Food security in Bangladesh has improved in recent years, but the country is now facing a double burden of malnutrition while also being highly vulnerable to climate change. Little is known about how this may affect food supply to different sectors of the population. To inform this, we used a national dietary survey of 800 rural households to define dietary patterns using latent class analysis. Nutrient adequacy of dietary patterns and their potential vulnerability to climate shocks (based on diversity of calorie sources) were assessed. We fitted mixed effects logistic regression models to identify factors associated with dietary patterns. Four dietary patterns were identified: rice and low diversity; wheat and high diversity; pulses and vegetables; meat and fish. The wheat and high diversity and meat and fish patterns tended to be consumed by households with higher levels of wealth and education, while the rice and low diversity pattern was consumed by households with lower levels of wealth and education. The pulses and vegetables pattern was consumed by households of intermediate socio-economic status. While energy intake was high, fat and protein intake were suboptimal for all patterns except for the wheat and high diversity pattern. All patterns had fruit and vegetable intake below the WHO recommendation. The wheat and high diversity pattern was least vulnerable to shocks, while the rice and low diversity pattern was the most vulnerable, relying mainly on single cereal staples. The diets showed "double vulnerability" where the nutrient inadequate patterns were also those most vulnerable to shocks.
Keywords: diet vulnerability; dietary pattern; farm production; latent class analysis; nutrition transition; staples.