The sample was selected, retrospectively, from a cohort of 68
glaucoma suspects and 27 patients with glaucoma who had been followed
prospectively over approximately 12 months (mean, 12.75 months; SD,
2.29 months). Informed consent was obtained from all volunteers. The
study was carried out in accordance with the tenets of the Helsinki
agreement and approval was obtained from the Medical Ethics Committee
of Manchester Royal Eye Hospital and Manchester University.
To be classified as a glaucoma suspect in the given eye, one or more of
the following inclusion criteria was met in addition to a normal W-W
field: POAG in the fellow eye; a presenting IOP of ≥26 mm Hg and a
vertical cup to disc (CD) ratio ≥ 0.6; a presenting IOP of ≥26
mm Hg and a positive family history for glaucoma; a presenting IOP of≥
21 mm Hg and a between-eye asymmetry in CD ratio ≥ 0.2; a
presenting IOP of ≥21 mm Hg and a CD ratio of ≥0.8; a presenting IOP
of ≥21 mm Hg and the presence of any focal disc abnormality, notching
or disc hemorrhage; or a presenting IOP of ≥30 mm Hg. A normal visual
field (HFA Program 30-2; Full Threshold algorithm) was defined by Total
and Pattern Deviation probability analysis at the baseline visual
fields. IOP was determined by Goldmann applanation tonometry.
For the POAG patients, a repeatable, glaucomatous W-W visual field
defect (HFA Program 30-2; Full Threshold algorithm) at baseline was
required to coexist with an abnormal optic disc, also consistent with a
diagnosis of glaucoma (including increase in cup size, increase in cup
disc ratio, disc asymmetry, changes in the lamina cribrosa, loss of
neuroretinal rim, pallor, evidence of peripapillary atrophy, vessel
changes, or disc margin hemorrhage),
15 and a presenting
IOP of >21 mm Hg. In patients where both eyes fulfilled the inclusion
criteria, one eye was arbitrarily selected.
The exclusion criteria for both groups comprised a visual acuity in the
designated eye of worse than 6/9; clinically significant cataract
determined by slit-lamp examination with dilated pupils; a history of
congenital color vision defect or optic nerve disorder not attributable
to glaucoma; previous intraocular surgery, ocular trauma or
inflammation; gonioscopic evidence of anterior chamber abnormality or
angle closure; a history of CNS disorder; systemic medication known to
affect the visual field; and ametropia of ≥ 6.00 DS and ≥2.50
DC.
The designated eye of each of the glaucoma suspects and each of the
POAG patients was prospectively monitored for six visits over the
follow-up period. Optic nerve head stability was assessed using two
separate criteria. At each visit, the optic nerve head in the
designated eye was evaluated by stereo observation using slit-lamp
biomicroscopy and by scanning laser tomography (Heidelberg Retinal
Tomograph [HRT]; Heidelberg Engineering, Heidelberg, Germany). The
stereo examination at each visit evaluated features of the optic nerve
head (as described above) for progressive damage.
15 Twelve
stereometric parameters were obtained from seven image series at each
visit to describe the optic nerve head topography (disc area, mean
height of contour, height variation in contour, cup volume, rim volume,
volume above and below surface, mean depth inside contour, and mean
retinal nerve fiber layer thickness). Stability was defined as no
significant change over the time to follow-up in the parameters using
repeated measures analysis of variance (
P > 0.01).
Although there are currently no standard criteria for defining change
using the HRT, the approach adopted was intended to ensure a
conservative definition of stability. The optic nerve head had to meet
both criteria to be classified as stable.
The visual field was determined for W-W perimetry using the Full
Threshold algorithm of the Humphrey Field Analyzer 640, with Program
30-2 (stimulus size III) and for SWAP using the Full Threshold
algorithm with Program 24-2 and the default stimulus parameters of a
blue size V stimulus and the 100 cdm
−2 broadband
yellow background. Program 24-2 was used for SWAP to provide an
examination of duration similar to W-W perimetry.
11 The
order of the type of perimetry (i.e., W-W or SWAP) was randomly
assigned between patients but remained constant for each patient at
each examination throughout the period of follow-up. Each visual field
was deemed to be reliable in terms of the responses to the catch trials
(≤33% fixation losses, ≤33% false-positive responses, and ≤33%
false-negative responses), and all visual field examinations were
obtained by a single examiner (SLH).
All patients had experienced W-W perimetry before commencing the study,
and the cohort was deliberately biased toward patients with early
visual field loss. The first visual field examination was repeated
within 2 weeks, and both examinations were designated as the baseline.
These fields were excluded from further analysis. The purpose of the
baseline examinations was to ensure that each patient was familiar with
the measurement procedure and allowed the examiner to subjectively
assess a patient’s reliability. The mean interval between the third
and the seventh examinations for the sample as a whole was 11.88 months
(SD, 1.32; median, 11.74 months; range, 9.84–15.24 months).
The W-W and SWAP visual fields were reviewed by one of the authors
(JMW), experienced in visual field interpretation, who was masked to
the outcome of any other clinical findings. The W-W visual fields were
assessed by inspection of the Overview, Change Analysis and Glaucoma
Change Probability Analysis print-outs of the HFA STATPAC statistical
software. The SWAP visual fields were assessed by evaluation of the
Overview print-out, alone, as the Change Analysis and the Glaucoma
Change Probability Analysis print-outs are not commercially available.
Linear regression analyses were also carried out for the W-W and SWAP
summary visual field indices over time to follow-up for each of the
glaucoma suspects and for each of the POAG patients. The regression
analyses determined whether a significant change (
P <
0.05) was evident in the magnitude of the Pattern Standard Deviation
(PSD) and Corrected Pattern Standard Deviation (CPSD) indices over the
time of follow-up. The PSD and CPSD indices were chosen as the most
likely to indicate change consistent with glaucoma.
16 A
second linear regression analysis, only applicable to the POAG
patients, determined if there was any significant difference in depth
and area of an existing cluster of abnormal locations over the duration
of follow-up. A cluster was defined as a nasal step or as two or more
adjacent nonedge locations in the Pattern Deviation plot exhibiting
abnormality at
P ≤ 0.01 significance. The decibel
depth of each pattern deviation cluster was determined from the STATPAC
print-out.
Patients were excluded from the sample if one or more of the following
was present: failure to complete the examinations in the follow-up
period; change in topical therapy or surgical intervention during the
follow-up period; change in optic nerve head topography and/or
progressive visual field loss for either or both W-W perimetry and
SWAP; or a learning effect lasting beyond the baseline fields for
either, or both, W-W perimetry and SWAP. Forty-five individuals
were excluded from the original cohort
(Table 1) , and the sample comprised 33 glaucoma suspects and 17 POAG patients.
There was no statistically significant difference in the duration of
follow-up between the two groups of patients (
P =
0.732).
The mean age of the 33 suspects was 66.6 years (SD, 10.0 years; range,
42.4–77.3 years; 7 women, 26 men). The descriptive statistics for the
group mean Mean Deviation (MD), PSD, and CPSD for W-W perimetry and for
SWAP are shown in
Table 1 . Seventeen of the suspects were receiving
medical therapy (topical β-blocker), which remained unchanged
throughout the course of the study. The mean age of the 17 stable POAG
patients was 64.9 years (SD, 11.4 years; range, 42.8–84.9 years; 8
women, 9 men). The mean age of the glaucoma suspects and the POAG
patients were not significantly different (
P = 0.332).
Descriptive statistics for the group mean MD, PSD, and CPSD for W-W
perimetry and for SWAP, and the stage of the glaucomatous visual field
defect for W-W perimetry, as defined by the criteria of Hodapp et
al.,
17 are shown in
Table 1 . All POAG patients were
therapeutically controlled using topical β-blocker agents, and in one
patient this was combined with miotic therapy. None of the POAG
patients had a change in medical therapy over the course of the study.
The LF(Ho) and LF(He) components, expressed in decibels, were
calculated between the third and the seventh visual field examinations
(i.e., using data from the 3rd and 7th examinations only) for W-W and
for SWAP. The determination of the LF(Ho) and LF(He) components has
been previously described.
13 Briefly, LF(Ho) and LF(He)
were derived from a two-factor ANOVA with replications,
18 based on the double determinations of sensitivity at the 10 standard
stimulus locations incorporated in Program 30-2 and in Program 24-2 of
the HFA, using the formula:
\[Y_{jkl}{=}{\mu}{+}L_{j}{+}V_{k}{+}LV_{jk}{+}E_{jkl}\]
where
Y is the threshold determination at each
individual location, μ is the overall mean value of sensitivity,
which is related to MD by an aggregate constant,
L j is the effect of the examination
locations,
V k is the effect of each visual
field examination
, LV jk is the
interaction of
L j and
V
k, and E
jkl is the
experimental error. The integers
j, k, and
l represent, respectively, the number of considered stimulus locations
with a double determination of threshold (
j = 1, 2,…
n), the number of examinations (
k = 2), and the number of determinations of threshold at each considered
location (
l = 2).
The respective values of LF(Ho) and LF(He) were determined by
partitioning the variance derived from the ANOVA that is attributable
to each component of LF. The mean square estimates (MSE) for
V k , attributable to LF(Ho), and
LV jk , attributable to LF(He), were
then reduced to their constituent variances to remove the accompanying
error variance
E jkl by the
hypothesis
19 :
\[\mathrm{MSE}(V_{k}){=}jl{\sigma}_{k}^{2}{+}l{\sigma}_{jk}^{2}{+}{\sigma}^{2}\]
\[\mathrm{MSE}(LV_{jk}){=}l{\sigma}_{jk}^{2}{+}{\sigma}^{2}\]
\[\mathrm{MSE}(E_{jkl}){=}{\sigma}^{2}\]
In instances where a negative component of LF
(σ
k or σ
jk ) was
obtained, the variance(s) that exhibited a negative component of LF was
combined with that of the error term,
E. The error term was
then recalculated, the new value more accurately reflecting the
magnitude of both the error term and the previously negative LF
component.
13 This value was then used in all subsequent
analyses.
The decibel unit of differential light sensitivity is a relative
measure referenced to the maximum stimulus luminance. Any decibel value
obtained using W-W perimetry is not directly comparable with that using
SWAP because of the differing maximum stimulus luminance used by each
procedure. Therefore, to enable a more direct comparison to be made
between the components of LF derived by each perimetric technique, the
log unit equivalents of the LF(Ho) and LF(He) were calculated. This was
achieved by converting the decibel value to log units (i.e., relative
to the maximum stimulus luminance), and this value was then expressed
as a reciprocal. Thus, the reciprocal log unit would approach zero (and
the apostilb value approach the maximum stimulus luminance) as the
decibel value of the long-term fluctuation approached zero.