Prevalence and correlates of household food insecurity in Delhi and Chennai, India

Abstract

India is home to nearly 200 million undernourished people, yet little is known about the characteristics of those experiencing food insecurity, especially among urban households. The objectives of this study were: (1) to report the prevalence of food insecurity in two large, population-based representative samples in urban India, (2) to describe socio-economic correlates of food insecurity in this context, and (3) to compare the dietary intake of adults living in food insecure households to that of adults living in food secure households. Data are from 4334 households participating in an ongoing population-based cohort study of a representative sample of Delhi and Chennai, India. The most recent wave of data (2017–2018) were analysed. Food insecurity was measured using the 9-item Household Food Insecurity Access Scale (HFIAS) and dietary intake using a 33-item semi-quantitative food frequency questionnaire. The overall prevalence of food insecurity was 8.5% (95% confidence interval [CI], 6.8–10.2); 15.2% (95% CI 12.0–18.4) of the poorest households (lowest wealth index tertile) were food insecure compared to 1.7% (95% CI 1.0–2.3) of the wealthiest households (highest wealth index tertile). Participants experiencing food insecurity were significantly younger and more likely to be from Delhi compared to Chennai. After adjustment for socio-economic factors (city, age, sex, education, wealth index, fuel used for cooking, and source of drinking water), participants experiencing food insecurity had significantly higher meat, poultry, roots and tubers (potato), and sugar sweetened beverage intakes, and lower vegetables, fruit, dairy, and nut intakes. Food insecurity is highly prevalent among the poorest households in urban India and is associated with intake of a number of unhealthy dietary items.

Introduction

There has been a recent uptick in global hunger according to the Food and Agriculture Organization (FAO), signalling a clear need for improved monitoring of who is experiencing food insecurity (FAO et al. 2018). Understanding hunger in India, which is home to approximately one-fifth of the world’s population, will be especially important for global progress toward achieving Sustainable Development Goal 2 (“Zero Hunger”). The Government of India adopted the National Food Security Act (NFSA) in 2013 in order to strengthen the targeted public distribution system (TPDS) and address hunger and food insecurity. The NFSA legislates that households below the poverty line (totalling over 65 million across the country) are entitled to 5 kg of subsidised cereals per month per person. Moreover, pregnant women, lactating mothers, and children (aged 6 months to 14 years) are entitled to meals under the Integrated Child Development Scheme and Mid-Day Meal scheme. Despite this ambitious programme, India has not witnessed a substantial decline in the prevalence of undernourishment according to estimates from the FAO: it has remained around 14–15% from 2014 to 2018 (FAO et al. 2018). In 2015–16, the national prevalence of stunting, wasting, and underweight among children under 5 years of age was 38.4%, 21.0%, and 35.7%, respectively (International Institute for Population Sciences 2017).

Food insecurity is one of the underlying drivers of persistent undernourishment in India (Bhutta 2016; Haddad et al. 2014). The food insecurity situation in India is not entirely driven by low production – indeed, yields of rice and wheat exceed current consumption (Department of Agriculture Cooperation & Farmers Welfare 2017) and requirements for buffer stocks (Hussain 2018), though when all macronutrients are considered (e.g. carbohydrates, protein, and fat), domestic crop yields are insufficient to meet requirements (Ritchie, Reay, and Higgins 2018). Persistent hunger in India is driven by uneven distribution (e.g. access), gender inequality, and declining purchasing power due to high food prices and shrinking agrarian and informal sector incomes (e.g. affordability) (Upadhyay and Palanivel 2011).

Hunger remains an enormous challenge for the country, yet only a handful of studies have reported on the prevalence of food insecurity and individual- and household-level characteristics of people that are food insecure. In a small, non-representative door-to-door survey (130 households, 68% classified as having low socio-economic status) conducted in 2009 in urban Tamil Nadu, the prevalence of food insecurity with hunger was high: 62% based on the US Department of Agriculture Household Food Security Scale (Gopichandran et al. 2010). A survey of 500 households in an urban slum in Kolkata conducted in 2010–2011 using the same scale found a much lower prevalence of food insecurity (15%) (Maitra 2017). In two small door-to-door surveys (250 to 410 households) conducted in urban slums in Delhi, 51% were food insecure based on a 4-item scale that had not been validated (Agarwal, Sethi, and Gupta 2009) and 77% were food insecure based on the Household Food Insecurity Access Scale (HFIAS) (Chinnakali et al. 2014). In the latter study, higher educational attainment of women handling food and greater number of earning members in the household were both significantly associated with a lower risk of food insecurity (Chinnakali et al. 2014). Finally, a study conducted in 2016 in Bangalore found that 16.7% of households were food insecure based on the HFIAS (Anand et al. 2019).

Policies and programmes to address food insecurity, and, in turn, undernourishment, are likely to differ between rural and urban areas. Almost 70 million Indians live in urban slums, and nearly 14% of the urban population lives below the poverty line according to the 2011 Government of India Census. The existence of considerable hidden urbanisation has also been demonstrated by the World Bank: according to the Agglomeration Index, an alternative measure of urban concentration, the share of India’s population living in areas with urban-like features in 2010 was 55%, compared to the official 2011 Census of 31% (Ellis and Mark 2016). Despite this significant urbanisation trend in India and globally, food insecurity research focused on urban areas to date is sparse (Tacoli 2019). Urban residents rely on food purchases (Parra et al. 2015), and much of the urban workforce in India is in the informal sector and dependent on low, irregular daily wages, thus impacting food purchasing power. Moreover, about half of urban slums are not notified and therefore residents do not have access to government schemes (e.g. TPDS) (Upadhyay and Palanivel 2011). Food safety is also a major concern as poorer urban residents have less choice in their sourcing of food and less access to clean water, and may therefore be at increased risk of foodborne illness.

In order to ensure that the urban poor are not left behind in actions toward achieving Sustainable Development Goal 2 (“Zero Hunger”), up-to-date estimates of the prevalence of urban food insecurity and the socio-economic characteristics and dietary intakes of those who experience food insecurity in urban areas of India are urgently needed. The objectives of this study were: (1) to report the prevalence of food insecurity in two large, population-based representative samples in urban India, (2) to describe socio-economic correlates of food insecurity in this context, and (3) to compare the dietary intake of adults living in food insecure households to that of adults living in food secure households.

Materials and Methods

Subjects

Data are from the Centre for cArdiometabolic Risk Reduction in South-Asia (CARRS) cohort study, a population-based study of adult men and non-pregnant women aged 20 years or older (Nair et al. 2012). Baseline data were collected in 2010–2012 from three South Asian cities–Karachi in Pakistan and Chennai and Delhi in India–and follow-up data were collected annually thereafter. In the fifth follow-up survey conducted between 2017–2018 in Chennai and Delhi, a module on household food insecurity was added to the questionnaire. Participants who completed this questionnaire during the fifth follow-up visit (n=4334 households and n=7458 participants) were included in this study (Appendix Figure 1).

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the Public Health Foundation of India Ethics Review Committee (Delhi), the Madras Diabetes Research Foundation Ethics Review Committee (Chennai), and the Emory University Institutional Review Board (Atlanta). Written informed consent was obtained from all participants.

Assessment of Household Food Insecurity

The HFIAS module was developed by USAID to assess household food insecurity over the past 4 weeks (Coates, Bilinsky, and Coates 2007). The module consists of nine conditions covering three domains including (i) anxiety and uncertainty, (ii) insufficient quality, and (iii) insufficient food intake and its physical consequences. For example, participants were asked, “Did you worry that your household would not have enough food?” and respondents answered “yes” or “no.” If a participant responded “yes” to any of the nine conditions, they were further asked about the frequency (rarely, sometimes, or often) that condition was experienced.

The responses to these nine conditions were used to calculate three indicators of household food insecurity: (1) the HFIAS score, (2) domain-specific food insecurity prevalence, and (3) overall food insecurity prevalence (the primary outcome). The minimum HFIAS score is 0, if the household responded “no” to the occurrence of all conditions, and the maximum score is 27, if the household responded “yes” to the occurrence of all conditions with a frequency of “often” for all conditions. Thus, higher HFIAS scores correspond to greater food insecurity, and lower HFIAS scores correspond to greater food security (Coates, Bilinsky, and Coates 2007). For the primary outcome of overall food insecurity, households were defined as “food secure” if the participant reported “no” to all items (e.g. HFIAS score equal to 0). Households were defined as “food insecure” if the participant answered “yes” to any of the items (e.g. HFIAS score of 1 or more).

Assessment of Covariates

While food insecurity was measured only at the fifth follow-up visit, some covariates were assessed at earlier time points. If a covariate was assessed at multiple time points (such as source of drinking water, which was measured at baseline and the fourth follow-up visit), we used the most recent data available. At baseline (October 2010–December 2012), sex, age, and total number of people residing in the household were measured; at the fourth follow-up visit (February 2016–February 2017), marital status, educational attainment, source of drinking water, monthly household income, and wealth index were measured; and at the fifth follow-up visit (March 2017–April 2018, concurrent with food insecurity), current employment and type of fuel used for cooking were measured. Wealth index, categorised into tertiles, was derived using principal component analysis based on the following variables: availability of a separate room for cooking, type of toilet facility used, monthly household income, and assets owned by the participants (e.g. television, fridge, washing machine, microwave oven, mixer-grinder, mobile phone, DVD player, computer, car, bicycle, and motorbike). Data for 739 participants were not available for the fourth follow-up visit and so information from their baseline visit was carried forward for marital status, educational attainment, source of drinking water, monthly household income, and wealth index. Those for which this carry-forward imputation was conducted (n=739) were younger and more likely to be male, unmarried/widowed, have no formal schooling, currently employed, and have a household income <10,000 INR (all p<0.05) compared to those with non-missing data at the fourth follow-up visit (n=6719) (Appendix Table 1).

Dietary intake was also assessed during the fifth follow-up visit (concurrent with food insecurity) via a 33-item qualitative food frequency questionnaire (FFQ) adapted from the 19-item INTERHEART study (Iqbal et al. 2008). A comparison of the food items included in the CARRS FFQ and INTERHEART FFQ is provided in Appendix Table 2. For each food item, participants were asked how frequently they consumed it in the past year, and, when consumed, the approximate amount consumed. Portion sizes were estimated using a food model booklet (Australian Food Model Booklet 2017), which was modified as per the Indian setting. FFQ data were transformed into grams per day. For food groups with 25% or more non-consumers (e.g., those reporting never consuming or consuming less than one time per month, food groups in this category included meat, poultry, seafood, egg, nuts, dessert, fried foods, and sugar sweetened beverages), the non-consumers were the referent group and we calculated tertiles among consumers to create a variable with four categories. For all other food groups, we calculated quartiles for the analysis.

Statistical Analysis

All analyses took into account the complex survey design and were age- and sex- standardised to the 2011 South Asia regional population. Analyses were conducted using Stata version 12.0 (StataCorp 2011). All variables were binary or categorical, and thus assumptions of normality did not apply. We confirmed the assumptions of binary logistic regression namely, binary dependent variable, independent observations, little multicollinearity among independent variables, and sufficient sample size (Kleinbaum et al. 2013). To assess socio-demographic and dietary correlates of food insecurity, we used univariable and multivariable binary logistic regression accounting for clustering within households. We tested all univariable associations between socio-demographic and dietary correlates and food insecurity, and included all variables that were statistically significant (p<0.05) in subsequent multivariable models. Based on the univariable models, the final three multivariable models included the following variables: (1) city, age, sex, education, wealth index, fuel used for cooking, and source of drinking water, (2) city, age, sex, education, wealth index, fuel used for cooking, source of drinking water, and one of the food groups (we ran multivariable models for each food group individually, adjusting for significant socio-demographic correlates), and (3) city, age, sex, education, wealth index, fuel used for cooking, source of drinking water, meat and organ meat, poultry, seafood, vegetables, roots and tubers, nuts, fruit and fresh fruit juice, rice and pasta, idli and dosa, legumes and pulses, dairy, desserts, deep-fried foods, and sugar sweetened beverages (this model contained all significant socio-demographic and dietary correlates).

Results

The overall prevalence of food insecurity was 8.5% (95% confidence interval [CI] 6.8–10.2%) with a mean ±SD HFIAS score of 0.5 ±2.1 (the maximum HFIAS score, indicating substantial food insecurity, is 27) (Table 1). Across the three domains of food insecurity, 5.2% (95% CI 4.0–6.4%) experienced anxiety and uncertainty, 4.1% (95% CI 3.2–5.1%) had insufficient food quality, and 1.5% (95% CI 1.0–2.0%) had insufficient food intake with its physical consequences.

Table 1.

Household Food Insecurity Access Scale (HFIAS) components among households in Delhi and Chennai (n=4334 households) in the past 4 weeks.^{a, b}

	Did Not Experience	Experienced Rarely	Experienced Sometimes	Experienced Often
Did you worry that your household would not have enough food?	94.1 (4105)	1.2 (42)	4.1 (166)	0.6 (21)
Were you or any household member not able to eat the kinds of foods you preferred because of a lack of resources?	92.6 (4043)	1.5 (64)	5.0 (196)	0.8 (31)
Did you or any household member have to eat a limited variety of foods due to a lack of resources?	93.4 (4069)	1.6 (74)	4.1 (165)	0.7 (26)
Did you or any household member have to eat some foods that you really did not want to eat because of a lack of resources to obtain other types of food?	94.2 (4101)	1.5 (65)	3.7 (147)	0.5 (21)
Did you or any household member have to eat a smaller meal than you felt you needed because there was not enough food?	97.2 (4224)	0.8 (28)	1.8 (79)	0.0 (3)
Did you or any other household member have to eat fewer meals in a day because there was not enough food?	99.2 (4307)	0.2 (5)	0.4 (21)	0.0 (1)
Was there ever no food to eat of any kind in your household because of lack of resources to get food?	99.6 (4319)	0.2 (3)	0.2 (11)	0.0 (1)
Did you or any household member go to sleep at night hungry because there was not enough food?	99.7 (4323)	0.1 (2)	0.2 (7)	0.0 (2)
Did you or any household member go a whole day and night without eating anything because there was not enough food?	99.6 (4319)	0.2 (3)	0.2 (11)	0.0 (1)

^aAll analyses took into account the complex survey design and were age- and sex- standardized to the 2011 South Asia regional population. Values shown are weighted percent (unweighted n).

^bQuestions were asked of one female and one male in most households, and so we used the answer of the female participant in most cases. In households where only a male was interviewed, we used his answer.

Participants experiencing food insecurity were significantly more likely to be from Delhi (Table 2). Indeed, in multivariable models, city was the strongest predictor of being food insecure: adjusted odds ratio (OR) 40.86 (95% CI 24.79–67.34). The second strongest predictor was educational attainment: food insecure participants were significantly more likely to have no formal schooling or only primary or secondary schooling versus Bachelor’s degree or higher: adjusted OR 3.22 (95% CI 1.83–5.69) and 2.13 (95% CI 1.28–3.53), respectively. Participants experiencing food insecurity were also significantly more likely to be aged 20 to <45 years compared to older age groups: adjusted OR 0.73 (95% CI 0.56–0.94) as compared to 45 to <60 years and 0.45 (95% CI 0.32–0.68) as compared to ≥60 years. The prevalence of food insecurity in the poorest households (lowest wealth index tertile) was 15.2% (95% CI 12.0–18.4) compared to 1.7% (95% CI 1.0–2.3) in the wealthiest households (highest wealth index tertile). Food insecure households also had significantly lower socio-economic status as measured by household wealth index, use of unclean cooking fuel, and use of public source of drinking water (Table 2).

Table 2.

Association of socio-demographic factors with food insecurity among adults 20 years or older living in urban India.

	Food securea (n=6886)	Food insecurea (n=572)	Univariableb	Multivariableb,c
City
Chennai	57.0 (4170)	5.8 (47)
Delhi	43.0 (2716)	94.2 (525)	17.15 (11.36–25.88)	40.86 (24.79–67.34)
Age
20–44.9 years	64.1 (3949)	75.5 (405)
45–59.9 years	27.7 (2103)	20.9 (129)	0.59 (0.48–0.75)	0.73 (0.56–0.94)
≥60 years	8.2 (834)	3.6 (38)	0.44 (0.31–0.64)	0.45 (0.32–0.68)
Sex
Male	45.8 (3172)	43.6 (249)
Female	54.2 (3714)	56.4 (323)	1.11 (1.04–1.18)	1.01 (0.90–1.13)
Marital status
Married	90.4 (6136)	90.8 (507)
Unmarried/Widowed	9.6 (750)	9.2 (65)	1.05 (0.78–1.41)
Education
Bachelor’s degree or higher	20.5 (1315)	4.1 (24)
Primary to secondary school	69.4 (4819)	64.6 (349)	3.96 (2.51–6.28)	2.13 (1.28–3.53)
No formal schooling	10.2 (751)	31.3 (199)	14.51 (9.00–23.43)	3.22 (1.83–5.69)
Currently employed
Yes	53.4 (3496)	48.1 (274)
No	46.6 (3390)	51.9 (298)	1.12 (0.97–1.29)
Household income
<10,000 INR	36.0 (2971)	58.2 (246)
10,000–20,000 INR	33.5 (2398)	29.2 (168)	0.61 (0.47–0.78)
>20,000 INR	20.5 (2067)	12.6 (70)	0.29 (0.20–0.41)
Wealth indexd
1	31.4 (2138)	60.8 (355)
2	31.9 (2301)	32.5 (182)	0.48 (0.37–0.61)	0.44 (0.33–0.59)
3	36.7 (2447)	6.8 (34)	0.08 (0.05–0.13)	0.08 (0.04–0.15)
Fuel type used for cooking
Gas	97.8 (6716)	92.9 (532)
Electricity	0.7 (42)	1.9 (9)	2.71 (1.07–6.84)	2.05 (0.75–5.60)
Kerosene	1.5 (124)	5.2 (31)	3.15 (1.90–5.24)	3.58 (1.89–6.77)
Source of drinking water
Public	28.3 (1998)	21.5 (134)
Private shared	2.1 (134)	7.0 (39)	4.34 (2.67–7.03)	0.53 (0.30–0.94)
Private own	17.5 (1119)	55.8 (310)	4.13 (3.13–5.44)	0.63 (0.42–0.94)
Bottled/purified tap water	52.2 (3634)	15.7 (89)	0.37 (0.26–0.52)	0.37 (0.22–0.62)

^aValues are weighted percent (unweighted n). “Food secure” was defined as reporting “no” to all nine items on the Household Food Insecurity Access Scale (HFIAS). “Food insecure” was defined as reporting “yes” to any one or more of the nine items on the HFIAS.

^bValues are odds ratio (95% confidence interval).

^cAdjusted for city, age, sex, education, wealth index, fuel used for cooking, and source of drinking water. Because monthly household income was included in the derivation of the wealth index, it was not included as an independent predictor in the multivariable model.

^dWealth index derived using the following variables: monthly household income, separate room for cooking, type of toilet facility used, and assets owned by the participants (television, fridge washing machine, microwave oven, grinder, mobile, DVD player, computer, car, bicycle and bike).

After adjustment for socio-demographic factors (city, age, sex, education, wealth index, fuel used for cooking, and source of drinking water), participants experiencing food insecurity had significantly higher meat, poultry, roots and tubers (potato), chapatti, and sugar sweetened beverages intakes, and lower vegetables, fruit, dairy, nuts, idli/dosa, and dessert intakes (Table 3). The lower intake of vegetables and fruits was an especially strong association: adjusted OR 0.39 (95% CI 0.27–0.57) and 0.31 (95% CI 0.22–0.44), respectively. Similarly, food insecure participants had an approximately 60% lower odds of consuming dairy: adjusted OR 0.43 (95% CI 0.30–0.61).

Table 3.

Association of food groups with food insecurity among adults 20 years or older living in urban India.

Quartiles of food group intake	Food securea (n=6886)	Food insecurea (n=572)	Univariableb	Multivariableb, c	Multivariableb, d
Meat and organ meat
0	47.9 (3300)	47.3 (264)
1	19.4 (1355)	15.6 (91)	0.84 (0.64–1.11)	1.16 (0.82–1.63)	0.94 (0.66–1.33)
2	16.4 (1114)	12.2 (77)	0.86 (0.64–1.16)	0.80 (0.56–0.92)	0.88 (0.61–1.29)
3	16.2 (1117)	24.9 (140)	1.57 (1.20–2.04)	1.65 (1.18–2.29)	1.73 (1.22–2.46)
Poultry
0	31.6 (2209)	27.2 (153)
1	23.4 (1657)	23.8 (140)	1.21 (0.94–1.59)	1.51 (1.10–2.06)	2.29 (1.63–3.22)
2	23.2 (1561)	37.0 (199)	1.84 (1.42–2.38)	1.56 (1.14–2.14)	2.39 (1.69–3.39)
3	21.9 (1455)	12.1 (80)	0.79 (0.57–1.10)	1.11 (0.74–1.67)	1.74 (1.13–2.68)
Seafood
0	39.3 (2580)	64.5 (358)
1	20.0 (1371)	28.8 (168)	0.88 (0.70–1.11)	1.06 (0.79–1.40)	1.09 (0.81–1.47)
2	20.5 (1456)	4.1 (25)	0.12 (0.08–0.19)	0.58 (0.35–0.93)	0.48 (0.29–0.79)
3	20.2 (1479)	2.6 (21)	0.10 (0.06–0.17)	0.68 (0.36–1.30)	0.56 (0.31–1.00)
Egg
0	26.0 (1800)	28.3 (157)
1	29.1 (2011)	28.0 (161)	0.92 (0.71–1.19)	1.34 (0.98–1.83)
2	25.8 (1867)	19.1 (116)	0.71 (0.54–0.94)	1.17 (0.83–1.64)
3	19.2 (1202)	24.6 (138)	1.31 (1.00–1.73)	1.11 (0.79–1.56)
Vegetables
0	24.6 (1865)	23.8 (143)
1	23.3 (1581)	35.2 (195)	1.61 (1.27–2.04)	0.56 (0.41–0.77)	0.56 (0.42–0.75)
2	25.2 (1678)	26.5 (146)	1.13 (0.88–1.47)	0.41 (0.29–0.58)	0.45 (0.32–0.63)
3	26.9 (1762)	14.6 (88)	0.65 (0.49–0.87)	0.39 (0.27–0.57)	0.46 (0.32–0.67)
Roots and tubers
0	24.2 (1844)	4.3 (29)
1	28.8 (2023)	12.6 (81)	2.55 (1.69–3.83)	1.42 (0.90–2.26)	1.95 (1.22–3.13)
2	24.9 (1606)	34.3 (188)	7.44 (5.04–10.99)	1.69 (1.10–2.61)	3.66 (2.26–5.94)
3	22.1 (1408)	48.8 (274)	12.37 (8.33–18.38)	1.58 (1.00–2.50)	4.21 (2.55–6.97)
Nuts
0	69.9 (4877)	90.1 (519)
1	10.1 (674)	5.3 (26)	0.36 (0.24–0.55)	0.73 (0.45–1.19)	0.73 (0.47–1.15)
2	10.2 (664)	2.1 (11)	0.16 (0.08–0.30)	0.34 (0.16–0.70)	0.29 (0.14–0.59)
3	9.8 (660)	2.5 (15)	0.21 (0.13–0.36)	0.60 (0.35–1.06)	0.41 (0.24–0.72)
Fruit and fresh fruit juice
0	21.9 (1637)	43.9 (249)
1	25.1 (1706)	22.9 (138)	0.53 (0.42–0.67)	0.53 (0.40–0.69)	0.59 (0.46–0.78)
2	26.2 (1756)	21.1 (120)	0.45 (0.36–0.57)	0.50 (0.37–0.66)	0.54 (0.41–0.70)
3	26.8 (1787)	12.2 (65)	0.24 (0.18–0.32)	0.31 (0.22–0.44)	0.26 (0.19–0.37)
Rice and pasta
0	26.2 (1722)	46.5 (263)
1	24.7 (1663)	41.9 (233)	0.92 (0.74–1.13)	1.04 (0.79–1.35)
2	23.2 (1671)	6.7 (43)	0.17 (0.12–0.24)	0.89 (0.57–1.39)
3	25.9 (1819)	4.9 (33)	0.12 (0.08–0.18)	0.98 (0.53–1.81)
Idli and dosa
0	25.5 (1629)	65.1 (373)
1	28.9 (1931)	30.5 (164)	0.37 (0.30–0.46)	0.63 (0.48–0.82)	0.58 (0.45–0.75)
2	20.6 (1475)	1.9 (15)	0.04 (0.02–0.08)	0.44 (0.19–1.04)	0.20 (0.10–0.39)
3	25.0 (1842)	2.5 (19)	0.05 (0.03–0.08)	0.64 (0.29–1.41)	0.36 (0.19–0.66)
Chapati
0	26.9 (1929)	4.8 (40)
1	25.4 (1860)	1.0 (6)	0.16 (0.05–0.46)	0.21 (0.07–0.63)	0.20 (0.07–0.62)
2	35.0 (2290)	64.8 (365)	7.69 (4.97–11.88)	0.75 (0.34–1.67)	3.65 (2.26–5.89)
3	12.7 (792)	29.5 (161)	9.80 (6.22–15.46)	0.64 (0.28–1.48)	3.82 (2.26–6.44)
Legumes and pulses
0	27.1 (1979)	15.1 (83)
1	21.8 (1493)	38.3 (222)	3.55 (2.71–4.63)	1.16 (0.82–1.64)	1.33 (0.96–1.86)
2	31.4 (2041)	31.6 (167)	1.95 (1.47–2.59)	0.82 (0.57–1.18)	0.90 (0.63–1.28)
3	19.7 (1364)	15.1 (100)	1.75 (1.25–2.45)	0.65 (0.43–0.99)	0.81 (0.54–1.22)
Dairy
0	23.8 (1757)	26.2 (154)
1	25.2 (1732)	26.2 (161)	1.06 (0.83–1.35)	0.71 (0.52–0.97)	0.81 (0.60–1.09)
2	26.3 (1720)	33.0 (169)	1.12 (0.88–1.42)	0.57 (0.42–0.77)	0.64 (0.48–0.86)
3	24.7 (1677)	14.5 (88)	0.60 (0.44–0.81)	0.43 (0.30–0.61)	0.40 (0.28–0.56)
Desserts
0	30.9 (2280)	42.0 (256)
1	24.0 (1650)	15.0 (84)	0.45 (0.35–0.59)	0.52 (0.38–0.72)	0.61 (0.44–0.84)
2	21.4 (1437)	24.2 (123)	0.76 (0.61–0.96)	0.59 (0.45–0.78)	0.90 (0.68–1.20)
3	23.7 (1519)	18.9 (109)	0.64 (0.50–0.82)	0.56 (0.42–0.75)	0.71 (0.52–0.97)
Deep-fried foods
0	34.3 (2459)	38.6 (229)
1	21.7 (1482)	22.2 (128)	0.93 (0.74–1.17)	0.58 (0.44–0.77)	0.87 (0.67–1.13)
2	22.0 (1469)	19.7 (111)	0.81 (0.64–1.03)	0.63 (0.47–0.85)	0.92 (0.69–1.23)
3	22.0 (1476)	19.6 (104)	0.76 (0.59–0.98)	0.68 (0.51–0.92)	0.91 (0.67–1.24)
Sugar sweetened beverages
0	57.8 (4087)	55.2 (325)
1	14.2 (1015)	13.6 (68)	0.84 (0.64–1.11)	0.83 (0.60–1.14)	1.01 (0.73–1.39)
2	13.4 (857)	15.7 (93)	1.36 (1.06–1.76)	0.79 (0.58–1.07)	1.33 (0.97–1.82)
3	14.6 (927)	15.5 (86)	1.17 (0.90–1.52)	0.69(0.50–0.94)	1.38 (1.02–1.89)

^aValues are weighted percent (unweighted n). “Food secure” was defined as reporting “no” to all nine items on the Household Food Insecurity Access Scale (HFIAS). “Food insecure” was defined as reporting “yes” to any one or more of the nine items on the HFIAS.

^bValues are odds ratio (95% confidence interval).

^cAdjusted for city, age category, sex, education, wealth index, fuel used for cooking, and source of drinking water

^dAdditionally adjusted for the following food groups: meat and organ meat, poultry, seafood, vegetables, roots and tubers, nuts, fruit and fresh fruit juice, rice and pasta, idli and dosa, legumes and pulses, dairy, desserts, deep-fried foods, and sugar sweetened beverages.

Discussion

The overall prevalence of food insecurity in this representative sample of Delhi and Chennai was 8.5%. Delhi and Chennai are the second and sixth most populous cities in India according to the most recent government census (Government of India 2011). This prevalence is lower than the overall prevalence in Asia for 2018 of 16.6%, and much lower than that for sub-Saharan Africa of 35.2%, estimated using the USDA Economic Research Service demand-oriented International Food Security Assessment model (Thome et al. 2018). Younger adults and those living in poorer households and in Delhi compared to Chennai were most likely to be food insecure. Even after adjustment for a wide range of socio-economic factors, participants experiencing food insecurity had significantly higher intakes of unhealthy foods associated with overweight and nutrition-related chronic non-communicable diseases (e.g. meat, roots and tubers (potato), and sugar sweetened beverages) and lower intakes of nutrient-rich foods (e.g. vegetables, fruit, dairy, and nuts). These findings highlight, to the best of our knowledge, for the first time in a low-income country, unique aspects of urban food insecurity and suggest a pattern similar to high-income countries–that is, food insecurity is associated with diets poor in quality, not necessarily quantity. Policies and programmes should therefore not only provide subsidised cereals to the urban poor, but should also include nutrient-rich, culturally appropriate foods such as vegetables and fruit.

One previous study that used the HFIAS to assess household food insecurity among 250 households in an urban slum in Delhi found a prevalence of food insecurity of 77% (Chinnakali et al. 2014), which is higher than the prevalence we found for the poorest households (15.2%) in this representative sample of Delhi in 2017–2018. Similar results were found in urban Tamil Nadu (the state where Chennai is located), where the prevalence of food insecurity was reported to be 75% in a small, non-representative sample in 2009 (Gopichandran et al. 2010) and 57% in 2015 (Dharmaraju et al. 2018), compared to 0.9% in our representative sample of Chennai in 2017–2018. In addition to calendar time, these studies differed in terms of the socio-economic status of the sampled population: 89% of the participants belonged to the low/lower-middle class in the study conducted in 2009 (Gopichandran et al. 2010), whereas only 24% of the participants belonged to low/lower-middle class in the study conducted in 2015 (Dharmaraju et al. 2018). As reported in these earlier studies (Agarwal, Sethi, and Gupta 2009; Gopichandran et al. 2010), and confirmed in our study, household food insecurity is inversely associated with the socio-economic status, and thus the overall prevalence will be influenced by the socio-economic status of the sampled population. A major strength of our study is that it used data from representative samples of Chennai and Delhi. Of note, a more recent study conducted in 2016 in Bangalore found that 16.7% of households were food insecure based on the HFIAS (Anand et al. 2019), which is much closer to the estimate we found in our study.

We found that individuals living in Chennai, a city located in the south Indian state of Tamil Nadu, were much less likely to experience food insecurity compared to individuals living in Delhi. This is consistent with recent state-level estimates of the epidemiological transition in India that suggest that Tamil Nadu is slightly more advanced in terms of the transition from undernutrition to overweight and nutrition-related chronic non-communicable diseases (India State level Disease Burden Initiative Collaborators 2017). Tamil Nadu has one of the best functioning public distribution system (PDS) programmes in India with more than 93% of Fair Price Shops (e.g. ration shops) run by self-help groups and cooperative societies, which is an important contributor to the programme’s success (Nandhi 2016), and approximately 80% of the state’s population covered by PDS (Dharmaraju et al. 2018). Moreover, under a special PDS programme in Tamil Nadu, participating households are also provided with fortified palmolein oil, pulses, and spices at a subsidised rate (Corporation Tamil Nadu Civil Supplies 2011). While a similar scheme was launched in Delhi, it was expensive compared to its counterpart in Chennai and has yet to be fully implemented (Nandi 2015).

In the present study, education was independently associated with the food insecurity: illiterate individuals had 3-fold greater odds [OR (95% CI) 3.22 (1.83–5.69)] of experiencing food insecurity compared to individuals with a Bachelor’s degree or higher, even after adjustment for wealth index. The relationship between food insecurity and education is likely complex, especially in urban slums of low-income countries: as the proportion of a household’s income spent on food increases, sacrifices may be made in terms of other household expenditures including education. This association is consistent with recent studies from a wide range of countries including Portugal (Álvares and Amaral 2014), the United States (Gowda, Hadley, and Aiello 2012), and Jordan (Bawadi et al. 2012). We also observed that adults aged 45 to <60 years and ≥60 years were much less likely to be food insecure compared to younger adults, aged 20 to <45 years. This is similar to the United States (Gowda, Hadley, and Aiello 2012) and suggests that the young, urban poor may be an especially vulnerable group.

One of the most important findings of this study is that even after adjustment for a wide range of socio-economic factors, participants experiencing food insecurity had significantly higher intakes of unhealthy foods (e.g. roots and tubers (potato), and sugar sweetened beverages) and lower intakes of healthy foods (e.g. vegetables, fruit, dairy, and nuts). While similar findings have been reported in the United States (Hanson and Connor 2014) and South Korea (Kim and Oh 2015), where food insecurity is associated with lower intakes of vegetables, fruit, dairy, and nuts, few studies have explored the association between food insecurity and dietary intake in the context of low-income countries. One study in rural Maharashtra, India conducted in 2012 among children aged 6–23 months found that food insecurity was inversely associated with dietary diversity (Chandrasekhar et al. 2017). Outside of India, in Addis Ababa, Ethiopia, a community-based survey conducted in 2012 among 550 households found a much higher prevalence of food insecurity using HFIAS (75%), but similarly found that low-income households and those headed by uneducated individuals were more likely to experience food insecurity (Birhane et al. 2014). Moreover, they found that food insecurity was related to lower consumption of vegetables, fruit, nuts, and animal-based foods (Birhane et al. 2014). Similarly, a study of rural pregnant women in Bangladesh (n=14,600) found that food insecurity was associated with lower consumption of vegetables, fruit, nuts, and animal-based foods (Na et al. 2016). The opposite finding in our study relating to higher meat consumption among food insecure households may relate to the affluence associated with meat consumption in some subpopulations in India (Robbins 1999) and elsewhere that may lead to increased purchasing despite higher prices, in addition to potential social desirability bias in reporting. More research is needed to explore this finding in India, especially in the context of the larger nutrition transition.

While this is the largest, representative study to evaluate food insecurity in urban India using a validated internationally recognised instrument, it is not without limitations. Given that our dietary assessment instrument was a qualitative FFQ that had not been validated against repeated food records or recalls, we were not able to calculate total caloric intake or intake of specific macro- or micronutrients. However, this is one of the first studies to evaluate the association of food insecurity with dietary intake in India, and our results may be more informative than total caloric intake in terms of translating these findings into food-based policies and programmes. Because food insecurity was self-reported, responses may have been subject to social desirability bias, though this may have been minimised by the fact that the HFIAS and FFQ questions were asked as part of a much larger survey on lifestyle behaviours and disease. In addition, our results may not be generalisable to other cities in India and are unlikely to be generalisable to rural areas in India.

Food insecurity is highly prevalent among the poorest households in urban India, especially in Delhi compared to Chennai, and is associated with unhealthy dietary intakes. Food insecurity is thus not just a rural issue. In order to achieve the Sustainable Development Goal 2, food must be accessible and available for all urban residents–across income levels. In the context of rapidly developing countries experiencing an accelerated nutrition transition resulting in the double burden of malnutrition such as India (Misra et al. 2011), policies and programmes focused on making food accessible and available must consider not only the quantity of food provided, but also the quality of food.

Biographies

Garima Rautela

Senior Research Investigator,

Centre for Chronic Disease Control (CCDC), India

Garima Rautela is currently employed as a Senior Research Investigator at Centre for Chronic Disease Control (CCDC), has a postgraduate degree in Food and Nutrition from University of Delhi, India and has done Postgraduate certificate course in diabetes education from International Diabetes Federation. She is qualified for lectureship through UGC-NET. She has more than 9 years of experience in working in several research projects in different capacities, from implementation, management and to analysis.

At CCDC, she is responsible for designing, collecting and analysing the diet data for an epidemiological survey of “Non-Alcoholic Fatty Liver Disease (NAFLD) for its association with cardiometabolic risk factors in North India. In the past, at Public Health Foundation of India (PHFI), she has worked as a study coordinator in Centre for cArdiometabolic Risk Reduction in South Asia (CARRS)- Cohort study, aiming to assess the prevalence and incidence of cardio-metabolic risk factors. Large diet data is available in CARRS (n=16288) including food frequency questionnaire, 24 hour recalls and knowledge and practices regarding salt consumption. She is analysing this data which will provide a greater insight into diet and CVD relationship in India.

Her broad research interests include public health and nutritional epidemiology especially investigating the role of nutrition in non-communicable diseases. Also, she is interested to learn more about the role of household food insecurity in the development of diet sensitive chronic diseases.

Mohammed K. Ali

Associate Professor of Global Health and Epidemiology

Rollins School of Public Health, Emory University

Dr. Ali has Master’s training in cardiovascular medicine, global health, and business and management with expertise in diabetes, cardiovascular diseases, and implementation/translation sciences, and has extensive experience in study design, implementation and management of large population-based studies, as well as quality assurance and control. He helped design and is on the Steering Committee for the National Heart, Lung, Blood Institute (NHLBI)-funded Center of Excellence for Cardio-metabolic Diseases in South Asia (CoE-CARRS), playing a significant role in the design and set up, development of treatment algorithms, trial logistics and coordination processes, and quality assurance plans for the ten clinic multi-center NHLBI-funded CARRS Translation Trial, and the three-city CARRS cohort study of 13,000 people in South Asia. Dr. Ali co-led the Expert Group on diabetes complications for the Global Burden of Diseases Study and is an investigator on a number of other studies in India.

Dr. Ali also works as a consultant for the US Centers for Disease Control and Prevention (CDC) where he helps manage a program that uses natural experiment designs to evaluate diabetes prevention and control policies in the US. He also serves as scientific advisor for the National Diabetes Prevention Program of the CDC’s Division of Diabetes Translation, co-managing a network of U.S. policy and health services researchers (NEXT-D Study). Dr. Ali has received departmental and national recognition for his teaching, and has mentored over 50 pre-doctoral, medical, and post-doctoral trainees.

Dr Prabhakaran Dorairaj

Vice President (Research and Policy) and

Director, Centre for Control of Chronic Conditions, PHFI

Professor D. Prabhakaran is a cardiologist and epidemiologist by training. He is an internationally renowned researcher and is currently the Vice President- Research & Policy, Public Health Foundation of India, Executive Director of Centre for Chronic Disease Control, New Delhi, India and Professor (Epidemiology) London School of Hygiene and Tropical Medicine, UK. He heads the Centre for Control of Chronic Conditions at PHFI which is a joint initiative of four leading institutions (Public Health Foundation of India, London School of Hygiene and Tropical Medicine, All India Institute of Medical Sciences, New Delhi and Emory University). This center conducts cutting edge research in the prevention of chronic diseases in India and the developing world.

His work spans from mechanistic research to understand the causes for increased propensity of cardiovascular diseases (CVD) among Indians, to developing solutions for CVD through translational research and human resource development. Prof. Prabhakaran is a Fellow of the Royal College of Physicians, UK, Fellow of the National Academy of Sciences, India, and an Adjunct Professor at the Emory University. He is member of Executive Council of the International Society of Hypertension (ISH) and Chair of the International Society of CVD Epidemiology and Prevention (ISCEP). He has received funding from NHLBI, Wellcome Trust, European Commission and several other international and national funding bodies. He has mentored over 40 post-doctoral and doctoral students so far. He has authored several chapters and over 375 scholarly papers with an H index of 63. He is the lead editor of the Cardiovascular Respiratory and Renal Disease Volume of the latest Disease Control Priorities Project.

K.M. Venkat Narayan

Ruth and O.C. Hubert Professor of Global Health and Epidemiology

Emory University

Rollins School of Public Health, Professor of Medicine

Emory University School of Medicine

Dr. Narayan is noted for substantial, multidisciplinary work in diabetes and non-communicable diseases (NCD) public health. He has been involved in several major national and international multi-center epidemiological studies, public health surveillance, translation research, and intervention studies. He is currently also exploring intriguing differences in the pathophysiology of type 2 diabetes in South Asian and other developing countries’ populations globally. With more than 400 publications (Relative Citation Ratio in top 1% on NIH Icite; H-index >60 on Thompson; >110 on google. Scholar), including several high-impact studies, his work exemplifies his leadership in diabetes and NCD public health.

Narayan is a member of the US National Academy of Medicine, and is Fellow of the Royal College of Physicians of Ireland, Fellow of the Faculty of Public Health Medicine, UK, Fellow of the American College of Physicians, and Fellow of the American Heart Association. He won the American Diabetes Associations’ Kelly West award for outstanding achievement in epidemiology in 2015, Danish Diabetes Academy Visiting Professor award 2015–17, Government of India, Jawaharlal Nehru Chair 2016, and Emory University’s Mentor of the Year award in 2011. Aside his professional accomplishment, Narayan was also an invited civilian attendee at the National Security Forum by the Secretary of Air Force, was Duke of Edinburgh Common Wealth Fellow, and CDC Leadership Management Institute Fellow.

Nikhil Tandon

Professor and Head of Department of Endocrinology & Metabolism,

All India Institute of Medical Sciences, New Delhi

Prof. Nikhil Tandon, an endocrinologist by training, is an accomplished clinician researcher, with research contributions span from investigating immunological mechanisms in autoimmune thyroid disease to studying the pathogenesis of chronic disease through epidemiological studies. His ongoing research includes studying the early life influence on adult chronic disease by being an investigator in the New Delhi birth cohort, translation research in the delivery of diabetes care and collaborative research in the area of genetics of type 2 diabetes. He has more than 250 peer reviewed publications which have been cited more than 3000 times.

He has been a member of various scientific committees of the ICMR, and the Department of Biotechnology, Government of India; has helped formulate the ICMR’s Management Guidelines for Type 2 Diabetes in India and is also a technical adviser to the National Programme for the Prevention and Control of Diabetes, CVD and Stroke. Dr Tandon is a Fellow of the National Academy of Science, India, and has been its Vice President. He is also a Fellow of the National Academy of Medical Sciences and has received the Fellowship of the Royal College of Physicians (London).

VM Mohan

Chairman & Chief Diabetologist

Dr. Mohan’s Diabetes Specialities Centre, Chennai

President & Director

Madras Diabetes Research Foundation, Chennai

Dr. V. Mohan is an eminent Indian Diabetologist who has been working in the field of diabetes for over 30 years in Chennai in southern India. Dr. Mohan is the Chairman and Chief of Diabetology at Dr. Mohan’s Diabetes Specialities Centre, a WHO Collaborating Centre for Non-communicable Diseases Prevention and Control and an IDF Centre of Education. He is also President and Director of the Madras Diabetes Research Foundation.

Dr. Mohan started working on diabetes as an undergraduate medical student when he joined his father late Prof. M. Viswanathan a pioneer in diabetes in India. Together with his father, Dr. Mohan setup the first private diabetes centre in India in 1971 and continued to work at this centre till 1991. Dr. Mohan and his late wife Dr. Rema Mohan subsequently established their own diabetes centres under the name of “Dr. Mohan’s Diabetes Specialities Centre”. Dr. Mohan and his colleagues currently oversee 20 diabetes centres and clinics in India and 1 International diabetes centre at Oman and over 350,000 diabetic patients have been registered at these centres.

In 1996, Dr. Mohan established the Madras Diabetes Research Foundation (MDRF) in Asia. His research combines clinical, epidemiological and genomic aspects of diabetes. He has published over 875 papers including over 550 original research articles in peer reviewed journals, over 140 chapters in textbooks on diabetes, endocrinology and Internal Medicine and several invited reviews and editorials.

Dr. Mohan’s work has been recognized by the World Health Organization (WHO) which has nominated his centre as a “WHO Collaborating Centre for Prevention and Control of Non-Communicable Diseases” as well as by the International Diabetes Federation (IDF) which nominated his centre as an “IDF Centre of Education”. He also serves on the Steering Committee of the Govt. of India’s National Program for Prevention and Control of Diabetes, Cardiovascular Diseases and Stroke (NPDCS) and on several other national scientific organizations. He is also a WHO Consultant on Diabetes and serves on several committees of the International Diabetes Federation.

Lindsay M. Jaacks

Assistant Professor of Global Health

Department of Global Health and Population

Dr. Jaacks is an Assistant Professor of Global Health at the Harvard T.H. Chan School of Public Health with over a decade of experience in nutrition research and formal training in nutrition (BS in Nutrition from Cornell University; PhD in Nutrition from The University of North Carolina, Chapel Hill [UNC]), epidemiology (formal minor in Epidemiology from UNC), and environmental health (Postdoctoral Fellowship at Emory University). Dr. Jaacks’s research focuses on improving our understanding of the global drivers of the epidemiological transition from communicable to non-communicable diseases. Specifically, she is interested in the complex interactions between nutritional and environmental exposures within the food system and the role that these interactions play in the etiology of obesity, diabetes, and cardiovascular disease. To date, she has published over 45 peer-reviewed articles.

She has worked on a number of epidemiological studies in China, India, and the United States. Current projects include the quantification of circulating levels of persistent organic pollutants among adults in urban India and estimation of their association with incident diabetes; the longitudinal assessment of circulating levels of persistent organic pollutants, pyrethroids, organophosphate metabolites, and perfluorinated chemicals among youth with diabetes in the United States and estimation of their association with cardio-metabolic risk factors; and analysis of mediation of the association between cookstove fuel use and blood pressure by dietary intake in rural Chinese women. She is a co-investigator for a large, multi-country (India, Rwanda, Guatemala, and Peru) household air pollution intervention trial.

Dr. Jaacks has served as a consultant for the UK Department for International Development on addressing overweight and obesity in low- and middle-income countries and for RTI International on policies to prevent diabetes in the United States. She is a Visiting Professor at the Public Health Foundation of India.

Appendix

Appendix Figure 1.

Flow diagram depicting the enrollment of households at the baseline and number of households followed at each follow-up visit.

Appendix Table 1.

Comparison of socio-demographic characteristics at baseline between participants with and without a fourth follow-up visit (FU4). Values are percent (n).

	Those with FU4 (n=6719)	Those missing FU4 (n=739)	P-value from chi-square test
City
Chennai	3806 (53.2)	411 (47.5)	0.6
Delhi	2913 (46.8)	328 (52.5)
Age
20–44.9 years	3893 (64.7)	461 (68.4)	0.02
45–59.9 years	2044 (27.5)	188 (23.6)
≥60 years	782 (7.8)	90 (8.0)
Sex
Male	3033 (45.0)	388 (51.4)	<0.001
Female	3686 (55.0)	351 (48.6)
Marital status
Married	6123 (91.9)	642 (88.8)	<0.001
Unmarried/Widowed	596 (8.1)	97 (11.2)
Education
Bachelor’s degree or higher	1205 (19.1)	100 (14.4)	<0.001
Primary to secondary school	4777 (70.1)	501 (68.0)
No formal schooling	737 (10.8)	138 (17.6)
Currently employed
Yes	3180 (48.5)	391 (52.8)	0.004
No	3539 (51.5)	348 (47.2)
Household income
<10,000 INR	4451 (65.5)	566 (73.9)	<0.001
10,000–20,000 INR	1179 (18.1)	101 (15.5)
>20,000 INR	1070 (16.4)	71 (10.6)

Appendix Table 2.

Summary of 33 food items used in this study and comparison of food groups in INTERHEART food frequency questionnaire (FFQ).

CARRS study	INTERHEART study
1) Meats	1) Meat/poultry: beef, pork, lamb, mutton, goat, veal, rabbit, chicken, turkey, duck, pheasant; their curries; Mexican meat soups/stews (menudo), liver, kidney, brain, spleen, heart and sausages
2) Poultry
3) Organ meats
4) Fish	2) Fish: fresh-water and sea-water fish; preserved fish such as salted fish, canned fish, dried fish; shellfish and crustaceans (clams, squid, prawns, mollusks); caviar
5) Shell fish and crustaceans
6) Eggs	3) Eggs: Includes preserved eggs, duck eggs, thousand year old eggs
7) Cooked green leafy vegetables	4) Leafy greens: spinach, bok choi; choi sum, collards, mustard or turnip greens; asparagus
8) Cooked other vegetables	5) Other cooked vegetables: any cooked vegetables not included in the preceding categories
9) Cooked Roots and tubers	6) Other raw vegetables: any raw vegetables not included in the preceding categories
10) Uncooked raw vegetables
11) Fruits (Sapota, mango, grapes and banana)	7) Fruit/juice: all fruits and their juices
12) Fruits (All other fruits)
13) Fresh fruit juices
14) Fruit juices
15) Boiled rice, fried rice, briyani, pulav, semolina, sago, pasta	8) Grains: whole wheat flour; whole wheat chappati, cracked wheat; brown/wild rice; corn/hominy/masa harina/corn flour/maize, oats - old fashioned & Scotch/cracked oats; couscous; pot barley, brown rusk; whole wheat pasta, semolina
16) White bread, idli, taftan, sheermal, dosa	9) Refined/milled grains: white flour; white flour chapati; white/polished/instant/ parboiled rice; jook or rice congee; pasta; noodles/ramen/somen; bulgur; pearl barley, sago; plain rusk; sheermal; taftan
17) Whole wheat roti, brown bread, whole grain porridge, pearl millet, barley, ragi, oats
18) Legumes and pulses	10) Legumes: dried beans, lentils, peas, daals; soups (split pea)
19) Milk & milk based drinks	11) Dairy products: milk, yogurt, cheese, curd, raita, lassi, custard, khoya, firni, kheer, milk puddings, and ice cream.
20) Milk products
21) Milk based desserts
22) Deep fried foods1 (chicken nuggets, pakoras, namakparey, namkeen, french fries)	12) Deep fried foods: french fries, potato chips, onion rings, samosas, papad, pakoras; sev; fried won ton, egg roll
23) Deep fried food 2 (samosa, egg rolls, kachori, cutlets, poori, patties)	13) Salty snacks: salt added in cooking and to food at the table and salty snacks such as chips, crackers etc.
24) Dessert1 (Cholate, tarts candy, cakes, pies, ice-cream and pastries)	14) Desserts/sweet snacks: the use of jam; cakes; pies; chocolate; candy; burfi/ladoo; rasgulla/gulab jamun; halwa; shameia, mohalabeia, Chinese sweet buns; nor mei; sweet bean desserts, Coke and other soft drinks
25) Dessert2 (burfi, ladoo, jalebi, gulabjamun, rasgullah, rasmalai)
26) Pickles & chutnies	15) Pickled vegetables (brine): pickled in brine such as dill pickles, relishes; olives; salted cabbage or leafy greens (mui choi); mango pickle, lemon pickle; salted root vegetables (choi po); pickled eggs, pickled meat
	16) Soy and other sauces: fish sauce, oyster sauce, tamari; fermented bean pastes (hoi sin); other salty sauces
27) Nuts	17) Nuts/seeds: peanuts, almonds, sunflower seeds, cashews, walnuts
28) Carbonated beverages	N/A
29) Tea
30) Coffee
31) Miscellaneous foods [biscuit, rusk, phen]
32) Other1
33) Other2
N/A	18) Sugar/sweetener: the use of white sugar, brown sugar, corn syrup, honey, molasses, maple syrup, treacle
	19) Tofu/soybean curd: textured vegetable protein, soy milk