Study reference grouped by condition
(author, year) Utility measures Methods Results
Glaucoma
Aspinall et al., 200844 EQ-5D Known groups
(severity) Convergence
EQ-index stratified by mild, moderate and severe visual field loss. EQ-index, mobility, self-care and anxiety statistically significantly correlated with VA. Mobility and self-care correlated with severity of visual field loss
Kobelt et al., 200645 EQ-5D Known groups
(severity)
EQ-5D utility decreased with increased severity, but difference between groups only statistically significant for severe disease after controlling for co-morbidity
Mittmann et al., 200134 HUI3 Known groups
(case–control)
Mean HUI3 values (SD): glaucoma patients 0.924 (0.086); no condition patients 0.953 (0.068)
Montemayor et al.,
200146
EQ-5D Convergence EQ-5D correlated with age (health status only) and VFA score. Not correlated with diagnosis, VA, mean deviation in the better or worse eye, corrected pattern standard deviation in the better or worse eye. VFA was the best predictor of EQ-5D
Thygesen et al., 200854 EQ-5D Convergence
Known groups (severity)
Better VA is correlated with higher EQ-5D (p = 0.005). EQ-5D was
consistent with the severity groups defined by Snellen scores
AMD
Cruess et al., 200747 EQ-5D Known groups
(case–control) Convergence
EQ-5D not significantly lower in subjects compared with control (14% relative difference, p = 0.064). Moves in the right direction.
No association between EQ-5D and VA stratification found Espallargues et al., 200522 EQ-5D, SF-6D and HUI3 Convergence Known groups (severity)
All preference-based measures were correlated and significant to 1% level with VF-14. EQ-5D was not correlated to a significant level with CS or VA. VAS was correlated with 5% significant level with CS and VA. SF-6D was correlated with CS (1% level) and VA (5% level). HUI3 and TTO were correlated with both CS and VA to 1% level. VA and CS were consistent with HUI3, SF-6D, TTO and VAS but not with the EQ-5D
Lotery et al., 200748 EQ-5D Known groups
(severity) Convergence
EQ-5D and VFQ-25 differentiated between groups (statistically significant). No apparent relationship was found between EQ-5D and severity of vision loss. This was found for the NEI-VFQ-25 (no p-value reported)
Payakachat et al.,
200949
EQ-5D Known groups (severity)
Subjects reported full health in EQ-5D but had visual problems, as elicited by the NEI-VFQ-25
Ruiz-Moreno et al.,
200856
EQ-5D Known groups (case–control)
Adjusted mean scores 0.68 vs. 0.79 p < 0.05 for neovascular- AMD vs. control
Soubrane et al., 200743 EQ-5D Known groups
(case–control and severity)
Adjusted mean scores of EQ-5D 0.65 vs. 0.75 p < 0.001 for
neovascular-AMD vs. control.
No significant difference across VA levels of neovascular-AMD (and does not follow degree of severity)
Kim et al., 201055 EQ-5D Known groups
(severity)
Significant differences were found in EQ-5D scores for people with unilateral and bilateral AMD
Study reference grouped by condition
(author, year) Utility measures Methods Results
Cataracts
Asakawa et al., 200836 HUI3 Known groups
(case–control, gender)
Adjusted mean differences in single-attribute vision utility scores for cataracts were negative, quantitatively important (difference > 0.05) and statistically significant
Datta et al., 200853 EQ-5D Convergence No visual variables were significantly associated with EQ-5D.
VF-14 was strongly associated with acuity, stereopsis and contrast sensitivity. Acuity was less important than either stereopsis or contrast sensitivity for EQ-5D, which may suggest that acuity is required for function tasks, but stereopsis and contrast sensitivity were more important determinant of generic QoL
Polack et al., 200737 EQ-5D Known groups
(case–control) Convergence
Cases were significantly more likely to report problems with mobility, self-care, usual activities and anxiety than controls No significant association between VA and EQ-5D across all dimensions, except for self-care which has a borderline (p = 0.05) association
Polack et al., 200838 EQ-5D Known groups
(case–control) Convergence
Significant difference (p < 0.001) across all EQ-5D dimensions
between cases and controls. Poorer VA was associated with higher odds or reporting any problem with mobility, self-care, usual activities and pain. There was no significant association for depression
Polack et al., 201039 EQ-5D Known groups
(case–control) Convergence
Significant difference between cases and controls using VF20 and self-rated health scale. Cases were significantly more likely to report problems with all five EQ-5D domains compared with controls after adjustment for age, gender and socioeconomic status. Inconsistent association between EQ-5D and VA level. Borderline trend with VA shown with self-care (p = 0.05), driven
by the higher prevalence of reported problems among cases with perception of light compared with those with moderate visual impairment. The lack of association with the remaining domains may reflect the fact that relatively few cases (< 25%) reported no problem, resulting in a lack of variation in the data
Diabetic retinopathy
Lloyd et al., 200842 EQ-5D
and HUI3
Known groups (severity) Convergence
EQ-5D index, EQ-VAS and HUI3 all show some inconsistency when compared with degree of severity. Pattern on VFQ-25 consistent. Between each level of VA, not every difference in utility was significant or consistent. Results show a significant trend with EQ-5D and HUI3 worsening as VA worsens. A regression was undertaken and VFQ-25 and LogMAR were identified as independent significant predictors of utility. The data from the EQ-5D, HUI and VFQ-25 suggest that relatively mild vision loss (6/12 to 6/18) can be associated with very substantial declines in utility, with lower scores than people with worse vision Smith et al., 200850 Convergence
(through regressions) Known groups (severity)
No clear pattern from mean values. OLS model used to estimate the impact on utility of a doubling of the visual angle. Utility values dropped by approximately seven points for each doubling (assuming linear relationship between acuity and utility). Doubling visual angle results in utility loss of about 0.03. A non-parametric ordinal logistic model was fitted and this estimated that anyone who suffered any degree of visual impairment were more likely to report non-perfect utility values (OR 1.44, 95% confidence interval 1.08 to 1.91)
Study reference grouped by condition
(author, year) Utility measures Methods Results
Conjunctivitis
Pitt et al., 200460 EQ-5D Known groups
(case–control)
Inconsistent results comparing SAC to controls. Only the pain domain and the EQ-5D were significantly worse in the SAC group compared with the control. In some cases, the remaining domains were worse in the control (but non-significant). RQLQ was statistically significant across all domains. VFQ-25 was statistically significant across the mean vision score and the general health score
Rajagopalan et al.,
200551
EQ-5D Known groups (severity)
EQ-5D showed significant differences in scale scores across the varying severity levels (EQ-5D, p < 0.05, and VAS, p < 0.0001).
Significant differences were seen across all IDEEL scales except treatment satisfaction. EQ-5D and IDEEL were consistent in their ranking of severity. Strength of difference analysis was provided and the IDEEL outperformed EQ-5D and SF-36 across all severity levels. Mean (SD) EQ-5D scores: control 0.87 (0.03), non-SS KCS 0.82 (0.02) and SS 0.74 (0.03). Mean (SD) EQ-5D VAS score: 88.93 (2.06), non-SS KCS 82.45 (1.19) and SS 66.94 (2.43) Smith et al., 200561 EQ-5D Known groups
(case–control)
EQ-VAS and all EQ-5D dimensions, except mobility, are statistically significant (p < 0.02) between SAC and control groups.
Interestingly, VFQ-25 showed significantly lower scores in all domains in the SAC group, except for the general health domain, which returned a lower (non-significant) value for the
control group
Other visual disorders
Boulton et al., 200640 HUI3 Known groups
(severity)
Statistically different (unknown to what level) mean HUI3 scores between groups
Clark et al., 200862 EQ-5D Known groups
(case–control)
Significant differences between cases and controls using NEI VFQ-25, but not with EQ-5D or TTO. Only the mobility domain had a significant difference. Patients had a significant difference using the VFQ-25; however, no difference was significant when stratified by visual impairment. Postoperation VA was statistically significantly different
Kempen et al., 200363 EQ-5D Known groups
(severity)
Does not distinguish between groups (non-significant) and direction of trend is counter-intuitive. VAS distinguished newly-diagnosed group. No statistically significant difference in EQ-5D and borderline between VAS
Langelaan et al.,
200741
EQ-5D Known groups (severity)
None were statistically significant at the 5% level. VA saw an appropriate movement in EQ-5D; however, VF moved in the wrong direction
Quinn et al., 200464 HUI3 Known groups
(severity)
HUI3 mean (SD) scores: All 0.59 (0.39). Blind or low vision in better eye 0.25 (0.37). Sighted in better eye 0.78 (0.25). No-ROP subjects 0.90 (0.16). Statistical significance of VA not given but appears to be statistically significant and appropriate. HUI3 showed a significantly lower score (p < 0.001) for the blind
group compared with the sighted group and the non-ROP group compared with the sighted group (p < 0.001)
van Nispen et al., 200952
EQ-5D Convergence (through regression)
LogMAR VA is a significant risk factor for lower QoL
IDEEL, impact of dry eyes on everyday life questionnaire; KCS, Keratoconjunctivitis sicca; LogMAR, logarithm of the minimum angle of resolution; NEI-VFQ-25, National Eye Institute Visual Functioning Questionnaire – 25; ROP, retinopathy of prematurity; RQLQ, rhinoconjunctivitis QoL questionnaire; SS, Sjögren’s syndrome.