Based on different grades of nutrition, BMI results indicated that 28.0% of women were underweight, 52.4% were normal, 17.8% were overweight and 1.8% were obese. The average value of BMI for underweight, normal, overweight, and obese women was 17.8, 21.2, 27.1, and 32.1, respectively. A large number of obese women were in the age group of 31–40 years. Data on the waist-hip ratio (WHR) showed that 23.6% of women had more than 0.85 WHR. The analysis of the waist-hip ratio showed that the risk of metabolic diseases (such as cardiovascular disease, diabetes, etc.) was higher in moderate group women (16.0%).
Dietary assessment was done using the frequency of food consumption and dietary intake survey by 24-hour recall method. The data on food frequency of subjects calculated as intake per day by the number of respondents were also recorded. The food frequency data showed that the major cereal consumed was wheat, followed by seasonal cereal grains (Jawar, Bajra, Makka, etc.). Rice was consumed once or twice a week. Among vegetables only a few seasonal vegetables were used, the use of green leafy vegetables was only once or twice a week. Milk was commonly used in preparing tea. Overall food frequency indicated that the varieties of foods used were very limited and most of the women depended on locally grown produce. The daily dietary intakes and intake differences as per the Recommended Dietary Allowance (RDA, 2010) of the farm women were given in Table 4. The nutrient value of food is calculated according to the Nutritive Value of Indian Foods.
The intake of calories by the sedentary, moderate, and heavy work groups was 0.3% more 2.7%, and 9.8% less respectively as compared to RDA. The intake of protein was 8.7g and 22.7g more by moderate and heavy group women respectively, while 1.4g less intake of protein by sedentary women according to recommended protein intake value. The visible fat consumption was higher than the recommendation in all the groups ranging from 28.7–44.6%, and it has been observed that more intake of fat by the moderate work group of farm women. The chances of metabolism disorder increase due to the intake of a higher amount of fat. Calcium intake was insufficient at 310.6, 314.3, and 319.4 mg/day for sedentary, moderate, and heavy work group women, respectively, however, it required 600 mg/day according to RDA. The lack of calcium affects the bone stability of the women participants. The diet was inadequate in iron with 19.7mg against the recommendation of 21.0mg/day for sedentary work group women, while it was sufficient in the moderate and heavy group with values of 21.2 and 23.7mg, respectively. However, Vitamin A was much below the desired intake level due to low-level intake of green leafy and other vegetables. It might be noted that most of the vitamin A intake of these participants comes through carotenoids. It was found that the intake of carotene for sedentary. The moderate and heavy work groups were 1432.5 (70.2% less), 1147.0 (76.1% less), and 1435.7 µg/day (70.1% less) respectively, while according to RDA for adults, it will be required at the level of 4800 µg/day. Vitamin C was also found to be inadequate.
These results suggest that the participants needed to increase the intake of protective foods such as vegetables and fruits to take care of such inadequacies in their diets. The overall result shows that poor intake of micronutrients in their diet according to their work activity due to which they faced tiredness, weakness, white nail, back pain, etc.
The actual intake of nutrients varies significantly across the physical status of the respondents, physical activities carried out, and their monthly income level (Table 5 to 7). It is evident from the tabular analysis that the intake of nutrients and physical status of the respondents have a positive association, with the exception of carotene (Table 5). With respect to the physical activities performed by the respondents (Table 6), it is explicit that the intake of nutrients was more for the respondents who perform heavy work and vice-versa, barring vitamin C. Income is a major economic variable influencing the nutrients intake. From the tabular analysis (Table 7), it is clear that respondents with a monthly income of < ₹ 3000 have more nutrient intake. This might be due to the higher level of physical work associated with the selected income class level i.e., < ₹ 3000.
Boxplot analysis (Fig. 3 and Fig. 4) indicates the distribution of the respondents with respect to outcome variables like BMI apart from calorie intake by the respondents. The visualization analysis supports the tabular analysis of higher calorie intake for the respondents who are associated with heavy work, followed by moderate and sedentary work styles. In terms of physical status, as expected, the median calorie intake (2281 kcal) for the respondents falling under the ‘over weight’ category stood at the highest (Fig. 3). On the contrary, the BMI registered the lowest for respondents who performed heavy work, followed by moderate and sedentary work, affirming the apriori relationship between BMI and work style (Fig. 4). Similarly, there existed a positive relationship between weight and BMI.
The deviation from the norm shows that barring protein, visible fat, and iron, the rest of the nutrients have witnessed a shortfall for the overall sample (Table 8). However, it differs across the physical status of the respondents as observed in the Table 8. The deviation (shortfall) was highest for the under weight category in terms of calorie intake. Folic acid, calcium, carotene, and vitamin C have shown a shortfall from the norm across the physical status of the respondents. Similarly, visible fat, iron, and protein showed a positive deviation from the norm irrespective of the respondent’s physical status.
Inequality (estimated by Gini coefficient) analysis captures the extent of distribution of the nutrient intake among sample respondents and it showed that there is not much variation among the nutrients level across the physical status of the respondents with the exception of carotene and vitamin C (Table 9). This implies that only a particular section of the respondents consumes foods that are rich in Vitamin C and Carotene. Accordingly, policies and strategies have to be prescribed for reducing the inequality level like creating awareness of the importance of the specific nutrient, sources of food items that are rich in Vitamin C and Carotene, and possible access to such foods.
CRT analysis was performed to get a visual depiction (Fig. 5) of the correlates between the physical status of the respondents and its determinants (explanatory variables). The CRT presents the order of importance of variables that determines the physical status of the respondents. The root node has been divided into node 1 and node 2, based on the BMI, which indicated that height and weight are the foremost important factors among the selected variables in deciding the level of physical status. It was further disaggregated into visible fat. The CRT produces the branches (child nodes) from the parent node based on the homogeneity level prevailing among the respondents for the classification criterion. In the classification process, branch formation gets terminated if there are no correlates providing any additional information (51). In our case, with the exception of fat, the rest of the nutrients didn’t turn significant in determining the respondents’ physical status in terms of weight. The estimates of the predictive power of CRT indicate that the model has predicted 99.6% correctly overall (Table 10). For the observed ‘Normal’ and ‘Over Weight’ the model predicted 100% correctly, whereas, for ‘Under Weight’ the prediction was 98.4%. The overall model prediction percentage indicated 99.6% reliability of the CRT results. Such analysis helps to know the probable variables in deciding the physical status of the respondents and thereby facilitates to take policy decisions.
In addition to the above, cluster analysis, a data-mining technique has been used to identify discrete groups of respondents with respect to BMI, the outcome variable of nutrient intake. This analysis was carried out to produce the final cluster centers to identify the discrete and thereby suggest policies. These are estimated as the mean value for each variable within each final cluster. The final cluster centers grouped by the analysis for each cluster reflect the typical features of the dataset. In our case, the cluster 1 respondents tend to have moderate BMI (Table 11). Also, the cluster 2 respondents tend to be with low BMI and respondents in cluster 3 (a large number of respondents were assigned: 141) tend to have higher BMI. The Analysis of Variance (ANOVA) table indicates the most contributing factor to the cluster production based on the magnitude of the F value. Alternatively, variables with a larger F value provide the greatest separation between clusters. In our case, carotene significantly influenced the cluster formation with an estimated F value of 827.47. Similarly, the Euclidean distance between the final cluster centers implies that the greater the distance between two clusters greater the dissimilarity between them. In our case, cluster 2 and 3 are most different. Clusters 1 and 2 are approximately equally similar to clusters 1 and 3.