The nutrition transition and adolescents' diets in low- and middle-income countries: a cross-cohort comparison

Abstract

Objective: To investigate changes in dietary diversity and dietary composition among adolescents in four developing countries.

Design: We analysed dietary diversity and consumption of seven food groups and foods with added sugars as reported by adolescents from two cohorts growing up 8 years apart, when they were aged about 12 years.

Setting: Ethiopia, India (Andhra Pradesh), Peru and Vietnam in 2006 and 2013.

Subjects: Adolescents (n 3659) from the older cohort (OC) born in 1995/96 and adolescents (n 7422) from the younger cohort (YC) born in 2001/02 (N 11 081).

Results: Controlling for other factors, dietary diversity increased in Peru (OC=4·89, YC=5·34, P<0·001) and Ethiopia (OC=3·52, YC=3·94, P=0·001). Dietary diversity was stable in India (OC=4·28, YC=4·29, P=0·982) and Vietnam (OC=4·71, YC=4·73, P=0·814); however, changes in dietary composition were observed. YC adolescents were more likely to consume eggs (India: +32 %, P=0·038; Vietnam: +50 %, P<0·001) and milk and dairy (India: +12 %, P=0·029; Vietnam: +46 %, P<0·001). Other notable shifts included meat consumption in Peru (+72 %, P<0·001) and consumption of fruit and vegetables in Ethiopia (+36 %, P<0·001). Compared with OC, the prevalence of added sugar consumption was greater among the YC in Ethiopia (+35 %, P=0·001) and Vietnam (+44 % P<0·001). Between 2006 and 2013, disparities in dietary diversity associated with household wealth and place of residence declined, although this varied by country. No marked gender disparities in dietary diversity were evident.

Conclusions: We found significant changes over time in dietary diversity among adolescents in four countries consistent with the hypothesis of the nutrition transition.

Keywords: Adolescents; Cohorts; Dietary diversity; Low- and middle-income countries; Non-communicable diseases; Nutrition transition.

Fig. 1

Differences in dietary diversity by adolescent gender (, girls; , boys), country and cohort (OC, older cohort (born 1994/95); YC, younger cohort (born 2001/02)), Young Lives study. Predicted margins, with their standard errors represented by vertical bars, adjusted for adolescent’s gender and age in months, caregiver’s age, gender and educational level, head of the household’s age and gender, household size, wealth tertile, urban residence, cohort, and interactions between cohort and gender, cohort and wealth tertiles, and cohort and urban residence. *P<0·1, **P<0·05.

Fig. 2

Differences in dietary diversity by household wealth (, poorest tertile, equivalent to the lowest household wealth tertile; , least poor tertile, equivalent to the highest household wealth tertile), country and cohort (OC, older cohort (born 1994/95); YC, younger cohort (born 2001/02)), Young Lives study. Predicted margins, with their standard errors represented by vertical bars, adjusted for adolescent’s gender and age in months, caregiver’s age, gender and educational level, head of the household’s age and gender, household size, wealth tertiles, urban residence, cohort, and interactions between cohort and gender, cohort and wealth tertiles, and cohort and urban residence. **P<0·05, ***P<0·01

Fig. 3

Differences in dietary diversity by place of residence (, rural; , urban), country and cohort (OC, older cohort (born 1994/95); YC, younger cohort (born 2001/02)), Young Lives study. Predicted margins, with their standard errors represented by vertical bars, adjusted for adolescent’s gender and age in months, caregiver’s age, gender and educational level, head of the household’s age and gender, household size, wealth tertiles, urban residence, cohort, and interactions between cohort and gender, cohort and wealth tertiles, and cohort and urban residence. **P<0·05, ***P<0·01

Fig. 4

(colour online) Heterogeneity in consumption of food groups contributing to protein consumption by household wealth (wealth tertiles defined based on household wealth, which was estimated as wealth index: , poorest tertile; , second tertile; , least poor tertile), country and cohort (OC, older cohort (born 1994/95); YC, younger cohort (born 2001/02)), Young Lives study. Predicted margins, with their 95 % confidence intervals represented by vertical bars, adjusted for adolescent’s gender and age in months, caregiver’s age, gender and educational level, head of the household’s age and gender, household size, wealth tertiles, urban residence, cohort, and interactions between cohort and gender, cohort and wealth tertiles, and cohort and urban residence

Fig. 5

(colour online) Heterogeneity in consumption of added sugars consumption by place of residence (, rural; , urban), country and cohort (OC, older cohort (born 1994/95); YC, younger cohort (born 2001/02)), Young Lives study. Predicted margins, with their 95 % confidence intervals represented by vertical bars, adjusted for adolescent’s gender and age in months, caregiver’s age, gender and educational level, head of the household’s age and gender, household size, wealth tertiles, urban residence, cohort, and interactions between cohort and gender, cohort and wealth tertiles, and cohort and urban residence