In this study, we examined the association between dietary consumption and nutrition related health outcomes using data extracted from nationally representative surveys on food consumption practices and prevalence of common diseases. Two aspects may affect the associations between dietary consumption and health outcomes, namely, the socioeconomic status that can influence the dietary consumption practices, and regional spatial associations that may converge on similar dietary practices and/or health outcomes. We adjusted for these two factors using appropriate statistical measures at the analysis stage. We used districts as the units of selection for our study to analyse the associations on a finer detail than doing at state or any other broad geographical regions. We also extracted the data separately for urban and rural regions and determined the dietary patterns and associations separately as there may be variations in the health outcomes.
Factor analysis revealed four major dietary patterns in both urban and rural regions. ‘Milk and wheat rich diet’ is predominantly consumed in the rural and urban regions of districts of northern most states of India. It is no surprise that the diet with higher consumption of milk coincides with the lactose tolerant North-Indian population[29, 30]. ‘Rice and meat rich diet’ is consumed in the districts of southern, eastern, and north-eastern states. It is interesting to note that this diet is characterised by lesser consumption of milk and coincided with lactose intolerant south Indian population[29, 30]. ‘Coconut and sea food rich diet’ is consumed in the districts of Kerala, Goa, coastal Karnataka, and other coastal regions. It is interesting to note that the food items that characterised these dietary factors found in our study are almost similar between rural and urban regions.
Exploratory factor analysis has been used by several authors as an empirical way of studying dietary patterns. Such studies have found association of certain dietary patterns with health outcomes. For example, a study by van Dam RM et.al (2003), in Dutch population found association of certain dietary patterns with blood pressure, blood glucose and cholesterol concentrations[31]. Another study by Menotti et.al (1999), in seven different countries found association of dietary patterns with mortality due to coronary heart disease[32]. Joy et.al (2017) and Satija et.al (2015) analysed the Indian Migration study data and found association of certain dietary patterns with non-communicable disease risk factors[8, 11]. But the data was collected from people working in only four factories and thus may not be representative of the national population.
The dietary patterns obtained in this study are similar in characteristics to those reported from other studies done in India either using the same data set or from regional studies with few minor variations. For instance, the preference of cereal consumption varies between rice (South and East India) and wheat in (North and West India) different regions of India. Similarly, milk (more in Northern districts) and meat consumption (Southern and Eastern districts) have distinct patterns in India. The variations between dietary patterns identified in this study could be related to the different methodologies used (factor analysis in our study, cluster analysis in Tak et. al., from the same dataset[10].
We then analysed the association between dietary patterns and health outcomes and found several direct associations of the dietary factors. While ‘Milk and wheat rich diet’ was associated with prevalence of anemia among men. One obvious explanation for the observed association is that the diet lacks iron rich red meat content. Another possible explanation could be the presence of factors that influence iron absorption. ‘Milk and wheat rich diet’ is characterised by higher consumption of milk. Milk proteins are known to affect the intestinal absorption of iron[33]. The effect of milk and other dietary components on iron absorption in ‘Milk and wheat rich diet’ needs to be further investigated.
Both ‘Rice and meat rich diet’ and ‘Coconut and sea food rich diet’ was associated with prevalence of overweight BMI and high blood glucose. A recently published multi-national study has reported an association between rice consumption and incidence of diabetes which is particularly strong among South Asians[34]. Another recently published meta-analysis has revealed increased risk of developing diabetes mellitus with increased consumption of meat[35]. Both the ‘Rice and meat rich diet’ and ‘Coconut and sea food rich diet’ are characterised by higher consumption of meat. Meat is also rich in iron and multiple studies have shown increased risk of diabetes with higher intake of iron and increased iron stores in the body[36]. The association of diabetes with rice, and meat could possibly explain the association of prevalence of high blood glucose with these two diets. ‘Coconut and sea food rich diet’ is also found to be protective against anemia and underweight BMI which could be possibly due to the higher intake of sea food which are higher in protein and iron content.
‘Refined oil and tur dal rich diet’ is associated with higher prevalence of overnutrition and lower prevalence of undernutrition health indicators. The reasons for which could be due to a higher intake of protein rich tur dal diet and possible health benefits from using refined oil.
Many of the associations seen in rural regions were not seen in urban regions. This points to the more complicated nature of nutrition related health outcomes in urban regions as compared to rural regions. They may also point to variables other than dietary patterns and DDI predicting the health outcomes which requires further studies.
The findings of the study have implications for policy decisions. For example, while choosing the food item for fortification with a nutrient, the dietary pattern, and its component food items of the regions with higher prevalence of the nutrient deficiency can be considered. It has been reported that the central government of India is planning to make fortification of rice with iron mandatory to reduce prevalence of anaemia[37]. This must be reconsidered in light of our study findings since it is the wheat rich diet that is associated with increased prevalence of anemia. The regions affected by higher prevalence of anemia may not benefit adequately from fortification of rice. Moreover, association of diet rich in rice with increased prevalence of high blood glucose must sound caution since increase in iron intake could further increase the risk of diabetes among regions consuming rice rich diet.
There are several limitations of the study that are worthy of mention. Factor analysis as a methodology may be suitable for identifying dietary patterns but may not be ideal to infer causal relationships on health outcomes. Hence, the associations obtained in this study may need to be confirmed by other studies.
Another important limitation of our study is that the data used in the study were from two different surveys of health outcomes and food consumption done in two different time points and by different agencies. However, since the analysis was done at district level and not at individual level and as it is corrected for spatial autocorrelation, we expect the interpretation of this study to be still relevant. Another limitation is that the data on food consumption is not available separately for gender and different age groups.
In conclusion, our study found four major dietary patterns among districts of India, and they are associated with several nutrition related outcomes even after adjusting for differences in socio-economic development. Mechanism of these associations including differences in specific nutrient intake between the dietary patterns need to be explored further.