Abstract
Purpose: :
Glaucoma treatment failure often takes several IOP measurements to establish, as clinicians generally accept a range of pressures with outliers being justified by other clinical indicators. We investigate whether a single IOP reading can, by itself, indicate treatment failure within acceptable statistical parameters.
Methods: :
IOP’s of 30 eyes were measured multiple times on consecutive visits using a single Goldmann applanation tonometer by a single examiner with a minimum of 12 data points per patient collected and analyzed by an independent examiner. Acceptable IOP control is defined as 95% of values measuring lower than a clinically determined goal. Assuming a Gaussian distribution, the mean of the target IOP range must therefore lay 1.28 standard deviations below the maximal goal IOP. Pressure values 2.33 standard deviations above the target mean have less than a 0.5% probability of lying within the target range. Thus, an acceptable range of variability extends 1.05 (2.33 - 1.28) standard deviations above the maximal target goal. Any value outside of these parameters should be considered unacceptable statistically and hence indicative of treatment failure. Having established a numerical parameter for treatment failure, secondary analyses will focus on rates and prevalence of disease progression comparing individuals at goal versus those not at goal.
Results: :
The highest observed intra-eye standard deviation of IOP measurements was 2.27. Conservatively, calculations were performed with a supranormal standard deviation of 2.57. This figure when applied to the target range formula yielded a value of 6.0 mm Hg greater than the mean goal IOP or 2.70 mm Hg above the maximal acceptable IOP as indicative of treatment failure. Correlation to rates of disease progression is ongoing.
Conclusions: :
Within the statistical constraints chosen of a sampling error of +/- 2 mm Hg and a 95% confidence interval, a patient with a single IOP reading at least 6 mm Hg above mean goal IOP can be said to have a 99.5% probability of failing therapy based on this single IOP measurement.
Keywords: intraocular pressure • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology