Purpose
To investigate the prevalence of and risk factors for depression among participants with glaucoma and the predictive value of glaucoma for depression.
Design
Cross-sectional study.
Methods
This study included 6760 participants in the National Health and Nutrition Examination Survey (NHANES) between 2005 and 2008, aged ≥40 years, who reported a presence or absence of glaucoma. Demographic and disease-related information was obtained by interview. Self-reported measures of vision were ascertained via items from the Visual Function Questionnaire (VFQ-25). Participants underwent visual acuity examination, fundus photography, and visual field testing with screening frequency-doubling technology (FDT N-30-5). The main outcome was presence of depression, as determined by a score ≥10 on the Patient Health Questionnaire-9 (PHQ-9).
Results
Prevalence of depression among participants with and without glaucoma was 10.9% (SEM 2.2%) and 6.9% (SEM 0.62%), respectively. While the presence of glaucoma was significantly associated with depression after adjustment for demographic factors (OR 1.80, 95% CI 1.16–2.79), this association was not significant after adjustment for self-reported general health condition (OR 1.35, 95% CI 0.822–2.23). Among participants with glaucoma, objective measures of glaucoma severity were not significant predictors for depression. However, several self-reported measures of visual function were significantly associated with depression.
Conclusions
Glaucoma is a significant predictor of depression after adjustment for demographic factors and multiple comorbidities, but not after adjustment for self-reported general health condition. Among participants with glaucoma, self-reported measures of vision were significant risk factors for depression, whereas objective measures of vision were not.
Glaucoma is a chronic, progressive, and irreversible disease that can result in severe visual disability. Over the past 2 decades, interest has increased among physicians and other health professionals regarding patients’ quality of life, which has led to a better understanding of glaucomatous disease burden and outcomes of treatment. It has been postulated that mental health may impact clinical factors such as glaucoma medication adherence and persistence.
Previous studies have estimated a 10% to 12% prevalence of depressive symptoms in subjects with glaucoma. The prevalence of depression was found to be as high as 32.1% in patients with severe glaucomatous disease. Furthermore, depression has been found to be associated with patients’ perception of vision; however, in contrast to subjective measures of visual perception, objective measures of function such as visual acuity or visual field results have not been linked to glaucoma diagnosis or depression severity. Common limitations of prior analyses have included small sample size, enrollment of patients from clinic rather than population-based settings, and lack of sufficient normal controls, with the latter being particularly problematic in large studies of depression and glaucoma.
The National Health and Nutrition Examination Survey (NHANES) is an annual national population-based study administered by the Centers for Disease Control and Prevention (CDC) designed to assess the health status of the US population, sampling approximately 5000 persons per year. It includes not only an extensive interview questionnaire related to a variety of physical and mental health conditions, but also a physical examination component, with an assessment of visual acuity and objective refraction. From 2005 to 2008, NHANES also included fundus photography and visual field testing of participants over the age of 40. This extensive national database can be used to determine the prevalence of diseases such as glaucoma and depression, to investigate the predictive value of the presence of glaucoma for depression, and to discern the prevalence of risk factors for depression among individuals with glaucoma.
Methods
Sample and Population
We used publicly available data from the 2005-2008 administrations of the NHANES, a cross-sectional series of interviews and examinations of the civilian, noninstitutionalized population of the United States. NHANES is administered by the CDC for the purpose of providing US health statistics and uses a stratified multistage sampling design that requires a weighting scheme to most accurately estimate disease prevalence in the US population.
Our analysis included 6760 subjects consisting of all survey participants aged 40 years and older who underwent both the interview and examination phases of the survey between 2005 and 2008. Thirty-five participants were excluded for not knowing whether or not they had glaucoma, and 2 participants were excluded for not answering the glaucoma question.
Measures
The primary predictor variable was the presence of self-reported glaucoma (n = 453). The main outcome was the presence or absence of depression, as determined by score on the Patient Health Questionnaire (PHQ)-9, a self-administered version of the depression module of the Primary Care Evaluation of Mental Disorders Questionnaire (PRIME-MD). For each of the 9 Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria for the diagnosis of depression, participants reported how often they were bothered by that symptom over the preceding 2 weeks, with scores on each question ranging from 0 (not at all) to 3 (nearly every day) for a total score range from 0 to 27. A prior validation study found that a score ≥10 achieved 88% sensitivity and 88% specificity for major depression. The range may be further subdivided into scores of 5, 10, 15, and 20, representing mild, moderate, moderately severe, and severe depression, respectively.
Potential confounders in our analysis included age, sex, ethnicity, annual household income, and education; comorbid medical conditions such as self-reported history of stroke, thyroid disease, emphysema, liver disease, cancer, congestive heart failure, diabetes, angina, coronary heart disease, myocardial infarction, and chronic bronchitis; comorbid eye conditions such as self-reported history of cataract extraction, diabetic retinopathy, and macular degeneration; self-reported general health condition (self-rated as excellent or very good, fair, or poor or very poor); body mass index; and spherical equivalent on objective refraction.
Potential mediators in our analysis of the relationship between glaucoma and depression included several self-reported measures of vision and visual disability, as well as objective measures of visual function. All self-reported measures of vision were ascertained from a subset of questions from the National Eye Institute Visual Function Questionnaire (NEI-VFQ-25), a reliable and validated instrument for self-reporting visual disability. These included a series of questions assessing the difficulty that participants experience in performing vision-related tasks, which include reading ordinary print in newspapers; doing work or hobbies that require seeing well close up; going down steps, stairs, or curbs in dim light or at night; noticing objects off to the side while walking; finding something on a crowded shelf; and driving during the daytime in familiar places. We rated each of these tasks on a scale from 0 (no difficulty) to 4 (unable to perform as a result of poor eyesight), for a total score range from 0 to 24. NHANES also ascertained whether or not participants had trouble seeing despite the aid of glasses or contact lenses; participants’ self-assessment of their eyesight, ranging from excellent to very poor; how much time participants spent worrying about their eyesight; and how often their vision limited the length of common daily activities, ranging from no time to all of the time.
Objective measures of visual function among participants with glaucoma included vertical cup-to-disc ratio (VCDR) as graded from fundus photographs, best-corrected visual acuity (BCVA) from objective refraction, and presence of visual field defects based on a N-30-5 frequency-doubling technology (FDT) screening protocol, which is a 19-point supra-threshold visual field screening test. In NHANES, abnormal FDT status was defined by a 2-2-1 algorithm: 2 fields in the first test below the 1% threshold level, at least 2 fields in the second test below the 1% threshold level, and at least 1 failed field in the same location on both tests. Examinations were considered unreliable if either of the 2 tests on each eye had at least 2 out of 3 false-positive or blind spot errors, or the technician supervising the test noted lack of fixation. The NHANES 2-2-1 algorithm for FDT N-30-5 had a previously demonstrated sensitivity of 54.8% and specificity of 91.9% in detecting subjects with glaucoma. In addition, we further stratified FDT results of the first visual field test administered for each eye into normal, early, moderate, or severe visual field defects based on the clinical classification scheme previously published and validated against the Glaucoma Staging System, which showed a Cohen kappa agreement of .679 and specificity of 95%. The classification of severe glaucoma was slightly modified for our study and defined as more than 9 probability ( P < 1%) defects (same as the original criteria), or more than 12 abnormal points with more than 6 P < 1% defects (modified from the original criteria where the cutoff was 0.5% rather than our 1%). This slight modification was necessary because of lack of P < 0.5% threshold data in the NHANES data set.
In addition, the use of topical glaucoma medications in the 30 days prior to the interview, including whether or not such medications included a topical beta blocker, was ascertained by self-report. Participants who reported using betaxolol, metipranolol, levobunolol, and timolol or timolol-containing combination medications were considered to be on topical beta blocker therapy.
Data Analysis
We compared the distribution of possible confounding and mediating variables between participants with and without self-reported glaucoma using design-adjusted Rao-Scott Pearson-type χ 2 and Wald tests for categorical and continuous variables, respectively. Multivariate logistic regression models were used to examine the independent association of self-reported glaucoma with questionnaire-assessed depression as defined by a PHQ-9 score ≥10, with confounders added sequentially to the model. Confounding comorbidities not found to be significant at the P < .1 level in multivariate models were excluded from the final model. These excluded confounders were body mass index; spherical equivalent on objective refraction; comorbid eye conditions including history of cataract extraction, diabetic retinopathy, and macular degeneration; and medical conditions including congestive heart failure, diabetes mellitus, angina, coronary heart disease, myocardial infarction, and chronic bronchitis. Similar multivariate logistic regression models were constructed to examine the independent association between depression and objective measures of glaucoma severity including visual acuity, FDT results, cup-to-disc ratio, and number of topical glaucoma medications used, as well as subjective measures of glaucoma severity, while adjusting for the same set of confounders. The subjective measures from the NEI-VFQ, which assessed the difficulty that participants had in performing daily activities, were analyzed with the “No difficulty” category as the reference. In order to most accurately calculate confidence intervals around estimates for the US national population, all data analysis was performed in Stata 12.0 (StataCorp, College Station, Texas, USA) using weighted data, and standard errors of population estimates were calculated using Taylor linearization methods.
Results
Population Characteristics
The combined 2005-2008 NHANES data yielded 6760 participants over the age of 40 who underwent both the interview and the examination, and who were able to self-report glaucoma status. Of these, 453 participants self-reported glaucoma, representing 5.06% (standard error of the mean [SEM] 0.33%) of the sampled US civilian noninstitutionalized population.
Demographic and general health characteristics of participants self-reporting glaucoma are compared to the control group of those who self-reported not having glaucoma in Table 1 . Unadjusted P values were calculated for a crude difference in means or proportions between these 2 groups. The mean ages of those self-reporting glaucoma and controls were 66.9 (SEM 1.03) and 56.5 (SEM 0.38) years, respectively ( P < .001). All demographic variables and almost all general health characteristics differed significantly between the 2 groups. Notably, the self-reported variable “general health condition” differed significantly between the 2 groups ( P < .001), with 34.5% (SEM 3.2%) of the glaucoma group reporting poor or very poor general health, compared to 18.9% (SEM 0.91%) of the control group.
| Self-reported No Glaucoma (n = 6307) Mean or % a (SE) | Self-reported Glaucoma (n = 453) Mean or % a (SE) | P Value b | |
|---|---|---|---|
| Demographics | |||
| Age, years | 56.5 (0.38) | 66.9 (1.03) | <.001 |
| Female | 53.0% (0.71%) | 52.4% (3.2%) | .03 |
| Race | |||
| Mexican | 5.6% (0.72%) | 4.0% (1.3%) | |
| Other Hispanic | 3.2% (0.62%) | 2.7% (1.1%) | |
| Non-Hispanic white | 75.9% (2.1%) | 72.5% (3.6%) | .004 |
| Non-Hispanic black | 10.1% (1.3%) | 16.7% (2.6%) | |
| Other and multiracial | 5.2% (0.61%) | 4.2% (1.4%) | |
| Education | |||
| <9th grade | 7.4% (0.61%) | 14.1% (2.6%) | |
| 9th grade – less than high school graduate | 11.6% (0.86%) | 13.4% (1.6%) | |
| High school graduate or GED equivalent | 26.1% (0.95%) | 28.2% (2.9%) | <.001 |
| Some college | 27.8% (0.95%) | 28.8% (3.1%) | |
| College graduate and beyond | 27.2% (1.6%) | 15.5% (2.1%) | |
| Annual household income | |||
| <$20 000 | 14.8% (0.865) | 23.4% (3.1%) | |
| $20 000 – $44 999 | 26.8% (1.3%) | 35.5% (2.9%) | |
| $45 000 – $74 999 | 22.0% (0.99%) | 21.0% (2.3%) | <.001 |
| $75 000 and up | 33.7% (1.8%) | 16.9% (2.7%) | |
| >$20 000 c | 2.8% (0.39%) | 3.3% (1.0%) | |
| General health status and comorbidities | |||
| General health condition | |||
| Excellent or good | 42.4% (1.4%) | 37.8% (3.3%) | |
| Fair | 38.8% (0.78%) | 27.7% (3.0%) | <.001 |
| Poor or very poor | 18.9% (0.91%) | 34.5% (3.2%) | |
| Medical comorbidities | |||
| Stroke | 4.3% (0.43%) | 10.2% (2.2%) | .002 |
| Emphysema | 2.6% (0.23%) | 7.4% (0.17%) | <.001 |
| Thyroid problem | 13.4% (0.65%) | 18.5% (2.2%) | .01 |
| Liver condition | 4.4% (0.31%) | 4.1% (1.1%) | .74 |
| Cancer | 12.2% (0.53%) | 22.1% (2.5%) | .001 |
a All means, proportions, and standard errors are weighted estimates of the US population characteristics, taking into account NHANES’ complex sampling design.
b All P values are unadjusted. P values were calculated using Wald test for continuous variables and design-adjusted Rao-Scott Pearson χ 2 test for categorical variables.
c Participants who were unable to provide a more specific annual household income were asked to indicate whether the household income exceeded $20 000 per year.
The unadjusted means and proportions relating to vision and depression in the glaucoma and control groups are presented in Table 2 . Participants with glaucoma had significantly worse visual function and vision-related characteristics by objective measures, including worse logarithm of minimal angle of resolution (logMAR) BCVA in the worse eye (0.254 vs 0.176 in controls, P < .001), larger VCDR in the eye with the smaller VCDR (0.44 vs 0.36 in controls, P < .001), and higher proportion of subjects with visual field defects in 1 or both eyes (29.6% vs 6.0% in controls, P < .001). Furthermore, a higher proportion of subjects in the glaucoma group self-reported poor or very poor vision, worrying about their eyesight most or all of the time, and being limited by poor vision in the performance of activities most or all of the time relative to those in the control group. The prevalence of depression also differed significantly between the 2 groups, measuring 10.9% (SEM 2.2%) in the glaucoma group compared to 6.9% (SEM 0.62%) in the control group ( P = .02).
