Factors Associated With Health-Related Quality of Life in Medically and Surgically Treated Patients With Glaucoma

Key Points

Question  What clinical factors are associated with reduced health-related quality of life in medically and surgically treated patients with glaucoma?

Findings  In this cohort study, reduced health-related quality of life was associated with worse diplopia, lower mean deviation on visual field testing, poor vision, and younger age. Previous glaucoma drainage device surgery was associated with poorer health-related quality of life on Adult Strabismus–20 domains compared with trabeculectomy or medical treatment.

Meaning  These findings suggest patients with glaucoma should be assessed for diplopia before and after surgery, given the negative effect of diplopia on health-related quality of life, and that potential psychosocial effects of glaucoma drainage devices should be discussed during preoperative counseling.

Importance  Health-related quality of life (HRQOL) is often reduced with glaucoma, but associated factors are poorly understood.

Objective  To determine factors associated with reduced HRQOL in medically and surgically treated patients with glaucoma.

Design, Setting, and Participants  Prospective cohort study at a tertiary referral glaucoma practice, with 160 consecutive, prospectively enrolled medically or surgically treated adult patients with glaucoma.

Main Outcome and Measures  All patients completed 2 HRQOL questionnaires (the 25-item National Eye Institute Visual Function Questionnaire and the Adult Strabismus–20 questionnaire [AS-20]). Thirty-six patients had undergone glaucoma drainage device surgery, 51 underwent trabeculectomy, and 73 were medically treated. Factors considered for association with HRQOL in multiple regression analyses were age, sex, best-eye and worst-eye mean deviation on Humphrey visual fields, treatment modality, best-eye and worst-eye visual acuity, and diplopia.

Results  The mean (SD) age of participants was 69 (13) years, 63% were female, 97% were white, 93% were not Hispanic, and the mean deviation (SD) was −13 (10) dB. Reduced HRQOL was associated with worse diplopia (Diplopia Questionnaire score) on 6 subscales (range of partial r² [r_p²], 0.207-0.069). Reduced HRQOL was associated with lower best-eye mean deviation on 5 of 6 subscales (r_p² range, 0.379-0.027), lower worst-eye mean deviation on 4 of 6 (r_p² range, 0.242-0.046), treatment group on 3 of 6 (r_p² range, 0.190-0.025), lower worst-eye visual acuity on 5 of 6 (r_p² range, 0.063-0.025), lower best-eye visual acuity on 2 of 6 (r_p² range, 0.032-0.017), and younger age on 2 of 6 (r_p² range, 0.021-0.014). In adjusted analyses, glaucoma drainage device was associated with worse HRQOL in 3 AS-20 subscales compared with trabeculectomy and 2 AS-20 subscales compared with medical. Differences ranged from −14.7 to −7.4, with half the absolute magnitude of the full range of the 95% CI ranging from 9.2 to 5.7.

Conclusions and Relevance  Our findings support the assertion that reduced HRQOL is common in surgically and medically treated patients with glaucoma. Overall, poor HRQOL in patients with glaucoma is moderately associated with worse diplopia, lower mean deviation on visual field testing in either eye, poorer visual acuity in either eye, treatment type, and younger age. Previous glaucoma drainage device surgery was specifically associated with poorer HRQOL compared with trabeculectomy or medical treatment. Psychosocial effects of glaucoma drainage device should be considered when counseling patients with glaucoma.

Reduced health-related quality of life (HRQOL) has been reported in patients with glaucoma and has been associated with paracentral and inferior visual field loss, worse mean deviation (MD), retinal nerve fiber layer thinning on optical coherence tomography, and lower socioeconomic status.^1-15 We reported in 2017 that diplopia and strabismus are common in patients with glaucoma and more frequent when treated with glaucoma drainage devices (GDD) than when treated with trabeculectomy or when medically treated.¹⁶ Associations with diplopia and strabismus may be important for HRQOL in patients with glaucoma because diplopia and strabismus can profoundly influence HRQOL.^17-19 Therefore, in this study of prospectively enrolled patients with glaucoma, we evaluated factors associated with reduced HRQOL using instruments that are sensitive to diplopia and strabismus (the Adult Strabismus–20 Questionnaire [AS-20]^20,21) in addition to the commonly used 25-item National Eye Institute Visual Function Questionnaire (VFQ-25).²² We assessed the 2 established unidimensional subscales of the VFQ-25 (visual functioning and socioemotional²³) and the 4 unidimensional subscales of the AS-20 (self-perception, interactions, reading function, and general function²¹), with specific attention to previous glaucoma procedure (GDD vs trabeculectomy vs medical treatment).

Approval was obtained from the institutional review board of the Mayo Clinic, Rochester, Minnesota, and all participants gave informed written consent prior to participating. All procedures and data collection were conducted in a manner compliant with the Health Insurance Portability and Accountability Act.

We prospectively enrolled 195 consecutive patients with glaucoma as part of a previous study¹⁶ at their regularly scheduled follow-up glaucoma visits between August 2014 and April 2015. As previously described,¹⁶ surgical patients were required to have at least 1 month (range, 1-229 months) of postoperative follow-up. Patients with multiple tubes, previous cataract surgery, failed glaucoma surgery, scleral buckle, or penetrating keratoplasty were not excluded. Medically treated patients may have previously undergone selective laser trabeculoplasty, trabectome, iStent, or cataract surgery. Patients were required to understand English.

Patients prospectively completed 3 questionnaires: the VFQ-25,²² the AS-20,^20,21 and the Diplopia Questionnaire (DQ).²⁴ The VFQ-25 is a 25-item general vision-related quality of life²² instrument and contains 2 robust subscales (visual functioning and socioemotional) by Rasch analysis.²³ The AS-20 is a 20-item instrument²⁰ with 4 subscales (self-perception, interactions, reading function, and general function) designed to detect both HRQOL and general functional vision concerns.²¹ The DQ²⁴ is a brief questionnaire designed to quantify the frequency of diplopia in standardized gaze positions over the previous week (scored from 0 to 100).²⁴ All questionnaires were self-administered, without supervision, following simple verbal and written instructions. Patients were instructed to respond as if wearing habitual refractive correction.

For this study, we analyzed a subgroup of 160 patients (mean [SD] age, 69 [13] years; 100 were female (63%); 155 were white (97%); 149 were not Hispanic (93%); MD [SD], −13  [10] dB) from the original study¹⁶ who completed VFQ-25, AS-20, and DQ questionnaires, and either automated visual fields or Goldmann visual fields within 3 years of the study visit as a measure of glaucoma severity (closest to study visit selected when more than 1 measure was available). Patients with visual field loss primarily owing to other ocular comorbidities (eg, aniridia and trauma) were excluded. Of the included 160 patients, 87 were treated surgically and 73 were treated medically. Of those treated surgically, 36 had undergone GDD placement and 51 had undergone trabeculectomy. Patients who underwent both trabeculectomy and GDD were analyzed in the GDD group. In the GDD group, 24 (67%) had a 350 Baerveldt, 3 (8%) had a 250 Baerveldt, and 11 (31%) had a FP7 Ahmed GDD. Six patients (17%) in the GDD group had multiple tubes. Two patients (6%) in the GDD group had previous penetrating keratoplasty, 2 (6%) had previous Descemet stripping endothelial keratoplasty, and 1 (3%) had a previous scleral buckle. In the trabeculectomy group, only 1 (2%) had previous penetrating keratoplasty and 1 (2%) had previous Descemet stripping endothelial keratoplasty. No patients in the trabeculectomy group had a previous scleral buckle procedure.

Severity of glaucoma was represented by visual field loss, recorded as an MD from the latest automated Humphrey visual field. In a preliminary analysis, we found a high correlation between MD and visual field index, and therefore, only MD was analyzed as a measure of glaucoma severity. If automated visual field data were not available (14 eyes of 9 patients), a matching method was used to impute an MD from the Goldmann visual fields. We created a reference set of paired Goldmann and Humphrey fields by using visual field data from 24 eyes of 14 patients in our cohort who had both Goldmann and automated visual fields, with MD ranging from −1.95 dB to −31.15 dB. We matched Goldmann visual fields in the HRQOL study to the closest Goldmann in the reference set and then used the corresponding MD from the matching Humphrey field for analysis. For 1 patient whose Goldmann field was worse than the worse field in the reference set (with no matching Goldmann), we imputed an MD of −32 dB.

The VFQ-25 responses were scored using published Rasch lookup tables,²³ yielding a Rasch-calibrated logit score for the visual functioning and socioemotional subscales. The VFQ-25 overall composite scores were also calculated to facilitate comparison with other studies reporting composite scores (despite being limited by multidimensionality). The AS-20 responses were also scored using published Rasch lookup tables,^21,25 yielding Rasch-calibrated scores from 0 to 100 for the self-perception, interactions, reading function, and general function subscales, where lower scores indicate worse HRQOL.

Factors associated with lower HRQOL score were assessed using univariate linear regression analysis for each of the 2 VFQ-25 subscales, VFQ composite score, and 4 AS-20 subscales. Factors considered were age, sex, diplopia score (0 to 100), best-eye MD, worst-eye MD, treatment group (analyzed as 3 separate contrast statements: GDD vs not GDD, trabeculectomy vs not trabeculectomy, and medical vs not medical), best-eye visual acuity (logMAR), and worst-eye visual acuity (logMAR). Factors found to be associated with HRQOL score in univariate analysis at a 2-sided P value of .10 or less were then entered into a stepwise multiple linear regression model for each of the 6 subscales and VFQ composite score. Spearman rank correlations were used to identify strong correlations (defined as r ≥ 0.6) between factors entering the multivariate models. Any highly correlated factors were evaluated in separate multiple variable models where the highest associated of paired correlated factors was removed for the subsequent models to account for potential masking of an effect by another correlated factor (collinearity).

Nonadjusted analyses were performed using Kruskal-Wallis comparisons with Dwass, Steel, and Critchlow-Flinger methods for multiple comparisons and Hodges-Lehmann estimating methods for 95% confidence intervals on nonparametric data. A separate adjusted analysis was performed using generalized linear models for each of the 6 subscales (2 VFQ-25 and 4 AS-20 subscales) and VFQ composite score to determine whether treatment group (medical, trabeculectomy, and GDD) was specifically associated with reduced HRQOL. We adjusted for all of these factors regardless of whether they were statistically significant in the univariate analyses. Tukey-Kramer methods were used to account for multiple comparisons between treatment groups in this adjusted analysis.

To evaluate the relative effects on HRQOL between AS-20 and VFQ-25 subscales, a standardized measure of effect was calculated as a proportion of the pooled standard deviation, analogous to an effect size. All statistical analyses were performed using SAS software, version 9.4 (SAS Institute Inc).

Patient demographics (age, sex, and race/ethnicity) were similar across the 3 groups with the exception of the median time from surgery to evaluation of diplopia and quality of life, which was longer in the trabeculectomy group (24 months; range, 1 month to 16 years) compared with the GDD group (5 months; range, 1 month to 6 years; P = .02; Table 1).

Results of univariate linear regression analyses are shown in Table 2. For all AS-20 subscales, VFQ-25 subscales, and VFQ-25 composite score, the following 6 factors were found to be significant and entered into the multiple linear regression model: diplopia score (0 to 100), best-eye MD, worst-eye MD, treatment group (analyzed as 3 separate contrast statements: GDD vs not GDD, trabeculectomy vs not trabeculectomy, and medical vs not medical), best-eye visual acuity (logMAR), and worst-eye visual acuity (logMAR). In addition, age at examination was entered as a factor for the self-perception, interactions, and socioemotional subscales.

Regarding correlation between included variables, the only factors found to have high correlations and also included in the multiple variable models were MD in the better and worse eye (r_s = 0.6244); therefore, separate multiple regression models were created for these 2 factors (Table 3), where the highest associated of the correlated factors (best-eye MD) was removed for the subsequent models to evaluate association with worst-eye MD.

Reduced HRQOL was associated with worse diplopia (DQ score) on all 6 subscales (VFQ-25 visual functioning and socioemotional, AS-20 self-perception, interactions, reading function, and general function; partial r² [r_p²], 0.207-0.069). Reduced HRQOL was associated with lower best-eye MD on 5 of 6 subscales (r_p², 0.379-0.027), lower worst-eye MD in 4 of 6 subscales (in the alternative models, r_p², 0.242-0.046), treatment group (GDD vs trabeculectomy vs medical) on 3 of 6 subscales (r_p², 0.190-0.025), lower worst-eye visual acuity on 5 of 6 subscales (r_p², 0.063-0.025), lower best-eye visual acuity on 2 of 6 subscales (r_p², 0.032-0.017), and younger age on 2 of 6 subscales of the AS-20 (r_p², 0.021-0.014, Table 3). The VFQ-25 composite scores were associated with lower best-eye mean deviation, worse diplopia, and lower worst-eye visual acuity (r_p², 0.255-0.076, Table 3). Partial r² values indicated the strongest associations with reduced HRQOL were treatment group (with GDD having the greatest effect), severity of diplopia represented by DQ score, and severity of glaucoma based on MD in either eye (Table 3).

Regarding treatment type, unadjusted treatment group comparisons revealed that GDD was associated with worse HRQOL compared with medical treatment on all AS-20 subscales (difference on self-perception, −12.5 points on a 0-100 scale; 95% CI, −20.8 to −4.2; interactions, −5.2; 95% CI, −10.3 to 0.0; reading function, −22.3; 95% CI, −34.9 to −9.8; and general function, −19.9; 95% CI, −31.8 to −7.9) and VFQ-25 subscales (socioemotional, −1.03 logits; 95% CI, −1.81 to −0.26; visual functioning, −1.56; 95% CI, −2.27 to −0.85) as well as the VFQ-25 composite score (−13.4 points on a 0-100 scale; 95% CI, −20.4 to −6.4; Figure). When comparing GDD with trabeculectomy, lower HRQOL was found in patients treated with GDD on the self-perception (−9.5; 95% CI, −19.0 to 0.0), interactions (−5.2; 95% CI, −10.3 to 0.0), and general function (−12.3; 95% CI, −24.6 to 0.0) subscales of the AS-20 (Figure, A). Lower scores on both VFQ-25 subscales (socioemotional, −0.62; 95% CI, −1.23 to 0.00; visual functioning, −0.80, 95% CI, −1.55 to −0.06) and VFQ composite scores (−7.3; 95% CI, −14.8 to 0.2) were also found when comparing GDD with trabeculectomy, although the differences did not reach significance (Figure, B).

In adjusted analyses, GDD was associated with worse HRQOL on the self-perception (difference, −12.3; 95% CI, −20.9 to −3.7), interactions (−9.3; 95% CI, −15.1 to −3.6), and general function (−9.4; 95% CI, −18.9 to 0.0) subscales of the AS-20 compared with trabeculectomy (Table 4). Comparing GDD with medical treatment, HRQOL was lower with GDD for the self-perception (−14.7; 95% CI, −23.9 to −5.5) and interactions (−7.4; 95% CI, −13.5 to −1.3) subscales of the AS-20 (Table 4). Adjusted analysis revealed lower scores on both VFQ-25 subscales and VFQ composite scores when comparing GDD with either trabeculectomy or medical treatment, although the differences did not reach significance (Table 4).

The differences between GDD, trabeculectomy, and medical are presented as effect sizes (eTable in the Supplement) relative to the pooled standard deviations. The AS-20 interactions subscale showed the greatest difference for GDD vs trabeculectomy, whereas the VFQ-25 socioemotional subscale showed the least difference of the subscales for GDD vs trabeculectomy.

Overall, in our study, more severe diplopia, more severe glaucoma in either eye (worse MD on visual field testing), poor visual acuity, treatment type, and younger age were associated with reduced quality of life. Regarding treatment type, patients with GDD had lower unadjusted HRQOL scores on all AS-20 and VFQ-25 subscales compared with medically treated patients. When comparing GDD with trabeculectomy, statistically lower unadjusted HRQOL scores were found on 3 of 4 AS-20 subscales.

Glaucoma drainage device was associated with reduced quality of life in this study even when accounting for visual acuity, visual field loss, diplopia, age, and sex in adjusted analyses. Patients appear to be self-conscious of the device because the specific AS-20 subscales of self-perception and interactions showed reduced HRQOL in patients with GDD compared with trabeculectomy or medically treated patients. It is possible that patients with a GDD feel the physical presence of the device or may have a psychological awareness of the device.

There are few studies with which to compare our data. A 2017 randomized prospective trial of GDD vs trabeculectomy (Tube vs Trabeculectomy [TVT] study²⁶) did not find a difference in HRQOL between patients who were treated with a trabeculectomy vs a 350 Baerveldt following previous trabeculectomy and/or cataract extraction. The strength of the TVT study is that it was a randomized clinical trial, reducing the influence of known and unknown confounders. Nevertheless, the TVT HRQOL study reported composite VFQ-25 scores and 10 subscale scores, which have been found to be multidimensional and not psychometrically robust. Pesudovs et al²³ reported that principal component analysis of the VFQ-25 revealed 2 major subscales, visual functioning and socioemotional, which we therefore reported in our study. Unlike the TVT HRQOL study, our analyses included data on visual acuity and visual field MD in each eye, not just the eye that had undergone surgery. The status of either or both eyes may affect HRQOL. An additional difference between our study and the TVT study is that our study prospectively included patients who had undergone multiple tubes, scleral buckling procedure, and penetrating keratoplasty. Similar to the TVT study, despite considering additional factors and despite using Rasch-scored VFQ-25 subscales, we also did not find statistically lower HRQOL scores in patients treated with GDD vs trabeculectomy, when assessing HRQOL using the VFQ-25. Nevertheless, we found lower HRQOL using specific AS-20 subscales (self-perception, interactions, and general function) in patients who had undergone GDD compared with trabeculectomy, suggesting that the AS-20 subscales were more sensitive than the VFQ-25 for detecting reduced HRQOL in patients with glaucoma.

In contrast to the TVT study, our study also compared the entire range of glaucoma treatment from medically controlled to trabeculectomy to patients with GDD, which allowed a more complete evaluation of HRQOL across the spectrum of disease. We intentionally included patients with severe glaucoma, using imputation methods to assign a value for MD in each patient, unlike some previous studies that excluded visual field data, or had minimal visual field data, on patients with severe disease.

In 2017, we reported a high prevalence of diplopia in patients with glaucoma,¹⁶ including those who have been treated with GDD, and therefore we also used an HRQOL instrument that is particularly sensitive to strabismus and diplopia. Considering relative effect sizes, the AS-20 self-perception, interactions, and general function subscales were particularly well able to detect differences between patients who had undergone GDD and patients who had undergone trabeculectomy. One interpretation of this result is that the effect of GDD on HRQOL may have a component related to diplopia and strabismus. Our use of instruments that are sensitive to strabismus and diplopia issues may have enhanced our ability to detect differences between GDD and trabeculectomy in our complex glaucoma patient population.

Our findings of reduced HRQOL in AS-20 self-perception and interactions subscales highlight negative self-perception in patients who undergo GDD implant. Ophthalmologists should be conscious of the potential contributors to poor HRQOL and openly discuss them with patients when gauging surgical options. We currently prefer to use translucent patching materials, such as cornea, which may mitigate a patient’s poor self-perception after GDD placement, but many patients in this study had opaque patching materials, which may partly explain why their self-perception, interactions, and socioemotional scores were so low.

Our study was not without limitations. We had insufficient ethnic and racial diversity to consider those factors. We did not evaluate personality type, depressive symptoms, economic status, topical eyedrop therapy, coexistent cataract, or coexistent age-related macular degeneration, which others have reported to influence HRQOL scores.^27-31 We did not collect longitudinal data on change in HRQOL, but we focused on a single point, including points as short as 1 month since surgery. We do not know whether the patients who underwent GDD surgery had poor HRQOL prior to that specific surgery and we acknowledge that, in our particular practice, GDDs tend to be used later in the surgical treatment algorithm. It is possible that longer duration of disease and/or a greater number of failed treatments influenced HRQOL. We did not use glaucoma-specific HRQOL instruments used by some research groups (eg, the Glaucoma Quality of Life–15^2,3,32 and the Glaucoma Activity Limitation–9),³³ but we used the VFQ-25, which has been used by many other groups in glaucoma research. We did not perform a de novo Rasch analysis, but we used the established method of Rasch-based scoring with published lookup tables, and it is possible that the HRQOL factor structure in our patients may have differed from the reference populations.^21,23 We used the AS-20 instrument^20,21 because our study was designed to evaluate the specific effect of diplopia and strabismus on HRQOL.

Reduced HRQOL is common in patients with glaucoma previously treated with surgery and nonsurgical treatments. Overall, poor HRQOL in patients with glaucoma is associated with diplopia, lower MD on visual fields testing in either eye, poor visual acuity in either eye, treatment type, and younger age. Previous placement of a GDD was associated with poorer HRQOL scores on specific subscales of the AS-20 compared with trabeculectomy and medical treatment. The psychosocial effects of GDD, reflected in AS-20 self-perception, interactions, and VFQ-25 socioemotional subscales, should be considered when counseling patients with glaucoma. These findings suggest patients with glaucoma should be assessed for diplopia before and after surgery, given the negative effect of diplopia on HRQOL, and potential psychosocial effects of glaucoma drainage devices should be discussed during preoperative counseling.

Corresponding Author: Cheryl L. Khanna, MD, Ophthalmology E4, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (khanna.cheryl@mayo.edu).

Accepted for Publication: January 1, 2018.

Published Online: February 22, 2018. doi:10.1001/jamaophthalmol.2018.0012

Author Contributions: Dr Holmes and Mr Leske had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: All authors.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Khanna, Holmes.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: All authors.

Obtained funding: Holmes.

Administrative, technical, or material support: All authors.

Supervision: Holmes.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: Supported by National Institutes of Health grant EY024333 (Dr Holmes), Research to Prevent Blindness, New York, New York (an unrestricted grant to the Department of Ophthalmology, Mayo Clinic), and Mayo Foundation, Rochester, Minnesota.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentations: This paper was presented at the Association for Research in Vision and Ophthalmology annual meeting; May 5, 2015; Denver, Colorado; and at the American Glaucoma Society, February 28, 2015; Coronado, California.