Table of Contents
This was a two-arm, mixed method, single-blinded superiority randomized controlled comparative effectiveness trial where we collected 12-month longitudinal data. The hypothesis was that a referral to TelePR would lead to lower 6-month rehospitalization, or death compared to a referral to standard, office-based pulmonary rehabilitation (SPR). The secondary outcomes included changes from baseline to post-PR sessions (i.e., day 1 and 8-weeks) and tertiary outcomes included update and adherence to PR after referral, longitudinal changes (6- and 12-months) in symptoms and whether patients would continue to exercise. Figure S1 provide overview of participant flow.
Outcomes were chosen based on the known effect of PR in COPD persons, and clinical relevance to patients and providers. Where available, we selected measures validated for use in both English/Spanish.
Participants included African American or Hispanic persons who were hospitalized for a COPD exacerbation at one of 9 hospitals in the New York City metropolitan area. Eligibility criteria included: COPD diagnosis, Hispanic or African American ethnicity/race, and current hospitalization for COPD-exacerbation. Potential participants were identified via the electronic medical record.
Participants were invited to enroll in a program to assist with management of their COPD, a referral to SPR and survey follow-up over the course of 1 year. Participants were shown an English/Spanish testimonial video depicting SPR or TelePR during recruitment. If they agreed, they signed consent form 1 which allowed for prospective collection of data over the course of 1 year from the EMR and via phone call, as well as a referral to PR. Participants were then randomized to receive the PR component of the program via either (1) a referral to SPR or (2) a referral to TelePR. As per the modified Zelen Randomized Consent Form (MZRCF) method , only those randomized to TelePR were informed that TelePR was being offered instead of SPR and then invited to sign consent form 2 (Fig. S2). If the patient declined to sign consent form 2, they would be followed over the course of 1 year and analyzed in the TelePR arm, and the patient would receive a referral to SPR (per consent form 1).
The MZRCF method allows researchers to obtain consent from patient participants for longitudinal follow-up in standard of care and then to randomly assign study participants to either the intervention (TelePR) or the control arm (SPR), for which additional informed consent must be obtained for those enrolled in the technology-based arm (i.e., nonstandard arm, consent form 2). We predicted that if a study participant was randomly assigned in a conventional way and did not receive the “high-tech” TelePR intervention, it was likely that they would refuse to continue in the study.
After enrollment, participants needed to receive medical clearance to be able to participate in PR. This required confirmation of COPD by a pulmonary function test (PFT) or a pulmonologist’s clinical diagnosis and a medical provider’s determination that the patient could exercise safety. These determinations required an in-person visit to the patient’s pulmonologist and, for those with cardiac medical comorbidities, an additional visit to a cardiologist for medical clearance.
Randomization was carried out in permuted blocks and stratified by enrollment site and by race/ethnicity. Both the biostatistician performing outcomes analyses, and the clinicians who were providing medical clearance were blinded to study allocation/randomization.
Participants in the TelePR arm had PR delivered via telehealth in either the participant’s home or community center (if preferred and depending on space in their homes). All the equipment including a full-size recumbent bicycle, weights, stretch bands, vital sign monitor and tablet computer with Wi-Fi card was delivered to patient homes. Prior to the first session, the RT met participants in their home/community center for training on device usage and to check oxygen supplementation devices. TelePR sessions were conducted by the RT with up to 3-participants simultaneously, via a secure HIPAA-compliant server using Zoom web-conferencing technology (Fig. S3). Participants received a Nonin® watch that transmitted vital signs directly to the platform for continuous monitoring during the PR sessions. A pulmonologist was on-call during each session in case of an emergency. The same educational videos used in SPR were shown while participants exercised. The exercise program (tracking, duration, progression etc.) was developed to parallel SPR.
After completion of 8-weeks of PR, all equipment was recovered from the TelePR participants. Both the TelePR and SPR participants were provided with an exercise-peddler and a list of community centers, gyms, and different SPR locations to encourage continuation of exercise.
Active control- standard pulmonary rehabilitation
For those enrolled in the SPR arm, two SPR sites were made available to participants within the study geographic area: Northwell Health Physician Partners Pulmonary and Sleep Medicine at Lake Success, New York and Glen Cove Hospital Outpatient Pulmonary Rehabilitation Program, Glen Cove, New York. We recruited patients within the metropolitan New York area – depending on where the patients lived – these centers were either within a 20 min or 1 h commute for patients. SPR facilities are equipped with exercise equipment, vital sign monitors, and supplemental oxygen devices, and are staffed by a team of RTs, other clinicians, and administrators. Educational lectures are given as part of the PR sessions.
Participants in both arms received $175 for their time completing longitudinal surveys. TelePR was provided at no cost to the participant. However, participants in SPR were required to pay co-payments based on insurance laws and SPR was charged to participants’ health insurance.
TelePR was provided at no cost to the patient; however, SPR was charged to participants’ health insurance, and co-payments were required. Both arms needed to have medical clearance appointments submitted to their insurance carriers and to pay co-payments when applicable. During the initial consent (consent form 1), all participants were made aware of this real-world requirement for SPR. It was not until the participant was randomly assigned to TelePR that the research team explained that the intervention could not be paid for by insurance and therefore would be paid by the study grant. Before the start of the program, the social worker and research team had discussed sliding-scale payment options with the health system and worked closely with the medical billing departments to assist with insurance navigation for participants.
Transportation costs were covered to and from clinic visits for medical clearance appointments and the SPR sessions. When possible, the social worker attempted to leverage existing insurance-subsidized transportation programs to offset the cost to the patient.
Exercise training content and progression for SPR and TelePR
Each PR session was approximately 60-min, consisting of 30-min of aerobic exercise on a treadmill (SPR) or bicycle (TelePR), 20-min of anaerobic exercise and 10-min of cool down. The bilingual respiratory therapist (RT) developed an individualized exercise program for each participant based on exercise capacity, and documented progress using standardized forms paralleling those used in SPR.
A bilingual social-worker maintained contact with each participant from the time of hospital discharge to completion of the PR program to assist with identifying insurance-subsidized transportation programs, insurance navigation for participants without insurance, and arranging clinic appointments.
The RCT outcome measures are presented within the RE-AIM (Reach, Effectiveness, Adoption, Implementation and Maintenance) framework below.
Table 1 provides a summary of the different dimensions of REAIM and the aspects that are being measured in the study.
The Reach of an intervention measures whether the intended/target population was reached by the program, and whether the intended population expressed interest in participating in the program. This is outlined in the CONSORT diagram. To increase the likelihood that our intervention would appeal to our target population, we convened a study-specific Community Advisory Board (CAB) comprised of patients and caregiver’s representative of our target communities, directors of PR clinics, and clinicians. The CAB provided input on initial study protocols, modifications to increase recruitment and retention, and the acceptability of the equipment that was used in TelePR including testing the bikes and technologies associated. Additionally, as part of the clinical trial, we conducted interviews and focus groups with participants who had different levels of adherence to PR to identify and address barriers they encountered.
Our primary outcome was a composite of COPD-related hospital readmissions/death within 6-months of discharge, based on the mean duration of 6-month follow-up in the systematic review used for sample-size calculation .
Our secondary outcomes included changes from baseline to post-PR sessions (i.e., day 1 and 8-weeks) in: perceived symptom control and Quality of Life (QOL) (CAT, MMRC) [11, 12]; self-efficacy to manage symptoms (COPD-Self Efficacy Scale) ; COPD knowledge (BCKQ) ; self-reported depression, fatigue, social support, and anxiety (PROMIS-forms) [15, 16]; Functional Capacity (TelePR: 6MWT (meters) and SPR: 2MST (steps), and perceived exercise capacity (measured via the Modified Borg Scale) [17,18,19].
We further assessed differences in uptake (participating in at least one PR session, which is described as ‘adoption’ in RE-AIM) and adherence (completion of the program and number of sessions attended of the 16-total PR sessions). In addition, we measured CAT, MMRC, PROMIS, and whether participants continued with exercise (yes/no self-report) at 6- and 12-months, and feasibility of equipment delivery and function in TelePR. We also recorded technical or other barriers to TelePR session completion, and participant satisfaction with the program (Table S1).
Sample-size and power calculation
The 276-person sample-size was based on an effect size of 0.3 odds ratio (OR), with 20% loss to follow-up and 80% power to detect superiority of referral to TelePR compared to SPR for the 6-month COPD readmission/mortality outcome using a two-sided chi-squared test at a significance level of 0.05 .
There were three levels of analysis: Intention to Treat (ITT), which included all the people randomized at the beginning of the study (excluding those who were later found to not met inclusion criteria for referral, such as immobility after hospital discharge or PFT results that did not indicate COPD) (ITT); people randomly assigned to a study arm and medically cleared (Sub-analysis 1), and people who were randomly assigned, medically cleared, and who had participated in at least 1 PR session (Sub-analysis 2).
The ITT analysis compared outcomes for those randomized to TelePR vs. SPR, excluding those who would not have received a referral to PR in real-world practice (i.e., those who were later found not to meet inclusion criteria because they were immobile or did not have COPD, or who became medically unstable). We then performed 2 sub analyses: Sub-analysis 1 for patients who ultimately received medical clearances after referral (i.e., patients who would be allowed to participate in PR) and Sub-analysis 2 for patients who were medically cleared and then participated in at least 1 PR session.
Logistic regression analysis compared the odds ratio of the primary outcome, in 3 sets of models: (Model 1) intervention only with no other covariates added to the model (unadjusted); (Model 2) intervention, adjusted for race and clinical site (stratification variables); (Model 3) intervention, adjusted for race, clinical site, and risk factors shown to be associated with the primary outcome in the literature. We reviewed each study contained in the Cochrane Systematic Review [10, 20] and we identified 19 unique risk factors for COPD exacerbation admission: (1) depression [21, 22], (2) SES [23, 24], (3) heart disease , (4) male sex [22, 23], (5) nursing home residence , (6) age , (7) lower QOL , (8) prior hospitalization , (9) longer hospital length of stay , (10) higher number of comorbidities [24, 26], (11) need for long-term oxygen treatment , (12) poor lung functions [22, 24, 27], (13) marital status , (14) cor pulmonale [21, 28], (15) hypoproteinemia , (16) elevated PCO2 , (17) anemia , (18) low serum magnesium level , and (19) elevated C-reactive protein level . Of the available patient data in our study, 6 of the risk factors (cor pulmonale, hypoproteinemia, elevated PCO2, anemia, low serum magnesium level, and elevated C-reactive protein level) were not reliable, because many people did not have these laboratory values in their EMR and so could not be included in the analyses. These analyses were performed for ITT group, Sub-analysis 1 and Sub-analysis 2. Therefore, there were a total of 9 analyses. We analyzed the data with a two-sided alpha = 0.05.
Continuous variables were summarized using mean, median and standard deviation; categorical variables were summarized using frequency and percentages. Two-sample t-test or nonparametric Wilcoxon test compared the continuous variables, and Chi-squared test or Fisher’s exact test compared categorical variables. To compare continuous variables between day 1 and 8-weeks, we used a paired t-test or nonparametric signed rank test, and a generalized linear mixed models (repeated measures analysis of variance “MMRMA”) to determine whether there was a difference in the change over time between the arms, and if the magnitude of change depended on treatment arm (treatment x time interaction). Unstructured covariance was used in all the models. Adherence was separately included as a covariable to examine its role on the primary outcome in the models.
Missing data was handled using multiple imputation with details described in Appendix S1.
Qualitative methods and analysis
Qualitative interviews and focus groups conducted among a sample of participants allowed for a deeper understanding of (1) the barriers to initiating PR despite a referral to PR and (2) the barriers to participating in > 1 PR session once started (Appendix S2). Participants for the sub-study were recruited from among the 209 participants who had been enrolled into the wider study and represent those randomized to either TelePR or SPR. Interviews were conducted either in person or by phone by a member of the study team using the interview guides. Focus groups of participants were conducted in person by members of the study team using the focus-group guides. All interviews and focus groups were audio-recorded and transcribed by a professional medical transcribing company. Thematic analysis was performed by 3 members of the study team using the constant comparison method to create a codebook of themes with definitions, exemplary quotes, and exclusion and inclusion criteria. Transcripts were coded in NVivo Pro 12. A Kappa coefficient of less than 0.85 required discussion among the coders to resolve discrepancies and reach agreement.
Adoption measures the proportion of people/settings who were willing to initiate the intervention. Because the intervention was in participants’ homes, the proportion of people who initiated TelePR sessions in their homes is the relevant metric for adoption and for describing barriers to adoption. In addition, we report the proportion of community centers that were willing to ‘house’ the TelePR sessions for participants who did not have space for equipment in their homes.
To increase adoption, we held meetings twice a year (estimated 10 meetings) with local clinicians informing them about the TelePR program - within the context of the clinical trial - including an email communication across the entire Northwell Health System by the chair of medicine, and presentations at division meetings. We worked with our CAB to identify methods to increase referral by hospital staff (physicians and respiratory therapists) for potentially eligible patients, and to increase adoption by the COPD patients and their families once approached by the study team. These meetings informed our recruitment materials targeted specifically to people from predominantly underserved Hispanic and African American people with COPD in the NYC metropolitan region. Details are provided in a separate manuscript .
Implementation measures whether the intervention was delivered as intended. We include measurements of fidelity to clinical trial protocols, describing in detail any adaptations that were made based on CAB feedback as well as due to early findings in the clinical trial execution. We also report fidelity to the TelePR intervention components, to address specific factors relevant to TelePR sessions being executed as intended.
Our CAB provided recommendations for successful implementation of TelePR. This included improvements to equipment functionality, safety features for frail, older patients using the ergonomic stationary bike, providing a micro-key to the telehealth tablet computer to make it easier for older patients to turn it on and off and to access the features needed; and a laminated how-to sheet attached to the equipment. To increase retention among those patients who were enrolled in the program, we distributed a monthly newsletter, and a dedicated social worker helped participants obtain medical clearance appointments and associated considerations if they were re-hospitalized during the program.
Maintenance measures the continued use of the program over time. For this study, because it was funded by clinical trial grant monies, we measured maintenance on two levels. First, individual level maintenance of exercise by joining a gym or by continuing in a standard PR program. This does not reflect the value of TelePR directly but measures motivation that TelePR provided to exercise and the benefits of exercise. Second, we describe inquiries from health systems and pulmonary organizations to continue the TelePR programs, and the logistical and financial considerations that were discussed.