The United States ranks only 25th worldwide in terms of per capita coffee consumption,
17 yet in this US-based prospective study, caffeinated coffee consumption was associated with an increased risk of EG/EGS. This trend did not robustly extend to caffeinated product consumption generally. The association between caffeine consumption was modified by a family history of glaucoma, where the increased risk with greater caffeine or caffeinated coffee consumption was stronger among those who might have been more genetically susceptible to developing EG/EGS. Because this is the first prospective study to evaluate long-term caffeine and coffee intake and risk of EG/EGS, our results, particularly for the secondary analyses, must be interpreted cautiously and confirmed with future analyses.
Compared with the null associations with baseline caffeinated coffee consumption or the most recent consumption, the positive association between long-term average consumption of caffeinated coffee in relation to EG/EGS support the hypothesis that habitual caffeinated coffee at ≥3 cups/day may contribute to the gradual ocular accumulation of exfoliation material. The Hcy-elevating effect of coffee consumption represents a biologically plausible link between coffee consumption and ES. After acute ingestion of filtered caffeinated coffee, there is an increase in plasma Hcy,
18–20 and polyphenols such as chlorogenic acid, a compound abundant in coffee but not found in other foods, may contribute to this effect.
33,34 The Hordaland homocysteine study,
53 a population-based study of 11,940 healthy subjects residing in Norway, found that smoking, coffee consumption, and low folate levels are associated with higher serum Hcy levels. Several studies show that Hcy levels are elevated in blood, aqueous humor, and tears taken from ES patients compared to those in controls.
21–28 Elevated Hcy levels, in turn, may contribute to vascular leakage and dysfunctional extracellular matrix remodeling in ES.
54 Alternatively, elevated Hcy levels are a consequence and not a cause of ES, and some unlinked factor(s) related to heavy coffee consumption may contribute to EG/EGS.
The strength of this study is the prospective design where intake of caffeine and coffee was assessed before disease occurrence, making recall bias highly unlikely. This was a large study with 360 incident cases, among 78,977 women and 41,202 men followed for up to 26 years, with high follow-up rates. Other strengths of our study include repeated dietary and lifestyle risk factor assessment during follow-up. In using detailed questions on extensively validated FFQs, we were able to incorporate caffeine intake from various sources and examine the association with various caffeinated products. Finally, we were able to control for numerous updated covariates, minimizing confounding bias.
Some limitations of our study must be considered. EG/EGS cases were ascertained based on nonstandardized eye examinations performed by eye care providers from throughout the United States. Nonetheless, using this approach, which was needed for practical reasons, we previously confirmed that incident EG/EGS is a strongly age-related condition that produces higher IOPs at diagnosis than incident POAG
14 ; furthermore, we showed that female sex is a risk factor for EG/EGS,
14 as observed in the Reykjavik Eye Study, indicating construct validity of the outcome measure.
13 Another limitation is that our study population is over 90% Caucasian, and thus, our results may not be generalizable to other more diverse populations. Our study was not designed to capture people with only ES and no signs of glaucoma, and we certainly had under-ascertainment of EG/EGS; yet, our objective was not to ascertain the absolute incidence of EG/EGS (where high sensitivity is important and under-ascertainment may be a detriment); instead, the goal was to ascertain the relative incidence of EG/EGS to identify potential risk factors. In this setting, a low sensitivity of disease identification in epidemiologic studies is acceptable if the disease outcome definition has a high specificity and the under-ascertainment is not systematically different by exposure groups. Indeed, the frequency of disease confirmation among self-reports and the number of reported eye examinations during follow-up were similar by caffeine intake, minimizing the possibility of bias. Finally, it is possible that residual confounding by other factors associated with high coffee consumption may account for these results.
In summary, in this first prospective population-based investigation of the relationship between caffeine and coffee intake and risk of developing EG/EGS, we found that heavier caffeinated coffee consumption was associated with increased risk of EG/EGS. The effect modification by family history on the association with caffeine deserves further study.