Remotely-administered resilience and self-compassion intervention targeting loneliness and stress in older adults: a single-case experimental design

Abstract

Objectives

Loneliness and chronic stress are prevalent issues for older adults that have been linked to adverse health outcomes. We conducted a remote resilience and self-compassion intervention targeting loneliness and chronic stress.

Methods

This study utilized a multiple-phase-change single-case experimental design with three consecutive 6-week phases: control, intervention, follow-up. Assessments and biomarker collection (blood pressure, inflammation, sleep actigraphy) were conducted at each phase. Participants completed a 6-week remotely-administered resilience and self-compassion intervention using techniques from cognitive behavioral therapy and resilience training. Repeated measures ANOVAs were conducted over the 12-week period from control (week 0) to intervention completion (week 12) and over the 18-week period from control (week 0) to follow-up (week 18) in supplemental analyses.

Results

Participants reported a reduction in stress (p < 0.001; η_p² = 0.15), depression (p =  0.02; η_p² = 0.08), and loneliness (p = 0.003; η_p² = 0.18), and an increase in self-compassion (p  =  0.01; η_p² = 0.13) from control to intervention completion (weeks 0−12). Post-hoc tests revealed that stress reduced significantly during the intervention phase (weeks 6−12) and loneliness reduced significantly during the control phase (weeks 0−6). Some improvements in blood pressure, inflammation, and sleep quality were noted in a subsample of participants.

Conclusion

Findings indicate that our remote resilience and self-compassion intervention for older adults targeting loneliness and chronic stress was efficacious.

Keywords:

• Loneliness
• chronic stress
• wisdom
• mindfulness-based interventions
• positive psychiatry

Introduction

A rise in social and physical isolation was borne by older adults during the COVID-19 pandemic (Peng & Roth, 2022), resulting in a greater risk of experiencing loneliness, chronic stress, anxiety, depression, and poor sleep quality (Caruso Soares et al., 2022; Choi et al., 2022; Sepúlveda-Loyola et al. 2020). Loneliness and chronic stress are significant psychosocial stressors that impact health and well-being across the lifespan. Previous work has tied loneliness and chronic stress to an increased risk of cardiovascular disease, psychiatric disorders, cognitive decline, and mortality (Kiecolt-Glaser & Glaser, 1999; Marin et al., 2011; Quadt et al., 2020; Steptoe & Kivimäki, 2012). Loneliness and chronic stress also affect health by increasing inflammation, further contributing to chronic diseases such as diabetes, cancer, and kidney disease as well as neurodegenerative disorders (Furman et al., 2019; Smith et al., 2020). Various studies have linked loneliness (state and trait) and chronic stress to increased inflammatory biomarkers and pro-inflammatory cytokines (Smith et al., 2020; Vingeliene et al., 2019). Higher levels of inflammatory biomarkers and pro-inflammatory cytokines such as high-sensitivity C-reactive protein (hs-CRP) and Interleuikin-6 (IL-6) have been associated with poorer physical and mental performance as well as aging-related diseases and disabilities in older adults (Puzianowska-Kuźnicka et al. 2016). Although many psychosocial interventions have targeted loneliness and chronic stress in older adults, few have attempted to address loneliness and chronic stress through a remote intervention utilizing positive psychiatry (Gardiner et al., 2018; Gouin et al., 2008; Jeste et al., 2023).

Previous literature has indicated that higher levels of resilience and wisdom in older adults may serve as a protective factor against loneliness and its associated decrements in mental and physical health (Vahia et al., 2020). Wisdom is a complex psychosocial trait that is comprised of several components: self-reflection, prosocial behaviors (such as compassion and empathy), emotional regulation, acceptance of diverse perspectives, decisiveness, spirituality, and social advising (Thomas et al., 2022). There has been some evidence that wisdom and associated psychosocial characteristics can be improved via positive psychiatry interventions (Gorenko et al., 2021; Lee et al., 2020; Treichler et al., 2020). A recent metanalysis of randomized control trials (RCTs) focused on enhancing wisdom and its components found interventions enhancing spirituality, emotional regulation, and prosocial behaviors to be largely effective in improving wellbeing (Lee et al., 2020). Further, a narrative review of positive psychiatry interventions in geriatric mental health found PPIs to be effective at improving overall wellbeing and health outcomes and emphasized the need for more wisdom and resilience interventions tailored for the geriatric population (Lam et al., 2020).

More recent research has invested in identifying the specific components of wisdom that may buffer the relationship between isolation and loneliness. Prior studies have found loneliness to be linked to decreased prosocial behaviors across adults of all age groups, suggesting that improving traits related to altruistic behavior such as compassion could protect against loneliness (Nguyen, 2021; Nguyen et al., 2020). Similarly, a recent qualitative study of individuals living in senior housing communities identified self-compassion and empathy as the key components of wisdom that may aid in coping with loneliness and its correlates (Morlett Paredes et al., 2021). A longitudinal study of adults found higher levels of self-compassion to be a predictor of better mental well-being and loneliness across the lifespan (Lee et al., 2021). Additionally, Ramsey et al. (2023) found self-compassion, rather than compassion towards others, to be associated with better physical health and fewer comorbidities.

To further test these concepts, our team conducted an efficacy and feasibility 18-week pilot of the Technology-Assisted Compassion Training for Seniors (T-ACTS) intervention, a remote wisdom-based positive psychiatry intervention focused on self-compassion, targeting loneliness and chronic stress in older adults (Jeste et al., 2023). The intervention aimed to reduce loneliness and stress by increasing prosocial behaviors and enhancing levels of resilience and self-compassion. Results showed promising intervention adherence and feasibility. Further, participants demonstrated small-to-medium effect size decreases in loneliness and perceived stress, as well as increases in resilience, happiness, and several other aspects of wisdom and positive aging perceptions. However, we were limited by a small sample size (n = 20) and could not make strong conclusions.

The present study aims to expand upon our original findings with a larger sample size (N = 43), including the original 20 participants, and an exploratory focus on biomarkers. Based on our previous work (Jeste et al., 2023), we hypothesized that participants would report a reduction in loneliness and perceived stress (primary outcomes), an increase in resilience, self-compassion, and happiness and a decrease in depression (secondary outcomes). Additionally, we explored the impact of our self-compassion and resilience training on changes in biomarkers, specifically blood pressure (systolic and diastolic), inflammation (high sensitivity C-reactive protein), and actigraphy-derived measures of sleep quality and quantity (total sleep time, sleep efficiency, wake after sleep onset, and intraindividual variability in each of these measures) in a subset of the sample.

Materials and methods

Study design

We used a multiple-phase-change single-case experimental design in which participants served as their own control (Pate & McCambridge, 2021). The study consisted of three consecutive 6-week phases. Phase one was the control period (baseline-week 6), phase two was the intervention period (weeks 6-12), and phase three was the follow-up period (weeks 12-18). Control and follow-up periods were 6 wk long to match the 6-week intervention. Participants had no contact with study staff during control and follow-up periods. Study assessments were conducted prior to the control period (baseline), before intervention (week 6), after intervention (week 12), and after follow-up period (week 18). Participants were transitioned to the next phase once they had completed the corresponding study assessment. This study design allows for participants to serve as their own control group. Further, post-hoc analyses using this study design provide investigators an understanding of where the largest changes occurred (e.g. during the control phase, intervention phase, follow-up phase). This contextual information is helpful as some outcomes may change due to a placebo effect (i.e. related to meeting study personnel and signing up for a psychological intervention), while other outcomes may revert to baseline values after the conclusion of the intervention if the intervention is not efficacious. A subset of participants (n = 13) additionally underwent biomarker collection at each study assessment.

The study protocol was approved by the University of California, San Diego (UCSD) Human Research Protections Program and Institutional Review Board (IRB), and all participants provided informed consent before participation.

Participants

Participants were recruited from the greater San Diego area. Recruitment flyers were posted in campus facilities, advertised in campus newsletters, and posted on social media sites. Independent retirement communities and assisted living facilities were targeted through recruitment presentations. Upon reaching or being contacted by a potential participant, recruitment staff introduced the project and gauged interest in study participation. Those who expressed interest were screened over the phone by recruitment staff. Inclusion criteria were age ≥65 years, fluency in English, and ability to provide informed consent. Additional screening criteria were scores >28 on the Modified Telephone Interview for Cognitive Status (TICS-m) and >36 on the UCLA Loneliness Scale—Third Edition (UCLA-LS). This was done to ensure that our participants were cognitively healthy and experienced some level of loneliness. Potential participants were excluded if they had a diagnosis of dementia or a serious mental illness (e.g. schizophrenia-spectrum disorders, bipolar disorder). Twelve participants were excluded after the screening process, 11 for not meeting the required UCLA-LS score and one for not meeting the age requirement.

Of the 51 participants initially enrolled in the study, 2 dropped out before baseline, 4 during the control period, and 2 after the first intervention session, resulting in a 4% attrition rate in participants that attended the first session. Of the 8 participants that dropped from the study, 7 voluntarily withdrew due to a loss of interest in the study and 1 was withdrawn by the Principal Investigator due to aggressive behavior with study staff. Reasons provided for loss of interest included the following: scheduling conflicts, unanticipated life events such as injuries, and finding the study assessments to be tedious. A subset of 13 participants agreed to provide biomarker samples.

Due to staffing changes and loss of funding, the present study was closed before completion, and six participants were unable to be seen for their 18-week follow-up study assessment. To maintain maximum statistical power, we included weeks 0-12 for the primary Results, but provide weeks 0-18 in the supplemental material (See Supplementary Table 1).

Remote assessment and intervention

Study assessments and intervention sessions were conducted via Zoom, a video-conferencing platform, on the participants’ personal laptop or tablet. HIPAA-compliant Zoom Pro Licenses provided by UCSD Health were utilized in order to safeguard protected health information. Participants without access to a suitable device were provided an Apple 7^th Generation iPad with internet access by study staff to use for the duration of the study.

Feasibility and acceptability

Feasibility and acceptability were assessed through intervention adherence rate and home exercise completion rate.

Measures and procedures

Sociodemographic characteristics

Age, sex, education, race/ethnicity, and marital status were collected at the baseline visit.

Outcome measures

The primary outcomes for this study were loneliness and chronic stress. Loneliness was measured using the UCLA Loneliness Scale—Third Edition (UCLA-LS). The UCLA-LS is a highly reliable 20-item self-report questionnaire that assesses an individual’s subjective feelings of loneliness (Russell, 1996). Items are rated on a 4-point Likert scaled of 1 (never) to 4 (always). Higher scores indicate higher levels of loneliness. Stress was measured using the Perceived Stress Scale (PSS-10) which is a validated 10-item self-report questionnaire that measures the degree of which individuals perceive their lives to be stressful. Items are rated on a 5-point Likert scale of 1 (never) to 5 (often) with higher scores indicating higher stress (Cohen et al., 1983; Taylor, 2015).

Resilience, self-compassion, depression, and happiness were secondary outcomes. Resilience was measured using The Connor-Davidson Resilience Scale (CD-RISC), a validated 10-item self-rated scale that assesses the ability to cope with adversity (Campbell-Sills & Stein, 2007; Connor & Davidson, 2003). Items are rated on a 5-point Likert scale from 0 (not true at all) to 4 (true nearly all the time), with higher scores indicating greater resilience. The Neff Self-Compassion Scale—Short Form (SCS-SF) was used to assess self-compassion (Neff, 2003). SCS-SF is a 12-item self-reported scale that measures the capacity to extend compassion to oneself in instances of inadequacy or suffering. It is rated on a 5-point Likert scale of 1 (almost never) to 5 (almost always) with higher scores indicating greater self-compassion. The SCS-SF has been validated as a measure of overall compassion (Raes et al., 2011). Depression and happiness were measured using the 8-item Center for Epidemiological Studies Depression Scale (CES-D-8) and 4-item CESD Happiness Subscale (CESD-HS), respectively (Joseph, 2007; Karim et al., 2015). The CESD-8 is an 8-item self-rated scale of depression and has been validated as a reliable measure of depression in older adults. The CESD-HS is a 4-item subscale of the CES-D that measures the presence of happiness rather than the absence of depression. Both scales are rated on a 4-point Likert scale from 0 (rarely or none of the time) to 3 (most or all of the time) with higher scores indicating higher depression or happiness.

Additionally, we included the Duke Social Support Index-Social Interaction Subscale (DSSI-SI) and 36-item Short Form Health Survey (MOS-36) to characterize our sample in terms of social interaction and overall wellbeing (Koenig et al., 1993; Ware & Sherbourne, 1992). The DSSI-SI is a 4-item subscale of the Duke Social Support Index that assesses levels of social interaction with higher scores indicating more frequent social interaction. The MOS-36 is a 36-item self-rated scale of functional health and wellbeing. The scale contains a combination of Likert scales and yes/no questions with higher scores indicating better health status for the overall scale and subscales.

Exploratory biomarkers

Due to staffing and funding limitations, only a small subsample completed biomarker collection (n = 13). Biomarker collection was added to the protocol later in the study in response to receiving additional funding. Those who provided biomarkers were enrolled after the addition of this study component, and no specific considerations or criteria were assessed to be eligible for this subset. Blood samples were collected to assay high-sensitivity C-reactive protein (hs-CRP) to assess levels of inflammation. Height, weight, and blood pressure were recorded at each blood draw using a standardized scale and blood pressure monitor.

Additionally, study staff provided participants with a Fitbit Charge 5 activity-tracking watch to wear for a 7-day period following the blood draw to assess sleep quality, sleep quantity, and intraindividual variability in sleep measures. Devices were redeployed if the device was worn for less than four nights. Sleep variables assessed were total sleep time (TST), sleep efficiency (SE), and wake after sleep onset (WASO). Sleep efficiency refers to the percentage of time in bed spent asleep, and wake after sleep onset refers to the total duration of awakenings after initial sleep onset. Mean sleep variables represent the mean sleep value over the week-long wear period, and variability variables refer to the standard deviation value over the wear period.

Intervention

We administered the Technology-Assisted Compassion Training for Seniors (T-ACTS), a manualized positive psychiatry intervention developed for a previous pilot study (Jeste et al., 2023). The intervention is highly individualized and customizable, consistently being updated in response to participant feedback during the initial pilot. The intervention aimed to enhance a specific component of wisdom, self-compassion, and resilience to promote well-being and reduce loneliness and chronic stress in older adults, using techniques based in Cognitive Behavioral Therapy (CBT) and resilience training. The T-ACTS intervention addresses resilience and self-compassion within three components of loneliness: cognitive (education about social relationships, loneliness, and compassion), affective (mindfulness exercises, mediation, self-compassion exercises, increased tolerance for stress), and behavioral (savoring and gratitude practices, social skills training, engagement in value-based activities).

During the 6-week intervention, participants attended weekly 60-minute one-on-one virtual sessions with a trained facilitator (i.e. bachelors- and masters-level study personnel supervised by a protocol-trained and state-licensed social worker). Facilitators assisted participants in setting individualized short-term goals to improve wellbeing and led them through concrete values-driven activities to reach them. Sessions were both educational and interactive, incorporating CBT-based homework assignments and exercises such as deep breathing exercises, mindfulness-based meditation, and daily gratitude journaling. Gratitude journaling was assigned to be completed daily, and CBT-based homework were assigned weekly. Deep breathing exercises and mindfulness-based meditations were provided as tools to use as needed throughout the intervention (e.g. in times of stress).

Statistical analysis

Descriptive statistics were provided for study demographics, feasibility, and adherence outcomes. Repeated measures ANOVAs were used to compare changes in study outcomes over three time periods: baseline (week 0), start of the intervention (week 6), and end of the intervention (week 12). Supplemental analyses with a smaller sample size included the end of a 6-week follow-up period as well (week 18). All repeated measures ANOVAs were tested for sphericity using Mauchly’s Test of Sphericity, and Greenhouse-Geisser corrections were applied to any that violated the assumption. If the repeated measures ANOVAs were found to be statistically significant, post-hoc Bonferroni-adjusted pairwise comparisons were completed to discern which phases were significantly different from each other. Given that three post-hoc tests were completed within each repeated measures ANOVA (weeks 0 vs 12, 0 vs 6, 6 vs 12), an unadjusted pvalue lower than 0.0167 was necessary to be considered statistically significant. Significance level (α) for Bonferroni-adjusted pvalue was set at 0.05 and Bonferroni-adjusted pvalue are provided in the Results. Partial eta squared was used as the measure of effect size. We defined a small effect as η_p² ≥ 0.01, a medium effect as η_p² ≥ 0.06, and large effect size as η_p² ≥ 0.14 (Cohen, 1988). Given the small sample size of our exploratory biomarker analysis, we provide means, standard deviations, and supplemental line charts. Analyses were conducted in in SPSS Version 28 (IBM Corp., Armonk, N.Y., USA).

Results

Sociodemographic characteristics

The sample included 43 participants with a mean age of 76.9 (SD = 7.5). The sample was 77% percent female and 91% non-Latinx White. Additionally, 93% of participants received some college education, and 30% were currently married. Participants had mean TICS-m scores of 36.1 (SD = 4.5) at screening and mean UCLA-LS loneliness scores of 43.0 (SD = 10.2) at baseline (Table 1).

Table 1. Sample demographics.

	N	Minimim	Maximum	Mean or %	Std. Dev.
Age (years)	43	66.8	93.4	76.9	7.5
Gender
Female	33			76.7%
Male	10			22.3%
Race
Caucasian	39			90.7%
Non-Caucasian	4			9.3%
Marital Status
Currently Married	13			30.2%
Currently Single	30			69.8%
Education
High School and Below	3			7.0%
Some College to Bachelor Degree	26			60.4%
Post-Graduate Degree	14			32.6%
TICS-m (at screening)	43	28.0	42.0	36.1	4.5
UCLA-LS (at baseline)	41	24.0	73.0	43.0	10.2

Note: TICS-m: Modified Telephone Interview for Cognitive Status; UCLA-LS: UCLA Loneliness Scale—Third Edition.

Feasibility and adherence

We obtained intervention adherence data from 100% of participants and homework completion data from 95% of participants. The intervention yielded a 96% adherence rate, meaning that 43 of 45 participants who attended the first session completed all six sessions. Of the 41 participants with homework completion data, 95% completed deep breathing exercises, 87% completed CBT-based homework, 82% completed daily gratitude journaling, and 68% completed self-guided meditations at least once during the intervention period. On average, participants completed 80% of assigned weekly CBT-based homework and 57% assigned daily gratitude journal entries. From the tools that were intended to be implemented as needed, participants completed breathing exercises 62% of the time (approximately 17 of the possible 28 days) and meditations 35% of the time (approximately 10 of the possible 28 days).

Outcomes

Primary outcomes

Over the 12-week period from baseline (week 0) to post-intervention (week 12), there was a large effect size decrease in perceived stress (F(2,80) = 7.26, p = 0.001, η_p² = 0.15) as well as a large effect size decrease in loneliness (F(2,80) = 8.89, p < 0.001, η_p² = 0.18).

Secondary outcomes

Over the 12-week period, there was a medium effect size decrease in depression (F(2,82) = 3.62, p = 0.03, η_p² = 0.08) and a medium effect size increase in self-compassion (F(2,78) = 5.57, p = 0.005, η_p² = 0.13). No other changes in outcome measures reached statistical significance.

Post-hoc analyses

Post-hoc pairwise comparisons revealed a significant decrease in perceived stress during the intervention phase (week 12 vs week 6; M_difference = −1.59, p = 0.04, 95% CI [-2.14, −0.03]) and from baseline to post-intervention (week 12 vs week 0; M_difference = −2.22, p < 0.001, 95% CI [−3.56, −0.88]), but not during the control phase (week 6 vs week 0). There was a significant decrease in loneliness during the control phase (week 6 vs week 0; M_difference = −2.44, p = 0.003, 95% CI [−4.14, −0.74]) and from baseline to after the intervention (week 12 vs week 0; M_difference = −2.95, p = 0.003, 95% CI [−5.04, −0.86]), but not during the intervention phase. There was also a significant reduction in depression ratings from baseline to post-intervention (week 12 vs week 0; M_difference = −1.79, p = 0.02, 95% CI [-3.37, −0.20]), but not during the control or intervention phases. Similarly, self-compassion significantly increased from baseline to post-intervention (week 12 vs week 0; M_difference = 2.65, p = 0.01, 95% CI [0.52, 4.78]), but changes did not reach significance during the control or intervention phases.

Exploratory outcomes

Over the 12-week period, there were decreases in systolic blood pressure (M₀ = 134.36, M₆ = 133.64, M₁₂ = 125.55) and hs-CRP (M₀ = 2.50, M₆ = 1.96, M₁₂ = 1.77). There was also an increase in diastolic blood pressure (M₀ = 72.27, M₆ = 71.55, M₁₂ = 79.09). In regard to sleep variables, there was an increase in mean sleep efficiency (M₀ = 86.74, M₆ = 87.84, M₁₂ = 92.08). Additionally, there were decreases in mean WASO (M₀ = 67.48, M₆ = 70.15, M₁₂ = 39.16), sleep efficiency variability (M₀ = 5.43, M₆ = 7.85, M₁₂ = 4.48), and WASO variability (M₀ = 30.68, M₆ = 59.16, M₁₂ = 27.59; Table 2). Due to limited sample size, repeated measures were not run for exploratory analyses, but line graphs were created to visualize individual changes over the 12-week period (See Supplemental Figures 1a-1i).

Table 2. Comparison of scores before control period (week 0), before intervention (week 6), and after intervention (week 12).

	N	Week 0		Week 6		Week 12
		(start of Control Period)		(end of Control Period/start of intervention)		(end of intervention)
		Mean	SD	Mean	SD	Mean	SD	F	df	p	ηp^2
Primary Outcomes
Perceived Stress: PSS-10	41	15.07	5.93	14.44	5.50	12.85	5.58	7.26	2	0.001	0.154
Loneliness: UCLA-LS	41	42.76	10.23	40.32	9.82	39.80	9.75	8.89	2	<0.001	0.182
Secondary Outcomes
Resilience: CD-RISC	43	27.79	4.53	28.21	5.05	29.07	4.39	2.42	2	0.10	0.054
Depression: CES-D-8	42	8.64	5.03	7.83	5.18	6.86	5.02	3.62	2	0.03	0.081
Happiness: CESD-HS	42	8.67	2.36	8.69	2.64	9.50	2.55	2.13	2	0.13	0.049
Self-Compassion: SCS-SF	40	41.47	8.66	42.62	9.39	44.13	8.76	5.57	2	0.005	0.125
Additional Measures
Social Support: DSSI-SI	40	8.57	1.50	8.55	1.38	8.42	1.47	0.27	1.7	0.73	0.007
Mental Health: SF-36 MCS	41	47.65	9.22	49.34	9.30	50.27	9.12	2.07	2	0.13	0.049
Physical Health: SF-36 PCS	41	45.42	9.74	46.08	10.36	46.29	9.39	0.38	2	0.68	0.009
Physiological Measures
BMI	12	28.77	6.58	28.56	6.14	28.59	5.89
Systolic Blood Pressure (mmHg)	11	134.36	17.28	133.64	9.18	125.55	19.36
Diastolic Blood Pressure (mmHg)	11	72.27	8.97	71.55	14.07	79.09	6.11
Hs-CRP (mg/L)	10	2.50	3.21	1.96	2.17	1.77	2.24
Sleep Measures
Mean Total Sleep Time (min)	9	431.51	91.87	447.88	92.74	439.70	124.82
Mean Sleep Efficiency (%)	9	86.74	8.40	87.84	11.21	92.08	2.30
Mean Wake After Sleep Onset (min)	9	67.48	48.07	70.15	64.07	39.16	19.07
Total Sleep Time SD (min)	9	66.79	31.61	64.97	16.47	71.41	23.51
Sleep Efficiency SD (%)	9	5.43	4.55	7.85	6.39	4.48	3.27
Wake After Sleep Onset SD (min)	9	30.68	23.17	59.16	52.19	27.59	28.93

Note: CES-D-8: Center for Epidemiological Studies Depression Scale; CESD-HS: Center for Epidemiological Studies Depression Scale - Happiness Subscale; CD-RISC: Connor-Davidson Resilience Scale; SCS-SF: Neff Self-Compassion Scale—Short Form; PSS-10: Perceived Stress Scale; SD-WISE: San Diego Wisdom Scale; DSSI-SI: Duke Social Support Index—Social Interaction Subscale; SF36-MCS: Short Form 36 Health Survey Questionnaire – Mental Component Score; SF36-PCS: Short Form 36 Health Survey Questionnaire – Physical Component Score; UCLA-LS: UCLA Loneliness Scale—Third Edition; SD: Standard Deviation.

Discussion

The present study is a follow-up trial of a previously piloted resilience and self-compassion-based intervention targeting loneliness and chronic stress (Jeste et al., 2023). We build upon our previous work by including an expanded sample of older adults (double the original size, including the original 20) and providing preliminary findings on biomarkers. Feasibility and adherence were consistent with the previous trial, reflecting low attrition (4%) and high intervention session adherence, demonstrating high practicability and acceptance by community-dwelling older adults. Home exercise participation differed by task, with the majority of participants completing all assigned home exercises (CBT-based homework and daily gratitude journaling) but fewer employing the ‘as needed’ tools of self-guided meditation and breathing exercises. The intervention was designed to be flexible and individualized. While this was likely appreciated by the participants with busy and demanding lives outside of the study, freedom to choose study exercises may have lessened the potency of the intervention. It is possible that more precise goals regarding the frequency of and adherence to home exercises would have improved efficacy, but could have also led to a higher number of dropouts. Future work should continue to determine how home exercises can be implemented in remote interventions without being too burdensome on participants’ time.

Changes in perceived stress and loneliness in the current sample differed from the earlier pilot. Jeste et al. (2023) found small-to-medium effect size decreases in loneliness and perceived stress in both the control and intervention period. The present sample demonstrated significant decreases in both perceived stress and loneliness over the entirety of the 12-week period. Reductions in perceived stress were more pronounced over the intervention period but not control period, indicating that the intervention was efficacious at reducing stress in comparison to treatment as usual. However, reductions in loneliness were significant over the control period but not the intervention. This may be due to an anticipatory effect where the expectation of treatment results in behavior changes related to the treatment target (Malani & Reif, 2015). Participants simply knowing that they will be receiving treatment to reduce their loneliness could have led them to implement prosocial behaviors preemptively, such as growing or maintaining their existing social networks or engaging in social activities. Past research has also found anticipation of positive events to be related to well-being, implying that when participants look forward to receiving the intervention, it may result in an increase in their subjective well-being, reducing loneliness in the process (Luo et al., 2017; Macleod & Conway, 2005).

Further, the current study found a significant decrease in depression and increase in self-compassion over the 12-week period, but changes in ratings did not reach statistical significance in the control or intervention periods individually. In other words, depression and self-compassion changed over each of the phases, but only the difference from week 0 to week 12 was large enough to reach statistical significance. Clearly, a larger sample size and a control group are needed for increased statistical power and to distinguish the effects of the intervention from anticipation alone. Additionally, changes in resilience over the 12-week period did not reach statistical significance as they did in the previous pilot, potentially suggesting that self-compassion may be the more prominent mechanism at work (Jeste et al., 2023). These results are consistent with prior research suggesting enhancement of wisdom and its components, specifically self-compassion, may improve depression and increase wellbeing (Al-Refae et al., 2021; Conversano et al., 2020; Kadri et al., 2022; Willard et al., 2022). Similarly, a recent study by Al-Refae et al. (2021) piloting a self-compassion- and mindfulness-based mobile intervention found that participants that completed the intervention had significantly greater improvements in self-compassion and depression than controls. In addition to reducing negative mental health conditions like depression and chronic stress, there is further clinical importance in increasing positive psychosocial factors such as self-compassion, resilience, and wisdom. There has been a surge in recent research exploring positive psychiatry which suggests that improving these psychosocial factors, rather than targeting mental illness individually, may be effective at enhancing both overall wellbeing and measurable outcomes like morbidity and longevity (Conversano et al., 2020; Jeste & Palmer, 2015).

Objective sleep values and blood-based biomarkers were collected in a subset of participants to explore the potential impact of the intervention on biological correlates of loneliness and chronic stress. Results demonstrated a preliminary reduction in systolic blood pressure and an increase in diastolic blood pressure. There was also a reduction in inflammation measured via hs-CRP which has been linked to loneliness and chronic stress in past literature (Van Bogart et al., 2021). Additionally, participants’ sleep appeared to become more efficient and less variable over the 12-week period which is consistent with previous research associating loneliness and chronic stress to worse and more fragmented sleep (Griffin et al., 2020; Shankar, 2020). These preliminary findings suggest that positive psychiatry interventions may alter both mental and physiological health in older adults, but a larger sample is clearly needed to better understand these relationships.

The current study contributes to a large body of literature providing support for the usefulness and efficacy of wisdom- and mindfulness-based interventions in reducing loneliness and improving overall wellbeing. Among the literature, interventions found to be efficacious most commonly utilized CBT-based therapeutic modalities (Gorenko et al., 2021). Participants in our intervention learned CBT-informed thought-challenging strategies, enabling them to enhance social skills and challenge negative appraisals about themselves and others, resulting in decreases in loneliness and stress. Further, our intervention curriculum contained aspects of positive psychiatry and supportive therapy including mindfulness-based emotional regulation techniques (i.e. breathing exercises and mediation) and value-driven goal-setting to increase resilience and self-efficacy which have been similarly successful in prior interventions (Fu et al., 2022). Our intervention also implemented behavioral strategies to increase prosocial behaviors and enhance social interactions which have yielded positive results in similar loneliness interventions (Freedman & Nicolle, 2020); Gardiner et al., 2018)

While the specific treatment procedures in the literature are fairly heterogenous, the present intervention contains key components that have been consistently effective throughout. Gardiner et al. (2018) states one common theme among successful interventions is adaptability, meaning flexibility in modality (remote administration, accessibility, etc) and in content in which receivers of the intervention are able to have some level of control over intervention activities. Moreover, a recent systematic review of loneliness interventions emphasizes the importance of the ability to tailor interventions to meet individual needs as the experience of loneliness is not one-size-fits-all (Fakoya et al., 2020). The T-ACTS intervention excels in both of these areas as it is remotely administered and highly accessible. Further, the one-on-one sessions are easily adaptable and collaborative, focused on creating strategies to meet the participant’s individual goals. During the COVID-19 pandemic, means of remote treatment approaches became increasingly necessary and more frequently utilized. Recent literature suggests that remotely-delivered interventions for loneliness are highly efficacious and superior to brief intervention or care as usual (Fu et al., 2022). The remote modality of our piloted intervention contributes to its feasibility and ease of dissemination.

The present study has several limitations. The sample was homogenous, consisting mostly of non-Latinx White, educated female participants which limits generalizability to older adults of differing racial and ethnic groups. We also did not implement randomization or a comparison group, but instead utilized a control period prior to the intervention to ensure changes could be attributed to the intervention. Further, we excluded the follow-up phase (week 12 to week 18) from primary analyses to maintain maximum statistical power, limiting our ability to evaluate long-term intervention outcomes. Supplemental analyses with the smaller sample size (n = 36) found significant reductions in stress and loneliness as well as a significant increase in resilience over the 18-week period, but changes in self-compassion and depression did not reach significance as they did in the primary Results (Supplementary Table 1). Further study is required to better understand the lasting effects of the intervention.

Moreover, the way in which home exercises were assigned limits the ability to examine intervention fidelity. Some exercises were not assigned, but rather provided as optional tools to use as needed. As a result, participants could choose to not complete certain exercises if they found them unhelpful, making it difficult to ascertain whether the intervention was followed as precisely as intended. Future intervention trials seeking to measure adherence should include precise goal setting for these as-needed exercises. Additionally, while the collection of physiological biomarkers related to loneliness and chronic stress is novel, the sample size was small, limiting its potential to be statistically meaningful.

The present study indicates feasibility and efficacy of a remote resilience and self-compassion intervention for older adults targeting loneliness and chronic stress. It also provides exploratory biomarker results on sleep and inflammatory correlates of loneliness and stress. Components of wisdom such as self-compassion may be improved through intervention, but it is not yet known if change in self-compassion was the primary psychosocial mechanism by which aspects of mental health were improved in our study. Additional testing with a control group and a larger sample is necessary to confirm the specific mechanisms responsible for intervention outcomes. Further, work is needed to improve the generalizability of the intervention and to understand the potential physiological effects or mechanisms by which resilience and self-compassion interventions work in older participants.

Acknowledgements

We would like to thank Barton W. Palmer for comments on an earlier draft of this manuscript. We also want to acknowledge Drs. Dilip V. Jeste and Ellen E. Lee for their contributions to funding, data collection, and manuscript conceptualization.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Raw data are available upon request.

Additional information

Funding

This work was supported, in part, by the Murray Galinson San Diego-Israel Initiative, by University of California San Diego’s T. Denny Sanford Institute for Empathy and Compassion, by the National Institute of Mental Health NIMH T32 Geriatric Mental Health Program MH019934 (PI: Elizabeth Twamley), NIA R01AG061941 (PI: Brent T Mausbach), and by the Sam and Rose Stein Institute for Research on Aging.