The NHS started in 1976 with the enrollment of 121,700 US registered female nurses aged 30 to 55 years, who replied to a mailed health questionnaire.40
The HPFS commenced in 1986 when 51,529 male health professionals agreed to join a similarly designed prospective cohort analysis.41
Participants were monitored using biennial questionnaires covering health status, lifestyle tendencies, and dietary habits. Follow-up rates were high (>85% of the total possible person-time). The Human Research Committees of Brigham and Women's Hospital and Massachusetts Eye and Ear Infirmary approved this study; furthermore, the research adhered to the Declaration of Helsinki tenets.
The NHS study period began in 1980 and the HPFS in 1986 (when we began to assess dietary habits) and lasted until 2008. A participant began to contribute person-time in 2-year increments after age 40 (as glaucoma risk increases after this age) and only during periods in which they indicated having had an eye examination. Participants contributed person-time from the return date of the first questionnaire until the earliest occurrence of either a report of glaucoma, cataract extraction (because it can be difficult to detect exfoliation material after this surgery), cancer (because this diagnosis can profoundly effect lifestyle behaviors), death, loss to follow-up, or 2008.
Participants were excluded from the study at baseline for the following reasons: (1) they did not complete the 1980 or 1986 semiquantitative food frequency questionnaire (SFFQ) (23,239 women; 0 men); (2) the SFFQ dietary data provided seemed implausible (plausible dietary information consisted of >50 of 61 items completed, yielding 500–3500 kcal/day; 5994 women; 1596 men); (3) a cancer diagnosis (excluding nonmelanoma skin cancer) preceded glaucoma (3624 women; 1033 men); (4) a prevalent diagnosis of glaucoma or GS was provided (3624 women; 1033 men); (5) we were unable to contact the participant within 2 years after baseline (740 women; 175 men); (6) an ophthalmic examination was never documented (5804 women; 4282 men); and (7) a previous cataract extraction occurred (173 women; 672 men). In the 1980-to-1982 risk period for NHS and the 1986-to-88 risk period for HPFS, 44,771 women and 29,033 men were eligible for the cohort at risk for glaucoma. At later periods, participants meeting the age and eye examination requirements were allowed to join the cohort. By 2008, a total of 78,977 women and 41,202 men contributed person-time to the cohort at risk for glaucoma.
From participants with self-reported glaucoma, we obtained permission to retrieve medical information to confirm the diagnosis. We sent the diagnosing eye care provider a request to complete a glaucoma questionnaire about maximum IOP, optic nerve features, status of filtration apparatus, and presence of exfoliation material or other secondary causes for elevated IOP. In lieu of completing the questionnaire, eye care providers could send complete medical records. We also requested all visual field (VF) reports supporting the glaucoma diagnosis. A glaucoma specialist (LRP) evaluated the questionnaire and medical record information as well as the VF data in a standardized manner masked to dietary data.
In our study, we could not test the hypothesis of an association between caffeine consumption and ES; however, we evaluated EG/EGS as an outcome, which is an outcome of public health importance. We defined EG/EGS as the presence of ES as documented in the records and at least one of the following occurring in the eye demonstrating exfoliation material on slit lamp examination: (1) a history of IOP >21 mm Hg, (2) a cup:disc ratio of ≥0.6, or (3) VF loss consistent with glaucoma on at least one reliable test. Patients with ES who did not meet these criteria were excluded from analysis because our case identification method was initiated with a questionnaire regarding receiving a glaucoma diagnosis and not on the basis of reporting specific ophthalmic signs.
There were 7330 reports in women and 3089 reports in men of glaucoma diagnoses occurring during the study period. The eye care providers confirmed this diagnosis in 66% of women and 56% of men, as follows: POAG with VF loss (27% in women and 27% in men), only elevated IOP or optic disc cupping (20% in women and 17% in men), and other types of glaucoma or GS (14% in women and 10% in men), and EG/EGS (5% in women and 2% in men). The remaining self-reports (34% in women and 44% of men) could not be confirmed, as the participants themselves (6% in women and 13% of men) or their eye care providers (4% in women and 5% in men) could not be contacted, participants did not give permission to review their records (12% in women and 10% in men), participants indicated the initial report was in error (10% in women and 15% in men), or participants' eye care providers refuted the diagnosis of glaucoma (2% in women and 1% in men).
Of the 330 confirmed EG/EGS cases in women and 74 cases in men, 30 women and 14 men were excluded from the analysis, as EG/EGS occurred among those with previous cancer or cataract extraction, which were censoring criteria; thus, we included 300 women and 60 men who met the criteria for incident EG/EGS.
Measurement of Caffeine Consumption
We collected dietary intake data repeatedly (in 1980, 1984, and 1986 and every 4 years thereafter in the NHS and in 1986 and every 4 years thereafter in the HPFS) during the study period, using the SFFQ. The 1980 SFFQ included 61 food and beverage items; the 1984 SFFQ was expanded to 116 items, and similar versions of it were used from 1986 (126 items) onward in the NHS and HPFS (131 items).
In the FFQ, participants indicated their average intake of a serving of food or beverage over the preceding year. In the NHS, the 1980 FFQ asked about consumption of caffeinated coffee (in cups), caffeinated tea (in cups), and chocolate (in 1-ounce servings). From 1984 in the NHS and from 1986 in the HPFS, the FFQ was expanded to include intake of decaffeinated coffee (in cups) and separate items for caffeinated soda and caffeine-free sodas. The questionnaire provided nine response possibilities for intake frequency for each item ranging from “never or less than once per month” to “6 or more times per day.” We used data from the US Department of Agriculture to convert participants' responses into average daily intake of caffeine in mg/day. We assumed 137 mg of caffeine per cup of coffee, 47 mg of caffeine per cup of tea, 46 mg of caffeine per can or bottle of cola beverage, and 7 mg of caffeine per serving of chocolate.
Validity of Food Frequency Questionnaire Assessment of Caffeine
The reproducibility and validity of the FFQ have been reported previously.41,42
Self-reported intake of caffeinated beverages (cups/day) according to the FFQ was highly correlated with intake from diet records over 4 weeks: the correlations were 0.78 for coffee, 0.93 for tea, and 0.84 for cola drinks in the NHS,43
and in HPFS, the correlations were 0.93 for coffee, 0.77 for tea, and 0.84 for cola drinks.44
We calculated cumulatively updated caffeine intake by averaging the intake from all available dietary assessments up to the start of each 2-year period at risk. As glaucoma is a slowly developing chronic condition, we chose to study cumulatively averaged caffeine intakes as they best represented long-term intake, and average measures have less measurement error than single assessments.45
All caffeine intakes at each questionnaire were adjusted for total energy intake by using the residual method.46
Next, we examined caffeinated beverages and chocolate in association with EG/EGS. Such analyses are useful as products other than caffeine present in caffeinated coffee may contribute to ES. Also, individuals vary their caffeine intake predominately by altering their intake of caffeinated beverages, and thus the net effect of caffeinated beverages on the risk of EG/EGS must be evaluated for possible public health recommendations. Thus, we examined the risk of EG/EGS in relation to categories of cumulatively updated intakes of specific beverages and chocolate, such as caffeinated coffee, tea, caffeinated soda, chocolate, and decaffeinated coffee.
We calculated incidence rates of EG/EGS by dividing the incident cases by person-years accrued for each caffeine or beverage intake category. We adjusted for age using 5-year categories and calculated Mantel-Haenszel age-adjusted incidence rate ratios (RR) and their 95% confidence intervals (CI). For multivariable analyses, we controlled for potential glaucoma risk factors by including them simultaneously in Cox proportional hazards analysis stratified by age in months and the specific 2-year period at risk.47
We conducted tests for trend by including the midpoint values within each intake category.
We first analyzed data from each cohort separately and performed tests for heterogeneity of the cohort specific results to check for appropriateness of pooling the results. Then, we pooled the results using meta-analytic methods incorporating random effects.48
In our multivariable models, we adjusted for the following covariates: (1) established risk factors such as family history of glaucoma (self-report of any glaucoma in biologic parents, siblings or children), major ancestry (Scandinavian Caucasian, southern European Caucasian, other Caucasian, other ancestry); (2) history of cardiovascular disease (myocardial infarction) and other cardiovascular risk factors49,50
(i.e., body mass index in kg/m2
, smoking [pack-years], cumulatively updated alcohol intake [g/day], total caloric intake [kcal/day], self-report [yes/no] of hypertension, and diabetes, high cholesterol); and (3) most recently reported continental US residence stratified by latitude (northern tier, above 42°N; middle tier, 37°–42°N; southern tier, below 37°N), which we previously found was a strong risk factor for EG/EGS.14
Updated information for covariates was obtained from the biennial questionnaires; cumulatively updated alcohol and total caloric intake was calculated using responses to the FFQs. Data missing for the covariates were handled with missing indicator variables in models.
Secondary Analyses and Effect Modification
First, we examined the influence of timing of exposure by investigating intake only at baseline or at the most recent questionnaire.
In subsequent analyses, we additionally adjusted for cumulatively updated total fluid intake based on intake of nearly 30 different types of beverages. Because drinking a large quantity of fluids, particularly in a short period of time, generally causes IOP elevation,51
we conducted this analysis to determine whether association with caffeine intake depended on total fluid intake. We also determined whether the association between caffeine/coffee intake and EG/EGS was confounded by intake of B vitamins such as folate and vitamins B12 and B6, which may also affect Hcy levels.52
Finally, we examined the associations between caffeine intake and EG/EGS separately among those with and without a self-reported family history of glaucoma. The questions pertaining to family history of glaucoma were first asked of all participants in the 2000 follow-up questionnaire. In stratified analyses described above, we have statistically tested for effect modifications by testing the significance of interaction terms in Cox regression models.