Feasibility, Safety, and Efficacy of Pediatric Cardiac Telerehabilitation in Adolescents With Heart Disease Open Access

Cardiac rehabilitation is exercise therapy for people with heart disease and is the standard of care for most adult patients, with thousands of programs available around the United States.¹ Adult cardiac rehabilitation programs decrease mortality and hospital readmissions and improve postoperative outcomes.^2–4 Similar to the adult population, children with cardiac disease have reduced exercise capacity and increased sedentary behavior when compared to healthy peers, which leads to reduced quality of life and increased risk of mortality and postoperative complications.^5–7 Findings over the past 30 years have demonstrated that cardiac rehabilitation improves aerobic capacity, quality of life, social emotional function, and overall physical health.^4,7–9 Despite the potential benefits of pediatric cardiac rehabilitation, availability of these programs is limited, with relatively few programs in the United States.¹⁰ Access to pediatric cardiac rehabilitation is further limited by a lack of recognition as a standard of care for children with heart disease, which is one of many barriers to more widespread adoption of these programs across institutions.^5,11 

Telerehabilitation, or cardiac rehabilitation delivered virtually, lowers these barriers by reducing travel requirements, increasing scheduling flexibility, and reducing space requirements.¹² However, due to the small number of programs available, data demonstrating the feasibility, safety, and efficacy of pediatric cardiac telerehabilitation (PCTR) programs are limited. To address this gap, we conducted a randomized controlled trial (RCT) to determine the feasibility, safety, and efficacy of PCTR in a clinically diverse population of adolescents with congenital heart disease.

We conducted a single-center, prospective, nonblinded RCT of a PCTR program in adolescent and young adult patients at a large pediatric heart center in New York City. Our primary outcome was feasibility, defined as ability to recruit more than 8 patients per month, greater than 75% of total sessions completed by enrolled participants, and greater than 50% of participants completing all 12 weeks of the program. Secondary outcomes included safety, as measured by adverse events, and efficacy of a virtual PCTR program as measured by improvements in formal cardiopulmonary stress test (CPET) parameters and psychosocial assessments. The study was approved by the Columbia University Human Subjects Protection Committee (IRB#-AAAT7872).

Eligibility criteria included ages 10 to 25 years with primary or secondary pulmonary hypertension (PH), congenital structural heart disease (CHD), or history of orthotopic heart transplant (OHT) who were determined by their primary cardiologist to be developmentally and physically capable of performing exercise safely. Diagnoses of isolated patent foramen ovale, patent ductus arteriosus, and coronary anomalies not requiring surgical intervention were not considered congenital heart disease for this study. Exclusion criteria were clinical instability as demonstrated by recent medication escalation (within 3 months prior to starting the program), hemodynamic instability as defined by exertional hypotension, exercise-induced arrhythmia defined as runs of nonsinus (ventricular or supraventricular) tachycardia of more than 3 consecutive ectopic beats, ST segment depression of greater than 1 mm in at least 2 contiguous leads or T wave inversion, and end-tidal carbon dioxide below 20 mm Hg with exercise, as an end-tidal carbon dioxide level below 20 mm Hg is associated with poor outcomes in patients with pulmonary hypertension.¹³ 

The initial sample size needed was determined to be 40 participants, with 30 in the experimental group (5 OHT, 10 CHD, 15 PH) and 10 in the control group (3 OHT, 3 CHD, 4 PH), based on relative availability of eligible patients and funding constraints.¹⁴ This feasibility trial design targeted a clinically diverse population including sicker patients typically excluded from prior studies^15,16 to assess recruitment, adherence, and safety metrics. Secondary outcomes, including exercise capacity, fear/anxiety of exercise, and quality of life, were exploratory and not powered for significance, as power analyses indicated the need for 70 participants for exercise time, 57 for fear/anxiety, and 113 for quality-of-life outcomes. The sample size will guide future studies aimed at addressing secondary outcomes with sufficient power.

The study team recruited participants from the heart center of Columbia University Irving Medical Center/NewYork-Presbyterian Morgan Stanley Children’s Hospital, which includes large PH, CHD, and transplant programs and is located in an ethnically and socioeconomically diverse area of New York City. Participants were also referred by their primary cardiologists. Exercise clearance was required from the primary cardiologist for all participants.

Eligible patients were recruited stepwise until goal numbers for control and intervention groups were met. English proficiency was not required. Following recruitment, participants were randomized to either PCTR or control groups in a 2:1 ratio (intervention:control) based on the order in which they were referred to the study. The ratio was chosen to maximize participation in the exercise intervention given the well-established benefits of exercise. Because of the known benefits of exercise, controls were offered the opportunity to participate in the intervention group after completing 12 weeks in the control group.

All participants performed a baseline CPET and baseline questionnaires within 3 months prior to enrollment. Testing was repeated upon completion of the intervention period for PCTR participants or 3 to 6 months after enrollment for controls.

Participants in the PCTR group were provided with an activity tracker with heart rate (HR) monitoring to be worn 24 hours per day (Fitbit Inspire 2, Google) and a single-lead, chest-based electrocardiogram (ECG) device (Wellue) to be worn while exercising. They also received a stationary bicycle, a weight set (2–3 pounds), and resistance bands. Controls received a Fitbit activity tracker and weekly check-ins by phone but did not receive a formal exercise prescription, guided exercise sessions, or additional encouragement to exercise. All participants were asked to wear their Fitbits for the duration of the study. Equipment and pricing are listed in Supplemental Material 1.

PCTR participants underwent “semisupervised” exercise sessions, during which, 1 day per week, the participants exercised one-on-one with a physical therapist (PT) or exercise physiologist (EP) on a virtual platform, Zoom; 2 days per week, they exercised independently. An adult had to be available in case of an emergency during each session. Compliance was measured by attendance of Zoom appointments, Fitbit activity data, Wellue single-lead ECG data, and self-reported activity performed. Efficacy was measured by CPET results, validated questionnaire data, and activity levels as determined by Fitbit data. We selected Fitbits as our activity monitor because they had been used in prior research investigating activity levels in pediatric patients with heart disease.¹⁷ 

The Exercise Sensitivity Questionnaire (ESQ-18) was used to assess the participants’ anxiety levels.¹⁸ The Participant Health Questionnaire modified for Adolescents (PHQ-A) was used to screen for depression.¹⁹ Self-efficacy was measured via the physical self-efficacy questionnaire for children.²⁰ Quality of life was assessed via the Pediatric Quality of Life Inventory (PedsQL) (version 4.0, SF15) using raw score data rather than the reverse scoring utilized in clinical practice.²¹ A member of the study team administered a 4-question survey of the caregivers relating to fear/nervousness of having their child exercise, if they encourage their child to avoid exercise, and if their child is worried that they will hurt themselves and worsen their heart problem when exercising. We determined the program feasibility and satisfaction using the “Feasibility, Acceptability and Satisfaction Program Questionnaire.”²² 

CPETs were performed using MedGraphics equipment (MGC Diagnostics Corporation) and used Mortara ECG software (Hillron Corporation) and pulse oximetry (Masimo Corporation). The exercise testing protocol consisted of a 2-minute baseline and then ramping Bruce protocol to volition, followed by a 2-minute active recovery and up to 10 minutes of passive recovery. Blood pressure was taken every 2 minutes throughout testing. Almost all tests were done on treadmill (pre- and post-) with the exception of 2 participants with pulmonary hypertension, who performed exercises using a 10-W ramping cycle ergometer test due to disease severity that consisted of a 2-minute baseline, 1-minute warm up, 10-W ramp to volition, 2-minute active recovery, and up to 10-minute passive recovery. A respiratory exchange ratio of 1.1 and HR of 85% of the maximum predicted were used to determine quality of the exercise effort. Data points were collected at the point of metabolic fatigue, commonly known as the anaerobic threshold (AT), and at maximal exercise. Participants on beta-blocker therapy were analyzed the same as participants not on beta-blocker therapy.

The PCTR protocol was conducted for 12 weeks and included 12 supervised sessions over Zoom and 24 additional days of independent exercise using HR guidelines and exercises provided during their one-on-one visit (Supplemental Material 5). Personalized exercise prescriptions were created based on HR, ECG data, and reported perceived exertion collected during the exercise test.²³ 

Exercise sessions with a PT and/or EP via Zoom consisted of at least 10 minutes of cycling with gradual increase to 30 minutes by the end of the study as tolerated as well as stretching and strengthening exercises using the resistance bands and weights. After each exercise session, the participant sent a digital file of their ECG to the study team and tracings were reviewed by the study cardiologist. Abnormal findings and symptoms were brought to the attention of the primary cardiologist, and adjustments to the exercise prescription were made as needed. All PCTR participants met with the study cardiologist midway through and upon completion of the program to assess symptoms and safety, answer questions, and provide encouragement to lessen anxiety surrounding exercise.²⁴ 

The EP working with the participants provided supplemental exercises or videos in areas of interest to increase self-efficacy and decrease exercise anxiety.² Areas of interest included yoga, basketball, dance, soccer, and weight training.

Gamification was introduced midway through the study to improve compliance.²⁵ Participants were provided with a points system and an interactive progress meter (Supplemental Material 2). As they completed workouts, made appointments, and maintained target HR zones during workouts, they earned points, which were redeemable at the end of the program for prizes of different values. All participants started gamification at the same time.

The normality of continuous variables was evaluated through histograms, normal probability plots, and Anderson-Darling tests. Descriptive statistics were presented as counts and percentages for categorical variables and median (25th and 75th percentiles) for continuous data with skewed distributions. Continuous data with skewed distributions between groups were compared using Wilcoxon rank-sum tests. Comparisons between categorical variables were performed using χ² tests, or the Fisher exact test when expected cell counts were less than 5. To evaluate the efficacy of the intervention, we analyzed changes before and after the intervention within the intervention and control groups using signed-rank tests and changes between these groups using Wilcoxon rank-sum tests. Further analysis of the primary exercise outcome measures was conducted using generalized estimating equations (GEEs), adjusting for the effects of age, sex, and diagnosis (including pulmonary arterial hypertension, CHD, and OHT) on between-group changes. The GEE model was designed to compare differences between the intervention and control groups over time by evaluating the interaction between the intervention and time (intervention × time). Separate adjusted GEE models were used to assess the impact of the intervention on key outcomes including AT exercise time (seconds) and AT HR (bpm). Covariates such as age and diagnosis were included to control for baseline differences, and an exchangeable correlation structure was applied to account for within-subject correlations across repeated measurements.

Fitbit data were collected using the third-party platform Fitabase and analyzed by determining the change in steps per day measured by finding the average steps per day from the first 14 days of the program and the average steps per day for the last 14 days of the program. Based on prior research²⁶ we categorized patients’ step data into 3 groups: effective improvement (≥1000 increase in steps per day), ineffective improvement (1–1000 increase in steps per day), and no improvement (no increase in steps per day). A multinomial GEE model with an exchangeable correlation structure was used to assess the effect of the intervention over time, incorporating interaction terms (intervention × time) and least-squares means for group comparisons. Data analysis was performed using SAS software version 9.4 (SAS Institute Inc), with statistical significance defined as a P < .05 and 95% CIs calculated for all estimates.

Referral, enrollment, and randomization pathways are depicted in Figure 1. Programs referred more boys than girls, but a higher percentage of girls were willing to participate than boys. Despite randomization, more girls than boys dropped out of the intervention group, resulting in more girls in the control group than in the intervention group and more dropouts from the intervention group overall (5 participants) relative to the control group (1 participant). The median age of the 43 participants was 15 years (IQR 13–17); 37% were female, 44% had PH, 35% had CHD, and 21% had undergone OHT. The CHD group included participants with Fontan physiology, pulmonary atresia, double outlet right ventricle, tetralogy of Fallot, Ebstein anomaly, atrioventricular septal defect, and anomalous origin of the left coronary artery (Table 1).

Eighty-seven percent of participants completed all 12 supervised sessions. More girls dropped out of the study than boys (5 girls, 1 boy). Reasons for dropping out as reported by participants included lack of interest in exercise (N = 2), considerable time commitment (N = 3), and irritation from Fitbit band (N = 1). Participants attended an average of 87% of the Zoom visits. Participant satisfaction was high, with average ratings of 4 or 5 out of 5 with 5 being the most satisfied for all criteria (Supplemental Material 3).

Four participants experienced minor adverse events—3 with exercise. One participant with PH on chronic anticoagulation dropped out due to recurrent nosebleeds with and without exercise and was excluded from the study. One participant with CHD demonstrated asymptomatic ventricular ectopy while weightlifting beyond what was recommended in his exercise prescription. The team reinforced the prescribed training regimen, after which the ectopy resolved. One participant with PH described chest pain with exercise when he exceeded his prescribed HR range. Symptoms resolved with adherence to the prescription. One control participant experienced skin irritation from the Fitbit wristband and chose to withdraw from the study.

The PCTR group demonstrated a significant improvement in anxiety with exercise as demonstrated by a median change in ESQ-18 score of −8 (P = .01; 95% CI −16 to 2) as opposed to no significant change demonstrated in the Fitbit-only group. The PCTR group also demonstrated improved quality of life as demonstrated by a significantly lower PedsQL score after intervention relative to before intervention, with a median change of −5 (P = .01; 95% CI −5 to 7) (Figure 2, Supplemental Material 4). Fitbit-only participants demonstrated significantly lower perceived physical ability score (PPA), which corresponded with improved self-efficacy. The PCTR group demonstrated an overall positive but not statistically significant change in PPA score (Supplemental Material 4).

The PCTR group demonstrated significant improvement in time to the AT relative to preintervention findings and demonstrated a statistically greater improvement than that achieved by the control group. HR and oxygen consumption (VO₂) at the AT were also significantly greater in the PCTR group after intervention than the control group (Table 2). The GEE models revealed that the intervention had a significant positive impact on AT exercise time and AT HR over time, as indicated by the significant interaction term (intervention × time) (P = .01 for both outcomes). The adjusted increase in AT exercise time was 89.8 seconds (95% CI: 21.9–157.7), and the increase in AT HR was 10.5 bpm (95% CI: 2.3–18.7). The observed improvements were driven by the interaction between intervention and time (Table 3).

There were no statistically significant differences in baseline (P = .06) or post-study (P = .40) median step counts between the PCTR and control groups. There were also no statistically significant changes from pre- to postintervention step count within the control (P = .86) or PCTR groups (P = .12). Similarly, there was no statistically significant difference in degree of change in number of steps between control and PCTR, or in the number of those who demonstrated “negative,” “ineffective,” or “effective” change (P = .34).

To our knowledge, this study represents the first RCT of PCTR in a clinically diverse group of adolescents with CHD. PCTR was safe and effective at improving exercise capacity regardless of cardiac diagnosis. Participants demonstrated excellent compliance of 87%, higher than the 50% compliance rate reported for in-person cardiac rehabilitation programs.^27,28 This was likely related to a combination of monetary incentives, gamification, and high accessibility of the study team members who maintained regular contact and check-ins with the participants. These results support superior compliance using telehealth vs in person care for exercise therapy.²⁹ 

A strength of this study was the recruitment of a clinically heterogeneous participant population concerning cardiac diagnosis. Prior work in pediatric cardiac rehabilitation has limited inclusion to single diagnoses,^7,30 which can make it easier to demonstrate the impact of exercise training on exercise capacity because the participants are more likely to demonstrate similar changes. By expanding the participant population, we increased the risk of weakening our findings due to medical variability and provided data on how outcomes differ among different subpopulations. By including higher-risk participants, such as those with a history of ectopy or high pulmonary artery pressure, we demonstrated that with careful risk stratification, customized exercise prescriptions, and appropriate monitoring, patients who have previously been excluded from telerehabilitation can safely participate.⁴ 

Our findings are consistent with the literature in adult patients that describes cardiac telerehabilitation programs as feasible and safe.⁸ Studies in pediatric populations with obesity but without heart disease have shown benefits from these types of virtual exercise programs in terms of physical fitness, body mass index, blood pressure, and left ventricular systolic function.^31,32 Noncontrolled studies of exercise interventions in the adolescent population demonstrated that exercise training could improve quality of life in children with CHD⁹ and exercise performance in participants with single-ventricle disease who have undergone Fontan palliation³³ or OHT¹¹ and PH.^34,35 A study by Aronoff et al¹⁵ found that patients who participated in PCTR demonstrated better compliance than those who participated in in-person sessions; however, they also found that in-person programs demonstrated a significant improvement in peak VO₂ where telerehabilitation did not. The results of this study, and others that have analyzed maximum oxygen consumption changes in response to exercise,³⁶ are consistent with our findings that peak VO₂ was not improved. Aronoff et al did not analyze submaximal parameters; therefore, it is unknown if submaximal parameters would have demonstrated a similar benefit. Additionally, the sample size in the telerehabilitation program was only 10 patients; thus, it may not have been powered to find significant differences in peak VO₂. Another strength of this study was the comparison between Fitbit plus weekly contact vs our guided exercise program. Activity monitors such as Fitbits provide feedback to the participant on their degree of physical activity or inactivity, which can itself significantly improve physical activity behavior.³⁷ Therefore, we had to outperform the impact of Fitbit alone. Health-related quality-of-life measures and exercise-related anxiety measures responded well to the PCTR program, whereas PPA seemed to respond more to Fitbit feedback alone. The PPA measure in the PCTR group did trend toward a positive change.

Markers of improved fitness at the AT were also more responsive to the PCTR program vs Fitbit alone. This finding aligns with the personalized exercise prescription given to the PCTR group, targeting an exercise intensity that was at or just below the AT.

The intragroup improvements in exercise capacity in response to PCTR demonstrated by our population are similar to those reported by Rhodes et al,⁷ who investigated the impact of in-person cardiac rehabilitation in participants with CHD. The Rhodes group described improvements in O₂ pulse and metrics at the AT. Changes in the AT appeared to demonstrate the greatest sensitivity for capturing the impact of exercise therapy on exercise capacity. These findings are similar to those described by the FUEL trial that tested the impact of udenafil on exercise capacity.³⁰ In FUEL, peak VO₂ did not significantly change, but the metrics at the AT improved significantly more than in the control group. The authors theorized that this finding was due to a critical ceiling in maximal cardiac output from the underlying disease. Improvements in the AT are likely attributed to peripheral muscle adaptation with improved metabolic efficiency.

Additionally, the results demonstrated positive changes in fear/anxiety with exercise and quality-of-life scores within the PCTR group, which is similar to those reported in children⁸ and young adults^36,37 with CHD who underwent in-person cardiac rehabilitation interventions. Kroll et al⁸ reported that certain components of health-related quality of life were significantly improved by cardiac rehabilitation. Early findings of the QUALIREHAB RCT³⁸ of 142 participants with CHD also demonstrated a significant improvement in overall quality-of-life measures, similar to our findings.

Our findings can enhance the clinician’s ability to advocate for insurance reimbursement for PCTR. Insurance companies often require data from RCTs demonstrating the clinical impact of a therapy to support insurance coverage for the treatment.³⁹ These are costly and take many years to complete, but studies such as this can help clinicians advocate for reimbursement and support the rationale for larger multicenter studies. The need for RCTs investigating the effectiveness of PCTR is particularly important for participants with CHD, which is not currently a covered indication for cardiac rehabilitation under the center for Medicare and Medicaid services⁴⁰ due to lack of RCTs demonstrating the clinical and financial benefit.³⁶ 

There is a similar paucity of data on the impact of pediatric cardiac rehabilitation in pre- and post–heart transplant recipients. The findings from this work support the results of prior studies³⁹ and the clear benefit of cardiac rehabilitation in this population. The “ACTION initiative” survey for pediatric cardiac rehabilitation found that although 90% of centers were interested in having a program, only 52% of the 28 pediatric transplant/ventricular assist device centers had a program available to their participants.³⁹ We hope that with more high-quality RCTs, there will be greater support for payer reimbursement and fewer barriers to building pediatric programs.

Despite the many strengths of this study, there were also limitations. The researchers were not blinded, as those performing the CPET also administered the program. The participants could not be blinded either due to the nature of the exercise intervention.

The final sample size was 43 participants, with 19 PH, 15 CHD, and 9 OHT, and a less than 2:1 intervention to control ratio due to unexpected participant dropouts and recruitment challenges, with more dropouts from the PCTR group as compared with the control group. Although the sample size was not powered to detect significant changes in secondary outcomes, this limitation was anticipated, and the collected data will be used to inform future research with appropriate power. More boys were willing to participate in the intervention group than girls. Of the 5 participants in the intervention group who dropped out, 4 were female. Boys who were randomized to the control group were also more likely to request to be in the intervention group after completing their 3 months as a control.

Additionally, although English language fluency was not a requirement, all participants spoke English, which may limit the generalizability of our results to patients who do not speak English. Our patient population was also generally of normal weight and body mass index, with some exceptions. However, a guided exercise program may show an even greater effect in a population with a higher rate of overweight or obesity.

Relying on Fitbits also posed several barriers. Devices were sometimes removed during sleep or bathing and not replaced after. One participant lost their Fitbit but was hesitant to notify the study team. Two participants lost their Fitbit charger, resulting in paused data collection until replacement chargers were delivered. Younger participants wore the device and recorded data more reliably, likely due to assistance from their parents. Steps were sometimes overcounted: for example, while participants were playing video games or a musical instrument, or undercounted, such as when performing certain activities that did not involve traditional “steps,” such as cycling. For this reason, we felt that reporting change in step counts from the pre- to post-study period was more clinically useful than using absolute step counts for intergroup comparisons. Additionally, Bluetooth range or internet was limited for some families, so children had to be able to exercise within a certain range of their phone or tablet. One family obtained a free SIM card to enable internet access for study purposes. Had there been families who did not have internet access and needed assistance, we had funding in our budget to assist with these internet access issues. In some districts schools may provide internet-enabled devices as well.

This first-of-its-kind RCT demonstrates that PCTR is feasible, may be performed safely, and is effective at improving submaximal exercise capacity, anaerobic threshold parameters, and quality of life in young patients with heart disease similar to those enrolled in our study. There was high compliance and patient-reported satisfaction. The results of this study support the need for larger-scale trials in this field and expanded access to cardiac telerehabilitation for this highly vulnerable population.

ASI

anxiety sensitivity index

AT

anaerobic threshold

BMI

body mass index

BPM

beats per minute

CHD

congenital heart disease

CMS

Centers for Medicare and Medicaid Services

CPET

cardiopulmonary stress test

DBP

diastolic blood pressure

ECG

electrocardiogram

EP

exercise physiologist

ESQ-18

Exercise Sensitivity Questionnaire

GEE

generalized estimating equations

HR

heart rate

O₂ pulse

oxygen pulse rate

OHT

post-orthotopic heart transplant

PCTR

pediatric cardiac telerehabilitation

PedsQL

Pediatric Quality of Life inventory

PH

pulmonary hypertension

PHQ-A

Participant Health Questionnaire Modified for Adolescents

PPA

perceived physical ability

PT

physical therapist

RCT

randomized controlled trial

SBP

systolic blood pressure

SpO₂%

percentage oxyhemoglobin saturation

VO₂

oxygen consumption