Sources of variability in dietary intake in two distinct regions of rural India: implications for nutrition study design and interpretation.

OBJECTIVE: Limited inter-person variability in nutrient intake hampers epidemiologic studies of diet-disease relationships. Despite conjecture about non-traditional study bases providing large inter-person differences, virtually nothing is known about variations in nutrient intake outside of Europe and North America. The current study was conducted in India to determine sources of variability in the intakes of nutrients thought to be of public health importance.
DESIGN: Adult subjects in Gujarat (North India; n=60) and in Kerala (South India, n=60) were administered 24 h diet recall interviews six and eight times, respectively, over a 1 y period. To assess subject (inter-person) and residual (intra-person) contributions to variance, regression models were fit to the data. From this, the variance ratio (VR, total within or intra-person: total between or inter-person) was computed.
SETTING: Rural communities in North and South India.
RESULTS: In both regions, inter-person variability was larger than that observed in the West. This was most pronounced in Gujarat, for which the VR was <1.0 for seven of the 12 energy-adjusted nutrient intakes in both men and women. In analyses of the combined data, region accounted for >20% of variance for fat, iron, copper, zinc, beta-carotene and riboflavin in both men and women. With the region term removed from this model, virtually all of the variability ascribed to region contributed to inter-person variability.
CONCLUSIONS: The relatively large inter-person variability observed here could contribute to improved 'resolution' of diet-outcome relationships in epidemiologic studies. While this applies to data from each region, it was especially evident in analyses of the combined data.