Anxiety disorders are pervasive among older adults and especially common in older military veterans. These disorders include generalized anxiety disorder (GAD), social anxiety disorder, panic disorder, agoraphobia, and unspecified anxiety disorders [1]. A meta-analytic review estimated that nearly 1 in 10 (9.1%) older military veterans met the criteria for one of these anxiety disorders [2]. Not only do late-life anxiety symptoms and disorders contribute to detrimental outcomes such as functional and cognitive decline [3-5], but the presence of an anxiety disorder (ie, GAD) is a risk factor for suicide ideation among older non–combat-exposed veterans [6,7]. Thus, access to mental health services for anxiety in older veterans is critical.

Numerous barriers impede older veterans with anxiety from receiving mental health services. These barriers include mental health stigma and beliefs, lack of knowledge about available mental health services, mobility limitations, transportation challenges, and residing in rural areas [8]. Another barrier is the impact of anxiety diagnoses on referral for treatment. Older veterans with anxiety are more likely to receive nonspecific anxiety disorder diagnoses and less likely to receive mental health services compared with younger veterans who receive specific anxiety diagnoses and more subsequent mental health services [9]. Thus, accessible and brief nonpharmacological interventions for late-life anxiety are needed to address the barriers to accessing mental health care faced by older veterans with anxiety disorders. Furthermore, conclusions from a recent systematic review and meta-analysis suggest that these brief interventions may show promise for reducing anxiety symptoms in older adults [10]. One such brief behavioral intervention, progressive relaxation [11], also known as progressive muscle relaxation (PMR), has been shown to be an efficacious behavioral intervention for the treatment of late-life anxiety [12]. PMR is often included as a component within cognitive behavioral therapy (CBT) for anxiety treatments, as well as being a component of digital mental health interventions such as self-management mobile apps. We focused on this skill and developed a 4-week video-delivered intervention that teaches PMR and diaphragmatic breathing and encourages the application of these skills to help patients engage in activities in which anxiety or other types of stress may arise. Activity engagement was selected as a treatment target as discomfort and distress from anxiety symptoms prompt individuals to use coping strategies such as escape from and avoidance of anxiety-evoking situations in the short term [13]. This process (ie, negative reinforcement) makes it more likely for avoidance to be relied upon in the long term and is particularly detrimental to older patients because reduced engagement in activities often leads to isolation and functional decline [7].

This intervention—Breathing, Relaxation, and Education for Anxiety Treatment in the Home Environment (BREATHE)—is organized with weekly video lessons, daily practice videos, and telephone coaching to encourage adherence to the practices. It was initially tested in a proof-of-concept study comparing the 4-week BREATHE intervention to a waitlist control in older adults with anxiety disorders [14]. The BREATHE intervention was found to be superior to the waitlist control in reducing anxiety, depressive, and somatic symptoms; however, the attrition in BREATHE (35%) warranted further investigation. To address these attrition issues, we sought feedback on BREATHE from older veterans and made iterative revisions to the intervention in a second study [15]. The revised BREATHE intervention was then tested in a small feasibility study with 10 older veterans with anxiety disorders [16] in which BREATHE was found to be feasible and acceptable in that 90% (n=9) completed the intervention and 100% felt that BREATHE somewhat or completely met their expectations. While BREATHE is a guided self-management approach to anxiety, it differs from other similar approaches that either rely on manuals and bibliotherapy or are entirely internet delivered (see the study by Cremers et al [17] for a review). BREATHE falls between these 2 approaches, uses familiar technology (web-based videos or DVDs), and focuses on a single skill (PMR) rather than multiple skills (eg, CBT).

In this study, we conducted a pilot randomized controlled trial (RCT) comparing 2 guided self-management interventions in older veterans with anxiety disorders. This study was conducted in part to determine the feasibility of conducting an RCT by documenting willingness to be randomized, engagement with the interventions assigned, and dropout. We also examined the preliminary efficacy of BREATHE compared with a psychoeducation intervention (Healthy Living for Reduced Anxiety) on anxiety symptoms and functioning in older veterans. We hypothesized that (1) BREATHE would result in greater reduction in anxiety symptoms compared with psychoeducation at 12 weeks and (2) BREATHE would result in significantly greater improvements in functioning compared with psychoeducation at 12 weeks. Our exploratory aims were to examine whether home practices and treatment engagement were related to patient characteristics or to intervention outcomes and examine participants’ perceptions of the home practices via qualitative interviews.

This study was reviewed and approved by the Stanford University Institutional Review Board (IRB-32454), the institutional review board of record for the US Department of Veterans Affairs (VA) Palo Alto Health Care System. Informed consent was obtained at the baseline visit. Participants were paid US $60 for the initial assessment and the 12 week assessment. Participants were paid US $10 for completing the telephone assessments at weeks 4 and 8. Data were deidentified prior to data entry and analysis.

Participants were recruited from a large VA health care system via posted flyers and brochures. Informational letters were also sent to patients who had at least one encounter during the past year that was related to a diagnosis of anxiety. Recruitment took place from February 2019 to March 2020 (before the COVID-19 pandemic) and from July 2020 to November 2020 (after the COVID-19 pandemic). Participants were eligible if they were aged ≥60 years and proficient in English and exhibited a diagnosis of an anxiety disorder (ie, GAD, panic disorder, agoraphobia, social anxiety disorder, or unspecified anxiety disorder). Participants were excluded if they were currently enrolled in other intervention research studies or were currently involved in individual therapy or group therapy more frequently than once per month. Those with possible cognitive impairment per a brief cognitive assessment or those who reported a diagnosis of bipolar disorder, psychotic disorder, schizophrenia, or other serious psychiatric disorders were excluded during the telephone screening process. If participants were taking psychotropic medications, they needed to be on a stable dose for 1 month before enrollment.

The following sections describe measures for inclusion and exclusion criteria assessment, outcome assessment, and measurement of covariates.

An a priori sample power analysis was calculated with an α of .05 and power of 0.80. On the basis of previous research, the controlled effect size (Hedges g) for relaxation therapy’s effect on anxiety was 0.90 (95% CI 0.44-1.44) [12]. The total estimated sample size was 30 (15 per group; Cohen f=0.41). Because of expected attrition and use of self-directed treatment, we estimated a smaller effect of the primary measure of anxiety and, thus, aimed for a sample of at least 26 participants per group (f=0.35).

Descriptive statistics were used to help characterize the sample; t tests (2-tailed) and chi-square analyses were used to test whether the 2 groups differed on any characteristics. Correlation analyses were conducted to examine whether baseline anxiety or medical comorbidity were associated with homework completion. Analyses were conducted using SPSS Statistics (version 29.0; IBM Corp) [39].

To examine the hypotheses regarding the primary outcomes of anxiety (GAS) and functioning (VR-12 MCS), mixed-effects models, also known as linear growth models or multilevel models [40], were used. Missing data points due to participant dropout were handled assuming that the data were missing at random and conditional on observed information. Mixed models were used to examine the change in outcomes across 4 time points (baseline [T1], 4 weeks and end of treatment [T2], 8 weeks [T3], and 12 weeks [T4]). Growth models with just time were estimated first followed by the fully specified models that included a between factor of treatment group (BREATHE vs psychoeducation); a within factor of time; and an interaction of treatment by time, which was the estimate of the treatment effect over the course of the study.

A total of 3 mixed-effects models with 2 time points (baseline [T1] and 12 weeks [T4]) were conducted on (1) anxiety as measured using the HAM-A, (2) perceived anxiety control as measured using the ACQ, and (3) lifestyle-adjusted functioning using the ACS. Sensitivity analyses were conducted to evaluate the change in anxiety symptoms (GAS) from baseline to 4 weeks.

Rapid qualitative analysis [41,42] was used to investigate perceptions of the home practices among the participants assigned to the BREATHE intervention. Interview transcripts were summarized using templates, and then domain summaries were created by 2 authors trained in qualitative techniques (CG and CC). The summaries were reviewed for themes.

Of the 98 participants assessed for eligibility, 56 (57%) were eligible and were all subsequently randomized. Figure 1 shows the participant flow throughout the study.

Randomized participants had a mean age of 71.36 (SD 6.19) years and ranged in age from 60 to 88 years. Table 2 shows the characteristics of participants at baseline. Participants were diagnosed with a current anxiety disorder at baseline using the SCID-5. A total of 16% (9/56) of the participants had more than one concurrent anxiety disorder, 29% (16/56) had co-occurring depression, and 14% (8/56) had subthreshold posttraumatic stress disorder (PTSD; other specified trauma disorder). The most frequently occurring disorders were other specified anxiety disorder (30/56, 54%), GAD (18/56, 32%), social anxiety disorder (11/56, 20%), agoraphobia (2/56, 4%), and panic disorder (2/56, 4%).

Completion of the assessment at 4 weeks and after treatment was 81% (22/27) for BREATHE and 93% (27/29) for Healthy Living (χ²₂=1.7; P=.19). The retention at week 12 was similar across groups, with 81.5% (22/27) of BREATHE participants and 82.8% (24/29) of Healthy Living participants completing the assessment at 12 weeks (T4). The groups did not differ with regard to which modality they used for the delivery of the videos (χ²=2.4; P=.30). Across both groups, 66% (37/56) used DVD delivery, 21% (12/56) used web delivery, 7% (4/56) used both, and 5% (3/56) had missing data on modality due to dropout from the study.

Most of the weekly coaching calls were completed for both interventions, with no differences found in the number of weekly calls completed for each intervention (t₅₄=0.92; P=.36). On average, BREATHE participants completed 3.48 (SD 1.05) of 4 (87%) weekly coaching calls. Healthy Living participants completed 3.72 (SD 0.92) of 4 (93%) weekly coaching calls.

With regard to engagement with the intervention as self-reported to coaches, the Healthy Living group completed significantly more lessons (mean 3.55, SD 1.09) than the BREATHE group (mean 2.85, SD 1.43; t₅₄=2.07; P=.04). In addition to the weekly lessons, the BREATHE group had daily home practices and 3 assignments to apply the skills in real life. Pearson correlation analyses found that greater baseline anxiety scores (GAS) were associated with fewer completed practices in the BREATHE group (r=–0.41; P=.03). In addition, greater severity of medical comorbidity (Cumulative Illness Rating Scale–Geriatric) was associated with fewer completed practices in BREATHE (r=–0.50; P=.009). On average, BREATHE participants completed 23.07 (SD 17.38) practices during the 4-week intervention. A total of 22% (6/27) of the participants completed 1 application of the skills, 19% (5/27) completed 2 applications of the skills, and 30% (8/27) completed ≥3 applications.

Participants shared feedback about the interventions through a postsurvey (Table 3). Participants generally agreed to strongly agreed regarding the ease of use and the usability of the DVD and website to access the videos. Most participants agreed to strongly agreed that the coaching calls were frequent enough and the program length was sufficient.

Treatment effects were examined using mixed-effects models conducted on total anxiety scores as measured using the GAS. The models were best fitted with a fixed effect of time and group and random intercept. Table 4 shows the means for the BREATHE and Healthy Living groups across 4 time points (baseline [T1], 4 weeks and end of treatment [T2], 8 weeks [T3], and 12 weeks [T4]). Notably, there was substantial variability for baseline GAS scores, with some participants reporting low symptoms in the previous week despite meeting the criteria for an anxiety disorder. There were no differences in the rates of decline in anxiety between the groups. That is, the effects of time (P=.07), group (P=.64), and interaction of treatment by group (P=.07) were not significant. An inspection of the subscales revealed that there was a significant time by group interaction for somatic anxiety symptoms (F_{3, 145.6}=2.81; P=.04) but not for affective or cognitive anxiety symptoms. This significant interaction indicated that the Healthy Living group experienced a significant decline in somatic anxiety over time compared with the BREATHE group.

No difference was found with regard to a difference in BREATHE or Healthy Living effect on mental functioning over time as measured using the VR-12. In addition, no difference was found between treatment groups with regard to depressive symptoms as measured using the PHQ-9 (Table 4), anxiety symptoms as measured using the HAM-A, lifestyle-adjusted functioning as measured using the ACS, or anxiety control as measured using the ACQ (Table 5).

As shown in Table 4, the BREATHE and Healthy Living groups did not differ in effects at week 4 and end of treatment with the exception of GAS somatic scale scores, which were lower for Healthy Living compared with BREATHE at this time point. We also investigated whether changes in scores from baseline to week 4 were associated with home practice completion. Using Pearson correlation analyses, we found the association with practice completion to be nonsignificant (r=0.33; P=.13). Reduction in GAS scores from baseline to week 4 (end of treatment) was not associated with completion of lessons (r_s=–0.09; P=.53) or coaching calls (r_s=–0.009; P=.95) across the BREATHE and Healthy Living combined samples.

Individual activity goals set by participants were examined with regard to the number of goals identified, types of goals, and change in individualized goal frequencies from baseline to 4 weeks. On average, participants selected at least 4 activities to focus on (mean 4.7, SD 1.2; range 2-7). Of 263 total activities identified, the 3 most frequently selected activity categories were social activities (72/263, 27.4%); high-demand leisure activities (70/263, 26.6%) such as hiking or swimming; and low-demand leisure activities (63/263, 24%) such as photography, reading, or playing a musical instrument. The remaining activities were instrumental activities (35/263, 13.3%) or other activities identified by participants (23/263, 8.7%).

Change scores for each activity goal were calculated from 0 to 4 weeks. Scores were then collapsed into 2 categories: increase in activity frequency (change score of >0) or no change or decrease in activity frequency (change score of ≤0). The BREATHE and Healthy Living groups did not differ in their distribution of increased activities at 4 weeks (χ²=0.2; P=.64). A total of 79% (23/29) of Healthy Living participants and 74% (20/27) of BREATHE participants reported an increase in frequency of one or more goals at 4 weeks. On average, participants in both groups attained ≥1 goal at 4 weeks (BREATHE: mean 1.6, SD 1.4; Healthy Living: mean 1.6, SD 1.3; consistent with an intention-to-treat approach, all participants were retained in these analyses. Those individuals who were lost to follow-up were coded as not attaining their goal).

Feedback regarding the experience of the practices in the BREATHE group, including the overall experience of the practices, mechanics and frequency of practices, challenges encountered, and application of skills to real-world situations, was analyzed to better understand participants’ experience with the BREATHE intervention (22/27, 81% completed the interviews). While Healthy Living participants relayed the same type of information, in this section, we focus on the BREATHE intervention as the condition of interest.

The findings of this pilot RCT of BREATHE compared with Healthy Living demonstrate the feasibility of conducting an RCT in that all participants were willing to be randomized and retention was similar between study arms at 12 weeks. Some variation in engagement with and acceptability of the interventions emerged, as discussed in the following sections, but does not decrease the feasibility of conducting an RCT. The hypotheses were not supported with regard to the greater reduction in total anxiety symptoms or improvement in functioning (as measured using the VR-12 MCS score) in BREATHE compared with the psychoeducation control of Healthy Living. On the anxiety symptom subscales, participants in the Healthy Living condition experienced a greater reduction in somatic anxiety symptoms, but no other differences were found for the other 2 subscales of cognitive and affective anxiety symptoms. There are several reasons for these findings. First, as observed in most previous late-life intervention studies [43], the inclusion of an active control may have diminished the potential effect of relaxation. Furthermore, a growing body of research has focused on the importance of well-being–focused interventions, including yoga [44], nutrition [45], and other alternative interventions [46], and the presence of some of these factors may have contributed to the benefits that participants achieved in the Healthy Living intervention. This intervention did include gentle stretching and some basic sleep and coping tips in addition to nutritional information, which all may relate to physical and mental health while aging. It is possible that a single-component intervention was not effective for engaging participants across a 4-week period. Recent studies have demonstrated that CBT for depression and anxiety is effective at not only reducing anxiety but also preventing relapse over a 10-year period [47]. In addition, the heterogeneity of anxiety disorders, as detected using structured interviews, and the range in anxiety severity based on baseline GAS scores may lead to attenuated findings with a single-component intervention rather than a multicomponent intervention to address varied facets of anxiety (ie, cognitive, somatic, and affective symptoms).

Our exploratory aim examined factors related to home practices in the BREATHE condition or related to intervention outcomes which revealed that greater self-reported anxiety symptom severity and greater medical comorbidity at baseline were associated with fewer progressive relaxation practices completed during the 4-week intervention. The qualitative findings helped us further probe the potential effect of practices. Qualitative findings suggested that, for some, practicing breathing, relaxation, or both breathing and relaxation was not viewed as enough to manage their anxiety and stress, thus dovetailing with the quantitative finding of those participants with greater anxiety practicing less. Accordingly, BREATHE or Healthy Living may be better suited for individuals with mild to moderate anxiety, which should be tested directly in a future study. It is possible that older veterans with greater distress would have preferred having access to group or individual psychotherapy and may have had previous exposure to these modalities as a function of the integration of mental health into the Veterans Health Administration [48].

Other challenges to practicing were recounted by participants, including difficulty finding the time to practice, having other things to do (eg, visitors and other activities), or worsening of health problems and experiencing acute health changes. Some participants reported a decline in practices after the coaching calls ended at the 4-week mark. Thus, the challenges with practices were 2-fold and included both adhering to the practices and maintaining a practice routine over time, which parallels the challenges of having sustained engagement with digital mental health interventions [49] and with any behavioral intervention requiring practice. Perhaps improvements and variations in the types of relaxation are needed to maintain interest in practices over time.

These findings led to the question of whether home practice is the issue itself. Integrating an ongoing relaxation practice into daily life while facing multiple medical conditions may have required extra effort and motivation from participants. The qualitative interview findings revealed that participants were able to personalize their practices based on their own needs, tailor their routine, and implement the relaxation procedures in their daily life when they needed it. Another possible future direction would be to provide multiple skills in a modular approach consistent with digital mental health interventions [50] or consider adding a session focused on motivational interviewing to the intervention.

Participants described using the skills in a myriad of settings, including when facing worries or anxiety, when in a stressful situation, or when simply waiting for an appointment, and they achieved one or more goals on average. Future studies should clarify the dose of PMR needed and whether preferences for psychotherapy compared with technology [51] play a role in practice completion and overall adherence. Alternatively, while these interventions were designed to be widely accessible and scalable, it is possible that a more mechanistic consideration of breathing approaches for specific individuals may be needed (ie, precision medicine approaches [52]). In addition, more engaging and adaptive technology–based interventions could be of benefit for late-life anxiety and lead to more sustained engagement with practices. Further refinement of goal setting to promote intervention engagement alongside skills to cope with anxiety is needed. On average, BREATHE and Healthy Living participants attained at least one goal at the end of treatment, but this could be strengthened using more directed, individualized outcome measures for the goals (eg, goal attainment scaling approaches [53]).

This study is not without limitations. One of the key limitations may be the heterogeneity of anxiety disorders and symptom presentation in a relatively small pilot study. Rather than including multiple types of anxiety disorders, focusing on specific symptoms (eg, worry that exceeds a particular threshold on a questionnaire) may be a better approach, as used with other studies [44,54]. As this sample included veterans, we likely encountered higher rates of co-occurring PTSD compared with other nonveteran samples. This particular comorbidity of PTSD may have made the anxiety presentation more complex and difficult to treat using guided self-management interventions. Additional limitations include the small sample size as the study was designed to be a pilot, the use of nonblinded assessors, and the absence of a validated measure of acceptability. Another limitation to our design was that only a portion of the outcome measures were assessed at the end of treatment (T2). This study took place earlier in the COVID-19 pandemic, which led to some modifications in data collection methods and limited our ability to collect complete data, in particular from the lifestyle-adjusted ACS.

These findings differ from those of earlier studies on progressive relaxation for late-life anxiety in that the effects of progressive relaxation were not as robust as those found in earlier studies [55]. The use of a primarily male, ethnically and racially diverse military veteran sample may have contributed to some differences compared with earlier studies, in which older nonveteran White women made up much of the samples. The benefits derived from the Healthy Living intervention fit with the evidence for complementary and integrated health approaches for late-life anxiety [44] and the Whole Health program in the Veterans Health Administration [56].

These findings suggest that guided self-management approaches to treating late-life anxiety in older veterans are feasible but that further refinement and study are needed to identify what works best for whom using a video-delivered format with remote coaching. Our findings suggest that a psychoeducation-based approach may help older adults with somatic anxiety symptoms. While progressive relaxation was deemed to be feasible and enjoyable to most participants, those with mild to moderate anxiety symptom severity and fewer health problems were more likely to adhere to the recommended home practice. Thus, based on the qualitative feedback, the BREATHE intervention in particular might not be a good match for all older adults with anxiety disorders due to some participants needing a higher-intensity treatment and others experiencing negative reactions to progressive relaxation, primarily “relaxation-induced anxiety” [57]. Further work is needed to delineate the role of intervention design factors and individual participant baseline characteristics on the effect of guided self-management approaches on late-life anxiety.