research paper mid day meal

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 12 July 2021

Intergenerational nutrition benefits of India’s national school feeding program

• Suman Chakrabarti   ORCID: orcid.org/0000-0002-5078-2173 1 ,
• Samuel P. Scott   ORCID: orcid.org/0000-0002-5564-0510 1 ,
• Harold Alderman 1 ,
• Purnima Menon 1 &
• Daniel O. Gilligan   ORCID: orcid.org/0000-0002-3530-0148 1  

Nature Communications volume  12 , Article number:  4248 ( 2021 ) Cite this article

18k Accesses

33 Citations

943 Altmetric

Metrics details

• Developing world

Matters Arising to this article was published on 27 October 2022

India has the world’s highest number of undernourished children and the largest school feeding program, the Mid-Day Meal (MDM) scheme. As school feeding programs target children outside the highest-return “first 1000-days” window, they have not been included in the global agenda to address stunting. School meals benefit education and nutrition in participants, but no studies have examined whether benefits carry over to their children. Using nationally representative data on mothers and their children spanning 1993 to 2016, we assess whether MDM supports intergenerational improvements in child linear growth. Here we report that height-for-age z-score (HAZ) among children born to mothers with full MDM exposure was greater (+0.40 SD) than that in children born to non-exposed mothers. Associations were stronger in low socioeconomic strata and likely work through women’s education, fertility, and health service utilization. MDM was associated with 13–32% of the HAZ improvement in India from 2006 to 2016.

Similar content being viewed by others

Basic determinants of child linear growth outcomes in sub-Saharan Africa: a cross-sectional survey analysis of positive deviants in poor households

A comprehensive intervention package improves the linear growth of children under 2-years-old in rural Bangladesh: a community-based cluster randomized controlled trial

Causes and consequences of child growth faltering in low-resource settings

Introduction.

Globally, 149 million children are too short for their age and over half of these children live in Asia 1 . Within India, 38% of children were stunted in 2015–2016 (ref. 2 ). Linear growth failure is a marker of chronic undernutrition and multiple pathological changes which, together, have been termed the ‘stunting syndrome’ 3 . Stunted children are at risk of not reaching their developmental potential, thus stunting has large implications for human capital and the economic productivity of entire societies 4 , 5 , 6 . The World Health Assembly set the ambitious target of reducing childhood stunting by 40% from 2010 to 2025 (ref. 7 ), a target that likely will not be met 4 . Thus, it is imperative to understand how countries can accelerate progress toward stunting reduction.

Though much focus has been placed on nutrition-specific interventions during the 1000-day period from conception to the child’s second birthday, investments across multiple life periods and which address underlying determinants are also important to achieve stunting reductions 8 , 9 . Interventions may work directly through maternal–child biological pathways or indirectly through socioeconomic mechanisms. In India, women’s height and educational attainment are among the strongest predictors of child stunting 10 , 11 , 12 , 13 , 14 , 15 .

In the Indian context, a candidate intervention which potentially improves both women’s height and education—and which, therefore, may lead to reductions in stunting among children born to these women—is the national school feeding program, the Mid-Day Meal (MDM) scheme 16 . Launched in 1995 by the Government of India, the MDM scheme provides a free cooked meal to children in government and government-assisted primary schools (classes I–V; ages 6–10 years). The mandated minimum meal energy content is 450 kcal and the meal must contain 12 g of protein. In 2016–2017, 97.8 million children received a free cooked meal through the scheme every day, making the MDM scheme the largest school feeding program in the world 17 .

Econometric evaluations of India’s MDM scheme have shown a positive association with beneficiaries’ school attendance 18 , 19 , learning achievement 20 , hunger and protein-energy malnutrition 21 , and resilience to health shocks such as drought 22 —all of which may have carryover benefits to children born to mothers who participated in the program. We are not aware of studies that have explored whether program benefits for the MDM or similar programs in other countries extend to the next generation. Filling this research gap is critical, as (1) stunting carries over from one generation to the next and is therefore optimally studied on a multigenerational time horizon 23 , 24 , 25 , 26 , (2) school feeding programs are implemented in almost every country 27 , and (3) social safety nets such as India’s MDM scheme have the potential for population-level stunting reduction as they are implemented at scale and target multiple underlying determinants in vulnerable groups 28 .

At a broader level, a substantial literature documents effects of cash transfer programs on education of girls in low- and lower-middle-income countries 29 . While transfer programs clearly address food security, their track record on improving anthropometry is mixed at best, possibly because evaluations focus on relatively short-term impacts 28 , 30 . However, even in the United States, a timely transfer—for example, the Supplemental Nutrition Assistance Program—has been shown to have health benefits over time 31 . Other studies document effects of cash transfers, health insurance, and other programs for children in beneficiary households on future adult outcomes such as incomes, achieved schooling 32 , nutritional status 33 , 34 , and mortality 35 .

The described literature suggests a potential pathway through which school feeding programs and other cash transfer or in-kind safety nets focused on education may have intergenerational effects on child nutrition outcomes. Current frameworks for understanding the intergenerational transmission of health disparities advocate for a multi-generation approach that addresses parental socioeconomic status (SES), child and adolescent health and development, and young adult’s capacity for planning and future parenting 36 . However, since interventions to improve maternal height and education must be implemented years before those girls and young women become mothers, empirical assessment of the effectiveness of such programs for reducing undernutrition among future offspring is challenging.

This paper studies the intergenerational nutrition benefits of India’s MDM scheme. We use seven population level datasets spanning 1993 to 2016, including multiple rounds of National Sample Surveys of Consumer Expenditure (NSS-CES), National Family Household Surveys (NFHS), and India Human Development Surveys (IHDS). We match cohorts of mothers by state, birth year, and SES with data on MDM coverage measured as the proportion of primary-school-age girls receiving MDM using data from the NSS-CES. Birth cohort fixed effects and controlled interrupted time series models are used to estimate the association of mother’s exposure to the MDM scheme with the nutritional status of her future children. We find that maternal cohorts living in areas with higher coverage of the MDM scheme are less likely to have stunted children than cohorts living in low coverage areas. This effect is robust to the inclusion of a broad set of controls at multiple levels and fixed effects. Controlled interrupted time series models confirm that the 14 states which rolled out MDM in the late-1990s experienced improvements in child height earlier than the rest of the nation, which scaled up MDM in the 2000s after the Supreme Court mandate. Plausibility is supported by our findings of MDM association with participants’ education, age at birth, number of children, use of antenatal care, and delivery in a medical facility.

Program description and motivation

The MDM scheme, initiated by the central government in 1995, was intended to cover all government schools under the National Programme of Nutritional Support for Primary Education 21 . Due to institutional challenges, only a few states scaled up the program immediately. NSS-CES data from 1999 show that only 6% of all girls aged 6–10 years received mid-day meals in school (Fig.  1 ). Between 1999 and 2004, program coverage increased in many states, largely due to an order from the Supreme Court of India directing state governments to provide cooked mid-day meals in primary schools 37 . In 2004, 32% of Indian girls aged 6–10 years were covered by the program. Finally, following a substantial increase in the budget allocation for the program in 2006, by 2011, 46% of girls aged 6–10 years benefited from the program. Coverage among boys was similar throughout this period. NSS-CES data show that substantial state variability in MDM rollout existed even ten years after the central mandate. A complete listing of state heterogeneity in program roll-out can be found in Supplementary Table  1 .

Coverage refers to the proportion of girls aged 6–10 years who received a MDM in school. Source for MDM program coverage data (green maps): NSS-CES 55 (2000), 61 (2005) and 68 (2012). Source for child stunting data (red map): NFHS4 (2016). MDM, mid-day meal. Source data are provided as a Source Data file.

Our empirical exploration of the intergenerational benefits of the MDM scheme was motivated by the observation that stunting prevalence was lower among children aged 0–5 years in 2016 in states where MDM coverage was higher in 2005 (Fig.  2 ). The ability of historical MDM coverage to predict the prevalence of stunting in 2016 suggests that a mother’s exposure to the program during primary school may have future returns for her children. However, the observed association may be biased because policy variables in observational data are unlikely to be independent of latent individual and institutional characteristics 38 .

Each circle represents an individual state in India, with the size representing the state population size. Fit line and shaded 95% confidence interval are also weighted by state population size. Sources: NFHS 4 (2016) for stunting data and NSS-CES 61 (2005) for MDM coverage data. MDM mid-day meal. Source data are provided as a Source Data file.

Birth cohort fixed effects analyses

To inform the birth cohort fixed effects analysis, we examined coverage and scale-up of the MDM scheme and HAZ of children by mother’s birth year and SES. The rate of MDM scale-up across SES deciles moved in tandem with child HAZ along the mother’s birth year axis (Fig.  3 ). Later-born mothers from poor households were more likely to be exposed to the program than either earlier-born mothers or mothers from non-poor households (Fig.  3a ). HAZ in children also increased with later mother’s birth year and was higher in non-poor households compared to poor households (Fig.  3b ). The observed trends provide motivation for using MDM rollout by mother’s birth year as a source of variation that is time varying and cohort specific 39 .

Bottom 3 deciles are the poorest households in the sample and top 4 deciles are non-poor. MDM exposure of women born between 1980 and 1998 ( a ) and HAZ of children under 5 years old in 2016 of mothers born between 1980 and 1998 ( b ). Source of MDM coverage data: NSS-CES 50 (1994), 55 (2000), and 61 (2005). Source of HAZ data: NFHS 4 (2016). HAZ height-for-age z -score, MDM mid-day meal. Source data are provided as a Source Data file.

In the birth cohort model, maternal MDM coverage was associated with future child HAZ (Fig.  4a ). After adjusting for maternal birth year, wealth, state, and state-specific-birth-year fixed effects, as well as a set of child-specific controls, HAZ in children born to mothers who lived in areas with 100% MDM coverage was 0.40 SD higher than HAZ in children born to mothers living in areas without the MDM ( p  < 0.05). The inclusion of ICDS and PDS access variables did not attenuate this association. The effect of the program varied by SES; children from poor households had the largest effect (0.5 SD, p  < 0.05) followed by children from middle SES strata (0.33, p  < 0.05), relative to children from the wealthiest SES strata. In robustness checks, program access coefficients were slightly attenuated but remained significant when adding birth year specific SES fixed effects but were not significant after adding birth year and state-specific SES fixed effects. Further, regressions on subsamples of stunted children showed higher precision but smaller coefficients for the benefits of MDM coverage on HAZ compared to children who were not stunted (Supplementary Fig.  5 ).

Panel a shows the relationship between MDM coverage and future child HAZ in the birth cohort model (Eq.  1 ) while panel b shows the relative association across wealth strata (Eq.  2 ). The circles represent the point estimates and whiskers are 95% confidence intervals. Point estimates are interpreted as the difference in HAZ due to 100% exposure to the MDM scheme during primary school years for the relevant sample. Point estimates in panel b for MDM × poor and MDM × middle are the relative effect of 100% MDM coverage for that SES stratum compared to the average effect of 100% MDM coverage for the wealthiest four deciles (MDM coverage). MDM coverage is the proportion of girls born between 1980 and 1998, within state-specific socioeconomic status deciles, who reported receiving at least 10 meals free of cost at school in the previous month. All models control for child age, sex, birth order, maternal antenatal care (4+ visits), institutional birth, residence (urban/rural), religion, caste, access to services from the Integrated Child Development Services (dummies for receiving take home rations, child health check-ups, pre-school education, weight measurements, and nutrition counseling) and the Public Distribution System (household has a Below Poverty Line card to obtain subsidized food). The models include fixed effects for mother’s birth year, state, household wealth, and for state × mother’s birth year. All models cluster standard error estimates at the district level. Sources: NFHS 4 (2016) for outcome and covariates. NSS-CES 50 (1994), 55 (2000), and 61 (2005) for MDM coverage data. HAZ height-for-age z -score, MDM mid-day meal, SES socioeconomic status. Source data are provided as a Source Data file.

Controlled interrupted time series analyses

The controlled interrupted time series model exploits variation in the timing of the expansion of the MDM program to estimate program benefits relative to a reference period (event time 0 = birth year 1992). MDM expansion between event time 0 and 4 (birth years 1992–1996 capturing the short run impact of the program) differed substantially across intervention and control states (Fig.  5a ). Trends in child HAZ were parallel between event time −4 and 0 across intervention and control states (Fig.  5b ). After event time 0, intervention states saw a larger change in child HAZ compared to control states. In regression models, the coefficient for parallel trends was not significant, confirming that trends in child HAZ were statistically similar across intervention and control states before the intervention (Fig.  5c ). The estimated association was similar across all three specifications, 0.038, 0.041, and 0.044 SD per year ( p  < 0.05). Relative to wealthier households, the effect estimate of the MDM in intervention states was larger among poor and middle-income households at 0.044–0.055 SD per year ( p  < 0.10) (Fig.  5d ). In robustness checks, effect coefficients were stable when excluding Gujarat, Odisha, and Chhattisgarh (some districts in these states adopted MDM after Tamil Nadu and Kerala) from treatment states (Supplementary Fig.  2 ).

All models exclude Kerala and Tamil Nadu. Panel a shows MDM coverage by event time across intervention and control states. The program begins between event time 0 and 1. Panel b shows the local polynomial of HAZ of children in 2016, born to women belonging to birth cohorts, before and after the start of the program in each state. The shaded gray area indicates the 95% confidence interval. Panel c shows the coefficient on γ 6 (parallel trends) and γ 7 (DID) from Eq. ( 3 ). Coefficients from three models are specified as Eq. ( 3 ) plus random effects and fixed effects for district and state. Panel d : γ 6 (parallel trends) and γ 7 (DID) from Eq. ( 3 ) with state fixed effects run on a subset of low (SES 1–3), middle (SES 4–6), and high (SES 7–10) households. The squares/diamonds represent the point estimate and whiskers are 95% confidence intervals. The DID coefficient can be interpreted as the difference in the average rate of change in HAZ, per-year, before versus after MDM started, in the intervention compared to control states All models control for child age, sex, birth order, maternal antenatal care (4+ visits), institutional birth, residence (urban/rural), religion, caste, access to services from the Integrated Child Development Services (dummies for receiving take home rations, child health check-ups, pre-school education, weight measurements, and nutrition counseling) and the Public Distribution System (household has a Below Poverty Line card to obtain subsidized food). All models cluster standard error estimates at the state level. Sources: NFHS 4 (2016) for outcome and covariates. NSS-CES 50 (1994), 55 (2000), and 61 (2005) for MDM coverage data. FE fixed effects, MDM mid-day meal, RE random effects, SES socioeconomic status. Source data are provided as a Source Data file.

Program pathways

When examining factors that the MDM may work through to influence child HAZ, full MDM coverage during primary school years was a meaningful predictor of all factors examined (Table  1 ). Full MDM coverage predicted 3.9 years of attained maternal education in years, delaying age in years at first birth by 1.6 years, having a fewer (−0.8) children, a higher probability of having at least four antenatal care visits (22%), and giving birth in a medical facility (28%) (all p  < 0.001). Full MDM coverage predicted higher adult height among direct beneficiaries (0.51 cm) but the association was not statistically significant.

Regression decomposition

Our findings can be put into context by considering changes in HAZ among children under 5 years of age reported in the National Family Health Surveys. HAZ improved by 0.4 SDs between 2006 and 2016, on average. Using Eq. ( 4 ), with an average MDM coverage of 32% in 2004 at the national level (NSS-CES 61) multiplied by the effect size of 0.166 SD (raw data model) to 0.401 SD (smoothed data model), we estimate the MDM explains 0.053–0.128 SD or 13.3–32.1% of average change in HAZ. Using Eq. ( 5 ), with an average of 2.6 years of exposure multiplied by the effect size of 0.044 SD per year, we estimate the MDM can explain 0.114 SD or 28.6% of average change in HAZ. The range estimated contributions are similar in magnitude and relatively substantial, considering that HAZ is dependent on a large set of determinants, of which each can individually only explain a small part of total variation in South Asian countries 11 .

A possible concern for our estimates is susceptibility to the effects of migration. Since we measure MDM exposure at the state level in the past and associate it with child nutrition in the future, attribution of the estimated role of MDM exposure would be weakened in the presence of substantial migration across states. A recent study allays this concern by providing estimates on migration in India. Although 30% of India’s population has ever migrated, two-thirds are intra-district migrants, more than half of whom are women migrating for marriage 40 . In 2001, only 4% of India’s population migrated across state borders 40 . Therefore, migration is not a major concern for misclassification of treatment status in our models.

Discordant SES matching between NSS-CES and NFHS

Overall, we find a 78% concordance between expenditure-based SES deciles measured in 2005 and asset-based SES deciles measured in 2012 at the state level using India’s IHDS (Supplementary Fig.  3 ). Given this 22% discordance, we cannot rule out that our estimates are somewhat biased due to imperfect classification by SES status. However, the degree of bias is likely to be small because mobility across deciles is limited (the IHDS shows that a household generally only moves up by one or two SES deciles over 7 years, if they move at all) and MDM coverage within states does not fluctuate greatly with small increments of SES classes (in 2005, coverage in the IHDS sample ranged between 53 and 62% in the bottom four SES deciles). Moreover, non-differential misclassification as a form of measurement error generally tends to bias estimates towards the null 41 .

Matching by caste and religion

To test the sensitivity of our estimates to demographic measures of socioeconomic position that are less likely to change over time, we matched MDM coverage by state of residence, caste, and religion. Similar to SES matching, adjusted full maternal MDM coverage using caste and religion matching was associated with an improvement in HAZ among children aged 0–59 months (Supplementary Fig.  4 ).

Selection bias from program placement in government schools

Using 2011 data (IHDS-2) we tested whether girls aged 11–17 years in government schools are shorter than those in private schools. We fit a model with state fixed effects that controls for child age, urban residence, occupation, household size, household expenditure, assets, and parental education. We find that girls in government schools are, on average, 0.89 cm shorter than those in private schools ( p  < 0.001). When we add a dummy variable indicating the receipt of MDM during primary school for these girls (identified in IHDS-1), we find that MDM is associated with a higher height of 1.3 cm on average ( p  < 0.001), while government school attendance is associated with 1 cm lower height ( p  < 0.001). This suggests that selection effects from program placement in government schools are likely to bias our estimates downward and that MDM is the driver of higher height among government school beneficiaries.

Testing fixed effects models with raw MDM coverage data

The MDM coverage estimate from a regression model using Eq. ( 1 ) and the raw coverage data is statistically significant but attenuated to 0.166 SD as expected (Supplementary Table  5 , model 1). We also specified a second set of regressions using only the 2004 NSS data, and matched MDM coverage by district and SES. Again, we find an attenuated but significant coefficient of 0.115 SD (Supplementary Table  5 , model 2) and, as expected, a larger coefficient of 0.189 SD among poor households ( p  < 0.05) (Supplementary Table  5 , model 3). As the district level exposure does not have temporal variation by birth year, this model is not directly comparable with the birth cohort model. However, it does demonstrate that MDM coverage variation by district and SES is strongly correlated with HAZ of children of mothers born between 1993 and 1997.

The MDM coverage estimate from a regression model using Eq. ( 1 ) and the log-linear smoothed coverage data are statistically significant but attenuated to 0.261 SD (Supplementary Table  6 , model 1). However, attenuation here is smaller in magnitude compared to those using raw data. The model using Eq. ( 2 ) shows that children from poor households had the largest effect (0.468 SD, p  < 0.01) followed by children from middle SES strata (0.296, p  < 0.05), relative to children from the wealthiest SES strata (Supplementary Table  6 , model 2).

Overall, we conclude that both the smoothed and raw data models provide evidence of an effect of maternal MDM coverage on child anthropometry, though the size of the effect depends on the preferred model. We have provided evidence that this effect is robust to varying model specifications, and that the effect of MDM coverage is largest among the poorest households. Moreover, the control interrupted time series models do not use smoothed coverage but provide qualitatively similar estimates.

We have shown that investments made in school meals in previous decades were associated with improvements in future child linear growth. The plausibility of this finding is supported by an association between MDM exposure and underlying determinants of child linear growth: women’s education, fertility, and health service use. As the analysis covers a large nationally representative sample of households, the results reflect a program implemented at scale, with all its flaws, and not a pilot program designed to provide proof of concept. This, of course, comes at a cost; we could not follow a randomized cohort of girls from primary school to childbearing. We put the magnitude of the association into context by using regression decomposition to estimate the share of the actual HAZ improvement explained by the predicted MDM effect on HAZ.

While others have examined the effects of school feeding programs on education and nutrition in beneficiaries themselves, to our knowledge our paper is the first to demonstrate an intergenerational transmission of benefits. This finding provides evidence that, when intergenerational effects are considered, the complete benefit of school feeding programs at scale for linear growth is much larger than previously understood. The result that a school feeding program is related to the nutritional status of children in the next generation also has important implications for other transfer programs. The literature generally focuses on investments in nutrition during the 1000-day period to reduce childhood stunting; our findings suggest that intervening during the primary school years can make important contributions to reducing future child stunting, particularly given the cumulative exposure that is possible through school feeding programs.

School meal programs are often motivated by their potential to increase schooling, particularly that of girls. While enrolment parity is within reach in primary schooling –between 2000 and 2015, the number of primary school-age children not in school declined globally from 100 million to 61 million 42 —there is a larger goal of primary and post primary school completion. Very little in the literature on school meal programs can quantify program contribution to total years of schooling completed. Moreover, evidence that the scale-up of school meals is associated with increased heights of women—in a population in which stunting has been historically linked with maternal undernutrition—provides a new perspective on the contribution of such programs. This reinforces an increased attention to seeking opportunities to improve nutrition in the “next 7000 days” 8 , that is, to find means of addressing undernutrition should efforts in the high priority period prior to a child’s second birthday not be fully successful. The results here show that school meals may contribute to education, nutrition (height), later fertility decisions, and access to health care; by doing so, school meals may reduce the risk of undernutrition in the next generation. In its current form, India’s MDM scheme has the potential to address multiple underlying determinants of undernutrition. Improving the quality of meals provided and extending the program beyond primary school might further enhance its benefits 8 , though we could not empirically test these hypotheses given the available data.

The MDM is mandated by the Supreme Court of India as a social protection program addressing food insecurity. The social protection role of addressing hunger and food insecurity may be a justification by itself for school-based transfers in many settings 43 . However, evidence such as presented here depict these programs as contributing to both food security and to improved outcomes in the next generation, thus contribute to the policy framework for school-based interventions.

Data sources

This paper relies on evidence from seven rounds of three publicly available nationally representative surveys (Supplementary Table  2 ). The primary analysis in this paper uses data on whether children born between 1980 and 1998 received free meals at school from the National Sample Survey of Consumer Expenditure (NSS-CES) (1993, 1999, and 2004 rounds) 44 , 45 , 46 , 47 . These data are combined with data on child height-for-age z -scores in 2016 from wave four of India’s Demographic Health Survey, the National Family Health Survey (NFHS) 48 . Both are large nationally representative surveys, which make it possible to match exposure to MDM by cohorts of girls born between 1980 and 1998 at the district level with data on mean child height in the same locations in 2016. The 2016 NFHS4 sample included 217,940 women with 196,310 children under 5 years of age. NSS-CES data from 2011 were also used for generating maps for coverage but not for the primary analyses. Our interest was examining next generation benefits on child stunting and our hypothesis was that intergenerational effects work through first generation improvements in education, height, fertility, and access to health services 14 , 43 , 49 , 50 , 51 , 52 . We expected larger influence of maternal coverage compared to paternal coverage given previous evidence showing larger program impacts on girls than on boys 18 . We support our main findings by using the 2004 and 2011 rounds of Indian Human Development Surveys for descriptive analyses and robustness checks 53 , 54 . IHDS provides a wide array of variables that are not available in the NSS or the NFHS and offers supportive evidence on the main estimates and model assumptions. Our study was a secondary analysis of existing public survey data; hence, no ethical approval was required for our study. All surveys complied with ethical norms with appropriate approvals and consent taken at the time of survey. Summary statistics for the primary and secondary outcomes examined in this paper are shown in Supplementary Table  3 . Summary statistics for the covariates from NFHS are shown in Supplementary Table  4 .

Identification strategy

In an ideal experiment, children would be randomly assigned access to free lunches from the MDM program in primary school and we would compare the average HAZ outcomes for the children of the MDM beneficiaries and of the MDM non-beneficiaries when the original children in the experiment reached adulthood. In the absence of randomized treatment allotment, we chose to use panel data techniques from repeated cross-sections 55 to exploit the strengths of the available data for identification—the fact that the data cover birth cohorts over a long period and that MDM coverage varies by state of residence and SES. SES was calculated using a principal component analysis of household assets, including cooking fuel, floor and wall materials, land and house ownership; and the possession of assets, including a mattress, pressure cooker, chair, bed, table, fan, TV, sewing machine, phone, computer, fridge, watch, bicycle, motorbike, car; and the possession of animals, including cows, goats, and chickens.

Year of birth, SES decile, and state of residence were used to determine an individual’s exposure to the program. In India, children are expected to attend primary school between the ages of 6 and 10 years. The NSS-CES provide data on the age of all household members and whether they received free meals at school in the past 30 days. Of all the girls aged 6–10 years in the 2005 NSS sample who reported receiving any free meals at school ( N  = 8873), 95.6% reported receiving at least 10 meals in the previous month. We used a minimum of 10 meals per month to ensure that our coverage estimates were for children who received the program with fidelity. Models were run separately using any MDM access (at least 1 meal) and comparable results were obtained. We use this information to calculate the percentage of all girls aged 6–10 years covered by the program for cohorts born between 1980 and 1998. This period gives us an approximately equal number of birth cohorts who were born before and after the introduction of the MDM scheme. Since the MDM scheme was introduced in 1995, those born after 1989 would be able to receive free meals in primary school. In addition, the NSS-CES provide measures of SES and state of residence, which allowed us to calculate coverage rates for all girls aged 6–10 years, specific to each SES strata in all Indian states.

For any cohort, MDM exposure is a function of the number of years an average child spends in primary school and when the program started in the school they attended. In an ideal data setting, to obtain an accurate coverage estimate for a birth cohort, we would have data from five cross-sections surveyed consecutively. For example, to obtain an estimate of MDM coverage for the 1994 cohort, we ideally would have coverage data on 6-year-old children measured in 2000, 7-year-olds in 2001, 8-year-olds in 2002, 9-year-olds in 2003, and 10-year-olds in 2004. We would then average these five coverage estimates into a single estimate, representing average MDM exposure assuming a typical five-year period in primary school for the 1994 cohort. The averaging is necessary because any single year does not accurately reflect exposure for all 5 years in primary school.

Each NSS-CES repeated cross-section, conducted within 5 year intervals, provides MDM coverage by child age as measured in the survey year. We used linear interpolation to estimate a smoothed continuous exposure indicator that varies by maternal birth year, state, and SES. For example, using coverage estimates for 6 year olds in 1999 NSS-CES (birth year 1993) and 6 year olds in the 2004 NSS-CES (birth year 1998), we first used linear interpolation to estimate the average rate of increase in MDM coverage for 6 year olds for the years (2000, 2001, 2002, and 2003) with no NSS-CES data (these correspond with birth years 1994, 1995, 1996, and 1997). Next, we performed similar interpolation for 7-, 8-, 9-, and 10-year-old children. This provided smoothed coverage estimates for children born in 1993 for the survey years 1999, 2000, 2001, 2002, and 2003—the years the 1993 cohort would have aged from 6 to 10 years. We take the average coverage for these 5 years as the final estimate of coverage experience of a specific birth year (Supplementary Fig.  1 ). This process of smoothing (i) estimates the relationship between maternal school meals exposure and annual child HAZ outcomes under an assumption of a linear trend in exposure and (ii) reduces probable bias due to measurement error present in the raw data by moving extreme values closer to the center of the distribution.

Throughout the paper, we use the term MDM coverage, which refers to an estimate of state-by-year average program exposure during primary school for the birth cohorts in the sample, under the assumption that coverage increases in a linear fashion within age groups of children in primary school surveyed in the years 1993, 1999, 2004, and 2011. It is almost certain that exposure in the interval between 2 years lies between the values in the end points; the assumption that the expansion is linear is a plausible pattern of program roll out. We assume that within 5-year intervals, the duration of primary school, age-specific trends in coverage would have increased gradually. Gradual rollout is typical of at-scale programs in developing countries with numerous implementation, financing, and bureaucratic challenges 56 . However, in sensitivity analyses, we subject the assumption of a linear scale-up to an additional robustness check where we smooth MDM coverage using a log-linear process.

Next, using birth year, SES deciles and state of residence, we match NFHS data with NSS-CES data for the percentage of girls covered by the MDM for cohorts born between 1980 and 1998. NFHS data provide anthropometric measurements for the last three births for each mother. We use data for all available children with valid anthropometric measurements. We calculate HAZ using the “zscore06” STATA routine which automatically excludes outlier measurements.

We specify the following model:

where Y iwst is the height-for-age z -score for child i belonging to SES strata w in state s in mother’s birth year t . MDM wst is a continuous indicator coded as the proportion of mothers covered by the MDM as children and ranges between 0 and 1. T t represents birth-year fixed effects which forces identification of within birth-year effects and controls for time-varying national level economic changes, programs, and policies. Examples of these are national programs such as the National Health Mission introduced in 2005 (ref. 57 ) and changes in national GDP, which has shown robust growth 58 .

We estimated Eq. ( 1 ) using MDM coverage at the state level disaggregated by wealth strata. \(\,{{\bf{W}}}_{w}\) represents the wealth-decile fixed effects and provides controls for all unobserved time-invariant factors associated with household wealth and MDM coverage. S s is the state fixed effects which controls for all for time-invariant differences across states with high and low MDM exposure. S s  *  T t or state-birth-year fixed effects controls for unobserved state-specific time-varying factors that could be correlated with the outcome such as the state’s political climate, varying degrees of implementation of welfare programs, agricultural policies, and educational subsidies. A concern for a model estimated without this parameter is that states that introduced free meals in primary school at different times and rates of coverage expansion could be systematically different. For example, states with residents who had lower education or poorer nutritional status on average may have been more likely to introduce the MDM. Similarly, states with better governance may have been better equipped to implement the MDM program at scale. In either case, the correlation between outcomes and MDM implementation could be confounded with unobserved state-specific time-varying factors.

C iwst represents a vector of individual, household and survey-specific controls, including child age, sex, birth order, mothers antenatal care status during pregnancy, birth in a medical facility, and household characteristics at the time the outcome was measured. The vector includes SES, caste, religion, and residence (urban or rural). P iwst represents a vector of individual and household-specific programmatic controls, including access to services from the Integrated Child Development Services (dummies for receiving take home rations, child health check-ups, pre-school education, weight measurements, and nutrition counseling) and the Public Distribution System (household has a Below Poverty Line card to obtain subsidized food) 59 , 60 . Controlling for these variables reduces possible confounding from government interventions that could benefit current child nutritional status. All standard error estimates were clustered at the district level. Clustering adjusts standard error estimates after accounting for intra-district correlations and assumes that residuals are independent across districts 61 .

The coefficients estimated by Eq. ( 1 ) are intent-to-treat (ITT) estimates because the MDM coverage variable measures “potential exposure” to the program on entire birth cohorts. Our ITT estimates are a policy-relevant parameter for an ex-post analysis of the effects of a large program on the entire population (birth cohorts) 21 , 22 . Our models, based on population representative MDM coverage, estimate the magnitude of improvement in child undernutrition that can be expected if a cohort is potentially treated.

Testing for differential benefits for the poor

IHDS data show that 80% of all MDM beneficiaries in 2004 attended government schools and that two-thirds of children attending government schools were from low-income households (bottom six SES deciles), suggesting that MDM was primarily implemented in government schools rather than in private schools as an incentive for children from poor households to attend primary school (and to improve nutrition); therefore, the estimates in Eq. ( 1 ) are likely to mask heterogeneity of response to the program. Masking is anticipated because outcome data from children sampled from non-poor households, who would be more likely to opt out of the government school system in favor of private schools, would influence average effect sizes 62 . We expect that mothers who were enrolled in government schools during their childhood would have worse nutritional outcomes and this might place a downward bias on our estimates. To investigate the existence of such heterogeneity, we compared associations across SES groups. We created SES deciles and grouped women in the bottom three (poor), middle three, and top four (non-poor) deciles to create two wealth strata. We estimated models for differential associations for poor, middle versus non-poor households by modifying Eq. ( 1 ) as follows:

where \({{\rm{Poor}}}_{{{wst}}}\) and Middle wst dummy variables for bottom three and middle (4–6) SES deciles, respectively, with the top four SES deciles serving as the reference non-poor group. \({\gamma }_{1}\) and \({\gamma }_{2}\) measure if poor and middle SES households benefitted more from MDM coverage compared to non-poor households. We expect \({\gamma }_{1}\) to be larger than \({\gamma }_{2}\) , and if these coefficients are statistically significant and of a large order, then we have evidence that MDM program benefits differ by SES. Note that SES here is current, and mother’s SES may have differed in childhood. To this end, we offer evidence in our sensitivity analyses that SES mobility is likely modest.

Controlled interrupted time series models using state rollout timing

The birth cohort model exploits variation in treatment measured as the proportion of children covered by the program within a birth year, state, and across SES strata. It allows us to express the relationship between MDM and HAZ as a function of coverage. However, it comes at the cost of potential for endogeneity because MDM coverage could potentially be associated with changes in living conditions that vary within cohorts defined by state, birth year, and SES strata. An alternate model exploits the differential timing of MDM rollout across Indian states as a robustness check on the birth cohort model. This alternative can reveal insights for the short-term cumulative benefits of the program 31 .

States implemented the program at different times; de-facto, the program was rolled out in the three phases (Supplementary Table  1 ). According to the NSS data, MDM coverage patterns by state and birth year show that Tamil Nadu and Kerala, i.e. “phase 1” states, had average coverage greater than 20% for maternal birth year 1988). These states initiated school feeding programs well before the central government funded MDM. Following the central government order, in phase 2, other states—Odisha, Himachal Pradesh, Uttaranchal, Haryana, Rajasthan, Sikkim, Tripura, West Bengal, Chhattisgarh, Madhya Pradesh, Gujarat, Maharashtra, Andhra Pradesh, and Karnataka—implemented the program at scale with coverage increasing by more than 10% between maternal birth years 1992–1996. In the remaining states (phase 3), MDM coverage was below 5% and increased by less than 10% between maternal birth years 1992–1996.

These roll-out patterns lend themselves to analysis using a controlled interrupted time series design (CITS) 63 , 64 , 65 . Conceptually, the CITS is a combination of the difference-in-differences and interrupted time series models. It includes a within group before–after comparison, and a between-group comparison, strengthening the control for potential confounders. The first difference is the change in the outcome trend within each group, comparing the period before MDM to the period after (slope change). The second difference is the difference in slope changes in the control group compared to the intervention group (difference-in-differences of slopes). The CITS reduces bias due to other interventions or events occurring around the same time as the MDM intervention and allows comparison groups to start at different levels of the outcome. Moreover, the CITS controls for the improvement in HAZ that would be expected without the MDM and tests for parallel trends within the model.

We exclude Tamil Nadu and Kerala from CITS analysis as they were early MDM implementers and both states have better nutrition outcomes compared to other states in India. We focus on maternal birth years 1988 to 1996, when we have a pre intervention period with no MDM across all states, and a post intervention period when some states introduced the program while others did not. Phase 2 states form the intervention group and phase 3 states serve as the control group. We parameterize the CITS model using Eq. ( 3 ).

In state s in mother’s birth year t , \({{\rm{Int}}}_{{{s}}}\) is a dummy for the intervention states, \({{\rm{T}}}_{{{t}}}\) is the event time, a discrete variable (for maternal birth years 1988–1996) that is centered at 1992 and ranges between −4 and 4. \({{\rm{Post}}}_{{{t}}}\) is a dummy for maternal birth years 1993–1996. In Eq. ( 3 ), \({\gamma }_{6}\) tests the null hypothesis of parallel pre-intervention trends; if not significant, we reject this hypothesis and conclude that pre-interventions differed between intervention and control groups. \({\gamma }_{7}\) is the coefficient of interest, and represents a “difference-in-difference of slopes” between the intervention and control states. If \({\gamma }_{7}\) is statistically significant, the change in HAZ slope for intervention states differs from the change in HAZ slope for control states. In other words, it tests for faster gains in child linear growth for states with MDM. To account for spatial heterogeneity, we run three specifications of Eq. ( 3 ) by adding district random effects, district fixed effects, and state fixed effects.

To explore heterogeneity, we investigate differential associations by household SES by running models within subsamples of poor (SES deciles 1–3), middle (SES deciles 4–6), and non-poor (SES deciles 7–10) households. For robustness, we check sensitivity of coefficients to exclusion of Gujarat, Odisha, and Chhattisgarh from the intervention group. These states had greater than 10% coverage at event time 0 and thus could arguably be placed in the phase 1 category.

To test the plausibility of our results we performed regression-based decomposition with our estimates from Eqs. ( 1 ) and ( 3 ) (ref. 66 ). From Eq. ( 1 ), we estimated the population level effect of the program between 2006 and 2016 with Eq. ( 4 ).

where \({\gamma }_{1}\) is the coefficient of MDM from Eq. ( 1 ), MDM 2004 is the MDM coverage in 2004 and ΔHAZ change in HAZ between 2006 and 2016.

From the controlled interrupted time series model, we estimated the effect of exposure to the program using Eq. ( 5 ).

where \({\gamma }_{7}\) is the coefficient from Eq. ( 3 ), \({{\mathrm{{PostEventTime}}}}\) is the average event time 5 years before and after the start of the program, and ΔHAZ is the change in HAZ between 2006 and 2016. Estimates from Eqs. ( 4 ) and ( 5 ) are proportions and are expected to be less than 1 because the predicted difference in HAZ explained by MDM must be less than the total change in HAZ observed between 2004 and 2016.

We next investigated plausible pathways that might support intergenerational links between the MDM program and child nutrition. We used Eq. ( 1 ) to investigate the association of the MDM with six factors that may be related to the MDM program and which, in turn, correlate with child HAZ: mother’s education and height, mother’s age at first birth, total number of children per mother, number of antenatal care visits attended by the mother during pregnancy and if the child was born in a medical facility. We recognize that this is a plausibility analysis and cannot isolate causality.

We matched MDM coverage by state of residence and SES decile between the NSS-CES and NFHS. This assumes that (1) mobility across SES strata over time is minimal and (2) SES deciles in NFHS correspond well with those in the NSS. We therefore use panel data from the IHDS to assess the concordance of expenditure-based SES deciles measured in 2005 and asset-based SES deciles measured in 2012. Since IHDS follows the same individuals over 7 years, we can track their mobility across SES strata over time and then compare their status on both SES measurements.

Matching by caste or religious group

To test the sensitivity of our estimates to demographic measures of socioeconomic position, we matched maternal MDM coverage by birth year, state of residence and households’ caste/religious groups in the NSS-CES and NFHS. The social groups used to match households were scheduled caste (Hindu), scheduled tribe (Hindu), Muslim, Christian, and others. Similar to the SES model, this model works by assigning a probability of exposure to MDM for maternal birth cohorts that varies by state, religion, and caste. While social groups do not follow strict income hierarchies across states, they have the advantage of being largely time invariant and thus do not introduce biases that result from income mobility.

To test the sensitivity of our estimates using smoothed coverage, we offer an additional alternative using raw coverage data from NSS. These coverage estimates are from cross-sections at specific points in time and are not smoothed using the age profiles of children in the NSS rounds. We first created a scatter plot of smoothed coverage estimates against the raw coverage data to gauge the degree and direction of the smoothing process (Supplementary Fig.  6 ).

The maps in Fig.  1 show a discrete jump in coverage from 6% in 1999 to 32% in 2004. The smoothed data attempt to fill in data gaps on coverage for the years 2000 to 2003. The scatterplot of the smoothed coverage data against the raw data shows that the smoothed data are less extreme than the raw data, which has many 0 and 100% coverage estimates. These extreme values present in the raw data likely reflect measurement error for cohort-specific coverage because they do not capture the transition of increasing coverage for the initial years of program implementation, so that coverage for any observation reflects only a single year during a time of program expansion despite the fact that a student will have spent more than a single year in school. To test our hypothesis that measurement error in the raw coverage data would attenuate results compared to those from the models using smoothed data in keeping with standard expectation with random errors in variables, we ran our primary birth cohort model with raw coverage matched by state, SES, and birth years. We also ran a second test of sensitivity with models that use only the 2004 NSS raw coverage data, matched by district and SES. These models specify the same level of coverage to SES groups within districts for birth cohorts 1993 to 1997.

Testing fixed effects models with MDM coverage smoothed using log-linear process

To test the sensitivity of our estimates using linearly smoothed coverage, we use an alternative log-linear smoothing process. This process assumes an exponential growth in MDM coverage within 5-year intervals. We then fit models with Eqs. ( 1 ) and ( 2 ).

Reporting summary

Further information on research design is available in the  Nature Research Reporting Summary linked to this article.

Data availability

The conclusions of this article are based on publicly available datasets 44 , 45 , 46 , 47 , 48 , 53 , 54 . Source data are provided with this paper. The cleaned and merged dataset is available on the Harvard Dataverse at [ https://doi.org/10.7910/DVN/JTN87W ] 67 .  Source data are provided with this paper.

Code availability

The analysis code that reproduces all the tables and figures in the manuscript is available on the Harvard Dataverse at [ https://doi.org/10.7910/DVN/JTN87W ] 67 .

Development Initiatives. 2020 Global Nutrition Report—Global Nutrition Report . https://globalnutritionreport.org/reports/2020-global-nutrition-report/ (2020).

Ministry of Health and Family Welfare Government of India. National Family Health Survey-4, 2015-2016, India Fact Sheet . http://rchiips.org/nfhs/pdf/NFHS4/India.pdf (2017).

Prendergast, A. J. & Humphrey, J. H. The stunting syndrome in developing countries. Paediatr. Int. Child Health 34 , 250–265 (2014).

Article   Google Scholar  

Development Initiatives. Global Nutrition Report 2017: Nourishing the SDGs (2017).

Black, M. et al. Early childhood development coming of age: science through the life course. Lancet 389 , 77–90 (2017).

de Onis, M. & Branca, F. Childhood stunting: a global perspective. Matern. Child Nutr. 12 , 12–26 (2016).

de Onis, M. et al. The World Health Organization’s global target for reducing childhood stunting by 2025: rationale and proposed actions. Matern. Child Nutr. 9 , 6–26 (2013).

Bundy, D. A. P. et al. Investment in child and adolescent health and development: key messages from Disease Control Priorities, 3rd Edition. Lancet https://doi.org/10.1016/S0140-6736(17)32417-0 (2017).

Black, R. E. et al. Maternal and child undernutrition and overweight in low-income and middle-income countries. Lancet 382 , 427–451 (2013).

Corsi, D. J., Mejía-Guevara, I. & Subramanian, S. V. Risk factors for chronic undernutrition among children in India: estimating relative importance, population attributable risk and fractions. Soc. Sci. Med. 157 , 165–185 (2016).

Headey, D., Hoddinott, J. & Park, S. Drivers of nutritional change in four South Asian countries: a dynamic observational analysis. Matern. Child Nutr. 12 , 210–218 (2016).

Alderman, H. & Headey, D. D. How important is parental education for child nutrition? World Dev. 94 , 448–464 (2017).

Cavatorta, E., Shankar, B. & Flores-Martinez, A. Explaining cross-state disparities in child nutrition in rural India. World Dev. 76 , 216–237 (2015).

Kim, R., Mejía-Guevara, I., Corsi, D. J., Aguayo, V. M. & Subramanian, S. V. Relative importance of 13 correlates of child stunting in South Asia: insights from nationally representative data from Afghanistan, Bangladesh, India, Nepal, and Pakistan. Soc. Sci. Med. 187 , 144–154 (2017).

Chen, Y. & Li, H. Mother’s education and child health: is there a nurturing effect? J. Health Econ. 28 , 413–426 (2009).

Raghunathan, K., Chakrabarti, S., Menon, P. & Alderman, H. Deploying the power of social protection to improve nutrition what will it take? Econ. Polit. Wkly 52 , (2017).

Ministry of Human Resource Development. Mid Day Meal Scheme. Department of School Education & Literacy, Government of India. http://mdm.nic.in/ (2017).

Afridi, F. The impact of school meals on school participation: evidence from rural India. J. Dev. Stud. 47 , 1636–1656 (2011).

Drèze, J. & Kingdon, G. School participation in rural India. London School of Economics and Political Science (1999).

Chakraborty, T. & Jayaraman, R. School feeding and learning achievement: evidence from India’s Midday Meal Program. IZA Discussion Paper 10086 (2016).

Afridi, F. Child welfare programs and child nutrition: evidence from a mandated school meal program in India. J. Dev. Econ. 92 , 152–165 (2010).

Singh, A., Park, A. & Dercon, S. School Meals as a Safety Net: An evaluation of the Midday Meal Scheme in India. Econ. Dev. Cult. Change 62 , 275–306 (2014).

Nabwera, H. M., Fulford, A. J., Moore, S. E. & Prentice, A. M. Growth faltering in rural Gambian children after four decades of interventions: a retrospective cohort study. Lancet Glob. Health 5 , e208–e216 (2017).

Addo, O. Y. et al. Maternal height and child growth patterns. J. Pediatr. 163 , 549–554. e1 (2013).

Hambidge, M., Mazariegos, M., Kindem, M., Wright, L. & Cristobal-Pereza, C. Infant stunting is associated with short maternal stature. J. Pediatr. Gastroenterol. Nutr. 54 , 117–119 (2012).

Stein, A. D. et al. Comparison of linear growth patterns in the first three years of life across two generations in Guatemala. Pediatrics 113 , e270–e275 (2004).

Bundy, D. A., Drake, L. J. & Burbano, C. School food, politics and child health. Public Health Nutr. 16 , 1012–1019 (2013).

Ruel, M. T. & Alderman, H. Nutrition-sensitive interventions and programmes: how can they help to accelerate progress in improving maternal and child nutrition? Lancet 382 , 536–551 (2013).

García, S. & Saavedra, J. E. Educational impacts and cost-effectiveness of conditional cash transfer programs in developing countries: a meta-analysis. Rev. Educ. Res. 87 , 921–965 (2017).

Leroy, J. L., Ruel, M. & Verhofstadt, E. The impact of conditional cash transfer programmes on child nutrition: a review of evidence using a programme theory framework. J. Dev. Effect. 1 , 103–129 (2009).

Hoynes, H., Schanzenbach, D. W. & Almond, D. Long-run impacts of childhood access to the Safety Net. Am. Econ. Rev. 106 , 903–934 (2016).

Cohodes, S. R., Grossman, D. S., Kleiner, S. A. & Lovenheim, M. F. The effect of child health insurance access on schooling: evidence from public insurance expansions. J. Hum. Resour. 51 , 727–759 (2016).

Almond, D., Currie, J. & Duque, V. Childhood circumstances and adult outcomes: Act II. J. Econ. Lit. 56 , 1360–1446 (2018).

Aizer, A., Eli, S., Ferrie, J. & Lleras-Muney, A. The long-run impact of cash transfers to poor families. Am. Econ. Rev. 106 , 935–971 (2016).

Wherry, L. R. & Meyer, B. D. Saving teens: using a policy discontinuity to estimate the effects of medicaid eligibility. J. Hum. Resour. 51 , 556–588 (2016).

Cheng, T. L., Johnson, S. B. & Goodman, E. Breaking the intergenerational cycle of disadvantage: the three generation approach. Pediatrics https://doi.org/10.1542/peds.2015-2467 (2016).

Drèze, J. & Khera, R. Recent social security initiatives in India. World Dev. 98 , 555–572 (2017).

Barrett, C. B. & Carter, M. R. The power and pitfalls of experiments in development economics: some non-random reflections. Appl. Econ. Perspect. Policy 32 , 515–548 (2010).

Verbeek, M. & Vella, F. Estimating dynamic models from repeated cross-sections. J. Econ . https://doi.org/10.1016/j.jeconom.2004.06.004 (2005).

Kone, Z. L., Liu, M. Y., Mattoo, A., Ozden, C. & Sharma, S. Internal borders and migration in India. J. Econ. Geogr . https://doi.org/10.1093/jeg/lbx045 (2018).

Chen, Q., Galfalvy, H. & Duan, N. Effects of disease misclassification on exposure-disease association. Am. J. Public Health . https://doi.org/10.2105/AJPH.2012.300995 (2013).

UNICEF. UNICEF Data Base on Primary Education (2018).

Alderman, H., Behrman, J. & Tasneem, A. The contribution of increased equity to the estimated social benefits from a transfer program: an illustration from PROGRESA/Oportunidades. World Bank Econ. Rev .  https://doi.org/10.1093/wber/lhx006 (2019).

National Sample Survey Office. India—Household Consumer Expenditure, July 1993–June 1994, NSS 50th Round. National Data Archive. DDI-IND-MOSPI-NSSO-50Rnd-Sch1.0-1993-94 (2019).

National Sample Survey Office. India—Household Consumer Expenditure, July 1999–June 2000, NSS 55th Round. National Data Archive. DDI-IND-MOSPI-NSSO-55Rnd-Sch1-July1999-June2000 (2019).

National Sample Survey Office. India—Household Consumer Expenditure, July 2004–June 2005, NSS 61st Round. National Data Archive. DDI-IND-MOSPI-NSSO-61Rnd-Sch1-July2004-June2005 (2019).

National Sample Survey Office. India—Household Consumer Expenditure, Type 1: July 2011–June 2012, NSS 68th Round. National Data Archive. DDI-IND-MOSPI-NSSO-68Rnd-Sch1.0-July2011-June2012 (2019).

DHS Program. India: Standard DHS, 2015–2016 (2016).

Bhutta, Z. A. et al. Evidence-based interventions for improvement of maternal and child nutrition: what can be done and at what cost? Lancet 382 , 452–477 (2013).

Rieger, M. & Trommlerová, S. K. Age-specific correlates of child growth. Demography 53 , 241–267 (2016).

Jayachandran, S. & Pande, R. Why are Indian children so short? The role of birth order and son preference. Am. Econ. Rev. 107 , 2600–2629 (2017).

Özaltin, E., Hill, K. & Subramanian, S. V. Association of maternal stature with offspring mortality, underweight, and stunting in low- to middle-income countries. JAMA 303 , 1507–1516 (2010).

Desai, S., Vanneman, R. & National Council of Applied Economic Research. India Human Development Survey (IHDS), 2005 (ICPSR 22626). Inter-University Consortium for Political and Social Research . https://doi.org/10.3886/ICPSR22626.v12 (2018).

Desai, S., Vanneman, R. & National Council of Applied Economic Research. India Human Development Survey-II (IHDS-II), 2011-12 (ICPSR 36151). Inter-University Consortium for Political and Social Research. https://doi.org/10.3886/ICPSR36151.v6 (2018).

Deaton, A. Panel data from time series of cross-sections. J. Econ . https://doi.org/10.1016/0304-4076(85)90134-4 (1985).

Cotlear, D. Going Universal. World Bank https://openknowledge.worldbank.org/bitstream/handle/10986/22011/9781464806100.pdf (2015).

D’Silva, J. Can India pull off its ambitious National Health Mission? BMJ https://doi.org/10.1136/bmj.f2134 (2013).

Bosworth, B. & Collins, S. M. Accounting for growth: comparing China and India. J. Econ. Perspect . https://doi.org/10.1257/jep.22.1.45 (2008).

Chakrabarti, S., Raghunathan, K., Alderman, H., Menon, P. & Nguyen, P. India’s integrated child development services programme; equity and extent of coverage in 2006 and 2016. Bull. World Health Organ . https://doi.org/10.2471/BLT.18.221135 (2019).

Chakrabarti, S., Kishore, A. & Roy, D. Effectiveness of food subsidies in raising healthy food consumption: public distribution of pulses in India. Am. J. Agric. Econ . https://doi.org/10.1093/ajae/aay022 (2018).

Kim, R., Mohanty, S. K. & Subramanian, S. V. Multilevel geographies of poverty in India. World Dev. 87 , 349–359 (2016).

Muralidharan, K. & Kremer, M. Public-private schools in rural India. School Choice International. https://doi.org/10.7551/mitpress/9780262033763.003.0005 (2013).

Bernal, J. L., Cummins, S. & Gasparrini, A. The use of controls in interrupted time series studies of public health interventions. Int. J. Epidemiol . https://doi.org/10.1093/ije/dyy135 (2018).

Shadish, W., Cook, T. & Campbell, D. Quasi-experimental designs that use both control groups and pretests. Experimental and Quasi-Experimental Designs (2002).

Bernal, J. L., Cummins, S. & Gasparrini, A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int. J. Epidemiol . https://doi.org/10.1093/ije/dyw098 (2017).

Headey, D., Hoddinott, J. & Park, S. Accounting for nutritional changes in six success stories: a regression-decomposition approach. Glob. Food Security 13 , 12–20 (2017).

International Food Policy Research Institute (IFPRI) & University of Washington. Intergenerational Nutrition Benefits of India’s National School Feeding Program. https://doi.org/10.7910/DVN/JTN87W (2021).

Download references

Acknowledgements

We acknowledge feedback from the participants of the following conferences where drafts of the paper were presented: Nutrition 2018 (organized by the American Society for Nutrition), North East Universities Development Consortium (NEUDC) 2018, and the National Institute of Nutrition (NIN) 2018 Centenary conference. Bill & Melinda Gates Foundation through Partnerships and Opportunities to Strengthen and Harmonize Actions Against Malnutrition in India (POSHAN), led by the International Food Policy Research Institute (IFPRI).

Author information

Authors and affiliations.

Poverty Health and Nutrition Division, International Food Policy Research Institute, Washington, DC, USA

Suman Chakrabarti, Samuel P. Scott, Harold Alderman, Purnima Menon & Daniel O. Gilligan

You can also search for this author in PubMed   Google Scholar

Contributions

S.C. conceived the idea for the research. S.C. and S.P.S. contributed to all other aspects including data analysis, interpretation, and manuscript preparation and writing. H.A., P.M., and D.O.G. provided inputs to analysis and contributed to manuscript writing. All co-authors read and approved the final version of the manuscript.

Corresponding author

Correspondence to Samuel P. Scott .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information, peer review file, reporting summary, source data, source data, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Chakrabarti, S., Scott, S.P., Alderman, H. et al. Intergenerational nutrition benefits of India’s national school feeding program. Nat Commun 12 , 4248 (2021). https://doi.org/10.1038/s41467-021-24433-w

Download citation

Received : 21 February 2021

Accepted : 07 June 2021

Published : 12 July 2021

DOI : https://doi.org/10.1038/s41467-021-24433-w

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Loading metrics

Open Access

Peer-reviewed

Research Article

The effect of the Mid-Day Meal programme on the longitudinal physical growth from childhood to adolescence in India

Contributed equally to this work with: Shivani Gharge, Dimitris Vlachopoulos, Annie M Skinner, Craig A Williams, Raquel Revuelta Iniesta, Sayeed Unisa

Roles Conceptualization, Formal analysis, Methodology, Writing – original draft, Writing – review & editing

Affiliation Department of Mathematical Demography and Statistics, International Institute for Population Sciences, Mumbai, Maharashtra, India

Roles Conceptualization, Methodology, Supervision, Writing – review & editing

* E-mail: [email protected]

Affiliation Children’s Health and Exercise Research Centre, Faculty of Health and Life Sciences, University of Exeter Medical School, University of Exeter, Exeter, United Kingdom

Roles Conceptualization, Methodology, Writing – review & editing

Roles Conceptualization, Methodology, Supervision, Writing – original draft, Writing – review & editing

• Shivani Gharge, 
• Dimitris Vlachopoulos, 
• Annie M Skinner, 
• Craig A Williams, 
• Raquel Revuelta Iniesta, 
• Sayeed Unisa

• Published: January 11, 2024
• https://doi.org/10.1371/journal.pgph.0002742
• Reader Comments

The study aims to examine the effect of the world’s largest school-feeding programme, the Mid-Day Meal (MDM) programme, on the changes in the underweight prevalence among school-children in India. Data from the Indian Human Development Survey (IHDS) Rounds 1 (2004–05) and 2 (2011–12) were utilized. The sample included individual-level information of children aged 6 to 9 years in IHDS-1 who then turned 13 to 16 years in IHDS-2. The sample was categorised into four groups based on their MDM consumption history (Group 1: no MDM support in IHDS-1 and IHDS-2, Group 2: MDM support in IHDS-1, Group 3: MDM support in IHDS-2, Group 4: persistent MDM support in IHDS-1 and IHDS-2). The dependent variable was underweight status as defined by the World Health Organisation Child Growth Standards Body Mass Index for age (BMI Z-score) < -2 SD of the median. Bivariate analysis was used to examine the prevalence of underweight and establish associations between underweight status and socio-demographic characteristics. Logistic regression was performed to assess the strength of the association of socio-demographic characteristics and MDM consumption patterns with underweight across poor and non-poor asset groups. The findings suggest that early and persistent MDM support among respondents reduced the likelihood of low BMI Z-scores compared to those without MDM support. Respondents from the poor asset group who received MDM support in at least one of the two survey rounds had higher odds of being underweight in comparison with those who did not receive MDM support at all. Girls and adolescents residing in the Eastern region of India were less likely to be underweight. The study shows that the MDM programme was effective in reducing the rate of underweight among school children. However, continuous programme upscaling with a special focus on children from poor households will significantly benefit India’s school-aged children.

Citation: Gharge S, Vlachopoulos D, Skinner AM, Williams CA, Iniesta RR, Unisa S (2024) The effect of the Mid-Day Meal programme on the longitudinal physical growth from childhood to adolescence in India. PLOS Glob Public Health 4(1): e0002742. https://doi.org/10.1371/journal.pgph.0002742

Editor: Julia Robinson, PLOS: Public Library of Science, UNITED STATES

Received: June 13, 2023; Accepted: November 30, 2023; Published: January 11, 2024

Copyright: © 2024 Gharge et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The Inter-University Consortium for Political and Social Research (ICPSR) data repository retains all the Indian Human Development Survey (IHDS) datasets that were utilised in this study. The data can be accessed at https://www.icpsr.umich.edu/web/DSDR/series/507 . Additional information about the IHDS project is available on the https://ihds.umd.edu website.

Funding: We want to thank the UK Department for Business, Energy and Industrial Strategy (BEIS), British Council, UK Research and Innovation Council (UKRI) and Indian Council of Social Science Research (ICSSR) for supporting this research. This research article has arisen from the work undertaken during the Newton Bhabha PhD Placement Programme 2020-2021. Shivani Gharge (SG) was the recipient of this award with File No: ICSSR-BC (UK)/NBF/Ph D-08/2020-IC. URL to sponsors’ website: https://www.britishcouncil.in/programmes/higher-education/newton-fund/phd-placements . The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Childhood to adolescence is a period of rapid growth and development, which is highly dependent on optimal nutrient and energy intake [ 1 ]. Inadequate dietary intake during this critical period of growth may lead to undernutrition. Undernutrition can cause irreversible stunting, higher risk of infections and compromise organ development including the brain, which may affect physical, emotional and social wellbeing [ 2 ]. Therefore, nutritional interventions targeting school-aged children have a significant impact on health and assist in uplifting physical and/or mental health benefits, thereby improving children’s chances for a better future [ 3 ].

Undernutrition across various socioeconomic strata is the most challenging issue faced by children and adolescents all over the world [ 4 ]. Body Mass Index (BMI) is a measure of acute nutritional status and is currently considered the gold standard to assess nutritional status in children and adolescents [ 5 – 7 ]. According to World Health Organization (WHO) child growth standards, children, and adolescents whose BMI Z-score is below -2 SD are considered underweight [ 5 , 6 ]. The prevalence of children and adolescents who were underweight declined from 9.2% in 1975 to 8.4% in 2016 among girls and from 14.8% to 12.4% among males [ 8 ]. Globally, around 75 million girls and 117 million boys were moderately or severely underweight in 2016 [ 9 ]. The prevalence of underweight among children and adolescents was highest in India in 2016, at 22.7% among girls and 30.7% among boys, and it has not decreased substantially in the last three decades [ 8 ]. According to the Comprehensive National Nutrition Survey (CNNS) India 2016–18 report, 10% of school-age (5–9 years) children and 47% of late adolescent girls aged 15–19 years were underweight in India [ 10 ]. Undernutrition significantly compromises optimal growth and development of children and is associated with a higher risk of infectious diseases, reducing the ability to learn, lowering school performance [ 11 , 12 ]. In the long term, it increases the risk of non-communicable diseases e.g., cardiovascular diseases, diabetes and osteoporosis, is associated with adverse pregnancy outcomes and importantly affects economic productivity [ 12 , 13 ].

In 1995, the Government of India launched the National Programme of Nutritional Support to Primary Education (NSPE) as a centrally sponsored scheme with the primary objective of improving enrolment, retention, and attendance with a simultaneous effort to improve the nutritional status of schoolchildren [ 14 ]. On November 28, 2001, the Supreme Court directed all the State Governments/Union territories to implement the Mid-Day Meal (MDM) Scheme, in which every child in every Government and Government aided school was to be served a cooked meal with a minimum content of 300 kilocalories and 8–12 gram protein per day for a minimum of 200 days per year [ 14 ]. The majority of the Indian states began providing cooked and warm meals by 2003, and eventually, around 120 million students were covered under the MDM by 2006, which is now regarded as the world’s largest school feeding programme [ 15 ].

The fundamental aim of this programme is to increase school enrolment, retention, and attendance of children in India by providing free cooked meals for lunch on working days to children in primary and upper primary classes in government, government-aided, local body, Education Guarantee Scheme. Alternate innovative education centres, Madrassa and Maqtabs supported under Sarva Shiksha Abhiyan, and National Child Labour Project Scheme schools run by the Ministry of Labour and Employment are also included [ 14 ]. Since the inception of MDM, the programme has undergone several transformations before reaching the current phase. The MDMs appear to have substantially contributed to overcoming classroom hunger, and have been a huge help to low-income families, relieving them of the responsibility of providing a one-time meal to their children [ 16 , 17 ]. Universal primary education has been achieved in the last decade, but enrolment has only improved marginally [ 18 ]. The MDM programme has led to an increase in school participation rate from marginalised households [ 18 ], an improvement in dietary and total energy intake (TEI) during school days only [ 16 , 17 ] and improvement in weight-for-age (WFA) and height-for-age (HFA) [ 19 , 20 ].

Benefits from the MDM programme have shown significant “improvement” in WFA and HFA in children during severe droughts [ 20 ]. However, MDM’s longitudinal effects on children’s physical growth as pupils transition from MDM beneficiaries to non-beneficiaries and vice versa, has not been investigated [ 21 ], especially at the national level. Therefore, this study sought to add new knowledge to the existing research and contribute to the larger literature on school feeding, with a special focus on the effect of school feeding on underweight. Employing a longitudinal data set from Rounds 1 and 2 of the Indian Human Development Survey (IHDS-1 and 2), the present study aimed to examine the changes in the prevalence of underweight (defined as BMI-for-age < -2 SD of the WHO Child Growth Standards median) in schoolchildren aged 6 to 9 years in IHDS-1 and 13 to 16 years in IHDS-2 as they transition from MDM beneficiaries to non-beneficiaries and vice versa. The secondary aim was to examine the prevalence of underweight status according to different socio-demographic characteristics and determine the predictors of underweight.

Ethics statement

This study employed a publicly available longitudinal secondary dataset that had no information that may lead to the respondents’ identification. The Inter-University Consortium for Political and Social Research (ICPSR) data repository retains all the IHDS datasets that was utilised in this study [ 22 , 23 ].

Materials and methods

Data source.

India has one of the largest education systems in the world, with more than 1.4 million schools, 9.7 million teachers, and 265 million children [ 24 ]. To understand the effect of the MDM programme on underweight in India, we utilise data from the Indian Human Development Survey (IHDS). The IHDS is an initiative through a collaborative programme by researchers from the National Council of Applied Economic Research, New Delhi, and the University of Maryland. The data for Round 1, IHDS-1, was collected in 2004–05, and for Round 2, IHDS-2, it was collected in 2011–12. The survey yields information pertaining to the key dimensions of human development indicators and a series of quantifiable variables measuring wider contexts. A nationally representative, multitopic survey encompassing 41,554 households from 1,504 villages and 970 urban neighbourhoods in Round 1 [ 22 ] and 42,152 households from 1503 villages and 971 urban neighbourhoods in Round 2 [ 23 ] across India. Around 83%, i.e., 34,621 households, were re-interviewed in Round 2 along with some split households residing in the same community. IHDS collects extensive information on sociodemographic characteristics, education, fertility, health, agriculture, energy use, and utilisation of large public programmes like the Integrated Child Development Services and Public Distribution System at the national level.

Study design

We utilized panel data from Rounds 1 (2004–05) and 2 (2011–12) of the IHDS, and our final sample size was 3,199 (1,638 girls and 1,561 boys). For the analytical purposes of our study, data was restricted to the individual-level information of children aged 6 to 9 years in IHDS-1 who then turned 13 to 16 years in IHDS-2 and were currently attending government, government-aided, EGS, and Madrassa schools. We included schoolchildren from the above-mentioned age-group because the MDM scheme was covering primary school children (classes I to IV) and since 2008 the programme covers all children studying in Government, Local Body and Government-aided primary and upper primary schools and the Education Guarantee Scheme schools or alternate innovative education centres including Madrassa and Maqtabs supported under SSA of all areas across the country [ 14 ]. The percentage of loss to follow-up for all the observations was 17% [ 22 , 23 ], and for our study sample it was 25%, which was very low; therefore, we excluded individuals lost to recontact for IHDS-2 from the study sample. Details of the final sample of re-interviewed respondents and the sample selection process are presented in Fig 1 (also see [ 22 , 23 , 25 , 26 ] for further details on the sample of re-interviewed households).

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

(A) Details of the final sample of re-interviewed respondents and the sample selection process are presented in this figure; (B) Abbreviations: IHDS, Indian Human Development Survey; N: Frequency; EGS, Education Guarantee Scheme; MDM, Mid-Day Meal.

https://doi.org/10.1371/journal.pgph.0002742.g001

Variable description

Outcome variable..

The main outcome was underweight defined as BMI-for-age (BMI Z-score). BMI Z-score is currently considered the gold standard to assess nutritional status in children and adolescents [ 5 , 6 ]. Nutritional status is defined as underweight (BMI Z-score < -2 SD), normal weight (BMI Z-score -2 SD to ≤ 1 SD), overweight (> 1 SD to ≤ 2SD) and obese (BMI Z-score > 2SD). Data on the height and weight of the respondents were recorded in IHDS-1, and followed up in IHDS-2. BMI-for-age Z-scores were calculated, and the results were classified as underweight if their BMI-for-age was more than two standard deviations below (< -2 SD) the WHO Child Growth Standards median [ 5 , 6 ].

Explanatory variables.

In IHDS-1, MDM consumption was defined as those who received grain, Dalia and/or a variety of meals. In IHDS-2, MDM consumption was defined as those who received school meals regularly and irregularly. The participants were categorized into four groups according to their response to the question on MDM consumption from IHDS-1 and IHDS-2; group one (No MDM support in IHDS-1 and IHDS-2), group two (MDM support in IHDS-1 only), group three (MDM support in IHDS-2 only), and group four, (MDM support in both IHDS-1 and IHDS-2).

The socio-demographic characteristics include sex, household size stratified as ≤ 4 members, 5 to 8 members and ≥ 9 members, asset groups (poorest, poor, middle and rich) [ 27 , 28 ], household adult’s education (all are illiterate, at least one completed primary, at least one completed secondary, at least one completed higher), place of residence (urban and rural), religion (Hindu, Muslim, Christian and other) and region (North, Central, East, Northeast, West, South).

Statistical analyses

Descriptive statistics were calculated to show the mean height (cm), weight (kg) and BMI Z scores of the study sample. Bivariate analyses were performed to estimate the prevalence of underweight (BMI Z score < -2 SD) among children and adolescents aged 6 to 9 years in IHDS-1 who then turned 13 to 16 years in IHDS-2 in India by transition in MDM consumption as well as for the categories of independent variables. Chi-square tests were used to determine whether independent variables, sex, household size, asset group, household adult’s education, place of residence, religion and region had significant associations with the prevalence of underweight status at p < 0.05. The asset group has not changed over time from IHDS-1 (2004–2005) to IHDS-2 (2011–2012) which is why we have considered only the sample from IHDS-2 (2011–2012) in our multivariate analysis ( Table 3 ). Two separate multivariate models were then fitted to discern the extent to which poor and non-poor asset groups, and in combination with MDM consumption pattern and sociodemographic factors, explain the associations with underweight (BMI Z score < -2 SD). Model 1 is controlled for MDM consumption pattern, socio-demographic variables and poor asset group category, whereas, Model 2 is controlled for MDM consumption pattern, socio-demographic variables and non-poor asset group category.

Socio-demographic characteristics of the study population

Descriptive characteristics of the study sample are shown in Table 1 . In this study sample of 3,199 respondents, 50% of the study participants were girls (IHDS-1: n = 1638, 50.45%; IHDS-2: n = 1638, 50.31%) and 50% were boys (IHDS-1: n = 1561, 49.55%; IHDS-2: n = 1561, 49.69%). Around 64% (n = 2088) and 61% (n = 2030) of the study participants had a household size of 5 to 8 members in IHDS-1 and IHDS-2 respectively. At least one adult member in 46% to 48% of the households had completed secondary level education. The highest proportion of study participants were Hindu by religion (IHDS-1: n = 2719, 85.87%; IHDS-2: n = 2710, 85.25%). Majority of the sample resided in rural areas with 81% (n = 2453) rural sample in IHDS-1 and 85% (n = 2452) rural sample in IHDS-2. The Northern region formed the bulk of the study sample, followed by the Eastern and Western regions.

https://doi.org/10.1371/journal.pgph.0002742.t001

Table 2 shows the underweight prevalence among school-aged children by socioeconomic characteristics and the change in MDM consumption status from Rounds 1 and 2 of the IHDS. The findings indicate that underweight prevalence is highest among school-aged children who have received MDM support at an early age of 6 to 11 years in IHDS-1 (IHDS-1: 22.36%; IHDS-2: 18.40%), and among those who have received persistent MDM support (IHDS-1: 22.31%; IHDS-2: 21.68%). The highest decline of around 4% in underweight prevalence can be observed among those respondents who received MDM support during their early ages. In group 1 (No MDM support at IHDS-1 and IHDS-2) there is a statistically significant relationship between underweight status and sex, household size and region. In group 2 (MDM support at IHDS-1 only), underweight status is dependent on sex, education of adult members in the household and region. In group 4 (MDM support at IHDS-1 and IHDS-2), underweight status is significantly associated with region.

https://doi.org/10.1371/journal.pgph.0002742.t002

Tables 3 and 4 show whether there is a significant increase or decrease in BMI Z scores from IHDS-1 to IHDS-2 within the four groups and whether that change is different between the groups. The ANOVA tests show that there was a significant decrease in BMI Z scores from IHDS-1 to IHDS-2 within group 3 (-0.46 ± 1.87 for IHDS-1 and -0.93 ± 1.39 for IHDS-2) and 4 (-0.86 ± 1.61 for IHDS-1 and -1.1 ± 1.37 for IHDS-2). In case of boys, there was a significant decrease in BMI Z scores from IHDS-1 to IHDS-2 within groups 2 (-1.00 ± 1.53 for IHDS-1 and -1.19 ± 1.26 for IHDS-2), 3 (-0.19 ± 2.08 for IHDS-1 and -1.02 ± 1.47 for IHDS-2) and 4 (-0.81 ± 1.73 for IHDS-1 and -1.21 ± 1.47 for IHDS-2). The ANOVA also confirms that the change in the BMI Z scores is also different between the groups. Respondents from group 2 (MDM support at IHDS-1 only) and 4 (MDM support at IHDS-1 and IHDS-2) are 36% and 39% significantly less likely to have a low BMI Z score as compared to respondents from group 1 (No MDM support at IHDS-1 and IHDS-2).

https://doi.org/10.1371/journal.pgph.0002742.t003

https://doi.org/10.1371/journal.pgph.0002742.t004

Table 5 shows the change in the asset group category of the respondents from IHDS-1 to IHDS-2 in percentages. The primary outcome was to observe whether there has been a major significant shift in the asset group category of respondents from IHDS-1 to IHDS-2. Chi-square test shows a significant association between asset group in IHDS-1 and asset group in IHDS-2. There is a small shift among respondents from the poorest asset group to the poor asset group and vice versa. Around 26% of respondents shifted from the poorest asset group category in IHDS-1 to poor asset group category in IHDS-2. Vice versa, around 31% of respondents who belonged to the poor asset group in IHDS-1, now fell into the poorest category in IHDS-2. However, the overall asset group category of respondents was consistent over time from IHDS-1 (2004–2005) to IHDS-2 (2011–2012).

https://doi.org/10.1371/journal.pgph.0002742.t005

Table 6 shows the results of the logistic regression analyses of the determinants of underweight children and adolescents among poor and non-poor asset groups from the IHDS-2 (2011–2012) after controlling for change in MDM consumption status, sex, household size, education of adult members in the household, religion, place of residence and region. Regression estimates reveal that study participants belonging to the poor asset group who received MDM support at an early age (OR: 2.03; CI: 1.06 to 3.87), at a late age (OR: 3.73; CI: 1.60 to 8.69), and also who received persistent MDM support (OR: 2.43; CI: 1.21 to 4.88) had higher odds of being underweight in comparison with those who did not receive MDM support at all. The magnitude of association between underweight status and change in MDM consumption status was lowest among adolescents receiving MDM support at IHDS-1. Adolescents from the non-poor asset group who received persistent MDM support (OR: 2.09; CI: 1.34 to 3.28) were more likely to be underweight as compared to those who did not receive any MDM support. Both poor (OR: 0.47; CI: 0.33 to 0.67) and non-poor (OR: 0.55; CI: 0.41 to 0.72) girls had lower odds of being underweight in comparison to boys. Adolescents residing in the Eastern region of India were less likely to be underweight in both poor (OR: 0.47; CI: 0.29 to 0.77) and Non-poor (OR: 0.31; CI: 0.16 to 0.58) asset group categories compared with their counterparts living in the Northern region.

https://doi.org/10.1371/journal.pgph.0002742.t006

In this study, we have assessed the impact of the world’s largest school feeding programme using a nationally representative data on the underweight prevalence according to the transition in MDM consumption among children and adolescents aged 6 to 9 years in IHDS-1 who then turned 13 to 16 years in IHDS-2 in India. The original and important findings indicate that the MDM programme had a positive and significant impact on lowering the underweight prevalence of the beneficiaries. However, after controlling for certain background characteristics, the odds of being underweight were significantly higher for those who received MDM support in at least one of the two survey rounds in comparison with those who did not receive MDM support at all. Children consuming MDM at younger ages (6 to 9 years) were less affected by underweight. Adolescent girls in the age group 13 to 16 years were less likely to be underweight than boys, regardless of their socioeconomic status.

Study participants from the poor asset group who received MDM support at an early age, late age, or who received persistent MDM support were more likely to be underweight as compared to those who did not receive MDM support at all. A review by Jomaa (2011) showed that previous studies have revealed mixed findings regarding how school feeding affects children’s weight, height, and BMI gains [ 29 ]. Prior research by Jacoby (2002) and Afridi (2010) focused on the nutritional intake outcomes of MDM beneficiaries, whereas Singh, Park and Dercon (2014) is the only study that focused on the outcome indicators of child nutrition i.e., the anthropometric z-scores on two measures, WFA and HFA. Also, the mean age of the study sample considered in these earlier studies was 4.7 to 8.5 years [ 20 , 30 , 31 ], whereas the current study covers a broader age group of children ages ranging from 6 to 9 years in IHDS-1 who then turned 13 to 16 years in IHDS-2 and attended government, government aided, EGS, and Madrassa schools. Also, the study participants when last surveyed in IHDS-2 were currently in school but may not be consuming MDM since the scheme was extended to cover children in upper primary classes (i.e., classes VI to VIII) in April 2008 [ 14 ].

The magnitude of the association between change in MDM consumption status and being underweight was lowest among adolescents receiving MDM support only in IHDS-1 (2004–05), indicating that MDMs were more effective in reducing underweight prevalence among children when consumed only at younger ages. At the time of IHDS-1 in 2004–05, children in our analytical group were in the age range of 6 to 9 years and would have received the MDM for a minimum duration of about 1 year and a maximum of 4 years. This result indicates that dietary interventions are more successful in childhood (age 6 to 9 years) than in adolescence (age 13 to 16 years) because there is potential for catch-up if conditions are made better, for instance through nutritional supplementation when children are still young [see, 32 – 34 ]. Importantly, the period between 5 and 9 years of age is a time of continued growth and development and children are affected by multiple forms of malnutrition [ 35 ]. Therefore, dietary interventions may directly alter the intake of children through the school lunch programme, since schools are key settings that provide access to a large proportion of children for prolonged periods [ 36 , 37 ]. Moreover, adolescents from the non-poor asset group who received persistent MDM support were also more likely to be underweight as compared to those who did not receive any MDM support. These results, however, differ from those of other earlier research, which revealed that children from higher-income families have access to a number of healthier food options through their home meals along with the MDM provided in schools, and hence participants from the Non-poor asset group category have a better nutritional status [ 27 , 38 ].

Adolescent girls aged 13 to 16 years were less likely to be underweight than boys, regardless of their socioeconomic status. These findings are in accordance with some of the previous studies worldwide and in India, which report a sex difference in the prevalence of underweight, with boys having a higher prevalence compared to girls both among children and adolescents [ 9 , 10 ]. Compared to their counterparts living in the Northern area, adolescents living in India’s Eastern region were less likely to be underweight in both the poor and non-poor asset group categories. The result from the current study is consistent with a study performed using the CNNS (2016–18) data which reported that adolescents from Eastern India were at decreased odds of thinness compared to adolescents from Northern India [ 39 ].

There are some limitations to the study which should be considered. First and foremost, as these factors influence children’s underweight status, data on school-related features, including adequate water and sanitation facilities in schools, maternal characteristics, inflammation and infectious disease history, and dietary diversity could have offered more insight. Secondly, the meals provided 300 kilocalories and 8–12 gram protein per day; however, we were unable to estimate the full nutritional composition (fats, carbohydrates and micronutrients) of the MDM due to the lack of data in the survey. This may influence the effectiveness of the MDM programme in reducing the prevalence of underweight. There was a drop period (between 9 to 13 years of age) where the participants were not examined between IHDS-1 and IHDS-2, and we could not evaluate the underweight or the MDM consumption status of the children during that period. Lastly, the duration of the research participants’ MDM support would have been another crucial component to include when studying the relationship between MDM consumption status and underweight.

Undernutrition is more common in early childhood and is also likely to persist through adolescence into adulthood, and therefore, this "second opportunity" for catch-up growth during adolescence should not be missed [ 28 ]. Schools providing cooked meals are mostly government or government-aided schools where the cost of schooling is generally lower, which attracts children from the lower economic strata. There is extensive literature that states that children attending government schools and belonging to lower socioeconomic strata are more likely to be undernourished, and thus, for the vulnerable sections of the country, a scheme like this can serve its purpose [ 7 , 28 , 40 ]. The estimates drawn from these large datasets could help policymakers determine the extent to which operational goals are met and set priorities to facilitate target-based decision making. In continuation of this study, we would be analyzing the data from the IHDS using a Generalized Estimating Equations (GEE) model. For the MDM programme to reflect effectively on its beneficiaries, extensive use of available data for monitoring every stage of the programme through longitudinal comparison of the indicators will appropriately demonstrate the effectiveness of the intervention.

This study has shown that the MDM programme administered to the beneficiaries was effective in reducing the rate of underweight (defined as BMI-for-age < - 2 SD). However, the MDM programme was not effective in reducing the underweight prevalence among beneficiaries from the poor asset group. Children consuming MDM at younger ages (6 to 9 years) were less affected by underweight. Adolescent girls in the age group 13 to 16 years were less likely to be underweight than boys, regardless of their socioeconomic status. However, continuous programme upscaling with a special focus on children from poor households will significantly benefit India’s school-aged children. Given the Indian context, this is one of the few attempts at a careful assessment of a programme using a nationally representative dataset, and these original findings, alongside with other research on the beneficial effects of school meals on school enrolment, attendance, and daily nutrient intake, offer empirical support for the advantages of the programme in India.

These findings could be taken to support a broad focus by the government, thus providing a basis for potential new policy recommendations, tackling the dual and triple burden of malnutrition, and implementing programmes in the early years to instill healthy lifelong eating habits. Moreover, this research should lead to further research focusing on individuals’ physical growth outcomes, thus giving way to sub- programmes focusing on the factors emerging from the current study. The wider applicability of these findings could help the central and state governments identify regional- and state-specific measures, thereby positively impacting growth outcomes and future adult health amongst Indian children and adolescents.

Supporting information

A . Results of logistic regression showing the determinants of underweight children and adolescents among poor asset groups from IHDS-2 (2011–2012). Note: (a) This model is adjusted for change in MDM consumption status and sociodemographic factors such as sex, household size, Education of adult members in the household, religion, place of residence and region. (b) Poorest and poor asset groups were combined to create Poor Asset group. (c) Respondent was considered underweight if BMI-for-age was more than two standard deviations below (< -2SD) the WHO Child Growth Standards median. (d) Ref. denotes reference category. (e) The z value is the ratio of the estimated coefficient to its standard error and it measures the number of standard deviations that the estimated coefficient is away from 0. (f) The P >|z| column represents the p-value for each coefficient. A significance level of 0.05 indicates a 5% risk of concluding that an association exists between the dependent and independent variables. In these results, the odds ratio of 2.03 for Group 2 is statistically significant at the significance level of 0.05, therefore, Group 2 beneficiaries are 2.03 times more likely to be underweight. (g) Christian and others category has very few respondents. (h) Abbreviations: MDM, Mid-Day Meal; HH, Household; IHDS, Indian Human Development Survey. B . Results of logistic regression showing the determinants of underweight children and adolescents among non-poor asset groups from IHDS-2 (2011–2012). Note: (a) This model is adjusted for change in MDM consumption status and sociodemographic factors such as sex, household size, Education of adult members in the household, religion, place of residence and region. (b) Middle and rich asset groups were combined to create Non-poor Asset group. (c) Respondent was considered underweight if BMI-for-age was more than two standard deviations below (< -2SD) the WHO Child Growth Standards median. (d) Ref. denotes reference category. (e) The z value is the ratio of the estimated coefficient to its standard error and it measures the number of standard deviations that the estimated coefficient is away from 0. (f) The P >|z| column represents the p-value for each coefficient. A significance level of 0.05 indicates a 5% risk of concluding that an association exists between the dependent and independent variables. In these results, the odds ratio of 2.09 for Group 4 is statistically significant at the significance level of 0.05, therefore, Group 4 beneficiaries are 2.09 times more likely to be underweight. (g) Abbreviations: MDM, Mid-Day Meal; HH, Household; IHDS, Indian Human Development Survey.

https://doi.org/10.1371/journal.pgph.0002742.s001

S1 Checklist. STROBE statement—Checklist of items that should be included in reports of observational studies.

Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/ , Annals of Internal Medicine at http://www.annals.org/ , and Epidemiology at http://www.epidem.com/ ). Information on the STROBE Initiative is available at www.strobe-statement.org .

https://doi.org/10.1371/journal.pgph.0002742.s002

• View Article
• PubMed/NCBI
• Google Scholar
• 5. World Health Organization. Growth reference data for 5–19 years. Geneva: World Health Organization. 2006. https://www.who.int/tools/growth-reference-data-for-5to19-years
• 10. Ministry of Health and Family Welfare (MoHFW), Government of India, UNICEF and Population Council. Comprehensive National Nutrition Survey (CNNS) National Report, 2016–18. New Delhi, India; 2019.: https://nhm.gov.in/WriteReadData/l892s/1405796031571201348.pdf
• 13. United Nations Children’s Fund. Strategy for improved nutrition of children and women in developing countries. New York; 1990. https://digitallibrary.un.org/record/227230/files/UNE_ICEF%2802%29_S771-E.pdf
• 14. Pradhan Mantri Poshan Shakti Nirman (PM POSHAN), Ministry of Education, Government of India [Internet]. About the PM POSHAN. [cited 2023 July 15]. https://pmposhan.education.gov.in
• 22. Desai S, Vanneman R. India Human Development Survey (IHDS), 2005: Version 12 [dataset]. 2008 [cited 23 Dec 2019]. Inter-University Consortium for Political and Social Research. https://www.icpsr.umich.edu/web/DSDR/studies/22626/versions/V12 Referenced in https://doi.org/10.3886/ICPSR22626.v12
• 23. Desai S, Vanneman R. India Human Development Survey-II (IHDS-II), 2011–12: Version 6 [dataset]. 2015 [cited 23 Dec 2019]. Inter-University Consortium for Political and Social Research. https://www.icpsr.umich.edu/web/DSDR/studies/36151/versions/V6 Referenced in https://doi.org/10.3886/ICPSR36151.v6
• 24. Ministry of Education, Department of School Education and Literacy and Government of India. Unified District Information System For Education Plus (Udise+) Report, 2021–22. New Delhi, India. https://udiseplus.gov.in/#/page/publications
• 25. Desai S, Dubey A, Joshi BL, Sen M, Sharif A, Vanneman R. India Human Development Survey Users’ Guide Release 03. University of Maryland and National Council of Applied Economic Research, New Delhi; 2010. https://www.icpsr.umich.edu/icpsrweb/content/DSDR/idhs-data-guide.html
• 26. Desai S, Dubey A, Vanneman R. India Human Development Survey-II Users’ Guide Release 01. University of Maryland and National Council of Applied Economic Research, New Delhi; 2015. https://www.icpsr.umich.edu/icpsrweb/content/DSDR/idhs-II-data-guide.html
• 32. Tanner JM. A history of the study of human growth. Cambridge university press; 1981 Aug 13.

• Corpus ID: 54947277

Future of Mid-Day Meals

• J. Drèze , A. Goyal
• Published 2003
• Economics, Agricultural and Food Sciences
• Economic and Political Weekly

172 Citations

School meals as a safety net: an evaluation of the midday meal scheme in india, mid day meals: a detailed study of indian states.

• Highly Influenced
• 18 Excerpts

How does mid-day meal scheme shape the socialization value in rural India?

An evaluation of mid -day meal scheme, full meal or package deal, mid day meal-not a sufficient deal, efficacy of mid-day meal scheme in india: challenges and policy concerns, an empirical study of the mid-day meal programme in khurda, orissa, implementation of mid-day meal scheme in government elementary schools of bihar, reaching out of noon meal scheme in india, 17 references, school participation in rural india, right to food in india, related papers.

Showing 1 through 3 of 0 Related Papers

• My Shodhganga
• Receive email updates
• Edit Profile

Shodhganga : a reservoir of Indian theses @ INFLIBNET

• Shodhganga@INFLIBNET
• YBN University

Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

The PMC website is updating on October 15, 2024. Learn More or Try it out now .

• Advanced Search
• Journal List

The Influence of Meal Frequency and Timing on Health in Humans: The Role of Fasting

Antonio paoli.

1 Department of Biomedical Sciences, University of Padova, 35131 Padova, Italy

2 Faculty of Sport Sciences, UCAM, Catholic University of Murcia, 30107 Murcia, Spain

Grant Tinsley

3 Department of Kinesiology & Sport Management, Texas Tech University, Lubbock, TX 79409, USA; [email protected]

Antonino Bianco

4 Department of Psychology, Educational Science and Human Movement, Sport and Exercise Sciences Research Unit, University of Palermo, 90144 Palermo, Italy; [email protected]

Tatiana Moro

5 Department of Nutrition and Metabolism, University of Texas Medical Branch, Galveston, TX 77550, USA; ude.bmtu@oromat

6 Sealy Center on Aging, University of Texas Medical Branch, Galveston, TX 77550, USA

The influence of meal frequency and timing on health and disease has been a topic of interest for many years. While epidemiological evidence indicates an association between higher meal frequencies and lower disease risk, experimental trials have shown conflicting results. Furthermore, recent prospective research has demonstrated a significant increase in disease risk with a high meal frequency (≥6 meals/day) as compared to a low meal frequency (1–2 meals/day). Apart from meal frequency and timing we also have to consider breakfast consumption and the distribution of daily energy intake, caloric restriction, and night-time eating. A central role in this complex scenario is played by the fasting period length between two meals. The physiological underpinning of these interconnected variables may be through internal circadian clocks, and food consumption that is asynchronous with natural circadian rhythms may exert adverse health effects and increase disease risk. Additionally, alterations in meal frequency and meal timing have the potential to influence energy and macronutrient intake.A regular meal pattern including breakfast consumption, consuming a higher proportion of energy early in the day, reduced meal frequency (i.e., 2–3 meals/day), and regular fasting periods may provide physiological benefits such as reduced inflammation, improved circadian rhythmicity, increased autophagy and stress resistance, and modulation of the gut microbiota

“Eat like a king in the morning, a prince at noon, and a peasant at dinner” (Moses ben Maimon or Maimonides. 1135-1404)

1. A Brief Historical Introduction

In Western culture, it is a common idea that the daily food intake should be divided into three square meals: breakfast, lunch, and dinner. Often dieticians suggest adding two snacks (morning and afternoon) to help appetite control, and indeed the mainstream media message is to eat “five to six times a day”. However, the number of meals is not a universal standard, and the traditional three square meals are, somewhat surprisingly, a recent behaviour. As an example, the Ancient Romans had only one substantial meal, usually consumed at around 16:00 h ( coena ), and they believed that eating more than once per day was unhealthy. Although they also ate in the morning ( ientaculum ) and at noon ( prandium ), these meals were frugal, light and quick [ 1 ]. Later, Monastic rules influenced common peoples’ eating behaviour. The term breakfast means “break the night’s fast”, pointing out that it is the first meal after the evening/night devoted to prayer [ 2 ]. In the early medieval times, monks were obliged to remain silent during meals while one of them read aloud a religious text. One of the most-read texts was the Collationes (compilation) by Giovanni Cassiano, and it is worth mentioning that the Italian term for breakfast is “colazione ”, which is derived precisely from the Latin word “ collationes ” [ 3 ]. Breakfast also became important during the industrial revolution as a meal consumed before going to work. Dinner in its current form and timing became popular after the widespread use of artificial light, which facilitated eating before dawn and after dark [ 3 ].

2. Meal Frequency

2.1. epidemiological studies about the effects of meal frequency on cholesterol, body weight and diabetes.

The origin of the firm belief that eating three meals per day is the better healthy choice is a mix of cultural heritage [ 4 , 5 , 6 ] and early epidemiological studies [ 7 ]. The available epidemiological studies have not primarily investigated cardiovascular diseases (CVDs), but rather some risks factors such as cholesterol and body weight [ 8 , 9 ]. These studies observed a worsening of blood lipids associated with a “gorging” (a reduced meal frequency, one or two meals daily) diet compared to “nibbling” (the consumption of frequent smaller meals or snacks). In these early studies, authors stated that a reduced meal frequency is associated with an increased risk of cardiovascular disease [ 10 ]. Subsequent studies seemed to confirm these previous findings, reporting a lower age-adjusted total and LDL (low-density lipoprotein) cholesterol in subjects who reported eating four or more meals daily, compared to those reporting one or two [ 11 ]. The association was also confirmed after adjustment for alcohol, smoking, systolic blood pressure, anthropometric measurements as WHR (waist to hip ratio) and BMI (body mass index), and macronutrient intake. In a 1989 paper, authors compared a very high frequency of meals (17) to a lower frequency (3) and found an improvement of total and LDL cholesterol with the higher frequency; however, this particular approach is clearly atypical in ordinary life [ 12 ]. A recent study within the European Prospective Investigation into Cancer (EPIC) project showed a lower concentration of total and LDL cholesterol in subjects reporting a higher (≥6 times/day) meal frequency compared to those who ate 1 or 2 times a day, even when adjusted for age, BMI, physical activity, smoking, total energy intake, and macronutrient distribution [ 13 ]. Again, a recently published cross-sectional analysis within the prospective Seasonal Variation of Blood Cholesterol Study in Worcester County, Massachusetts (SEASONS) showed that a frequency higher than four times per day leads to a lower risk of obesity compared to a frequency lower than three times per day, even after adjustment for age, sex, physical activity, and total energy intake [ 14 ].

Another large cohort study, the Malmo Diet and Cancer study, reported that eating more than six meals per day reduces the risk of obesity compared to less than three meals daily; moreover, after adjustment for diet and lifestyle, frequent eaters had lower waist circumference [ 15 ]. Regarding diabetes, a 16-year follow-up study showed an increased risk of type 2 diabetes mellitus in men who ate 1–2 times a day compared to those who ate three meals a day (relative risk RR 1.26) after adjustments for age, BMI, and other relevant factors [ 16 ]. These data are in contrast to another study that found no correlation between increased meal frequency and type 2 diabetes risk in women after six years follow up (3 times a day: RR 1.09, ≥6 times a day: RR 0.99) [ 17 ]. Despite the numerous studies examining risk factors, only one prospective cohort study investigated the relationship between meal frequency and coronary heart disease (CHD) risk. Cahill et al. [ 18 ] found that men eating 1–2 meals per day hadan RR for CHD of 1.10, men eating 4–5 meals per day hadan RR of 1.05, and men eating ≥6 times hadan RR 1.26, as compared to who ate three times a day after adjustment for total energy intake, diet composition, and other risk factors. In general, conflicting results are depending on the outcome investigated and the methodology used.

However, as also suggested by other authors [ 19 , 20 ], the correlation between a reduced meal frequency and a higher risk of CHD in these studies appears to be weak considering the cross-sectional nature of these studies, making it difficult to establish the causality or temporality of this association.

2.2. Meal Frequency and Weight Control: One, Two, Three, or More Meals?

Obesity is a rapidly growing epidemic worldwide; its prevalence has nearly doubled in more than 70 countries since 1980. In 2015, a total of 107.7 million children and 603.7 million adults were obese [ 21 ]. Seventy-five percent of the world’s population live in countries where overweight and obesity kills more people than underweight [ 22 ]. Obesity is one of the main risk factors for cardiovascular disease, along with dyslipidemia and hypertension [ 23 ]. As a part of the strategies proposed for reducing energy intake (diets, drugs, and bariatric surgery) [ 24 ] and for increasing energy output (exercise and non-exercise movement) [ 25 ], meal timing and frequency could exert a significant influence on weight control and weight loss. [ 26 , 27 ]

A very recent and extensive study published by Kahleova and colleagues [ 28 ] investigated 50,660 adult members of Seventh-day Adventist churches in the United States and Canada. The results showed that eating one or two meals daily was associated with a relatively lower BMI compared with three meals daily. Interestingly, they found a positive relationship between the number of meals and snacks (more than three daily) and increases in BMI. Furthermore, the change in BMI was related to the length of the overnight fast: the longer the overnight fast, the lower the BMI. Authors suggested that the positive effects of such nutritional regimen are due to the combination of timing, meal frequency, and long overnight fasting; they hypothesised different underlying reasons as an effect of satiety hormones (leptin or ghrelin), an improvement of peripheral circadian clock (and therefore an improvement of key metabolic regulators such as cAMP response element-binding protein), and a reduction of oxidative damage together with a higher stress resistance [ 28 ]. These data suggest that 1–2 meals are better than three or more, but how can we integrate these results with previous, older research? Both older studies [ 9 , 10 , 12 , 29 , 30 ] and more recent research [ 31 ] seem to suggest that a higher meal frequency can reduce weight gain risk; however, recent large prospective studies seem to support that frequent snacking increases the risk of weight gain [ 32 , 33 ] and type 2 diabetes [ 16 , 17 ]. Additionally, research investigating acute metabolic responses to differing meal frequencies may support the benefits of a lower meal frequency. Taylor and Garrow evaluated the effects of isocaloric diets consisting of two or six meals per day on energy expenditure measured in a metabolic chamber. The results showed no differences during the day whilst night expenditure was significantly higher with two meals compared with six meals [ 34 ]. On the contrary, other studies demonstrated a significantly higher basal energy expenditure in the morning compared to the evening [ 35 , 36 , 37 ]. However, diurnal differences in the total energy expenditure are not consistently found in all studies [ 38 ]. Other studies suggest that weight gain and its metabolic consequences with a higher meal frequency are due to not only to the higher sugar derived energy intake [ 39 ] and associated metabolic issues, but also to increased food stimuli, hunger and desire to eat [ 40 , 41 ]. Thus, a regular meals pattern has potential positive effects on health outcomes regardless of meal frequency.

Often infrequent meal pattern, i.e. a reduced meal frequency, is associated with an irregular eating approach that could cause weight gain, increase hunger-related hormones, and ultimately lead to a metabolic disturbance that may increase cardiovascular risk [ 42 ]. On the contrary, a lower frequency but with regular timing may decrease weight gain risk [ 28 ].

2.3. Intervention Studies and Reciprocal Influences of Meal Frequency and Macronutrients

In addition to the effects of changing meal frequency per se, it must be considered that these changes could also modify the overall macronutrient intake. This was demonstrated by McGrath and Gibney, who convinced subjects who usually eat six times daily to reduce their frequency while persuading lower frequency eaters (three times daily) to increase their frequency to six times. The increase of meal frequency induced a significant reduction of total and LDL cholesterol but was coupled with a reduction of carbohydrate intake [ 30 ].

The reductions of cholesterol observed by McGrath and Gibney can be considered in light of the current debate about the real relationship between traditional disease markers such as total cholesterol and LDL cholesterol and CHD [ 43 ], as some have challenged the common idea that higher blood levels of cholesterol increase stroke and other cardiovascular events [ 44 ]. It is reasonable to assume that the mechanisms involved in cholesterol reduction may be related to cholesterol synthesis mechanisms. We now know that insulin activates a key enzyme in cholesterol biosynthesis, hydroxymethylglutaryl-CoA (HMGCoA) reductase (the target for statins) [ 45 ]. Even though the discussion about the mechanisms underlying this control (AMP-activated protein kinase, increased rate of transcription, or insulin-induced genes) [ 46 , 47 , 48 ], exceeds the aims of this review, it appears consequential that an increase in blood glucose and, of consequence, of insulin will lead to increased endogenous cholesterol synthesis [ 49 , 50 , 51 ]. It was demonstrated that a higher meal frequency (nibbling) reduced insulin concentrations as compared to three meals daily [ 12 ], likely caused by a reduction in cholesterol synthesis [ 29 ].

Apart from insulin’s action, another effect of a high meal frequency could be the increased cholesterol removal (reverse cholesterol transport) in the postprandial phase after a meal containing fat [ 52 ] and the inhibitory effects of cholesterol and fats on HMGCoA reductase [ 53 ]. We cannot dismiss the effects of macronutrient composition on meal frequency, blood lipids, and insulin effects. An increase in the number of snacks can also increase the amount of dietary protein [ 54 ]. Data suggest that whilst there is no correlation between number of snacks and hunger [ 54 ], or at least not a positive one [ 55 , 56 ], there is a greater fullness-related response with higher protein intake [ 41 , 55 ]. Thus, when discussing meal frequency, it is essential to also consider, from an ecological perspective, that changing meal frequency could also change the percentage of energy from particular macronutrients during the day. Moreover, substituting carbohydrates/sugars with protein in the snacks could change the outcome of low or high-frequency meals studies [ 39 ]. Finally, it is important to underline that meal frequency alone could not explain its effects on health’s outcomes. The contrasting data about meal frequency and health could be explained by the fact that, often, a reduced number of meals reflects an incorrect distribution: skipping breakfast, light lunch, and a high-calorie dinner or a very low number of meals (i.e., 1–2) could lead to poor metabolic control [ 16 ]. Moreover, the meal frequency effects are strictly related to meal timing and macronutrients uptake. At the moment the available data about the effects of nibbling (small, frequent meals) compared to gorging (large, infrequent meals) on isoenergetic conditions [ 57 ] provide conflicting results, probably due to the above-mentioned confounding factors.

3. Meal Timing

3.1. epidemiological data on meal timing: breakfast or not breakfast, this is the question.

When considering meal frequency and timing, which meals are maintained or removed is not a minor issue. Generally speaking, those who consistently eat breakfast have a lower risk of weight gain compared to those who skip breakfast; moreover, those eating their largest meal at lunch or dinner have a greater risk of an increased BMI [ 28 ]. Moreover, Cahill et al. in 2013 discovered an interesting association between coronary heart disease (CHD) risk and frequency of consuming breakfast. Authors reported data coming from 51,529 healthy males (monitored from 1992 up to 2008) and concluded that “ eating breakfast was associated with significantly lower CHD risk ” [ 18 ]. Both dinner and breakfast skipping increased 24-h energy expenditure, concomitant with a longer fasting period, but skipping breakfast may elicit higher postprandial insulin concentrations and increased fat oxidation, suggesting a metabolic inflexibility that may lead to low-grade inflammation status and impaired glucose homeostasis [ 58 ]. In general, available data suggest that if there are health-promoting effects of reducing meal frequency, there may be differential effects of skipping breakfast versus dinner (i.e., evening fasting before an overnight fast vs. an overnight fast followed by continued morning fasting). Moreover, it has been suggested that late eating is related to increased risk of obesity and CHD [ 59 ] and also that a “grazing” eating pattern is related to higher total energy intake and later night-time food consumption [ 60 ]. Finally, there is a consensus about the association between breakfast consumption and CHD. Cahill et al. [ 18 ] published a large prospective study from the Health Professionals Follow-up Study on 26,902 American men aged 45 to 82 years. They found that men who skipped breakfast had a 27% higher risk of CHD compared with men who regularly ate breakfast (RR 1.27; 95% confidence interval CI 1.06–1.53). Additionally, eating late at night led to a 55% higher CHD risk (RR 1.55; 95% CI 1.05–2.29) compared to an earlier dinner.

3.2. Intervention Studies and Meal Timing: Inner Clock Mechanisms

Furthermore, Jakubowicz et al. [ 61 ] demonstrated that an isocaloric diet differing in the distribution of calories during the day (i.e., high calorie in the morning vs. high calorie in the evening) could influence weight loss, serum ghrelin, insulin resistance indices, and subjective appetite feeling in overweight/obese women. The results confirmed the positive effects of consuming more calories earlier in the day, including through breakfast consumption, and the correlation between meal timing and body weight. However, it should be noted that some evidence has failed to support the importance of breakfast consumption for body weight change in free-living adults. Dhurandhar et al. [ 62 ] conducted a randomized controlled trial that assigned 309 overweight and obese adults to either eat breakfast or skip breakfast for 16 weeks. Despite high compliance with the assigned programs, they found that breakfast consumption did not produce weight loss relative to breakfast skipping. On the contrary, regarding cardiovascular health, Uzhova and collaborators found that skipping breakfast was associated with an increased risk of non-coronary and generalized atherosclerosis independent of conventional CVD risk factors in a sample of middle-aged asymptomatic individuals [ 63 ]. Moreover, Betts and colleagues showed that both lean and obese adults expend less energy during the morning when remaining in the fasted state than after consuming breakfast [ 64 , 65 , 66 ].

The opposite (i.e., the negative effects of late dining) is not so conclusive. Even though a recent meta-analysis demonstrated an association between evening energy consumption and higher BMI, they concluded that because of high heterogeneity it is difficult to draw conclusions about the effect of large evening dinner on weight control [ 67 ].

An important consideration related to early versus late feeding is the influence of feeding on the internal circadian clock [ 68 , 69 , 70 , 71 ]. The body circadian timing system is composed by a central clock in the hypothalamic suprachiasmatic nucleus and by different peripheral tissue clocks. The circadian clock system is involved in many metabolic rhythms including glucose and lipids. Whilst central clock dictates food intake, energy expenditure and insulin sensitivity, peripheral/tissues clocks carry out an additional control. For instance, the peripheral clock in the gut regulates glucose absorption and peripheral clocks in the adipose tissue and liver regulate their insulin tissue sensitivity while another peripheral clock in the pancreas regulates insulin secretion. Also lipids biosynthesis and catabolism are regulated in different tissue by a local molecular clock as demonstrated by recent studies on metabolomics and lipidomics.

It is well-known that disruption of central and or peripheral circadian clocks could promote obesity and CHD in many organisms [ 72 ]. Almost all species have developed an internal cellular clock mechanism, sensitive to the light and dark phases of a day, which allows animals to anticipate and to adapt to the changes in the environmental conditions linked to light and darkness. Early research performed in the 1970s identified the suprachiasmatic nucleus (SCN) as the main biological clock. The SCN regulates not only sleep-wake cycles but also many other physiological variables such as body temperature, blood pressure, hormone secretion, and behavioural variables. These circadian rhythms allow the organism to adapt to the environment and to be prepared for the different demands of daily life. For example, the morning increase of cortisol prepares the cardiovascular system for the upcoming day’s activities, and thus disruption of the circadian cortisol rhythm and the consequent cardiovascular impairment could lead to an increased risk of cardiovascular events in the early morning [ 73 ]. Another important marker of the internal clock is melatonin. Melatonin is strongly regulated by light/dark cycle with high levels during the night in all vertebrates. This fundamental rhythmic endocrine signal for darkness in the body is controlled by the master clock in the SCN and mainly by the Period gene (Per1) that has been shown to cycle rhythmically in the pineal gland [ 74 ]. For instance, McHill et al. [ 75 ] found that, on average, obese individuals consumed most of their calories an hour closer to melatonin onset (biological marker of impending sleep onset) compared to lean individuals.

Also different physiological functions exhibit circadian rhythm: for example glucose tolerance changes during the day showing a poorer glycaemic control in the evening and at night in healthy adults. These changes are influenced by diurnal rhythms in β-cell responsiveness, insulin clearance, and peripheral insulin sensitivity, whilst hepatic insulin sensitivity seems to be less important. However, the circadian rhythm and the inner clock mechanism could be affected by different factors such as light exposure, sleep/wake, physical activity, and food intake. Actually, meal timing is one of the main factors that might influence these physiological functions and, therefore, various health outcomes and body weight control [ 76 ]. Meal timing influences either the central master clock (SCN) or peripheral cellular clocks, including Bmal1, Clock, Per1/2, Cry1/2, Rev-erbα/β, Rorα/β, Dbp, Dec1/2, CK1ε/δ, and NPAS2 [ 74 , 77 ].

It is important to underline that peripheral tissues show proper circadian rhythms and cellular clocks. Central and peripheral clocks work together and they are also influenced by food availability. Indeed, regular feeding patterns may synchronize human peripheral clocks and delayed meals could instead influence plasma glucose rhythm but not insulin rhythm [ 78 ].

Many genes whose expression is not cyclic may start to follow a circadian rhythm under the pressure of nutritional challenge that modulates PPARs (besides their circadian rhythm) activating many genes by cyclic chromatin recruiting.

Even though the mechanisms underlying the effects of meal timing on health outcomes remain obscure, some hypotheses ( Figure 1 ) can tentatively be presented:

• (1) Food timing that is out of sync with light/dark cues could induce higher caloric intake due to impaired satiety mechanisms through leptin and ghrelin [ 79 ]. Even other hormones involved in metabolism control are affected by circadian misalignment as thyroid hormones [ 80 ].
• (2) Alteration of gene expression in genes that are associated with evening eating preference and weight loss resistance e.g.,SIRT1, CLOCK 3111T/C, and Perilipin1 [ 81 , 82 ]
• (3) Modification of resting energy expenditure: feeding time may affect energy expenditure/basal thermogenesis as core body temperature is controlled by circadian clocks. For example, Rev-erbα is a cellular circadian clock that controls the rhythmic expression of uncoupling protein 1 (UcP1), a fundamental factor for brown adipose tissue thermogenesis [ 83 ].
• (4) Differences throughout the day in diet-induced thermogenesis (DIT): DIT decreases from morning to night [ 35 , 36 , 84 ], and some have suggested that “Such circadian thermogenesis could reasonably explain increases in the body mass of persons who skip breakfast” [ 85 ].
• (5) Circadian clocks influence also insulin resistance through glucose absorption, muscle, fat tissue, and liver insulin sensitivity [ 86 ] and food intake or nutritional challenge influence, in turn, circadian clock. Indeed shift workers, transcontinental travelers and people with irregular work schedules often show gastrointestinal symptoms as alterations in bowel habits, constipation, and diarrhoea. These examples indicate that some intestinal functions are rhythmically regulated and that their disruptions lead to health disorders. It was demonstrated that Clock (a peripheral cellular clock) regulates nutrient absorption through the expression of many nutrient transport proteins in the intestine e.g., GLUT2, GLUT5, and Pept1 (a major protein involved in the transport of small peptides from the intestinal lumen to intestinal epithelial cells). However, other external factors could influence the internal clock. For example, NAD + (nicotinamide adenine dinucleotide) levels are influenced by nutritional status and/or physical activity. NAD + influences the SIRT1-dependent deacetylase that activates, through deacetylation, the clock genes BMAL1 (brain-muscle-arnt-Like-protein 1) and PER2 (Period gene 2). Nicotinamide phosphoribosyltransferase (NAMPT) a downstream of BMAL1, has an oscillatory behaviour, therefore modulating the intracellular concentration of NAD+. Thus, in a feedback loop, NAD + concentration regulates SIRT1 that modulates nuclear factors such as PPARγ (peroxisome proliferator-activated receptor gamma) and cofactors as PGC-1α (peroxisome proliferator-activated receptor gamma coactivator 1-alpha) with many effects on different tissues e.g., on hepatic glucose homeostasis (PGC-1α) or adipose tissue lipid mobilization (PPARγ). In general, a regular availability of food (regular meal timing) influences the release, from the gut, of different signals. It has been suggested that signals coming from intestine inform the dorsomedial hypothalamus (DMH) about food availability. Thus, DMH might influence other tissue and regulate food anticipation, digestion, and absorption. Thus, even though circadian genes expressed by gut play an important role, there is some evidence that food, per se, is an important regulator of food entrainment through Clock activity.

Effects of external factors on the inner central clock that influence different downstream mechanisms and peripheral clocks (CNS: central nervous system).

4. Reducing Meal Frequency: The Case for Time Restricted Feeding

The importance of fasting: what’s new.

If the potential health-promoting effects of less frequent eating are considered sufficient for implementation of this dietary strategy, is consuming one daily meal equivalent to the consumption of two daily meals? In this case, the answer is not merely “less is better”: reducing food intake to only one meal per day may worsen the positive effect of lower meal frequency [ 87 , 88 ]. Therefore, the intake of two (or three) meals per day is perhaps the best option, and the difference between two or three could depend on the length of the daily fasting period they produce.

Much research in recent years suggests a positive health effect of a wide temporal fasting window during the day, i.e., limiting daily food intake to a ~6–8 h time window seems to induce, in humans, many health benefits compared to the normal daily meal distribution (i.e.,three to five meals, spread from breakfast to late dinner), even in isocaloric conditions [ 89 ]. It is clear that fasting, in general, exerts many positive effects on health [ 90 ], with some features in common with the caloric restriction (CR) approach (protects against diabetes, cancers, heart disease, and neurodegeneration; reduces obesity, hypertension, asthma, and rheumatoid arthritis).

During a typical CR protocol, the daily energy intake is chronically reduced by 20–40%, but meal frequency is maintained. It is well known that CR is a viable tool for health improvement: both animal studies [ 91 ] and human research [ 92 , 93 ] showed that this approach could improve many health-related variables.

However, we have to consider the experimental setting of the ab libitum diet and the CR condition to which it is compared in animal experiments. Often, in animal models, the CR condition influences fasting duration. In these experiments, animals in the ad lib diet have unrestricted access to food, not only in quantity but also in frequency, whilst the CR group can only eat within a specific window, usually determined by the researcher’s schedule. In these settings, meals are often spaced out, creating prolonged fasting windows that could influence the outcomes [ 94 ]. This is an important issue because fasting is a different approach than traditional CR. We consider fasting as an abstention from food and caloric beverages for a specific interval of time, usually longer than the normal 8 h of sleep. Alternatively, starvation refers to extreme forms of fasting, which result in nutrient deficiencies and other chronic health problems related to the absence of appropriate nutrient intake. Starvation is, actually, a dysregulated condition that leads to a pathological loss of homeostasis related to the reduction in fundamental organ and tissues performance [ 95 ]. When considering the different types of fasting programs, we can divide them into two main categories: long-term fasting (LTF) that induces ketosis, and short-term fasting (STF) that does not lead to ketosis. LTF, i.e., fasting with accompanying ketosis, is performed for approximately three days or more. After this period, glucose reserves become depleted and glycogen stores are no longer sufficient to either aid in normal fat oxidation (via oxaloacetate in the Krebs cycle) or to supply energy to the brain and central nervous system (CNS) [ 96 ]. Thus, an alternative energy source is needed to maintain the metabolism of the brain. This energy is supplied by the ketone bodies (KBs) acetoacetate (AcAc), 3-hydroxybutyrate (3HB), and acetone, which are generated from acetyl-CoA via a process called ketogenesis, which occurs mainly in the mitochondrial matrix of hepatocytes [ 96 , 97 ]. Ketosis exerts many positive effects on metabolism and numerous cellular pathways, such as increasing stress resistance, lipolysis, mitochondria efficiency, and autophagy (e.g., one of the ketone bodies, b-hydroxybutyrate (D-bHB), is a natural inhibitor of class I and IIa histone deacetylases that repress transcription of the FOXO3a -forkhead box O3 - gene). Moreover, ketone body metabolism reduces the ROS (reactive oxygen species) toxicity through the NADPH system [ 98 ]. However, in the context of meal timing and frequency, we want to emphasize the role of STF, which utilizes fasts of insufficient duration to induce ketosis unless used in conjunction with a ketogenic diet. There are several types of STF programs [ 99 ]: intermittent fasting (IF) performed as alternate day fasting (ADF) or whole-day fasting for 1–2 days per week, periodic fasting (PF) lasting three or more days every 2–3 weeks, and TRF (Time restricted feeding) whichallows subjects to consume ad libitum energy intake within a defined window of time (from 3–4 h to 10–12 h) [ 100 , 101 ], resulting in a fasting window of 12–21 h per day. For our purposes, we will discuss the TRF because if the number of meals is reduced to two (i.e., breakfast and lunch), and the last meal is consumed between 14:00 h and 16:00 h, this leads to a 12 to 16 h of fasting per 24-h period. It is also worth noting that a substantial amount of research has been conducted during the month-long period of Ramadan fasting observed by practicing Muslims [ 102 ]. Ramadan fasting can be considered a form of TRF since food intake is disallowed when it is light outside. However, some notable factors make it difficult to appropriately compare Ramadan fasting to other forms of TRF: the light/dark cycle of eating and fasting is reversed as compared to natural circadian rhythms, the length of fasting window varies based on geographical location and year (Ramadan is set according to the lunar calendar), and different implementations of Ramadan fasting exist (i.e., some eat before the sun rises and after the sun sets, while others only eat after the sun sets). Finally, nearly all studies are observational and last only 4 weeks since this is the duration of Ramadan fasting.

Despite the fact that the duration of fasting during Ramadan (about 16 h) would not typically result in ketosis, it is sufficient to stimulate many of the pathway linked to long term fasting approach, e.g., autophagy [ 103 ]. Autophagy, an intracellular process that mediates protein degradation, organelle turnover, and recycling of cytoplasmic components, is a fundamental process to combat cellular stress and preserve normal cell function. In heart and blood vessels, specifically, autophagy plays a fundamental role not only during cardiac embryonic development but also for a normal cardiovascular function. It has been suggested that many of peptides and hormones involved in cardiovascular system physiology are also regulated by autophagy, thus “it is possible to speculate that dysregulation of autophagy could be associated with hypertension, obesity, diabetes mellitus, and end organ damage” [ 104 ]. As fasting stimulates autophagy, it is likely that these two factors are both related to the demonstrated cardioprotective effect. Indeed, Godar et al. 2015 [ 105 ] demonstrated that ADF protects mice from in-vivo ischemia-reperfusion injury, but only in wild-type animals. In mice with impaired autophagy (heterozygous null for Lamp2 coding for lysosomal-associated membrane protein 2), there was not a protective effect, but rather a worsening effect. Another study performed on rats showed that ADF has a cardioprotective effect reducing cerebral infarct size and infarct expansion in a rat model of myocardial infarction (MI) [ 106 ].

Fasting affects substrate metabolism, the cardiovascular system and inflammation, as well as exerting potentially powerful effects on circadian rhythms.Increasing the fasting window stimulates fat metabolism and the metabolic switch between glucose oxidation and fat oxidation. Indeed, at between 12 and 36 h of fasting there is an increase of TG (triacylglycerol) lipolysis highlighted by the increase of plasma FFA and glycerol [ 107 ]. The metabolic switch typically occurs in the third phase of fasting when glycogen stores in hepatocytes are depleted and the accelerated adipose tissue lipolysis produces an increase in plasma fatty acids and glycerol (21). The fasting period associated with IF and TRF seems to have various positive effects on the cardiovascular system as well: they enhance parasympathetic activity (mediated by the neurotransmitter acetylcholine) in the autonomic neurons that innervate the heart and arteries, resulting in a reduced heart rate and blood pressure [ 90 , 108 ]. Furthermore, TRF could also act on inflammation levels. It is well known that inflammation is related to CHD and atherosclerosis. We demonstrated in humans [ 100 ] that an isocaloric TRF approach may reduce many markers of inflammation such as tumour necrosis factor a, interleukin 6, and interleukin 1b, and, at the same time, may increase adiponectin (an anti-inflammatory cytokine). As demonstrated for late eating, fasting also seems to be involved in circadian rhythm regulation or dysregulation. It has been demonstrated that TRF could protect mice against obesity, hyperinsulinemia, hepatic steatosis, and inflammation when fed with a high-fat diet (HFD). The ad libitum HFD rodents also showed altered circadian rhythmicity compared to the TRF rodents. Moreover, TRF improved CREB (cAMP response element-binding protein), mTOR (mechanistic target of rapamycin), and AMP-activated protein kinase (AMPK) pathway function and oscillations of the circadian clock, as well as improving motor coordination [ 109 ]. These results could be explained through the considerable crosstalk and the tight interaction between the cellular clocks and the signalling induced by fed/fasted state. For example, we know that fasting, similar to a ketogenic diet, induces the phosphorylation of AMPK, a fundamental actor in mitochondrial biogenesis and function. On the other hand, the fed state stimulates the mechanistic target of rapamycin pathway (mTOR), which promotes anabolic processes during increased energy availability, which could interfere with AMPK pathway. This connection supports the tight relationship between fed/fast state and molecular pathways.

Finally, we have to underline that ketogenic diet, caloric restriction and fasting have many pathways and targets in common as shown in Figure 2 .

Mechanisms involved in health effects of the ketogenic diet (KD), caloric restriction (CR), and fasting. The size of the arrows is related to the relative effect of KD (orange), CR (blue), and fasting (green) on the different pathways involved (IGF-1: insulin-like growth factor-1; Murf2: Muscle-specific RING finger-2; Nf-kB:nuclear factor kappa-light-chain-enhancer of activated B cells).

5. Meal Frequency and Timing: The Microbiota Connection

We cannot conclude this exploratory review without discussing the role of meal frequency on microbiota. In recent years, this field of research has experienced rapid growth. The collective microbiomal organ provides many fundamental functions such as metabolic, immunological, and infection control. In the last years the gut microbiota has been recognized as an important factor for host general health, immunity, and also energy homeostasis. Changes in microbiota population might cause the development of many metabolic diseases attributable to the modification of the relationship between the bacteria and the host. Up to 100 trillion bacteria constitute the human gut microbiota with 150 times more genes (the microbiome) than the human genome. Abnormalities in gut microbiota composition might have many effects on metabolism in adipose tissue, muscle and liver. Moreover, the gut microbiota has been associated to many metabolic diseases such as obesity, diabetes, chronic low-grade inflammation and, last but not least, cardiovascular disease [ 110 ].

Indeed, it has been demonstrated that the composition of microbiota might be a risk factor for CVD. Mice studies have demonstrated the link between gut microbiota dysbiosis and the development of hypertension and vascular dysfunction [ 111 ] while in human demonstrated a relationship between negative changes in gut microbiota and primary hypertension has been demonstrated [ 112 ]. Gut microbiota converts choline (derived from dietary phosphatidylcholine) to trimethylamine (TMA). In the liver, TMA is converted in trimethylamineN-oxide (TMAO) that promotes atherosclerosis and increases thrombosis risk through the agonist-induced platelet activation [ 113 , 114 ]. In conclusion, available data strongly support the critical role of gut microbiota as a regulatory element in many CVD risk factors.

The microbiota exerts also many actions on the central nervous system, so many that it has been coined the “gut–brain axis.” Diet composition (e.g., fat and fibre content) influences gut microbiota. No data are available in humans concerning meal frequency, whilst some preliminary information about food timing and microbiota are available. Changes in gut microbiota may be stimulated by changes of diurnal feeding and sating rhythms, and it is known that a desynchronization of the suprachiasmatic nucleus, the master clock of the brain, together with a parallel desynchronization of the tissue circadian clocks in skeletal muscle, fat and liver may influence the risk of chronic and metabolic diseases [ 115 ]. There is a multifaceted relationship between microbiota and food timing: first, intestinal epithelia cells’ internal circadian clock influences daily glucocorticoid production under the control of the pituitary-adrenal axis, and this rhythm is influenced by microbiota status; second, an alteration of microbiota could lead to a disrupted corticosteroid circadian rhythm influencing food uptake. Moreover, microbiota composition has its variability during the day that could be disrupted by a variety of conditions, for example, jet-lag [ 116 ] or high-fat diets. Not only can diet composition exert negative effects on microbiota, but meal timing can also: consuming food outside the normal feeding phase (eating during light time for rodents and during late night in humans) may disturb normal peripheral and central clocks [ 115 ]. This desynchronization of internal clocks, and thus the modification of microbiota, is associated with increased risks of metabolic and cardiovascular diseases. Recently it has been demonstrated that a chronic circadian misalignment in mice and a time shift jet–lag in humans induces a dysbiosis; this dysbiosis has been demonstrated to be able to promotes glucose intolerance and obesity in a germ-free mice throughfaecal transplantation [ 117 ]. On the other hand, maintaining a correct eating phase (diurnal for humans) and increasing the fasting period (i.e., reducing meal frequency) could positively affect the gut microbiome, reducing gut permeability and improving systemic inflammation. Finally, further studies are needed to explore properly the connection between microbiota and meal frequency and timing.

6.Concluding Thoughts

In order to gain a comprehensive picture of the physiological and health effects of meal timing and frequency, multiple lines of research must be integrated and an exploratory review seems to be, in our opinion, the appropriate approach in order to understand, at a glance, the influence of fasting, meal frequency, and timing on cardiovascular diseases. In addition to considering existing evidence of meal frequency and timing per se, research on breakfast consumption, night-time eating, caloric restriction and intermittent fasting can help provide much more awareness about the effects meals manipulation on health outcomes. While a recent meta-analysis reported that high versus low meal frequencies result in negligible differences in body weight and composition changes [ 117 ], many of the experimental trials of meal frequency have not adequately considered some of the determinants highlighted in this article which could influence these outcomes (i.e., duration of daily fasting periods and different spacing of meals within the same meal frequency, influence of eating styles on food choices and macronutrient intake, etc.). Additionally, beyond body weight and composition, it is likely that different eating patterns may exert some degree of differential effects on physiological processes, even in isocaloric conditions ( Figure 2 ). Furthermore, the existence of different chronotypes should be taken into account: being larks and owls [ 118 , 119 ], or morning types (M-types) and evening types (E-types), might probably influence also eating behaviour and food metabolism. Even though this classification is not new [ 120 ] only in recent years has the association between chronotypes and eating behaviour been investigated. A recent paper by Maukonen and colleagues analysed the associations between chronotype and intakes of energy and macronutrients in the morning and the evening in 1854 participants from the National FINRISK 2007 and FINDIET 2007 studies. They found that, in the morning, E-types showed lower total energy and lower macronutrient intakes except for sucrose (increased intake) compared to M-types.In the evening, E-types had higher intakes of energy, fats, and sucrose than M-types. These data suggest that even chronotype might influence meal patterns [ 121 ]

Based on the evidence presented in this review, several interesting health-promoting recommendations can be shared with the audience. There may be physiological benefits to consuming a greater proportion of calories earlier in the day, which often involves breakfast consumption, as compared to consuming a large number of calories later at night. There may also be benefits to extending the daily fasting period beyond a standard overnight fast or implementing occasional fasting periods. In order to reconcile these two strategies, an individual could eat from breakfast until mid- to late-afternoon each day ( Figure 3 ). However, it should be considered that this style of eating may not be desirable or feasible for many individuals, as it represents a paradigm shift from traditional eating patterns in many parts of the world.

Effects of different meals timing and frequency on different variables. At the centre of the picture the reciprocal influences of brain, heart and gut was showed. AMPK: AMP-activated protein kinase.

Additionally, due to the increased access to food associated with evening leisure time, compliance with this recommendation may not be realistic for some. In those cases, it may be beneficial to implement one of the health-promoting strategies (i.e., shift the consumption of most calories earlier in the day or implement a fasting window longer than an overnight fast). The lifestyle approach should include physical activity. Unfortunately, whilst there are few papers on physical exercise and internal clock [ 122 ], no data are available about the reciprocal influence of meal time and frequency and physical exercise in humans. This topic is worthy of further investigation.

While a complete picture of the impact of meal timing and frequency in various populations remains to be elucidated, it is likely that manipulation of these variables may be useful in improving health in the human population ( Figure 4 and Figure 5 ). The scientific literature provides sufficient data to suggest that there is a substantial influence of fasting, meal frequency, and timing on health outcomes. These findings underline that not only the food quality but also frequency and timing are crucial for optimal health.

Effects (green: positive; red: negative; blue: neutral) of meal timing on different CVD risks factors and diseases. CHD: coronary heart disease; CVD: cardiovascular disease; TRF: time restricted feeding.

Effects (green: positive; red: negative; blue: neutral) of meal frequency on different CVD risks factors and diseases.CHD: coronary heart disease; CVD: cardiovascular disease.

Author Contributions

All authors contributed equally to the review.

This research received no external funding.

Conflicts of Interest

The authors declare no conflict of interest.

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to  upgrade your browser .

•  We're Hiring!
•  Help Center

Mid Day Meal Scheme

• Most Cited Papers
• Most Downloaded Papers
• Newest Papers
• Injection Molding Follow Following
• Plea Bargaining Follow Following
• Cultures of Punishment Follow Following
• Policy Analysis and Management Follow Following
• Nuclear Power Plants Safety Follow Following
• Advanced Nuclear Energy Systems Follow Following
• Networks and Public Governance Follow Following
• Philosophy of Ecology Follow Following
• Tempo Follow Following
• Medical Marijuana Follow Following

Enter the email address you signed up with and we'll email you a reset link.

• Academia.edu Journals
•   We're Hiring!
•   Help Center
• Find new research papers in:
• Health Sciences
• Earth Sciences
• Cognitive Science
• Mathematics
• Computer Science
• Academia ©2024