Abstract
Purpose: :
Recently published practical recommendations suggest 3 visual field (VF) tests per year over 2 years are required to estimate rapid progression (Humphrey MD loss of -2dB per year) in a patient with average VF variability [1]. We examine the hypothesis that estimates would be improved if the tests were done at the beginning and end of the 2 year period.
Methods: :
Computer simulated ‘patients’ (n=100000) were given a MD loss of -2dB per year with added measurement variability as previously described [1]. Linear regression of MD against time was used to estimate rates of loss and to identify these rapid progressors with a criterion of -1 dB per year at P<0.05 over a 2 year period. One group of ‘patients’ was measured every 4 months (7 VFs) whilst the ‘wait and see’ group were measured 3 times at baseline (in the first 3 weeks) and 3 times at the end of the follow-up (6 VFs). The time to detection was recorded (starting with the 4th visit at 1 year). We also generated stable patients (0dB loss per year) to examine the effect of the follow-up patterns on false positive (FP) detection. The simulations were performed in the statistical programming environment R.
Results: :
With the regular 4 monthly follow-up, 83% of rapid progressors were correctly detected as changing by 2 years. This power of detection significantly improved to 95% with the ‘wait and see’ follow up (P<0.0001). In addition the slopes (rates of loss) were better estimated, with SD of slopes from ‘wait and see’ being 0.41dB compared to 0.56dB in the regular follow-up (P<0.0001). When corrected for overall numbers truly detected, average detection time was slightly quicker (difference of 3.1 months) with the regular follow up when compared to ‘wait and see’. However, the sequential testing of regular follow-up incurred a large increase in FPs: 5.9% compared to a much reduced 0.4% in ’wait and see’ because of the improved estimation of the slopes (P<0.0001).
Keywords: visual fields • perimetry • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology