The IHS was conducted between December 2006 and July 2008 in three regions of India: New Delhi in the north (All India Institute of Medical Sciences and Centre for Chronic Disease Control); Mumbai in the west (Healis-Sekhsaria Institute for Public Health); and Trivandrum in the south (Regional Cancer Center). These centers were selected to capture regional variability in diet and lifestyle and to utilize cancer registries meeting the International Agency for Research on Cancer standards [38
Sampling in the IHS was stratified by gender, religion, and type of residence (urban/rural) (Figure ) and designed to estimate the mean and range of various foods and nutrients within each strata with the following precision: mean intake of nutrients would lie within 5% of the true intake value with 95% confidence and mean intake of foods would lie within 10 percent of the true value with 95% confidence [39
]. To obtain our desired precision for dietary data and satisfy requirements for estimating the participation rate, approximately 200 households per stratum were required.
India Heath Study Centers with the sampling categories.
Participants were eligible if they were aged 35-69 years old, resided in the study area for a minimum period of one year (to minimize participant migration), had no prior history of cancer or cardiovascular event, could speak English or the primary regional language, had no physical ailments that prevented them from fully participating in the study, and were willing to provide biological samples. Females could not be pregnant. We recruited an approximately equal numbers of subjects for each five-year age category, and one male and one female per household for equal gender distribution and cost efficiency.
Human ethics committees from each study center and the Special Studies Institutional Review Board of the United States National Cancer Institute reviewed and approved the study protocol prior to study commencement. Due to the nature of the international collaboration, the Indian Health Ministry Screening Committee for projects involving foreign assistance and/or collaboration, which is part of the Indian Council of Medical Research reporting to the Government of India, also reviewed and approved the study. We obtained written informed consent from all participants.
, India's capital, is a metropolitan city spread over 1,483 km2
. The IHS was conducted in the South District, the second largest district with 2.3 million residents (16% of New Delhi's total population), covering 16.7% of the city's total area [41
]. Of the three subdivisions in this district, Hauz Khas was randomly selected and of the 19 wards, Wards 13 and 11 were randomly chosen. Within these wards, a total of 23 census enumeration blocks were randomly selected for sampling households (Figure2).
We initially attempted to identify households using the 2001 census information, but the houses listed in the census did not match with the existing properties; therefore, we selected an equal number of houses from each lane of a census enumeration block (contains three to five lanes). Of the 1,298 households identified, we successfully interviewed 626 households.
(formerly Bombay), the densely populated capital of the state of Maharashtra is divided into suburban districts and the island city, which comprises 15.9% of the greater metropolitan area (76.8 km2
) with 3.3 million people [41
IHS participants were recruited from an ongoing study of mortality, the Mumbai Cohort Study (39, 40) in three representative areas (Parel, Naigaum, Sewri) from Ward F-South (Figure ). If the cohort member from a selected household had died, moved outside the study area, or was not eligible for the IHS, then a new eligible person, who may or may not have been a cohort member, was recruited from the same or neighboring household. Of the 851 households identified, 687 households were successfully interviewed.
district (or Thiruvananthapuram), the capital of the state of Kerala, is located on the south-west coast of India, and spans 2,192 km2
. Trivandrum's 3.4 million people live in four taluks or sub-divisions, which are split into urban areas (34%) and rural panchayats, village councils (76%) [41
]. We recruited participants from six urban and 49 rural wards in two taluks (Figure ).
The wards were selected to maximize the number of Hindu, Christian, and Muslim participants as a proxy for dietary practices. Wards are divided into three or four polling stations and households were identified with voter lists from 2006 that contained the name of the head of household, address, and the number of adult individuals in the household. Of the 4,915 households identified, 1,720 households were successfully interviewed (925 urban, 795 rural).
Before the study recruitment began, the three centers used multiple approaches to introduce the study aims and objectives to the communities. Trained field staff, along with the principal investigators, talked to community and religious leaders, and/or held public meetings in communal areas. In order to standardize the study protocol and implementation, as well as develop a highly-trained and professional staff, we conducted two intensive, week-long protocol development and training sessions. We also instituted a quality control component where principal investigators of the study sites would pay unannounced visits to the field. The community-based mobile clinics or camps, where participants underwent medical examinations and provided biological samples, were staffed by medical doctors and nurses. This was useful to bring attention and credibility to the study's recruitment efforts, as well as to achieve compliance for the different components.
Once households were identified, field interviewers visited the homes to determine eligibility, provide information, and schedule a visit. During the first in-home morning visit, the interviewer administered demographic, residential history, physical activity, tobacco and alcohol use, and occupational history questionnaires, as well as a computerized diet history questionnaire (Table ). Participants usually completed all questionnaires within one hour. In Mumbai, anthropometric measurements were also completed during the first visit. Interviewers in New Delhi and Trivandrum scheduled a second visit and left biospecimen containers with directions for collecting a first morning urine sample and toenail clipping. Biological samples were not collected in Mumbai as the study center did not have laboratory or storage facilities.
Field visits and details on questionnaires
Diet was assessed using a computer-based, interviewer-administered, meal-based comprehensive diet assessment tool known as the N
iet in I
tudy of H
ealth (NINA-DISH) [42
]. This software was developed for the IHS by modifying software developed by Novo Nordisk Pharma India (Bangalore, India). The diet history (DH) component included three sections: a set of defined questions similar to a food-frequency questionnaire [44
], an open-ended section for each mealtime to collect additional unique regional foods, and a food preparer questionnaire (amount and type of oils, spices, onion, garlic, chilies, and coconuts purchased per household). A subset of participants completed four 24-hour dietary recalls providing information on all foods consumed during the day.
All participants completed the validated, short-form of the International Physical Activity Questionnaire (IPAQ) about total time spent in physical activity for recreation, occupation, household work, and transportation in the last 7 days [45
]. Total weekly physical activity (metabolic equivalents of task (MET-hr/wk) was calculated as the weighted sum of the reported time spent at each intensity using a MET value specific to each category (walking: 3.3 METs; moderate: 4 METs; vigorous: 7 METs).
Biological sample collection and processing
The second visit was conducted either in mobile clinics within the participant's neighborhood or at the individual's home between six and eight in the morning. The participants completed medical history and reproductive questionnaires, and provided 15 ml of blood, 100 ml of first morning urine sample, and toenail clippings from all toes. As an incentive, we offered to provide the participants with the results from the blood analyses. Blood pressure and anthropometric (weight, standing and sitting height; waist, hip, and thigh circumference; and triceps, sub-scapula and supra-patella skin fold) measurements were also taken. Biological samples were transported to the laboratory in coolers within three-hours of collection. Laboratory technicians processed the samples into fractions as soon as they reached the laboratory (i.e., plasma, serum, blood clot, buffy coat, red blood cells) and stored them in equal aliquots at -80° Celsius. Toenail clippings and Guthrie cards with blood spots were stored in a dry environment at ambient temperature.
Fasting glucose levels were determined with the glucose oxidase/peroxidase method [48
] (New Delhi: Randox Laboratories Ltd., Antrim, UK; Trivandrum: Span Diagnostics Ltd., Surat, India). In New Delhi only (all reagents from Randox Laboratories Ltd., Antrim, UK), lipid profiles were analyzed using the following methods: total cholesterol by cholesterol oxidase/p-aminophenazone method, triglyceride by glycerolphosphatase oxidase-peroxidase aminophenazone method, and HDL by precipitation method using phosphotungstate/magnesium-precipitation of apolipoprotein B containing lipoproteins followed by estimation of cholesterol in supernatant by enzymatic method. LDL was estimated using the Friedwald formula [49
Follow up projections
We calculated the expected numbers of cancer cases over five years per 100,000 people for the eight most common cancers at each of the four IHS location (i.e., Mumbai, New Delhi, Trivandrum rural and urban) by gender (i.e., total cohort of 800,000). For incidence, we used truncated crude rates (35-69 years) and age-specific rates (35-39, 40-44, 45-49, 50-54, 55-59, 60-64 and 65-69) from the Trivandrum Cancer Registry (2005-2006) [50
] and the National Cancer Registry Program (2001-2004) for Mumbai and New Delhi [51