Is the observed lowering of intraocular pressure due to treatment?

OBJECTIVE: Use Bayes' theorem to estimate the intraocular pressure (IOP) lowering effect of medical treatment initiated for glaucoma and determine if IOP comparisons to the baseline IOP of the same eye is clinically useful. MATERIALS AND METHODS: The probability that treatment with prostaglandin is responsible for an observed 20% decrease in IOP with prostaglandin treatment was calculated using Bayes' theorem using the following available information: the probability of a 20% decrease in IOP given treatment with prostaglandin, the probability of a treatment effect using prostaglandin and the overall probability of a 20% decrease in IOP. The calculations were repeated to account for a possible 2 mmHg overestimation of effect caused by measurement error in performing applanation tonometry. RESULTS: The probability that treatment is responsible for an observed 20% decrease in IOP following initiation of treatment with a prostaglandin was 99%. After adjusting for measurement error this probability was 98%. Obtaining two IOP measurements marginally increased the probability. CONCLUSION: Following initiation of treatment with prostaglandin, Bayes' theorem allows us to infer that treatment effect is the most likely explanation for an observed 20% decrease in IOP from the baseline; this inference remains even after adjusting for known measurement error. The high probability of a treatment effect is due to the high prior odds of treatment effect and the high likelihood ratio for prostaglandin producing such an effect. If data is available, similar calculations can be used for other percentage decreases, other medications and for the monocular trial.