Ambulatory Heart Rate Variability Monitoring in Depressed and Suicidal Adolescents Receiving Intensive Outpatient Treatment: Feasibility Study

Introduction

Suicide risk in adolescence fluctuates over short time scales and is shaped by affective and contextual processes [1]. Continuous naturalistic assessment of psychophysiological markers may improve understanding of proximal risk states and inform timely intervention strategies [2]. High-frequency heart rate variability (HF-HRV; 0.15‐0.40 Hz), reflecting parasympathetic nervous system activity, has been linked to emotion regulation capacity and affective flexibility [3]. Lower resting HF-HRV and altered HF-HRV reactivity have been associated with suicidal ideation and behavior beyond depressive and anxiety symptoms [4-6].

However, most studies linking HF-HRV to suicide risk have relied on laboratory or cross-sectional designs. Ambulatory monitoring may better capture real-world fluctuations in suicidal ideation, which often emerge outside clinical settings and may be influenced by sleep and daily demands [1,7]. Chest-worn electrocardiogram (ECG) sensors provide superior signal quality compared with wrist-based photoplethysmography devices [8], yet no prior studies with adolescents used chest-worn devices for more than one week. Chest-worn devices may pose feasibility challenges including discomfort, visibility under clothing, interference with daily activities, and susceptibility to user error, especially among distressed adolescents. Hence, the long-term feasibility for their use in adolescents receiving intensive psychiatric care remains unclear.

This study evaluated the feasibility of continuous HF-HRV monitoring over 2‐3 weeks using a research-grade chest-worn ECG sensor in adolescents receiving intensive outpatient treatment for depression and suicidality. Our target benchmarks, specified a priori, were ≥75% for both compliance (sensor wear time) and data quality (valid HRV data). We also examined whether feasibility varied over time, by sleep and wake periods, or by school attendance.

Methods

Participants and Procedures

Participants were 121 adolescents (ages 13‐18) enrolled between September 2021 and October 2023 from intensive outpatient and partial hospitalization programs specializing in depression and suicidality. Eligibility required endorsement of suicidal thoughts in the past month per self-report or clinical evaluation.

Following baseline questionnaires and clinical interviews, participants were asked to wear a chest-worn ECG sensor (Movisens ECGMove4; 1024 Hz) on an elastic strap continuously for 14 days, with the option to continue for an additional 7 days. As an alternative, 6% of participants elected to use disposable chest electrodes. Participants were instructed to charge the ECG sensor daily (~10‐15 min) and to remove it for activities that risked damage. Participants received escalating daily compensation ($5‐10/d) contingent on ≥20 hours of wear. ECG data were analyzed in 60-second epochs using automated R-peak detection and artifact handling in DataAnalyzer software (Movisens GmbH, Germany). The software also provided automatic detection of sensor wear (per epoch) based on acceleration data. Compliance was defined as the proportion of intended monitoring minutes during the scheduled monitoring period during which the sensor was worn. Data quality was defined as the proportion of valid HRV values produced by the DataAnalyzer software during periods when the sensor was worn.

Participants also wore research-grade wrist actigraphs (CentrePoint Insight Watch, Ametris, Pensacola FL) to collect daily sleep data. Sleep periods (including naps) were identified using standard algorithms (Cole-Kripke and Tudor-Locke) in ActiLife software version 6.13.5 (Ametris, Pensacola, FL). Actigraphy data were available for 97.7% of ECG measurement minutes (of which 35.2% were during sleep).

Participants also completed morning and evening daily diary surveys via SMS or email links to a secure online platform (n=1485 days [91.8%] completion). Evening diary items included school attendance using predefined categories (in-person, virtual, no school, nonschool day).

Data Analysis

Descriptive statistics summarized feasibility outcomes. Bivariate correlations, χ² tests, and t-tests explored demographic and clinical differences in feasibility metrics. Linear mixed-effects models (restricted maximum likelihood; SPSS version 29.0) examined time (days), sleep vs wake, and school attendance effects on compliance and data quality.

Ethical Considerations

This study was approved by the University of Pittsburgh Institutional Review Board (STUDY20110313). Adolescents provided assent, and parents provided written informed consent.

Results

Demographic and clinical characteristics are shown in Table 1. Of 121 enrolled adolescents, 100 (82.6%) provided ECG data. Among those without ECG data, 9 declined due to discomfort and 12 agreed but never wore the sensor. Participants without ECG data had marginally lower SES (t(119) = 1.96, P=.05) and were more likely to identify as non-White (χ²(1)=5.22, P=.02); they did not differ on age, sex at birth, ethnicity, age at depression onset, level of suicidal ideation, or suicide attempt history (see Table 1).

Overall feasibility benchmarks were met or exceeded (see Table 2). Participants wore the ECG sensor an average of 14.86 days (SD 7.25) and for 76.21% of the intended minutes (SD 21.01). When worn, valid HRV data was obtained 81.42% of the time (SD 17.33). There were no significant differences in compliance or data quality based on demographic or clinical characteristics (all Ps≥.14).

Mixed-effects models indicated that compliance and data quality declined over time (compliance: B=−0.017, SE=0.002; data quality: B=−0.015, SE=0.002; both Ps<.001) and were higher during sleep than wake (compliance: B=0.062, SE=0.007; data quality: B=0.094, SE=0.006; both Ps<.001). There were no significant differences by school attendance (Ps≥.48).

Discussion

Continuous chest-worn ECG monitoring of HF-HRV was feasible in adolescents receiving intensive outpatient treatment for depression and suicidality, producing high data yield across monitoring periods up to three weeks. Feasibility was particularly strong during sleep, consistent with reduced movement and improved electrode contact. Compliance and data quality remained high during school days, suggesting integration of chest-worn monitoring into adolescents’ daily routines is achievable. Declines in feasibility over time likely reflect wear fatigue and chest strap degradation, suggesting an optimal monitoring window of approximately two weeks. Integration with real-time compliance and data quality monitoring (eg, Bluetooth-linked apps) may further improve compliance and data yield.

Ambulatory HF-HRV monitoring via chest-worn ECG is feasible in adolescents at high suicide risk. Potential demographic differences in monitoring participation warrant replication in larger, more diverse adolescent samples. Future studies should also examine feasibility in broader adolescent psychiatric populations and evaluate whether HF-HRV functions as a dynamic marker of short-term suicide risk.