According to the Centers for Disease Control and Prevention, more than half of American adults – six in ten adults in the USA – have one or more chronic diseases(1). Many of these, including CVD, high blood pressure, type 2 diabetes, some cancers, and chronic lung and kidney disease, are related to poor diet quality(1). Concurrent with these persistently high levels of obesity and chronic diseases, trends in food intake over time reflect poor eating choices(Reference Bentley2). Data from the National Health and Nutrition Examination Survey (NHANES) show that, in the 2015–2016 NHANES, Healthy Eating Index-2015 (HEI-2015) Scores, a measure of how food choices align with United States Department of Agriculture (USDA) Dietary Guidelines, have remained below fifty-nine out of 100 possible points(3). In addition, too many energies, added sugars, and saturated fats and too few whole grains, fruits, and vegetables are consumed, according to the USDA(Reference Bentley2).
Research has shown that food choices are influenced by many factors. These include prices and income, time constraints, environmental factors such as the proximity of grocery stores and restaurants, community characteristics, family structure, psychological factors, nutritional information and participation in Federal food and nutrition assistance programmes(Reference Kalenkoski and Hamrick4). Current studies indicate that factors related to total food expenditures and time use are major determinants of diet quality(Reference Darmon and Drewnowski5–Reference Venn and Strazdins8).
Research papers in the nutritional sciences, economics and time use fields often tackle the same issues but do so independently of each other. Nutritional scientists focus on detailed dietary data to assess diet quality. Economists focus on the effects of prices and income on food purchases and consumption. Time use researchers examine how people spend their time on various activities, including time spent preparing and eating food.
From the best of the authors’ knowledge, no previous study has explored the relationships between time and money expenditures and food nutritional quality jointly. Aguiar and Hurst estimated a household production function with time and money on the right-hand side of the equation as we did, but their dependent variable is food consumption rather than diet quality(Reference Aguiar and Hurst9,Reference Aguiar and Hurst10) . A few studies have analysed the relationship between money expenditures and food nutritional quality, without functioning time(Reference Darmon and Drewnowski5,Reference Rehm, Monsivais and Drewnowski6,Reference Rao, Afshin and Singh11) . Rao et al.(Reference Rao, Afshin and Singh11) summarises findings from ten studies reporting twenty-nine cost differences between more- and less-healthy diet patterns. Twenty-seven of the reported estimated differences showed that healthier dietary patterns were more expensive than less-healthy patterns. Money expenditures used in all these studies do not correspond to actual amounts spent on food but to an estimated cost of the diet. Food quality, on the other hand, is defined in a variety of ways. For example, Drewnowski et al.(Reference Drewnowski, Darmon and Briend12) defined food quality in terms of consumption of relevant food categories such as vegetables and fruits, while Waterlander et al.(Reference Waterlander, de Haas and van Amstel13) considered the energy density, and others look at the HEI, a standard measure to assess compliance with the Dietary Guidelines for Americans(Reference Carlson, Dong and Lino14,Reference Krebs-Smith, Pannucci and Subar15) .
In the economics literature, Carlson et al.(Reference Carlson, Dong and Lino14) examined the relationship between food nutritional quality and diet cost in the USA using dietary recall (24 h) data from the NHANES 2003–2004 and a nationally representative food prices database for the same year. Diet quality was measured using the HEI-2005 calculated using dietary recall data. Because the NHANES does not report food expenditures, the study estimated the diet cost indirectly using data on quantities obtained in the NHANES and prices from the food prices database. In this study, diet cost was found to have a very low association with food nutritional quality (a $1/d increase in diet cost is found to be associated with a 1·9-point increase in the HEI-2005).
Several studies examining the relationship between the HEI and diet cost have also been carried out by public health and nutrition experts. For example, Beydoun et al.(Reference Beydoun, Fanelli-Kuczmarski and Allen16) analysed the association between the HEI-2010 and diet cost using the same approach and price database used by Carlson et al.(Reference Carlson, Dong and Lino14) with data from a sample of individuals living in Baltimore, MD in 2004. Findings from this study suggested a small positive association between the HEI and food expenditures: a $1/d increase in food expenditures (about a 20 % increase in the average diet cost per day) was found to be associated with a 1·65 higher HEI-2000. Similar studies have also been conducted by Rehm et al.(Reference Rehm, Monsivais and Drewnowski17) and Rehm et al.(Reference Rehm, Monsivais and Drewnowski6) using the same price database as Carlson et al.(Reference Carlson, Dong and Lino14) and Beydoun et al.(Reference Beydoun, Fanelli-Kuczmarski and Allen16) but corresponding to different years. A limitation of these studies is the fact that they only focus on the correlation between HEI and diet cost.
Few studies that have examined the relationship between time spent preparing food and diet quality have failed to factor in food expenditures(Reference Mancino and Gregory18,Reference Monsivais, Aggarwal and Drewnowski19) . These studies differ in terms of data sources as well as the measure of diet quality and time expenditures. Mancino and Gregory(Reference Mancino and Gregory18) used data from a survey of 400 women in the Chicago area in 2007. Diet nutritional quality was using the HEI-2005 calculated using a 24-h dietary recall, whereas that time expenditures were measured using the number of minutes a person spent in food preparation on the day the survey was conducted. The study found no evidence of a relationship between time spent cooking and diet quality. Monsivais et al. (Reference Monsivais, Aggarwal and Drewnowski19) used data from a survey of households conducted in Seattle in 2008–2009. Diet quality in this study was measured by frequency of consumption of six food groups reflecting healthier and less healthy intakes: fruit, green salad, vegetables other than salad or potatoes, fruit juice, sugar-sweetened beverages and sweetened grain-based snacks. Time expenditures included time spent preparing, cooking and cleaning up after meals. The study found that spending more on food preparation at home was associated with increased consumption of fruits and vegetables which resulted in higher diet quality.
Because the data that are collected and utilised in these three different fields capture different aspects of this production of diet quality, the household production function for diet quality in which time and money expenditures are inputs cannot be estimated with existing data. Therefore, there is a need to create a survey instrument that captures important pieces of information from all three disciplines. The purpose of this paper is to describe such a comprehensive survey instrument. Knowledge from the nutrition, economics and time use disciplines is incorporated into the design of this survey. The paper also describes the results of a pilot test of this survey on a sample of faculty and staff at an institution of higher education as well as some preliminary analyses exploring the role of time and money expenditures on dietary quality.
Methods
Questionnaire development
A comprehensive survey instrument was constructed by pulling already-tested questions from three nationally representative US surveys. These include the Consumer Expenditure Survey(20), the American Time Use Survey and its Eating and Health Time Use Module(21), and the University of Michigan Health and Retirement Study(22). The first two surveys are collected by the U.S. Bureau of the Census for the U.S. Bureau of Labor Statistics. The last is supported by the National Institute on Aging and the Social Security Administration. Questions from the Consumer Expenditure Survey used in this study are related to expenditures on food at home, food away from home and alcoholic beverages. Questions from the Eating and Health Time Use Module are about time spent eating, time spent purchasing and preparing foods, secondary eating (i.e. eating while doing other activities), respondent’s height and weight, self-reported health status, and food and nutrition assistance programme participation. Finally, FFQ and questions about food consumption and nutrition (including vitamins and other supplements) were obtained from the 2013 Health and Retirement Study: Health Care and Nutrition Study(22) which were originally obtained from the Harvard FFQ for adults(23). In addition, standard questions about the sociodemographic characteristics of the individual and the household, including gender, family size, education, income, race and ethnicity, are included in this survey instrument.
To pre-test the survey instrument, all the researchers involved in the study were asked to complete the survey. Feedback from the pre-test was used to correct errors and to modify some questions to ensure consistent use of units and language.
Recruitment and data collection procedures
The target population for the pilot test of the survey were faculty and staff aged 18 years and older from a public university located in West Texas, USA. Participants were recruited through campus announcements sent via email from July 2017 to November 2017. The data collected are cross-sectional. The recruitment announcements were sent once per week and provided general information about the study such as the purpose of the study, the target population, the survey they needed to complete, the approximate time it would take to respond, the probability of receiving an incentive, the link to the survey and the contact information for the research assistant. All interested participants were invited to follow the survey link, which was provided through the campus announcement, to complete the questionnaire using Qualtrics software (Qualtrics), an online survey tool that allows researchers to develop, distribute and analyse online surveys in real time. All participants who completed the questionnaire were asked if they would like to participate in a drawing to win one of five gift cards.
Diet quality assessment
The FFQ was used to obtain usual intake of 171 food items over the past 12 months. It included food items from the following food groups: dairy foods; fruits and vegetables; eggs, meats and fish; cereals, breads and starches; beverages (sugar-free, not sugar-free and other beverages); and sweets, baked goods and miscellaneous. Each food item was assessed by using eight categories that ranged from never to ≥2 times/d or nine categories that ranged from never to ≥6 times/d for some beverage and food items. Additionally, each food item from the FFQ specified the serving size according to the USDA Food and Composition Database. According to Emmett et al.(Reference Emmett, Jones, Northstone and Pounis24) and Kristal et al.(Reference Kristal, Shattuck and Williams25), the average energy and nutrients intakes can be estimated over the time frame of the questionnaire. The procedure explained by the Women’s Health Initiative(Reference Neuhouse26) and by the General Nutrition Assessment(27) of calculating nutrient intakes from FFQ was followed in this study. The following steps were taken to calculate the daily nutrient intake:
-
1. Estimate annual servings of a food by multiplying the reported weekly frequency and portion size from the USDA Food Composition Databases.
-
2. Obtain daily servings by dividing annual servings by 365 d.
-
3. Obtain daily portion size by dividing the daily serving (in g, ounces or millilitres of the food item) by the standard grams, ounces or millilitres per serving size according to the USDA Food and Composition Database.
Once a daily portion size for all food items was calculated for each participant, energies (kcal), MUFA (g), PUFA (g), saturated fats (g), Na (mg) and added sugar (g) were calculated using the Food Processor Nutrition Analysis Software version 11.1 which uses the latest version of the USDA Food Composition Database.
Diet quality was then determined using the HEI-2015(Reference Krebs-Smith, Pannucci and Subar15). The HEI-2015 contains thirteen dietary components: total fruits, whole fruits, total vegetables, greens and beans, dairy, total protein foods, seafood and plant proteins, fatty acids, refined grains, Na, added sugar and saturated fats. The intake of the dietary components is scored based on its respective density in comparison with the standards from USDA’s Healthy US-Style Eating Patterns(Reference Krebs-Smith, Pannucci and Subar15). Minimum and maximum scores are 0 and 100 points, respectively. A nutritionist calculated the scores for each component following the scoring standards of HEI-2015(Reference Krebs-Smith, Pannucci and Subar15) shown in detail in Table 1.
Table 1 HEI-2015 components and scoring standards*
HEI, Healthy Eating Index; eq., equivalent; oz., ounce.
* Intakes between the minimum and maximum standards were scored proportionately.
† Includes 100 % fruit juice.
‡ Includes all forms except juice.
§ Includes legumes (beans and peas).
|| Includes all milk products, such as fluid milk, yogurt and cheese, and fortified soya beverages.
¶ Includes legumes (beans and peas).
** Includes seafood, nuts, seeds, soya products (other than beverages) and legumes (beans and peas).
†† Ratio of PUFA and MUFA to SFA.
Money inputs
Money input expenditures include expenditures on groceries for food-at-home consumption; expenditures on meals or snacks at restaurants, cafeterias, carry outs, etc.; the value of ‘free’ meals at work; and the value of or the expenditures on meals eaten at school. Money expenditures are weekly expenditures for all household members.
Time inputs
For the calculation of time-input expenditures, the following were included as food-at-home-related activities: grocery shopping (including travel time and time spent shopping), food preparation, eating and clean-up. For food-away-from-home activities, meals at fast-food chains, eating places and restaurants (including travel time and time spent shopping) were included. All time use variables are measured on a weekly basis although, in several instances, survey respondents were given the option to provide time use information on a daily or weekly basis (e.g. time spent preparing, eating and cleaning up at home). The time use section only gathered information about the time spent on food-related activities by the survey respondent, not the household.
Theoretical model
The household production model originally developed by Becker in 1965 provides a useful framework for the analysis of the relationships between the quality of food consumed by households and total food expenditures and total time spent on food-related activities(Reference Becker28–Reference Variyam, Blaylock and Smallwood30). The household is assumed to derive utility from both market and household-produced goods such as health. The framework takes into account individuals’ time and budget constraints, market prices for all goods, the times used to produce/consume goods, the diet quality production process (using purchased food and household time) and the health production process (with diet quality as one of the inputs). Moreover, it is assumed that decision-making follows a multi-stage decision process. At a higher stage, they allocate money and time to broad groups of goods and activities (e.g. money and time devoted to food and recreation) and subsequently make decisions to allocate money and time in subgroups (e.g. money and time for food at home and away from home). Thus, as shown in the appendix, the following model can be used to explore the role of total expenditures on food (i.e. money expenditures) and total time spent on food production/consumption (i.e. time expenditures) on dietary quality:
$${\rm{DQ}} = {\bf{g}}({{\rm{E}}_F},{t_{{\rm{DQ}}}},{\bf{C}}),$$
where DQ is the diet quality, EF is the total food expenditures, tDQ is the total time needed to produce diet quality, C are household characteristics and g(.) denotes a function.
Data analyses
Two types of analyses were conducted using the survey data. First, survey respondents’ behaviour was evaluated. Second, the relationships between food nutritional quality and time and money expenditures were analysed. This second type of analyses was conducted to test the hypothesis that more time spent in food-related activities (grocery shopping, preparing, eating and cleaning) and higher food expenditures are associated with higher dietary quality among participants.
Survey respondents’ behaviour was assessed using basic descriptive statistics on completion rates, survey duration and sociodemographic characteristics. Data obtained from completed surveys were subsequently used to analyse the relationships between food nutritional quality and time and money expenditures using empirical versions of equation (1) (main equation of interest). The models were specified as multiple linear regression models with the natural logarithm of the HEI as the dependent variable and the natural logarithm of total time spent on food-related activities by the respondent, the natural logarithm of total household expenditures and the respondent’s and household’s sociodemographic characteristics (household composition, education, gender, race, ethnicity and age of the respondent) as explanatory variables. Two versions of the model in equation (1) were considered, one including separate values of time and money expenditures related to food at home and food away from home (model 1), as well as a model where time and money expenditures were aggregated (model 2). Total household income and variables related to hours worked per week (employment status) were not included as explanatory variables in the initial specifications of equation (1) as they are assumed to affect only decisions at a higher stage in the multi-stage budgeting process.
To complement the analysis, two additional models were considered with the natural logarithm of HEI as the dependent variable. Model 3 includes as explanatory variables the number of meals consumed away from home by the survey respondent (per week) and sociodemographic characteristics. This model aimed to analyse the process of transforming food consumed to diet quality. Finally, model 4 included as explanatory variables all the sociodemographic characteristics considered in the previous two models as well as variables affecting the highest stage in the multi-stage budgeting process: employment status and total household income. Hence, this model can be interpreted as a reduced form version of equation (1). These models also serve as robustness tests for the analyses of the effects of sociodemographic characteristics on the nutritional quality of food as well as to further explore the role of food-away-from-home consumption on the nutritional quality of food.
All models were estimated using ordinary least squares, although there are potential endogeneity issues with the time and money expenditures variables as well as with the variables related to the number of meals away from home.Footnote * Thus, estimated regression coefficients were not given a causal interpretation but rather an interpretation as measures of association between the variables. Preliminary efforts to estimate the models using instrumental variable approaches were also not successful (e.g. Davis & You(Reference Davis and You31)), likely due to the small sample size. Finally, to evaluate the heteroscedasticity of the residuals, the White heteroscedasticity test was implemented. Multicollinearity among explanatory variables was evaluated using the variance inflation factor(Reference Variyam, Blaylock and Smallwood30).
Results
Survey response behaviour
A total of 217 respondents started the survey, but only 142 (65·44 %) answered all the questions. The average survey duration for respondents that completed the survey was 226·27 min (3·77 h); with a median duration of only 30 min and an interquartile range of 18·78 min. In addition, 85 % of households completed the survey in 1 h or less. The high value for the average time of survey duration was due to the presence of few participants that completed the questionnaire in very large times (five participants had a time duration value higher than 35 h).
Table 2 compares the descriptive statistics related to the sociodemographic characteristics of respondents who completed the survey and respondents who started but did not answer all the survey questions. The difference in means was significantly different than zero (α = 0·05) for only two variables: white race and full-employment status. The proportions of white and fully employed individuals were higher among the group of respondents who completed the survey.Footnote †
Table 2 Descriptive statistics of faculty and staff that participated in the survey
** Indicates statistical significance at the 5 % level.
*** Indicates statistical significance at the 1 % level.
The average age of respondents who completed the survey was 34 years, and the average household size was 2·56 individuals. Average household income was about $69 000. Most of the respondents were white (86 %), female (74 %), employed full time (75 %) and had a college degree or above (75 %).
Regression analysis
Table 3 presents descriptive statistics related to the main variables of interest in the regression analyses: food nutritional quality and money and time expenditures. These statistics reflect an average HEI of 66·11 with a minimum of 36·54 and a maximum of 91·38. Average weekly household food expenditures were $132·36. A majority of these food expenses were for food-at-home, as the average weekly expenditures on food-at-home were $99·76. The average numbers of restaurant and fast-food meals per week were 1·40 and 2·48, respectively. The average time spent on food-related activities was 1176·05 min/week (19.60 hours). As with the case of money expenditures, the majority of the time devoted to food activities is composed of food-at-home activities, with the average time devoted to these activities at 961·45 min/week (16.02 hours). Most of the time spent on food activities at home is related to food preparation (328·88 min or 05.48 hours/week) and eating (360·23 min or 06.00 h).
Table 3 Descriptive statistics of healthy eating index, meals away from home, food expenditures and time use
In all the estimated regression models (Tables 4 and 5), White tests for heteroscedasticity failed to reject the null hypothesis that the residuals were homoscedastic (α = 0·05). Similarly, variance inflation factor values for all the variables in all models were below 3 (values above twenty are suggested as indicative of multicollinearity problems)(Reference Greene32). Thus, both heteroscedasticity and multicollinearity were not found to be problematic in this data set. The R 2 of the models ranged from 0·115 to 0·170. When evaluating the overall significance of the regressions using F-tests, only two models were significant at the 5 % level (model 2 and model 4). Model 3 was significant at the 10 % level, and model 1 was very close to being significant. Overall, the models are appropriate given the exploratory nature of these analyses and the small sample size.
Table 4 Regression estimates, determinants of the natural logarithm of the healthy eating index including separated and aggregated values of time and money expenditures related to food at home and away from home among 135 faculty and staff
Ln, natural log.
* Indicates statistical significance at the 10 % level.
** Indicates statistical significance at the 5 % level.
*** Indicates statistical significance at the 1 % level.
Table 5 Regression estimates, determinants of the natural logarithm of the healthy eating index including number of meals consumed away from home per week and sociodemographic characteristics among faculty and staff
Ln, natural log.
* Indicates statistical significance at the 10 % level.
** Indicates statistical significance at the 5 % level.
*** Indicates statistical significance at the 1 % level.
Tables 4 and 5 presents estimation results. Models 1 and 2 are the main models exploring the associations of time and money expenditures with the nutritional quality of food. When considering aggregate measures of time and money spent on food activities (model 1), only food expenditures are found to be associated with the HEI. Estimated coefficients indicate that a 1 % increase in food expenditures is estimated to be associated with a 0·06 % increase in the HEI. This would mean that, for the average household, an increase in weekly food expenditures of $13·24 (a 10 % increase) would be associated only with an increase of 0·40 points in the HEI (a 0·6 % increase). Results from the model that included separate variables for time and money expenditures including food-at-home and food-away-from-home activities (model 2) suggested that only food-at-home expenditures and time spent on food-away-from-home are associated with the HEI. A 1 % increase in food-at-home expenditures is estimated to be associated with a 0·09 % increase in the HEI. On the other hand, a 1 % increase in the time spent on food-away-from-home activities is estimated to be associated with a 0·03 % decrease in the HEI. It is important to highlight the fact that adding disaggregated measures of money and time expenditures has a large impact on the predictive power of the model, as the R 2 increases by about 50 % when going from model 1 to model 2. When considering the number of meals away from home as an explanatory variable in the HEI regression model (model 3) instead of food and money expenditures, only the number of fast-food meals is associated with a lower HEI. More specifically, each additional fast-food meal per week is found to be associated with a 1·6 % decrease in the HEI.
Results in model 4 suggest that the HEI index is associated with household income and respondents’ race and gender. A 1 % increase in income is associated with a 0·04 % increase in the HEI. Asian respondents are found to have a HEI that is about 13 and 20 % higher than the HEI of white households and household of races other than white, respectively. Female respondents are found to have a HEI that is about 6·5 % higher than the HEI of male respondents. The estimated coefficients on the sociodemographic characteristics were very similar across all model specifications.
Table 6 presents the estimated associations between a 10 % increase in time and money expenditures at home and away from home and the scores of the components of the HEI. These values were obtained using the coefficients of the auxiliary regression models for the individual components of the HEI (complete results available from the authors upon request). These results confirm those obtained using model 1. Increases in money expenditures are found to be positively associated with increases in the scores of the majority of the individual components of the HEI (eight out of thirteen effects are positive and significant). On the other hand, increases in time expenditures on food away from home are associated with decreases in a large number of the scores for the individual components (five out of thirteen effects are negative and significant, and only one is positive and significant). The largest estimated effect of money spent on food at home corresponds to seafood and plant proteins, which indicates that a 10 % increase in money expenditures is associated with an increase of 0·17 points in the score of this HEI component. The largest effect (in absolute value) of time expenditures on food away from home was for total vegetables and seafood and plant proteins (–0·05); thus, a 10 % increase in time spent on food away from home is associated with a 0·05 decrease in the scores of these HEI components.
Table 6 Association between a 10 % increase in time and money expenditures related to food at home (FAH) and away from home (FAFH) and the components of the healthy eating index among 135 faculty and staff†
† Calculations at the average values of time and money expenditures and the index scores. Average money and time expenditures on food at home are $99·76 and 961·45 min/week. Average money and time expenditures on food away from home are $31·85 and 215·33 min/week.
‡ Max denotes the maximum score value, and avg. the average value in the sample.
§ S and NS indicates if the underlying regression model is statistically significant or NS, respectively (10 % level). The underlying regression models use the natural log of the score as the dependent variable and as explanatory variables money and time expenditures both at home and away from home as well as sociodemographic characteristics: age, gender, race (White and Asian, dummies), ethnicity (Hispanic, dummy), education (College education and above, dummy) of the household head; household size; and, presence of children under 18 years in the household (dummy).
* Indicates statistical significance at the 10 % level.
** Indicates statistical significance at the 5 % level.
*** Indicates statistical significance at the 1 % level.
Discussion
Combining questions from three survey instruments that collect data on food nutrition, food expenditures and time spent in food-related activities resulted in a 30-page survey instrument. The median completion time of the survey instrument was about 30 min. A concern with the survey instrument is its length, thirty pages, which could have discouraged some survey respondents. However, about 45 % of respondents who did not finalise the survey stopped answering questions with 8 % of progress or less which suggests that, for a large proportion of survey participants, survey length was not the main factor for survey abandonment. Moreover, although the median survey length of 30 min is at the lower bound of what some authors consider as a ‘long survey,’ it may be worth reducing its length if possible(Reference Galesic and Bosnjak33). This is important as previous studies have found negative associations between survey length and both survey completion and the quality of the responses(Reference Galesic and Bosnjak33). Despite the length of the survey, the instrument has enabled us to test the hypothesis that more time spent in food-related activities and higher expenses in food increase dietary quality.
Regression analyses of the pilot survey data identified a small positive association between food expenditures and HEI. Such small association suggests that preferences and eating habits are more important than time and money when making food choices. Similar results were observed in studies carried out by Chrisinger et al.(Reference Chrisinger, Kallan and Whiteman34), Carlson et al.(Reference Carlson, Dong and Lino14), Dubois et al.(Reference Dubois, Griffith and Nevo35) and Beydoun et al.(Reference Beydoun, Fanelli-Kuczmarski and Allen16) For instance, the study conducted by Beydoun et al.(Reference Beydoun, Fanelli-Kuczmarski and Allen16) found that the estimated degree of association between food expenditures and diet quality was very small even after controlling for other factors.
Regression analyses using the data from the pilot study also revealed a positive association between food-at-home expenditures and the HEI score. This result runs contrary to a recent study carried out by Tiwari et al.(Reference Tiwari, Aggarwal and Tang36), where frequent cooking at home was associated with lower food expenditures per capita but higher HEI scores. However, the findings of this pilot study reinforce the affirmation by multiple scientists in the area of nutrition, public health, economics and time use fields that foods prepared at home are associated with higher quality of diets(Reference Monsivais, Aggarwal and Drewnowski19,Reference Fulkerson, Farbakhsh and Lytle37,Reference Seguin, Aggarwal and Vermeylen38) . On the other hand, the analyses of this study found that an increase in the time spent on food away from home is associated with a decrease in HEI scores. Similarly, the most recent ‘America’s Eating Habits: Food Away from Home Report’ found that food away-from-home consumption is associated with lower HEI scores(Reference Saksena, Okrent and Anekwe39).
Study results also identified a negative association between the number of fast-food meals per week and the HEI score. This result coincides with the findings from the ‘Americas Eating Habits: Food Away from Home Report’ that fast-food consumers tend to have the worst diets based on their HEI scores(Reference Barnes, French and Mitchell40). The negative association likely stems from the fact that fast food tends to be higher in energy (calories), empty calories, saturated fat, Na(Reference Tiwari, Aggarwal and Tang36,Reference Barnes, French and Mitchell40) ; and lower in vegetables, whole grains, fibre, vitamins and minerals(Reference Barnes, French and Mitchell40).
Regarding the relation between sociodemographic characteristics and the HEI score, pilot test data analyses results showed that Asians had higher HEI scores than other ethnicities. This finding is similar to a study conducted among South Asians Americans where it was found that their HEI score was 68 and significantly higher than the average HEI scores of American adults which was 53(Reference Saira, Robert and Bahram41). In contrast, other studies comparing HEI scores among different ethnicities have found higher HEI among Hispanics(Reference Chrisinger, Kallan and Whiteman34,Reference Hiza, Casavale and Guenther42) . Pilot test data analyses results also found that females had higher HEI scores than males, and this result is consistent with other studies(Reference Hiza, Casavale and Guenther42,Reference Reedy, Lerman and Krebs-Smith43) . For instance, a previous study that analysed the HEI in the adult national representative population found a HEI score of 59 and 54 for females and males, respectively(Reference Hiza, Casavale and Guenther42).
Analyses of the survey data identified several limitations of the current version of the survey as well as some considerations for further survey refinement. The first limitation is the aggregate nature of the HEI as a measure of food nutritional quality for both food at home and food away from home. The second limitation is a mismatch between time and household expenditures. Whereas the time use questions on food activities were asked at the individual level, most of the questions related to food expenditures were asked at the household level. The third limitation is that the survey did not include a question related to the number of meals consumed at home but prepared away from home (e.g. frozen meals, take-out or delivery foods) or the number of meals prepared at home but consumed away from home (e.g. prepared lunch or meal) and the time spent packing and eating these meals. Finally, the issue of what is a meal should be considered more carefully to appropriately incorporate, for example, time and money spent preparing, buying and eating snacks.
Conclusion and implications
The feasibility of developing and implementing a comprehensive survey instrument to analyse the relation between diet quality, time use and food costs has been demonstrated; however, pilot testing of the survey instrument revealed some limitations of the current version of the survey which would have to be considered in future research. Results of the analyses using data from the pilot test suggest that time and money expenditures are associated with diet quality, but their role is small relative to sociodemographic characteristics such as race and gender, but more work is needed with different and/or larger target populations. Moreover, a larger sample might allow the exploration of time use and food expenditures among population groups with high diet quality as well as potential interactions between time and money.
Acknowledgements
Acknowledgements: The authors would like to acknowledge the Community and International Nutrition Research (CINR) lab members: Ruhul Amin, Bong Nguyen, Hyun-Jung Lee and Theophilus Rufus. Ruhul Amin collaborated in the development of the questionnaire; and Bong Nguyen, Hyun-Jung Lee and Theophilus Rufus assisted in the dietary analysis. Financial support: This research received support from Texas Tech University’s Transdisciplinary Research Academy (TTUTRA). Conflict of interest: None. Authorship: C.E.C., C.M.K. and M.M. conceptualised the research study and the questionnaire; C.E.C. and C.M.K. conducted the literature review; A.F.M.-C. oversaw data collection and dietary quality analyses; C.E.C. and C.M.K. conducted data analyses and provided statistical expertise; C.E.C., C.M.K., A.F.M.C. and M.M. wrote the manuscript, reviewed and commented on subsequent drafts. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving study participants were approved by the Institutional Review Board (IRB) at Texas Tech University. A waiver for consent form was approved by the IRB TTU.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980020001615
