Cardiovascular diseases (CVD) are the leading cause of death worldwide [1]. The burden of CVD, considering the number of disability-adjusted life years and deaths, and especially the CVD burden attributable to modifiable risk factors increase globally. Therefore, countries are recommended to implement cost-effective public health programs and interventions which target modifiable risks, promote healthy aging across the lifespan, and reduce disability and premature death due to CVD [2].

Cardiac rehabilitation is a multidisciplinary model of health care that consists of 4 phases. Phase I starts during in-hospital treatment and focusses on early mobilization. Phase II can either be performed as in-clinic or outpatient cardiac rehabilitation, depending on the availability of outpatient cardiac rehabilitation and the patient’s individual needs and preferences. Outpatient cardiac rehabilitation phase III (OUT-III) can follow both in-clinic and outpatient cardiac rehabilitation and consists of weekly visits at outpatient cardiac rehabilitation facilities. While during phase II patients are being introduced to the management of cardiovascular risk factors and subsequent lifestyle changes, phases III targets at further improving or at least maintaining results of lifestyle changes achieved during phase II [3,4]. Home- and telehealth-based interventions are increasingly being used in cardiac rehabilitation, and depending on the chosen cohort, outcomes for home-based rehabilitation may well compare with center-based cardiac rehabilitation programmes in terms of hospitalizations, quality of life (QoL), and cost [5].

The success of home- and telehealth-based cardiac rehabilitation is associated with various clinical, psychological, and physiological parameters that influence and are influenced by the patients’ adherence to the cardiac rehabilitation programme. Various studies have analyzed the effect of wearables and home-base cardiac rehabilitation on the QoL. In a recent review including 57 articles, Jones et al [6] conclude that home-based cardiac rehabilitation leads to an improved QoL and exercise capacity.

So far, there is little knowledge concerning the relationship of patients’ intrinsic motivation and digital tools during home-based cardiac rehabilitation. In a small-scale study with 23 patients, Lu et al [7] identified patients’ intrinsic motivation as a key facilitator for successful rehabilitation with remote activity sensing. The rehabilitation program, however, did not focus on cardiac patients and none of the patients were reported to have a cardiac disease. Nonetheless, considering the broader context of health behavior change, which ties in closely with multiple of the above-mentioned cardiac rehabilitation components, there is a considerable body of research supporting the importance of (intrinsic) motivation [8]. Many current psychological models of health behavior, such as the Health Action Process Approach (HAPA), place motivation and resulting intentions (motivational phase) as a prerequisites of health behavior engagement, distinguishing it from a second, volitional phase, which translates intentions into actual behavior [9]. In the motivational phase, the HAPA proposes risk perception and outcome expectancy as determinants of intentions. Consecutively, action planning, self-efficacy, and action control are proposed to affect the enactment of these health behavior intentions. Therefore, all of these are promising targets for health interventions [10-12]. Various studies analyzed self-efficacy and behavioral driving models during home-based rehabilitation [13-16], indicating that there is a huge potential for addressing psychological drivers in digital technologies used in cardiac rehabilitation. When implementing digital tools to support cardiac rehabilitation, user-friendly interfaces are crucial, since poor usability is a main barrier of digital technologies in cardiac rehabilitation, as recently highlighted by McGowan et al [17].

The multicentered EPICURE study analyzed the effect of digital tools in outpatient cardiac rehabilitation including home training. Both patient groups concluded the OUT-III according to the latest national and international recommendations for center- and home-based exercise training [18,19]. Patients aged ≥18 years who concluded OUT-III at 1 of the 5 participating institutions, who had any documented cardiovascular disease (not further specified), and who were able to provide their informed consent were eligible for the EPICURE study if they gave their written informed consent [18]. The study duration was 6 to 12 months, depending on the duration of OUT-III which is flexibly adjusted according to the underlying disease [19]. EPICURE’s primary outcomes did not support the hypothesis that the additional use of digital tools during home training lead to further improvement in the exercise capacity (P_max) during and after OUT-III. P_max was assessed during exercise stress test on an ergometer under supervision of health care professionals in the cardiac rehabilitation facilities. Patients who self-reportedly used either 1 or multiple of the following technologies during OUT-III were categorized as a patient using digital tools: phone-based assessments by the attending cardiac rehabilitation facility and digital training diaries (with and without adherence monitoring done by the cardiac rehabilitation facility and with and without wearables).

This information was reported retrospectively by the patient and no protocol had been put in place to assign patient as users of digital tools (eg, minimum number of phone-based assessments) [18]. So far, the relationship between clinical, psychological, and physiological parameters as recorded in the EPICURE study had not been analyzed in detail.

The present paper takes a deeper look into the EPICURE data to better understand the relationship between the use of digital tools and aforementioned psychological, clinical, and physiological parameters, and the relationship between digital tools, psychological, clinical, and physiological parameters and the improvement of exercise capacity during cardiac rehabilitation. Results are discussed in relationship to the primary outcome of the EPICURE study as published in [18] and concerning their accordance with the state-of-the-art.

Data were taken from the multicentric EPICURE study [18]. The dataset contained 149 patients, who were asked to answer a questionnaire concerning their cardiac rehabilitation, and to grant access to data acquired during cardiac rehabilitation. No blinding or randomization was applied since patients were recruited retrospectively. Patients were assigned to 2 groups, depending on whether they reported to have used any of the following digital tools during OUT-III: phone-based assessments, digital training diaries with and without adherence monitoring and with and without wearables. At baseline (before OUT-III), patients using digital tools were significantly younger, fitter in terms of the maximum power during ergometry (P_max), had a lower BMI and body weight, and reported a higher QoL in all four aspects of the MacNew questionnaire prior to OUT-III [20]. More details concerning the EPICURE study population at baseline are shown in Table 1. The data in Table 1 are presented as mean (SD) and P-values were calculated with a t test. MacNew quality-of-life scores were assessed based on a German version of the MacNew quality-of-life questionnaire with subcategories regarding physical limitations in daily living (physical), emotional state of the patient in regard to the cardiovascular disease (emotional), effects of the cardiovascular disease on the social life and social activities (social) and an overall score over all subcategories (global) [21]. All questionnaires were recorded retrospectively post OUT-III, except for MacNew, which was recorded before and after the study duration. The structure of the analyzed data is described in the following.

The maximum mechanical power P_max (W) as achieved by patients during the exercise stress test at their regular assessments in the study centers pre- and post OUT-III was analyzed. The stress test was conducted through ergometry following the Austrian guidelines for exercise testing [26]. The difference ΔP_max (W) between P_max at the end of OUT-III minus P_max at the beginning of OUT-III was determined. If no data from the assessment before OUT-III was available, data from the assessment post phase II were taken instead since, in Austria OUT-III is expected to immediately follow phase II and, therefore, the patients P_max and questionnaire results are not expected to change between the end of phase II and start of OUT-III [27-29].

For each of the parameters described in Dataset section, we performed 2 analyses, (1) comparison of the 2 groups with and without digital tools and (2) correlation with ΔP_max. Where data for both, pre- and post OUT-III, were available, the change in the respective parameter during OUT-III was determined and group analysis and correlation were applied on data pre OUT-III, data post OUT-III, and the change during OUT-III.

Normal distribution of the data was tested by the Shapiro-Wilk test. A Student t test was applied to test for global differences between pre- to post cardiac rehabilitation (dependent t test) and for differences between the groups (independent t test). A value of α<.05 was considered significant. Where appropriate, we included confounding variables in the ANCOVA (analysis of covariance) models as covariates, to control for theses confounders. Confounding variables were all variables which significantly differed between patients without versus patients with digital tools determined by t tests. Changes pre- versus post cardiac rehabilitation were analyzed with boxplots and 2-tailed t tests based on matched pairs. The correlation coefficients “r” were calculated using Pearson’s correlation coefficient and the significance of the influence was calculated using ANCOVA. If not stated otherwise, the cofounders considered in the ANCOVA were the age, P_max, body mass index, and the global MacNew score, all measured at the start of OUT-III.

The EPICURE study protocol was approved by the ethics committee of Upper Austria (vote nr. 1165/2019) and registered at ClincalTrials.gov (Identifier: NCT04458727). All analysis performed during this work are covered by this ethics approval. All participants received oral and written information before the study entry and provided written informed consent to participate in this study. This informed consent also covered the analysis of secondary study objectives performed in this paper. All data were recorded in a pseudonymized way and anonymity was ensured for all participants during presentation of the findings. Patients did not receive any financial incentives for participating in the study.

Table 4 shows results per subsection of the Intrinsic Motivation Scale (IMS) as well as the aggregated score over all 4 categories. A significant difference in the item group concerning perceived competence was found between patients with versus without digital tools (P=.05), however, no significant correlation with ΔP_max was identified (P=.57). In addition, the IMS score for interest and enjoyment and the total IMS score showed borderline significant differences between the groups. In all significant and borderline significant cases, the group with digital tools achieved higher scores.

As summarized in Table 5, we did not identify a significant difference between the patients without and with digital tools concerning the self-efficacy (P=.95). In addition, there was no significant correlation between self-efficacy and the change in exercise capacity ΔP_max (P=.16). We did not identify any significant differences between patients without and with digital tools regarding the intention, control, and planning scores. However, the planning score was significantly correlated with ΔP_max (P=.04, r=0.14). Table 5 also shows results concerning the patients’ expected health benefit of physical activity (outcome expectations) and cardiac risk of nonadherence to physical activity (risk perception). Borderline significance (P=.06) was found between outcome expectations and ΔP_max. Patients with digital tools reported to expect significantly higher cardiac risks, if they were adherent to physical activity (P=.03).

As shown in Figure 2, patients with versus without digital tools significantly differed in five out of 17 EARs related questions. The left of Figure 2 shows the percentage of patients per group, answering with Yes or No as indicated left to the bars. Yes or No are aligned alternatively, depending on positive or negative phrasings of the respective questions, so that end-to-end bars relate to positive behavior. The right of Figure 2 shows the difference of the score per item between patients with versus without digital tools, including the corresponding P value.

Concerning adherence-related items, patients without digital tools reported to do their exercises significantly less often when tired (item 9). Patients with digital tools reported to forget to do their exercises significantly less often (item 15).

Concerning reasons for nonadherence, item 12 ("I do my exercises because I enjoy them") was significantly more important in patients with digital tools. Pain-related items were more important in the group without digital tools (item 14 and item 17), although high scores of item 14 represent low adherence, while high scores for item 17 relate to high adherence.

The overall EARs score showed no significant difference between the patients without (11.80 [2.7]) and patients with (12.80 [2.79]) digital tools (P=.5). No significant correlation between the overall score and ΔP_max was found (P=.21, r=0.07). Also, the adherence specific score as calculated by items 1, 3, 8, 10, 11, and 15 resulted in no significant difference between patients without (3.79 [1.71]) and patients with (4.38 [1.58]) digital tools (P=.57). No correlation between the adherence specific score and ΔP_max was found (P=.27, r=0.01). These scores were calculated with the sum of the selected items (0…no, 1…yes).

In this work, we analyzed the secondary hypotheses of the EPICURE study (primary outcomes have been published previously [18]), that is, the effects of digital tools in OUT-III. Our main findings are that, according to the MacNew questionnaire, QoL improved significantly during OUT-III, with similar improvement in patients without and with digital tools indicating that the use of digital tools did not significantly influence the increase of QoL during OUT-III. Furthermore, we found that patients with and without digital tools significantly differed in the following aspects:

The previously published EPICURE paper [18] summarized that, overall, exercise capacity improved both in patients with and without digital tools. The exercise capacity was measured during ergometry in 5 different Austrian rehabilitation centres. No data concerning the inter-rater reliability between the centres was available for analysis. When comparing changes with t tests, patients with digital tools improved significantly more than patients without digital tools. However, this change did no longer reach statistical significance when correcting for confounders with an ANCOVA analysis. In this paper, wherever suitable, P values were based on ANCOVA corrected for confounders.

Most questionnaire data were recorded at the time of study entry of the patients, that is, post OUT-III. Therefore, except for those questionnaires that were routinely recorded during cardiac rehabilitation (eg, MacNew questionnaire prior to OUT-III), all questions dealing with data before the end of OUT-III need to be interpreted with care, since the answers might have been influenced by the patients’ experiences during OUT-III and different answers might have been given in a prospective setting.

During sample number calculations in the study preparation phase, we assumed that approx. A total of 50% of patients would report to have performed cardiac rehabilitation with and without digital tools, respectively. However, questionnaire data revealed that approx. One third did and 2 thirds did not use digital tools, which reduced the power of our analyses. Since especially the number of patients using digital tools was rather low, additional studies with larger sample size may be indicated. In addition, not all patients had valid data at the start of OUT-III. For these patients data from the end of phase II was used with the assumption that the relevant physiological and psychological characteristics did not change in the short period between phase II and OUT-III. For future work, a prospective study is warranted that randomizes patients to either a group with or without digital tools, preferably including stratification on those baseline parameters that were found to have significant influence on cardiac rehabilitation outcomes, that is, age, P_max, BMI, body weight, and QoL.

We conclude that digital tools used during OUT-III including home-training can increase perceived confidence and improve joy during cardiac rehabilitation. In addition, digital tools can help to not forget exercises and to do the exercises even when tired. Concrete planning is correlated with better improvement in exercise capacity and should be fostered during OUT-III.