Smartphones for smarter eating: Elucidating eating behaviors, stress, and heart rate variability
- Author(s): Godfrey, Kathryn
- Advisor(s): Afari, Niloofar
- et al.
Rationale: Binge eating puts individuals at risk for dropout of weight loss treatments and weight regain after treatment. However, treatments for binge eating have not been successful at influencing weight. To improve obesity treatment, research needs to examine binge eating with new theoretical approaches, interdisciplinary paradigms that span physiological, psychological, and behavioral bases, and designs that enable study of eating behaviors within real world settings. The current study examined stress and binge eating, with a design that integrated ecological momentary assessment (EMA) of stress and binge eating behavior with psychophysiological monitoring of the autonomic nervous system (ANS). The ANS is crucial for self-regulation, especially responding to and balance the sympathetic and parasympathetic nervous systems, and low ANS flexibility is related to numerous psychological and physical health stressors. Measures of heart rate variability (HRV) are indicators of ANS flexibility and can be obtained through noninvasive, ambulatory methods. The specific aims were to: 1) examine if lab-based HRV at baseline and when stressed by an experimental protocol is related to binge eating behaviors recalled from the previous four weeks and during a seven day at-home data collection period; 2) analyze if self-reported stress precedes binge eating during a seven day at-home data collection period; and 3) describe the experience of wearing a portable HR monitor and using a smartphone with EMA and the feasibility for clinical use in assessment and intervention.
Design: 32 male and female participants with obesity completed a single lab visit to measure HRV and assess binge eating in the previous four weeks. HRV was measured through a lab protocol containing 5 minute recordings during a baseline period and a mental stressor. A subsample (n=16) of participants also completed a seven day at-home protocol for EMA assessment of stress and binge eating using a smartphone. During the seven days, participants self-reported stress using the 4-item version of the Cohen Perceived Stress Scale before each eating episode and reported their eating behaviors after they finished eating. Participants wore a HR monitor for one day of the seven days of the at-home protocol. At the end of the at-home protocol, these 16 participants underwent a semi-structured interviews and completed self-report questionnaires assessing their experience in the study and exploring the potential feasibility and clinical utility of systems using the study devices. Multiple linear regression, longitudinal multilevel mixed effects models, and qualitative, thematic content analysis were performed.
Results: The sample was comprised of mostly female, non-Hispanic/Latino white or African American single participants, and a range of household incomes. At the first lab visit, many HRV measures (RMSSD, HF, LFn, HFn, LF/HF ratio) were significantly different between the baseline and stressed conditions (p = 0.01 to p < 0.001). Significant relationships were found between HRV variables at baseline and both loss of control (SDNN B = -1.26, p = 0.03, lnHF B = -0.06, p = 0.04) and overeating (LFn B = 0.01, p = 0.04) from the previous four weeks. No significant associations were found for HRV variables under stress, nor among HRV and binge eating behaviors from the at-home portion. Analyses of the at-home EMA data revealed that higher self-reported stress was linked to increased probability of overeating and loss of control overeating (p = 0.011 to p < 0.001) but not of eating non-nutritious, high calorie foods, or breaking dietary rules. Results from adherence data, self-report questionnaires, and semi-structured interview suggest that participants were adherent to study procedures and found them to be straightforward. Participants expressed enthusiasm for elements of the study and for clinical applications of the study system and provided numerous suggestions for improvement.
Conclusions: Findings confirm the link between stress and binge eating behaviors in obesity and provide insights for future research and clinical applications. Measures of ANS flexibility in the lab and increased self-reported stress during the at-home EMA portion were associated with more severe binge eating behaviors. Continuing this line of could inform the development of technologies that detect stress and provide just-in-time adaptive interventions when individuals are at risk for binge eating, improving the capacities and reach of evidence-based interventions for binge eating and weight management.