Household cooking and eating practices across food system typologies in 135 countries from 2018 to 2022

Abstract

Cooking meals at home is often recommended as a strategy to consume a healthy diet and avoid ultra-processed foods and unhealthy dietary patterns. However, cooking has traditionally been a highly gendered activity, and as food environments, economies, and social norms evolve globally, understanding of cooking behaviors in different contexts is needed. We analyzed repeated cross-sectional data from 135 countries from the 2018–22 Gallup World Poll (n = 638 192). We conducted unadjusted and adjusted negative binomial regression models, overall and stratified by gender, to estimate differences over time and by five food system typologies in the mean number of days cooking and eating lunch and dinner at home per week. Across most food system typologies, mean number of days cooking but not eating meals at home gradually increased between 2018 and 2019, peaking in 2020 (among females) and 2021 (among males) and then declined in 2022, trending towards prepandemic levels. Adjusted for covariates, participants from emerging and diversifying food systems cooked meals at home 13% (95% CI: 1.10, 1.16, P < .001) more compared to participants in rural and traditional food systems. Compared to males, females cooked meals at home 214% more frequently (P < .001). Across the globe, COVID-19 shifted cooking behaviors and exacerbated preexisting gender disparities in frequency of cooking meals at home. Future research is needed to better understand the role of cooking behaviors in the nutrition transition at global, national, and local levels.

Contribution to Health Promotion.

• Cooking and eating meals at home are an important area of focus for health promotion, as an opportunity to improve the health of families and children globally

• Food systems have dynamically transitioned over the past half-century, shifting food access, preparation, and consumption practices, often at the expense of nutrition

• A variety of evidence-based policies and interventions are needed at national, regional, and local levels to create enabling and equitable environments that encourage practices that support healthy diets across all food systems

INTRODUCTION

A growing body of research suggests a potential protective effect of meals prepared and consumed at home (Mills et al. 2017a, Wolfson et al. 2020, Tucker et al. 2024). Cooking frequently at home has been shown to be associated with greater dietary diversity (Komatsu et al. 2018) and in some settings, lower risk of developing diet-related health outcomes (Zong et al. 2016, Méjean et al. 2018, Zhang et al. 2019). However, across the globe, food prepared and eaten away from home represents an increasing share of people’s food consumption (Godbharle et al. 2022, Landais et al. 2022), and the availability and quality of foods purchased or consumed outside of the home varies significantly from food consumed in the home (Lee et al. 2016, Matsumoto et al. 2021). While these trends are rapidly evolving, the strength of associations between cooking meals at home or consuming food away from home and dietary quality varies around the globe depending on food system factors that intersect to shape individuals’ access to and choices about food, such as the availability, accessibility, and promotion of foods within the food environment, as well as household and individual-level factors such as income and purchasing power, intrahousehold dynamics, cultural and religious beliefs, as well as consumer knowledge and preferences (High Level Panel of Experts on Food Security and Nutrition 2017).

The dynamic transition of food systems across rural and urban settings and country economic stratifications has resulted in changes to populations’ physical, economic, and social access to food (Ambikapathi et al. 2022, Downs et al. 2022). Categorizing food systems, termed food systems typologies, can provide insight into how food systems are transitioning from rural to hyper-industrialized and through which macro drivers such as globalization and modernization (Marshall et al. 2021). The utility of typologies becomes critically important in the context of the nutrition transition—the well-defined phenomena of population shifts whereby as countries transition from traditional to industrialized food systems, driven by economic development, the growing female workforce, urbanization, and demographic changes, there are significant changes in dietary patterns, physical activity, and disease epidemiology (Popkin 1993). The changes in dietary patterns are a product of what food is available, how it is prepared and by whom, and where it is consumed.

Throughout history, women have predominantly been responsible for domestic work, including the task of food preparation, due to gendered norms regarding employment outside of the home, lack of access to education, agency, and decision-making power about the allocation of household resources (Bianchi et al. 2000, Ilahi 2000). Women often serve as the custodians of food rituals and traditional knowledge in households and communities. However, in recent decades, the increasing participation of women in the labor force has resulted in a shift of responsibilities, with less time for women to spend on unpaid household work such as cooking (Ortiz-Ospina et al. 2024). Yet, large gender disparities in household cooking frequency still exist, with women cooking far more frequently than men (Wolfson et al. 2021). This is particularly importantly as the COVID-19 pandemic exacerbated gender disparities in how men and women experienced the combined health and economic shocks of the pandemic, with considerable variation across countries (Kalbarczyk et al. 2022).

Global trends in cooking and eating meals at home have been explored by the Gallup World Poll (GWP)—an annual global survey of multiple topics which for the past five years has measured frequency of cooking and eating meals at home in over 160 countries (Gallup World Poll 2021). Data from the GWP show that frequency of cooking meals at home varies considerably around the globe and that there are consistent trends in gender disparities in home cooking frequency, with women cooking much more frequently than men; though how these trends differ across diverse food systems is unknown (Gallup World Poll 2024). Additionally, the COVID-19 pandemic disrupted food and economic systems in ways that may have increased frequency of consuming meals prepared at home among all individuals, though this trend may have differed in different regions and food system typologies (Carducci et al. 2021). Given the GWP’s representativeness, the objective of this study is to determine whether cooking and eating meals at home between 2018 to 2022 is associated with food system transitions (as measured by five distinct typologies) and whether these associations differ by gender in 135 countries.

We hypothesized that prior to the COVID-19 pandemic countries with more industrialized food systems would cook and consume lunch and dinner at home less frequently as compared to countries with more rural and traditional food system patterns. We also hypothesized that cooking and eating at home would increase during 2020 and 2021 in high-income countries, while low- and middle-income countries would remain relatively unchanged in their behaviors due to differences in COVID-19 mitigation. Additionally, consistent with our prior work (Wolfson et al. 2021), we hypothesized that women within all food typologies would report a higher frequency of cooking and eating at home as compared to men.

MATERIALS AND METHODS

Data source

The GWP is a repeated cross-sectional survey that includes noninstitutionalized individuals, aged 15 years and older, in over 160 countries worldwide (Gallup World Poll 2021). In each country, nationally representative samples of ~1000 individuals are recruited using a probability-based, multicluster sampling strategy (in large countries such as China, India, and Russia samples include ~2000 individuals) (Gallup World Poll 2021). The GWP is administered via face-to-face or telephone interviews (Gallup World Poll 2021). In countries in which face-to-face interviewing was used, a 3-stage sampling procedure was employed (Gallup World Poll 2021). In the first stage, clusters of households are stratified by population size and geography and selected proportionally when possible. In the second stage, sample households are determined by a random route procedure. In the third stage, respondents are randomly selected by a computer-assisted personal interviewing system or Kish grid method (Gallup World Poll 2021). The selection of clusters with appropriate sampling techniques depends on the detail of population information from each country. In countries in which telephone interviews are conducted, random digit dialing or a nationally representative list of phone numbers was used. In 2020 and 2021, due to the COVID-19 pandemic, the GWP prepared a contingency methodology based entirely on telephone interviewing, with the goal of protecting the safety of interviewers and respondents, retaining high levels of representativity, and ensuring high quality of data collection. In 2022, face-to-face interviewing was resumed. The GWP survey includes more than 100 global questions as well as region-specific items. The questionnaire is translated into all major languages and individuals are asked the same core questions in their preferred language. The GWP methodology and codebook are available for more details (Gallup World Poll 2021).

Measures

Outcomes

Frequency of cooking at home was based on two questions from the GWP asking an individual’s (rather than household) frequency of cooking (i) lunch and (ii) dinner. The question was phrased as, “Thinking about the past 7 days, on how many days did YOU, personally, COOK [lunch or dinner] at your home?” The introduction to the questions about home cooking read “By ‘cooked at home’ we mean a meal prepared at home mostly from ingredients, such as vegetables, meats, and grains.” Frequency of eating at home was based on two separate questions, asking an individual’s (rather than household) frequency of eating (i) lunch and (ii) dinner that was cooked at home. The question was phrased as, “Thinking about the past 7 days, on how many days did YOU, personally, EAT [lunch or dinner] that was cooked AT YOUR HOME, either by you or someone else?” For this study, we created two outcomes by calculating the mean number of days per week of (i) cooking or (ii) eating meals at home (i.e. lunch and dinner combined) for each individual.

Individual-level covariates

We used self-reported sociodemographic indicators from GWP. This included respondents’ age in years, gender (man or woman), number of children <15 years old in the household, educational levels, residency, marital status, employment status, affordability of food, and wealth quintile. Age was categorized as 15–29 years, 30–49 years, 50–69 years, 70–89 years, and 90–100 years. Educational levels were categorized as completion of primary school, some secondary school or completion of secondary school. Residency was categorized as living in a rural area or on a farm, a small town or village, a large city, and a suburb of a large city. Marital status was categorized as single or never married, married or having domestic partner, and separated, divorced, or widowed. Employment status was classified based on full-time, part-time, out of the workforce or unemployed. Affordability of food was based on one question, “Have there been times in the past 12 months when you did not have enough money to buy food that you or your family needed?”. Wealth quintiles (poorest 20%, second 20%, middle 20%, fourth 20%, richest 20%) are calculated based on self-reported household monthly income in local currency. Local income is converted to International Dollars using the World Bank’s individual consumption purchasing power parity conversion factor, making income estimates comparable across all countries.

Country-level measures

We used country food system typology classifications from the 2020 Food Systems Dashboard (Marshall et al. 2021, Food Systems Dashboard 2024). The exhaustive and mutually exclusive five typologies are: (i) rural and traditional; (ii) informal and expanding; (iii) emerging and diversifying; (iv) modernizing and formalizing; and (v) industrial and consolidated. Though static in nature, these typologies elicit the notion of the nutrition transition. The typologies are created based on the following indicators: agriculture value added per worker (in constant 2010 USD), number of supermarkets per 100 000 inhabitants, urban population as a percent of total population, and percent of dietary energy from cereals, roots, and tubers. To construct the typology, countries were first ranked from highest to lowest on each indicator, under the hypothesis that higher values were associated with more industrialized food systems, and lower values more traditional food systems. The ranking was inverted in the case of the share of dietary energy from cereals, roots, and tubers, which is theorized to be lower in more industrialized food systems. A score for each country was assigned based on the sum of its ranks on each of the four indicators. Using the distribution of scores, five quintiles were formed, with the lowest quintile representing the most industrialized food system type and the highest quintile representing the most traditional food system type. The 155 countries included in the 2020 Food Systems Dashboard typologies represent 97% of the global population, 93% of global land area, and 97% of global gross domestic product (Food Systems Dashboard 2024). See Table 1 for key characteristics of food system typologies.

Table 1.

Food system characteristics by typology

	Rural and traditional	Informal and expanding	Emerging and diversifying	Modernizing and formalizing	Industrial and consolidated
Production	Farming is mainly done by smallholders, and agricultural yields are typically low. Among most farmers, production is typically focused on staple crops, some of which they keep to eat, and a limited number of cash crops.	Agricultural productivity is higher on average than in rural and traditional food systems. Medium- and some large-scale farms are beginning to emerge.	Increased number of medium- and large-scale commercial farms coexist alongside large numbers of small-scale farms. These small-scale farms are more linked to markets than in more traditional food system types.	Agricultural productivity is generally higher than in emerging, informal, and traditional systems. Larger farms rely more on mechanization and input-intensive practices.	Farming is a small proportion of the economy. There are a small number of large-scale, input-intensive farms that serve specialized domestic and international markets (e.g. horticulture, animal feed, processed food ingredients, biofuels).
Supply chains	Supply chains are short due to smaller urban populations, resulting in many local, fragmented markets. The lack of refrigeration and storage facilities results in large food losses for some crops, which may make producers less likely to diversify into perishable foods. The quantity and diversity of foods available varies seasonally, often with a pronounced lean season.	Modern food supply chains are common for grains and other dry foods. While there are still small-scale processors, many large-scale processors and centralized distribution centers have also emerged. Modern chains are also emerging for fresh foods, though traditional supply chains continue to dominate for these foods due to weak cold chains and inadequate market infrastructure. Processed and packaged foods are available in both urban and rural areas. Food processing may include a combination of locally sourced and imported ingredients.	Modern supply chains for fresh foods, including fruits, vegetables, and animal-source foods, are developing more rapidly. Urban areas source both dry and fresh foods through longer supply chains and rely on food imports more than traditional and informal food systems. Processed and packaged foods are more available in rural areas.	Food supply chain infrastructure is more developed, which results in fewer food losses on the farm and beyond the farm gate. Food and beverage manufacturing represent a smaller percentage of overall manufacturing because countries in this type have more manufacturing in nonfood sectors. Dietary energy is derived from diverse food sources and better national distribution chains enhance the role of food imports in enabling more year-round availability of diverse foods.	Market consolidation is common—large-sized food retailers procure directly from processors and urban wholesalers procure directly from farmers, which reduces the number of intermediaries along the supply chain. Supply chains are long, with national and international sourcing of nearly all types of foods.
Retail and markets	Food is mainly sold in informal market outlets, including independently owned small shops, street vendors, and central/district markets. Supermarkets are rare outside of capital cities.	Demand for convenience foods increases as the formal labor force grows and includes more women. Urbanization and income growth also play a role. Supermarkets and fast food are rapidly expanding and, compared to rural and traditional food systems, are more accessible. However, most people continue to obtain most of their food from informal market outlets, especially for fruits, vegetables, and animal source foods.	Supermarkets are common even in smaller cities and towns, and their market share is growing rapidly. Processed foods, including ultra-processed foods, are common in urban areas and also found in many rural areas. Most fresh food continues to be acquired through informal markets, but the share of supermarkets is rising and significant.	Multiple supermarket chains exist within cities and larger-sized towns, but their growth is slower than in late transitional systems. These supermarkets and other modern retail outlets hold a large share of processed and dry goods sales, have captured a larger market share of fresh foods, and low-income consumers are much more likely to shop in them.	Supermarket density is high in urban and metropolitan areas. In general, only small towns lack a supermarket and most medium-sized towns have multiple outlets. The formal food sector has captured nearly all of the food eaten domestically, including fresh foods. There is growth in luxury food retail, as well as “fast-casual” restaurants, which market higher-quality fast food.

Source: Food Systems Dashboard (Marshall et al. 2021, Food Systems Dashboard 2024).

In addition to food system typologies, we further classified each country by 2022 World Bank country income level (low-income, lower-middle income, upper-middle income, and high-income) (World Bank 2022), and GWP region classification (Europe, Former Soviet Union, Asia, Americas, Middle East and North Africa and Sub-Saharan Africa).

Study sample

For the present analysis, we triangulated countries that had data from five waves of the GWP surveys covering years 2018–22 and had a food system typology and World Bank country income classification, providing an analytic sample of 135 countries and 638 192 participants (see Supplementary Table 1).

Statistical analysis

We conducted descriptive analyses for trends over time in individual- and country-level factors associated with cooking and eating meals at home, overall and stratified by country typology. To examine differences in each outcome (mean frequency of cooking meals at home and eating meals cooked at home) across the study period and by food system typology we used unadjusted negative binomial regressions with an interaction between year and typology. Then we used postestimation margins commands to report predicted mean frequency of cooking meals at home and eating home-cooked meals in each year overall and by typology. To understand the relationship between the two outcomes variables (frequency of cooking meals at home and frequency of eating home-cooked meals) and individual- and country-level covariates, we performed negative binomial regressions adjusted for the covariates described above overall and stratified by gender. Finally, we used postestimation predicted margins commands after adjusted models and conducted pairwise comparisons to evaluate the difference in predicted means between each year and each typology, for each outcome, overall and stratified by gender. Additionally, in supplemental analyses we repeated the above analyses separately for frequency of cooking and eating lunch, and frequency of cooking and eating dinner. For all analyses, we utilized GWP provided separate survey weights, depending on if data was collected face-to-face or by telephone, to correct for unequal selection probability, nonresponse (gender, age, and, where reliable data are available, education or socioeconomic status), and double coverage of landline and cellphone users when using both cellphone and landline frames. GWP also weights its final samples to match the national demographics of each selected country (Gallup World Poll 2021). All analyses were conducted in 2023 using Stata/SE, 18.0 (StataCorp 2023).

RESULTS

Study sample

Table 2 presents the characteristics of the analytic sample overall and by food system typology. Of the 638 192 participants, 51% were female, 56% were married, 51% had completed some secondary education, 41% worked full-time, 56% lived in a rural area or small village, and 53% had children in the household under 15 years. Thirty-seven percent of participants did not have enough money to buy food for themselves or their families in the past 12 months, with the greatest proportion living in rural and traditional food systems (61%). At the country level, 90% of respondents were from middle- or high-income countries, with geographical spread across World Health Organization regions, including 20% from Sub-Saharan Africa and 19% from Asia. The number of participants by country in each survey wave is provided in Supplementary Table 1.

Table 2.

Participant characteristics by food system typology, 2018–22 (N = 638 192)

	Overall		Rural and traditional	Informal and expanding	Emerging and diversifying	Modernizing and formalizing	Industrial and consolidated
Overall N	638 192		123 286	119 471	106 079	137 837	151 519
	N	%	%	%	%	%	%	P
INDIVIDUAL-LEVEL MEASURES
Age
15–29 years	190 721	33.0	46.0	41.0	31.0	27.0	21.0	<.001
30–49 years	233 306	36.0	36.0	38.0	37.0	38.0	33.0
50–69 years	156 635	23.0	14.0	18.0	25.0	26.0	31.0
70–89 years	54 280	7.0	2.8	3.2	6.6	8.1	14.0
90–100 years	3244	0.5	0.4	0.3	0.4	0.4	0.8
Gender, Female	334 734	51.0	51.0	51.0	52.0	50.0	51.0	<.001
Marital status
Single	196 100	32.0	30.0	38.0	35.0	30.0	30.0	<.001
Married	350 979	56.0	61.0	53.0	53.0	57.0	55.0
Divorced, separated, or widowed	87 712	12.0	9.0	9.3	12.0	13.0	14.0
Households with children under 15 years	320 041	53.0	80.0	71.0	51.0	42.0	28.0	<.001
Education
Completed primary school	161 618	34.0	64.0	43.0	34.0	23.0	15.0	<.001
Some secondary school	334 972	51.0	32.0	49.0	52.0	56.0	61.0
Completed secondary school	141 602	15.0	5.0	8.0	14.0	21.0	24.0
Employment
Working full-time	272 388	41.0	36.0	38.0	39.0	46.0	46.0	<.001
Working part-time	98 704	16.0	22.0	19.0	14.0	10.0	13.0
Out of the workforce	41 311	7.0	8.0	8.0	8.0	7.0	5.0
Unemployed	222 388	36.0	34.0	35.0	38.0	37.0	36.0
Urbanization
Rural area	146 385	24.0	42.0	28.0	27.0	14.0	13.0	<.001
Small town or village	213 352	34.0	35.0	32.0	31.0	33.0	37.0
Large city	200 626	30.0	15.0	29.0	30.0	42.0	33.0
Suburb of a large city	77 829	12.0	8.0	11.0	12.0	11.0	17.0
Not enough money for food	224 472	37.0	61.0	54.0	37.0	27.0	13.0	<.001
COUNTRY-LEVEL MEASURES

	Overall		Rural and traditional	Informal and expanding	Emerging and diversifying	Modernizing and formalizing	Industrial and consolidated
Number of countries	135		28	27	22	28	30
	N	%	%	%	%	%	%	P
Wealth quintile
Lowest	101 693	20.0	20.0	20.0	19.0	20.0	20.0	.999
Second	111 047	20.0	20.0	20.0	20.0	20.0	20.0
Middle	122 578	20.0	20.0	20.0	20.0	20.0	20.0
Fourth	135 913	20.0	20.0	20.0	20.0	20.0	20.0
Fifth	163 110	20.0	20.0	20.0	20.0	20.0	20.0
Country Income
Low-income	62 213	10.0	46.0	4.0	0	0	0	<.001
Low-middle income	184 731	29.0	54.0	71.0	23.0	7.0	0
Upper-middle income	172 750	27.0	0	25.0	69.0	47.0	3.0
High-income	218 498	34.0	0	0	8.0	46.0	97.0
Region
Europe	164 279	26.0	0	0	14.0	36.0	66.0	<.001
Former Soviet Union	64 917	10.0	4.0	11.0	20.0	15.0	3.0
Asia	122 498	19.0	36.0	25.0	13.0	11.0	14.0
Americas	92 122	14.0	0	18.0	24.0	17.0	14.0
Middle East and North Africa	65 157	10.0	3.0	12.0	14.0	20.0	3.0
Sub-Saharan Africa	129 201	20.0	58.0	34.0	16.0	0	0

Notes: P-value based on chi-squared tests for categorical variables with country food system typology type as a continuous variable. Bolded P-values indicate statistical significance. Data are survey-weighted. Percentages are column percentages.

Unadjusted results for frequency of cooking at home and eating home-cooked meals

Table 3 presents the unadjusted mean number of days spent cooking and eating meals at home by food system typology and year. Overall, the unadjusted mean number of days cooking meals at home ranged from 2.83 days (95% CI: 2.79, 2.88) in 2018 to a high of 3.09 (95% CI: 3.04, 3.14) in 2021. Frequency of cooking meals then declined to 2.89 (95% CI: 2.84, 2.93) in 2022. Frequency of cooking meals at home differed across food system typologies in all years (2018, 2020, 2021, and 2022) (P < .001), and 2019 (P = .011), with the highest being in 2020 in the industrial and consolidated typology (3.48, 95% CI: 3.45, 3.51) and lowest being in 2018 in the modernizing and formalizing typology (2.83, 95% CI: 2.79, 2.87). The unadjusted mean number of days eating meals cooked at home declined from a high of 5.55 (95% CI: 5.50, 5.59) in 2018 to a low of 4.99 days (95% CI: 4.95, 5.04) in 2020. Across food system typologies, there was a statistically significant difference in the frequency of eating meals cooked at home in 2018, 2019, 2021 (P < .001), and 2022 (P = .002), but not in 2020 (P = .863).

Table 3.

Unadjusted mean number of days cooking and eating lunch and dinner at home by food system typology, 2018–22

	Overall	Rural and traditional	Informal and expanding	Emerging and diversifying	Modernizing and formalizing	Industrial and consolidated	P
Mean number of days per week cooking lunch and dinner at home (95% CI) by year
2018	2.83 (2.79, 2.88)	2.84 (2.79, 2.89)	2.89 (2.84, 2.93)	3.25 (3.21, 3.31)	2.83 (2.79, 2.87)	3.12 (3.08, 3.15)	<.001
2019	2.99 (2.95, 3.04)	3.03 (2.98, 3.07)	2.97 (2.92, 3.01)	3.20 (3.15, 3.25)	2.86 (2.83, 2.90)	3.17 (3.14, 3.20)	.011
2020	2.99 (2.94, 3.05)	3.19 (3.13, 3.25)	3.07 (3.02, 3.12)	3.20 (3.16, 3.25)	3.26 (3.22, 3.30)	3.48 (3.45, 3.51)	<.001
2021	3.09 (3.04, 3.14)	3.21 (3.15, 3.26)	3.15 (3.10, 3.21)	3.24 (3.19, 3.29)	3.24 (3.20, 3.28)	3.44 (3.41, 3.47)	<.001
2022	2.89 (2.84, 2.93)	2.91 (2.87, 2.95)	3.07 (3.03, 3.12)	3.13 (3.08, 3.18)	2.94 (2.90, 2.98)	3.26 (3.22, 3.29)	<0.001
Mean number of days per week eating lunch and dinner prepared at home (95% CI) by year
2018	5.55 (5.50, 5.59)	5.35 (5.31, 5.39)	5.29 (5.25, 5.33)	5.54 (5.51, 5.58)	5.10 (5.07, 5.14)	4.94 (4.92, 4.97)	<.001
2019	5.41 (5.37, 5.45)	5.17 (5.14, 5.21)	5.20 (5.16, 5.24)	5.31 (5.27, 5.35)	5.02 (4.98, 5.05)	4.71 (4.68, 4.74)	<.001
2020	4.99 (4.95, 5.04)	5.10 (5.05, 5.15)	4.81 (4.77, 4.85)	5.02 (4.98, 5.06)	5.09 (5.06, 5.12)	4.95 (4.92, 4.97)	.863
2021	5.28 (5.23, 5.33)	5.20 (5.15, 5.25)	5.02 (4.98, 5.06)	5.01 (4.97, 5.05)	4.89 (4.85, 4.92)	4.78 (4.75, 4.81)	<.001
2022	5.19 (5.15, 5.24)	5.00 (4.96, 5.04)	5.03 (4.99, 5.07)	5.25 (5.21, 5.29)	5.04 (5.00, 5.07)	4.66 (4,63, 4.69)	.002

Notes: P-value based on unadjusted negative binomial regressions for continuous variables with year and country food system typology type as continuous variables. Bolded P-values indicate statistical significance. Data are survey-weighted.

Frequency of cooking meals at home: adjusted results

Supplementary Table 2 presents associations between food system typology and year with frequency of cooking meals at home from fully adjusted models. Adjusted for all covariates, overall, participants from emerging and diversifying and industrial and consolidated food systems cooked meals on average 13% (95% CI: 1.10, 1.16, P < .001) and 5% (95% CI: 1.03, 1.08, P < .001) more in 7 days, respectively compared to participants in rural and traditional food systems. Compared to males, females cooked lunch and dinner 214% more in 7 days (P < .001). Compared to individuals <15–29 years, individuals 30–49 years and 50–69 years cooked 13% and 7% more frequently in 7 days, respectively (P < .001). Individuals with higher education cooked 2% less frequently in 7 days than those who completed primary school (P = .001). Overall, having enough money for food and living in a nonrural area were associated with less frequent cooking at home (P < .001), whereas having children in the home, being employed part-time, unemployed or out of the workforce, being either married, separated, divorced, widowed or having a domestic partner (compared to being single or never married) were associated with more frequent cooking at home (P < .001). Individuals in high-income countries, in Asia or the Americas cooked more frequently at home (P < .001), whereas those living in the Middle East and North Africa cooked less frequently (P < .001).

Postestimation margins and pairwise comparisons suggest that across most food system typologies, the predicted mean number of days cooking meals at home gradually increased between 2018 and 2019, peaking in 2020 and 2021 (depending on gender) and then declined in 2022, trending towards prepandemic levels (Fig. 1a). For example, the overall predicted mean number of days cooking meals at home in 2018 ranged from 2.83 days (95% CI: 2.79, 2.87) (modernizing and formalizing) to 3.20 days (95% CI: 3.15, 3.25) (emerging and diversifying). In 2020, rural and traditional (3.27, 95% CI: 3.20, 3.34, P < .001 from 2018), informal and expanding (3.29, 95% CI: 3.24, 3.35, P < .001 from 2018) and emerging and diversifying peaked (3.26, 95% CI: 3.21, 3.31), whereas modernizing and formalizing (3.36, 95% CI: 3.32, 3.41, P < .001 from 2018) and industrial and consolidated (3.43, 95% CI: 3.38, 3.47, P < .001 from 2018) peaked in 2021. While all food system typologies appear to decline as of 2022, cooking meals at home in emerging and diversifying food systems has declined beyond 2018 levels, though still remain higher than other typologies (3.12, 95% CI: 3.07, 3.17, P < .05) (Supplementary Table 4).

Figure 1.

Predicted mean number of days cooking lunch and dinner at home by typology, overall (Panel A) and by gender (Panels B and C), 2018–22 (N = 638 192). Based on postestimation margins after negative binomial models using survey weights and adjusted for age, gender (overall model only), education, marital status, urbanization, employment status, ability to afford food, children in the household under 15, wealth quintile, country income, and region.

When stratified by gender, females from modernizing and formalizing and industrial and consolidated food systems cooked meals on average 5% (95% CI: 0.93,0.97, P < .001) and 6% (95% CI: 0.92,0.96, P < .001) less in 7 days as compared to females in rural and traditional food systems. Interestingly, males from all food system typologies cooked on average 9-35% more than males in rural and traditional food systems (P < .005 for informal and expanding; P < .001 for expanding and diversifying, modernizing and formalizing and industrial and consolidated). With the exception of females in 2022, across the study period, stratified adjusted models suggest that on average days spent cooking lunch and dinner at home over 7 days was greater than in 2018 than in subsequent years (P < .001) (Fig. 1b and c). In addition, pairwise comparisons suggest that across most food system typologies, the predicted mean number of days cooking lunch and dinner at home peaked in 2020 for both females and males, except for females in emerging and diversifying typologies (Supplementary Table 4).

Frequency of eating meals cooked at home: adjusted results

Results from adjusted models for frequency of eating meals prepared at home overall and stratified by gender are presented in Supplementary Table 3. Postestimation margins showing the predicted mean frequency of eating meals prepared at home in each year by food system typology are shown in Fig. 2. In the adjusted model, overall, participants from all food system typologies ate lunch and dinner cooked at home on average 4%–10% less in 7 days (P < .001) as compared to participants in rural and traditional food systems. Further, participants ate meals cooked at home on average 4%–8% less frequently in 7 days in survey waves 2019–22, as compared to 2018 (P < .001). Full model results are available in Supplementary Table 3. Compared to males, females ate meals that were cooked at home 3% more in 7 days (P < .001). Compared to individuals <15–29 years, individuals 30–89 years more frequently eat meals prepared at home (P < .001). Individuals who completed secondary school ate meals cooked at home less frequently than those who completed primary school (P ≤ .001). Having children in the home, living in a nonrural area, having enough money for food and being divorced, separated or widowed were associated with less frequent eating of meals prepared at home (P < .001) whereas, being employed part-time, unemployed or out of the workforce, being either married (compared to being single or never married) were associated with more frequently eating meals cooked at home (P < .001). Individuals in upper-middle and high-income countries in Asia, the Middle East and North Africa, and Sub-Saharan Africa ate home-cooked meals less frequently (P < .001), whereas those living in Americas ate home-cooked meals more often compared to individuals living in Europe (P < .001). Unlike cooking at home, pairwise comparisons suggest that across most food system typologies, the predicted mean number of days eating lunch and dinner cooked at home was highest in 2018 and has slowly declined (Fig. 2a and Supplementary Table 5).

Figure 2.

Predicted mean number of days eating lunch and dinner prepared at home by typology, overall (Panel A) and by gender (Panels B and C), 2018–22 (N = 638 192). Based on postestimation margins after negative binomial models using survey weights and adjusted for age, gender (overall model only), education, marital status, urbanization, employment status, ability to afford food, children in the household under 15, wealth quintile, country income, and region.

When stratified by gender, both females and males from all food systems ate lunch and dinner on average 4%–10% less frequently in 7 days as compared to females and males in rural and traditional food systems (Fig. 2b and c). Across the study period, stratified adjusted models suggest that on average females spent 4-8% fewer days eating lunch and dinner cooked at home over 7 days in 2019–22 as compared to 2018 (P < .001), whereas males spent 2% (2019, 95% CI: 0.97, 0.99, P = .004; 2021, 95% CI: 0.96, 1.00, P = .02) to 5% fewer days (2020, 95% CI: 0.93, 0.97, P < .001; 2022, 95% CI: 0.94, 0.97, P < .001) compared to 2018 (Supplementary Table 3). In addition, pairwise comparisons suggest that across most food system typologies, the predicted mean number of days eating lunch and dinner cooked at home was highest in 2018 for females and have slowly declined (Fig. 2b and Supplementary Table 5). Pairwise comparisons in males indicate a varying pattern across typologies (Fig. 2c).

Supplementary analyses on lunch and dinner separately are presented in Supplementary Tables 6–9 and are generally consistent with the findings from lunch and dinner combined in the main analyses.

DISCUSSION

This study examined trends from 2018 to 2022 in the frequency of cooking and eating meals at home on a global scale across five food system types, using data from nationally representative samples in 135 countries. We find that the mean number of days cooking meals at home and eating meals prepared at home, and trends over time for both cooking and eating meals at home, differed by food system typology and by gender. Three key findings emerged. First, overall, females cooked meals at home twice as frequently as males; however, there was little difference in eating meals prepared at home between females and males. Across most typologies, males did cook meals at home more frequently in 2020 and 2021 compared to 2018, however, the increase was not enough to close the disparity in cooking frequency between males and females. Furthermore, cooking frequency among males declined in 2022, but not to prepandemic levels. Second, on average, participants in emerging and diversifying food systems cooked meals at home more frequently compared to participants in rural and traditional food systems. However, these home-cooked meals were consumed less frequently in comparison. Third, across food system typologies, cooking meals at home became more frequent during the COVID-19 pandemic before declining again in 2022 towards prepandemic levels. The mean number of days cooking meals at home increased significantly from 2018 across food system types until 2020 and 2021 and have since declined to at or below prepandemic levels. There was greater variation in eating meals cooked at home with only emerging and diversifying, modernizing and formalizing, and industrial and consolidated showing a similar trend.

Stark gender inequities are both a cause and consequence of unsustainable and unjust food systems, their value chains, the food environment, and consumer behavior (Njuki et al. 2023). Importantly, these inequities are exacerbated in how men and women are differentially exposed to shocks and stressors and their capacity to respond to them, which include some of the largest threats facing humanity—climate change, conflict, and COVID-19 (Bryan et al. 2023). Our findings corroborate the notion that women consistently devote more time to unpaid domestic work, such as household cooking than men (Flagg et al. 2014, Sliwa et al. 2015), which appears to be greatest in less globalized and modernized food systems. Prior evidence has shown consistent gender disparities in frequency of cooking around the globe (Wolfson et al. 2021). The present findings add to this evidence base showing that these disparities exist in all food system typologies, though there were somewhat different trajectories across food system typologies during the COVID-19 pandemic. According to UN Women, estimates suggest women spend an average of nearly 18% of their day (4.3 h) on unpaid care and domestic work compared to men who spend 6.5% of their day (1.6 h). In other words, women spend nearly three times as much of their day on unpaid care and domestic work compared to men (UN Women 2014). During the pandemic, this disparity became even more striking due to women’s economic and social insecurity in formal and informal markets and participation in childcare because of school closures (OECD 2021). While long-term impacts have yet to be fully estimated, the 2023 Global Gender Gap Report suggests it will take 130 years to reach parity across four key dimensions, including economic participation and opportunity, educational attainment, health and survival, and political empowerment—a 30-year increase from their 2020 estimation (World Economic Forum 2023). Importantly, the present findings show that while more frequent cooking at home was not sustained as many food systems appear to have reverted to similar levels of cooking at home from 2018, a gender gap remains between females and males.

It is important to understand how socioeconomic factors affect the decision and ability of women to engage in unpaid and domestic care work such as food preparation. The key, often overlapping, dimensions considered in the literature include the level of economic development and the nature of growth; educational attainment; household income; social dimensions, such as social norms influencing marriage, fertility, and women’s role in and outside of the household; and political context (laws, social protection, benefits). In the current study, adjusted analyses indicated that a higher frequency of cooking and eating meals cooked at home was associated with lower educational attainment, living in rural areas, being employed part-time, unemployed or out of the workforce, and not having enough money for food. These sociodemographic factors relate to food and economic insecurity and food choice attributes such as perceived time. With increasing household incomes, the share of expenditure on food decreases, while time scarcity and the demand for higher-quality food increases (Ambikapathi et al. 2022). This relationship has been explored primarily in high-income countries or those within the industrial and consolidated typology (Feudtner et al. 2014, Mills et al. 2018). However, there are few population-based studies in the literature that have examined these factors in low- and middle-income countries, especially countries within the emerging and diversifying typology in relation to both home meal preparation and consumption (Mills et al. 2017b). It is also worth noting that cooking at home does not necessarily equate to higher diet quality. In part this is due to a lack of consensus on what constitutes home cooking (Wolfson et al. 2016, Raber and Wolfson 2021, Yuan et al. 2022), which is particularly important as food systems evolve and new products such as highly processed, shelf-stable, convenience products (e.g. ultra-processed foods) become more widely available and more ubiquitous in transitioning food environments. Additionally, as social, economic, and food environment conditions change, households may engage in unhealthy coping mechanisms such as food rationing and consumption of poor-quality foods, especially during times of crisis (González-Monroy et al. 2021, Ismail et al. 2023). Low-income households, while still cooking at home, may rely on nutrient-poor foods such as staple grains (e.g. wheat, rice, soy and maize) and shelf-stable ultra-processed foods due to the income and expenditure constraints of a healthy diet (i.e. nutrient-dense foods such as fruits and vegetables, nuts and animal-sourced products) (Wolfson et al. 2020, 2024, Headey et al. 2023).

Policy and practice implications

Global dialogues on improving food systems to ensure safe, affordable and nutritious diets have accelerated since the United Nations Food Systems Summit in 2021. The Summit highlighted the need for comprehensive and sustainable pathways, rooted in scientific evidence to guide the transformation of food systems. In anchoring to the Sustainable Development Goals, the adoption of food systems transformation pathways in over 120 countries aid in identifying and designing policy instruments that incentivize transformative changes across sectors (UN Food Systems Hub 2024). Findings from the present study suggest that a variety of evidence-based policies and interventions are needed at national, regional and local levels to create enabling environments and encourage practices that support healthy diets across all food systems. Importantly, the consideration of equity should be paramount to ensure certain population groups are not further marginalized with the implementation of such policies and interventions (Friel et al. 2015, EAT–Lancet Commission 2023). At a population level, reducing economic and physical barriers to healthy diets by improving purchasing power (e.g. create income-generation opportunities, cash transfer schemes), addressing food prices (e.g. vouchers and subsidies) or investing in supply chain infrastructure (e.g. cold storage), can improve the accessibility, affordability and availability of nutritious foods in markets (Durao et al. 2020). At an individual level, behavioral interventions include skills-building (including food preparation, home/community gardens), and nutrition education have been widely adopted (Iacovou et al. 2013). Reviews on cooking interventions suggest that increased frequency of home cooking and enhanced culinary skills result in improved dietary quality of adults (Reicks et al. 2014, 2018). It is important to note however, these nutrition-related policies and interventions may not necessarily improve gender disparities in home cooking; therefore complementary policies and interventions that address structural constraints imposed by lack of public services and restrictive labor policies should explicitly focus on promoting a redistribution of time spent on unpaid domestic work between men and women (Alonso et al. 2019).

Strengths and limitations

Our study has several strengths, including the use of empirical data from nationally representative samples in nearly 140 countries across a 5-year period and the ability to analyze data from diverse geographical regions. To our knowledge, this is the first study to pair food system typologies with global data on cooking and eating meals at home over a 5-year period. Additionally, by combining lunch and dinner as an outcome, we considered different normative values on main meals across the globe. It is important to note, however, these data relied on multiple cross-sections of self-reported behaviors. Participants may have over-reported their home cooking if they perceived it to be healthy and more socially desirable. Importantly, in this survey, participants were prompted to consider scratch cooking, rather than takeaways, packaged or preprepared foods for the cooking and eating home cooked meal questions. The introduction to the questions about home cooking read “By ‘cooked at home’ we mean a meal prepared at home mostly from ingredients, such as vegetables, meats, and grains.” However, similar to most surveys on food preparation, “cooking” was not further defined (Yuan et al. 2022), which could have resulted in differing interpretations of the meaning of cooking (Wolfson et al. 2016, Yuan et al. 2022) and could have biased results if cooking is interpreted differently or if there are different cultural norms about whether to include heating up leftover food, for example. In a global context clearly defining what is meant by a home cooking or a home-cooked meal is a particular challenge (i.e. which foods constitute home cooking), and is critical to enable consistent methodology, informed comparisons and conclusions in research, and effective public health promotion. Therefore, future research should develop clear terminology and adopt a common conceptual framework around home cooking (Raber et al. 2016) as well as explore potential associations between dietary intakes and patterns of meal sourcing to determine which interventions most effectively change patterns of consumption that improve nutrition.

Finally, food system typologies are everchanging and are nuanced within countries. We used static typologies that classified countries at a national level as of 2020 and therefore, it is possible that some countries shifted in their classification after 2020. In large countries, it is also possible that multiple typologies coexist subnationally, such as Brazil, China, and India. However, these food system typologies have not been established for lower administrative levels due to lack of data required for each indicator to construct typologies.

CONCLUSION

From 2018 to 2022, across the globe and across different food system typologies, the frequency of cooking at home increased over time, peaking in 2020 and 2021 at the height of the COVID-19 pandemic, before declining to prepandemic levels in 2022. Previously documented disparities in cooking frequency based on gender persisted, and from 2018 to 2022, in all food system typologies, females cooked more than males. The ongoing global agenda for food systems transformation, which aims to accelerate the achievement of Sustainable Development Goals, should consider policy measures such as promoting gender equity in household food preparation, while ensuring that all people, regardless of the food system in which they live, have the ability to have food security and consume a healthy diet regardless of how frequently they cook at home or eat home-prepared meals. However, this will only be possible if there is a global food supply that is accessible, sustainable, safe, affordable, and above all, nutritious.

Author contributions

B.C., J.F. and J.A.W. conceptualized the study. B.C. and J.A.W. developed the analysis plan and B.C. conducted the analyses and interpreted the results. B.C. wrote the first draft of the manuscript, and all authors revised and approved the final manuscript as submitted.

Conflict of interest

None to declare.

Funding

This research received no support from any funding agency, commercial or not-for-profit sectors. Bianca Carducci was supported by the Canadian Institutes for Health Research Banting Fellowship and Columbia Climate School Postdoctoral program. Julia Wolfson was supported by grant K01DK119166 from the National Institutes of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health.

Data availability

Data described in the manuscript will not be made available because it requires permission by Gallup World Poll. Analytical code will be made publicly and freely available without restriction on GitHub.

Ethics statement

This study was conducted in accordance with the requirements of the code of Federal Regulations on the Protection of Human Subjects (45 CFR 46), consistent with the Declaration of Helsinki, including its relevant subparts and approved by the Gallup Institutional Review Board. Verbal informed consent was obtained from all adults 18 and older including parental consent from minors.