Background

Take-out food has become increasingly popular among urban residents globally in recent years [1]. This increasement in the preference for take-out food options has been further exacerbated by the COVID-19 pandemic, which has prompted a shift towards convenient online ordering and offline delivery services, collectively known as Online-to-Offline (O2O) platforms [2, 3]. With the proliferation of internet access and the widespread adoption of smartphones, elements such as availability, pricing, payment, marketing, and convenience have migrated online, reshaping how food is marketed, sold, and delivered [4]. Meanwhile, the offline component of O2O services remains grounded in the tangible aspects of the food consumption journey, encompassing food preparation, packaging, the actual delivery process, and the dining experience [5]. Due to the diverse menu options and wide price range, the customer base for O2O take-out food also exhibits a high level of diversity [1]. As the prevalence of O2O take-out food continues to grow, it has the potential to exert a profound influence on urban life and dietary habits [6]. Notably, the prevalence of overweight and obesity has also risen during the pandemic [7, 8]. This simultaneous rise in O2O take-out food consumption and obesity rates necessitates a critical examination of the impact of enhanced convenience on public health.

Previous studies have outlined the harmful effects of the consumption of fast food, especially in relation to gaining too much weight, obesity, and other related health problems [9–14]. Recent investigations have also found a discernible nexus between O2O take-out food consumption and the proximate accessibility of fast-food establishments [15]. Nevertheless, the extant literature remains relatively scarce in its comprehensive exploration of the specific impact that O2O take-out food has on weight-related concerns [16]. This empirical investigation seeks to enhance our understanding of O2O take-out food consumption behaviors by delving into both the frequency and individual preference aspects. It aims to elucidate the intricate connections between these factors and the propensity of urban residents to encounter issues related to excess weight and obesity.

Literature review

Overweight and obesity are recognized as common chronic diseases worldwide, affecting both adults and children [17]. Obesity, characterized by excessive energy intake surpassing expenditure, is a multifactorial chronic condition [18, 19]. The health implications of overweight and obesity are profound. The WHO provides statistics indicating that mortality, morbidity, and disability resulting from major chronic diseases are projected to account for approximately 73% of all deaths and 60% of the global disease burden [20]. Obesity plays a significant role in the development of numerous chronic diseases, including cardiovascular disease (CVD) and type 2 diabetes, which have detrimental effects on individual and public health [21–24].

Extensive studies have delved into the links between fast-food consumption, the abundance and accessibility of fast-food outlets, and their impact on the rates of obesity and overweight [9, 10, 13, 25–37]. Meanwhile, the density and accessibility of recognized health-food providers like supermarkets, fresh fruit vendors, and vegetable markets are found to be beneficial for weight management [38–40]. While fast-food restaurants are often linked to poor nutrition and higher Body Mass Index (BMI), other dining options are perceived more favorably in terms of health. However, a comprehensive literature review reveals that eating outside the home, in general, is associated with higher energy and fat consumption, alongside a decrease in micronutrient intake [41]. This finding underscores the need for a nuanced understanding of the differences between fast-food restaurants and other types of out-of-home dining venues, as suggested by Bezerra and colleagues [42].

Moreover, the fast-food landscape is diverse, extending beyond Western-style offerings to include traditional snacks and meals in Asia, Latin America, and Africa [43]. For instance, a large segment of Singapore’s population frequently eats at hawker centers, which are vibrant, open-air complexes where a variety of local foods are available [44]. Interestingly, a comparison of Western fast food with local Asian offerings revealed little difference in energy and fat content [45], suggesting that the health effects of non-Western fast-food styles warrant further study.

The advent of digital technology has transformed the food environment. Granheim and colleagues highlighted how the digitalization of food availability, pricing, payment, and marketing has particularly affected food purchasing behaviors, ushering in a new era of online shopping and delivery services [46]. Yet, these services often promote unhealthy food options, as noted by Duthie and colleagues [47], indicating potential public health implications. While recent research has begun to profile the socio-demographic characteristics of online food order consumers [48, 49], the direct effects of this digital food environment on body weight and nutrition remain underexplored.

This empirical study aims to bridge these gaps by comprehensively examining O2O take-out food consumption behaviors, including frequency, preferences, and the variety of food types consumed. It differentiates between Western-style fast-food, non-western fast-food, and non-fast-food options to provide a clearer picture of the health implications associated with these choices.

Study design and participants

This study specifically targeted adult urban residents residing in the metropolitan areas of Shanghai and Suzhou, located within the Yangtze River Delta region of China. The O2O industry in China has undergone rapid expansion, encompassing 421 million customers in 2019 [50]. The significance of Shanghai and Suzhou in the Yangtze River Delta region is underscored by their substantial contribution, accounting for one-third of all O2O orders in China [51]. Consequently, selecting Shanghai and Suzhou as the study’s focal points ensured access to a robust dataset, facilitating a comprehensive analysis of consumer behaviors associated with online-ordered take-out food.

The survey spanned the period from April to July 2022, involving 746 adult participants recruited through a cross-sectional online survey. The survey tool is a structured questionnaire, provided in the Mandarin Chinese version. Due to travel restrictions and the risk of COVID-19 quarantine in China in 2022, we hired a professional survey company in China, Wenjuanxing™, to distribute the questionnaire online. The company specializes in distributing questionnaires to specific demographic groups (adults aged 18 to 60) according to our requirements and has guaranteed the reliability of the samples and the quality of the responses. Therefore, this study did not place significant emphasis on the response rate. It is worth noting that in China, university students are particularly enthusiastic about ordering O2O take-out food [16]. Therefore, given that the survey company only screened for adult participants, it was expected that a significant portion of the sample would consist of university students.

Importantly, the study was ethically approved by the University Ethics Committee of Xi’an Jiaotong–Liverpool University under Proposal Number 21-02-34. The approval process involved a thorough review of the research proposal, ensuring compliance with ethical guidelines and standards. Any deviations from the approved protocol or supporting documents require prior approval from the University Ethics Committee (UEC) before implementation. Participants in the study were informed of their rights and provided with an informed consent form at the beginning of the online survey. It was made clear that participation was voluntary, with the option to withdraw at any time without explanation. Participants were also informed of their right to access and request the destruction of their provided information. Completion of the survey indicated the participant’s informed consent, and data analysis was conducted anonymously to ensure confidentiality and privacy.

The Socio-Ecological Model on Obesity and Type-2 Diabetes, which is a multi-level and multi-sectoral framework initiated for the obesity and diabetes research [52, 53], served as the conceptual framework for the study (See Fig 1). To capture the multifaceted dynamics influencing overweight and obesity, a survey questionnaire was meticulously developed following the conceptual framework. The questionnaire comprehensively addressed various aspects, including urban customers’ energy intake and expenditure, as well as individual demographic and socioeconomic factors.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Conceptual framework of the research.

https://doi.org/10.1371/journal.pone.0315189.g001

Independent variables

The independent variables of this study are food intake behaviors including O2O and non-O2O take-out food consumption behaviors. Assessing these dietary behaviors necessitates a thorough examination of the frequency of food intake, a pivotal determinant [54–57]. To address this, survey questions pertaining to frequency were formulated based on the Food Frequency Questionnaire (FFQ)—a widely utilized tool for gauging the habitual frequency of consuming specific types of food over a specified timeframe [55, 58, 59]. Specifically, to assess the frequency of O2O take-out food consumption, respondents were queried, ‘On average, how frequently do you consume online ordered take-out food per week? (‘0~2 times’ = 1; ‘3~4 times’ = 2; ‘5~7 times’ = 3; ‘8~10 times’ = 4; ‘Over 10 times’ = 5)’ The survey also captured the frequency of specific conventional dietary behaviors that pertain to food consumption outside the O2O take-out sphere. This specifically encompassed the frequency of fruit and vegetable intake, which is associated with a lower risk of being overweight and obese [39, 60, 61]. Furthermore, the survey inquired about the frequency of fast food consumption, acknowledging its correlation with an elevated risk of overweight and obesity [9–14]. Fast food refers to ready-to-eat food that is quickly prepared and served using standardized methods. In this study, the primary difference between fast food consumption in regular diets and O2O take-out food consumption lies in whether the food is obtained through online ordering. The survey questions are ‘On average, how frequently do you consume fruits and vegetables in regular meals per week?’ and ‘On average, how frequently do you consume fast-food ordered in regular meals per week?’ respectively. The options are ‘0~2 times’ = 1; ‘3~4 times’ = 2; ‘5~7 times’ = 3; ‘8~10 times’ = 4; ‘Over 10 times’ = 5.

Consumers’ preferences toward specific food categories are likely to significantly shape their dietary behaviors [62], especially when the prevalence of online food delivery services simplifies their access to diverse food options. In the tools of assessing diet, an individual’s preference and inclination towards certain types of food is often used by Dietary screeners, and its effectiveness has been verified by previous research [63, 64]. Therefore, the survey also assessed O2O take-out food consumption behaviors by investigating preferences for ordering specific food categories. Participants were asked to express their agreement with statements such as ’I prefer to order Western-style fast food.’ This included examining preferences for ordering fast food through O2O take-out services, with a meticulous distinction between Western-style fast food and Chinese local fast food. Participants rated their agreement with statements about their preference for ordering from each category, using a five-point Likert scale from ’strongly disagree’ to ’strongly agree.’ Additionally, the survey documented preferences for ordering take-outs of fruits and vegetables. Other types of take-out foods addressed in preference-related inquiries encompassed barbecue, snacks, coffee, milk tea, and other beverages—items widely popular on O2O take-out food platforms in China [65].

Dependent variable

This study uses a binary variable—whether or not one is overweight or obese—as the dependent variable, with Body Mass Index (BMI) as the criterion for determining overweight and obesity status. Participants were asked to self-report their height and weight. BMI was determined by dividing weight (in kilograms) by height squared (in meters) [66]. The overweight and obesity in this study were based on the guidelines set forth by China’s National Health and Medical Commission in 2013. Specifically, the normal range for adult BMI was defined as 18.5–23.9, while BMIs falling between 24–27.9 were considered overweight, and those above 28 were classified as obese [66].

Control variables

The control variables of this study include energy expenditure and individual factors. The energy expenditure was measured in the form of physical activity, which is evaluated using the International Physical Activity Questionnaire (IPAQ) short form [67], widely applied in physical activity-related research [67, 68]. Since the survey was conducted in China, the Chinese version of the IPAQ was used. The reliability of this version has also been validated by previous studies [69]. A respondent was defined as physically ’active’, based on the ’Minimally Active’ standard in the IPAQ [67]. The ‘Minimally active’ is defined as a) three or more days of vigorous activity of at least 20 minutes per day, b) five or more days of moderate-intensity activity or walking of at least 30 minutes per day, or c) five or more days of any combination of walking, moderate-intensity, or vigorous-intensity activities achieving a minimum of at least 600 MET-min/week (MET is the weighting unit specified in IPAQ) [67]. Respondents who did not meet the standard were classified as ’inactive.’

The individual factors of the respondents included gender (male/female), age, marital status (married/single, divorced, widowed), educational level (college degree and above/below college degree), occupations (students/non-students), and income level. Respondents from Suzhou and Shanghai were also indicated and compared in the study.

Statistical analysis

The O2O food options were not previously included in the FFQ and Dietary screeners. Therefore, the consistency and reliability of those new options need to be assessed. Cronbach’s alpha was employed to evaluate the internal reliability of the relevant survey questions [70–73]. The standardized Cronbach α coefficient of the items need to surpass the threshold of 0.6 and every Corrected Item-Total Correlation (CITCs) of the items need to surpass the threshold of 0.3 [70–74]. Additionally, Exploratory Factor Analysis (EFA) was used to determine the construct validity of the measurements [71, 72, 75, 76]. The Kaiser-Meyer-Olkin (KMO) value in the EFA need to surpass the threshold of 0.6 and all items demonstrated factor loading coefficients need to exceed 0.4 [71, 72, 75, 76].

Binary logistic regression models were developed to identify potentially significant variables associated with being overweight or obese. Given that those O2O and non-O2O take-out food consumption behaviors can either independently impact weight-related health outcomes or concurrently influence these outcomes when considered together, three binary logistic regression models were developed to comprehensively elucidate the intricate interplay between them. In addition to individual characteristics and energy expenditure variables, Model 1 includes only the non-O2O take-out food consumption variables as energy intake factors; Model 2 includes only the O2O take-out food consumption variables; Model 3 includes both sets of variables. To mitigate the risk of multicollinearity and ensure independent variable reliability, multicollinearity diagnostics were applied, confirming that the variables in the models do not exhibit high intercorrelations. The Events per Variable (EPV) for the three models will be calculated separately. If the EPV falls within the range of 5 to 15, it indicates that the study size is appropriate for analysis [77, 78]. All analyses were performed using IBM SPSS Statistics 27.

Descriptive statistics of participants

Table 1 presents the descriptive summary of the respondents, encompassing demographic characteristics, overweight and obesity status, physical activity status, and geographic locations.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive statistics of the sample.

https://doi.org/10.1371/journal.pone.0315189.t001

The demographic breakdown of the survey participants included 42% male and 58% female respondents. Participants’ ages ranged from 18 to 58 years, with the mean age being 26 years and a standard deviation of 6.93. Notably, students comprised a considerable segment of the respondents at 52%, with a significant concentration of college students within the 21–25 years age bracket. Moreover, reported monthly income levels were modest, as over 68% of respondents earn under 6,000 RMB (approximately USD$820) per month. In terms of marital status, a majority (over 71%) were single, divorced, or widowed. The incidence of obesity or overweight status among the participants was 17.31%. Importantly, a notable 66.62% of respondents described themselves as being physically ’active’.

Cronbach α reliability analysis and exploratory factor analysis (EFA) results

Following the application of the Cronbach α reliability test and the EFA for establishing validity, a set of five items demonstrated satisfactory levels of reliability and validity. Please refer to Tables 2 and 3 for a detailed analysis process.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Reliability statistics of O2O take-out food consumption behaviors.

https://doi.org/10.1371/journal.pone.0315189.t002

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Validity analysis (EFA) of O2O take-out food consumption behaviors.

https://doi.org/10.1371/journal.pone.0315189.t003

Those five items meeting reliability and validity test standards are as follows: (1) Frequency of take-out food consumption, (2) Preference for ordering Western fast food, (3) Preference for ordering Chinese fast food, (4) Preference for ordering vegetables, and (5) Preference for ordering fruits. Consequently, these items were employed in the binary logistic regression models for assessing the behavior of O2O take-out food consumption in this study. The descriptive statistics of these five items are shown in Table 4.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Descriptive statistics of the items for assessing the behavior of O2O take-out food consumption.

https://doi.org/10.1371/journal.pone.0315189.t004

Regression model results

The results of the binary logistic regression model are shown in Table 5, and the dependent variable of this model is whether an individual is overweight and obese or not (Yes = 1). The number of overweight and obesity samples is 124. The EPV of the three models are 12.4, 9.5 and 8.3 respectively, indicating that the study size is appropriate for analysis.

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Binary logistic regression model results.

https://doi.org/10.1371/journal.pone.0315189.t005

In Model 1, including conventional dietary behaviors only, energy expenditure, and individual factors, the frequency of fast-food consumption per week and the frequency of fruit and vegetable consumption per week were significantly positively and negatively correlated with overweight and obesity, respectively, with odds ratios of 1.016 (95% CI: 1.004~1.028, p-value = 0.011) and 0.972 (95% CI: 0.957~0.987, p-value = 0.032).

In Model 2 with O2O take-out food behavior variables only, energy expenditure, and individual factors, four out of the five items related to O2O take-out food consumption behaviors exhibited statistical significance. Specifically, there was a significant positive correlation between the frequency of take-out food consumption and being overweight or obese, with an odds ratio of 1.082 (95% CI: 1.064~1.101, p-value<0.001). Similarly, a significant positive correlation was observed between the preference for ordering Western fast food and being overweight or obese, with an odds ratio of 1.062 (95% CI: 1.041~1.085, p-value<0.001). Conversely, the preference for ordering Chinese fast food and the preference for ordering vegetables displayed a significant negative correlation with overweight or obesity, with adds ratios of 0.971 (95% CI: 0.952~0.990, p-value = 0.003) and 0.976 (95% CI: 0.960~0.993, p-value = 0.005), respectively.

In Model 3, wherein both O2O take-out food consumption behaviors and conventional dietary behaviors were concurrently considered, subtle alterations in the outcomes are observed. Notably, the frequency of O2O take-out food consumption and the preference for O2O Western fast food maintained a statistically significant positive correlation with being overweight or obese, exhibiting odds ratios of 1.087 (95% CI: 1.068~1.106, p-value<0.001) and 1.071 (95% CI: 1.046~1.095, p-value<0.001), respectively. Additionally, the two conventional dietary behavior variables, namely the weekly frequency of fast-food consumption and the frequency of fruit and vegetable consumption, exhibited continued significant positive and negative correlations with being overweight and obese, respectively, with odds ratios of 1.036 (95% CI: 1.015~1.058, p-value<0.001) and 0.970 (95% CI: 0.948~0.993, p-value = 0.011). However, the remaining three variables of O2O take-out consumption behaviors did not demonstrate statistically significant relationships with being overweight or obese.

Moreover, the findings of all three models exhibited similarities concerning the individual factors. Specifically, the results indicated that men were more susceptible to being overweight or obese compared to women. At the same time, individuals who were physically active were less likely to be overweight and obese. Interestingly, residents in Suzhou were significantly more likely to be overweight and obese than their counterparts in Shanghai.