dietary behaviour and to further examine the associations of different dietary
compositions with selected characteristics.
Design: Latent class analysis was applied to data from the recent cross-sectional
National Family Health Survey that collected information on the intake frequency of
selected foods. Different responses regarding intake frequency were condensed into
a set of five meaningful latent clusters representing different dietary patterns and
these clusters were then labelled based on the reported degree of diet mixing.
Setting: Indian states.
Subjects: In total, 90 180 women aged 15-49 years.
Results: Three clusters were predominantly non-vegetarian and two were vegetarian.
A very high or high mixed-diet pattern was observed particularly in the southern and
a few north-eastern states. Many women in the very high mixed-diet cluster
consumed mostly non-green/leafy vegetables on a daily basis, and fruits and other
non-vegetarian diet on a weekly basis. In contrast, those in the low mixed-diet cluster
consumed more than three-fifths of the major vegetarian diet ingredients alone on a
daily basis. The affluent group that represented the low mixed-diet cluster were
primarily vegetarians and those who represented the very high mixed-diet cluster
were mostly non-vegetarians. The significant interrelationships of different
characteristics highlight not only socio-economic, spatial and cultural disparities
related to dietary practices, but also the substantial heterogeneity in diet mixing
Conclusions: The results of this study confirmed our hypothesis of heterogeneous
dietary behaviour of Indian women and yielded useful policy-oriented results which
might be difficult to establish otherwise.