The “Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)” guideline was followed in conducting this systematic review and meta-analysis [21]. We used the “International Prospective Register of Systematic Reviews (PROSPERO)” to register our work (CRD42024495179).

Since only published data were used and analyzed, no ethics approval or written informed consent from the participants analyzed within the included studies was required.

We conducted systematic literature research on the databases Cochrane Library, APA PsycInfo, and Web of Science using only free text. On PubMed we used a combination of free text and Medical Subject Headings (MeSHs) and on Embase, we used free text combined with terms from Emtree. The keywords we have chosen for the free text search are as follows: “eHealth,” “e-Health,” “mHealth,” “telehealth,” “digital health intervention,” “mobile,” “app,” “web,” “web-based,” “online,” “phone,” “internet,” “internet-based cognitive behavioral therapy,” “web-based cognitive behavioral therapy,” “online cognitive behavioral therapy,” “digital cognitive behavioral therapy,” “cardiovascular disease,” “heart disease,” “coronary heart disease,” “coronary artery disease,” “ischemic heart disease,” “heart attack,” “heart failure,” “cardiac failure,” “acute coronary syndrome,” “myocardial infarction,” “peripheral arterial disease,” “peripheral occlusive disease,” “cardiac rehabilitation,” “secondary prevention,” “stress management,” “stress reduction,” “distress,” “stress,” “self-management,” “self-efficacy,” “quality of life,” “risk factor modification,” “risk factor reduction,” “randomized controlled study,” and “randomized controlled trial.” By using the database for MeSHs of PubMed, we were able to identify the following suitable MeSHs for our systematic literature research: “Telemedicine”[Mesh], “Mobile Applications”[Mesh], “Internet”[Mesh], “Computers[Mesh],” “Cell Phone”[Mesh], “Cognitive Behavioral Therapy”[Mesh], “Cardiology”[Mesh], “Cardiovascular Diseases”[Mesh], “Depression”[Mesh], and “Anxiety”[Mesh].

First, a combined search term was created for the systematic literature research on PubMed using the predefined keywords, MeSH, the operators “AND” as well as “OR” and the phrase search for multiple word keywords (eg, “cardiac rehabilitation”). This created search term was then converted with the “Polyglot Search Translator” for Embase [22]. Within this converted search, MeSHs were replaced by terms from Emtree, if MeSHs and terms from Emtree were not identical. Second, we have created a combined search term like the previous one but containing only keywords for a free text search. Afterward, we converted the created search term with the “Polyglot Search Translator” for the Cochrane Library, APA PsycInfo, and Web of Science [22]. In Table S1 in Multimedia Appendix 1 [22-29], we presented the used search term for each database searched.

In addition to searching the databases, we also searched the following clinical trial registers for completed and ongoing trials: the International Clinical Trials Search Portal, the International Standard Randomized Controlled Trial Number registry, the German Clinical Trials Register (GermanCTR), and ClinicalTrials.gov. The search terms used for the systematic research on all clinical trial registers are also shown in Table S1 in Multimedia Appendix 1.

We used the literature management software EndNote (version 20.6 on Windows 11; Clarivate Analytics) to store the identified publications and clinical trial register entries, to detect and delete duplicates, and to conduct the entire study selection process (title/abstract screening and full-text screening). The study selection process started after we imported all records including the study register entries, and after all duplicates had been removed.

Inclusion criteria were created following the PICOS (Population, Intervention, Control, Outcome, Study Design) scheme. Thus, to consider a study eligible, the following criteria had to be met:

Exclusion criteria were defined regarding the population, intervention, outcome, and publication type as follows:

Considering the inclusion and exclusion criteria, a title and abstract review was undertaken for publications as the first part of the study selection process. All reviewed publications were either directly excluded with a note or taken into consideration for a full-text screening, which then decided on study inclusion or exclusion with a note. All notes regarding the exclusion of a publication within the first or second part of the study selection process were documented in EndNote. The results of the selection process were discussed between 2 researchers.

The first part of the screening process for registered clinical trials (records from clinical trial registers) was similar to the screening process described above. The title and all available information were read and assessed for eligibility. Only eligible studies were taken into consideration for the second part of the screening process. We searched on PubMed and Google for publications linked to the study identification number. Once a match was found, the corresponding report was linked to the study in EndNote, and a full-text screening of this report was performed. In cases where we were unable to find suitable publications, we requested study reports from the contacts listed in the study register entry. No study report was requested for ongoing studies. These studies were included or excluded based on available and other researched information.

Data extraction was performed separately for eligible studies with already published results and for ongoing studies using 2 different Microsoft Excel files. Regarding the completed studies, we decided to extract data as follows: first author (last name), publication year, registration number, important information for assessing the risk of bias (study design, randomization method, blinding, and intention-to-treat-analysis), sample sizes and demographic data (age and gender distribution), investigated CVD (acute coronary syndrome, coronary artery disease, congestive heart failure, or peripheral arterial disease), number of participants with other diseases (eg, heart diseases different from the investigated diseases, diabetes, cancer), participation criteria, delivered intervention (type, period, and number of sessions), information regarding the control group, and outcome of interest including the assessment method and the results of interest measured at baseline and after the intervention period.

As far as the ongoing studies are concerned, we have decided to extract general study information (official name and registration number), inclusion criteria, study start, delivered interventions, and outcomes of interest in the Microsoft Excel file created for this purpose. All data were extracted based on study register entries or researched study protocols, for those available.

Quality assessment was performed by evaluating the risk of bias. For this purpose, we decided to use the Risk-of-Bias 2 tool that addresses the following domains: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other biases [39].

Microsoft Excel was used for data synthesis, summarizing the means and SDs for all reported mental well-being outcome parameters as follows: (1) anxiety rated with the GAD-7 and the CAQ; (2) depression rated with the CDS, the PHQ-9 and the MADRS-S; (3) anxiety/depression rated with the HADS; (4) stress rated with the DASS-21; and (5) quality of life with rated with the SF-12 and the AQoL-8D [16,17,30-37]. These values were summarized at 2 time points (baseline and postintervention) for the intervention group and for the control group.

A meta-analysis was performed for outcome parameters reported by at least 2 studies. In some cases (eg, CAQ-avoidance) a meta-analysis was also conducted for the subscales. Thus, the following psychological well-being outcome parameters were investigated within a meta-analysis: CAQ (CAQ-total, CAQ-avoidance, CAQ-attention, and CAQ-fear) and GAD-7 for assessing anxiety, MADRS-S and PHQ-9 for assessing depression, and SF-12 (SF-12 Physical Health and SF-12 Mental Health) for assessing quality of life. The meta-analyses for anxiety and depression were conducted based on the measurement instrument used in at least 2 studies, as most studies reported anxiety and depression using multiple scales [23-27]. For the meta-analysis of quality of life, the assessment methods used within the studies were not comparable in that the studies that used the SF-12 assessment method reported quality of life with 2 scores, the Physical Health and the Mental Health score [23,24,26], whereas the study using the AQoL-8D assessment method delivered one total score [28]. For this reason, no scale-independent meta-analysis was conducted for the psychological well-being outcome parameter anxiety, depression, or quality of life.

The conducted meta-analysis is based on a random effects model and only postinterventional values were used. All analyses were performed using standard mean difference (SMD) to indicate the effect size and I² statistics to evaluate heterogeneity.

The sensitivity analysis was performed using a different effect size for each outcome parameter. For this, we decided to use the effect size “mean difference.” In view of the small number of studies (≤5 studies) within each meta-analysis, we decided not to conduct subgroup analyses for any outcome parameter.

All analyses were performed with the software IBM SPSS Statistics for Windows (version 29.0.0.0; International Business Machines Corporation).

Our systematic literature search initially yielded 2990 records, including publications and clinical trial registrations. After removing duplicates, we identified a total of 7 ongoing studies [40-48] and 6 completed studies [23-29] that met our eligibility criteria and were included in the analysis. The whole study selection process is detailed in the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) flow chart [21] in Figure 1.

An overview of the 7 eligible and currently ongoing studies can be found in Table 1 [40-48]. The study characteristics of the 6 completed studies [23-29] are summarized in Table 2. As far as the completed studies are concerned, the total sample sizes vary ranging from 34 to 239 participants [25,28]. The highest mean age of the participants in the intervention group was 63.6 (SD 13.9) years [29], whereas the highest mean age of the participants in the control group was 64.0 (SD 12.0) years [26]. Control groups within the completed studies received usual care [23,25], participated in a web-based discussion forum [26,29], or were placed on the waiting list [24,28]. One study examined patients with atrial fibrillation in addition to patients with congestive heart failure or acute coronary syndrome. As the total number of patients with congestive heart failure and acute coronary syndrome represented the majority of patients, we decided to include this study in our systematic review and meta-analysis [26]. Depression as a physical well-being outcome of interest was reported in all 6 studies [23-26,28,29], anxiety in 5 studies [24,25,27-29], quality of life in 4 studies [23,24,26,28], and stress/distress in one study [24].

All 6 completed studies, which are summarized in Table 2, underwent assessment for potential bias. There was neither a study with a high risk of bias in any of the 5 domains nor with an overall low risk of bias as presented in Figure 2. Two studies were found to have some concerns regarding the second domain, which assessed deviations from the intended interventions [24,25]. Four studies were assessed with some concerns regarding the domain addressing missing outcome data [23,25,26,29] and the domain related to the selection of reported results [24,26,28,29]. The domain dealing with outcome measurement was evaluated in one study with some concerns [24].

The meta-analyses included all 6 completed studies and were conducted on the basis of the outcome parameters reported in 7 study reports [23-29]. Meta-analysis was not undertaken for outcome parameters of interest that were assessed using the following questionnaires, as they were only reported in one study each: AQoL-8D [28], HADS [25], CDS [23], and DASS-21 [24].

One study reporting CAQ values for the intervention group was excluded from the meta-analysis of the parameter CAQ due to the lack of detailed values (mean and SD) for the control group [29].

Analysis of the SF-12 score for physical health was based on 3 studies investigating a total of 158 participants in the intervention group and a total of 160 participants in the control group [23,24,26]. One of the analyzed studies reported a lower SF-12 score for physical health in the intervention group in contrast to the control group [24] suggesting poorer physical health [35], while the other studies reported higher values in the intervention group compared with the control group [23,26]. This analysis showed a nonsignificant result (P=.08), an overall small effect (SMD=0.21), and negligible heterogeneity characterized by the I² value of 0.12. The results of this meta-analysis are shown in Figure 6.

The SF-12 score for mental health was analyzed based on the same 3 studies as the SF-12 score for physical health and also included a total of 158 participants in the intervention group and a total of 160 participants in the control group [23,24,26]. All 3 studies reported higher values in the intervention group [23,24,26] indicating improved psychological well-being [35]. This analysis, shown in Figure 7, yielded a significant value (P<.001) with an overall small effect size (SMD=0.38) and no heterogeneity that was reflected in an I² value of 0.00.

The sensitivity analysis conducted on the outcome parameters linked to anxiety all resulted in nonsignificant findings. Two studies each were involved in the analysis of the CAQ-total score, the CAQ-attention score, the CAQ-avoidance score, and the CAQ-anxiety score [24,27], while the sensitivity analysis of the GAD-7 score included a total of 3 studies [24,27,28].

Regarding the outcome parameters measuring depression, the sensitivity analysis of the PHQ-9 score that included 5 studies [23,24,26,28,29] showed significance (P<.001), while the analysis of the outcome parameter MADRS-S, which was based on 2 studies [25,26] yielded a non-significant result.

Sensitivity analysis of the SF-12 Physical Health score and the SF-12 Mental Health score, which express quality of life, included 3 studies each [23,24,26]. The SF-12 Mental Health score yielded a significant result (P<.001), while the SF-12 Physical Health score yielded a nonsignificant value.

The mean difference and its CI, the SE, the P value, and the I² values expressing heterogeneity determined in the sensitivity analysis for each outcome parameter are presented in Table S2 in Multimedia Appendix 1.

This systematic review and meta-analysis examined the effects of eHealth stress management interventions on psychological health parameters, including anxiety, depression, and quality of life, as measured by psychometric instruments. Significant postinterventional differences between the intervention and control groups were found for depression, indicated by the PHQ-9 score, and for quality of life, measured by the SF-12 Mental Health score. No significant effects were observed for anxiety, as measured by the CAQ or GAD-7 score, nor for depression, as measured by the MADRS-S, nor for quality of life, as expressed by the SF-12 Physical Health score. It is important to note that the analysis of the MADRS-S included fewer studies [25,26] compared with the PHQ-9 analysis [23,24,26,28,29]. Aside from this, the result regarding the MADRS-S showed a high level of heterogeneity, which could potentially be attributed to differing group sizes or intervention periods within the studies [25,26]. Since no heterogeneity was observed in the analysis of the PHQ-9 score, it can be assumed that the overall effect regarding depression, as measured by the MADRS-S, was negatively influenced by the high heterogeneity. This is likely not the case for the PHQ-9, due to the absence of heterogeneity. The analysis of SF-12 Physical Health and SF-12 Mental Health was conducted using the same set of studies [23,24,26]. Our meta-analysis finding that internet-based CBT has no significant effect on the SF-12 Physical Health score is consistent with the findings of a previous meta-analysis that investigated the impact of CBT in patients with CVD [49].

Since all conducted analyses included studies that investigated internet-based CBT [24-29] or an eHealth intervention with a CBT approach [23], and both the PHQ-9 score and the SF-12 Mental Health score showed significant changes with the intervention, a positive effect can be attributed to internet-based CBT. Previous meta-analyses have also demonstrated that CBT can reduce depression [50,51] and improve mental health-related quality of life in patients with CVD compared with a control group [49]. Another already published meta-analysis, which included studies on both CBT and internet-based CBT, showed positive effects on anxiety, depression, stress, and quality of life in patients with cardiac conditions [52]. These findings align well with results from a prior research article examining the impact of internet-based CBT on patients experiencing increased stress and stress-related disorders [53]. We were unable to confirm these results for anxiety and stress in relation to internet-based CBT. Our analyses for anxiety did not show significant results, and the number of studies that assessed stress using the same assessment method was insufficient to conduct a meta-analysis. However, it is important to note that we examined the effects of eHealth stress interventions on anxiety, depression, and quality of life depending on the measurement method used. We identified differences between the 2 groups in terms of anxiety levels after the treatment period using the estimated overall effect sizes. In the sensitivity analysis, the same outcome parameters yielded significant results on depression measured by the PHQ-9 score and quality of life measured by the SF-12 Mental Health score. We conclude that both effect sizes lead to comparable results.

Our research highlights the use of internet-based CBT for enhancing psychological health parameters in patients with CVD. Recent studies investigated its application and impact on psychological outcome parameters in patients with other diseases [54,55]. Research on effectively reducing stress, depression, and anxiety is important, especially since the success of treatment and adherence can be negatively impacted, such as among patients with cancer [56].

In addition to internet-based CBT, other internet-based and computer-based stress management interventions have shown potential to positively impact psychological health parameters by reducing stress, anxiety, and depression, as demonstrated in a previous meta-analysis [57]. Mindfulness-based interventions via the web generally demonstrate stress-reducing effects, as evidenced by a conducted meta-analysis involving patients without diagnosed psychological or physical illnesses [58]. Therefore, further exploration of these interventions beyond internet-based CBT, as indicated by ongoing studies included in this work [40,41,44,47,48], can yield additional insights in this field. It could be also beneficial to explore the implementation of web-based exercise programs for stress reduction in the secondary prevention of heart disease. Physical activity has the potential to significantly enhance heart health by improving parameters of heart rate variability, which may be adversely affected by stress [59]. Mind-body exercises such as yoga but also interventions like HRV-biofeedback have been shown to have positive effects on stress reduction in individuals with CVD, as indicated by recent research [60]. Internet-based approaches or variations of these or other effective stress management interventions should therefore be considered, as several studies have demonstrated the positive impact of eHealth interventions on patients with CVD [61] and high patient acceptance for eHealth cardiac rehabilitation programs in secondary prevention is given [62].

This meta-analysis faced some limitations. First, we decided to include a study in which not all participants experienced atrial fibrillation, though a considerable number of them did. The results reported in that study cannot be attributed exclusively to patients with coronary artery disease or congestive heart failure which represented the majority of the patient population examined within the study [26]. For future research, it would therefore be of interest to examine patients with atrial fibrillation only, to better understand the impact of eHealth stress management interventions specifically in relation to this condition. Second, high heterogeneity between the studies was also found within some analyses. The causes of the heterogeneity could be attributed to the number of study participants, the duration of the intervention period, or the differences regarding the control group. However, due to the small number of studies and the resulting lack of subgroup analysis, the cause of heterogeneity could not be determined. Third, all analyses conducted included less than 15 studies which is due to the limited number of available and comparable studies. This means that the results presented should be interpreted with caution and that further research is needed in this area. Fourth, important supplementary information, such as the total number of sessions and the number of sessions completed by all participants, was not considered in our review, as it proved difficult to compare this information meaningfully due to the differing intervention durations or the varied structure of the intervention modules. This indicates that future research should also focus on this aspect to draw meaningful conclusions from the quantity and frequency of the sessions. Fifth, since we had only one already completed study examining the impact of an eHealth stress management intervention on stress [24], we were unable to assess its influence on stress within the meta-analysis. More studies are needed that examine the effects of eHealth stress management methods on stress in patients with CVD.

Overall, this meta-analysis validates the positive influence of eHealth stress management interventions on psychological health parameters in patients with CVD and marks the importance of further clinical approaches for digital health strategies.