Subjects
This study included data collected in the first two phases of the OHTS study through December 30, 2008. The design of the OHTS has been described previously (www.clinicaltrials.gov, registration number NCT00000125).1 Briefly, the study was conducted in three phases: the first phase (OHTS Phase 1) was a randomized clinical trial conducted from February 28, 1994, to June 2, 2002. Between February 1994 and October 1996, 1636 participants with OHT were randomized to receive either topical ocular hypotensive medication (medication group) or close observation (observation group). The second phase (OHTS Phase 2) was conducted from June 3, 2002, to December 30, 2008. During this phase, both groups received treatment: the original medication group continued to receive treatment (early medication group), and hypotensive treatment was offered to the original observation group (delayed medication group). In the third phase (OHTS Phase 3), treatment was no longer determined by the study protocol and was not included in our analysis.
All participants in the OHTS signed a statement of informed consent approved by the institutional review board of each participating clinic. The study adhered to the tenets of the Declaration of Helsinki and was in compliance with the Health Insurance Portability and Accountability Act.
Visual field data
We included 2,749,398 test points from 58,115 visual fields of 1,188 patients (2,369 eyes) that participated in the OHTS and met the following criteria: 1) a series of at least six reliable visual fields which were performed over at least six years of follow-up and 2) each eye was required to have at least two qualifying visual field tests with normal GHT, normal PSD (P < 5%), and less than 33% fixation losses, false positive results, and false negative (as per the OHTS criteria).13
Since the early visual field testing paradigm used the 30-2 pattern and did not use the SITA algorithm, to maintain consistency the 52 locations of the 24-2 grid (54 minus 2 points for the blind spot) were retained from the 30-2 grid. A correction factor of +1.0 dB was applied to threshold sensitivities measured using the Full Threshold algorithm to permit a comparison to those measured with the SITA family of algorithms.14 The number of VF tests and length of follow-up for the early and delayed medication groups are presented in Figure 2.
Schematic model to predict the points most likely to progress on the 24-2
The vulnerability regions described by Hood et al. include the temporal half of the superior and inferior quadrants of the disc (i.e., 45 to 90 degrees and -45 to -90 degrees in figure 1a.), defined as the SVZ and IVZ, respectively. Although the SVZ and IVZ represent a relatively small (45°) region of the disc, defects in these regions can still vary in location, depth, and width, as well as homogeneity. Thus, the corresponding VF defects seen on a 24-2 VF can show a wide range of patterns.13 Based on previous work by Janosnious et al.8 we identified for each hemifield the points on the 24-2 that are expected to have the greatest overlap (top quartile) with the corresponding disc vulnerability zone. As depicted in figure 1b (orange circles), the points are located in the nasal and paracentral region. We hypothesize that these locations would show progression more commonly than other VF points.
Identifying Vulnerability Regions on the 24-2 using the OHTS dataset
Data cleaning, analysis, summarization, visualization, and manuscript preparation were performed using the R statistical programming language.15–21 In order to identify the regions vulnerable for progression on the 24-2 VF, two independent endpoints were devised to define pointwise progressive loss using both event-based and trend-based analysis. In an event-based analysis, each observation is compared to a reference, and a binary event occurrence is determined. In this study, threshold sensitivity cutoffs were defined by the 5% limit of variability within all OHTS patients using the qualification test and test retest for each point (Figure 3).
A pointwise event was defined as a single field point having three consecutive threshold sensitivity observations below the respective pointwise cut off, and the time of the event was defined as the time of the first observation in the triplet (Figure 4). As follows from this definition, one eye was permitted to have multiple pointwise events (e.g., two separate points satisfying the above criteria).
Trend analysis with pointwise linear regression utilizes all eligible data and their relationship to time to determine the rate of change. For every eye, the rate of change in threshold sensitivity was determined at each of the 52 test points by extracting the slope from a linear regression model of threshold (dB) versus time. Figure 5 provides an example of this analysis for one eye. The mean slope (dB/year) was calculated and compared between the two groups for each VF point.