SMARTWATCH MEASUREMENTS AND CONTROLLED DEPRESSION: COMPARATIVE ANALYSES ON PHYSICAL ACTIVITY AND SLEEP

Background

Digital technology engagement has the potential as an adjunct to supporting depression management by improving physical wellbeing. However, there is a need for long-term (e.g., 12-month) longitudinal studies that characterize the digital phenotype of those who managing their depressive symptoms compared to non-depressed controls.

Methods

Participants in a 12-month decentralized wellbeing study at the Mayo Clinic wore a commodity smartwatch as a digital health technology (DHT) and completed electronic surveys on mental and overall functioning at baseline and on a quarterly basis. Participants included in this study wore the smartwatch for ≥ 50% daily for ≥ 70% of days in the year. The Center for Epidemiological Studies – Depression (CES-D) scale was used to measure depression symptom severity. Individuals with stable, controlled depression were those with baseline CES-D total score < 16 and reported current antidepressant use. Non-depressed controls had a CES-D total score < 16 and no reported antidepressant use. A 4 point or greater decrease in the CES-D total score from baseline to 12-months was considered a clinically meaningful improvement in depressive symptoms. To better understand the associations between perceived smartwatch helpfulness and recorded metrics in controlled depression, an inductive thematic analysis (ITA)-derived perceived helpfulness score was utilized, with higher scores indicating more behavioral change and improved wellbeing themes and lower scores indicating more discomfort and stress themes. Linear mixed-effects models were used to examine differences in daily functioning between those with controlled depression and non-depressed controls, with participants included as random effects and sex, age, marital status, and enrollment duration as covariates. Additional analyses were done to also examine differences in CESD improvement groups and perceived helpfulness groups.

Results

Of 531 participants, 244 (46%) reported either taking antidepressants or no other medication and completed the study survey. Of these, 220 of 244 (90%) met the smartwatch adherence criteria and 110 (45%) were non-depressed at baseline. Among these 110 participants, 55 had controlled depression and 55 were non-depressed controls. Of the controlled depression group, 8 (15%) improved their depressive symptoms at endpoint, compared to 16 (30%) in non-depressed controls. Compared to non-depressed controls, controlled depression had these daily associations: 0.90 less miles walked (p < 0.001), 16 mins less exercise (p < 0.017), 25.6 mins more sleep (p < 0.042), 25 mins more light sleep (p < 0.001), and 6.2 mins more times of sleep interruption (p < 0.038). When both groups improvement in their CESD, non-depressed controls walked significantly more miles per day (2.17, p < 0.002) and vigorously exercised 7.3 mins more per day compared to participants with controlled depression. Finally, for every increase in perceived smartwatch helpfulness, participants with controlled depression vigorously exercised 3 mins more per day.

Conclusions

The digital phenotype of controlled depression is distinct from those who are non-depressed controlled, even when both have comparative lower depressive symptoms and both improve over time. Individualized digital data engagement may promote behavior activation in both groups and future studies should test whether these associations can serve as a scalable guide for improving the physical wellbeing of individuals with controlled depression.

References

Dunn AL, Trivedi MH, Kampert JB, Clark CG, Chambliss HO. Exercise treatment for depression: efficacy and dose response. Am J Prev Med. 2005 Jan;28(1):1-8. doi: 10.1016/j.amepre.2004.09.003. PMID: 15626549. Pearce M, Garcia L, Abbas A, Strain T, Schuch FB, Golubic R, Kelly P, Khan S, Utukuri M, Laird Y, Mok A, Smith A, Tainio M, Brage S, Woodcock J. Association Between Physical Activity and Risk of Depression: A Systematic Review and Meta-analysis. JAMA Psychiatry. 2022 Jun 1;79(6):550-559. doi: 10.1001/jamapsychiatry.2022.0609. PMID: 35416941;