Edited by: Igor Pravst, Institute of Nutrition, Slovenia
Reviewed by: Andrew Hill, University of Leeds, United Kingdom; John David Eastwood, York University, Canada; Naomi Kakoschke, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia; Laura Maria König, University of Bayreuth, Germany
This article was submitted to Eating Behavior, a section of the journal Frontiers in Nutrition
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Self-reported weight gain during the COVID-19 shelter-at-home has raised concerns for weight increases as the pandemic continues. We aimed to investigate the relationship of psychological and health markers with energy balance-related behaviors during the pandemic-related extended home confinement. Ratings for stress, boredom, cravings, sleep, self-control, and beliefs about weight control were collected from 1,609 adults using a questionnaire between April 24th–May 4th, 2020, while COVID-19 associated shelter-in-place guidelines were instituted across the US. We calculated four energy balance behavior scores (physical activity risk index, unhealthy eating risk index, healthy eating risk index, sedentary behavior index), and conducted a latent profile analysis of the risk factors. We examined psychological and health correlates of these risk patterns. Boredom, cravings for sweet/savory foods, and high sleepiness ratings related to high risk of increasing unhealthy eating and sedentary behavior and decreasing physical activity and healthy eating. Having greater self-control, control over cravings, or positive mood was related to lowering all aspects of energy intake and energy expenditure risks. Although individuals in risk pattern classes showed similarity in physical activity and healthy/unhealthy eating habits, they exhibited different patterns of positive mood, craving control, food cravings, boredom, and self-control. Psychological and health variables may have a significant role to play in risk behaviors associated with weight gain during the COVID-19 related home confinement. Emerging behavioral patterns may be meaningful in developing targeted weight management interventions during the current pandemic.
香京julia种子在线播放
In March 2020, the novel severe acute respiratory distress coronavirus 2 (SARS-CoV-2) infection emerged as a global COVID-19 pandemic. As a consequence, widespread shelter-at-home was implemented in the US to prevent the spread of this infection, primarily between March 15th and May 7th, 2020. This public health action markedly disrupted everyday activities and increased unstructured time for people, making weight management a concern (
With shelter-at-home restrictions and inability to practice normal life, numerous possible challenges can affect energy intake and energy expenditure, the two components of energy balance. With regards to energy intake behaviors, COVID-19 disruptions may have introduced multiple influences on people's dietary behaviors which may have produced increased unhealthy eating and/or healthy eating. In particular, during the lockdown people had easy access to snacks and craving inducing energy dense convenience foods (
COVID-19 lockdown and related social distancing drastically impacted the life of people in the US. People lost their jobs and shifted to work from home schedule while actively taking care of family and dealing with the fear of infection. Travel, social life, and leisure activities were also severely restricted, unlike prior to pandemic. These major life adjustments were accompanied by severe physiological and psychological costs, as reported in multiple studies. In particular, the recent lockdown caused dramatic increases in these state-like psychological variables, such stress, anxiety, low-sleep quality etc. (
With regards to the trait-like psychological variables, some of these factors are known to be known to be protective toward these extreme behavioral alterations. For example, lack of self-control (
Overall, this study aimed to investigate the relationship between relevant demographic characteristics, state- and trait-like psychological markers and energy balance-related behaviors, during the pandemic-related shelter-in-place. Specifically, we examined associations between stress, boredom, cravings, sleep, self-control, BMI, and beliefs about weight control. In addition, we evaluated differences in risk behaviors between demographic groups. Using a Latent Profile Analysis, we also aimed to identify and characterize patterns of health behavior change during the pandemic. We hypothesized that sleep time and quality, craving control, self-control, and beliefs that one can control their weight would be negatively associated with energy balance-related behaviors during the pandemic. In contrast, we expected that boredom, stress, and food cravings would be positively associated with energy balance-related behaviors during the pandemic.
The study design has been described in full detail elsewhere (
We recruited 1,779 men (43.38%) and women (56.62%) between the age of 18 and 75 years. Inclusion criteria included: (1) access to the internet, and (2) living in the U.S. The questionnaire was administered through Amazon Mechanical Turk (Mturk, © 2005–2018, Amazon Mechanical Turk, Inc., Seattle, WA) (
A small compensation ($1.66) was given to eligible participants completing the survey through Mturk. This amount was estimated based on the minimal amount required to complete a similar survey and in line with the median hourly wage earned by an MTurk responder. Participants recruited via social media, email, and word of mouth volunteered to complete the survey and did not receive any monetary compensation. Of note, while the participation using this recruitment method was completely voluntary, it is possible that the compensation offered to Mturk workers for completion of survey may have been a motivational factor for them to participate in our study.
Participant recruitment and data collection occurred during the 11 days from April 24th, 2020 to May 4th, 2020, while shelter-in-place guidelines were instituted across the US. Of the 1,779 participants who initially responded to the call to complete the questionnaire, 1,609 participants were included in the data analysis. Of the 170 people excluded from the analysis (MTurk
The Qualtrics questionnaire included the following 7 item categories: demographics, weight behaviors, sleep, and other health behaviors, eating behaviors, physical activity behaviors, psychological factors, and food purchasing behaviors. Questions within these categories were aimed at understanding change in practices and beliefs during the COVID-19 shelter-at-home. Similar to other studies, we asked whether these practices “increased,” “decreased,” or “stayed the same” during the COVID-19 shelter-at-home (
Eating behaviors were determined by asking participants if their consumption of the following items increased, decreased, or remained the same during COVID-19 shelter-in-place: fruits (during meals), vegetables (during meals), caffeine, non-diet drinks (includes, Coke, Pepsi, flavored juice drinks, sports drinks, sweetened teas, coffee drinks, energy drinks, electrolyte replacement drinks), and diet soda and other diet drinks. To determine change in consumption of processed and ultra-processed foods, we presented a list of foods as described by the NOVA classification system (
We also collected information on the change in consumption of snack items (cake, cookies, ice-cream, other desserts; chips, popcorn, pretzels, and crackers; gummy snacks, fruit candy, sour gummy, or other fruity candies; fruits; vegetables; chocolate; yogurt/cheese). Change in consumption of restaurant/take-out/fast food/delivery food and alcohol intake was also recorded. We did not collect data on quantities consumed for the specific food items using the traditional methods of self-reported dietary data collection because they are prone to reporting errors and appears to underestimate energy and nutrient intake (
Change in sitting, walking, moderate physical activity, and vigorous physical activity during the COVID-19 outbreak in their area were assessed using “I am doing more,” “I am doing the same,” and “I am doing less” options. Change in sedentary behaviors was determined by asking questions on change in time spent on watching television, social media, or other leisurely activities such as video games, computer, email etc. since COVID-19 outbreak. Given the lack of validated questionnaires to capture the perceptual change in behaviors, we developed and used face-valid items for both the physical activity and eating behavior measures. We intentionally wrote these items to target if the energy balance behaviors “increased,” “decreased,” or “remained the same” to capture self-reported change.
The validated CoEQ comprised 21 items and included questions on general appetite and overall mood (independent of craving), frequency and intensity of general food craving, craving for specific foods (e.g., dairy, starchy, sweet, or non-sweet foods), and individuals' perceived control over resisting craved food items. Participants responded about their experience over the previous seven days. These items were assessed using a 10-point visual analog scale (VAS). Subscales created form the questionnaire were used to calculate scores for: craving control, craving for sweet foods, craving for savory foods, and positive mood (
To assess sleep duration, participants were asked to report the average number of hours spent sleeping per day since the COVID lockdown in their area. To quantify sleep quality, we used the Stanford Sleepiness Scale (
The Multidimensional state boredom scale (
All participants reported report their current stress levels using a visual analog scale. The scale ranged from 1 through 10, with 1 = no stress at all and 10=highest stress possible.
The Capacity for Self-Control Scale (
The Implicit Theory of Weight Measure (
SAS version 9.4 (Cary, NC) and MPlus version 8.0 with Mixture software (
For change in each behavior related to energy intake or energy expenditure, we assigned scores to responses “I am doing more,” “I am doing the same,” and “I am doing less” such that, 1 = healthy change, 2 = no change, and 3 = unhealthy change. The α's for high-sedentary behavior score, low-physical activity behavior score, high-unhealthy eating behavior score, and low-healthy eating behavior score were 0.54, 0.63, 0.74, and 0.86, respectively. Note that scores on low-physical activity behavior and low-healthy eating behavior were calculated such that higher scores reflected less physical activity and less fruit and vegetable consumption.
We first conducted ANOVAs to test differences of health-risk behaviors between demographic groups. We then calculated intercorrelations between energy balance behavior scores and health and psychological risk and protective factors. We then characterized item level changes (increased, decreased, or stayed the same) for each health/psychological risk factor (see
ANOVAs were conducted to evaluate differences of risk behaviors between demographic groups. Participants' scores for four energy balance behavior scores are presented for each demographic variable in
Scores for four energy balance behavior categories by demographic profile of participants.
Males |
2.42 (0.43) | 2.13 (0.53) | 1.97 (0.34) | 1.97 (0.48) | ||||
Females |
2.54 (0.43) | 2.11 (0.58) | 2.06 (0.38) | 1.94 (0.55) | ||||
White |
2.47 (0.44) | 2.09 (0.55) | 2.03 (0.35) | 1.97 (0.50) | ||||
Black |
2.49 (0.42) | 2.15 (0.54) | 1.96 (0.42) | 1.85 (0.52) | ||||
Other |
2.52 (0.39) | 2.08 (0.58) | 2.07 (0.42) | 1.95 (0.54) | ||||
Asian |
2.59 (0.39) | 2.36 (0.59) | 1.99 (0.43) | 1.88 (0.59) | ||||
Hispanic |
2.52 (0.42) | 2.12 (0.58) | 2.00 (0.41) | 1.91 (0.57) | ||||
Not Hispanic |
2.49 (0.43) | 2.12 (0.56) | 2.03 (0.36) | 1.96 (0.51) | ||||
Married |
2.44 (0.43) | 2.05 (0.55) | 2.02 (0.37) | 1.95 (0.51) | ||||
Not married |
2.53 (0.43) | 2.19 (0.56) | 2.03 (0.36) | 1.95 (0.53) | ||||
18–39 |
2.53 (0.43) | 2.13 (0.57) | 2.03 (0.38) | 1.95 (0.54) | ||||
40–64 |
2.44 (0.43) | 2.11 (0.55) | 2.02 (0.34) | 1.95 (0.48) | ||||
>64 |
2.34 (0.40) | 2.02 (0.38) | 1.96 (0.32) | 2.01 (0.43) | ||||
<30,000 |
2.45 (0.43) | 2.20 (0.53) | 2.01 (0.36) | 1.72 (0.56) | ||||
30,000–59,999 ( |
2.46 (0.43) | 2.14 (0.55) | 2.01 (0.36) | 1.97 (0.52) | ||||
60,000–89,999 ( |
2.53 (0.43) | 2.08 (0.57) | 2.00 (0.39) | 1.97 (0.51) | ||||
>90,000 ( |
2.50 (0.43) | 2.07 (0.56) | 2.07 (0.35) | 1.93 (0.53) |
Scale intercorrelations were calculated to highlight associations between psychological and health risk and health protective factors. Correlations are shown in
Correlations between psychological and health risk/protective factors.
Boredom (1) | 3.74 (1.53) | — | ||||||||||
Self-control (2) | 3.39 (0.78) | −0.62 |
— | |||||||||
Beliefs about weight control (3) | 2.61 (1.19) | 0.22 |
−0.21 |
— | ||||||||
Positive mood (4) | 5.51 (1.97) | −0.57 |
0.53 |
−0.14 |
— | |||||||
Control of cravings (5) | 5.52 (2.46) | −0.39 |
0.45 |
−0.06 |
0.25 |
— | ||||||
Cravings for sweet foods (6) | 4.19 (2.33) | 0.33 |
−0.32 |
0.12 |
−0.21 |
−0.74 |
— | |||||
Cravings for savory foods (7) | 4.49 (2.03) | 0.33 |
−0.28 |
0.05 | −0.16 |
−0.65 |
0.60 |
— | ||||
Sleepiness rating (8) | 2.90 (1.45) | 0.42 |
−0.45 |
0.05 |
−0.53 |
−0.27 |
0.17 |
0.17 |
— | |||
Hours of sleep (9) | 7.31 (1.45) | 0.01 | 0.00 | 0.09 |
0.13 |
−0.05 | 0.06 |
0.03 | −0.08 |
— | ||
Body mass index (10) | 25.99 (5.95) | −0.02 | −0.06 |
−0.03 | −0.01 | −0.17 |
0.07 |
0.12 |
0.05 |
−0.11 |
— | |
Stress (11) | 4.59 (2.50) | 0.46 |
−0.36 |
0.19 |
−0.63 |
−0.28 |
0.27 |
0.23 |
0.40 |
−0.09 |
0.03 | — |
Next, we conducted a LPA to characterize classes of participants' patterns of risky health behaviors during the COVID-19 pandemic using composite variables for physical activity, sedentary behavior, healthy food consumption, and unhealthy food consumption. A model with four classes demonstrated the best fit with the data, Log Likelihood (LL) = −3744.75, degrees of freedom (df) = 23, Aikake Information Criterion (AIC) = 7535.49, Bayes Information Criterion (BIC) = 7659.10, Sample-size adjusted BIC (ABIC) = 7586.03, Entropy = 0.826. The classes' patterns of endorsed risky health behaviors are shown in
Average scores of engagement in obesogenic risk behaviors by latent classes. Class 3 is considered the General Low Risk Group; Class 4 is considered the General High-Risk Group. Class 1 is the Medium General Risk, Medium Sedentary Risk Group, and Class 2 is the Medium General Risk, High Sedentary Risk Group.
Examining the characteristics of participants in all risk profiles (
Psychosocial risk factors across class determined by latent profile analysis.
Low-physical activity score | 2.04a | 2.10a | 1.89b | 2.64c | 63.85 |
0.11 [0.08, 0.14] |
High-sedentary behavior score | 2.16a | 2.84b | 1.58c | 2.89d | 2918.05 |
0.85 [0.84, 0.86] |
High-unhealthy eating score | 1.96a | 2.02b | 1.94a | 2.33c | 51.69 |
0.09 [0.06, 0.12] |
Low-healthy eating score | 1.95a | 1.74b | 1.88a | 2.80c | 284.24 |
0.35 [0.32, 0.38] |
Boredom | 3.38a | 3.99b | 3.33a | 4.42c | 32.52 |
0.06 [0.04, 0.08] |
Self-control | 3.53a | 3.35b | 3.43a | 2.94c | 27.32 |
0.05 [0.03, 0.07] |
Beliefs about weight control | 2.64a | 2.57a | 3.00b | 2.46a | 4.17 |
0.01 [0.00, 0.02] |
Positive mood | 5.81a | 5.44b | 5.66ab | 4.44c | 21.81 |
0.04 [0.02, 0.06] |
Control of cravings | 5.93a | 5.43b | 5.76ab | 4.06c | 26.21 |
0.05 [0.03, 0.07] |
Cravings for sweet foods | 3.85a | 4.35b | 4.01ab | 5.07c | 13.68 |
0.03 [0.01, 0.04] |
Cravings for savory foods | 4.12a | 4.67b | 4.34ab | 5.48c | 21.68 |
0.04 [0.02, 0.06] |
Sleepiness rating | 2.65a | 2.96b | 2.78ab | 3.68c | 24.70 |
0.04 [0.03, 0.06] |
Hours of sleep | 7.23a | 7.41b | 7.24ab | 7.26ab | 2.04 | 0.00 [0.00, 0.01] |
Body mass index | 26.20a | 25.67a | 25.59a | 26.45a | 1.38 | 0.00 [0.00, 0.01] |
Stress | 4.29a | 4.70b | 4.31ab | 5.49c | 11.61 |
0.02 [0.01, 0.04] |
Age | 40.14a | 36.31b | 39.09abc | 35.75bc | η2 = 0.02 [0.01, 0.04] | |
Male | 51.23% | 37.77% | 43.75% | 33.73% | 32.12 |
0.14 |
Married | 51.66% | 44.65% | 57.70% | 36.89% | 15.34 |
0.10 |
Hispanic | 10.30% | 12.40% | 8.97% | 11.31% | 1.90 | 0.03 |
White | 79.45% | 74.92% | 88.61% | 72.78% | 18.10 |
0.06 |
Black | 6.51% | 7.80% | 6.33% | 5.92% | ||
Asian | 8.54% | 11.93% | 5.06% | 15.38% | ||
Other | 5.50% | 5.35% | 0.00% | 5.92% | ||
< $30,000 | 22.36% | 17.71% | 15.79% | 16.36% | 14.84 | 0.06 |
$30,000–59,999 | 29.00% | 25.93% | 34.21% | 29.09% | ||
$60,000–89,999 | 20.85% | 23.35% | 14.47% | 25.45% | ||
>$90,000 | 27.79% | 33.01% | 35.53% | 29.09% |
The primary purpose of this paper was to investigate the relationship between relevant psychological markers and energy balance-related behavior scores, during the COVID-19 related shelter-in-place. Generally, we report that increased boredom, higher self-reported cravings for sweet/savory foods, and high sleepiness ratings during the lockdown were related to increased unhealthy eating and sedentary behavior and decreasing physical activity and healthy eating during the lockdown. Whereas, having psychological traits such as greater general self-control, control over cravings, or positive mood was related to lower self-reported energy intake and energy expenditure during the lockdown. Individuals with the highest risk pattern reported having higher sleepiness, more boredom, less positive mood, and more cravings for sweet and savory foods.
Our hypothesis that self-reported change in boredom during the lockdown, a state like-psychological variable, may be related to dietary intake risk was based on prior research suggesting that high boredom increases the desire for and intake of unhealthy foods and snacks (
The relationship observed between self-reported sleepiness ratings, sleep duration, and diet quality in the current study confirms results from prior studies. We, and others, have previously shown that higher sleepiness (
Similar to the findings by Buckland et al. where lower craving control predicted high energy dense sweet and savory food intake during COVID-19 lockdown, we also showed that greater control on food cravings, representing a state-like psychological characteristic, was related to unhealthy eating score (
In everyday life, general self-control, a trait psychological characteristic, is associated with positive weight management behaviors, including healthier eating, successful weight loss, and increased physical activity, as well as with better psychological well-being (
One predictor of weight management behaviors is the belief that a person's body weight is malleable (
Given the heterogeneity in energy balance-related behaviors, an assessment of risk profile groups gave us a better insight into the unique characteristics of individuals who may be more prone to weight gain during the pandemic. Not surprisingly, individuals with the highest risk not only engaged in all energy balance-related behaviors but also reported to have psychological and health markers known to promote obesity. Although similar in risk level, we observed subtle but unique differences between the two moderate risk groups. The most striking difference between the two groups was sedentary behavior. As theorized by previous work, a complex interplay between personal circumstances, environmental variables, and social factors determines sedentary behavior (
The results of this study must be interpreted in light of several limitations. This study was cross-sectional and non-experimental; thus, causality and temporality cannot be inferred. As such, we cannot conclude if reported alterations in behaviors truly lead to weight gain. Additionally, while there is evidence of behavior changes with body mass index status, due to the self-reported nature of height and weight data collected, we did not test the difference in health behaviors between BMI groups. We also asked participants to report their perception of behavior change (increased, decreased, remained the same), rather than asking them to report behaviors before and during the lockdown period and calculating the change score for each variable. While we did this to minimize self-reporting bias and/or recall bias, the data is still self-reported, and our results may be subject to biases. Moreover, a recent report demonstrated that perceptual increase in physical activity is driven by the amount of vigorous physical activity performed, suggesting that an increase in intensive physical activity is important for perceiving a change in one's physical activity (
Additionally, while we did not disclose the specific purpose of the study to the participants, our results could also be driven by participant's expectation and not their actual behavior. With regards to the questionnaires, while validated instruments were used as possible, some necessary questions were developed by the investigators to capture the current unique environment. Moreover, we did not use a validated tool for dietary intake, such as food frequency questionnaires. Thus, care should be taken to integrate these findings with the broader literature. For our psychological and health behavior constructs, some variables were contextual or state like, while some were trait like. However, this should not have impacted our findings because whether it is a state like characteristic or trait like characteristic, we were interested in how it influenced energy-balance-related behaviors and how they differed between the risk classes. Moreover, despite the diversity and size of our sample, a convenience sampling approach was used, which may limit generalizability. Furthermore, the degree of shelter-in-place guidelines and the number of COVID-19 cases in participants' area of residence likely differed, creating differences in flexibility with stepping outside the house. The time frame of data collection may have influenced our results as well. As such, at the time of data collection, although most states had implemented shelter-in-place guidelines, a few states were considering lifting the restrictions. This one snapshot of time also assumes that thoughts and behaviors were static throughout the entire shelter-in-place time, which is likely an oversimplification.
Altogether, this study describes state- and trait-like psychological factors that relate to energy balance-related behavior categories during the COVID-19 shelter-at-home restrictions in the U.S. Our analysis provides important insights into the complex interplay of factors related to risk of increasing unhealthy eating and sedentary activities and decreasing healthy eating and physical activity. These results also contribute to improving our understanding of the patterns of risk groups and their unique characteristics, specifically highlighting that the lockdown did not adversely impact energy balance behaviors in all individuals. Our risk classes identified risk groups that represented 15–20% of our sample population. Health entities such as World Health Organization have several nutritional and lifestyle recommendations to follow during lockdown for the general public Thus, based on our findings, such public health efforts may be better spent targeting at-risk population subgroups in need of weight management interventions during the current pandemic rather than targeting people who are already managing the transition well. Our results also suggest that self-reported changes in state-like psychological variables impacted energy balance behaviors in a similar manner during COVID-19 lockdown, as they did during pre-COVID time. Thus, an effort to reduce stress and boredom, improve sleep hygiene, and strategies to control food cravings (all state-like psychological variables) using public health platforms may be beneficial in addressing a potential negative impact of lockdown on energy balance behaviors. Additional research is also needed on collecting longitudinal data to understand whether the high-risk behaviors revert back to normal as the pandemic crisis is passed.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
This study protocol (HS-2020-0105, HS-2020-0100) was reviewed and approved by the Institutional Review Board at San Diego State University, California. All participants gave an online informed consent before initiating the study questionnaire. The ethics committee waived the requirement of written informed consent for participation.
SB, JC, and MD conceived and designed the experiment and acquired the data. MD and LH analyzed the data. SB, JC, LH, and MD interpreted the results and wrote the paper. All authors contributed to the article and approved the submitted version.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Supplementary Material for this article can be found online at:
Body mass index
Amazon Mechanical Turk
The Control of Eating Questionnaire.