Study Design and Description of Database

This was a retrospective cross-sectional study using data from the 2005–2006 NHANES database. The NHANES program began in the early 1960s and was designed to assess the health and nutritional status of adults and children in the United States (US). The survey includes interviews, physical examinations, and blood draws. In 2005–2006, the NHANES performed additional examination and laboratory testing that included Humphrey matrix frequency doubling perimetry testing (N30-5 FDT), optic disc photographs, and serum IgE levels for a panel of common environmental allergens. The survey examines a nationally representative sample of about 10 000 persons per iteration and produces weighted estimates meant to be representative of the US population. Data accumulation was performed by the National Center for Health Statistics (NCHS) with approval from the NCHS Ethics Review Board (Protocol #2005-06). Data from NHANES are publicly available and anonymous, so this study received Institutional Review Board exemption from the University of California, Los Angeles.

Baseline Characteristics

Participants with optic disc photographs and FDT visual fields available in the 2005–2006 NHANES sample were included in the study. Participants with the following characteristics were excluded: (1) missing optic disc photographs or FDT visual fields; (2) age <40 years, because optic disc photographs and FDT visual fields were not collected for participants under age 40; and (3) documented alternative explanation for cup-to-disc ratio (CDR) findings (dysplastic disc or marked anisometropia) or for a visual field defect (retinal vascular disease, macular degeneration, or cerebrovascular disease). The presence of dysplastic disc, retinal vascular disease, and macular degeneration was assessed based on fundus photographs taken as part of the NHANES 2005–2006 study. The presence of anisometropia was assessed based on an objective refraction conducted as part of the NHANES 2005–2006 study. Cerebrovascular disease was based on self-report from the NHANES interview. Basic demographics that were analyzed included age and ethnicity, as both factors are known risk factors for glaucoma. Steroid use was also included as a covariate owing to its potential to be a confounder in the potential association between allergies and glaucoma. Steroid use was determined through the NHANES personal interview. Participants were asked during the household interview whether they had taken a prescription medication during the past month. Those who responded “yes” were asked to show containers of all their prescription medications to the interviewer. The interviewer entered the name of each prescription medication into a computer. For medications where no container was available, the interviewer asked the participant to verbally report the name of the medication. Steroid use was coded as a binary variable of steroid use vs no steroid use. Participants who reported use of any oral, ophthalmic, topical, inhaled, or injected steroids were counted as using steroids, and the remainder of participants were counted as not using steroids. Glaucoma medication use was not included as a covariate because it was presumed that only participants with glaucoma would be taking glaucoma medications, and that comparisons of glaucoma medication use in participants with vs without glaucoma would thus not be reliable.

The exposures of interest were total serum IgE levels and IgE levels specific to a panel of indoor allergens. Indoor allergens were chosen based on a nationally representative sample of household dust. Analysis of serum samples was conducted using the Pharmacia Diagnostics ImmunoCAP 1000 System. Total serum IgE was measured by the extent of its reaction to anti-IgE covalently coupled to the ImmunoCap reaction vessel, after washing, incubating with a developing agent, and comparison of the resulting eluate fluorescence with the response of calibrators. Allergen-specific IgE was measured by its reaction to specific allergens covalently bound to the reaction vessel. The IgE antibodies to specific indoor allergens that were tested included: Dermatophagoides farinae (American dust mite), Dermatophagoides pteronyssinus (European dust mite), cockroach, cat, dog, mouse, and rat. For participants where allergen-specific IgE measurements exceeded the total IgE value owing to cross-reactivity or conformationally equivalent epitopes of antigens for the allergens, total IgE measurements were coded as missing. For allergen-specific IgE levels, participants were considered negative if their serum IgE levels were <0.35 kilounits per liter (kU/L), while levels ≥0.35 kU/L were considered positive. For total IgE levels, values ≤119 kU/L were considered negative, while those >119 kU/L were considered positive. Cutoff values for allergen-specific and total IgE levels were based on the standard lower limit of detection for the laboratory assay used by NHANES. Quality control and quality assurance protocols met 1988 Clinical Laboratory Improvement Act mandates.

Outcome

The primary outcome of interest was the prevalence of glaucoma as defined by the Rotterdam criteria, which defines open-angle glaucoma based on optic nerve appearance and visual field defects. Within NHANES, optic nerve appearance was assessed based on optic nerve photographs, and glaucomatous visual field defects were assessed by FDT visual fields. A clinical diagnosis of glaucoma was defined as having 2 or more abnormal points in at least 1 eye on the N30-5 FDT on 2 tests in the same eye, along with CDR in 1 eye or CDR asymmetry between eyes ≥97.5% of the normal NHANES population. The Rotterdam criteria were chosen to assess glaucoma because of the ability to produce an objective clinical definition of glaucoma with data available within NHANES based on these criteria, as reported by a recent study. Information on the incidence and subtype of glaucoma was not available in the 2005–2006 NHANES database.

The FDT examination was conducted in the dark by trained health technicians. A practice test was run prior to the actual test to determine whether the participant understood the procedure. The test always started with the right eye, followed by the left eye; and then the test was repeated for each eye. Nineteen visual field locations were tested and each location was tested until the participant responded. During each test, reliability checks were performed at random times, which included 3 false-positive tests and 3 blind spot tests. Participants were then classified as having normal, positive, insufficient, or unreliable FDT examinations; those with insufficient or unreliable examinations were excluded from the study population. The CDR was determined by using retinal images. The examination consisted of 2 45-degree nonmydriatic digital images of the retina taken by trained technicians. The first image was centered on the macula and the second on the optic nerve. Trained graders at the University of Wisconsin assessed vertical CDR in each eye using digital images that were evaluated as they initially appeared on the back of the digital camera. The images were reviewed by at least 2 of 9 trained graders. If the graders did not agree, a third grader reviewed the image. If there was a disagreement, an adjudicator made the final decision.

Statistical Analysis

Sampling within NHANES occurs in 4 stages, starting with primary sampling units (PSUs) and followed by segments within PSUs (usually city blocks), households, and individuals. Based on this multistage probability sampling design, data are weighted and adjusted for nonresponse to produce weighted estimates meant to be representative of the US population. All data analyses in this study were based on weighted estimates using sample weights provided by NHANES.

Descriptive statistics were used to describe baseline characteristics for the study population. Age was analyzed as a continuous variable, and ethnicity and steroid use were analyzed as categorical variables. Total and allergen-specific IgE were analyzed as categorical variables, where participants with serum levels above the described cutoff values were considered positive and participants with serum levels below the cutoff values were considered negative. Glaucoma was assessed as a binary outcome based on the Rotterdam criteria.

Logistic regression modeling was used to examine the association between serum IgE levels and glaucoma. Separate unadjusted models were created for each allergen-specific IgE subtype as predictors of glaucoma. Partially adjusted models were then created including age, ethnicity, and steroid use as potential confounders for each IgE subtype and glaucoma. Because of the concern for cross-reactivity and epitope similarities, a final analysis was performed including all allergen-specific IgE subtypes, total IgE, and sensitization to any indoor allergen as predictors for glaucoma in the same model. All analyses were conducted with SAS version 9.3 (SAS, Cary, North Carolina, USA).