would be to fit a statistical model with a random intercept and serial correlation using a software program for mixed models.
Cox TA. Latanoprost and Physostigmine in Human Eyes. Arch Ophthalmol. 1998;116(3):398-399. doi: