Patients with coronary vascular dysfunction were screened for enrollment in 6 Dutch hospitals: Radboud University Medical Center Nijmegen, Maastricht University Medical Center, Zuyderland Hospital Heerlen, St Elisabeth Hospital Tilburg, Catharina Hospital Eindhoven, VieCuri Hospital Venlo, and Maasstad Hospital Rotterdam.

Adult women were eligible if diagnosed with CAS, defined as coronary epicardial vasospasm or microvascular spasm determined by CFT with acetylcholine according to the criteria of the Coronary Vasomotor Disorders International Study Group [5,16].

The upper age limit for inclusion was 70 years to enhance digital literacy. Further exclusion criteria included obstructive coronary artery disease (CAD), defined as ≥70% coronary artery stenosis, fractional flow reserve <0.80, or both; and practical inability to participate in the study, for example, due to a language barrier.

The study was designed as a single-sequence, 2-period, crossover study, allowing a within-patient comparison between control and intervention. After signing the informed consent form, participants had 1 habituation week to get used to the Garmin Vivosmart 4 wearable.

During this week, the researchers had a one-on-one digital or in-person meeting with the patient, setting up the connection and addressing potential questions and issues. The patients were instructed to wear the wearable during the daytime; nighttime use was optional. The 4-week control period was initiated when Bluetooth successfully connected the wearable.

After the control period, the consecutive 4-week intervention period commenced, in which the Wavy app was updated to the full access version. The researchers guided patients in installing the Wavy app, and patients were personally informed about the available features mentioned in the Wavy Assistant App section below. Considering that the study used a prototype version of the app, technical support was a priority and was readily accessible throughout the study period. Incoming data were closely monitored to promptly identify and address any technical issues. If no data were received for 2 consecutive days, researchers proactively initiated technical support. This support was provided through one-on-one virtual meetings, which included assisted screen sharing to resolve issues efficiently.

All participants provided written informed consent before participation. The study procedures comply with the ethical standards of the Declaration of Helsinki (2013). Moreover, the central commission of human related research region Arnhem-Nijmegen reported no objections to conducting the study, and local ethics approval was acquired for each center (NL77493.091.21). Pseudoanonymized data safeguarded patients privacy. Patients received no financial compensation for participation. Reporting is done according to the digital health implementation iCHECK-DH [26].

We screened 241 women with CAS, who were diagnosed by the CFT with acetylcholine. Of those, 137 (56.8%) provided informed consent. A total of 48 (19.9%) patients participated in the smaller pilot iteration rounds, aiming to develop and improve the Wavy app further, of which 28 (58.3%) continued to this study. The resulting 117 (48.5%) patients were included in the main trial described in this paper. Technological barriers and other issues with the app prohibited proper app use for 15 (12.8%) patients, leaving 102 (87.2%) patients to finish the trial (Figure 4). The trial enrollment was split into 3 periods, with group A starting in August 2022, group B in September 2022, and group C in November 2022.

Baseline characteristics are described in Table 1. The mean age was 58 (SD 7.3) years. From the 102 patients, the CFT diagnosed 53 (51.9%) patients with epicardial spasm and 49 (48%) patients with microvascular spasm as their endotype of CAS according to the Coronary Vasomotor Disorders International Study Group guidelines. The mean study duration was 59 (SD 9.2) days. The average number of days with a successful HRV measurement was 49 (SD 10.6); on the remaining days, no measurement was registered. On days with HRV registration, 61% of the maximum possible measurements were successfully recorded. In total, 1236 breathing exercises were performed with an average duration of 1 minute and 43 second (SD 43 seconds).

The primary outcome, average SAQ-SS, following the control and intervention phases was 50.4 (SD 15.4) and 50.9 (SD 15.0), respectively. These values indicated a fair to good quality of life in both groups, and comparison exhibited no statistically significant difference (P=.59; Figure 5). The mean SF-36 summary scores and individual components did not differ significantly (P=.92; Figure 6). Furthermore, the mean perceived stress score also showed no statistical difference (Figure 7).

However, HRV analysis showed a significant increase in RMSSD, indicating a relaxing effect during and after breathing exercises, compared to the measurement 15 minutes before, with a median (IQR) RMSSD of 18.59 (8.38-36.58) milliseconds before and 34.93 (24.19-50.64) milliseconds during the relaxation exercise (Z=–13.72; P<.001). Compared to the previous measurement, this effect is attenuated at 15 minutes after the exercise with a normalized median of 30.72 (IQR 18.97-47.76) milliseconds (Z=–11.75; P<.001). A visual overview is displayed in Figure 8.

Regarding the user experience, the Wavy app, which was still in development, experienced technical instability. With measurements taken every 15 minutes, a total of 96 time points could be collected each day, and 61.5% (59/96) of the possible data points were retrieved. Technical issues were prevalent, with most patients reporting technical issues, including difficulties logging in, connecting the wearable to the phone, and crashing. These issues resulted in increased technical support from the researchers. Nevertheless, 68 (66.7%) of 102 patients reported that they would use such an app in the future, provided it were a stable version.

The study was conducted to assess the impact and potential of the prototype Wavy app that strives to aid in lifestyle management for patients with CAS. As per the inclusion target, 102 (87.2%) of 117 patients completed the study. The primary outcome, the SAQ-SS, and secondary measures (SF-36 and Perceived Stress Score-10 items) did not exhibit significant improvements after a 4-week intervention period. However, the study was hampered by technical difficulties related to the app being in its development phase.

Notably, the results of the HRV analysis indicated increased HRV levels during the breathing exercise, with normalized values after the breathing exercise, thus showing a significant positive effect of breathing exercises on stress levels.

In reviewing the literature, no comparable studies of HRV biofeedback–based stress management on cardiac symptoms and quality of life were found for this specific patient group. However, various studies investigated the effect of HRV biofeedback–based stress management on mental stress and anxiety levels. Goessl et al [27] conducted a meta-analysis of 24 studies investigating the stress-reducing effects of HRV biofeedback. The analysis spanned 10 clinical and 14 community settings, with the majority of research (n=18) originating from the United States and the European Union [28,29]. In total, 13 studies that were mentioned by Goessl et al [27] included a comparison group, constituting standard care, sham, or other means of stress reduction. In some studies, the HRV biofeedback training could be performed at home with a device, while others included a personal trainer; the frequency of HRVB sessions varied between 1 and 50. When comparing the HRV-based stress management to the comparison condition, the biofeedback resulted in an effective reduction of anxiety and stress [27]. While Wavy did not achieve the desired effectiveness in improving perceived stress scores, we found a significant HRV improvement, indicating stress reduction.

In contrast to our intervention, which involves providing breathing exercises upon the detection of increased stress to enhance HRV and subsequently reduce stress, Held et al [30] explored the reduction in HRV induced by exposure to a stressful working condition in 26 participants with clinical anxiety and 14 control participants. Notably, the magnitude of HRV changes observed in both groups closely resembled the inverse HRV results presented in our paper. The decline in HRV during a cognitive stressful task in both healthy individuals and individuals with anxiety aligns with the HRV increase observed during the breathing exercises for these patients. Consistent with the findings of Held et al [30], the magnitude identified in our HRV analysis indicates a relevant stress reduction.

In comparison to other remote HRV biofeedback studies, compliance was decent, with 1236 breathing exercises performed (approximately 3 per week per patient). In contrast, a similar study by Hirten et al [30] failed to demonstrate the feasibility of remote monitoring due to poor compliance. Remote, wearable-based studies face significant challenges in acquiring a representative sample of available data points. For instance, Galarnyk et al [31] reported that only 5.6% of all recorded beat-to-beat interval data were usable, with 43% of participants never activating their device. In this study, despite similar technical challenges, 61.5% (59/96) of the total achievable data points—each representing approximately 30 seconds of recording every 15 minutes—were successfully collected. This outcome underscores the importance of active data monitoring and proactive technical support in maximizing data retention. Additionally, it highlights the role of an initiative-taking patient population, which likely contributed to the relatively high data capture rate observed in this study.

Although stress management is included in the current guidelines for the management of coronary vasospasms, our study is the first to evaluate a stress reduction program in women with proven CASs. The fact that 66.7% (68/102) of patients expressed willingness to engage in further app use after stabilization underscores the willingness of the patients to apply a home-based, digital form of stress management.

This study design enabled each patient to serve as their own control, reducing the influence of confounding variables. Although factors such as disease progression or medication changes could have introduced some variability, the 10-week study period makes the impact of these factors unlikely to be substantial.

As the app was still in the development phase, we encountered more technical issues than anticipated. While nearly all challenges were addressed through active data monitoring and personalized support, including screen sharing sessions, these recurring efforts placed an additional burden on participants and likely impacted their overall user experience. Consequently, it is difficult to draw definitive conclusions about the program’s effectiveness in alleviating symptoms. Stress is one of the potential triggers of angina; therefore, a full disease management app should extent their functionalities to address other lifestyle factors, for example, blood pressure management.

Despite these technical challenges, HRV analysis demonstrated a positive effect of the breathing exercises on stress reduction. Additionally, unlike many stress management interventions reported in the literature, which often span longer durations, our study used a relatively short intervention period. Considering evidence suggesting that extended intervention durations may yield greater improvements in self-reported stress and anxiety, the limited time frame of this study may have been a key limitation [27].

Because the SAQ-SS was developed primarily for obstructive CAD, its applicability to nonobstructive coronary disease is debated, and this may have been a potential limitation [32]. However, other studies support this psychometric for a more widespread population, including nonobstructive CAD [25,33]. A more sensitive CAS-specific outcome measure may yield more insightful results in future research; however, this has yet to be developed.

In conclusion, the current digital stress reduction prototype, Wavy app, did not affect perceived anginal burden in female patients with coronary vasospasms as measured by various questionnaires. However, considering the technical limitations of the device in its development stage, it is too early to dismiss the potential of this app to provide biofeedback to complement existing treatment strategies. The significant improvement in HRV during and after the breathing exercise, indicating the potential stress-reducing effect of the stress management app, is encouraging. For a complete disease management tool, development should extend beyond stress as its sole focus, as stress is just one of the possible symptom-provoking factors.

Future research is warranted with a technically stable device and longer follow-up duration to establish the true effect of stress reduction with breathing exercises on stress symptoms in the growing CAS population.