The global prevalence of dry eye disease: A Bayesian view.

ABSTRACT

PURPOSE: To provide estimates for the prevalence of dry eye disease globally and in sub-groups defined by diagnostic criterion, sex, geographic location and age, using a Bayesian approach. METHODS: Modelling prevalence as a Beta distribution, estimates were inferred from Bayesian posterior distributions obtained by combing an uninformed prior with likelihood functions generated from all relevant studies reporting dry eye prevalence between 1997 and 2021. RESULTS: Global prevalence of dry eye disease was estimated at 11.59% (standard deviation (SD) = 0.04). For symptomatic disease, the estimate was 9.12% (SD = 0.04), with women 9.5% (SD = 0.05) and men 6.8% (SD = 0.06); prevalence was lowest in North America, 4.6% (SD = 0.03) and highest in Africa, 47.9% (SD = 1.8). For signs, prevalence was 35.2% (SD = 0.3), with woman 34.7% (SD = 0.7) and men 37.6% (SD = 0.7); North America showed the lowest regional prevalence, 3.5%, (SD = 0.4) with Eastern Asia the highest, 42.8% (SD = 0.4). Using TFOS DEWS II diagnostic criteria resulted in a global prevalence of 29.5% (SD = 0.8), with women 28.1% (SD = 1.2) and men 24.9% (SD = 1.4). Prevalence was lowest during the fifth decade, increasing approximately linearly with age thereafter. Estimates for other categories are given in accompanying tables. CONCLUSION: A simple, flexible, yet powerful means of combining data from multiple sources to yield prevalence estimates across a range of circumstances is described, that is compatible with published guidelines for conducting meta-analysis. Estimates can be readily updated as new information emerges, or according to need. Understanding the specific characteristics of studies chosen for inclusion is critical to the validity of the outcome. Although dry eye disease is evidently common, affecting about one in 11 people world-wide, data are sparse for the young and all geographical locations except Eastern Asia.