This week, JMIR (Journal of Medical Internet Research) Cardio published our paper ‘Moderation of the Stressor-Strain Process in Interns by Heart Rate Variability Measured With a Wearable and Smartphone App: Within-Subject Design Using Continuous Monitoring‘. In this blogpost, I’ll attempt to break down the paper’s key findings in relatively lay language.
This paper is part of my PhD trajectory, which is titled the Wearable and app-based resilience Modeling in employees (WearMe) study and is supervised by promotors Prof. Dr. Robbert Sanderman (UMCG, UT), Prof. Dr. Cees van der Schans (UMCG, Hanze UAS) and co-promotors Dr. Wim Kamphuis (TNO) and Dr. Hilbrand Oldenhuis (Hanze UAS). The goal of the WearMe study is to explore possibilities to model the mental resilience of employees with a stressful job using wearables and apps. If successful, such insights could be used in the development of automated interventions that give employees personalized and just-in-time feedback on their health and well-being and thus help in the prevention of stress-related problems.
In 2019, we published the protocol for this study, which is also available online. In the protocol, we present a rationale for how the process of resilience may work on a daily basis, based on existing psychological theories and models. For the just published study, we specified the original conceptual model by operationalizing ‘resources’ as ‘Heart Rate Variability’ (HRV) and ‘recovery’ as ‘Total Sleep Time’ (TST), both of which can be measured with wearables. This brings us to the following conceptual model and the four hypotheses that were tested in our latest publication:
Based on population studies, individuals with a low resting HRV tend to be more sensitive in interpreting demands as stressful. Hypothesis 1 therefore expects that demands are associated with stress, and that having a high resting HRV has a buffering effect. Similarly, we expect stress to be related to mental exhaustion, whereas having a high resting HRV has a buffering effect. That second hypothesis is based on population studies that found that individuals with a low resting HRV tend to have less optimal emotion regulation. Hypothesis 3 suggests that stress precedes decreased TST. Finally, hypothesis 4 states that mental exhaustion predicts a decreased HRV the next morning, unless there is sufficient TST to aid the recovery process.
To test these hypotheses, we recruited 26 students in Applied Psychology, Social Work and Physiotherapy that were about to start on their first internship. Throughout a 15 week period, participants wore a Fitbit Charge 2 to measure sleep and physical activity, took a supine 2-minute resting HRV measurement using a Polar H7 chest strap and the Elite HRV smartphone app upon awakening and filled in a short (12 item) evening questionnaire with several stress-related outcomes. This resulted in data of 2,379 unique person-days, of which 1,004 (42.2%) contained complete data.
Since variables such as HRV can differ a lot amongst individuals, it was necessary to standardize ((observation – mean) / standard deviation) the collected data within-person. The analyzed data therefore where not the actually observed values, but values that represent the number of standard deviations that each daily observation differs from the respective individual’s own average value. As a result, there were no between-person differences in the analyzed data (due to the applied standardization technique, the average value of each person would be zero by design). This allowed us to report the results of relatively standard multiple linear regression analysis rather than that of multi-level analyses (which we did as well just in case and gave the exact same results).
The results confirmed our first two hypotheses and partially confirmed hypothesis 4, but not hypothesis 3. When a participant reported a highly demanding day in the evening questionnaire, he/she also tended to reported high stress. This relationship was particularly strong on days where the participant woke up with a low resting HRV, and not as strong when the participant had a relatively high resting HRV. Similarly, stress was associated with mental exhaustion, but having a favorable HRV buffered against that association. Finally, when the participant felt mentally exhausted during the evening, he/she would wake up with a slightly lower resting HRV the next morning, as expected. TST was unrelated to stress and did not buffer against the impact of mental exhaustion on HRV. These aggregated results are visualized in the figure below.
As can be seen, the combination of these results has a cyclical nature. Here, having a decreased resting HRV lowers individuals’ (psycho)physiological resources that normally help them during the appraisal and coping of stressors. This increases the likelihood of becoming mentally exhausted at the end of the day, which could lower the HRV again and thus cause a negative feedback loop. This is consistent with the Conservation Of Resources Theory, which states that an initial loss of resources could cause a loss spiral.
To our knowledge, this is the first study to report these within-day and within-subject relationships. As such, replication of these results in future studies is warranted needed to improve confidence in the presented conclusions. Furthermore, the population of this sample consisted of young and mostly female adults. Therefore, research in different samples is needed to improve the generalizability of these findings.
This study looked at within-day associations between wearable-measured HRV and stress-related questionnaire items. These insights are important to better understand how wearable-measured HRV may be related to these subjective outcomes, the harmful impact of stress is usually the result of more long-term accumulation. Therefore, we also explored the relationships between 5-week trends in resting HRV and the day-to-day fluctuations in resting HRV with changes in stress, somatization, anxiety and depression in a follow-up study of which we will submit a paper for peer-review soon. You can subscribe to this blog via RSS or e-mail (down below) or follow me on Twitter to be notified of future updates.