Increase in body mass index during the COVID-19 pandemic among people who smoke: An analysis of multi-site electronic health records

The effects of the COVID-19 period among people who smoke (compared by sex) are largely unknown. The purpose of this study was to compare body mass index (BMI) increase among men and women who smoked during the pandemic. We used a retrospective longitudinal, observational study design of secondary data. We used electronic health records from TriNetX network (n = 486,072) from April 13, 2020-May 5, 2022 among adults aged 18–64 who smoked and had a normal BMI prior to the pandemic. The main measure was a change of BMI from < 25 to ≥25. Risk ratio was determined between men and women with propensity score matching. Overall, 15.8% increased BMI to ≥25; 44,540 (18.3%) were women and 32,341 (13.3%) were men (Risk Ratio = 1.38, 95% CI: 1.36, 1.40; p < .0001). Adults with diabetes, hypertension, asthma, COPD or emphysema or who were women, were more likely to develop BMI≥25 during the pandemic. Women who smoked were more likely to have an increase in BMI than men who smoked during the COVID-19 period.


Introduction
During the early phase of the coronavirus disease 2019 (COVID-19) pandemic, emergencies were declared which included self-quarantines, travel restrictions, and lock downs. In the U.S., residents were encouraged to stay indoors, wear masks, wash hands frequently, and keep a social distance of six feet between people. Multiple waves of COVID-19 resulted in prolonged high levels of stress associated with the seriousness of the disease. Anecdotal reports indicated that stress, anxiety, depression, less physical activity, and more sedentary activities resulted in weight gain, described as "quarantine 15" [1]. For example, in a survey of people with a mean body mass index (BMI) of 27 (n = 111), 22% self-reported gaining 5-10 pounds during the pandemic [1]. Respondents in another survey (n = 764) reported mean weight gain of 0.62 kg [2]. In an anthropometric study of weight and COVID-19 (n = 11,534), the mean weight increase for men was 3.41 kg; while the mean weight increase for women was lower, at 3.09 kg; however, the weight gain did not correspond to a significant difference in BMI increase between men and women [3]. The number of people developing BMI�25 during the COVID-19 pandemic was particularly concerning due to the link between obesity and increased COVID-19 morbidity/mortality [3][4][5][6] through the effects of adipose tissue on the immune system such as reducing the response to antiviral agents through poor T-cell and macrophage actions; chronic activation of the immune system; and altered secretion of inflammatory mediators [4]. Additionally, during the 2009 H1N1 pandemic, obesity was a risk factor for severity, hospitalization, increased risk of transmission and mortality, therefore, extrapolating to COVID-19, obesity is a concern with similar poor health outcomes [4]. Examining overweight/obesity and smoking during the COVID-19 pandemic are important as: 1. smoking is associated with risk of central obesity [7]; 2. smoking cessation is associated with weight gain [8]; and 3. smoking increases morbidity/mortality with COVID-19 [9,10].
Mechanistically, smoking increases COVID-19 morbidity/mortality by the direct toxic effects of smoking injury to the nose and lower airway tissue enhancing infectivity; and, an antiviral inflammatory cascade that increases the virus's pathogenesis [11]. In addition, disturbances in the angiotensin-converting enzyme 2, or in the renin-angiotensin system may be altered with smoking [11], resulting in inadequate lung protection from infections [11].
The purpose of the present research was to examine increases in BMI during the pandemic (April 13, 2020 to May 5, 2022) among people who smoked and had a normal BMI prior to the pandemic. This observational study assesses the change in BMI with repeated observations over two time periods using harmonized electronic health records (EHR).

Ethical statement
The research received acknowledgment from the West Virginia University Institutional Review Board as non-human subject research.

Study design
The research had a longitudinal, observational study design of secondary data.

Data source
The data were extracted from TriNetX (Cambridge, MA), a system in which real-time, deidentified healthcare organization (HCO) data are available to researchers. Before distribution, the data are coded for anonymity and are extracted from structured EHR data and narrative texts [12]. Data analysis is available through TriNetX cloud-based internal custom platform from Java 1.9.0_171, R 3.44 (R core Team, Vienna, Austria) and Python 3.6.5 [13]. This population-based study included data of EHR of all who visited one of the 58 reporting HCO's. Funding: This study was financially supported by the National Institute of General Medical Sciences in the form of a grant (5U54GM104942-05) awarded to RCW. This work was also financially supported by the National Institutes of Health (NIH) in the form of a grant (1OT2OD032581-01) awarded to US. This work was also financially supported by the National Institute on Minority Health and Health Disparities through the Texas Center for Health Disparities (NIMHD) in the form of a grant (5U54MD006882-10) awarded to HW and US. This work was also financially supported by NIH funded AIM-AHEAD program in the form of a grant (NIH/ 1OT2OD032581-01) awarded to US. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests:
The authors have declared that no competing interests exist.

Inclusion/Exclusion criteria
The inclusion criteria at baseline were records with complete data on biological sex (male/ female), age (with restrictions to �18 years and <65 years), that the EHR indicated current smoking (ICD-10-CM codes F17 or Z72.0) and had a pre-existing BMI �18 to <25 on March 13, 2020. Records with a BMI�25 before March 13, 2020 were excluded. Additional inclusion criteria were that the participant had a follow-up appointment in which BMI was determined by May 5, 2022, and that the participant reported that they had continued to smoke.

Measures
The outcome variable was moving to the category BMI�25 from lower BMI between April 13, 2020 to May 5, 2022 (yes, no). Biological sex was the main independent variable (male, female) of interest for the comparison.
Propensity score matching variables The TriNetX system provides characteristics and signal strengths to determine if matching is statistically warranted. Standard differences are provided to determine matching success with <25% standard difference recommended on matched variables. For this study, several epidemiological factors were matched: hypertensive disease, diabetes mellitus, asthma, emphysema, and other chronic obstructive pulmonary diseases (COPD), race and age.

Time frame
The study included all available EHR that excluded BMI�25 before the March 13, 2020 baseline. The index date for the onset of BMI�25 was April 13, 2020. The development of BMI�25 was analyzed with a short-term horizon, the pandemic period from April 13, 2020 to May 5, 2022 [14].

Statistical analysis
The internal TriNetX software was used for change in BMI category and risk of BMI�25 during the pandemic using logistic regression analyses.

Results
Before matching, the study sample included 531,416 records of people who smoked in April 2020.

Discussion
We compared the increase in BMI from normal BMI to BMI�25 during the pandemic among people who smoked. Overall, 15.8% developed BMI�25. Women were 38% more likely than men to move from normal BMI to BMI�25. Two substantial percent changes to BMI�25 were among people who had diabetes (39.2% developed BMI�25) and people who had asthma (37.3% developed BMI�25). These results are concerning as, in addition to COVID-19, there are long-term consequences of BMI�25 and diabetes, as well as BMI�25 and asthma exacerbations.
Several factors are potentially associated with the findings. The COVID-19 quarantine, fear, economic downturn, and closures likely increased stress and that psychopathology could have influenced eating behavior [15]. Researchers indicated many symptoms of psychopathology did not vary by gender during that time, but eating pathology was increased in women (Beta = -0.013 [95% CI: -0.042, -0.004; p = 0.049) [15]. Previous researchers indicated that adults who smoked were more likely to weigh less than adults who did not smoke [7]. Nicotine is considered to increase metabolic rate, decrease metabolic efficiency, decrease caloric absorption and increase the acute anorexic effect [7]. Observing a change in BMI from normal BMI to BMI�25 in people who smoke carries additional concerns for potential COVID-19 outcomes and other health outcomes.

Strengths and limitations
This study's strength is the use of national, real-time data from 58 HCO's with the ability to match and determine BMI outcomes. The study included diverse HCOs across the U.S. The study findings have implications for COVID-19 prognosis as other researchers have indicated that BMI�25 is a significant factor for severe COVID-19 [4,16].
A study weakness is the use of data that were designed for clinical purposes rather than research. For example, there is a lack of race/ethnicity data in some HCO EHR as race/ethnicity may not have been of clinical relevance to that HCO. The propensity score matching approach assumes "no unmeasured confounders." However, the database did not have information on variables that are not routinely collected in clinical visits such as physical activity and calorie intake. The observed BMI change could have resulted from inactivity, increased calories, or a combination of both during the pandemic.
There is also the lack of information in the EHR's concerning the number of cigarettes smoked on a daily basis and historical pack-years smoked. Other limitations are that we did not match/control for medications; or match for other chronic conditions that may affect BMI beyond hypertension, diabetes, asthma, emphysema, and chronic obstructive pulmonary disease. Using the TriNetX platform for analysis, we were limited by the query procedures available in the platform. Although it is important to examine biological sexspecific effects as conclusions from one sex cannot be assumed for the other [17], the emphasis of the study was change BMI�25 through the factors of smoking and sex. We recognize that the development of overweight/obesity requires time and there are factors other than the pandemic that may have influenced weight gain in the study's time window (0.08 year to 2.16 years from declaration of COVID restrictions in the U.S.). Furthermore, the baseline measurement time window can vary from a minimum of xx to a maximum of xx.

Public health considerations
Smoking and BMI�25 increase COVID-19 morbidity/mortality, therefore it is important to have the epidemiological information for efforts in reducing COVID-19 harm. It is important to address the risks associated with BMI�25 and COVID-19 as well as guiding the public toward a normal BMI and tobacco cessation. The messaging needs to be person-centered and appropriate to improve overall whole health.