Physical Activity Has Positive Impact on Mood In a Case Study of Viral Infection and Fatigue

Claire Tait-Doak 2617797

Faculty of Behavioural and Movement Sciences, Vrije Universiteit Amsterdam

P_BSTRHEA Stress & Health

prof. dr. Eco de Geus

Introduction

Now in the latter years of the COVID-19 pandemic, much attention has been paid to the broader impacts of post-viral effects associated with COVID-19 infection, but also a renewed interest in chronic fatigue syndrome following viral infection.

The short- and long-term symptoms of a viral infection have become associated with increased risk of depression symptoms (Raison et. al., 2006) and exercise intolerance (Weldon et al., 2023). Physical exercise is well known for improving mood, and this effect may be even stronger for those with low mood (Lane & Lovejoy, 2001).

This creates a paradoxical problem for patients, where movement may help alleviate depressive symptoms, but too much physical exertion can easily worsen viral fatigue, and may result in relapse of the post-viral condition. Studying illness and physical activity is challenging in lab-based experimentation, but ambulatory measurement can provide valuable insight into the daily realities of patients and how they respond to activity (Burg et al., 2017).

               In the current case study, we aim to measure the bidirectional effect between physical activity and mood in a participant with active influenza infection, and the impact of physical exertion. We hypothesise that a participant with active influenza infection will have a positively increased mood following periods of physical activity as compared to periods of inactivity. Conversely, we hypothesise that periods of more positive mood are associated with subsequent periods of increased activity. In an additional hypothesis, we explore whether periods of excess activity are followed by periods of markedly low activity in a patient with current viral illness.

Methods

Sample & Variables

               The current study (N = 1) concerns a 41 year old female participant with active influenza infection. The participant has a history of COVID-19 infection with cardiac arrythmia and long-term chronic fatigue syndrome, especially following excessive or sustained exertion.

Measurements were taken during days 9-15 of an approximately 40-day illness. Ambulatory data was collected using an Actigraph accelerometer. Mood data was collected through the m-Path smartphone app. The participant notes that she was predominantly confined to bed, but with occasional moments of regular activities including attending lectures, physical exercise, and socialising.

Actigraph

Ambulatory data was collected using an Actigraph wGT3X BT, which is a triaxial accelerometer. Vector Magnitude is a measure of movement intensity, calculated as the square root of the sum of the squares of each of the three axes measured by the device: the horizontal axis (X), the vertical axis (Y), and the depth axis (Z). Measurements are recorded in 60-second intervals. Measurement data from the device were converted into Vector Magnitude (VM) and Steps (S).

The device was attached to an adjustable elastic belt worn around the torso at hip level, with the device at the front left-side. The participant wore the device during the day, except when showering, and removed it at night.

M-Path

Mood and other affect-based data was collected through the m-Path app which the participant used on her own smartphone. Through this app, the participant was asked a series of four questions asking to rate her current Mood (M), stress level (ST), state of focus (F), and sense of being in control (C). Each trait is measured on a scale of 0-100 (see Table 1). Prompts to answer the four questions were sent to the phone approximately every 90 minutes, between the hours of 0800 and 2230 in order to measure the four traits at ten equal intervals during the day. Measures were taken during seven consecutive days during which the accelerometer was also worn, where the aim was to record mood at 70 intervals (or ‘beeps’) over the course of the experiment.

Data Analysis

Data analysis was performed using R Studio RStudio v. 2024.04.2/764. Data from both the m-Path app and the Actigraph device was merged into a single dataset, and non-wear data from the device was removed. Data was sorted by date and time. The dataset consisted of movement data measurements interspersed with recorded mood data at the time of each prompt.

Every m-Path prompt was designated an interval number (interval 1, 2, 3,…n), and all Actigraph data preceding that prompt, but after the previous prompt, was assigned the same interval number. Averages were calculated for past VM (VMP) and past S (SP) for every interval, and averages for each subsequent interval was assigned as future VM (VMF) and future S (SF) for each preceding one. Actigraph data was then removed from the dataset.

Descriptive statistics were calculated for all variables M, F, ST, C, VM, and S. Pearson correlations (r) were calculated for all variables, which also included VMP, SP, VMF, and SF.

Three linear regression models were performed. Two were a bidirectional analysis for our main hypotheses, where in model 1, past activity was the predictor for current mood. In the second model, current mood was the predictor for future activity. The third regression model was used as exploratory analysis for our additional hypothesis, where past activity predicted future activity.

Results

Descriptives

The total number of prompts the participant answered in the m-Path app was 65, of a goal of 70, therefore there was 7.14% missingness in the survey data. The Actigraph data had a missingness of 45.83%, which is approximately what is expected when the device is removed at night. The wear time of the device was therefore ~13 hours per day (54.17%).

In Table 2, the descriptives for all variables are shown: M, ST, F, C, VM, and S. Specifically, M had a range of 3 to 76 (M = 34.80, SD = 18.08). VM had a range of 20.06 to 3229.78 (M = 297.47, SD = 517.85).

Confirmatory Analysis

               In model 1, a linear regression analysis showed that VMP significantly predicted M, F(1, 60) = 11.38, p = .0013, R² = .159, b = 0.0136, SE = 0.0040, t = 3.37 (see Figure 1). In model 2, a second linear regression analysis showed that M significantly predicted VMF, F(1, 60) = 7.35, p = .0088, R² = .109, b = 9.30, SE = 3.43, t = 2.71 (see Figure 2). Together, these findings indicate a bidirectional association between physical activity and mood.

Exploratory Analysis

               In model 3, an additional linear regression analysis showed that periods of high activity significantly predicted higher subsequent VM, F(1, 58) = 25.22, p < .001, R² = .303, b = 1025.27, SE = 204.15, t = 5.02. However, a visual exploration of the data shows distinctly lower activity following a period of very high activity where this effect is not as strong for periods of moderate activity.

Pearson correlations between all study variables are shown in Figure 4. Mood was positively correlated with both activity and control variables, while stress showed a small negative correlation with mood.

Discussion

In our study, we set out to explore the bidirectional relationship between activity and mood in a patient with active viral illness. We found that physical activity was a predictor of increased mood in a 41 year old patient, despite the presence of infection. This explained 16% of the variance in mood, which satisfies our first hypothesis, and is consistent with literature on how exercise affects mood in healthy participants (Lane & Lovejoy, 2001).

We also found that a more positive mood effectively predicted a subsequent increase in activity, accounting for almost 11% of the variation. However, visual inspection of the scatterplot (Figure 2) suggested that this relationship may have been sensitive to outliers. Although this suggests a bidirectional relationship between mood and physical activity, the evidence for confirming our second hypothesis should therefore be interpreted with caution.

Additionally we found that, contrary to our expectations, high levels of activity predicted higher future activity. However, visual exploration of the data (Figure 3) showed a pattern more consistent with reduced activity following periods of very high exertion, suggesting that the regression model may have been influenced by extreme observations. This pattern is consistent with literature on exercise intolerance during viral infection (Weldon et al., 2023).

Limitations

               While the current study has strengths as a field-based ambulatory assessment of real-world interplay between mood, physical activity, and viral illness, it also has several important limitations.

Mainly, our study was a single-case design, and had no baseline comparison, therefore we are unable to draw conclusions this relationship compared to the participant’s mood and activity during normal health.

The study was also affected by a key confounder where both mood and physical activity may have been influenced by fluctuations in illness severity across the measurement period, which was not directly measured in this study.

Future Study

Future study may benefit from a quasi-experimental design where participants may take part in specific activities as experimental conditions, and as a within- or between-subjects design with a larger number of participants with both illness and non-illness conditions. The experiment would also benefit from some more illness specific scales in the m-Path application, such as fatigue and illness severity.

References

Burg, M. M., Schwartz, J. E., Kronish, I. M., Diaz, K. M., Alcantara, C., Duer-Hefele, J., & Davidson, K. W. (2017). Does stress result in you exercising less? Or does exercising result in you being less stressed? Or is it both? Testing the bi-directional stress-exercise association at the group and person (N-of-1) level. Annals of Behavioral Medicine. https://doi.org/10.1007/s12160-017-9902-4

Lane, A. M., & Lovejoy, D. J. (2001). The effects of exercise on mood changes: The moderating effect of depressed mood. Journal of Sports Medicine and Physical Fitness, 41(4), 539–545.

Raison, C. L., Capuron, L., & Miller, A. H. (2006). Cytokines sing the blues: Inflammation and the pathogenesis of depression. Trends in Immunology, 27(1), 24–31. https://doi.org/10.1016/j.it.2005.11.006

Weldon, E. J., Hong, B., Hayashi, J., Goo, C., Carrazana, E., Viereck, J., & Liow, K. (2023). Mechanisms and severity of exercise intolerance following COVID-19 and similar viral infections: A comparative review. Cureus, 15(5), e39722. https://doi.org/10.7759/cureus.39722

Appendix: Tables and Figures