Association Between Noninfectious Uveitis and Psychological Stress

Key Points

Question  Is there an association between psychological stress and uveitis?

Findings  In this cross-sectional, case-control study of 120 patients with uveitis, having uveitis was associated with a significantly higher score on the 10-question Perceived Stress Scale in comparison with individuals serving as nonuveitic controls.

Meaning  These findings suggest that uveitis is associated with higher levels of stress compared with controls without ocular disease; consequently, patients with uveitis may benefit from increased focus on the psychological effect of this disease, especially if future longitudinal studies show that these findings have a cause-and-effect association.

Importance  Uveitis involves dysregulation of the ocular immune system. Stress has been shown to affect immune function, but it is unclear whether there is an association between stress and uveitis.

Objective  To determine whether having uveitis is associated with psychological stress.

Design, Setting, and Participants  A cross-sectional, case-control study including a self-administered survey, medical records review, and diurnal salivary cortisol test was conducted at a university-based uveitis clinic and comprehensive eye clinic. Participants included 146 consecutive adults with noninfectious uveitis and age-matched controls with no eye disease. The study was conducted from December 1, 2017, to March 14, 2018.

Main Outcomes and Measures  Participants completed the self-administered, Cohen 10-item Perceived Stress Scale (PSS-10), a demographics questionnaire. Responses to each question were categorized on a 5-point Likert scale, with total scores ranging from 0 (no stress) to 40 (high stress). In addition, participants submitted 3 salivary cortisol samples. Those with uveitis were classified as having recently active or controlled disease through medical records review. The prespecified primary analysis was a linear regression of PSS-10 score and uveitis correcting for age, sex, educational level, employment, and median income. Secondary analyses included comparing PSS-10 scores in patients with recently active and controlled uveitis, determining predictors of stress, and comparing diurnal salivary cortisol between uveitis and control groups.

Results  Of 146 eligible patients, 17 declined participation and 9 consented but were excluded because they did not complete both questionnaires, resulting in 120 patients (80 uveitis; 40 controls) in the final analysis. Eighty participants (66.7%) were women, and 70 (58.3%) were white. Median age was 40 years (interquartile range, 29-59 years). Having uveitis was associated with a 4.3-point increase in PSS-10 score (95% CI, 1.8 to 6.9; P = .002). There was no significant difference in PSS-10 scores between patients with recently active and controlled uveitis (1.0 point greater for patients with active uveitis; 95% CI, −2.0 to 3.9; P = .52). Factors associated with increased PSS-10 score in patients with uveitis included female sex (coefficient, 4.0; 95% CI, 1.6 to 6.5; P = .002), current immunomodulatory therapy (coefficient, 2.5; 95% CI, −0.3 to 5.2; P = .08), history of depression (coefficient, 3.8; 95% CI, 0.8 to 6.8; P = .02), and having posterior or panuveitis (coefficient, 2.6; 95% CI, 0.8 to 4.4; P = .006). Of the 70 participants (58.3%) who had testable samples for cortisol analysis, diurnal salivary cortisol levels did not significantly differ between uveitis and nonuveitis groups.

Conclusions and Relevance  These findings suggest that patients with uveitis have higher levels of psychological stress compared with controls, yet no significant difference was identified in the stress of patients with active vs controlled uveitis. Consequently, comprehensive treatment for noninfectious uveitis may be able to address the psychological results of this disease.

Uveitis is a significant cause of ocular morbidity that accounts for 10% to 15% of blindness in the United States.¹ Most uveitis cases worldwide are presumed to be immune-mediated and noninfectious.² In many instances, uveitis is associated with autoimmune disease.^2,3 Animal models and human peripheral blood specimens suggest that uveitis may involve dysregulation of regulatory T lymphocytes in the eye.^4-9 The negative consequences of uveitis on vision are well known, but the effect on mental health has not been as well elucidated.^10-12

An abundance of data support a connection between psychological stress and altered immune function. It has been shown that short-term stress can activate the hypothalamic-pituitary-adrenal axis, leading to redistribution of leukocytes to tissues.^13-18 Chronic stress can eventually lead to attenuated immune response over time.^13,16,18-20 Hypothalamic-pituitary-adrenal axis-immune system coordination may be mediated by release of the stress hormone cortisol, which can be measured in saliva.^18,21 Psychological stress has been associated with autoimmune disease activity, including diseases that may be associated with ocular inflammation.^22-27 Several studies have evaluated stress in uveitis, predominantly in its anterior subtype.^28-32 However, they have been retrospective and underpowered, and have produced inconsistent results.^28-34

This study aimed to investigate the psychological stress levels of patients with noninfectious uveitis compared with a nonuveitic population, as well as whether demographics and clinical characteristics are predictors of psychological stress. The self-administered Cohen 10-item Perceived Stress Scale (PSS-10) questionnaire was used as a subjective, patient-reported measure of stress levels. Diurnal salivary cortisol measurement was used as an objective, biochemical measure of patients' stress levels. Overall, this study explored a psychosocial outcome of uveitis to highlight the nonocular effects of this disease.

This study included consecutive patients seen between December 1, 2017, and March 14, 2018, at the Francis I. Proctor Foundation at the University of California, San Francisco (UCSF) and the UCSF Comprehensive Optometry Clinic. Eligible patients were aged 18 years or older, English-speaking, and able to provide written informed consent. Patients were included in the uveitis group if they had a confirmed diagnosis of noninfectious uveitis and had a prior laboratory workup excluding infectious causes. Controls were individuals seen in the optometry clinic for general eye care who did not have any ocular abnormalities requiring prescription treatment. Controls were recruited by matching for 10-year age strata. Ethical approval of the study design was obtained from the UCSF Institutional Review Board. Participants provided written informed consent; there was no financial compensation.

Participants completed the Cohen PSS-10, a validated self-administered questionnaire, to assess psychological stress.^35-37 Responses to each question were categorized on a 5-point Likert scale, with total scores ranging from 0 (no stress) to 40 (high stress). An additional questionnaire was created to record self-reported comorbid conditions, uveitis history, smoking and drinking status, current pain level on a 100-mm visual analog scale, and demographics, such as race/ethnicity, employment, and educational level. Participants were given the option to complete the questionnaires at the clinic or at home and return them by mail with a prepaid, preaddressed envelope.

Participants’ medical records were reviewed for best-corrected visual acuity, diagnosis of comorbidities, and use of psychiatric medication. The comorbid conditions recorded were heart disease, cancer, depression, anxiety, type 2 diabetes, hypertension, and systemic autoimmune diseases. Patients with uveitis were subgrouped as having recently active or controlled disease based on medical records review. The recently active subgroup was defined by any of the following in either eye within the past 90 days: greater than 0.5+ anterior chamber cell according to Standardization of Uveitis Nomenclature Working Group criteria,³⁸ greater than 0.5+ vitreous haze according to the National Eye Institute vitreous haze scale,³⁹ presence of an active retinal or choroidal lesion, or presence of cystoid macular edema as recorded by spectral-domain optical coherence tomography. The controlled subgroup was defined by absence of these criteria. Information on disease cause, anatomic location of inflammation, current treatments, and recent ocular injections was collected.

Anatomic location of inflammation was reported as anterior, intermediate, anterior and intermediate, posterior, and panuveitis using the Standardization of Uveitis Nomenclature classification system.³⁸ Visual acuity using a Snellen eye chart was recorded for each eye, then converted to a logMAR scale.⁴⁰ Low vision of count fingers, hand motion, light perception, and no light perception were recorded as logMAR 1.7, 1.8, 1.9, and 2.0, respectively. Median household income by zip code served as a proxy for patient socioeconomic status and was determined using the 2010 US census.⁴¹

Patients performed an at-home saliva collection for measuring salivary cortisol levels as a biometric indicator of stress. At enrollment, patients were given a collection kit with instructions to collect their saliva 3 times during 1 day within 3 weeks of their appointment and return the samples to UCSF through priority mail. Patients were instructed to collect saliva on wakening, 30 minutes after wakening, and before bed. Patients recorded the exact times of collection and returned the recorded times with the kit. Samples were stored at −20°C on arrival at UCSF. Samples were thawed and analyzed in duplicate on a 96-well plate using an expanded-range, high-sensitivity salivary cortisol enzyme-linked immunoassay kit (Salimetrics). The enzyme-linked immunoassay kit has been tested for the off-target effects of prednisolone, prednisone, and cortisone, which have cross-reactivities of 0.568%, less than 0.004%, and 0.130%, respectively.⁴² Prior data suggest that topical corticosteroids should not affect systemic cortisol levels.⁴³ Oral corticosteroids should minimally cross-react with the enzyme-linked immunoassay but could affect the hypothalamic-pituitary-adrenal axis.

The sample size was determined for the prespecified primary analysis, which was a linear regression of the PSS-10 score and uveitis correcting for significant covariates from the most-recent normative data set of Americans answering the PSS-10 questionnaire.⁴⁴ Using an α level of .05, a sample size of 120 patients gave us greater than 90% power to detect a 4-point difference (one-half of an SD in the normative data set) in the PSS-10 score between the uveitis and control groups. Participants were recruited in a ratio of 1 recently active subgroup patient to 1 controlled uveitis subgroup patient to 1 control patient to allow for greater than 90% power to detect a 4-point difference in the PSS-10 score between patients with recently active and controlled disease. Participants were included in the analysis if they provided informed consent and completed both questionnaires within 3 weeks of enrollment.

The Fisher exact test, Mann-Whitney test, and t test were used to compare baseline characteristics. Visualization of the distribution of PSS-10 scores was performed using the Epanechnikov kernel function with a bandwidth of 1.5951. The prespecified primary analysis was a linear regression model predicting the PSS-10 score, with the main predictor being diagnosis of uveitis, adjusting for age, employment, sex, income, and educational level. A subanalysis comparing PSS-10 scores between the recently active uveitis, controlled uveitis, and nonuveitis groups used linear regression with the same covariates as the primary analysis. Predictors of PSS-10 scores in patients with uveitis were assessed in a multivariable model using backward stepwise linear regression retaining variables with a P < .10 (likelihood ratio test). Association between PSS-10 score and logMAR best-corrected visual acuity in the better-seeing eye was assessed using Spearman rank correlation.

Measures of diurnal salivary cortisol levels were compared between the uveitis and control groups using linear regression predicting the diurnal cortisol measure, with the main predictor being uveitis diagnosis, correcting for oral corticosteroid use as a dichotomous variable. Samples were used to calculate the following measures: awakening cortisol response (ACR) (difference between cortisol at wakening and 30 minutes after wakening), a measure of how much cortisol is released from waking to its daily peak; diurnal slope (difference between evening cortisol and 30 minutes after wakening cortisol over time between samples), an estimate of the rate of decline in cortisol release from peak to nadir; and normalized area under the curve with respect to ground (AUC_g) (total AUC of all data points divided by total waking hours), which is an estimation of total daily cortisol release.

Statistical tests were 2-sided and a P value of <.05 was considered significant. For the secondary analysis comparing recently active uveitis, inactive uveitis, and nonuveitis controls, a P value of <.02 was considered significant. Statistical analyses were conducted in Stata, version 14 (StataCorp).

Of 146 patients eligible for the study, 17 declined participation and 9 consented but did not complete both questionnaires, resulting in 120 consecutive patients (80 uveitis; 40 controls) included in this analysis. Eighty participants (66.7%) were women, and 70 (58.3%) were white. Median age was 40 years (interquartile range, 29-59 years). Demographic characteristics were predominantly balanced between the uveitis and control groups (Table 1). The uveitis group had worse converted logMAR best-corrected visual acuity in the better-seeing eye (uveitis group: median, 0.1; interquartile range [IQR], 0-0.25]; control group: median, 0; IQR, 0-0; P < .001) and greater mean (SD) pain score on a 100-mm visual analog scale (16.2 [23.5] vs 6.9 [17.0] mm; P = .02). Patients with uveitis were more likely to have an autoimmune condition (37 patients [46.3%]) compared with the controls (3 patients [7.5%]) (P < .001). However, the groups did not differ significantly in other comorbidities, with a median of 1 comorbid condition (IQR, 1-1) in the control group and 1 comorbid condition (IQR, 0-1) in the uveitis group (P = .25).

The mean PSS-10 score was 13.1 (5.6) in the control group and 17.0 (7.5) in the uveitis group (Figure 1). The prespecified primary model found that having uveitis was associated with a 4.3-point increase in the PSS-10 score (95% CI, 1.8-6.9; P = .002; 1000 permutations). Additionally controlling for the presence of a comorbid systemic autoimmune disease also found that having uveitis was significantly associated with an increase in the PSS-10 score (coefficient, 3.7; 95% CI, 0.9-6.4; P = .01).

The mean PSS-10 score was 18.0 (7.8) in the recently active uveitis subgroup and 16.0 (7.3) in the controlled uveitis subgroup (Figure 2). In comparison with nonuveitic controls, having recently active uveitis was associated with a 4.8-point increase in the PSS-10 score (95% CI, 1.9 to 7.7; P = .001) and having controlled uveitis was associated with a 3.8-point increase in the PSS-10 score (95% CI, 0.9 to 6.8; P = .01). There was only a 1.0-point, nonsignificant difference in PSS-10 score between recently active and controlled uveitis subgroups (95% CI, −2.0 to 3.9; P = .52).

A hypothesis-generating analysis was performed using backward stepwise linear regression of PSS-10 score in patients with uveitis (Table 2) with demographic and clinical predictors (eTable in the Supplement). Backward stepwise linear regression found that the following predictors were associated with significant increases in the PSS-10 score: female sex (coefficient, 4.0; 95% CI, 1.6 to 6.5; P = .002), a history of depression (coefficient, 3.8; 95% CI, 0.8 to 6.8; P = .02), having posterior or panuveitis compared with having anterior or intermediate uveitis (coefficient, 2.6; 95% CI, 0.8 to 4.4; P = .006), and current immunomodulatory therapy (coefficient, 2.5; 95% CI, −0.3 to 5.2; P = .08). A 1-standard drink per week increase in average alcohol consumption was associated with a decrease in the PSS-10 score (coefficient, −0.5; 95% CI, −0.8 to −0.1; P = .01). Spearman correlation between converted logMAR best-corrected visual acuity in the better eye and PSS-10 score found a Spearman r of 0.25 (P = .03).

Eighty-five patients (70.8%) completed the saliva collection: 56 (70.0%) in the uveitis group and 29 (72.5%) in the control group. Four patient sample sets with inadequate volume for testing were excluded. Six patients’ sample sets were excluded because the first 2 samples were taken less than 15 or greater than 45 minutes apart and would not accurately represent the diurnal cortisol peak. After excluding samples that had a pH outside the testing range (3.5-9.0),⁴² 44 patients (55.0%) in the uveitis group and 26 patients (65.0%) in the control group had complete cortisol test data for analysis, summing to a total of 70 patients (58.3%) who completed a per protocol saliva collection with analyzable samples.

Three measures of diurnal cortisol levels were calculated. Raw cortisol data indicated controls had a mean (SD) awakening cortisol response of 118 (303) ng/dL, diurnal slope of −41.0 (57.0) ng/dL• h, and a normalized AUC_g of 275 (154) ng/dL. Patients with uveitis had a mean ACR of 92 (166 ng/dL), diurnal slope of −23.0 (17.0) ng/dL• h, and an AUC_g of 220 (135) ng/dL (Figure 3). All comparisons were corrected for oral corticosteroid use. The awakening cortisol response did not differ in patients with and without uveitis, with a mean ACR response 4 ng/dL lower in the uveitis group (95% CI, 117 ng/dL lower, 110 ng/dL higher; P = .95). Patients with uveitis had a mean 15-ng/dL• h higher diurnal slope than patients without uveitis (95% CI, 4 ng/dL• h lower, 34 ng/dL• h higher; P = .11). Mean normalized AUC_g was 33 ng/dL lower in patients with uveitis compared with nonuveitic controls (95% CI, 103 ng/dL lower, 36 ng/dL higher; P = .34). As a sensitivity analysis, we included baseline cortisol level in the regression model for each diurnal cortisol level measure, and the difference between the uveitis and control groups remained nonsignificant for all 3 measures.

The results of this cross-sectional, case-control study suggest that uveitis is associated with greater psychological stress compared with nonuveitis as measured by the PSS-10 questionnaire, yet there was no statistically significant difference in PSS-10 scores between patients with controlled and those with recently active uveitis. Multivariable analysis found that female sex, a history of depression, posterior or panuveitis, and immunomodulatory therapy were associated with increased PSS-10 scores. No measure of diurnal cortisol levels (awakening cortisol response, diurnal slope, or AUC_g) was significantly different between the uveitis and nonuveitis groups.

Although no formal clinically meaningful difference has been proposed for the PSS-10 questionnaire, the difference in PSS-10 scores found in our study is comparable to that noted in 2 randomized trials of psychological interventions where a 5-point change was thought to be meaningful.^45,46 Because patients in the uveitis group were being treated in our clinic, it is possible that the association between uveitis and the PSS-10 scores was lowered owing to better control of their disease and that patients with less-optimal control of their inflammation could have worse PSS-10 scores.

Prior studies have found that ocular inflammation is associated with negative outcomes in mental health. Patients with a uveitis score lower on the Medical Outcomes Study 36-Item Short Form Questionnaire and have higher rates of depression.^{11,12,31,32,47,48} However, the role of psychological stress in noninfectious uveitis is less clear. Most prior studies on this subject focused on recurrent acute anterior uveitis to determine if high stress is a risk factor for disease flares. One study compared symptoms of distress in 60 patients with idiopathic, recurrent acute anterior uveitis and a group of patients without ocular disease.²⁸ If any recurrent acute anterior uveitis relapsed in any patient within 12 months, they were reinterviewed to assess for recent changes in distress. Similar to our study, that study found no significant difference in distress between patients in remission and those experiencing acute recurrence of uveitis, but noted a significant difference between the patients with uveitis and controls. The authors postulated that the increase in distress among patients with uveitis was not related to disease activity, but rather represented a response to the general experience of having a chronic disease. However, only 15 patients in this study experienced a recurrence, so the sample of patients with active uveitis was small.

Two other studies of patients with recurrent acute anterior uveitis suggested that patients with uveitis have higher levels of psychological distress, lower disease-specific quality of life, and less variety in coping strategies.^30,33 While all of these studies were limited by small size and low generalizability, overall, they support the idea that psychological stress may play a role in uveitis. Still, it is unclear whether stress is a trigger for uveitis activity or a result of the additional anxieties that come with having a chronic, vision-compromising disease. Our study cannot determine whether stress is a trigger or a result of uveitis. However, the findings contribute to the growing body of evidence suggesting that patients with uveitis appear to have higher levels of stress compared with nonuveitic controls and would benefit from additional attention to their psychological well-being.

We also identified predictors of greater stress. Traditional demographics, such as age, race/ethnicity, educational level, work status, and income did not explain as much of the variability in PSS-10 scores compared with prior studies using the questionnaire.⁴⁴ However, median household income by zip code is a weak proxy for socioeconomic status, and this variable may have not fully captured the association with a key confounder that is a known contributor to psychological stress.⁴⁴ The predictors of the PSS-10 score that we found are consistent with reported predictors of health- and vision-related quality of life in patients with uveitis.^49,50 Prior analyses suggest that visual acuity is associated with poorer quality of life, but consistent with our study, it is a weak to moderate predictor.^49-52 The low number of predictors associated with PSS-10 scores illustrates the complexity of the concept being evaluated. Psychological stress is a subjective and heterogeneous experience that can be affected by many factors, not all of which can be captured by standardized survey data.

The strengths of this study include the use of a control group, collection of an objective biometric measure of stress, and a relatively large and adequately powered sample size. These findings should be validated by additional assessments to better understand the connection between stress, immune dysregulation, and the eye.

This study has limitations. The design is cross-sectional and the questionnaire is subject to recall bias. Our primary analysis is associative—not causative—and therefore we cannot declare that stress is a risk factor for the development or recurrence of uveitis. A longitudinal study would help to better characterize the association between stress and uveitis. In addition, it is challenging to find an appropriate control for this kind of investigation. It is possible that patients could have been receiving other systemic medications, which could be a confounding factor for stress.

In addition, our study found no significant differences in measures of diurnal salivary cortisol between the uveitis and control groups. This aim required patient participation in an at-home saliva collection. Despite frequent reminders, adherence was low, with 70.8% of all patients mailing their samples and only 70 (58.3%) completing the saliva collection per protocol with analyzable samples. Moreover, per-protocol participation was evaluated by patient self-report, which likely overestimates the number of collections completed correctly. There were also mail delays affecting cortisol analysis despite use of priority postage, especially around US holiday times. Further studies are needed to investigate whether cortisol levels are different in patients with uveitis.

The findings of this study suggest that patients with uveitis exhibit higher levels of stress compared with nonuveitic controls; and the findings suggest recommending heightened awareness and attention to psychological stress in this population.

Accepted for Publication: October 9, 2018.

Corresponding Author: Nisha R. Acharya, MD, MS, Francis I. Proctor Foundation, University of California, San Francisco, Medical Sciences S-309, 513 Parnassus Ave, San Francisco, CA 94143 (nisha.acharya@ucsf.edu).

Published Online: December 6, 2018. doi:10.1001/jamaophthalmol.2018.5893

Author Contributions: Ms Berlinberg and Dr Acharya had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Berlinberg.

Study concept and design: Acharya.

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

Drafting of the manuscript: Berlinberg, Gonzales, Acharya.

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

Statistical analysis: Berlinberg, Doan, Acharya.

Obtained funding: Acharya.

Administrative, technical, or material support: Gonzales, Doan, Acharya.

Study supervision: Acharya.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Acharya reported receiving personal fees from AbbVie Inc, unrelated to this study. No other disclosures were reported.

Funding/Support: Dr Acharya is currently supported by an National Eye Institute U10 EY021125-01 grant. The University of California, San Francisco Department of Ophthalmology is supported by the National Eye Institute and Research to Prevent Blindness Foundation grants.

Role of the Funder/Sponsor: The supporting organizations 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.