Sampling procedure
The Botucatu Eye Study (BES) was a population-based, cross-sectional ophthalmic survey of people living in households in the city of Botucatu, Brazil. Botucatu is a municipality in the state of São Paulo in Brazil, located 225 km from São Paulo, the state capital and 898 km from Brasília. It lies at 22°53'09"S, 48°26'42"W, has a temperate climate, and covers a total area of 1486.4 km2. The population in 2001 was 108,306 with 70% of people having less than 10 years of formal education and manufacturing being the most common mode of employment. The principal aims of this project were to estimate the prevalence and causes of visual impairment and the prevalence and risk factors of ocular disorders among all age groups of people in this region.
The population census data (Instituto Brasileiro de Geografia e Estatsitica (IBGE), 2001) for the Botucatu census sectors was used as the sampling frame. The eligible population consisted of permanent, non-institutionalized residents of Botucatu over the age of 1 year.
A sample size calculation was then made based on the population size of Botucatu and the assumption that the prevalence of visual impairment (as defined below) in Botucatu was thought to be approximately 2% [2]. In order to determine this level of visual impairment with a precision of 0.5% and level of significance of 5%, it was calculated that 2931 participants would be required for the study. It was determined that the average number of household residents was 3.3. Sampling 1 000 households would therefore yield and estimated eligible population of 3 300.
The IBGE divided the city into census sectors and households were identified within each block and enumerated in sequence to cover the entire urban area. The households were to be examined were selected using a random start point with a household cluster sampling technique. Households were identified systematically according to local census data: the first house was selected randomly; the next house was the seventh house on the even-numbered side of the street and so on, successively until the pre-determined number of households in each census sector had been reached. The number of households in each sector was determined based on the proportion of people in that sector compared to the total Botacatu population.
Each of the selected households received a letter of invitation to participate in the study, which was followed by a field worker visit. Those who agreed to participate were contacted by telephone to schedule an appointment at the Botucatu School of Medicine University Hospital, where all of the examinations were conducted. All persons of the household were eligible to participate in the study if it was their usual place of residence (by self-report) and they had attained the age of 1 year. If there was no answer when the examiners contacted the household or if people refused to participate in the research, the first house to the right was selected. If the next household refused to participate, the first house to the left of the initial house was selected, and so on, successively. After inviting 1,000 households to participate in the survey (approximately 3 300 people), 2 485 individuals or 75.3% of the potential sample participated. The demographics of the examined population were compared to those of the entire Botacatu population at the time of the census.
Data Collection
The study was carried out over a 4 month period by a single survey team consisting of 5 members, including 4 ophthalmologists. All study personnel underwent training and all procedures were standardized prior to commencement. Specific observations were performed by 1–2 members of the team in order to minimize interobserver variability. A medical and ophthalmic history was obtained from each patient in Portuguese by qualified health care workers. This included the collection of demographic, ocular and general health details.
Each participant then received a comprehensive vision and eye examination where uncorrected visual acuity (VA) was measured for the right eye followed by the left with a consistently illuminated illiterate E Snellen chart at 5 meters. The VA was then retested with the patients existing refraction. If the corrected VA was less than 20/20 an objective refraction using a streak retinoscope, trial frames and lenses was performed by the same two ophthalmologists (SAS, EH). This was subjectively refined using the Jackson cross-cylinder technique and the best corrected visual acuity (BCVA) recorded using the result of this refraction. If the subject was unable to read the largest letter at 5 m with the subjective refraction, testing was repeated at 1 m. If they were unable to read the largest letter at 1 m then the VA was recorded as count fingers (CF), hand movements (HM), light perception (LP) or no light perception (NLP). The spherical equivalent (SE) was calculated as the spherical error plus half the cylindrical error. The intraocular pressure (IOP) was measured using a noncontact pneumotonometer (CT-60 computerised tonometer, Topcon, Tokyo, Japan) and the mean of 3 measurements was recorded. If the IOP was higher than 25 mmHg then the measurements were repeated. Where possible, children under the age of five years had their visual acuity measured with the same chart; pre-verbal children were able to indicate the direction of the optotype. In those too young to do this a pupil examination, dilated fundus examination and cycloplegic refraction was able to exclude all but cortical causes of visual impairment or blindness. The intraocular pressure was not measured in children under 5 years unless clinically indicated.
Slit lamp biomicroscopy (BQ-900, Haag Streit, Bern, Switzerland) was performed, followed by dilatation with cyclopentolate 1%, 1 drop every 5 minutes (total of 3 drops for each eye) for those under the age of 13 years, cyclopentolate 1% 1 drop in each eye for participants aged 14–39 years, and tropicamide 1% and phenylephrine 10% for dilatation of those participants 40 years and over. The subjects then underwent fundus examination utilizing the slit lamp with a 90D Volk lens and then using a Schepens indirect binocular ophthalmoscope with a 20D Volk lens.
If best corrected VA was < 20/40 a determination as to the major cause of visual impairment in the best eye was made by the examining ophthalmologist at the end of the examination. Cataract was determined if the level of lens opacity correlated with the visual acuity, glaucoma was diagnosed based on intraocular pressure and optic nerve head appearance, and refractive error was the cause if the vision corrected to better than 20/30 with spectacles. If the etiology of visual impairment was not apparent the subject was scheduled for further investigations at the Botucatu School of Medicine University Hospital, after which a consensus diagnosis for the major cause of visual impairment was reached by two ophthalmologists.
Definitions of low vision and blindness
Visual impairment was defined according to the following World Health Organization (WHO) categories [6]. Presenting blindness was defined as unaided VA (or with spectacles if worn) < 20/400 (3/60) in the better eye; and WHO blindness was defined as best corrected VA < 20/400 (3/60) in the better eye. Presenting low vision was defined as VA < 20/60 (6/18) but ≥ 3/60 (20/400) in the better eye, unaided (or with spectacles if worn). WHO low vision was defined as best corrected VA < 20/60 (6/18) but ≥ 20/400 (3/60) in the better eye.Presenting visual impairment was defined as the combined set of presenting low vision and presenting blindness, and WHO visual impairment was defined as the combined set of WHO low vision and WHO blindness. Field defects were not taken into consideration. Vision after subjective refraction was considered best-corrected vision for the purposes of the study.
Ethics
The BES was approved by the Institutional Review Board and Research Ethics Committee of the Botucatu School of Medicine. Informed consent was obtained from participants prior to the commencement of the study. Where children were under the age of 16 years, their parents gave consent on their behalf. The study was then conducted in accordance with the tenets of the Declaration of Helsinki. The treatment of conditions found during the course of the study was offered by referral to the participants own ophthalmologist. If the individual did not have an ophthalmologist, they were referred to the Botucatu Medical School's Ophthalmology service.
Statistical analysis
The prevalence of low vision and blindness were calculated based on the sampling design, which was approximated as a one-stage cluster design where each household was considered to be the primary sampling unit. Households were randomly selected and thus point prevalences were unbiased. Age was categorized into decades from 1–9 years through to those people 70 years and over. Ethnicity was categorized by self-identification into people of European, African-Brazilian, East Asian or other racial (including multiracial) descent. Occupation was categorized into professional, tradesperson, service worker (for example; secretary), manual worker (for example; construction worker), retirees, and students. Medical history was categorized into healthy, other, diabetes, hypertension and diabetes and hypertension. Univariate and then multivariate analyses were performed to determine whether age, gender, ethnicity, occupation or medical history was significantly related to blindness. Logistic regression models were constructed to investigate the combined predictors of refractive error in a multivariate fashion. Covariates were added in a stepwise pattern and retained if the R2 increased, and beta coefficients were calculated for each predictor. Odds ratios (ORs) and 95% confidence intervals (CIs) for the predictors were calculated. All P values were 2-sided and were considered statistically significant when the values were <0.05. The statistical software packages used were SPSS version 13.0 (SPSS Inc., USA) and SAS Version 9.1 (SAS Institute Inc., Cary, NC, USA).