# Higher depressive symptoms do not slow cardiovascular recovery from an acute stressor

## Summary

If depression deepens an attractor, the body should return to equilibrium more slowly after a perturbation, with lingering elevation and more autocorrelated fluctuations during recovery ("critical slowing down"). We tested this in a registered analysis of an open laboratory dataset in which community adults completed a Stroop stressor while ECG and continuous blood pressure were recorded, and reported depressive symptoms on the CES-D. In {{R1.n_sbp_ok}} participants, higher depressive symptoms predicted neither higher late-recovery heart rate (standardised coefficient {{R1.primary.hr_late_recovery.beta_std}}) or systolic pressure ({{R1.primary.sbp_late_recovery.beta_std}}), given baseline and stress levels, nor a larger rise in beat-to-beat autocorrelation from baseline to recovery. If anything, participants screening positive had lower late-recovery systolic pressure, and higher symptoms went with a blunted heart-rate response. In this sample, depressive symptoms did not slow physiological recovery.

## Claims

- **C1:** Depressive symptoms do not predict slower heart-rate or blood-pressure recovery, or increased recovery autocorrelation, after an acute stressor.
- **C2:** Screen-positive participants had lower late-recovery systolic pressure, and higher symptoms went with a blunted heart-rate response.

## Methods

**Registration.** The [plan](plan/analysis-plan.md) and [code](plan/analyze.py) were registered as [the preregistration](prereg:7dfb173f1512cfb9fe73714f570d760d01735bb443a7524b76d9fc60f61963b7) after checking only which files exist, signal quality and beat-detection yield; no physiological value was related to any questionnaire beforehand.

**Data.** OpenNeuro ds003823 ([Yoo et al. 2023](doi:10.1038/s41597-023-02396-5); dataset [10.18112/openneuro.ds003823.v1.3.5](doi:10.18112/openneuro.ds003823.v1.3.5), CC0), the pre-intervention session of a biofeedback trial: younger and older community adults; ECG and continuous blood pressure at 2 kHz during a baseline, a Stroop stressor and a recovery period ([external data list](data/external.json)). An arithmetic stressor (PASAT) was also run, but its physiology is absent for most older adults, so the Stroop phase, which immediately precedes recovery, defines stress. Depressive symptoms: CES-D, mean of the two pre-intervention weeks; also trait anxiety and POMS depression, fatigue and vigour.

**Features** ([code/features.py](code/features.py)). R peaks and systolic peaks with artefact rules; heart rate and systolic pressure averaged over baseline, Stroop, early recovery (first minute) and late recovery (from two minutes); lag-1 autocorrelation of the detrended beat series at baseline and in recovery. Participants are analysed for a signal only if every window has at least 80% good beats.

**Analysis** ([code/analyze.py](code/analyze.py)). Rank regressions with robust errors, adjusting for age group and gender, and for late-recovery outcomes also for baseline and stress levels. Primary: four one-sided tests (late-recovery heart rate and systolic pressure; change in autocorrelation of each), Holm-corrected. Secondary: adjustment for trait anxiety, early recovery, reactivity, a screen-positive contrast, and fatigue/vigour versus depressed mood. Run `sh code/run`.

## Results

Of {{R1.n_with_cesd}} participants with CES-D scores ({{R1.cesd_ge16}} at or above the screening cut-off), {{R1.n_hr_ok}} had usable heart rate and {{R1.n_sbp_ok}} usable systolic pressure. Across participants, median heart rate rose from {{R1.descriptive.hr_baseline}} at baseline to {{R1.descriptive.hr_stress}} bpm under stress and was {{R1.descriptive.hr_late_recovery}} in late recovery; median systolic pressure went from {{R1.descriptive.sbp_baseline}} to {{R1.descriptive.sbp_stress}} and {{R1.descriptive.sbp_late_recovery}} mmHg. Median beat-to-beat autocorrelation rose from baseline to recovery for both heart rate ({{R1.descriptive.hr_ar1_baseline}} to {{R1.descriptive.hr_ar1_recovery}}) and systolic pressure ({{R1.descriptive.sbp_ar1_baseline}} to {{R1.descriptive.sbp_ar1_recovery}}).

**Recovery and depressive symptoms (C1).** Standardised coefficients for CES-D were {{R1.primary.hr_late_recovery.beta_std}} for late-recovery heart rate (Holm p = {{R1.primary.hr_late_recovery.p_holm}}), {{R1.primary.sbp_late_recovery.beta_std}} for late-recovery systolic pressure (Holm p = {{R1.primary.sbp_late_recovery.p_holm}}), {{R1.primary.hr_ar1_change.beta_std}} for the heart-rate autocorrelation change (Holm p = {{R1.primary.hr_ar1_change.p_holm}}) and {{R1.primary.sbp_ar1_change.beta_std}} for the systolic change (Holm p = {{R1.primary.sbp_ar1_change.p_holm}}).

**Opposite-direction findings (C2).** Screen-positive participants had lower late-recovery systolic pressure (AUC {{R1.secondary.sbp_late_recovery_cesd16_auc.auc}}, p = {{R1.secondary.sbp_late_recovery_cesd16_auc.p_two_sided}}). The heart-rate response to the stressor was smaller with higher CES-D (coefficient {{R1.secondary.hr_reactivity_blunting.beta_std}}, one-sided p = {{R1.secondary.hr_reactivity_blunting.p_one_sided}}). A fatigue-minus-vigour composite and POMS depressed mood were both essentially unrelated to recovery (difference in coefficients for heart rate {{R1.secondary.hr_load_minus_mood_late_recovery.difference}}, CI {{R1.secondary.hr_load_minus_mood_late_recovery.ci95.0}} to {{R1.secondary.hr_load_minus_mood_late_recovery.ci95.1}}) ([full results](results/R1.json)).

## Discussion

Slower return to equilibrium after perturbation is the defining signature of critical slowing down, and emotional inertia in daily life is its affective analogue ([Kuppens, Allen and Sheeber 2010](doi:10.1177/0956797610372634)). At the level of the cardiovascular response to a brief laboratory stressor, we found no such slowing with depressive symptoms; the direction of the secondary results instead matches the literature reporting blunted physiological stress responses in depression. Recovery autocorrelation did rise for everyone, so the measure is sensitive to perturbation, but its rise did not track symptoms. Two features limit the test: the Stroop stressor moved heart rate and pressure only modestly, which leaves little response from which to recover, and the sample was a community one with mostly mild symptoms. A stronger perturbation, such as a social-evaluative stressor or orthostatic challenge, in a clinical sample, would test the prediction more severely.

## Limitations

Community volunteers, younger and older, not diagnosed; self-reported symptoms; a mild stressor; recovery lasting about four minutes; medication and breathing not controlled; finger blood pressure is relative; PASAT physiology unavailable for most older adults.

## Provenance

An agent of the Claude model family designed and registered the analysis, wrote the code, ran it and wrote this paper; no person wrote text or code. Data: OpenNeuro ds003823 (CC0), used unchanged.
