Dry eye disease (DED) is a common inflammatory condition that interferes with quality of life through ocular pain and irritative symptoms.¹ Patients with DED are more likely to report trouble with daily tasks such as reading and driving, which can affect general and emotional well-being.^2,3 Depression is more prevalent among patients with DED, although the underlying mechanisms are not well understood.⁴ Patients with DED may be more at risk for depression involving pathways through decreased quality of life, shared genetic factors, changes to central pain processing, and anticholinergic effects of antidepressant medications.^4,5 Inflammation has also been implicated in the pathogenesis of both DED⁶ and depression,⁷ but little is known about whether the same inflammatory processes affect both conditions. Given that DED and depression are both prevalent conditions that affect daily well-being, studies are needed to better understand their association.

Purpose: To describe and demonstrate methods for analyzing longitudinal correlated eye data with a continuous outcome measure. Methods: We described fixed effects, mixed effects and generalized estimating equations (GEE) models, applied them to data from the Complications of Age-Related Macular Degeneration Prevention Trial (CAPT) and the Age-Related Eye Disease Study (AREDS). In CAPT (N = 1052), we assessed the effect of eye-specific laser treatment on change in visual acuity (VA). In the AREDS study, we evaluated effects of systemic supplement treatment among 1463 participants with AMD category 3. Results: In CAPT, the inter-eye correlations (0.33 to 0.53) and longitudinal correlations (0.31 to 0.88) varied. There was a small treatment effect on VA change (approximately one letter) at 24 months for all three models (p = .009 to 0.02). Model fit was better with the mixed effects model than the fixed effects model (p < .001). In AREDS, there was no significant treatment effect in all models (p > .55). Current smokers had a significantly greater VA decline than non-current smokers in the fixed effects model (p = .04) and the mixed effects model with random intercept (p = .0003), but marginally significant in the mixed effects model with random intercept and slope (p = .08), and GEE models (p = .054 to 0.07). The model fit was better with the fixed effects model than the mixed effects model (p < .0001). Conclusion: Longitudinal models using the eye as the unit of analysis can be implemented using available statistical software to account for both inter-eye and longitudinal correlations. Goodness-of-fit statistics may guide the selection of the most appropriate model.

Purpose: To describe and demonstrate appropriate statistical approaches for estimating sensitivity, specificity, predictive values and their 95% confidence intervals (95% CI) for correlated eye data. Methods: We described generalized estimating equations (GEE) and cluster bootstrap to account for inter-eye correlation and applied them for analyzing the data from a clinical study of telemedicine for the detection of retinopathy of prematurity (ROP). Results: Among 100 infants (200 eyes) selected for analysis, 20 infants had referral-warranted ROP (RW-ROP) in both eyes and 9 infants with RW-ROP only in one eye based on clinical eye examination. In the per-eye analysis that included both eyes of an infant, the image evaluation for RW-ROP had sensitivity of 83.7% and specificity of 86.8%. The 95% CI's from the naïve approach that ignored the inter-eye correlation were narrower than those of the GEE approach and cluster bootstrap for both sensitivity (width of 95% CI: 22.4% vs. 23.2% vs. 23.9%) and specificity (11.4% vs. 12.5% vs. 11.6%). The 95% CIs for sensitivity and specificity calculated from left eyes and right eyes separately were wider (35.2% and 30.8% respectively for sensitivity, 25.4% and 17.3% respectively for specificity). Conclusions: When an ocular test is performed in both eyes of some or all of the study subjects, the statistical analyses are best performed at the eye-level and account for the inter-eye correlation by using either the GEE or cluster bootstrap. Ignoring the inter-eye correlation results in 95% CIs that are inappropriately narrow and analyzing data from two eyes separately are not efficient.

The aim of this study is to evaluate the current practice of statistical analysis of eye data in clinical science papers published in British Journal of Ophthalmology (BJO) and to determine whether the practice of statistical analysis has improved in the past two decades. All clinical science papers (n=125) published in BJO in January–June 2017 were reviewed for their statistical analysis approaches for analysing primary ocular measure. We compared our findings to the results from a previous paper that reviewed BJO papers in 1995. Of 112 papers eligible for analysis, half of the studies analysed the data at an individual level because of the nature of observation, 16 (14%) studies analysed data from one eye only, 36 (32%) studies analysed data from both eyes at ocular level, one study (1%) analysed the overall summary of ocular finding per individual and three (3%) studies used the paired comparison. Among studies with data available from both eyes, 50 (89%) of 56 papers in 2017 did not analyse data from both eyes or ignored the intereye correlation, as compared with in 60 (90%) of 67 papers in 1995 (P=0.96). Among studies that analysed data from both eyes at an ocular level, 33 (92%) of 36 studies completely ignored the intereye correlation in 2017, as compared with in 16 (89%) of 18 studies in 1995 (P=0.40). A majority of studies did not analyse the data properly when data from both eyes were available. The practice of statistical analysis did not improve in the past two decades. Collaborative efforts should be made in the vision research community to improve the practice of statistical analysis for ocular data.

Objective: To evaluate perinatal risk factors for retinopathy of prematurity (ROP), in a large, broad-risk cohort of premature infants. Study design: Secondary analysis of data from the Postnatal Growth and ROP (G-ROP) Study, a retrospective cohort study of infants undergoing ROP examinations at 29 North American hospitals in 2006–2012. Results: Among 7483 infants, 3224 (43.1%) had any ROP and 931 (12.4%) had severe ROP (Type 1 or 2 ROP). In multivariable logistic regression analysis, significant risk factors for any ROP were lower birth weight (BW, odds ratio (OR) = 5.2, 1250 g), younger gestational age (GA, OR = 32, 29 weeks), 1-min Apgar score <4 (OR = 1.2), race (OR = 1.6, White vs. Black), outborn (OR = 1.5), and delivery room intubation (OR = 1.3); and for severe ROP were lower BW (OR = 20, 1250 g), younger GA (OR = 30, 29 weeks), male (OR = 1.5), Hispanic ethnicity (OR = 1.8), race (OR = 1.6, White vs. Black), outborn (OR = 1.6), and delivery room intubation (OR = 1.6). Together, these factors predicted well for any ROP (area under ROC curve (AUC) = 0.87) and severe ROP (AUC = 0.89), but BW and GA were the dominant factors for ROP (AUC = 0.86) and severe ROP (AUC = 0.88). Conclusions: Based on the largest report to date with detailed ROP data from infants meeting current screening guidelines, ROP risk is predominantly determined by the degree of prematurity at birth, with other perinatal factors contributing minimally.