Purchase this article with an account.
J. Huang, J.A. Young, S. Reddy, C.C. Teng, E. Nissan; A Statistical Approach to Determining the Number of IOP Measurements Necessary to Establish Treatment Therapy . Invest. Ophthalmol. Vis. Sci. 2006;47(13):430.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Maintanance of adequate intraocular pressure (IOP) control is the mainstay of glaucoma therapy. To achieve this goal in any particular patient, medication regimens often need to be altered. However, prematurely judging a medication to be ineffective or delaying recognition of a truly ineffective medication results in poor patient care. We propose a statistical approach to determine, for an individual patient, the minimum number of measurements required to determine treatment failure.
The intraocular pressures of twenty eyes of 10 patients were obtained by a single examiner on consecutive visits using Goldmann applanation tonometry. For each eye, the IOP measurements were sequentially evaluated to establish a running mean, variance and one–tailed t–statistic. A "goal" IOP was arbitrarily chosen at a level 2 mmHg below the calculated mean for the entire sample set. A one–tailed Students t–test compared IOP measurements from a single eye to an ideal target range (defined by no more than 5% of IOP’s above goal and using the variance of the measured sample as an estimator of the variance of the target range) given a 95% confidence interval.
Detection of statistically significant treatment failure was successfully accomplished in all twenty eyes with a mean of 2.9 measurements to detection (range 2 to 7 measurements). Distance between target mean IOP and goal IOP demonstrated a mean of 2.84 (range 0.95 to 5.7) and correlated with sample IOP variance (coefficient of correlation = 0.71).
Our study quantifies the idea of a target IOP range rather than a single goal IOP. The mean of this range is necessarily less than the goal IOP and the degree of this difference is a function of the variance of the individual patient’s IOP measurements. We were successful in producing a model which yields a minimum number of IOP measurements required to demonstrate a statistically significant difference between a patient’s measured IOP’s and that patient’s target range.
This PDF is available to Subscribers Only