COVID-19 Pandemic Associations on Mental and Physical Health in African Americans Participating in a Behavioral Intervention

Abstract

The COVID-19 pandemic has had disproportionate effects on communities of color, with higher death rates among African Americans (AA). The purpose of this study was to assess associations in African Americans’ mental and physical health with the COVID-19 pandemic. Data for this study came from a larger nutrition intervention of AAs in the Southeastern United States, the Nutritious Eating with Soul study. Data collected before and after March 15, 2020 (the day when local South Carolina schools and businesses closed), were analyzed to assess the association of the pandemic on participants’ stress, control of healthy eating, physical activity, and body mass index. Repeated measures analysis of covariance using full maximum likelihood estimation to handle missing data was used. At the onset of the COVID-19 pandemic, 150 participants were enrolled in the study (48.2 ± 10.6 years old, 79% female, 75% with college degree or higher). Participants’ reporting of stress did not show statistically significant change over time. Cognitive control increased 1.43 points (F = 20.60, p < 0.0001) and body mass index increased 0.72 kg/m² (F = 10.68, p = 0.001). Future longitudinal studies should investigate how the COVID-19 pandemic continues to present challenges to understanding and improving health among African Americans. The study is registered at www.clinicaltrials.gov NCT03354377.

Introduction

Compared to other race and ethnicity populations, African Americans have disproportionately higher rates of obesity and cardiovascular disease [1, 2]. These disparate health conditions are also prevalent comparing populations living in the southeastern areas of the USA to other areas [3]. Healthy behaviors such as stress management, healthy eating, and physical activity are known to mitigate obesity and cardiovascular disease. However, African Americans have reported how a lack of knowledge or self-efficacy, social support, and environmental influences make it difficult to begin and maintain behavior changes [4].

As researchers develop, deliver, and redesign interventions and programs to facilitate positive health behavior changes in African Americans, external circumstances and events may occur, impacting participants’ trajectory of behavior changes [5]. For example, one type of event that may impact positive behavior change is natural disasters. Research has demonstrated the negative side effects of natural disasters on a community’s social, emotional, and physical health [6, 7]. Hurricanes have been documented to produce negative effects on stress levels and health [8], and the physical damage caused by hurricanes exacerbates existing disparities African Americans may face regarding accessing healthy foods [9, 10].

Similar to a natural disaster, the COVID-19 pandemic was an external circumstance which caused an abrupt and unplanned closure of businesses, schools, and places of worship [11]. These unexpected changes and closures led to stay-at-home orders, creating major shifts in peoples’ daily routines and behaviors such as working remotely and not traveling to visit friends and family, which led to high levels of stress [12]. Furthermore, many people changed regular patterns in food consumption [13] and engaged in less physical activity [14]. Even more, COVID-19 mortality has been higher in African Americans compared to other race and ethnic groups [15]. Research has also found that African Americans struggled to maintain health during the COVID-19 pandemic [16]. For example, African Americans engaged in less exercise [17], experienced an increase in food insecurity [18] which may impact healthful eating patterns [19], and also have had difficulty managing stress and mental health [20, 21].

The Nutritious Eating with Soul (NEW Soul) study began in 2018 and recruited African American adults living in the Southeastern United States to participate in a 2-year dietary lifestyle intervention focused on soul food [22]. Participants were randomized to two groups, a vegan or low-fat omnivorous diet, for 2 years. Before the COVID-19 pandemic closures, participants attended in-person classes learning about nutrition education and receiving social support for following their assigned diet. Closures caused by the COVID-19 pandemic forced intervention delivery from in-person to online.

Adaptations to continue data collection of assessing participants on a delayed schedule were made which made a unique opportunity to conduct a sub-study related to how the pandemic was affecting a population of African American participants’ health related to stress, physical activity, and healthy eating during the early months of the COVID-19 pandemic. The primary research question for this study is as follows: How did the mental and physical health (i.e., stress, control of healthy eating behaviors, self-efficacy for healthy eating, physical activity, weight management) of African Americans participating in a dietary intervention change during the early months following the onset of the COVID-19 pandemic?

Methods

Study Design and Recruitment

Participants in this sub-study came from the NEW Soul study, described in detail previously [22] and described briefly here. The parent study was delivered to two separate 2-year cohorts, separated by 1 year. African American adults (N = 159) were randomized to a vegan or low-fat omnivorous diet and attended classes in a university-based teaching kitchen where they learned nutrition information, observed and participated in cooking demonstrations, and engaged in discussions and activities to facilitate adherence to their assigned diet. Participants completed study assessments at five regular time points throughout the study. The first cohort of the study occurred May 2018–August 2020, and the second cohort of the study occurred June 2019–June 2021.

Intervention delivery transitioned from in-person to online due to COVID-19 closures on March 15, 2020, the day local schools and the university where this study was hosted closed. Briefly, in-person intervention components of nutrition education, facilitated discussion around successes and challenges to healthy eating, and cooking demonstrations were maintained during online intervention sessions. Online intervention sessions were hosted using Zoom.

For this sub-study, data came from eligible participants (n = 150) active in the study on March 15, 2020. All participants provided signed informed consent at baseline to enroll in the study. An amendment was submitted and approved by the University Institutional Review Board to conduct this sub-study.

Data Collection

Table 1 displays the timeline for data collection for the two cohorts. Data were collected as part of already scheduled assessments during study. To assess the association that the COVID-19 pandemic was having on the health of NEW Soul participants, two time points were identified for each cohort to assess changes from before and after the onset of the COVID-19 pandemic. For participants in the first cohort, data collection for their final 24-month assessments was underway prior to March 15, 2020. Fifteen participants had completed their 24-month surveys prior to March 15 and were assigned to a pre-COVID onset group. These 15 participants were sent an additional survey with the same questions (see the “Measures” section) in June 2020 as the regularly scheduled surveys to be included in the post-COVID onset group. Participants received a $25 Amazon e-gift card for completing the additional survey. For participants in cohort 2, the participants who completed their 6-month survey and assessments in November–December 2019 were assigned to a pre-COVID onset group. Twelve-month assessments for cohort 2 were conducted in July 2020, and participants who completed their 12-month survey and assessments were assigned to the post-COVID onset group.

Table 1.

Timeline for data collection pre-COVID and post-COVID onset

	Pre-COVID onset	Post-COVID onset
Cohort 1	March 2020, as part of planned final 24-month assessments	June 2020, additional survey sent to 15 participants
Cohort 2	November–December 2019, part of planned 6-month assessments	July 2020, as part of planned 12-month assessments

Measures

1. Psychosocial outcomes

   Due to the negative impact of the COVID-19 pandemic on stress [12] and eating behaviors [23, 24], stress and eating-related psychosocial measures were selected for this sub-study. Participants’ stress was measured using the 10-item Perceived Stress Scale where participants respond to situations using a 5-point Likert scale (i.e., 0 to 4) [25]. Scores 0–13 correspond to low stress, 14–26 to moderate stress, and 27–40 to high stress.

   Self-efficacy for healthy eating was assessed with a shortened Weight Efficacy Lifestyle Questionnaire [26], where participants responded to ten situations using a 4-point Likert scale (i.e., 1 means not at all confident and 4 means very confident). Higher scores correspond to higher self-efficacy for healthy eating.

   Lastly, cognitive control, disinhibition, and susceptibility to hunger were assessed using the Three-Factor Eating questionnaire [27]. Cognitive control refers to the degree of restraint involved in food consumption and is measured using 21 items. Disinhibition refers to the loss of control of food intake during negative emotional states or palatable foods and is measured using 16 items. Susceptibility to hunger refers to one’s sensitivity to hunger cues and is measured using 14 items. The first 36 items are true or false (true = 1 and false = 0), the next 13 items are measured on a 4-point Likert scale (i.e., 1 means rarely and 4 means very much), and the final question is scored on 6-point Likert scale (i.e., 1 means eat whatever you want, whenever you want it, and 6 means constantly limiting food intake, never “giving in”). Higher scores in each subsection correspond to higher levels of cognitive control, disinhibition, or susceptibility to hunger.

2. Physical activity

   Physical activity was assessed using the International Physical Activity-Short Form (IPAQ-SF), a reliable and widely used measure in intervention studies [28]. Respondents reported physical activity over the past 7 days including the number of days and time spent each day in moderate- and vigorous-intensity physical activity, and walking. Respondents also reported sedentary time with the typical number of hours spent sitting on a weekday. Under normal assessment protocol procedures, a sensor measure of physical activity using an ActiGraph GTM accelerometer was used. Due to social distancing restrictions, the study team requested access to physical activity data from participants who reported wearing a Fitbit before and after March 15. An API database was created to sync participant data from Fitbit to the team.

3. Body mass index

   Height was assessed using a wall-mounted stadiometer during in-person baseline lab assessments for cohort 1 in April and May 2018 and cohort 2 in May and June 2019. Body weight was measured during in-person lab assessments using a calibrated scale. For cohort 1, the most recent weight assessment prior to March 15, 2020, was in March and April 2019 as part of 12-month assessments. For cohort 2, the most recent weight assessment was in November 2019 as part of 6-month assessments. After March 15, 2020, weight was measured remotely in June 2020. Remote weight assessments were completed using the FitIndex Bluetooth Body Fat Scale (www.fit-index.com) which was mailed directly to participants’ (n = 75) homes. Instructions were sent to participants to set up their scale and create an account that linked to the research team for data collection [29].

Data Analysis

First, descriptive statistics summarized study participant demographics and mental and physical health using survey responses for stress, cognitive control, disinhibition, susceptibility to hunger, self-efficacy, and physical activity. Participants who completed weight assessments prior to March 15 were provided with FitIndex Bluetooth Scales to measure weight and assess body mass index. Second, repeated measures analysis of covariance using Full Maximum Likelihood Estimation to handle missing data was used to examine changes in each of the outcomes from the two time points before and after March 15 (one model for each outcome). Full Maximum Likelihood Estimation allows for unbiased estimates under a missing at random assumption when one time point was missing. All models controlled for participants’ age, cohort, sex, education, and class attendance (dichotomous high attendance (> 80% of classes attended from January 1, 2020) or low attendance (< 30% of classes attended since January 1, 2020)). A covariate of cohort was included to account for the different lengths in time for participants in each cohort between completing assessments in the pre-COVID onset and post-COVID onset groups.

Similar to weight and collecting a standardized measure for each participant after March 15, physical activity was to be assessed with a sensor measure; however, an unexpectedly lower number of participants (n = 18) linked their Fitbit to the API database. Of the 18 participants, only 9 had complete data during the desired time. Due to the lower number of participants with complete Fitbit data, these results were not included in the analysis for this study. All data were analyzed using SAS v.9.4.

Results

The demographic characteristics of participants active in the sub-study (n = 150) are presented in Table 2. Participants were mostly female (79%) and African American (100%), and had an average age of 48.2 ± 10.6 years.

Table 2.

Demographics of active participants (N = 150) in the NEW Soul study on March 15, 2020

	NEW Soul study (N = 150)
Characteristic	Total
	Mean ± SD	n	%a
Age, years	48.2 ± 10.6
Sex
Male		32	21
Female		118	79
Diet group
Vegan (intervention)		71	47
Omnivorous (control)		79	53
Attendanceb
High		87	58
Low		63	42
Education
High school or equivalent, some college		38	26
College or advanced degree		112	75
Employment
Employed for wages		112	75
Retired		13	9
Other (self-employed, home maker, student, out of work, unable to work)		25	17
Child (< 18 years old) household makeup
0		60	67
1		20	22
2 +		10	11
Missing		60
Adult (18–64 years old) household makeupc
0		19	19
1		29	28
2 +		54	53
Missing		48
Older adult (65 + years old) household make up
0		63	80
1		13	16
2 +		3	4
Missing		71

^aDue to rounding, percentages may exceed or not sum to 100

^bHigh or low attendance refers to participants’ attendance leading up to the transition of intervention delivery from in-person to online. High attendance is > 80% and low attendance is < 30%

^cDoes not include the participant in the study

Results of changes from pre-COVID onset to post-COVID onset in the psychosocial measures, physical activity, and body mass index are displayed in Table 3. For the Three-Factor Eating Questionnaire, cognitive control increased 1.43 points (F = 20.60, p < 0.0001). Self-efficacy for diet behaviors non-significantly decreased 0.11 points (F = 2.62, p = 0.11). The IPAQ met-minutes per day decreased, but was not significant (F = 0.03, p = 0.86). Body mass index significantly increased 0.72 kg/m² (F = 10.68, p = 0.001).

Table 3.

Pre-COVID onset and post-COVID onset changes in psychosocial and physical health measures

Measure	Possible range of measure	n in models	Pre- LSM ± SE	Post- LSM, SE	Time effect, F	p
Perceived Stress Scale	0 to 40	113	19.65 ± 0.38	19.58 ± 0.43	0.03	0.87
Self-efficacy for Diet Behaviors	0 to 40	113	2.84 ± 0.07	2.73 ± 0.08	2.62	0.11
Three-factor Eating Questionnaire
Cognitive control	0 to 21	113	8.96 ± 0.39	10.39 ± 0.45	20.60	< 0.0001
Disinhibition	0 to 16	113	4.06 ± 0.32	4.61 ± 0.36	3.51	0.06
Susceptibility to hunger	0 to 14	113	3.58 ± 0.27	3.59 ± 0.28	0.00	0.95
Physical activity
IPAQ-SF (MET-minutes per day)	–	113	2636.25 ± 302.10	2575.70 ± 349.16	0.03	0.86
Body mass index (kg/m2)	–	116	36.18 ± 0.77	36.90 ± 0.80	10.68	0.001

LSM, least squares mean; SE, standard error. Each model adjusted for age, cohort, sex, education, and a dichotomous attendance level (high/low). Models accounted for missing data using Full Information Maximum Likelihood Estimation

Discussion

The NEW Soul study had a unique opportunity to assess the mental and physical health associations of COVID-19 on a population of African Americans participating in a dietary lifestyle intervention. The purpose of this study was to describe associations in participants’ stress, control of healthy eating behaviors, self-efficacy, physical activity, and body mass index with the onset of the COVID-19 pandemic.

Findings revealed no time effect regarding participants’ reported levels of stress, eating behaviors, and physical activity. First, stress levels exhibited no change over time after the pandemic began. This finding was very surprising as it does not align with previous research that determined individuals experienced higher levels of stress after the onset of the COVID-19 pandemic [12]. One possible explanation of this unexpected finding is that participants’ pre-COVID stress levels were already at a moderate level with scores between 14 and 26 on the Perceived Stress Scale [25]. Therefore, a ceiling effect may have been observed where participants would be less likely to report even higher levels of stress given reported stress as already at a moderate level prior to March 15, 2020.

Second, regarding healthy eating and weight management, cognitive control and body mass index increased over time; however, no significant time effect was found in participants’ reporting of disinhibition or susceptibility to hunger. Even more, only body mass index changed in the hypothesized direction, which aligns with previous research about risk factors for increased weight gain as a result of COVID-19 [30, 31]. Considering the importance of energy balance in losing or maintaining weight, with many people confined to their home unable to engage in regular outlets of physical activity, it follows that weight loss and management may prove difficult [32].

In addition, we attempted to collect a standard measure of participants’ body weight by sending Bluetooth scales directly to participant homes. This methodology seemed appropriate given a timely report about the use of Bluetooth scales in research [33]. However, only half of study participants elected to receive a scale. Additional analyses to understand factors affecting participant uptake, willingness, and engagement with Bluetooth scales determined that older participants and females were more likely to sign up to receive a scale for remote weight assessment [29].

Participants reported an increase in cognitive control regarding eating behavior. This finding was unexpected given how the pandemic altered daily life for many leading to disordered eating behaviors [34] and higher consumption of unhealthy foods [35]. For example, individuals who were working from home and commuting less may have experienced increased opportunities to engage in snacking and mindless eating. One possible explanation for this unexpected finding may have been that participants spending additional time at home were able to better control food intake than when under normal circumstances of living a busier schedule with more commitments and travel. However, research supporting this idea is limited.

Lastly, although also not significant, physical activity levels trended in the expected direction as participants reported spending less time doing physical activity via the IPAQ-SF and the Fitbit data. COVID-19 restrictions prevented the distribution of accelerometers for a sensor measure of physical activity; however, we attempted to test an innovative way of remotely collecting sensor-measured physical activity through participants’ Fitbit devices. Fewer participants than expected synced their Fitbit accounts to the research database, and even fewer of synced accounts had usable data. Overall, this study’s finding does align with a previous study that identified negative trends in physical activity levels due to COVID-19 [14].

This study was not without limitations. One limitation relates to collecting physical health data of weight and physical activity using standard procedures and equipment. For remote weight collection using Bluetooth scales, even though the same scales were ordered and distributed to participants, measurement consistency in each participants’ weight assessment in their home (i.e., weighing early in the morning, ensuring participants wore light clothing) could not be controlled. Physical activity data were primarily collected using self-report methods which often leads to response bias and over-reporting [36]. Second, this brief, longitudinal study lacked another point in time for data collection following the onset of the COVID-19 pandemic which continues to be an ongoing issue affecting individuals and research projects. Had in-person restrictions continued into fall 2020, at least another weight measurement from the Bluetooth scales would have been collected. For this study, data within a relatively short and early time window after the COVID-19 pandemic began were analyzed. Using data and findings from the overall 2-year NEW Soul intervention may further explain the impact of COVID-19 on participants. Lastly, due to different timelines for data collection between the cohorts, there was substantial loss to follow-up between pre-COVID onset and post-COVID onset groups. These differences in timelines and loss to follow-up may bias results. This limitation of loss to follow-up also substantially underpowered any analyses of differences in findings by cohort or socio-demographics.

Despite these limitations, this study had notable strengths. First, this longitudinal study design allowed for the assessment of relevant health behaviors disrupted by COVID-19 where cross-sectional studies are limited by temporality [37]. Given that the cohorts had staggered pre-post COVID-19 onset data collection time points, analyses controlled for participants’ cohort. This study also incorporated objective measures of weight and used validated surveys to assess stress, self-efficacy, and eating behaviors. Second, this study worked with all African American adults, a population at higher risk of complications from COVID [15, 38].

This study did not identify the hypothesized associations between the onset of the COVID-19 pandemic and health behaviors in African Americans. It is important to note that this work was exploratory during the early months of the COVID-19 pandemic. Given that the effects of the COVID-19 pandemic continue to persist and present a unique set of challenges to researchers and African Americans, further work is needed to understand impacts on psychosocial health and best strategies to positively impact behaviors related to stress, healthy eating, and physical activity in African Americans.

Author Contribution

Gabrielle Turner-McGrievy and John Bernhart obtained funding for the study and contributed to the study conception and design. Material preparation, data collection, and analysis were performed by John Bernhart, Ashley Fellers, Mary Wilson, and Shiba Bailey. The first draft of the manuscript was written by John Bernhart. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National, Heart, Lung, and Blood Institute of the National Institutes of Health under award number R01HL135220. The content is the sole responsibility of the authors and does not necessary represent the official views of the National Institutes of Health. The work was also supported by the Office of Research at the University of South Carolina under award number 115700–20-54014.

Data Availability

The data underlying this article cannot be shared publicly due to protections of participant confidentiality.

Declarations

Ethics Approval

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of the University of South Carolina.

Consent to Participate

Informed consent was obtained from all individual participants included in the study.

Consent for Publication

Informed consent explaining the study team’s intent to publish was obtained from all individual participants included in the study.

Competing Interests

The authors declare no competing interests.