Effect of the COVID-19 Epidemic on Body Mass Index: Findings from a Large Survey during Wuhan Lockdown

,


Introduction
To decrease the infection risk of COVID-19 in the population, many countries have initiated home quarantine policies at the early outbreak stage.Such policies were very efective in China.However, quarantine policies may negatively infuence population health because of changing lifestyles of the people in the long run.
Some studies have shown that home quarantine may increase individuals' risks of being overweight/obese.Homestay reduces access to outdoor sports activities [1] and increases sitting [2] and screen time [3].Concerning food intake during home quarantine, people may be more likely to change their dietary habits or patterns [4,5], especially in populations with lower socioeconomic status [6] and vulnerable groups (e.g.pregnant women) [7][8][9].Tese are the main risk factors for overweight/obesity [10].Weight gained during a short time may be difcult to lose and therefore last for a very long time [1].Obesity is a global epidemic that is associated with many serious illnesses.A recent metaanalysis showed that COVID-19 patients with obesity presented with a more severe disease condition and higher fatality rate on average [11,12].To deal with the current situation (i.e.continuous emerging variant strain virus and the increasing number of both new and fatal cases of , it is of immense practical signifcance to study the efects of home quarantine policies on overweight and obesity on a population level during the COVID-19 pandemic, especially in China.Wuhan was one of the most afected cities and has implemented the most comprehensive home quarantine policy on a global scale during the early epidemic period.Te novel nucleic acid detection system implemented in May 2020 marked this city's key milestone in COVID-19 epidemic elimination.Our data were collected through a telephone survey among Wuhan residents and our study was conducted on a large scale.Tus, the survey location and timing were typical, and the sample selection and data collection procedure were rigorous and valid.Especially, the survey population was unique in that 30% of residents left Wuhan City before the lockdown (as refected within our representative random sample), which naturally divided the whole city population into two comparable groups.
A survey within a large-scale population in Wuhan city was conducted to confrm the efect of the COVID-19 lockdown on trends in population BMI, to explore efect diferences according to population characteristics, and to examine the possible mediating factors, including food habits afected and changes in physical activity associated with these efects.We aimed to provide research evidence for coping with COVID-19 and fghting the obesity epidemic.

Study Design and Participants.
Te survey was conducted in Wuhan City, China, from July 3 rd to July 31 st , 2020 by the National Research Center at Renmin University of China.Te respondents were residents living in Wuhan aged 18 years and older before the lockdown.Random digit dialing based on available cell phone registry information was used for sampling.Te sampling frame was the Wuhan cell phone number segment dataset, available from the Ministry of Industry and Information Technology.A computer-aided telephone interview platform was used to administer the survey by trained interviewers.All respondents were confrmed with regard to eligibility criteria (age and family location) before conducting the formal interviews.Each telephone call was recorded and reviewed by the survey supervisor to ensure data quality.A total of 11,223 questionnaires were collected, of which 11,206 were eligible for the fnal data analysis after excluding a few respondents who lacked information on key variables.
Te questionnaire included the following components: (1) evaluations about government countermeasures during the COVID-19 epidemic; (2) risk assessments about the COVID-19 epidemic and self-protection measures; (3) queries as to whether the respondent lived in Wuhan City during the lockdown; (4) queries as to whether the respondent underwent nucleic acid detection for their COVID-19 status; (5) the COVID-19 status of the respondents' family members; (6) the COVID-19 status within their community; (7) respondent's psychological and physical health status during the epidemic period; and (8) demographic characteristics.
All participants provided oral informed consent before the telephonic survey.Te study was approved by the biomedical ethics review board of the Health Science Center at Xi'an Jiaotong University (approval number: 2020-1235).It was conducted in accordance with the Declaration of Helsinki and its later amendments.

Study Variables
(1) Te main outcome measures evaluated in this study were self-reported weight and height, which were used for calculating BMI.BMI was classifed according to Chinese cut-of points, classifying underweight and obesity as a BMI of <18.Tird, univariate and multivariable ordinary least squares (OLS) regression analysis was conducted to explore the association between living in Wuhan during the lockdown and BMI.Adjusted variables in the multivariable OLS regression analysis included sex, age, marital status, educational attainment, hometown location, household income per capita in 2019, respondent's health status as afected by the COVID-19 pandemic, experiencing at least one kind of chronic disease, changes in smoking habits during the lockdown, and changes in drinking habits during the lockdown.Analyses stratifed analysis by sex, age, and BMI grouping were likewise performed in the multivariable OLS regression analyses.
Finally, food habits afected and changes in physical activity mediating the association between living in Wuhan during the lockdown period and BMI were explored according to previously reported methodology [14].Briefy, there were several steps for the mediation analysis.Firstly, the previous statistical relationship analysis (model 3) between BMI and Wuhan lockdown was the basis for further mediation exploration.Secondly, multivariate regression analysis (model 4 and model 5) was conducted to examine the efect of the Wuhan lockdown on each mediator, including food habits affected or physical activity change.Tirdly, model 6 was constructed based on model 3 by adding two mediators, including food habits afected and physical activity change.Te mediation efect was examined using the "sgmediation" code in Stata.Te bootstrap method was used to calculate the confdence intervals of the associated products of the mediation analyses.Mediation analyses were stratifed by sex, age, and BMI grouping.When examining the mediating efects in the analysis, sex, age, marital status, educational attainment, hometown location, household income per capita in 2019, health status as afected by the pandemic, chronic disease, and changes in smoking and drinking habits were controlled.
All statistical analyses were carried out using Stata statistical software (version 16.0, College Station, TX, USA). 1 presents the demographic characteristics of the study sample; 55.5% of the respondents were male, and 24.7% were aged over 45 years.A total of 63.7% of the respondents were married and 36.3% had a bachelor's degree; 73.7% of respondents originated from urban "hometowns," and 47.7% of the population had a household income per capita of over 10,000 RMB in 2019.

Demographic Characteristics. Table
Te demographics of the sample were compared between residents living in and residents who left Wuhan during the lockdown.Te results showed that, compared with the population living in Wuhan during the lockdown, the residents who left Wuhan were more likely to be male (P � 0.003) and were also younger (P < 0.001), more likely to be unmarried (P < 0.001), more highly educated (P < 0.001), more likely to have lived in a rural area (P < 0.001), and more likely to have household income per capita of over 10,000 RMB in 2019 (P < 0.001).
Table 2 shows diferences in health status between residents living in Wuhan and those who left Wuhan during the lockdown.Te proportion of people whose health status worsened during the pandemic was higher among those staying in Wuhan (P < 0.001).A higher proportion of residents with at least one chronic disease, including hypertension, diabetes, cardiovascular disease, and other chronic diseases, had stayed in Wuhan (P < 0.001).Concerning comparative evaluations of lifestyle changes, a lower percentage of people were those who never smoked or quit smoking (P � 0.035), a higher percentage had never drunk or had quit drinking (P < 0.001), a higher percentage had experienced little or no impact on food habits (P < 0.001), and a higher percentage had experienced a decrease in physical activity during the lockdown period (P < 0.001) among those who stayed in Wuhan.

BMI Distribution.
We compared BMI distribution between residents who stayed in and who left Wuhan during the lockdown.Te results showed the mean BMI values of 23.05 ± 0.04 kg/m 2 and 22.34 ± 0.06 kg/m 2 , respectively, for residents who stayed in or left Wuhan (P < 0.001).
Figure 1 shows BMI distributions based on Chinese adults' overweight or obesity standards.Compared with an underweight prevalence of 6.8% among people living in Wuhan during the lockdown, the underweight proportion was higher (9.7%) in those who left Wuhan during the lockdown.Te percentages of overweight or obese people were 35.5% and 27.9% for residents living in and those who left Wuhan during the lockdown, respectively.
Table 3 presents BMI distributions by social and demographic characteristics.Te mean BMI was higher among the population that stayed in Wuhan as compared to those who left Wuhan during the epidemic when Health & Social Care in the Community    Univariate and multivariate OLS regression analyses were conducted on the association between BMI and living in Wuhan during the epidemic period according to sex, age, and BMI grouping (File S1).Living in Wuhan was associated with a BMI increase in male and female residents within both univariate and multivariate OLS regression analyses (P < 0.001).After stratifying by age group, this association remained in the subpopulations aged 18-29 or 30-44 years (P < 0.001) when adjusting for sex, marital status, educational attainment, hometown location, household income per capita in 2019, health status as afected by the pandemic, chronic disease, and changes in smoking and alcohol drinking habits.BMI increase was also statistically signifcantly associated with living in Wuhan during the lockdown period among overweight or obese residents (P < 0.05), but not in underweight residents.4, which illustrates that food habits were more strongly afected by residents living in Wuhan.Moreover, this efect existed within the total population and various sex, age, and BMI groupings (all P < 0.05).Model 5 uses physical activity change as the dependent variable.Te result shows that more residents living in Wuhan experienced a decrease in physical activity during the pandemic compared to people who had left Wuhan; this efect was found among men (30-44 years) and those with a normal weight and overweight/obesity.Model 6 was constructed based on model 3 by adding the following two mediators: food habits afected and physical activity change.Tis mediation model presents results on the direct efects of living in Wuhan during the lockdown period on BMI increases within the study population and according to sex, age, and overweight/obesity groupings.

Mediation Analysis. Model 4 used food habits afected as a dependent variable, as shown in Table
Te indirect efect proportions of living in Wuhan leading to a decrease in physical activity and consequently to an increase in BMI were 2.3% and 4.1%, respectively, among the study population and in the 30-44 years age group.In addition, an indirect infuence of living in Wuhan afecting food habits, which led to a BMI decrease, was found in the 45 and above age group.

Discussion
Some studies have explored the infuence of the COVID-19 pandemic on weight gain or overweight/obesity.However, few studies focused on the early stage of the outbreak when home quarantine policies were initiated strictly around the world.Moreover, there were methodological problems in some of these studies, especially with sample selection and data collection.For example, some studies were more qualitative or needed to recruit a sufcient sample size [15,16].Other quantitative studies with large sample sizes recruited respondents through the Internet or social media, and the questionnaires were self-administered; these could, therefore, not ensure sufcient sample representativeness and survey quality [17][18][19][20][21][22].In addition, the home quarantine policy in most regions of China was incomplete, and conclusions based on these studies should be drawn with caution.
After the outbreak of the COVID-19 epidemic, the Chinese government took decisive action by instituting a strict lockdown in Wuhan City for 76 days, from January 23rd to April 8th, 2020.Consequently, the early spring festival lockdown split the entire population into two groups naturally, which provided a historically unique survey feld.However, conducting a face-to-face survey during the epidemic was impractical, and obtaining a sample frame for a web survey was impossible.Fortunately, using a Cloud-CATI platform developed at our research center, the survey was initiated quickly and smoothly during this special occasion.Approximately two months after the Wuhan lockdown was lifted, a random sample was selected by an RDD method to survey over 10,000 residents living in Wuhan.Tis study, to our knowledge, is the largest scale random sampling survey regarding the COVID-19 epidemic outbreak that has been conducted to date.
Our study showed that the mean BMI was 0.71 kg/m 2 higher in residents living in Wuhan compared to residents who had left Wuhan.Te population living in Wuhan during the lockdown had a 1.9-2.2kg higher BMI on average than those who had left Wuhan in reference to an average height of 169.7 cm and 158.0 cm in men and women, respectively, as reported by Chinese nutritional researchers and with respect to chronic disease status distributions as of 2020.Tese results were consistent with those of other studies conducted in China and abroad [15-17, 19, 20].Because there are inevitable diferences in study characteristics across populations, including study region, demographic and lifestyle characteristics, sampling methods, and study periods, the results might difer across investigations.However, our results have undoubtedly added more evidence about the association between choices in residential locations 6 Health & Social Care in the Community  Health & Social Care in the Community during the Wuhan lockdown and higher BMI levels among the residents of Wuhan City.
Te mean BMI among residents living in Wuhan was 0.42 kg/m 2 (P < 0.001) higher than that of residents who had left Wuhan during the lockdown within a multivariate OLS regression analysis adjusted for sex, age, marital status, educational attainment, hometown location, household income per capita in 2019, health status as afected by the pandemic, chronic disease, and changes in smoking and alcohol drinking habits.Tis result was slightly higher than the 0.31 kg/m 2 diference reported in a prior meta-analysis [23], indicating that we detected direct efects of living in Wuhan on BMI.Direct efects were found in both the male and female populations but not in those aged over 45 years or underweight or normal weight, which was consistent with the results of other studies [5,24].
It is well known academically that BMI increases are associated with changes in food habits and physical activity.Hence, this study explored the mediation efects of changes in food habits and physical activity on the association between living in Wuhan during the lockdown period and BMI.Te results showed an indirect path of living in Wuhan with regard to decreases in physical activity leading to increased BMI in the total study population and in those aged 30-44 years, which meant that living in Wuhan had an indirect efect on BMI that was attributed to changes in physical activity in addition to the direct efects seen in the younger population.Moreover, an indirect path of living in Wuhan, acting through changes in food habits, which, in turn, led to BMI decreases, was found in individuals aged over 45 years.Tis meant that living in Wuhan decreased BMI in the middle-aged and older population because of its efects on food habits.Te result was similar to a prior study fnding of decreased weight occurring during the COVID-19 epidemic in an investigation conducted in Italy.Hence, we conclude that the middle-aged and older population should be monitored more closely with regard to their malnutrition status during the pandemic lockdown [25,26].
Tere are several strengths of this study.First, a random digital dialing method survey on a large scale was employed in Wuhan, which implemented the most comprehensive home quarantine policy for the most prolonged period of any city in China.Te results were relatively high quality.Second, this is the frst survey with a random sampling method to examine the association of weight gain with the COVID-19 lockdown in China based on large-scale population analysis adjusting multiple confounders.Tird, this study provided new evidence for the COVID-19 lockdown increasing residents' BMI, which was necessary for the prevention and early intervention of noncommunicable chronic disease.
Tis study has several limitations.First, a random digital dialing method was used in the survey.Still, it was impossible to avoid nonresponse bias because of the low response rate inherent to CATI feldwork and survey-based epidemiologic research.However, considering the large scale of this survey, our results (based on surveying over 10,000 residents) may be regarded as validated.Next, the respondents' weight before the COVID-19 epidemic was not collected, hence precluding the possibility of comparative evaluations conducted before and after the lifting of the lockdown.We adjusted many weight-associated covariates to increase comparability between the residents living in Wuhan and those who left during the epidemic.In addition, because of the questionnaire length limit, some factors relevant to weight (such as psychological factors) were not included in this evaluation, which may be partly associated with higher BMI among those living in Wuhan during the epidemic.Furthermore, the food habits measurement was not added based on this telephone survey.However, this study focused on the degree of infuence of food habits by COVID-19; hence, it is necessary to use innovative approaches and tools to assess them in the future [27,28], especially in China with more than one billion Internet users.Based on a large-scale survey administered in Wuhan City, the results provide direct evidence for the efects of the pandemic lockdown on BMI and overweight/obesity.
Te fndings should be paid close attention for the following reasons: (1) Te survey result showed a moderate self-reported proportion of overweight/obesity in Wuhan City.However, the continuously increasing trend with respect to overweight/obesity in China and around the globe should not be neglected.Based on this background, the BMI increase caused by the pandemic lockdown in Wuhan City puts forward a new challenge for the future control and prevention of overweight/obesity.(2) Currently, the COVID-19 pandemic is ongoing in China and around the world.Moreover, meta-analyses have shown that individuals with obesity are more at risk for contracting COVID-19 and hospitalization, intensive care unit admission, and COVID-19-associated mortality.Hence, the prevention of obesity is signifcant to COVID-19 control and prevention.(3) According to the latest statistics, there have been over 3.4 billion doses of COVID-19 vaccinations administered nationwide, and over 1 billion people have been vaccinated.However, vaccines are less efective for people with obesity because of their weakened immune response [29].Hence, based on our results, more educational workshops and prevention work should be conducted to avoid possible increases with regard to chronic disease burden and difculties in COVID-19 prevention due to shifting distributions of overweight and obesity.

Conclusions
Tis study showed a relationship between the COVID-19 lockdown and overweight/obesity in Wuhan city in China, especially among the young or overweight/obese, partly due to decreased physical activity.In China's social and economic development context, enhancing physical activity and nutrition intervention during the lockdown period is benefcial for preventing overweight/obesity.It further contributed to decreasing and delaying the development of noncommunicable chronic diseases among the population and reducing the related burden in China.

Figure 1 :
Figure 1: Weight status among residents living in and residents who left Wuhan during the epidemic period.Wuhan, China, 2020.
Table 4  is adjusted for sex, age, marital status, educational attainment, hometown location, household income per capita in 2019, health status as afected by the pandemic, chronic disease, and changes in smoking and drinking habits.b Te dependent variable in model 4 is food habits afected, adjusted for the same covariates as in model3.c    Te dependent variable in model 5 is physical activity change, adjusted for the same covariates as in model 3. d Te dependent variable in model 6 is BMI, adjusted for two more variables (food habits afected and physical activity change) as compared with model 3.
a "other regions" include all regions outside of Wuhan City.

Table 2 :
Comparison of lifestyle and health status between residents living in and leaving Wuhan during the lockdown.Wuhan, China, 2020.

Table 3 :
Comparison of BMI distribution between residents living in and those who left Wuhan, by social and demographic characteristics.Wuhan, China, 2020.

Table 4 :
Mediation analysis on factors mediating the association between BMI (body mass index) and living in Wuhan, by total, sex, age group, and weight status during the lockdown period.Wuhan, China, 2020.Group