Recovery After Stress—Autonomic and Subjective Arousal in Individuals With Psychosis Compared to Healthy Controls

Abstract

Background and Hypothesis

Heightened stress levels in individuals with psychosis (PSY) are associated with psychotic symptom occurrence and may be partially attributed to well-established deficits in resting-state heart rate variability (HRV) and emotion regulation. In healthy participants, resting-state HRV and self-reported emotion regulation skills have been linked to recovery after a stressor; however, it is unclear whether stress recovery is altered in PSY. Thus, we compared the autonomic and subjective recovery of PSY to healthy controls (HC) and investigated the predictive value of resting-state HRV and emotion regulation skills.

Study Design

We assessed resting-state HRV and self-reported emotion regulation one week prior to a combined physical and cognitive stress induction. After the stress exposure, we assessed the autonomic (decrease in heart rate [HR], increase in HRV) and subjective (decrease in subjective stress and negative affect) recovery in PSY (n = 50) and HC (n = 50) over 60 min.

Study Results

Repeated-measures ANOVA revealed the expected interaction of time × group for subjective stress but not negative affect or autonomic stress. Resting-state HRV predicted recovery of HR, and emotion regulation skills predicted recovery of HRV but not of the other parameters.

Conclusions

Although subjective stress recovery was delayed in PSY, the absence of autonomic recovery deficits suggests that a prolonged stress response may not contribute to heightened stress levels to the expected extent. Improving resting-state HRV and emotion regulation may support autonomic recovery, but further investigation is required to test the impact of such improvements on psychotic symptoms.

Introduction

When a stressor ends, a fast and complete return to initial levels of an organism’s autonomic and subjective arousal (ie, recovery) is considered adaptive functioning.^1,2 In contrast, elevated stress levels (ie, hyperarousal) after stress exposure can interfere with an appropriate reaction to further environmental demands.³ In individuals with psychosis (PSY) autonomic arousal, as evidenced by a higher heart rate (HR^4,5) and lower heart rate variability (HRV^6,7), and subjective arousal, as indicated by elevated levels of self-reported stress and negative affect,^5,8–10 are consistently higher than in healthy controls (HC). This well-documented hyperarousal is regarded as an etiological factor in multiple theoretical models, which posit that psychotic symptoms evolve from dysfunctional stress processing (eg, ¹¹). This link is further supported by numerous studies demonstrating that stress and negative affect precede or intensify psychotic symptoms.^12–17 However, the origins of hyperarousal in PSY and thus the factors that could serve to mitigate it effectively, currently remain unclear. One explanation could be that hyperarousal results from a deficient stress recovery process.

Resting-state HRV and the ability to volitionally and effectively employ emotion regulation skills are both considered essential characteristics for the regulation of autonomic and subjective arousal, enabled via inhibitory influences of the prefrontal cortex on subcortical structures, such as the amygdala.¹⁸ It has been shown that healthy participants with low resting-state HRV demonstrate deficits in cardiovascular recovery,¹⁹ while participants with high resting-state HRV showed fast adaptation of subjective arousal ratings.²⁰ Likewise, the ability to regulate emotions has been associated with a better recovery of both autonomic (state HRV^21,22; HR²³) and subjective arousal,²⁴ whereas deficits in regulating emotions have been related to a prolonged autonomic stress response.²⁵ PSY show deficits in both of these regulatory indices: First, a large effect of reduced resting-state HRV has been found in PSY in contrast to HC.²⁶ Second, PSY have reported to use more emotion regulation strategies that are classified as dysfunctional (eg, suppression of emotions) and fewer emotion regulation strategies classified as functional (eg, reappraising the situation) than HC.²⁷ To sum up, reduced resting-state HRV and deficits in emotion regulation are evident in PSY, which, in turn, may hinder adaptation and lead to deficient stress recovery.

However, until now, few studies have investigated stress recovery processes in PSY. In one such study, patients with schizophrenia showed less recovery of autonomic parameters than HC in the first 7 min after stress induction.²⁸ A second study utilized a similar stressor but added a deep breathing intervention before the recovery period and found autonomic arousal to be less responsive in PSY than in HC.²⁹ A third study used the experience sampling method and documented delayed subjective stress recovery in individuals with early, but not those with chronic psychosis when compared to HC.¹⁰ In conclusion, there is some evidence that alterations in recovery processes exist in PSY. However, the few studies available neither investigated autonomic and subjective arousal simultaneously nor did they examine predictors of recovery, which leaves us with limited insight into possible deficits in stress recovery in PSY.

To address these gaps, we investigated the recovery of autonomic and subjective arousal after stress induction. We hypothesized (I) that PSY would show a reduced autonomic recovery (ie, reduction of HR and increase in state HRV) and subjective recovery (ie, reduction of subjective stress and negative affect) in contrast to HC. We expected these group differences to be evident (a) in the total recovery over the course of one hour after the stressor, (b) in the initial recovery in the first 7 min after the stressor (cf., ²⁸), and (c) in the time of return to the baseline levels. Furthermore, we hypothesized (II) that (a) higher resting-state HRV and (b) the self-reported habitual use of functional emotion regulation (ie, emotion regulation skills) would be predictive of a better recovery.

Methods

Participants

The study was conducted at the Universität Hamburg, Germany. We included individuals with the diagnosis of a psychotic disorder and a group of HC, matched for age, gender, and education level. We recruited participants from in- and outpatient treatment settings and via postings in internet forums.

General inclusion criteria were (1) age 18–65, (2) sufficient German language skills, and (3) the capacity to provide informed consent. General exclusion criteria were (1) acute suicidality, (2) the diagnosis of substance dependence in the last 6 months, (3) any neurological or cardiological diseases, (4) intake of cardiovascular medication, and (5) particular cold sensitivities. For females, additional exclusion criteria due to the assessment of salivary cortisol (reported elsewhere) were (6) oral contraceptive intake or (7) a current pregnancy. During the COVID pandemic, we excluded individuals at high risk of severe outcomes from COVID infection as recommended by national health authorities.

For PSY, additional inclusion criteria were (1) the diagnosis of a psychotic disorder according to the 5th version of the Diagnostic and Statistical Manual of Mental Disorders (DSM³⁰), (2) the experience of paranoid delusions current or in the past (for the analysis of stress-related increases in paranoia, reported elsewhere), and (3) the intake of maximum one antipsychotic (for the analysis of influences on HRV³¹, reported elsewhere). For HC, additional exclusion criteria were (1) a current Axis-I disorder, (2) psychotropic drugs in the past 6 months, and (3) a first-degree relative with a psychotic disorder.

Procedure

The regional ethical review board approved the procedure. The study consisted of a prescreening via telephone, a first assessment day for the diagnostic procedure and baseline measures, and a second assessment day with the stress induction and recovery phases (Figure 1). In the prescreening, potential participants were interviewed regarding the criteria for participation; the absence of acute suicidality and diagnostic criteria were then verified on the first assessment day. Appointments took place between 2 and 6 pm to control for circadian changes.³² Participants were instructed to refrain from alcohol and intense physical training 24 h, and from eating, brushing teeth, caffeine, and nicotine 2 h before participation.

Fig. 1.

Design of the Study. HRV, heart rate variability; HR, heart rate.

First Assessment Day.

All participants provided informed consent and underwent the Structured Clinical Interview for DSM-IV with updated sections for the DSM-5 criteria for psychotic disorders³³ to verify the diagnostic criteria. In PSY, the delusion subscale of the Psychotic Symptom Rating Scales (PSYRATS³⁴) was administered focusing on paranoid delusions. Subsequently, participants filled out questionnaires pertaining to sociodemographic variables and emotion regulation skills and resting-state HRV was assessed (t0).

Second Assessment Day.

Seven days later, participants returned to the laboratory for the stress induction (130 min).

Baseline (t1).

After a short rest phase (5 min), we assessed a baseline of all autonomic and subjective parameters (t1; 15 min).

Stress Induction (t2).

The stress induction (5 min) consisted of a combination of the bilateral foot Cold Pressor Test (CPT) and the Paced Auditory Addition Test (PASAT) with a recording of the verbal responses to increase commitment.³⁵ This combination has been shown to evoke increases in subjective stress, HR, and cortisol and to yield recovery during 60 min.³⁵ It has advantages for a simultaneous assessment of state HRV as participants can remain in a body position that closely resembles the baseline recording position.³² Participants had to place both feet in cold water (2–3°C, constantly moved by a pump) for 3 min. During the CPT and 2 min after, participants had to summate successive numbers that were presented auditorily every 3 s (eg, “5”…“2”). While calculating (5 + 2) and answering (“7”), the last number (ie, 2) had to be kept in working memory to summate it with the next number (eg, “4”) sequentially (2 + 4, answer “6”). After the explanation of the PASAT, a short example audio was presented, and the stress induction did not start until participants calculated the example correctly. During stress induction, participants were instructed to resume, if they removed their feet from the water or stopped calculating. All participants followed the instructions. Directly after, participants completed state measures of subjective stress and negative affect (t2).

Recovery Phase (t3–t6).

In the subsequent recovery phase of 60 min participants could sit still, relax, and read. Throughout this phase, autonomic arousal was measured and participants were repeatedly asked to rate state measures of subjective stress and negative affect.

Completion of the Study.

The recovery phase was followed by a second stress induction with a recovery phase (reported elsewhere). Subsequently, participants were debriefed and received monetary compensation (10€/h).

Stress Recovery Measures

Autonomic Arousal.

We recorded the electrocardiogram with a sampling rate of 256/s using a NeXus-10-Mark-II with the software Biotrace. We processed the raw data with the software Kubios HRV Premium (version 3.3.1; Kubios Oy, Kuopio, Finland). Automatic R-peak detection was manually checked and corrected for artifacts by two independent raters to evaluate interrater reliability, which was very good (intraclass correlation coefficients ≥ 0.95). Where manual correction was not feasible, automatic correction was employed (N_segments = 24). Following the recommendations for the analysis of HRV parameters indicating vagal tone,³² we calculated the root mean square of successive differences (RMSSD) in intervals of 5 min (Figure 1) and applied logarithm transformation (ie, lnHRV) to obtain normal distribution, which was not met beforehand (Supplementary Appendix A1). To check for reliability, we additionally analyzed high-frequency HRV with Fast Fourier Transform (HF-HRV) analog to the RMSSD. We analyzed HR as average beats per minute.

Subjective Arousal.

Subjective stress was measured via the extent the participants agreed to the statement “I feel stressed by the situation” (cf., ³⁶). Since the internal consistency for the four-item scale including ratings of helplessness, controllability, and relaxation (based on ³⁷) was poor (range of Cronbach’s α = 0.45–0.69 for the different measurement points), we used the item of perceived stress due to its more consistent associations with autonomic parameters³⁸ and its sensitivity to change.^39,40 Negative affect was assessed as the mean level of the current experience of fear, sadness, anger, and shame.⁴¹ Cronbach’s α was ≥ 0.87. Both subjective parameters were rated on 11-point Likert scales ranging from “not at all” to “very much”.

Assessment of Recovery Predictors

Resting-state HRV and emotion regulation skills were assessed one week before the stress induction (t0). Resting-state HRV was measured in a 5 min interval after a 5 min resting phase and was logarithm transformed (ie, lnHRV). Emotion regulation skills were operationalized by the sum score of the Emotion Regulation Skills Questionnaire.⁴² The questionnaire assesses the frequency of employing various skills for adaptively dealing with negative emotions (eg, being aware of emotions, modifying emotions, accepting emotions). It comprises 27 items referring to the previous week, which are rated on a five-point Likert scale ranging from “not at all” to “very much”.

Statistical Analysis

We analyzed differences in the recovery between PSY and HC (I) as follows:

• a) for total recovery, we conducted two-way mixed ANOVAs with the respective stress parameter at different measurement points (HR, lnHRV, subjective stress, negative affect) as the within-subject factor and group (PSY vs HC) as the between-subjects factor. Total recovery analyses included t2 (stress induction), t4 (+13 min after stressor), and t6 (+60 min) for autonomic recovery, and t2, t4, t5 (+33 min), and t6 for subjective recovery (Figure 1). Degrees of freedom were corrected by Greenhouse-Geisser, if necessary. Significant effects were followed up by post-hoc tests with Bonferroni adjustments. Since the current phase of the menstrual cycle,⁴³ body mass index (BMI) and physical exercise,⁴⁴ the consumption of nicotine and alcohol,³² and antipsychotics³¹ can affect HRV measurements, we tested these variables as possible covariates for HRV analyses.

• b) for initial recovery, we conducted two-way mixed ANOVAs with HR and subjective stress at t2 (stress induction), t3.1 (+4min), t3.2 (+5min), t3.3 (+6min), and t3.4 (+7min).

• c) to analyze whether PSY and HC differed in the time of return to baseline, we estimated the recovery separately for each participant. Return to baseline was defined as the measurement point when values after the stressor did not differ from the individual’s baseline measure (ie, were smaller than baseline value (t1) + 0.5 SD_{t1_totalsample}). We then tested whether the number of individuals returning to baseline at each measurement point was differently distributed between the two groups by Mann-Whitney-U-tests.

To analyze whether resting-state InHRV and emotion regulation skills predicted recovery (II), we calculated relative changes of HR, state lnHRV, subjective stress, and negative affect during the recovery phase. Relative changes were defined as the absolute change in relation to the respective parameter at the stress induction (ie, (t6 – t2)/t2; cf., ^45,46). Therefore, a negative relative change represents a decrease and a positive relative change an increase in subjective and autonomic stress parameters during the recovery phase. Then, we conducted linear regressions with resting-state InHRV predicting relative changes in HR, subjective stress, and negative affect (IIa) and with emotion regulation skills predicting relative changes in HR, state lnHRV, subjective stress, and negative affect (IIb).

All statistical analyses were conducted with IBM SPSS (version 25; IBM Corp., Armonk, NY).

Results

Sample Characteristics

We pre-screened 255 individuals, of which 131 were invited to participate. The final sample that fulfilled all inclusion criteria and consented to participate consisted of n = 50 PSY and n = 50 HC. In that sample, initial HR recovery data (t3.1-t3.3) was missing for one HC participant and resting-state InHRV at t0 was missing for seven participants (n_PSY = 4, n_HC = 3) due to defective electrocardiogram measurements. For descriptive data and baseline group comparisons, see Table 1 and Supplementary Appendix A2.

Table 1.

Sociodemographic and Clinical Data for Individuals With Psychosis (PSY) and Healthy Controls (HC)

	PSY (n = 50)		HC (n = 50)		Group Comparison
	M (SD)a	Range	M (SD)a	Range
Age	37.88 (10.53)	18–65	38.38 (12.80)	18–65	t(98) = −0.21, p = .832
Gender (female) n	22		27		χ 2 (2) = 1.83, p = .400
Education (years)	16.04 (4.91)	10–41	16.76 (4.54)	9–28	t(98) = −0.76, p = .448
Education (level): high/medium/low n	27/ 16/ 0		33/16/0		U = 1005.50, Z = −0.46, p = .649
History of a mental health disorder n	50		15
Diagnosis of psychotic disorder n	50		0
• Schizophrenia n	30
• Schizoaffective disorder n	16
• Delusional disorder n	2
• Schizophreniform disorder n	1
• Brief psychotic disorder n	1
Current acute paranoid delusion n	24
PSYRATS (sum score delusion)	8.52 (6.92)	0–21
Psychotropic medication n	40
• Antipsychotic medication n	37
• Chlorpromazine equivalent	305.0 (200.8)	20–774
Emotion regulation skills	3.36 (0.66)	1.48–4.89	3.95 (0.45)	2.63–5.00	t(86.44) = −5.17, p < .001
Resting-state lnHRV(t0)	3.39 (0.69)	2.07–4.85	3.49 (0.76)	1.69–5.10	t(91) = −0.68, p = .500
Nicotine consumption (cigarettes/day)	6.80 (9.30)	0–30	2.28 (6.17)	0–35	t(85.12) = 2.86, p = .005
Physical activity (hrs per week)	2.20 (3.18)	0–16	5.04 (4.76)	0–25	t(98) = −3.51, p = .001

Note. PSYRATS = The Psychotic Symptom Rating Scales, subscale delusion. Emotion regulation skills are indicated by the sum score of the emotion regulation skills questionnaire. lnHRV = heart rate variability measured by the root mean square of successive differences (RMSSD, logarithm transformed). ^a or n if indicated. For additional sample characteristics, see Supplementary Appendix.

Manipulation Check

Paired samples t-tests showed HR, t(99) = −9.29, P < 0.001, subjective stress, t(99) = −8.29, P < 0.001, and negative affect, t(99) = −3.86, P < 0.001, were significantly higher at the stress induction (t2) compared to baseline (t1). LnHRV did not significantly differ between t2 and t1, t(99) = −1.14, P = 0.259. This pattern was evident in both groups and initial responses to the stressor (ie, Δt2–t1) did not differ significantly between groups (Supplementary Appendix A3).

Group Comparisons for Autonomic and Subjective Recovery (I)

The trajectories of the stress parameters are depicted in Figure 2 (Supplementary Appendix A4 for values at each measurement point). Results for lnHF-HRV (Supplementary Appendix A5) did not differ from results for lnRMSSD (reported below as lnHRV).

Fig. 2.

Autonomic and Subjective Arousal in Individuals with psychosis (PSY) and Healthy Controls (HC). Error bars depict SD. HRV = heart rate variability measured by the root mean square of successive differences (RMSSD), depicted without logarithm transformation. Subjective parameters can range between 0 and 10. Heart rate is measured in beats per minute.

Test for Potential Covariates.

We found no significant correlations of lnHRV with BMI, consumption of alcohol, phase of menstruation cycle, or chlorpromazine equivalents, but with nicotine consumption and physical exercise. Therefore, we conducted an ANCOVA with the covariates smoking and exercising to test their effect on lnHRV and their interaction with time. Since smoking and exercising differed significantly between groups (Table 1), we used an ANOVA to interpret the main effects of group and time and their interaction, as recommended (cf., ⁴⁷).

Total Recovery (Ia)

Autonomic Recovery (t2, t4, t6).

For HR and lnHRV (Figures 2a and b), we found main effects of time but no main effects of group or interaction effects of time × group (Table 2). Post-hoc analyses revealed that lnHRV increased and HR decreased over time. For lnHRV, an ANCOVA showed a significant main effect of smoking but not of exercising. Interaction effects of time × smoking and time × exercising were not significant.

Table 2.

Results of ANOVAs for Total and Initial Recovery

(Ia) Total recovery		F (df, Error df)	P	Comparison	M	P
HR	- Time	104.67 (1.28, 125.10)	<.001	t2−t4	−9.49	<.001
				t4−t6	−1.57	.001
	- Group	1.72 (1, 98)	.192
	- Time × group	0.27 (1.28, 125.10)	.662
lnHRV	- Time	4.90 (1.78, 174.66)	.011	t2−t4	0.05	>.99
				t4−t6	0.13	.032
	- Group	2.05 (1, 98)	.156
	- Time × group	0.61 (1.78, 174.66)	.526
	- Smoking	8.54 (1, 96)	.004
	- Exercise	2.78 (1, 96)	.099
	- Time × smoking	0.37(1.77, 170.13)	.667
	- Time × exercise	1.21 (1.77, 170.13)	.299
Subjective stress	- Time	61.94 (1.97, 193.08)	<.001	t2–t4	−2.20	<.001
				t4–t5	−0.80	<.001
				t5–t6	−0.04	>.99
	- Group	12.31 (1, 98)	.001	PSY−HC	1.43	.001
	- Time × group	3.65 (1.97, 193.08)	.029	t6−t4 (PSY)	−1.02	.026
				t6−t4 (HC)	−0.66	.371
Negative affect	- Time	19.08 (2.13, 209.15)	<.001	t2−t4	−0.53	<.001
				t4−t5	−0.25	.004
				t5−t6	−0.14	>.99
	- Group	10.41 (1, 98)	.002	PSY−HC	1.23	.002
	- Time × group	1.15 (2.13, 209.15)	.320

(Ib) Initial recovery		F (df, Error df)	P	Comparison	M	P
Initial HR	- Time	86.21 (1.90, 184.36)	<.001	t2−t3.1	−9.70	<.001
				t3.1−t3.2	−0.14	>.99
				t3.2−t3.3	0.14	>.99
				t3.3−t3.4	0.49	>.99
	- Group	2.56 (1, 97)	.113
	- Time × group	0.17 (1.90, 184.36)	.830
Initial subjective stress	- Time	39.05 (2.55, 250.28)	<.001	t2−t3.1	−1.51	<.001
				t3.1−t3.2	−0.19	>.99
				t3.2−t3.3	−0.40	.022
				t3.3−t3.4	−0.16	>.99
	- Group	6.45 (1,98)	.013	PSY−HC	1.25	.013
	- Time × group	3.20 (2.55, 250.28)	.031	t2−t3.1 (PSY)	−0.82	.171
				t2−t3.1 (HC)	−2.20	<.001

Note. lnHRV = heart rate variability measured by the root mean square of successive differences (RMSSD, logarithm transformed). Tests for pairwise comparisons with Bonferroni correction. M = mean differences. All effects including the covariates of smoking and exercising (centered) were tested by an ANCOVA, main effects of group and time and their interaction on lnHRV were tested by an ANOVA.

Subjective Recovery (t2, t4, t5, t6).

For subjective stress (Figure 2c), repeated-measures ANOVA revealed a main effect of time and group, and an interaction effect of time × group (Table 2). Post-hoc analyses showed that subjective stress was significantly higher in PSY than in HC and decreased over time. Between t4 and t6, subjective stress decreased significantly in PSY, but not in HC, who recovered significantly until t5. For negative affect (Figure 2d), repeated-measures ANOVA revealed a main effect of time and group, but no interaction effect. Negative affect was significantly higher in PSY than in HC and decreased over time.

Initial Recovery (Ib)

Autonomic Recovery (t2–t3.4).

For initial recovery of HR (Figure 2a), the main effect of time, but neither the main effect of group nor the interaction effect of time × group were significant (Table 2). Post-hoc analyses showed that HR decreased significantly between t2 and t3.1, but then did not change significantly during the following measurement points (ie, between t3.1 and t3.2 and later).

Subjective Recovery (t2–t3.4).

For subjective stress (Figure 2c), initial recovery analyses revealed a significant main effect of time and group, and an interaction effect of time × group (Table 2). Post-hoc analyses showed that subjective stress was significantly higher in PSY than in HC and decreased over time. HC showed a significant decrease in subjective stress between t2 and t3.1, while in PSY the decrease was later, with a significant change between t2 and t3.2.

Return to Baseline (Ic).

We found no significant group differences in the time of return to baseline (Table 3). For lnHRV, return to baseline was not calculated since there was no significant stress-induced decrease beforehand.

Table 3.

Return to Baseline of Autonomic and Subjective Arousal in Individuals with Psychosis (PSY) and Healthy Controls (HC)

	t3.1	t3.2	t3.3	t3.4	t4	t5	t6	No Return	Group Comparison
HR									U = 1179.00, Z = −0.48, P = .632
PSY (n)	40	1	1	1	2		2	3
HC (n)	41	0	0	1	4		2	1
Subjective stress									U = 1091.00, Z = −1.20, P = .232
PSY (n)	26	1	2	1	3	4	5	8
HC (n)	28	4	3	1	7	2	2	3
Negative affect									U = 1080.00, Z = −1.68, P = .093
PSY (n)					37	1	4	8
HC (n)					43	3	2	2

Note. Return to baseline ≙ value ≤ t1 + 0.5 SD_{t1_totalsample}. HR= heart rate.

Resting-State lnHRV and Emotion Regulation Skills as Predictors of Relative Changes During Recovery (II)

(IIa). Resting-state lnHRV (t0) significantly predicted relative changes during the recovery phase ((t6 − t2)/t2) in HR, β = −0.26, t(92) = −2.52, P = 0.014, that is, a higher resting-state lnHRV predicted a stronger decrease in HR from the stressor to the end of the recovery. Resting-state lnHRV did not significantly predict relative changes in subjective stress, β = 0.00, t(92) = 0.02, P = 0.988, or negative affect, β = 0.01, t(92) = 0.06, P = 0.953.

(IIb). Emotion regulation skills significantly predicted the relative change in lnHRV during recovery ((t6 − t2)/t2), β = 0.23, t(99) = 2.34, P = .021, that is, higher emotion regulation skills at t0 predicted a stronger increase in lnHRV from the stressor to the end of the recovery. Emotion regulation skills did not significantly predict relative changes in HR, β = −0.10, t(99) = −0.95, P = .345, subjective stress, β = −0.08, t(99) = −0.75, P = .453, or negative affect, β = −0.02, t(99) = −0.16, P = .873.

Exploratory Analyses of Disorder Severity

To explore whether recovery from stress differed within PSY based on the sum score of the delusion subscale of the PSYRATS, we compared subgroups created by median split. No significant interaction effects time × subsample emerged. Nonsignificant trends in the total autonomic recovery indicated a tendency of a slower recovery in individuals with high compared to individuals with low PSYRATS scores (HR P = .057, lnHRV P = .067; Supplementary Appendix A6).

Discussion

After a stress induction with a combined physical-cognitive stressor, we compared autonomic and subjective recovery in individuals with psychosis to healthy controls and investigated the predictive value of resting-state HRV and emotion regulation skills on recovery. We found no indication of general recovery deficits in individuals with psychosis. However, evidence for a delayed recovery process for subjective stress emerged, which has also been reported for individuals with early psychosis in daily life.¹⁰ Across the total recovery phase, the lowest level of subjective stress was reached earlier by the healthy controls (ie, at 33 min) than by the individuals with psychosis (ie, at 60 min). Correspondingly, during the initial recovery (in the first 4 min), the healthy controls showed a steeper decline in subjective stress than the individuals with psychosis. A successful recovery is commonly defined as the return to the level of arousal preceding a stressful event, as it should enable the organism to deal effectively with following environmental challenges.^1–3 The majority of participants from both groups reached their baseline for subjective stress at the first measurement point after the stressor (ie, at 4 min), but both groups showed further decreases thereafter. Descriptive group differences of individuals failing to return to baseline during the recovery phase (ie, eight individuals with psychosis and three healthy controls) were not significant. However, given the heightened baseline levels of subjective stress in individuals with psychosis, the return to baseline may not have been sufficient to be considered a full recovery. Future studies need to investigate whether insufficient subjective recovery could give rise to an inadequate response to a subsequent stressor and hence result in sustained, or accumulating, subjective stress.

Contrary to the findings for subjective stress, the assumption of a deficient recovery was not supported by the autonomic parameters. As physiological and subjective stress parameters have been shown to deviate from each other in psychosis research,⁴⁸ our results further emphasize the necessity to incorporate various stress parameters to obtain a comprehensive picture of stress sensitivity and explore divergent patterns in autonomic and subjective parameters. Our findings are contrary to previous studies reporting a deficient recovery of HRV in patients with schizophrenia.^28,29 When interpreting the absence of the expected group differences in autonomic recovery, the rather atypical observation of comparable autonomic stress levels in both groups across the entire testing session is noteworthy. This finding was surprising since a decreased resting-state HRV has been established as a stable finding independent of medication status.²⁶ Nonetheless, antipsychotic medication can impact HRV negatively⁴⁹ and the exclusion of individuals receiving more than one antipsychotic could have resulted in less extreme differences from healthy controls. Furthermore, individuals with a history of a mental disorder or former psychopharmacological treatment were not excluded from the control group and, in contrast to previous studies, groups were matched for educational level. Our results question the existence of general deficits in autonomic regulation in psychosis and confirm the high heterogeneity within individuals with psychosis.²⁶ The relevance of influencing factors for autonomic processes aside the fulfillment of the criteria of a psychotic disorder is supported by the associations of smoking and physical exercise with the global level of autonomic arousal. Furthermore, exploratory analyses indicated a nonsignificant trend toward a slower autonomic recovery in individuals with more severe paranoid delusions, which requires further examination in larger samples. A future objective should be to determine moderating factors between psychosis and autonomic arousal and thereby identify possible starting points for interventions.

Higher resting-state HRV and emotion regulation skills predicted a faster autonomic recovery after stress exposure, further supporting these factors’ roles in the regulation of autonomic arousal. Our findings are in line with the neurovisceral integration model linking emotion regulation to autonomic adaptation¹⁸ and support the notion that emotion regulation skills could be a connecting piece between low resting-state HRV and high-stress levels in psychosis.³⁶ As autonomic arousal is related to the perception of threat⁵⁰ and can precede paranoia,^14,17 a promising perspective is to test whether enhancing autonomic recovery through emotion regulation trainings⁵¹ and HRV-biofeedback⁵² can prevent paranoid symptoms.

Unexpectedly, recovery of negative affect was not related to emotion regulation skills. Notably, although the stress induction was successful, participants’ mean negative affect was low (ie, M = 2 on a scale from 0 to 10) during stress exposure. Therefore, the need to employ strategies to regulate distinct emotions may not have been central. After emotionally more evocative situations, such as social stressors, emotion regulation skills may help to mitigate the elicited negative affect and paranoid thoughts⁵³ while difficulties in emotion regulation may further exacerbate the affective and symptomatic response.^54,55 Thus, an investigation of recovery processes after different stressors is needed.

Some limitations should be noted. Firstly, we did not find a stress-induced reduction in HRV, and thus, we could not determine a return to baseline for HRV. Secondly, the experimenter was not blinded to group membership. Although the stress induction was standardized, an influence of the experimenter’s behavior cannot be ruled out. However, the stressor elicited a similar stress reaction in both groups, which renders this bias unlikely. Thirdly, the assessment of subjective stress may have been less valid, as it was based on one item only. Fourthly, hypotheses and analyses were not preregistered. Finally, as our clinical sample was limited to individuals taking a maximum of one antipsychotic, the results may not generalize to other samples. For instance, individuals receiving multiple antipsychotics have been found to show a more pronounced reduction of HRV⁵⁶ and increased HR⁵⁷ than individuals receiving monotherapy. However, only a third of medicated patients receive antipsychotic monotherapy.⁵⁸

In conclusion, we found no indication of deficits in the autonomic stress recovery process in individuals with psychosis but of delayed recovery in the subjective experience of stress when compared to healthy controls. Noteworthy, we did not generally observe heightened levels of autonomic arousal in individuals with psychosis taking maximum one antipsychotic only, but found associations of HRV with other modifiable factors like smoking and exercising. Thus, the frequently found heightened autonomic arousal may partly trace back to third variables. Our results offer the prospect of decreasing the putative building-up of subjective stress levels by improving subjective recovery and encouraging interventions increasing resting-state HRV and emotion regulation skills, respectively, to enhance autonomic recovery.

Funding

The study was funded by the German Research Foundation, Grant CL 757/1-1.

Conflicts of Interest

The authors declare no conflicts of interest with respect to the authorship or the publication of this article.