The details of various aspects of design of the APEDS have been described previously.
3 20 21 22 Approval of the Ethics Committee of the Institute was obtained for the study design, which was conducted during the 5-year period 1996 to 2000, in compliance with the tenets of the Helsinki Declaration.
Briefly, a multistage sampling procedure was used to select the study sample of 10,000 persons, with 5,000 each older and younger than 30 years based on the assumption that a 0.5% prevalence of an eye disease in either of these groups may be of public health significance. One urban and three rural areas from different parts of AP were selected. Approximately 2950 persons were sampled in each of the four areas with the intent of including ∼2500 participants in each area, so as to reflect the urban–rural and socioeconomic distribution of the population of this state. These four areas were located in Hyderabad (urban, stratified by socioeconomic status and religion), the West Godavari district (economically well off, rural), and the Adilabad and Mahabubnagar districts (poor, rural). To obtain a sample representative of the entire population of the city of Hyderabad, we stratified the urban blocks by socioeconomic status and religion, because these variables might influence ocular morbidity. Because details of socioeconomic status were not available, we stratified blocks based on our knowledge of Hyderabad gained from various sources, including a surveyor with 27 years’ field experience in Hyderabad. The socioeconomic strata were extreme low (monthly income per person, ≤ 200 rupees (US$ 4.31), low (201–500 rupees), middle (501–2000 rupees), and high (>2000 rupees). We assumed that 0.7% of the Hyderabad population was homeless (no accurate data were available) and included those people in the lowest socioeconomic stratum. We stratified blocks by two major religious groups, Hindu and Muslim, based on location, because people of the same religion tend to live in the same areas. For practical purposes, we assumed that socioeconomic status and religion were homogeneous within each block. We chose 23 blocks (clusters) and one cluster of homeless people by stratified random sampling with an equal probability of selection. The selected blocks were mapped, and the number of households listed. We randomly selected every third to fifth household depending on the total number of households in each block, to obtain a similar number of households in all blocks. We selected 2954 people from Hyderabad with the purpose of achieving a recruitment rate of at least 85% from these blocks.
From three rural areas in different parts of the state, 70 rural clusters were selected with the purpose of having a study sample representative of the socioeconomic distribution of the rural population of the state. We sampled 8832 subjects from these three rural areas, of whom 7771 participated in the study. The major difference between the urban and rural samples was that the former was selected from blocks stratified by socioeconomic status and religion, whereas the latter were selected from villages stratified by four different castes (forward caste, backward caste, scheduled caste, and scheduled tribe) assuming that the different castes roughly reflect the different socioeconomic strata in these rural areas.