Purpose: The stress experienced by healthcare learners and practitioners can impact learning, delivery of care, and mental health. Heightened awareness regarding this impact of stress has sparked the growing use of mobile health technologies for real-time tracking of stress in health professions education. The purpose of this study was to evaluate what can accurately be interpreted by the scores generated by mobile technology, regarding psychological stress levels in medical learners.
Methods: In a quasi-experimental within-subjects design, 10 Emergency Medicine residents experienced two rest periods and two stress-inducing simulation scenarios. Heart rate variability (HRV) parameters produced by a mobile HRV application were compared to a reference standard analysis software, and with traditional stress measures (salivary cortisol and self-report measures) using Pearson correlation coefficients. To determine whether HRV parameters from the mobile application differentiate between rest & psychological stress conditions, a Multivariate Analysis of Variance (MANOVA), with condition (rest, stress) as the independent variable, was calculated for the HRV, cortisol and self-report measures.
Results: The mobile application's time-domain HRV parameters correlate strongly with the reference software (r values: 0.93 to 0.99, all p < 0.01), salivary cortisol levels (r = -0.54 to -63, all p <. 0.01) and self-reported stress (r = -0.46 to -0.49, all p < 0.01). These time-domain HRV parameters also accurately differentiated between rest and stress periods (eta2 = 0.43-0.70, all p < 0.01). In contrast, the frequency-domain parameter (LF/HF) of the mobile application showed weaker associations with the reference software (r = 0.10, p = 0.58) and other measures of stress (r = 0.11 to -0.16, NS), and did not differentiate between rest and stress periods (eta2 = 0.07, p = 0.25).
Conclusion: This study provides some validity evidence for the use of a subset of time-domain HRV metrics, captured through a mobile application, for the detection of psychological stress responses in simulated clinical settings. The results also highlight the heterogeneity in HRV metrics produced by various programs. Despite the promise of mobile technologies for the detection of stress in learners and health professionals, further validity research is needed to support their use to detect stress in the health professions.