PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • Study design and data collection: The Indian Migration Study (IMS) was set up to examine the effects of rural to urban migration on obesity and diabetes. The study was nested in a larger sentinel surveillance study of cardiovascular risk factors in industrial settings [33], and used a sibling-pair comparison design in which urban factory workers who had migrated from rural areas were recruited together with their rural-dwelling sibling who had not migrated. Details of the design and major findings have been reported elsewhere. [14] Briefly, the study was based in factories in four Indian cities (Lucknow - Hindustan Aeronautics Ltd; Nagpur - Indorama Synthetics Ltd, Hyderabad - Bharat Heavy Electricals Ltd and Bangalore – Hindustan Machine Tools Ltd) situated in the north, centre and south of the country. Factory workers and their co-resident spouses were recruited if they were rural-urban migrants, using employer records as the sampling frame. Each participant was asked to invite one non-migrant full sibling of the same sex and closest to them in age still residing in their rural place of origin. A 25% random sample of urban non-migrant factory workers and spouses was invited to participate in the study, and also asked to invite a sib who resided in the same city but did not work in the factory. Information sheets were translated into local languages and explained to participants by trained interviewers and signed (or thumb print used if illiterate) to indicate informed consent. Ethics committee approval was obtained from the All India Institute of Medical Sciences Ethics Committee. Field work began in March 2005 and was completed by December 2007. Diet was assessed by an interviewer-administered semi-quantitative food frequency questionnaire (FFQ). The questionnaire assessed portion size and frequency of intake of 184 commonly consumed food items, asking about consumption over the last year. A standard portion size was assigned to each food (e.g. tablespoon, ladel, bowl), and participants were shown examples of these vessels and asked to report portions consumed as multiples of this. Frequency was recorded as daily, weekly, monthly, yearly/never. A single FFQ was used to cover the four regions of the study. Nutrient databases were developed for the study by collecting recipes from participants in rural and urban areas of each region, and using Indian food composition tables to calculate nutrient content of each recipe [34]. Where nutrient values were unavailable from the Indian food composition tables (e.g. for foods such as manufactured ‘western’ snacks and sweetened drinks), the United States Department of Agriculture nutrient database (USDA, Release No. 14) [35] or McCance and Widdowsons Composition of Foods were used [36]. Because of variation in food preparation, region, rural/urban, and cooking oil specific databases were used for calculation of average daily dietary intake. Energy, carbohydrate, fat, saturated fat, protein, as well as proportion of energy from carbohydrate, fat and protein, and energy density, were considered. The recipes were also used to generate databases of the food group composition of each food item, and used to calculate average daily food group intake. For this analysis the following food groups were considered: fruit; vegetables (including vegetables added to preparations, and salads); legumes (pulses, lentils, whole gram preparations); sugar (sugar and jaggery used in preparations and added to beverages); meat (including meat added to mixed dishes); fish; and dairy (including dairy products added to beverages and preparations). To measure socio-economic position, a subset of 14 of 29 questions were used from the Standard of Living Index (SLI), a household level asset based scale devised for Indian surveys,[37] selecting those believed to be most informative for the study population, and weighting them to give a maximum score of 38. Weights of items for the SLI developed by the International Institute of Population Sciences in India, and based on a priori knowledge about the relative significance of the items, were used in these analyses. Analyses of IMS data to date have shown large differences in chronic disease profile. For example, BMI in rural, migrant and urban men was 21.9, 24.0 and 24.3 respectively (p<0.0001) and in women 22.5, 25.2, and 25.9 respectively(p<0.0001)[14]. Initial analyses of the dietary data also showed that consumption was mainly of traditional foods in rural, migrant, and urban groups, and that there were low levels of western foods consumption [38]. Rural-urban migrants, their rural siblings, and urban non-migrants and their urban siblings were included in analyses. Distributions of macronutrient intake were checked for outliers. Median values and lower and upper quartiles were calculated (because of the skewed distribution of some variables) for energy intake, macronutrient intake (protein, fat, saturated fat, and carbohydrate, and percent energy from each of these), and energy-density, by sex and migration status. Nutrients were transformed to the natural log scale and statistical tests for the significance of the trend in mean values across rural, migrant, and urban groups were calculated using linear regression models and Wald test statistics, adjusted for age and factory site. Analyses were carried out separately in men and women, as we anticipated that there may be gender differences in the effect of migration, and also because of the statistical dependency between husbands and wives produced by the study design. Robust standard errors were used to account for clustering in sibling pairs. Similar analyses were conducted to examine variation in food groups by migration status. Vegetarianism was assessed by whether the participant reported consumption of any meat or fish in the FFQ. Differences in the proportion of vegetarians in rural, migrant and urban groups were assessed via Wald test statistics from a logistic regression model, adjusting for age, factory, and again accounting for clustering in sibling pairs by calculating robust standard errors. For meat and fish, differences in intake were considered for participants who were non-vegetarians only. Further analyses examining differences (suitably transformed) in intake between migrant and non-migrant siblings, taking advantage of the sibling-pair design, were also conducted as this comparison provides a high level of control for known and unknown genetic and early life confounders. Only rural-urban migrant pairs were included in these analyses as urban non-migrant sib-pairs were not informative for assessing migration effects. Because not all siblings were matched on sex, sex-specific z-scores of intake were calculated using the rural distribution as reference. Within pair differences in z-scores for energy, macronutrients, energy density, and food groups were considered and modelled using linear regression. Although the sibling closest in age was recruited, differences in age were inevitable so the within-pair difference in age was considered as a potential confounder. Factory was not controlled for because it did not vary within pairs. The estimated intercept in these models is to be interpreted as the estimated mean difference in z-scores between same-age siblings and therefore as the age-adjusted effect of migration. Mean differences and 95% confidence intervals were graphed to aid interpretation. Differences in energy-adjusted intake were calculated using the residuals method [39], calculating z-scores of residuals in each sibling, and taking the difference of these scores. The additional effect of the amount of time the migrant sibling had spent in urban areas (with categories: <10 years or 10+ years) was also studied by extending each of the linear regression models used to study the macronutrients and the food groups, with significance assessed via Wald tests. Similarly, effect differences by factory were studied by including factory indicators in the models. All analyses were conducted using STATA 11 statistical software (StataCorp. 2009. Stata Statistical Software: Release 11. College Station, TX: StataCorp LP).
rdf:type