Methods

Selection of population

The selection and recruitment of the population has been described in detail elsewhere.13 In summary, Proyecto VER is a population-based survey of non-institutionalised Hispanics age ⩾40 years living in Pima and Santa Cruz counties, Arizona, USA. Based on the 1990 census, a random sample of block groups in two strata, Nogales (Santa Cruz) and Tucson (Pima), that contained at least 5% Hispanic residents age ⩾40 years were selected. The probability of selection within the strata was proportional to the size of the proportion of the Hispanic population in the block group. Every household was mapped, and listed for each of the block groups chosen. Households were randomly selected for enrolment as described elsewhere.13 Eligibility criteria for members in selected households included self-described Hispanic, age ⩾40 years. US citizenship was not a requirement (and was not asked about), but the person must have resided at the address for at least 6 months. All household members found to be eligible were potential participants.

Procedures

Once eligible, household members were recruited, and written (in Spanish and English) informed consent was obtained to conduct an interview and conduct a clinical examination at a central clinical site. Those eligible persons who refused to participate were asked to respond to a short questionnaire, which included questions on self-perception of vision. The data from the short questionnaire were used to adjust prevalence rates for differential non-response, and to determine potential biases in associations.

The majority of home interviews (80%) were conducted in Spanish by trained, bi-lingual interviewers. All instruments were developed in English, translated into Spanish, then back-translated to ensure that content and meaning were preserved. Detailed questions on family history of Native American ancestry, education, family income, insurance status, use of health services and eye services, and whether a physician had diagnosed diabetes were asked. Questions on acculturation developed for the Hispanic Health and Nutrition Examination Survey (HHANES) were also asked, and a scale developed from the answers.

At the clinic site, blood was drawn for the determination of haemoglobin A1c (HbA1c). All blood was stored in a refrigerator for no more than 5 days and shipped on ice to the reference laboratory (Dr M Steffes, University of Minnesota). HbA1c was determined using HPLC methods and standard controls, as described elsewhere.14 An affirmative answer to the question of physician-diagnosed diabetes, or an HbA1c value of ⩾7.0% was defined as definite diabetes.

Pterygium was diagnosed at the clinic ophthalmological examination, where one ophthalmologist carried out more than 99% of all examinations. The following definition was used: a fleshy triangular fold of tissue that grows from the conjunctiva, and must have evidence encroachment on the cornea; it is usually on the nasal side, and may be bilateral. The location of the pterygium, right or left eye, was noted. If the patient requested surgery, surgery was arranged by the ophthalmologist. The surgeon obtained permission to send tissue from excised pterygium for analyses of presence of human papillomavirus (HPV) using standard testing.

Blood pressure was measured according to a standardised protocol.15 Hypertension was defined as either positive answer to the question of being told one has hypertension and currently under treatment, or systolic pressure >160 mmHg, or diastolic blood pressure >90 mmHg.

Statistical analyses

Prevalences of pterygium are shown by age and sex. Chi square analyses were used to determine differences in these univariate distributions. Multivariate models, using GEE models for adjustment of the correlation between eyes, were created to examine the joint contribution of risk factors, adjusting for age and sex, to the risk of pterygium.

Results

The 4774 persons participating in this study were largely Mexican-Americans, with 65% indicating they were born in Mexico. Similar to the findings in the US census, this community is economically less well off than the non-Hispanic white populations: 66% of the participants reported an average annual income of less than $20,000. Only 35% had completed high school.

The prevalence rate of pterygium in at least one eye, by age and sex, is shown in table 1. At each age group, men had a prevalence of pterygium twice that of women. Pterygium was equally common in left as in right eyes (10.3% versus 11.8%, respectively). Of the 16% who had pterygium, 36% had bilateral disease. This prevalence was as high as has been reported for those in populations with high occupational sun exposure (the Maryland Watermen1). The prevalence is higher than has been reported in other populations in more tropical zones, such as Singapore5 (table 2).

In addition to sex, several risk factors for pterygium were investigated (table 3). As has been found elsewhere, current smoking appeared to be protective for pterygium. Diabetes was also protective. The presence of hypertension and alcohol use was not associated with the presence of pterygium. The presence of cortical cataract was associated with pterygium, but only in men.

We investigated the association of pterygium with history of at least one grandparent (or closer) having Native American ancestry, but there was no association. Those with low education, and low income, were at increased risk of pterygium. Those who had bilateral cataract surgery were less likely to have pterygium.

In a multivariate model, adjusting for all factors, male sex, low education and low income remained statistically significantly related to pterygium (table 4). Those who smoked, and those with diabetes, were significantly less likely to have pterygium. None of the ten samples tested were positive for human papillomavirus.

Discussion

The prevalence of pterygium was high in this population of Latinos living in Arizona. The overall rate in men was similar to that observed in a Barbadian black population (23%).6 As has been reported elsewhere, there was a higher risk among men.357 However, studies from Indonesia4 and Barbados6 did not find women at lower risk. Female sex may be a marker for lower occupational status or recreational exposure to sunlight. In previous work, we found that women had significantly lower ocular sunlight exposure than men.16 However, McCarty et al found an increased risk for pterygium among men even adjusting for UV-B exposure.7 Similarly, occupational exposures did not explain the male preponderance in the Tanjong Pagar study, suggesting other, unknown factors may contribute to the male excess (or protect females).5

Smoking was protective for pterygium in this study. Smokers were less likely to have pterygium in other studies as well,467 although it was a risk factor for pterygium among Chinese people.5 The mechanism by which smoking might be protective is unclear, given the consistent finding of cigarette smoking in promoting carcinogenesis, and in upregulating cytokines and proteins responsible for cell proliferation and migration. However, other studies have found cigarette smoking to have protective effects in other diseases, such as uterine fibroids and endometriosis.17 However, the biological rationale for these findings is the anti-oestrogenic effect of smoking, an effect that is hard to link to a protective effect for pterygium. A more likely hypothesis may be the effect of smoking in inducing immune suppression, thought to explain the beneficial effect of smoking on ulcerative colitis and other immunologically mediated disorders.17 The relative consistency of the protective effect of smoking and pterygium deserves further investigation.

Several markers of socioeconomic status were associated with pterygium in the Latino population, including low income and less education. Low education was been identified as a risk factor in other studies as well,56 which again might point to occupations where sunlight exposure is more common. In the study of Singaporean Chinese people, Wong et al found that once occupation was entered into the model, the effect of education was no longer significant.5 We did not collect specific occupational data, nor detailed sun exposure, in our Latino population and cannot confirm these associations. However, more than one-half of our Latino population had less than high school education and were currently earning less than $20 000 per year, suggesting jobs in the labour/agricultural sector of Arizona.

Both hypertension and diabetes were very prevalent in the Latino population, as was pterygium. Therefore we explored the relationship of pterygium to these co-morbid conditions. There was a slight protective effect of diabetes. Such an association is counter-intuitive, as both pterygium and diabetes have been associated with pro-inflammatory cytokines. Notably, investigators have reported increased levels of vascular endothelial growth factor (VEGF) in pterygium epithelium and vascular endothelium,181920 although the upregulation could also be explained by exposure to UV-B.12 Alternatively, persons with diabetes may be more likely to have indoor jobs and less sun exposure, or the finding may be spurious. We found no association between between hypertension and prevalence of pterygium.

A major limitation of our study is the absence of detailed ocular sun exposure history, which would provide further evidence to support ocular UV-B as the primary risk factor for the high rates of pterygium in our population. Other risk factors, such as education and income, suggest sun exposure as a possible explanation, but direct measurements would be preferable. We also recognise the difficulty in differentiating small pterygia from pingueculae, such that some misclassification might have occurred. The main strength of our study is the large population-based sample of Latinos, with all eye examinations carried out by one experienced ophthalmologist.

In summary, this population-base study of Latinos in southern Arizona has provided estimates of a high prevalence of pterygium, especially in men. Further research on the effects of smoking in this population is of interest.