The Quality of Life of Seniors with Eye Diseases during COVID-19

Purpose To assess health-related quality of life (HRQoL), vision-related quality of life (VRQoL), depression and anxiety symptoms, and social support and community integration of seniors with eye diseases and to identify important predictor variables of the outcomes. Methods A cross-sectional survey was performed in seniors with eye diseases (n = 90). Demographic characteristics and questionnaire scores were summarized. Linear regression analysis with backward stepwise selection was used to predict the value of the outcomes of eye disease. Results Preference-based HRQoL of the study patients with eye diseases during the pandemic was likely good with a mean utility value of 0.88. VRQoL and sleep quality appeared to be good as well. Depression and anxiety symptoms appeared to be low, while community integration and social support were moderate. Furthermore, the presence of retinal disease, number of nonocular comorbidities, and education appeared to have significant negative effects on social support and community integration. The presence of retinal disease and the number of nonocular comorbidities both appeared to negatively impact VRQoL. The use of a mobility aid appeared to negatively affect depressive symptoms and sleep quality. Conclusions Overall quality of life and wellness among seniors with eye diseases appeared to be good during the COVID-19 pandemic. However, the presence of retinal disease and the number of nonocular comorbidities both appeared to negatively impact VRQoL and social support and community integration. Education appeared to impact social support and community integration negatively. The use of a mobility aid appeared to negatively affect depressive symptoms and sleep quality.


Introduction
Coronavirus disease 2019 (COVID-19) is a highly infectious disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [1].Tis disease has been reported to be able to cause severe acute respiratory infection with an incubation period of 1 to 14 days [2].Many of the common symptoms of COVID-19 include fever, dry cough, and fatigue [3].Te SARS-CoV-2 virus has been reported to be transmitted via respiratory droplets; however, it can also be spread through various discharges, feces, aerosol, and conjunctiva [4].
Te burden brought on by the COVID-19 pandemic has afected many groups of individuals on a global level.One vulnerable group of individuals who have been greatly impacted by the pandemic includes patients with ophthalmological conditions.Patients in this group are particularly vulnerable in the context of the COVID-19 pandemic because of their age and preexisting comorbid conditions; specifcally, elderly patients aged 65 and above with eye diseases like glaucoma, age-related macular degeneration, or diabetic retinopathy require regular followups and commonly sufer from additional comorbidities [5].Tese additional conditions include anxiety and depression which have been shown to be exacerbated throughout the ongoing pandemic [6,7].Even if these individuals do not currently have such comorbidities, they are extremely susceptible to developing mental health issues [8].
Furthermore, the negative impact of the COVID-19 pandemic has been shown to be pronounced among elderly individuals through public media outlets portraying COVID-19 as a disease that is particularly devastating to the elderly.Tis spread of fearful sentiment, in turn, has resulted in the development of social stigma and discrimination causing additional distress to elderly individuals, their families, and their caregivers [8].Tis information compounds their worry about being infected with the virus and not having access to proper healthcare.In addition to the vulnerabilities imposed by their age group, because of the proximity between the patient and healthcare personnel during ophthalmological examinations, the risk of transmission can be perceived as being relatively high among seniors with ophthalmologic conditions.As such, the delicate balance between the risk of exposure to COVID-19 and visual loss in delaying cases is a psychological stressor to both patients and clinicians [9,10].
Te present pandemic has led ofcials to rethink the management of patient lists and to restrict the patients to be assessed or treated based on the urgency of their condition in accordance with ministerial guidelines [11,12].Tese restrictions may have potentially resulted in many delays in clinical visits which, in turn, may increase the risk of visual loss not only by delaying necessary care but also by making patients less likely to follow their physician's guidance for their conditions.Moreover, due to the inability to attend clinical visits, patients themselves might decide to become nonadherent and interrupt their treatment or postpone their visit for fear of contracting SARS-CoV-2.Furthermore, the COVID-19 pandemic conditions have made it even more difcult to access medications that are high in demand and have made it difcult to make changes to treatments when necessary [13].Additionally, visual loss can accompany depressive symptoms and deteriorate the quality of life (QoL) [14] in addition to the difculties already present due to their ocular diseases and potential comorbidities.Tus, it is necessary to characterize and document the quality of life and mental well-being of patients with various ophthalmologic conditions during the COVID-19 pandemic.
In documenting the QoL and wellness of patients during the COVID-19 pandemic, we can quantify the collateral impact of COVID-19 beyond the direct impact of the virus.Furthermore, this will help to improve the future quality of care during non-COVID-19 conditions and even during potential future pandemic situations.Terefore, the goal of this study is to characterize the preference-based healthrelated quality of life (HRQoL), vision-related QoL (VRQoL), depression and anxiety symptoms, sleep quality, and social support and community integration of seniors aged 65 and above with various eye diseases during the COVID-19 pandemic.Furthermore, this study also aims to identify important predictor variables for the aforementioned measures.

Study Design and Sampling
Procedures.Te study followed a cross-sectional design.A convenience sample of 90 patients who were identifed as having an underlying ocular disease was recruited from four ophthalmic practices at the Ivey Eye Institute, St. Joseph's Healthcare, London, Ontario.All patients were sequentially recruited from November 2021 to May 2022 using convenience sampling.Te eligibility of the patients who were attending their regular ophthalmology visits was determined by the ophthalmologist on staf.Inclusion criteria included patients who were aged 65 and above and diagnosed with an eye disease by an experienced ophthalmologist.Exclusion criteria included patients who were unable to provide valid informed consent, who had signifcant communication barriers or lack of English profciency that prevents participants from completing the questionnaires, or who had irreversible vision loss that prevented them from completing the questionnaires.
All participants received a complete explanation of the purpose and procedures involved in the study, and patient concerns were addressed prior to study participation.Verbal and electronically written informed consent was obtained from all participating patients.Te study was initiated after approval by the Western University Health Science Research Ethics Board and Lawson Health Research Institute's Clinical Research Impact Committee.

Data Collection.
All data were collected through electronic questionnaires accompanied by face-to-face interviews for assistance.All data were recorded electronically using the UWO Qualtrics questionnaires that were set up in advance.Data from Qualtrics were then imported to a password-protected and encrypted spreadsheet on the password-protected local computer in the principal investigator's (MM) ofce at St. Joseph's Healthcare.Data were coded to protect participant confdentiality.Te code key with identifying data (master data) was also stored in a password-protected and encrypted spreadsheet on the password-protected local computer in the principal investigator's ofce.For analysis purposes, the deidentifed study data not containing patient information were stored in a password-protected and encrypted spreadsheet on the St. Joseph's Hospital OneDrive.Data quality checks were performed at random.
Following clinical examination of the patients, the ophthalmologists at the Ivey Eye Institute identifed and referred participants to the research assistant on duty based on the inclusion and exclusion criteria.Te questionnaires were presented to the patients using a combination of selfadministered and interviewer-assisted modes.Participants provided informed consent and completed the 30-minute questionnaire.Te questionnaire included the Time Trade-Of (TTO) questionnaire, the 25-item version National Eye Institute Visual Function Questionnaire (NEI VFQ-25), Hospital Anxiety and Depression Scale-Anxiety Subscale (HADS-A), Center for Epidemiologic Studies-Depression Scale (CES-D), Pittsburgh Sleep Quality Index (PSQI), and Community Integration Questionnaire (CIQ).Demographic characteristics on patients' age, socioeconomic status (SES), ethnicity, education level, living arrangement, city of residence, and use of mobility aid were also collected, all of which were provided by patients themselves in the electronic questionnaires.
Te TTO was used to obtain utility scores to calculate patients' preference-based HRQoL.Preference-based HRQoL is a frequently used measure calculated with utility values on a scale from 0 to 1, where 0 represents a health state equal to death and 1 represents a state of perfect health [15].In the current study, the utility score using the TTO method was calculated by dividing the number of years a patient was willing to trade in return for perfect vision by the estimated number of years of life remaining and subtracting this number from 1.
VRQoL measures the impact of vision on an individual's daily living, as well as one's satisfaction and attitudes towards their vision.Te NEI VFQ-25 is divided into 12 subscales: general health, general vision, near vision, distance vision, driving, peripheral vision, color vision, ocular pain, role limitations, dependency, social function, and mental health.Te NEI VFQ-25 was also shown to have high validity and reliability [16].
Te HADS-A is a 7-item self-report subscale for measuring symptoms of anxiety.Each item on the questionnaire is scored from 0 to 3, with total scores ranging from 0 to 21.Higher scores represent higher levels of psychological distress [17].Te HADS-A has been used in a previous study on the impact of low vision on the QoL, depression, anxiety, and social support [18].A score ≥8 on the HADS-A has a sensitivity of 0.9 and specifcity of 0.79 for identifying patients with anxiety [19].Te HADS-A has been demonstrated to have adequate internal consistency with a Cronbach's alpha value of 0.87 when administered in older adults [20].
Te CES-D is a brief self-report scale designed to measure self-reported symptoms associated with depression experienced in the past week.Te CES-D includes twenty items comprising six scales refecting major facets of depression.Te possible range of CES-D scores is 0 to 60, with the higher scores indicating the presence of more depressive symptomatology.A CES-D score ≥16 has high sensitivity and specifcity rates for identifying subjects with depressive disorder.Te CES-D has been demonstrated to be reliable with coefcient alpha estimates of 0.90 in clinical older adults.It has also been demonstrated to have high construct validity when administered to older adults [21].
Te PSQI is a self-report questionnaire that assesses sleep quality over a 1-month time interval.Te measure consists of 19 individual items, creating 7 components that produce one global score [22].Higher PSQI scores indicate worse sleep quality.A PSQI score >5 has sensitivity and specifcity rates of 89.6% and 86.5%, respectively, for identifying cases with sleep disorder [23].Te PSQI has been shown to have a high test-retest reliability and a good validity [24].
Te CIQ is a 15-item inventory designed to measure levels of community integration.Te overall score ranges from 0 to 29 and can be further divided into three subscores, corresponding to integration in the home, social integration, and productivity.A higher CIQ score represents greater integration [25].Previous research has demonstrated adequate test-retest reliability and internal consistency [26].
All patients were interviewed under standardized conditions.Te interviews were conducted by four interviewers who all received standardized training prior to administering the questionnaires.All questionnaires were completed by the patient in an electronic format on an electronic tablet through the UWO Qualtrics link containing the questionnaires.While completing the questionnaires, a research assistant was present to assist by answering any questions or concerns about the questionnaires, if patients had any, as well as to administer the questionnaires.

Statistical Analysis.
All statistical analyses were performed using STATA 17.0.Te descriptive statistics were computed for all demographic variables while univariate analysis was computed for all questionnaire outcome measures.To understand the central tendency and distribution of continuous variables, means and standard deviations were calculated.
Associations between predictor variables were also assessed using Pearson correlations between continuous predictor variables.Te Pearson correlation coefcient threshold absolute value of 0.6 was used as a cutof indicating a strong association between the variables [27].If a threshold above 0.6 was identifed, then a signifcance test was conducted to confrm the association.Chi-square tests were used to assess the association between pairs of categorical predictor variables.Te signifcance of the relationship between the predictor variables was defned at p < 0.050.Ttests and one-way ANOVA were used to assess the associations between pairs of continuous and categorical predictor variables.Again, statistical signifcance was determined at p < 0.050.
Bivariate analysis was performed to assess the unadjusted efect estimates and check whether each predictor variable and outcome were associated.Each of the predictor variables (age, number of nonocular comorbidities, number of ocular comorbidities, use of a mobility aid, SES during COVID-19, living arrangements, education, presence of retinal disease, presence of glaucoma, and presence of cataracts) was individually investigated for association with all six questionnaire outcomes, using simple linear regression analyses.
Linear regression models were also created with the questionnaire scores as the dependent variables using backward stepwise multiple regression.In backward stepwise regression, all predictor variables are frst used in the model.Following this, tests are then performed to determine the least signifcant predictor variable that is to be removed.Predictor variables continue to be removed until all Journal of Ophthalmology remaining predictors are determined to be relevant predictors of the outcome in the model.Regression coefcients were deemed to be signifcant if the associated p values were <0.050.
To assess the backward stepwise linear regression models' abilities to accurately predict each outcome, leaveone-out cross-validation (LOOCV) of each model was performed.In LOOCV, a single observation is used for the testing set while n − 1 observations are used for the training set [28].Tis process is repeated until each observation has been a part of the testing set.LOOCV evaluates a model based on prediction and is used for estimating the test error.Te root mean square error (RMSE) and mean absolute error (MAE) were determined for each multivariable model.
Further, model assumptions for each multivariable model were tested.Tat is, the constant variance of the residuals was tested using residuals versus ftted value plots.Te normality of the residuals was assessed using quantilequantile (Q-Q) plots.Linearity between the predictors and outcomes was assessed using component-plus-residual plots.Te variance infation factor (VIF) was used to test for multicollinearity.A VIF of 10 was used as the rule of thumb to indicate an acceptable level of multicollinearity [29].

Participants and Participant Characteristics.
A total of 128 patients were approached by the attending clinicians and asked to participate in the study.Of the 128 patients, 115 agreed to participate; however, 25 of these patients did not pass the inclusion/exclusion criteria.As such, a total of 90 participants consented and were included in the study.To summarize the characteristics of these included participants, univariate analyses were conducted.Te participants' characteristics can be seen in Table 1.Te mean age of the participants was 77.8 years with a standard deviation of 8.0 years.In terms of the ethnicities of the participants, 86 identifed as being white, two participants identifed as being black, one participant identifed as being Arab, and one participant did not indicate his or her ethnicity.Of the included participants, 37 participants completed high school or less and 53 participants had additional training or higher education.It was also noted that 23 participants had an income of $25,000 or less.Moreover, 67 participants lived at home with their family, spouse, or caregiver, while 22 participants lived at home alone and only one participant lived in a nursing home.Finally, 15 participants reported using a mobility aid such as a cane, walker, wheelchair, or motorized scooter.Of the 90 total participants, 65 patients reported to have other nonocular comorbidities with the most common being hypertension, hyperlipidemia, and diabetes.Te mean number of nonocular comorbidities was 1.6 with a standard deviation of 1.7.Of the 77 participants who reported their eye disease(s), the mean number of ocular comorbidities was 1.2 with a standard deviation of

Associations between Continuous Predictor Variables.
Te results of the associations between continuous predictor variables are presented in Table 2. Pearson correlations did not reveal any strong linear associations between the continuous predictor variables: age, number of nonocular comorbidities, and number of ocular comorbidities.

Associations between Pairs of Categorical Predictor
Variables.Te chi-square test results of the associations between categorical predictor variables are presented in Table 3. Signifcant associations were observed between the presence of retinal disease and glaucoma.Among patients who had a retinal disease, 90% did not also have glaucoma.Signifcant associations were also observed between the presence of glaucoma and cataracts.Among patients with glaucoma, 90% did not also have cataracts.

Associations between Pairs of Continuous Predictor
Variables.Finally, the t-test and one-way ANOVA results for the associations between continuous and categorical predictor variables are presented in Table 4. Te use of a mobility aid was signifcantly associated with age, the number of nonocular comorbidities, and the number of ocular comorbidities.On average, patients who used a mobility aid were older, had more nonocular comorbidities, and had more ocular comorbidities.Te presence of retinal disease was signifcantly associated with age.Patients who had retinal disease appeared to be older, on average.Te presence of glaucoma was also signifcantly associated with age; however, it appeared that on average patients without glaucoma were older.Finally, the presence of cataracts was signifcantly associated with the number of nonocular comorbidities.On average, patients with cataracts appeared to have a greater number of nonocular comorbidities.

Preference-Based HRQoL.
Te TTO utility score measures the preference-based quality of life on a scale from 0 to 1, in which a score of 0 represents a state of death and 1 represents perfect visual health.In our study population (n � 90), the average TTO utility score was 0.88 with a standard deviation of 0.23 (Table 5).Tus, the majority of the patients were willing to trade 12% of their remaining life for perfect vision.Te bivariate analyses for preference-based HRQoL with the demographic and clinical variables are presented in Table 6.No variables were signifcantly associated with preference-based HRQoL.
On average, patients who have completed more than high school in their education have an average diference in their TTO score of 0.05 (95% CI: −0.05, 0.15) as compared to patients who have completed high school or less.Te use of a mobility aid, on average, increases the TTO score by 0.02 (95% CI: −0.11, 0.16).For each increase in the number of ocular comorbidities, the TTO score increases by 0.07 (95% CI: −0.06, 0.19).On average, the presence of glaucoma increases the TTO score by 0.04 (95% CI: −0.08, 0.15).Te presence of retinal disease changes the TTO score by −0.06 (95% CI: −0.17, 0.05).Te presence of cataracts changes the TTO score by −0.09 (95% CI: −0.22, 0.05).
Te backward stepwise regression did not produce a model with any variable predictive of the TTO utility score.Tis is similar to the results of the bivariate analysis in that no variable was found to be signifcantly associated with the TTO score.
3.6.VRQoL.Te NEI VFQ-25 score measures vision-related quality of life on a total scale from 0 to 100, in which a score of 0 represents the worst possible score and 100 represents the best.In our study population (n � 90), the average NEI VFQ-25 score was 84.71 with a standard deviation of 11.61 (Table 5).
Te bivariate analyses for the VRQoL with the demographic and clinical variables are presented in Table 7. Te following variables were signifcantly associated with VRQoL: education (p � 0.027), number of ocular comorbidities (p � 0.042), and the presence of retinal disease.Patients who have completed more than high school in their education have an average diference in their NEI VFQ-25 score of 5.46 (95% CI: 0.62, 10.29) as compared to patients who have completed high school or less.For each increase in the number of ocular comorbidities, the NEI VFQ-25 score changes by −6.31 (95% CI: −12.23, −0.24).Te presence of retinal disease changes the NEI VFQ-25 score by −7.56 (95% CI: −12.58, −2.55).

Presence of Depressive Symptoms.
Te CES-D score measures self-reported symptoms associated with depression experienced in the past week on a scale from 0 to 60, in which higher scores indicate the presence of more depressive symptomatology.In our study population (n � 90), the average CES-D score was 6.79 with a standard deviation of 6.39 (Table 5).
Te bivariate analyses for the presence of depressive symptoms with the demographic and clinical variables are presented in Table 9. Te following variable was signifcantly associated with the presence of depressive symptoms.Use of a mobility aid, on average, increases the CES-D score by 4.35 (95% CI: 0.85, 7.86).
On average, patients who have completed more than high school in their education have an average diference in their CES-D score of 1.98 (95% CI: −0.72, 4. Patients with a SES of $10,001-$25,000, $25,001-$50,000, $50,001-$75,000, $75,001-$100,000, $100,001-$125,000, and greater than $150,000 are expected to have CES-D scores with a diference of −4.42  Upon assessment of the backward stepwise multivariable linear regression model, Q-Q plot did not confrm the assumption of normality.However, the plot of the residual versus ftted value and the VIFs confrmed the assumption of homoscedasticity and multicollinearity, respectively (Appendix A).Te backward stepwise multivariable regression model revealed that the presence of retinal disease and the use of a mobility aid were predictive of CES-D score (Table 8).On average, adjusting for the use of a mobility aid, the presence of retinal disease increased the CES-D score by 2.50 (95% CI: −0.43, 5.43); however, this increase was not signifcant (p � 0.094).Adjusting for the presence of retinal disease, the use of a mobility aid signifcantly (p � 0.028) increased the CES-D score by 4.20 (95% CI: 0.46, 7.94).

Presence of Anxiety Symptoms.
Te HADS-A subscale measures symptoms of anxiety in the past week each using a scale from 0 to 21 in which higher scores represent higher levels of anxiety.In our study population (n � 88), the average HADS-A score was 2.83 with a standard deviation of 2.56 (Table 5).
Te bivariate analyses for the presence of anxiety symptoms with the demographic and clinical variables are presented in Table 10.No variables were signifcantly associated with the presence of anxiety symptoms.
On average, patients living at home with others are expected to have a HADS-A score diference of 0.71 (95% CI: −0.52, 1.94) as compared to patients living alone or in a nursing/retirement home.Te use of a mobility aid, on average, increases the HADS-A score by 0.58 (95% CI: −0.87, 2.04).On average, for each increase in the number of nonocular comorbidities, the HADS-A score increases by 0.11 (95% CI: −0.21, 0.42).For each increase in the number of ocular comorbidities, the HADS-A score increases by 0.53 (95% CI: −0.83, 1.89).On average, the presence of glaucoma changes the HADS-A score by −0.12 (95% CI: −1.32, 1.07).Te presence of retinal disease increases the HADS-A score by 0.84 (95% CI: −0.33, 2.02).Te presence of cataracts changes the HADS-A score by −0.71 (95% CI: −2.15, 0.73).
Upon assessment of the backward stepwise linear regression model, the Q-Q plot did not confrm the assumption of normality, but the plot of the residual versus ftted value confrmed the assumption of homoscedasticity (Appendix A).Te backward stepwise regression model revealed that the use of a mobility aid was the only variable predictive of HADS-A score (Table 8).Once again, this model displays that, on average, the use of a mobility aid increases the HADS-A score by 0.58 (95% CI: −0.87, 2.04); however, this change is not signifcant (p � 0.428).

Sleep Quality.
Te PSQI score measures sleep quality over a 1-month time interval on a global scale from 0 to 21, in which higher scores indicate worse sleep quality.In our study population (n � 90), the average PSQI score was 6.58 with a standard deviation of 3.00 (Table 5).
Te bivariate analyses for sleep quality with the demographic and clinical variables are presented in Table 11.Te following variables were signifcantly associated with sleep quality: use of a mobility aid (p � 0.044) and SES during COVID-19.Te use of a mobility aid, on average,  On average, for every year increase in age, the PSQI score changes by −0.01 (95% CI: −0.09, 0.08).Patients who have completed more than high school in their education have an average diference in their PSQI score of 0.82 (95% CI: −0.48, 2.11) as compared to patients who have completed high school or less.Patients living at home with others are expected to have a PSQI score diference of −1.17 (95% CI: −2.62, 0.29) as compared to patients living alone or in a nursing/retirement home.On average, for each increase in the number of nonocular comorbidities, the PSQI score increases by 0.22 (95% CI: −0.15, 0.59).For each increase in the number of ocular comorbidities, the PSQI score increases by 0.36 (95% CI: −1.18, 1.90).On average, the presence of glaucoma changes the PSQI score by −0.56 (95% CI: −1.94, 0.81).Te presence of retinal disease increases the PSQI score by 0.57 (95% CI: −0.82, 1.95).Finally, the presence of cataracts changes the PSQI score by −0.28 (95% CI: −1.97, 1.42).
Upon assessment of the backward stepwise linear regression model, the Q-Q plot did not confrm the assumption of normality, but the plot of the residual versus ftted value confrmed the assumption of homoscedasticity (Appendix A).Te backward stepwise regression model revealed that the use of a mobility aid was the only variable predictive of PSQI score (Table 8).Once again, this model displays that, on average, the use of a mobility aid signifcantly (p � 0.044) increases the PSQI score by 1.73 (95% CI: 0.05, 3.41).

Social Support and Community Integration.
Te CIQ score measures social support and community integration on a scale from 0 to 29, in which a higher score represents more complete community integration and a higher level of social support.In our study population (n � 88), the average CIQ total score was 14.46 with a standard deviation of 4.07 (Table 5).
Te bivariate analyses for social support and community integration with the demographic and clinical variables are presented in Table 12.Te following variables were significantly associated with social support and community integration: number of nonocular comorbidities (p � 0.047) and presence of retinal disease (p � 0.001).On average, for each increase in the number of nonocular comorbidities, the CIQ score changes by −0.50 (95% CI: −1.00, −0.01).On average, for every year increase in age, the CIQ score changes by −0.10 (95% CI: −0.21, 0.02).Patients who have completed more than high school in their education have an average diference in their CIQ score of −1.23 (95% CI: −2.98, 0.53) as compared to patients who have completed high school or less.Patients living at home with others are expected to have a CIQ score diference of −0.67 (95% CI: −2.67, 1.33) as compared to patients living alone or in a nursing/retirement home.Use of a mobility aid, on average, changes the CIQ score by −1.22 (95% CI: −3.54, −1.10).Te presence of glaucoma increases the CIQ score by 0.80 (95% CI: −1.14, 2.73).Te presence of cataracts changes the CIQ score by −0.07 (95% CI: −2.45, 2.31).On average, for each increase in the number of ocular comorbidities, the CIQ score changes by −1.51 (95% CI: −3.70, 0.68).
Upon assessment of the backward stepwise multivariable linear regression model, the component-plus-residual plot, Q-Q plot, residual versus ftted value plot, and the VIFs confrmed the assumption of linearity, normality, homoscedasticity, and multicollinearity, respectively (Appendix A).Te backward stepwise multivariable regression model revealed that the presence of retinal disease, number of nonocular comorbidities, and education were predictive of CIQ score (Table 8).Adjusting for all other predictors, on average, the presence of retinal disease signifcantly (p < 0.001) changed the CIQ score by −3.10 (95% CI: −4.77, −1.43).For each increase in the number of nonocular comorbidities, the CIQ score signifcantly (p � 0.004) changes by −0.68 (95% CI: −1.15, −0.22).Finally, having an education of more than high school signifcantly (p � 0.033) changes the CIQ score on average by −1.79 (95% CI: −3.44, −0.15).13 demonstrate that the backward stepwise regression models for VRQoL and social support and community integration had generally similar MAEs as compared to a similar previous study by Uruthiramoorthy et al. [30].Te MAE of the VRQoL model in the current study is slightly higher than the models in the previous study suggesting that the model in the current study could be nearly as predictive as the previous models.However, the multivariable model for social support and community integration in the current study had an even lower value of MAE as compared to the study by Uruthiramoorthy et al. [30] indicating that it may be more predictive of the outcome.

Discussion
In this cross-sectional study, the preference-based HRQoL, VRQoL, depression and anxiety symptoms, and social support and community integration of seniors with various eye diseases were assessed.A total of 90 participants were included in the study, and the results showed that the  Te present study revealed that the preference-based HRQoL of our study patients with eye diseases during the pandemic is likely quite good with a mean utility value of 0.88.Moreover, it does not appear that any of the potential predictors had a signifcant level of impact on the HRQoL of seniors during the pandemic.Te analysis of the present study also showed that these seniors with eye disease appeared to have good vision and only a slight loss in visual performance impacting their VRQoL during the pandemic.However, it was also shown that the presence of retinal disease and the number of nonocular comorbidities had a signifcant negative impact on the participants' VRQoL.
Moreover, study participants appeared to have a low presence of depressive symptomatology.Te backward stepwise multivariable regression analysis revealed that the use of a mobility aid did appear show a signifcant increase in depressive symptomatology.Te level of anxiety appeared to be normal, and the quality of sleep appeared to be good overall among patients with eye diseases during the pandemic.Of note, the use of a mobility aid was found to negatively afect sleep quality but not the presence of anxiety symptoms.Previously, sleep quality has been found to be associated with physical disability in older adults which may explain this relationship between the use of mobility aid and poor sleep quality [31].On the other hand, an analysis of the levels of social support and community integration found that participants likely had moderate social support and community integration during the pandemic.We speculate that this fnding was a result of those with a higher education had a better understanding of the pandemic conditions and consequences.Tis higher education and better understanding likely then led individuals to socially isolate more and avoid others which, in turn, had a negative impact on their social support and community integration.We also speculate that through this mechanism, as a result of poorer social support and community integration, both HRQoL and VRQoL may have been negatively impacted (albeit not signifcantly).Furthermore, it was revealed that the presence of retinal disease, number of nonocular comorbidities, and education appeared to have signifcant negative efects on social support and community integration.Based on the key measures of QoL such as preference-based HRQoL, VRQoL levels of depression and anxiety, and access to social support and community integration, the above fndings indicate that the QoL and wellness of the elderly with eye diseases appear to be good.
However, it is important to note that there are studies conducted during this period that highlighted the need for further research to ascertain the impact of COVID-19 on the QoL of the elderly with eye diseases.For instance, a systematic review by Zaher et al. [32] showed that there was a 36% increase in the fear of vision loss due to an increased risk of missing appointments and a 48% increase in the fear of contracting the virus due to ofce visits.Terefore, a future survey study focusing on gathering data about the fear of vision loss due to an increased risk of missing appointments and fear of contracting the virus due to ofce visits among eye disease patients would shed light on the overall QoL of patients during ongoing pandemic.Further, a prospective cross-sectional comparative study by Shalaby et al. [33] found that the pandemic afected visually impaired people which results in diminished QoL.Moreover, majority of seniors who have vision loss tend to experience depressive symptoms [34].Te results of our study resonate that community integration and social support appear to be the lowest among the study population.Tis fnding concurs with the view that containment measures of the pandemic such as lockdowns, social distancing, and curfews may have potentially impacted the lives of seniors with eye diseases.
Interestingly, the results of the preference-based HRQoL in this current study seem to agree with those of previous studies performed in the same location and setting.A previous study by Tomas et al. [35] investigated the preference-based HRQoL among patients with glaucoma and diabetic retinopathy.Another previous study by Uruthiramoorthy et al. [30] investigated the preference-based HRQoL among patients with glaucoma.Both studies reported relatively high mean utility values from 0.89 to 0.94 which is like the value of 0.88 found in the current study using the same TTO technique.Tese similarities may be suggestive of the fact that perhaps the QoL of the patients at this location is relatively high due to the services and facilities provided by the Ivey Eye Institute or due to the environment of London, Ontario, in general.
Furthermore, the study by Uruthiramoorthy et al. [30] found that living arrangements and the use of a mobility aid were signifcant predictors of VRQoL.Additionally, the study by Uruthiramoorthy et al. [30] found age, sex, income, living arrangement, and use of mobility aids to be predictors of social support and community integration.Te diference seen between the predictor variables in the two studies may be due to a multitude of factors such as the COVID-19 pandemic conditions, diferent time periods, general eye disease patients versus glaucoma only patients, or the senior patient population.
While the current study provides valuable information on the QoL of seniors with eye diseases, there were some limitations.A potential limitation stems from the use of convenience sampling which is associated with sampling bias [36].Of note, many participants in our study were identifed as being white and not requiring the use of a mobility aid, both of which are characteristics associated with a higher QoL [37,38].Moreover, the mean number of nonocular comorbidities was relatively low at 1.6, which may also have been associated with a higher QoL.As a result, this means that the results may not be representative of the actual population and that they are likely not entirely generalizable to other populations of seniors with eye diseases [39].
Tis study was also limited by the number of included participants.As with most studies, more participants would allow for greater power and a greater ability to detect differences that are present.As such, a greater number of participants may have resulted in the detection of a signifcant impact on outcome measures like preference-based HRQoL.Additionally, there is an absence of data referring to the present ocular diseases such as the visual acuity of the participants which may have infuenced HRQoL or VRQoL.Furthermore, participants were divided into more general subgroups regarding their ocular diseases without specifying the disease to maintain numbers within groups for data analysis.However, this may also have infuenced the perceived quality of life and the impact of the pandemic.Two other limitations of this study are inherent to the cross-sectional design that was used.Te frst of which is that in a cross-sectional design one cannot necessarily determine whether the exposure did precede the outcome [40].As such, in the current study, it is not entirely certain as to whether the study participants' wellness was due to the presence of the COVID-19 pandemic conditions.Te second of which is length bias which is systematic error due to selection of disproportionate of long duration cases [40].With this limitation in mind, it may be that many participants recruited in the current study were patients who have had their eye diseases for a long duration of time and been properly managed for a long time as well.As such, these patients may have been more adapted to their conditions and reported better QoL and wellness.
Te fnal potential limitation of the study stems from healthy volunteer bias.Historically, volunteers in medical research tend to have a lower risk of mortality and other health problems compared to those who are not volunteers [41].Patients in the current study participated on a voluntary basis.As a result, this may mean that the results of the current study were biased in favor of higher QoL due to the study participants having healthier lifestyles than those who did not volunteer to participate.However, it was noted that of the 128 patients who were approached, 115 agreed to participate while 25 of these patients did not pass the inclusion/exclusion criteria.Tis suggests that the probability of such selection bias is likely low.
Te current study has major implications, especially when it comes to focusing healthcare resources on the most vulnerable groups of the population.However, future studies in this area could focus on the relationship between QoL and eye diseases using eye disease-specifc questionnaires.For example, instruments such as the Ocular Surface Disease Index for patients with DED and the Glaucoma Quality of Life-15 for patients with glaucoma could be studied [42,43].Generic preference-based measures of health to measure HRQoL such as the EuroQoL-5D and 36-Item Short Form Survey could be used [44,45].Moreover, future studies with larger sample sizes focused on the QoL of seniors with eye diseases after the pandemic is completely over would be worthwhile to conduct so that a comparison can be made to better understand the impact of the pandemic conditions on this population of patients.

Conclusions
Overall, the current study has found that the QoL among seniors with eye diseases appeared to be good.Measures of preference-based HRQoL and VRQoL appeared to be high.Indications of depression and anxiety symptoms were likely low, while community integration and social support appeared to be moderate.Te presence of retinal disease and the number of nonocular comorbidities both appeared to negatively impact VRQoL and social support and community integration.Education appeared to impact CIQ negatively.Te use of a mobility aid appeared to negatively afect depressive symptoms and sleep quality.
69) as compared to patients who have completed high school or less.Patients living at home with others are expected to have a CES-D score diference of −0.05 (95% CI: −3.13, 3.03) as compared to patients living alone or in a nursing/retirement home.For each increase in the number of ocular comorbidities, the CES-D score increases by 2.10 (95% CI: −1.33, 5.53).On average, the presence of glaucoma changes the CES-D score by −1.99 (95% CI: −4.95, 0.98).Te presence of retinal disease increases the CES-D score by 2.74 (95% CI: −0.18, 5.67).Te presence of cataracts changes the CES-D score by −0.43 (95% CI: −4.06, 3.20).

Table 1 :
Demographic characteristics of the included participants.

Table 2 :
Pearson correlation coefcients (p values) for the associations between continuous predictor variables.

Table 3 :
Chi-square tests (p value) for the associations between categorical predictor variables.

Table 4 :
Results of associations between continuous and categorical predictor variables.

Table 5 :
Summary of questionnaire scores for all participants.

Table 6 :
Unadjusted efects of variables with Time Trade-Of preference-based HRQoL.

Table 8 :
Coefcient estimates (95% confdence interval) for the backward stepwise linear regression models of the questionnaire outcomes.

Table 9 :
Unadjusted efects of variables with depressive symptoms.

Table 10 :
Unadjusted efects of variables with anxiety symptoms.

Table 12 :
Unadjusted efects of variables with social support and community integration.

Table 13 :
Backward linear regression model assessments from leave-one-out cross-validation.