Global Impact of COVID-19 on Weight and Weight-Related Behaviors in the Adult Population: A Scoping Review

Objective: To provide an overview of what is known about the impact of COVID-19 on weight and weight-related behaviors. Methods: Systematic scoping review using the Arksey and O’Malley methodology. Results: A total of 19 out of 396 articles were included. All studies were conducted using online self-report surveys. The average age of respondents ranged from 19 to 47 years old, comprised of more females. Almost one-half and one-fifth of the respondents gained and lost weight during the COVID-19 pandemic, respectively. Among articles that examined weight, diet and physical activity changes concurrently, weight gain was reported alongside a 36.3% to 59.6% increase in total food consumption and a 67.4% to 61.4% decrease in physical activities. Weight gain predictors included female sex, middle-age, increased appetite, snacking after dinner, less physical exercise, sedentary behaviors of ≥6 h/day, low water consumption and less sleep at night. Included articles did not illustrate significant associations between alcohol consumption, screen time, education, place of living and employment status, although sedentary behaviors, including screen time, did increase significantly. Conclusions: Examining behavioral differences alone is insufficient in predicting weight status. Future research could examine differences in personality and coping mechanisms to design more personalized and effective weight management interventions.


Introduction
Since the COVID-19 pandemic emerged about a year ago, it has infected more than 72 million people and claimed above 1.5 million lives [1]. As of 8 December 2020, approximately 152 countries/territories have experienced some form of lockdown or confinement that curtailed social mobility to prevent the spread of the COVID-19. This includes changes in social norms such as working from home, hosting smaller social gatherings and reducing air travel. However, the impact of such measures on weight-related lifestyle behaviors and weight changes remains unclear. While some studies reported an increase in time for physical activities and preparing homemade food [2,3], others have reported an increase in sedentary behaviors [4], decreased physical activity [4], increased consumption of junk food and weight gain [5]. The COVID-19 pandemic is a novel disease of which its impact on the global adult obesity situation is unclear. More than 13% and 39% of the global adult population are obese and overweight, respectively. Current evidence highlights two worrying trends between COVID-19 and obesity, which could well form a vicious cycle: (1) COVID-19 associated with weight gain and (2) worse patient outcomes in patients with concurrent obesity and a COVID-19 infection [6][7][8].
Due to the novelty of this disease, the range, nature and magnitude of its impact on weight management in healthy adults remain unclear. Existing systematic reviews tend to focus on the outcomes of patients with obesity diagnosed with COVID-19, but the authors could not find systematic reviews on the effects of COVID-19 on weight and weight-related behaviors [9][10][11]. Therefore, a scoping review is timely and appropriate in mapping the current evidence on the impact of COVID-19 on weight management in healthy adults, specifically to identify literature gaps (not research gaps) to inform future research directions [12]. Although COVID-19 prevention measures such as reduced social mobility will gradually be weaned off with time, measures like working from home will most likely be a new norm. Therefore, conducting a scoping review would provide an overview of the current evidence on the impact of COVID-19 on weight management, identify research gaps and determine the need to conduct further systematic reviews to answer specific research questions [13]. The aim of this review was to investigate what is known about the changes in weight and weight-related behaviors in healthy adult populations during the COVID-19 pandemic.

Materials and Methods
This systematic scoping review was conducted according to the five-phased methodology developed by Arksey and O'Malley [14]. Scoping reviews are useful for exploring relatively new evidence and phenomenon that remains ambiguous in terms of what research questions to evaluate in a systematic review or primary research. Specifically, it is valued for identifying the breadth, key concepts and key conceptual factors of evidence available on a certain topic while identifying current knowledge gaps to guide the direction of future inquiries (e.g., conducting a systematic review). This differs from the objectives of conducting a systematic review that aims to analyze current evidence and answer specific research questions to guide decision-making, practice and policies [15]. The study findings are illustrated according to the preferred reporting items for systematic reviews and meta-analyses extension for scoping reviews (PRISMA-ScR) checklist (Table S1).
Phase 1: Research questions This study's research question was developed based on the population, intervention, comparison and outcome (PICO) framework to identify changes in weight and weightrelated behaviors during the COVID-19 pandemic in healthy adult populations. Thus, the research question of this study was, "what is known about the changes in weight and weight-related behaviors in healthy adult populations during the COVID-19 pandemic?" Due to the limited number of studies that reports the impact of the COVID-19 pandemic on weight and weight-related behaviors, we included studies that examined populations with a majority of adults (i.e., mean age is >18) and excluded studies that reported exclusively on populations that were <18 years old. Studies on community-dwelling populations without diseases except being overweight or with obesity during this pandemic were included.
Phase 2: Literature search A systematic three-step search strategy was used to identify relevant literature that was published up to 8 October 2020. First, search terms were generated iteratively through searches on CINAHL and PubMed using the keywords "weight", "obesity," and "COVID-19". MeSH terms were also identified and used as search terms. Second, seven databases (CINAHL, Cochrane Central, Embase, PsycInfo, PubMed, Scopus and Web of Science) were searched for relevant articles published from the inception of the COVID-19 pandemic to 8 October 2020. The search terms used were "obes*", "overweight", "weight", "COVID", "COVID-19", "SARS-COV2", "SARS-CoV-2", "2019-nCoV", "2019 coronavirus", "behavio*". More information on the different combinations of the search terms used according to different databases is shown in Table S2. Lastly, the references of the included studies were searched for additional articles.
Phase 3: Study selection Studies were included if they: (1) described the changes in weight or weight-related behaviors (e.g., dietary or physical activity) during the COVID-19 pandemic and (2) were on community-dwelling adults without mention of other diseases except for obesity and being overweight. Studies were excluded if they focused on: (1) biological changes due to a COVID-19 infection; (2) obesity as a risk factor of COVID-19 infections and outcomes; (3) did not discuss weight-related changes related to COVID-19; and (4) were non-primary studies, e.g., simulation/modeling studies.
A total of 396 articles were retrieved. After removing 144 duplicate articles, the remaining titles and abstracts were screened for eligibility, which 77 articles were eligible for full-text screening. After excluding articles with reasons shown in Table S3, 18 articles remained and were included in this scoping review.
Phase 4: Data charting A data extraction form was created by HSJC and pilot tested on 5 studies. While doing so, common weight-related changes were identified, namely change in dietary behaviors, physical activity behaviors and other lifestyle behaviors. Therefore, the data extraction form was modified to expand the heading "weight-related changes" to the specific ones mentioned earlier. An excel spreadsheet was created to consolidate the extracted data according to the following headings-authors, year of publication, country of origin, study design, survey type, recruitment period, aim of study, follow-up, total number of participants, age, race, baseline BMI, BMI categories, BMI categories' cutoff scores, proportion of participants overweight, weight change, weight measurement instruments, diet change, diet measurement instruments, physical activity change, physical activity instruments, other weight-related lifestyle behavior changes, predictors of weight, diet, physical exercise and other weight-related lifestyle behavior changes, the significance of change (statistically significant or not) and important results. Countries of origin were recoded into World Health Organization (WHO) regions, and articles were regrouped into those that evaluated changes in weight, diet and physical activities.

Results
Phase 5: Collating, summarizing and reporting the results The 19 included articles represented 61,764 respondents, where the sample sizes of the articles ranged from 90-13,515 (median = 1844), mean/median age ranged from 19 to 47 years old with a median of 33.7 years old. 52.6% of the articles were from the European region (i.e., Belgium, Croatia, Italy, Poland, Spain, UK), 83.9% were cross-sectional descriptive studies, and all outcomes were collected through online self-report surveys (the usual method of data collection during the pandemic due to social distancing policy). The majority of the studies recruited participants during the months of April and May (72.2%) and comprised of more females than males (of the 17 studies that reported the proportion of female participants). Ten articles reported the participants' mean baseline BMI that ranged from 20.7 kg/m 2 to 27.7 kg/m 2 ; nine reported the proportion of participants who were overweight at baseline (25-60%), and only five articles reported cutoff score used to classify one's BMI as overweight (four studies used 25 kg/m 2 , only one used 23 kg/m 2 from China). More information on the study characteristics is detailed in Table 1.

Changes in Weight
Eleven out of 19 articles mentioned changes in weight where ten articles mentioned weight gain that ranged from 12.8% to 48.6% and six articles mentioned weight loss that ranged from 13.9-19.4% (Table 3) [2][3][4][5]18,19,21,25,26,28,29]. Two articles reported the combined proportion of participants who lost weight and did not perceive a change in weight [3,25]. It should be noted that these results were all derived from self-reports of perceived weight changes across different durations of confinement, cultural dietary norms (i.e., two studies focused on changes in Mediterranean diet change), and populations with different sociodemographic characteristics. Six studies examined the predictors of weight gain which included being in the middle-ages [4,26], female (n = 3; two studies reported odds ratio (OR) = 1.23-2.73) [2,4,18], higher baseline BMI (n = 3; two studies reported odds ratio OR = 1.07-1.12) [2,4,18], increased total food consumption [5], consumption of junk food (n = 2; OR = 1.76-3.12) [2,4], eating in response to sight and smell of food, stress eating and snacking after dinner [29], physical exercise (n = 5, three studies reported OR= 0.51-0.76), sedentary behavior ≥6 h/day (OR = 1.85), taking active breaks (OR = 0.72) [4], low water consumption (OR = 1.58) [4] and less hours of sleep a night [29]. However, one study did not find gender as a significant predictor of weight gain [26]. Alcohol consumption [18], screen time [5,29], education level [18,26], place of living and employment status [27] were also not significant predictors of weight gain. On the other hand, while not assessed for associations with weight changes, other lifestyle behavior changes were identified, including a general increase in sleep hours per night (30% to 54.8% of the respondents indicated an increase) [2,4,22,28,30], screen time per day (49.1% to 84.1% of the respondents indicated an increase) [5,22,28], stress/anxiety/boredom (42.7%) and concerns over weight, shape and eating [5]. However, there were contradictions regarding the changes in cigarette smoking per day [2,18].
Being female was a significant predictor of increased appetite [2], increased consumption of homemade meals and healthy eating [3]. Age was a significant predictor of night snacking (OR = 0.97) [2], junk food consumption (OR = 0.98) [2], adherence to a Mediterranean diet (respondents aged 18-30 years had a higher MEDAS score compared to the younger and elder population) [2] and higher adherence to a healthy diet [3,22]. However, there were mixed findings regarding age as a predictor of dietary behavior. While one study reported a decrease in the likelihood of adopting a healthy diet with age (OR = 0.65, 0.33, 0.22 for 40 s, 50 s, more than 60 years old) [22], another study reported lower adherence to healthy diets in those aged 21 to 50 years old compared to those above 50 years old [3]. Only one study assessed the change in appetite that was shown to predict junk food consumption (OR = 4.04) and healthy eating (OR = 1.72) [2]. It was also associated with a change in work habits (e.g., working from home), BMI and being female. This study also did not find BMI and age as significant predictors of healthy eating. While one study reported that those from the North of Italy were less likely to have increased appetite (OR = 0.53) and have significantly higher adherence to a Mediterranean diet [2], another study reported that those from the North of Spain were less likely to adopt healthy eating habits (OR = 0.67) [3]. While one study reported that an increased BMI predicted an increase in appetite (OR = 1.07), junk food consumption (OR = 1.03) and lower adherence to a Mediterranean diet [2], another study reported that being overweight (OR = 1.31) or obese (OR: 1.64) were significant predictors of adherence to a healthy diet [22]. Adherence to an unhealthy diet was predicted by a decrease in physical activity (OR = 2.62), living in macroeconomic regions (OR = 1.43-1.47), increased screen time (OR = 1.54) and decreased consumption of homemade food (OR = 3.06) [5,22].

Authors Change in Dietary Behaviors Predictors of Dietary Behaviors Change Non-Significant Predictors
Rodríguez-Pérez 2020 •
From reviewing the included studies, up to approximately half of the respondents perceived weight gain during the COVID-19 pandemic period, while up to a fifth of the respondents had reportedly lost weight. There were also inconsistencies in the changes in dietary habits in terms of the consumption of healthy or junk foods and that of physical activity in terms of frequency, duration and energy expenditure. This could be associated with various individual characteristics and prepotent lifestyle habits that influenced weight-related lifestyle changes during the pandemic period. Respondents with a higher baseline BMI was shown to be more likely to experience weight gain, possibly due to a predisposition to eating in response to visual and olfactory food temptations, stress and emotional eating, as mentioned in the results section [2][3][4][5]18,22,26,29]. In general, the percentage increase (59.6%) in total consumption was more than that of a decrease (33.5%), and the adherence to a healthy diet increased slightly more than those who decreased [3][4][5]22,26]. This could explain the higher proportion of participants who gained weight despite having a higher adherence to a healthy diet due to a higher overall calorie consumed. However, further research is needed to support this speculation by using more objective calculations of energy intake and expenditure instead of using self-reported questionnaires that examine perceived intake change using Likert scales or "yes/no/no change" response options. Additionally, more than 50% of the respondents were reported to have increased eating episodes with friends and families in response to cravings, food stimuli and emotions [29]. Moreover, contrary to our speculation that COVID-19 decreases social eating, one study reported a 59% increase in social eating, specifically with family and friends [31]. This could be influenced by one's personality traits and circumstances. For example, the frequency of social eating could have reduced during the COVID-19 pandemic, but while a more extroverted person may replace it with social eating with friends and families, one who is more introverted may not. In this case, the introverted individual could lose weight due to reduced total calorie consumption, but the extroverted individual could gain weight due to increased total calorie consumption. This is supported by a study that reported personality traits such as neuroticism, extraversion, agreeableness and conscientiousness to be significantly associated with health behaviors and self-efficacy in weight loss [32]. Future studies could consider exploring personality factors such as the Big five personalities when examining weight-related behavior trends to develop more personalized and targeted interventions.
Among the studies that examined changes in weight, diet and physical activity concurrently [2,4,5,24], weight gain was reported alongside an increase in total food intake in 36.3% to 59.6% of the respondents and a decrease in physical activity from 67.4% to 61.4% of the respondents. However, only one study examined and reported the association between increased eating and decreased physical activity [4]. Community-dwellers who were in the middle-ages and of the female sex were found to be more likely to gain weight, possibly due to an increased appetite, junk food consumption and total food consumption [2,4,18,26]. However, both predictors were also found to be associated with healthy eating, which suggests that weight gain could be associated with overconsumption (even for overconsumption of healthy food) or that these predictors only predicted a small change in weight status. Concurrently, one study reported that an increase in appetite predicted 1.7 to 4 times higher likelihood of junk food consumption and healthy eating, while another study reported a higher likelihood of healthy dietary adherence in individuals who were overweight [2,22]. Moreover, respondents, who were working from home, consumed less water, had less sleep at night, and stress eat could be more likely to gain weight. Other well-established predictors of weight gain were supported, including decreased physical activity, increased sedentary behavior and higher baseline BMI. However, there were mixed findings in terms of the proportion of respondents who increased versus decreased physical activity [2][3][4][5]16,17,21,22,24,28,30]. Therefore, while COVID-19 measures are to be in place for the next few years before they can be reasonably eradicated or be safe enough for the measures to be removed, health authorities could implement health promotion strategies to remind the citizens to be mindful of their total consumption (not only to eat more healthy foods) and stay physically active. This is especially for those who have a higher baseline BMI, of middle-age and of female sex as they are more likely to experience weight gain amidst a pandemic. Strategies could include teaching the public population on techniques to reduce appetite (e.g., taking small frequent meals), reduce snacking (e.g., distracting thoughts of snacking by performing physical activities), improve sleep (e.g., doing mindfulness exercises) and slotting physical exercises into their daily routines (e.g., taking the stairs instead of the lift).
Studies included in this review did not illustrate significant associations between weight gain and factors such as alcohol consumption, screen time, education, place of living and employment status, although sedentary behaviors and screen time did increase significantly [5,[16][17][18]26,[28][29][30]. This could suggest a moderating effect of screen time on the relationship between sedentary behaviors and weight gain, supported by a study where screen time seemed to be associated with weight gain only if it reduces physical activity, especially in adolescents [33]. An increase in screen time could also affect one's sleep schedule and quality, an observed effect of the COVID-19 lockdown that is associated with weight change and depressive symptoms [34,35]. On the other hand, non-significant findings between socioeconomic status and weight gain contradict a study with a sample of 17,724 participants [33], possibly due to the relatively small sample sizes (N = 3027 and N = 1097) [18,26].

Limitations
Our attempt at identifying the impact of COVID-19 on weight and weight-related behaviors was challenging because while some studies reported the statistical significance of the changes before and after the COVID-19 pandemic, others merely mentioned changes in proportion. Therefore, some changes could have been exaggerated or confounded by other variables such as seasonal changes in temperate countries that cause weight change. Moreover, the time period by which the changes occurred was unclear. It is possible that there exists a behavior change trajectory in coping with the pandemic, where such changes could normalize back to baseline once an individual gets used to the current circumstances-resulting in a minimal net weight change. However, such observation requires a longitudinal study design, which was only used in two studies that reported weight gain in 28.4% to 40% of the respondents [5,19]. Another limitation is in the selfreported nature of all the studies, where reported weight changes could be inaccurate due to different calibrations and types of weighing scales used. Furthermore, some studies estimated weight changes based on the participants' perceived weight change by asking them if they gained, lost or maintained their weight. While we extracted potential predictors of weight and weight-related behavior changes, statistical conclusions could not be achieved because of the heterogeneity of data analysis methods used. While some reported the odds ratio and the statistical significance of each variable in a model tested, others only reported the proportion of respondents who expressed changes. Moreover, the included studies were not consistent in control variables, all of which could have given rise to the mixed findings on the aforementioned predictors. Lastly, we did not search for literature in other languages, such as Chinese literature, from Chinese databases as both authors were generally English-speaking. Searching for articles from Chinese databases could have provided a more geographically balanced overview of the topic of inquiry.

Conclusions
While existing studies suggested a higher proportion of people, who gained as compared to those who lost weight, findings regarding the predictors of diet and physical activity changes remain mixed. Moreover, none of the included studies examined other influencing factors of weight-related behaviors, such as personality factors, which could be strong determinants of weight change. Future research could focus on the predictors of different weight-related adaptations (i.e., increase or decrease in weight-related behaviors) and use more objective outcome measures to enhance the development and accuracy of predictive models for weight management interventions. Health promotion initiatives could also consider exploring the respondents' needs and preferences in designing weight management programs instead of just prescribing recommendations to follow. Nonetheless, these findings highlighted two behavioral health adaptations-an increase and decrease in the adoption of a healthier lifestyle-to cope with the pandemic measures. This could inform further research, practice and policies in enhancing healthy coping behaviors in a post-COVID-19 era of new norms.