comparing two methods of caring for black and hispanic ... · heart failure (hf) is a progressive,...
TRANSCRIPT
1
Comparing Two Methods of Caring for Black and Hispanic Adults with Heart Failure after They Leave the Hospital
Renee Pekmezaris, PhD1,2,3,4; Christian N. Nouryan, MA1,2,3,4; Rebecca Schwartz, PhD1,2,3; Stacy Castillo, RN5; Amgad N. Makaryus, MD1,4,5; Deborah Ahern, NP5; Meredith B. Akerman, MS1,4; Martin L. Lesser, PhD1,4; Lorinda Bauer, RN5; Lawrence Murray, LMSW6; Kathleen Pecinka, RN, MSN7; Roman Zeltser, MD1,4,5; and Paola DiMarzio, PhD, MPH1,2,3,4
1. Department of Medicine (RP, CNN, PDM), Northwell Health, Manhasset, NY 2. Department of Occupational Medicine, Epidemiology and Prevention (RP, CNN, PDM, RS, MLL), Northwell
Health, Great Neck, NY 3. Zucker School of Medicine at Hofstra-Northwell, Hempstead, NY (RP, CNN, RS, PDM, MLL) 4. The Feinstein Institute of Medical Research, Manhasset, NY (RP, PDM, CNN, MLL, MBA) 5. Heart Failure Center, Nassau University Medical Center, Hempstead, NY (SC, ANM, DA, LB, RZ) 6. Community Advisory Board, Northwell Health (LM) 7. Queensborough Community College, Bayside, NY (KP)
Original Project Title: Telehealth Self-management Program in Older Adults Living With Heart Failure in Health Disparity Communities PCORI ID: AD-1304-6294 ClinicalTrials.gov ID: NCT02196922
_____________________________________
To cite this document, please use: Pekmezaris R, Nouryan CN, Schwartz R, et al. Telehealth Self-management Program in Older Adults Living With Heart Failure in Health Disparity Communities. Washington, DC: Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/10.2019.AD.13046294
2
Table of Contents
ABSTRACT .................................................................................................................................................................. 4
BACKGROUND .......................................................................................................................................................... 6
PARTICIPATION OF PATIENTS AND OTHER STAKEHOLDERS IN THE DESIGN AND CONDUCT OF RESEARCH AND DISSEMINATION OF FINDINGS ............................................................................................. 9
METHODS ............................................................................................................................................................... 12
Study Design......................................................................................................................... 12
Aim 1 Methods ..................................................................................................................... 12
Prestudy Usability (Stakeholder Perspective, Focus Groups 1 and 2) ................................. 13
In-home Patient Usability Perspective (Focus Group 3) ...................................................... 13
Aim 2 Methods ..................................................................................................................... 14
Comparator 1: TSM .............................................................................................................. 17
Comparator 2: COM ............................................................................................................. 19
RESULTS .................................................................................................................................................................. 26
Specific Aim 1: To Assess Usability and Adapt TSM to Facilitate Acceptability and Feasibility in a Population of Underserved Black and Hispanic Patients ............................. 26
Specific Aim 2: To Measure ED Utilization, Inpatient Utilization, and QoL of Patients Receiving TSM vs COM in a Population of HF Patients Discharged from Acute Care (Table 1) .......................................................................................................................................... 28
ITT Analyses (Tables 2-4) ...................................................................................................... 30
Post Hoc Analysis 1: Patients Not Receiving Home Care (Tables 2-4) ................................. 38
Post Hoc Analysis 2: Patients Hospitalized at Least Once (Tables 2-4) ............................... 39
Post Hoc Analysis 3: ED and Hospitalization by Heart Failure Class (Table 5) ..................... 40
Post Hoc Analysis 4: Utilization by Subpopulations (Black/Hispanic; Table 6) .................... 42
Post Hoc Analysis 5: Adherence and Utilization of Study Intervention (TSM Group Only; Table 7) ................................................................................................................................ 43
DISCUSSION ........................................................................................................................................................... 46
Decisional Context ............................................................................................................... 46
3
Study Results in Context ...................................................................................................... 46
Implementation of Study Results ........................................................................................ 47
Generalizability .................................................................................................................... 48
Subpopulation Considerations (Post Hoc Analyses) ............................................................ 48
Study Limitations ................................................................................................................. 49
Future Research ................................................................................................................... 50
CONCLUSION ......................................................................................................................................................... 51
REFERENCES .......................................................................................................................................................... 52
ACKNOWLEDGMENTS ........................................................................................................................................ 56
PUBLICATIONS ...................................................................................................................................................... 57
4
ABSTRACT Background
In the United States, black and Hispanic populations experience a high prevalence of heart failure (HF). To address this disparity, we sought to compare health care utilization and quality of life (QoL) for black and Hispanic patients admitted to a “safety net” hospital for HF. Patients were randomly assigned to receive a telehealth self-monitoring (TSM) intervention or comprehensive outpatient management (COM). Objectives
Aim 1. To assess usability and adapt TSM to facilitate acceptability and feasibility in a population of underserved black and Hispanic patients.
Aim 2. To compare inpatient and emergency department (ED) utilization and QoL of underserved HF patients randomized at discharge to TSM or COM. Methods
We used a mixed methods approach, including a randomized controlled trial (RCT), to assessTSM usability and effectiveness. Medical history, hospital utilization, demographics, depression, anxiety, and QoL were recorded at days 1 and 90. TSM patients transmitted their vital signs daily and attended a weekly video visit.
We analyzed binary outcomes for ED visits and hospitalizations using the standard chi-square or Fisher exact test. We also used Poisson or negative binomial regression, repeated measures analysis of variance, or GEE (generalized estimating equation) as appropriate. We based the choice of Poisson, overdispersed Poisson, or negative binomial on standard goodness-of-fit statistics (the deviance statistic). We computed associated 95% confidence intervals (CI) for these proportions and their differences using exact methods.
Results Aim 1. We used the ADAPT-ITT framework (Assessment, Decisions, Administration,
Production, Topical experts, Integration, Training staff, and Testing) to tailor the TSM intervention. The adaptation, based on data from 3 focus groups, theater testing, and a small pilot study, resulted in an acceptable and feasible intervention for the target patient population.
Aim 2. Of the 104 patients randomized, 31% were Hispanic, 69% black, and 41% female. Overall, 72% of patients reported incomes of < $10 000/year. Intention-to-treat (ITT) analyses revealed no significant utilization differences between TSM and COM groups for (1 or more) all-cause ED visits (RR (relative risk) = 1.37; 95% CI, 0.83-2.27) or hospitalizations (RR = 0.92; 95%
5
CI, 0.57-1.48) and mean length of stay (TSM: 5.2 vs COM: 3.6 days) within 90 days of discharge. The average number of all-cause hospitalizations greater than 90 days was significantly lower for COM patients (TSM = 0.78 vs COM = 0.55; p = 0.03). Finally, while QoL improved for both groups over time (TSM baseline = 62.7; 90-day = 36.3; COM baseline = 59.2; 90-day = 27.8; p = 0.5), COM patients reported a greater reduction of anxiety (TSM baseline = 50%; 90-day = 28%; COM baseline = 57%; 90-day = 13%; p = 0.05). TSM adherence was low: 50% of participants provided fewer than 10 transmittals of vital signs during the 90-day period. Conclusions
While the ADAPT-ITT framework was successfully utilized to tailor a TSM intervention for HF patients from black and Hispanic low-income communities, TSM was not associated with reduced all-cause 90-day ED and inpatient utilization. The mean number of all-cause hospitalizations was significantly lower for the COM group. QoL and depression did not differ between the 2 groups, while COM patients reported a greater reduction of anxiety over time. Limitations and Subpopulation Considerations
This single-center study may not be generalizable to other underserved HF patients. Future studies should address methods to improve compliance to improve TSM treatment effect size on all-cause- and heart failure-related utilization outcomes.
6
BACKGROUND
Heart failure (HF) is a progressive, chronic disease characterized by a variety of
cardiovascular issues leading to cardiac dysfunction, with typical symptoms such as dyspnea,
fatigue, and congestion.1 HF is among the most commonly diagnosed chronic diseases of
Medicare beneficiaries, with the risk doubling every 10 years.2 HF is characterized by recurrent
periods of clinical exacerbation marked by high rates of emergency department (ED) and
inpatient hospital utilization, typically followed by discharge to home care. Once the home care
episode is complete, the patient returns to outpatient care, until the next exacerbation. This
costly, inefficient management cycle leads to poor health outcomes, including decreased
quality of life (QoL) and functional status and exorbitant health care costs, largely arising from a
30-day all-cause (Medicare) rehospitalization rate of 24.7%.3
Blacks experience a disproportionately high prevalence of HF at an early age (< 50)
compared with whites.4-6 Likewise, Hispanics with HF are diagnosed younger and die earlier
than non-Hispanic whites.7 In black and Hispanic patients, the etiology of HF is more likely to
be attributed to hypertension, obesity, and systolic dysfunction than in white patients, in which
it is more likely to be attributed to coronary artery disease.2,4 Reasons for this greater disease
burden in black and Hispanic patients are complex, resulting from the interaction of factors
such as hypertension, diabetes, obesity, reduced health care access, socioeconomics, and
cultural factors.8-10 Given the larger burden of HF and unfavorable disease outcomes, a tailored
and focused management approach in black and Hispanic populations is warranted.
The use of telehealth, defined as an electronic exchange of medical information, is a
promising approach to improve patient outcomes.11,12 In remote home telemonitoring,
outpatients living in the community communicate with clinicians to optimize HF treatment.
Since exacerbations of HF are common as the disease progresses, the use of telehealth to
monitor physiologic indicators, such as weight, heart rate, lung sounds, and blood pressure,
7
facilitates improved management through timely treatment adjustments. Without leaving
home, patients can be monitored by clinicians using peripheral devices (eg, blood pressure cuff,
pulse oximeter, weight scale) and video technology, allowing for a “live” visit, during which the
clinician can listen to heart and lung sounds using a stethoscope. Telehealth management
programs can also provide the patient with self-monitoring educational tools to facilitate
follow-up care in partnership with the health care provider.13
A recent systematic review found that remote monitoring of HF patients significantly
reduced the odds of HF-related death and hospitalization compared with usual care.14 A study
conducted by Clark et al found that telemonitoring reduced rates of admission by 21% and all-
cause mortality by 20%, while improving QoL.15 Other studies found significant reductions in
utilization across multiple disease categories.16-19 These interventions also demonstrated
improvements in health-related QoL, HF knowledge, and improvement of self-care behaviors.20
Nonetheless, other large randomized clinical trials did not find rehospitalization or mortality
benefits of home telemonitoring.21,22 Finally, although these studies included minorities and
underserved patients, they did not specifically analyze the impact of telehealth interventions on
these populations.
Given the great discrepancy in prevalence rates of HF between whites and ethnic
minorities, coupled with disparities in access to innovative and effective treatment,
telemonitoring may have greater effect on populations that are underserved. It is therefore
imperative that researchers and clinicians bring effective interventions to these at-risk
populations. Although many telehealth randomized controlled trials (RCTs) in HF have been
published, there is limited literature on the use of telehealth in underserved populations. The
few telehealth studies conducted in underserved populations have focused primarily on
diabetes and blood pressure management.23,24 The findings from these studies supported the
efficacy of telemonitoring; however, an RCT study that tested the effectiveness of telephone
case management in Hispanics of Mexican origin with HF found no significant differences in HF
8
readmission rate, HF days in the hospital, HF cost of care, all-cause hospitalizations or cost,
mortality, QoL, or depression.25
To this end, this study directly compared health care utilization and QoL of underserved
black and Hispanic patients receiving a telemonitoring intervention (Telehealth Self-monitoring,
or TSM) tailored for underserved HF patients to those receiving comprehensive outpatient
management (COM) at an HF clinic.
9
PARTICIPATION OF PATIENTS AND OTHER STAKEHOLDERS IN THE DESIGN AND CONDUCT OF RESEARCH AND DISSEMINATION OF FINDINGS
This study utilized a community-based participatory research (CBPR) approach to assess
usability and adapt TSM to facilitate acceptability and feasibility in a population of black and
Hispanic underserved HF patients. To accomplish this, a community advisory board (CAB)
advised the study team on all aspects of study design, implementation, evaluation, and
dissemination. The CAB comprised black and Hispanic HF patients (5); nonprofessional
caregivers (3); patient advocates (2); health care practitioners, including a geriatrician,
cardiologist, HF nurse practitioner, and registered nurse (4); a health policy expert (1); a health
disparities expert (1); a public health expert (1); a health access specialist (1); a social disparities
expert (1); a telehealth installation and patient orientation specialist (1); a health law expert (1);
a Hispanic community leader (1); and a pharmacist (1). The research team initially recruited the
core CAB membership through letters of invitation; CAB clinicians identified HF patients
representative of the study population and their caregivers, and the CAB itself recommended
the expansion of its stakeholder membership at formal meetings.
Initially, both the CAB and research team meetings occurred monthly; once the
formative phase of the study was complete, CAB meeting frequency decreased to quarterly to
maintain transparency, open dialogue, and engagement with stakeholders. The formative
phase adaptation relied on features of the ADAPT-ITT model,26 which involves stakeholders
across multiple phases: (1) providing input into a needs assessment, (2) decision making
regarding program choice, (3) administering the intervention with theater testing (a
demonstration of a virtual visit with an off-site nurse measuring vitals for members of the CAB
remotely), (4) producing a draft of the proposed intervention, (5) including topical experts in
10
adaptation, (6) integrating all of the input into the new intervention, (7) training staff on revised
intervention, and (8) conducting a full intervention pilot test (Figure 1).
Stakeholder impact was significant in this study, both in facilitating successes and
addressing challenges. In general, there were no major challenges regarding patient and
stakeholder engagement, with the exception of some turnover of patient stakeholders who
became very ill and were unable to continue to serve on the CAB.
While the CAB did not revise the research question as originally proposed, there were
numerous instances in which stakeholder input improved study rigor and quality. Two of these
instances are notable: (1) Stakeholders identified the importance of the provision of
reassurances regarding identity protection to undocumented patients in implementing an
intervention that utilizes a camera; and (2) stakeholders strongly recommended that patients
enrolled in the program be assured that TSM would be provided in addition to their clinic care
and that their involvement in telehealth monitoring would not replace their clinic care, which,
for many underserved patients, is their only connection to medical care.27
11
Figure 1. TSM sample screen in Spanish, home monitor, and peripherals
12
METHODS
Study Design
To generate patient-centered evidence that is generalizable to the proposed
community, we used a mixed methods approach to (1) utilize CBPR methodology to conduct a
process evaluation to assess telehealth acceptance and usability; and (2) employ an RCT study
design to directly compare TSM with COM in underserved black and Hispanic patients with HF.
Aim 1 Methods
This study utilized a CBPR approach to assess usability and adapt TSM to facilitate
acceptability and feasibility in the target population. To accomplish this, a CAB advised the
study team on study design, implementation, evaluation, and dissemination. The study team
used the ADAPT-ITT framework to engage key community stakeholders in the process of
adapting the intervention in the context of 2 consecutive focus groups, theater testing and pilot
testing, with subsequent focus group participants.27 The data presented herein were collected
during 3 focus groups. CAB members (described on p. 9) attended the first 2 focus groups.27
The CAB advised the study team on all aspects of study design, implementation, evaluation, and
dissemination. More specifically, the CAB was responsible for tailoring the program and
identifying factors impacting acceptance/feasibility among this population to reduce the impact
of such factors on usability.
The primary goal of the first CAB focus group was to obtain specific feedback regarding
intervention adaptation needs. The goals of the second CAB focus group were to theater test
with HF patient stakeholder CAB members and ensure the adaptation was successfully
implemented. Although feedback from all CAB members was incorporated, the study team
gave particular weight to patient stakeholder feedback. The third focus group involved TSM
patients who were randomized to the telemonitoring study arm and completed a 3-month pilot
to identify “on-the-ground” barriers.
13
All 3 focus group discussions were audio-recorded and professionally transcribed and
lasted approximately 2 hours each. Structural coding was used to mark responses to topical
questions in the interview guide.28 Following a review of the a priori topics, the facilitator
developed a codebook to categorize the data and identify salient themes and relationships.27
The main themes that emerged from the text identified specific recommendations for
intervention adaptations. Both CAB members and patient stakeholders were compensated $50
for participation, and each focus group was conducted in a private conference room.
Prestudy Usability (Stakeholder Perspective, Focus Groups 1 and 2)
The first CAB focus group (n = 14) was conducted during the formative phase of the
project. A general discussion of community needs was followed by a dialogue about specific
adaptation needs regarding both TSM equipment and study design. CAB members were
presented with the intervention (in English and Spanish) and a telemonitoring nurse, remotely
connected, demonstrated core components and key adjustable characteristics of the initial
intervention. A qualitative consultant led the focus group discussions, with content guided by
predetermined topics outlined in an interview guide, including instructions to prioritize patient
stakeholder contributions above medical and professional stakeholders.29
A second CAB focus group, conducted a month later (immediately after adaptation but
prior to intervention implementation) and led by the same qualitative researcher, involved a
discussion framed by an interview guide as well as theater testing of the TSM equipment with
the CAB patient stakeholders. A draft of the adapted intervention was discussed to ensure all
CAB feedback was effectively incorporated.
In-home Patient Usability Perspective (Focus Group 3)
We offered focus group participation to the initial 10 patient participants randomized to
the TSM arm in aim 2 of the study (4 agreed to participate) after they completed the program
14
so that we could obtain direct user feedback regarding implementation/usability barriers.
Discussion topics included ease of intervention use (eg, transmitting vital signs, televisits),
intervention usefulness, barriers to intervention implementation, and adjustment
recommendations. Patient participants in focus group 3 had been recently hospitalized for HF.
As with the stakeholder perspective, the qualitative research consultant guided the focus group
discussionaccording to predetermined topics outlined in an interview guide.
Aim 2 Methods
Forming the study cohort. A total of 104 black and Hispanic patients from underserved
communities with a primary diagnosis of HF, a New York Heart Association (NYHA) class of 1 to
3, and a Folstein Mini Mental Status Exam (MMSE) score of at least 21, recently discharged
from the hospital, were enrolled in the study and followed for 90 days. The population of
patients targeted for the study was very low income, with limited education and receiving
outpatient care at the HF clinic at a “safety net” hospital. Spanish- and/or English-speaking
patients 18 years and older were included in the study. Patients were excluded if they (1) had
an NYHA class of 4, (2) had an MMSE score below 21, (3) did not speak English or Spanish, (4)
were already receiving telehealth monitoring, or (5) did not self-identify as black or Hispanic.
Patients were enrolled during their hospital stay at a large safety net hospital in the New York
metropolitan area from March 31, 2014, to June 30, 2016. Patients were randomly assigned to
either TSM or HF COM.
Figure 2 displays the results of eligibility screening and allocation of patients to
comparator arms. Of the 364 patients assessed for eligibility, 247 did not meet inclusion criteria
(the most common reasons were not being black or Hispanic, residing in a nursing home (NH),
being physically unable to use equipment, and receiving early discharge/leaving hospital AMA
[against medical advice]), and 13 patients declined participation for the following reasons: did
not want to be involved with study (9); did not want machine in the home (1); were still
15
thinking about the study (1); already had a visiting nurse (1); and did not want to be disturbed
(1). The remainder (n = 104) were randomized (46 allocated to TSM and 58 to COM).
Study setting. Patients were recruited as inpatients with a primary diagnosis of HF from
Nassau University Medical Center (NUMC), the county safety net hospital in East Meadow, New
York.
16
Figure 2. CONSORT diagram (patient flow)
17
Interventions. Our comparators were TSM and COM. Patients were randomized and
followed for 90 days. The target health condition was HF; we chose patients for enrollment
based on a primary diagnosis of HF during hospitalization. All communication with patients in
both comparator groups was available in the patient’s preferred language (English or Spanish).
After consent, patients were randomized to TSM or COM and asked to provide contact
information for their physician, caregiver, and/or family member. Medical history was recorded
for patients in both groups. Those assigned to TSM received telemonitoring equipment shortly
after hospital discharge (within 7 days). Both cohorts received a phone call within 72 hours of
discharge from the research nurse coordinator (RNC) to check on their health status and
schedule a clinic appointment within 7 days of discharge, as per HF clinic guidelines.30
Both cohorts received weekly check-in phone calls for the first 30 days of the study, as
per HF clinic standard of care, and were provided with a patient education booklet during their
hospitalization. A cardiologist was on call and available to address urgent problems when
patients called the clinic after hours.
Comparator 1: TSM
TSM was defined as a weekly video visit, combined with daily telemonitoring of vital
signs utilizing an FDA-approved computerized monitoring device (American TeleCare®
LifeView), which connected the patient’s residence—via a wireless air card, broadband, or a
standard telephone line—to the provider station (American TeleCare® Provider Station) in
English or Spanish. TSM had 2 components: (1) a daily vital signs self-monitoring component,
wherein patients were instructed to measure and transmit standard key indicators of possible
condition exacerbation to a secure server for daily clinician review; and (2) a telehealth visit,
wherein patients were instructed to attend a regularly scheduled weekly video visit with the
RNC (conducted in real time in the preferred language of the patient), and queried regarding
their behavior and condition during that week. The TSM intervention was adapted for use with
18
the targeted underserved communities using the ADAPT-ITT framework to optimize
acceptability and feasibility.27
Vital signs monitoring. Patients were trained by an RNC, fluent in English and Spanish,
to utilize equipment peripherals to measure vital signs. During daily transmissions, patients sat
in front of the patient station and were prompted by the screens, in the patient’s language of
choice, to use the peripheral devices to transmit key indicators, which were then stored on the
server. Key indicators included blood pressure, oxygen saturation rate, weight, and pulse/heart
rate. Figure 1 presents an example of a Spanish language daily vital signs monitoring screen for
body weight. The RNC was required to review indicators within 24 hours post-transmission on
weekdays and within 72 hours after weekends. If values for key indicators were outside the
reference range, the patient and/or patient’s caregiver was (1) instructed to remeasure values;
(2) contacted by his or her health care practitioner to discuss/revise the treatment plan (ie,
diuretic or antihypertensive); (3) asked to perform a telemonitoring visit; or, for urgent matters,
(4) instructed by the RNC to call 911. Vital signs feedback reports for patients were generated in
their primary language to allow them to easily identify out-of-range values.
Weekly telemonitoring video visit. Once a week, TSM patients were instructed to
attend a weekly scheduled televisit with the RNC. During the televisit, the patient and the RNC
were able to view and interact with each other on a monitor and the RNC was able to listen to
the patient’s heart and lung sounds. During the televisit, the patient and RNC also discussed
vital sign values and symptoms of HF, as well as behaviors which may have contributed to
symptoms. If a patient did not transmit his or her vital signs the RNC would call him or her or, if
necessary, his or her contacts in order to ascertain if more training was needed or if there was
another reason why the vital signs had not been transmitted. On several occasions, the
bilingual installer visited patients a second and third time to review the use of the machine with
the RNC participating remotely.
19
Comparator 2: COM
The COM comparator comprised 1 clinic visit within a week of discharge and 4 weekly
check-in phone calls during the first month with the RNC. The weekly check-in calls were
conducted by the RNC and specifically evaluated patient status regarding medication
adherence, presence of symptoms, daily weights, and overall adherence with dietary and
activity recommendations, as per ACCF (American College of Cardiology Foundation)/AHA
(American Heart Association) recommended HF 2013 management guidelines.31 Patients who
did not have a weight scale at home were provided with one. Patients with symptoms
indicative of worsening HF or who had vital signs outside of a range determined at start of care
by their cardiologist at the clinic were asked to attend an HF clinic visit or were managed based
on the RNC’s clinician assessment.
After 30 days of weekly telephone contact, COM patients were followed by their
primary cardiologist at the HF clinic at the discretion of the clinician (or request of the patient).
COM patients were queried about ED and hospital utilization on a weekly basis for the
remainder of the study period to maintain a comparable frequency of contact.
Follow-up. Patients were followed for 90 days. If a patient could not be reached for
follow-up, the RNC would call his or her caregiver, family member, and doctor to attempt to re-
establish contact. The exposure period was not exceeded.
Study outcomes. Both cohorts were assessed for inpatient utilization (primary
outcome), emergency department utilization, QoL (using the Minnesota Living With Heart
Failure Questionnaire [MLHFQ]), and anxiety and depression (using the Patient Health
Questionnaire-4 [PHQ-4]) at days 1 and 90. All scales were available in English and Spanish, as
needed.32,33 We selected outcomes based on previous literature and results from our previous
telehealth study34; we presented the proposed outcomes to the CAB to ensure they were
relevant for and important to both patients and clinicians.
20
Inpatient and ED utilization. Hospital and ED utilization were determined through
patient self-report (subsequently confirmed through medical records review) and/or ongoing
review of enrolled subject medical records. However, hospital utilization could not be
confirmed (beyond self-report) if it occurred outside NUMC or the Northwell Health System.
The researcher who recorded study outcomes was also required to access vital signs data for
the TSM group; therefore, masking to treatment arm was not possible. In addition, given the
nature of the intervention, after randomization, it was infeasible to blind patients or staff to
treatment arms. Every ED visit and hospital admission was reviewed by 2 cardiologists and 1 NP
for classification as HF related (yes/no) or cardiovascular related (yes/no) and based on chart
review. It is important to note that all-cause hospital utilization included, in addition to HF and
cardiovascular disease (CVD) conditions, a broad array of other conditions that were likely not
indicative of HF exacerbation.
QoL. The MLHFQ ranges from 0 to 105, with higher scores indicating poorer QoL.
Physical MLHFQ subscales range from 0 to 40; emotional MLHFQ subscales range from 0 to 25.
Anxiety and depression screen. Both cohorts were also screened for depression and
anxiety using the PHQ-4.33 PHQ-4 scores range from 0 to 12, where higher scores are more
severe and combined scores of 3 or greater on the first and second items or third and fourth
items indicate a positive screen for anxiety or depression, respectively.
Institutional Review Board (IRB) approval. The Northwell Health Institutional Review
Board approved the study protocol (IRB No. 13-518A).
Randomization process. The Department of Biostatistics created a randomization
schedule to randomize patients to TSM or COM using permuted block randomization with
stratification by NYHA class to ensure equal representation of HF class across the 2 groups.
Concealed allocation was achieved by maintaining random allocation in the biostatistics office
only. Randomization was achieved by having a staff member phone the biostatistics unit for a
randomization assignment. No study staff or patients were privy to the randomization tables
21
and were, therefore, unable to know the next randomization assignment. Toward the end of
the study, we were unable to randomize any further subjects to the telehealth group since the
TSM vendor, American Telecare®, became insolvent. We therefore modified the randomization
procedure to continue the randomization of 12 additional patients to the control group in order
to maintain statistical power and preserve the integrity of randomization.
Data collection and sources. Prior to hospital discharge, the RNC reviewed charts of all
black or Hispanic patients with an admitting diagnosis of HF (confirmed by attending physician)
and approached each eligible patient to offer study participation. In discussing the study with
the patient, the RNC reviewed the overall goals of the study, the study design, the study follow-
up time frame (90 days), and the randomization procedures. The nurse also gathered basic
information regarding exclusion criteria (eg, NYHA class 4). If the patient met inclusion criteria,
the nurse obtained consent in the patient’s primary language. Once the patient had given
consent, the nurse collected the patient’s home address and phone number as well as his or
her caregiver’s phone number (if applicable) and explained that she would contact them weekly
for data collection. Reason for study withdrawal/loss to follow-up was determined by phone
calls and in-person follow-up at clinic visits by the RNC.
Analytical and statistical approaches. The primary analysis of this RCT was based on the
ITT principle that, for comparison purposes, included all subjects in the study arm to which they
had been randomized. In addition, any TSM patients who were hospitalized prior to receiving
the TSM equipment were counted as hospitalizations according to the ITT principle. Descriptive
statistics are presented as means and standard deviations or medians and as
frequencies/percentages wherever appropriate, including ranges of values, as needed.
Primary outcomes included ED visits, inpatient utilization, and length of stay (LOS). We
defined inpatient utilization in 4 ways: (1) whether a patient had had at least 1 inpatient
hospitalization more than 90 days (2) the number of hospitalizations experienced by a patient
during the 90-day period, (3) the 100-day mean adjusted for exposure time, and (4) the
22
cumulative LOS (inpatient days) experienced by a patient during the 90-day period (ie, for
multiple hospitalizations, the sum of all LOSs was computed as the cumulative LOS).
We defined ED utilization in 3 ways, including whether a patient had had at least 1 ED
visit during the 90-day period (binary), the number of ED visits experienced during 90 days, and
the 100-day mean adjusted for exposure time. We analyzed binary outcomes for ED visits and
hospitalizations using the standard chi-square or Fisher exact test. We computed associated
95% CIs for these proportions and their differences using exact methods.
We separately compared the number of ED visits as well as hospitalizations within each
group using Poisson regression (SAS PROC GENMOD statistical software). Due to excess zeros
(ie, zero inflation), methods for overdispersed Poisson were required for ED visits and
hospitalizations. We evaluated the 100-day mean using Poisson regression, adjusted for
number of days of exposure (follow-up). We analyzed cumulative LOS (inpatient days) using
negative binomial regression.35 We analyzed HF-related hospitalizations, ED visits, and LOS in
the same fashion described above, as we did for CVD-related outcomes. There were no
adjustments of P values for multiple testing.
We used repeated measures analysis of variance with a mixed models approach to
compare changes at enrollment and at 90 days between groups for the following psychosocial
outcomes: PHQ-4, and MLHFQ physical and emotional subscales. We dichotomized the
subscales of anxiety and depression, as measured by the PHQ-4, to presence or absence and
used GEEto analyze these binary data.
We calculated days out of hospital as a proportion of days spent out of hospital during
the 90-day period. We compared the 2 groups using the Mann-Whitney test.
In addition, we conducted post hoc analyses, comparing TSM with COM among patients
not receiving home care and hospitalized patients only. We also compared outcomes by TSM
adherence (TSM group only). We calculated daily adherence only for the TSM group. Patients in
23
the TSM group were asked to transmit all vital signs daily (90 transmittals/90 days). We
arbitrarily decided that the minimally acceptable level of adherence would be 1 transmittals per
week (12 transmittals per 90 days), which corresponds to the frequency of real-time virtual
nursing visits. In order to build some leniency into this criteria, we reduced the minimally
acceptable adherence level to fewer than 10 vital signs transmittals during the 90-day period
(low adherence), while we defined high adherence as at least 10 transmittals for 90 days.
Moreover, we performed post hoc analyses comparing TSM with COM by racial/ethnic groups
and by NYHA class (2 vs 3). It should be noted that, for all post hoc analyses (subgroup analyses
and other end points), we did not adjust P values for multiple testing.
We based the sample size and power calculation on previous literature (Informatics for
Diabetes Education and Telemedicine, or IDEATel), which demonstrated a lower telemonitoring
adherence level in minorities than in whites, and on preliminary data from our RCT TSM
study.34 Accordingly, we hypothesized that 75% of patients in the COM group would be
hospitalized at least once during the 90-day study period and that TSM would reduce the rate
to 50% (a relative reduction of 33%). The power to detect a statistically significant difference
between the groups was 97%. We based the calculation on a chi-square test (alpha = 0.05, 2-
tailed) with a sample size of 104 (n = 52 patients per group). We considered a result statistically
significant at the p < 0.05 level of significance. We performed all analyses using SAS 9.4 (Cary,
NC: SAS Institute Inc.; 2013).
We hypothesized the following: (1) The TSM group would experience fewer ED visits and
hospitalizations and, when hospitalized, would experience a lower LOS than the COM group;
and (2) TSM would have a greater effect on patients who were ineligible for home care (eg,
uninsured, undocumented). We based these hypotheses on Feltner’s systematic review36 and
meta-analysis of transitional care interventions establishing that home-visiting programs are
effective interventions in reducing all-cause and HF-specific readmission up to 6 months after
an index hospitalization for HF patients. This is important because access to home care is much
less prevalent for underserved populations. For example, postdischarge, homebound Medicare
24
patients typically receive live nursing visits every few days for a 60-day episode of care. It
therefore stands to reason that TSM patients not receiving home care would fare better than
COM patients not receiving home care, as the care standard of COM patients is much lower. In
addition to all-cause hospital and ED utilization, we also explored HF- and CVD-related
utilization as well as respective LOS data for each of these subgroups.
We accomplished prevention of missing utilization data by a combination of patient self-
report and continual review of medical records data to ascertain unreported or inaccurate
utilization. However, if a patient was hospitalized outside of the Northwell Health System
(including NUMC), investigators had only self-report as a data source. When a subject was
unwilling or unable to complete the QoL instrument at the end of the 90-day study period, the
RNC would repeatedly attempt to contact the subject either by phone, via his or her caregiver
or family member, or by meeting him or her at the clinic or hospital during a visit.
Baseline demographic and clinical data was complete for 100% of subjects, and 81% of
all subjects had complete outcome data for the 90-day study period. Some subjects dropped
out prior to 90 days for any of the following reasons: transferred to hospice or skilled nursing
facility, died, moved away, or requested termination of their study participation. The primary
analysis compared hospitalization rates directly, with no adjustment for covariates. In cases in
which subjects dropped out prior to 90 days of follow-up, we modified the analysis to adjust for
number of days of exposure. Thus, there was no need to impute missing hospitalization
outcomes. In secondary analyses, for which gender and HF class were adjusted, there were no
cases of missing data for these 2 important variables.
Conduct of the study. Changes from the protocol as originally proposed include the
following: (1) Because eligible patients were unable to be enrolled on weekends and holidays,
as the budget could only accommodate 1 RNC who worked weekdays, we obtained IRB
approval to call patients by phone at home during the subsequent week to offer study
25
participation; and (2) because some patients were scheduled to receive telehealth
management through their home care agency, we obtained IRB approval to deem these
patients ineligible for enrollment.
26
RESULTS
Specific Aim 1: To Assess Usability and Adapt TSM to Facilitate Acceptability and Feasibility in a Population of Underserved Black and Hispanic Patients
Two major themes emerged from qualitative analyses of the focus group data. The first
identified equipment changes to maximize usability. This theme arose from the CAB prestudy
usability focus groups (stakeholder perspective, focus groups 1 and 2) as well as the patient
usability (patient perspective, focus group 3). Specifically, suggestions that arose from the CAB
focus group included proper Spanish translation (some words were translated incorrectly and
our CAB provided proper terminology), font size (determined to be too small on most screens),
and elimination of medical jargon (some words were considered too technical for the target
population). Feedback from the patient perspective focus group, composed of the first 10
patients enrolled in the TSM arm, included providing more detailed instruction about the
proper way to turn on the scale (patients needed to pause before standing on the scale or it
wouldn’t function correctly).
Focus group participants mentioned they were pleased to have easier access to a health
care practitioner through the machine as well as being able to check their vitals to see if
anything was wrong: It saves me time. . . . We don’t have to come here to the hospital, we can
[stay at] the homes, talk to the doctor about what’s going on. They teach me [about] my [blood]
pressure, about my heart, everything.
One Spanish-speaking patient stakeholder first described in English the equipment as
follows: This machine is easy to use. . . . You can check your condition in your house; you can
maintain your health in your house. It is something that is beneficial.
He then described his sentiments more fully in Spanish, which were translated by
another CAB member to the rest of the group: The convenience of not having to go to the
hospital [and] to be checked in his own home at his own convenience . . . he really likes that. . . .
27
It’s not like [he has] to go to the hospital or call the doctor or try to make an appointment or
whatever. He’s a phone call away, a direct communication with a health practitioner.
In the pilot patient focus group, all of the participants indicated that there were
sporadic issues regarding connectivity; one participant indicated that it was possible that the
issue was less about connectivity and more about her own ability to navigate the weekly calls
with the nurse while using the equipment: And I don’t think it was the machine, I think it was
me. Because I think that I logged on, went to the area of log on where I was doing my vitals. If
you do that, you're going to shut her out, she can’t see you. . . . She had to call me back.
The second theme that arose during the prestudy CAB involved suggested changes to
the RCT study structure in order to maximize participant engagement. For instance, having the
patient meet the RNC during his or her hospital stay to create a “familiar face”: If I get a chance
to first meet [a person] and I’ve seen him and I know he’s real, you know, and we’ve met and
we’ve established our contact, then I’m kind of okay with Skyping you the next time around
because I know you’re real.
The CAB recommended that the study nurse strongly reiterate to the patient that
participation in the study does not replace his or her regular provider, in the primary care or HF
clinic setting, to ensure doctor and clinic access would not be affected in any way through
participation in the study: May I suggest that as you describe the [study procedures in the]
consent form . . . that that piece (the study does not replace her/his regular provider) be very
loud and clear, because I think it would be very reassuring to [the] patient . . . to know that it’s
not like . . . off they go with this [computer] screen and they’ll never see another nurse [again].
The CAB also recommended that, before enrollment and during the introduction of the
study, the study nurse reiterate to patients that their immigration status would not be recorded
and would in no way affect their participation or be communicated to any third party. Similarly,
reassuring the patients that video visits would not be recorded was also seen as crucial when
enrolling undocumented patients. This was communicated to the RNC.
28
Finally, the CAB suggested that adding a family member, if available, might help with the
televisits— specifically, holding the stethoscope so the RNC could hear heart and lung sounds,
adding a pharmacist to the study to include his or her perspective, and increasing the time
allotted for patient/caregiver equipment training. We successfully incorporated all of these
suggestions into the study design.
Only one issue arose regarding study structure in the pilot group. This participant felt
that the time period (90 days) for observation of the intervention group was not long enough:
Well, I think that three months isn’t enough to be honest. . . . I think it should be at least 6
months because the person is recovering . . . so if they could have something that is at least 6
months to help them [monitor] themselves.
Regarding the discussion of time period (90 days) for observation, the study team could
not extend the time because of the nature of the PCORI award; however, the CAB asked the
study team to recommend to future investigators to extend the study period of HF
telemonitoring.
Specific Aim 2: To Measure ED Utilization, Inpatient Utilization, and QoL of Patients Receiving TSM vs COM in a Population of HF Patients Discharged from Acute Care (Table 1)
Table 1 displays detailed baseline characteristics for both groups. In all, 104 patients
took part in the study: 32 were Hispanic, 72 were black, 43 were female, and 61 were male. In
terms of insurance status, the majority were either uninsured (23%) or receiving Medicaid
(33%); the remainder were insured through Medicare (21%), were dual eligible (6%), had
private insurance (9%), or were of unknown insurance status (8%). In terms of education, 4%
reported having had no school, 19% completed elementary school, 8% completed middle
school, 14% completed some high school, 32% graduated high school, 16% had some college
education, 3% graduated college, and 2% attended or graduated from graduate school. In terms
of clinical characteristics, 70% of patients were classified as NYHA class 3 and 30% were NYHA
29
class 2; 61% had reduced ejection fraction, 10% borderline, and 29% preserved. None of the
patients screened declined participation due to technology, although 2 did not permit
installation after randomization to TSM.
30
ITT Analyses (Tables 2-4)
ED visits. Table 2 presents ED utilization by group; 32.6% of TSM patients were seen at
least once in the ED during the 90-day period, compared with 44.8% of COM patients. This
difference was not statistically significant (RR = 1.37; 95% CI, 0.83-2.27; p = 0.21). The number
of ED visits was also similar for the 2 groups, with a TSM ED utilization mean of 0.63 (SD = 1.18)
vs a mean of 0.69 (SD = 0.99) for COM patients during the 90 days. Poisson regression revealed
that the average number of ED visits per 100 patient days of follow-up was 0.86 for patients in
the TSM group and 0.79 for patients in the COM group. Thirty-day ED utilization was 4 of 46
(8.7%) for TSM and 14 of 58 (24.1%) for COM (p = 0.07). We observed no significant differences
between groups for the number of patients with at least 1 HF- and CVD-related ED utilization
visit during the 90 days.
Inpatient utilization. Table 3 presents inpatient utilization by group; 41.3% of patients in
the TSM group were hospitalized at least once during the 90-day study period, compared with
37.9% of COM patients. This difference was not statistically significant (RR = 0.92; 95% CI, 0.57-
1.48; p = 0.73). In terms of number of hospitalizations, TSM patients had a significantly higher
utilization mean of 0.78 (SD = 1.3) vs 0.55 (SD = 0.9) for COM patients during the 90 days (p =
0.03). Thirty-day hospitalization was 7 of 46 (15.2%) for TSM and 9 of 58 (15.5%) for COM (p =
0.97). We observed no significant differences between groups for the number of patients with
at least 1 HF- and CVD-related hospitalization during the 90 days.
LOS. Table 3 presents all-cause LOS data by group; the mean total LOS for TSM patients
was 5.2 days (SD = 8.1) vs 3.6 days (SD = 6.9) for COM patients (p = 0.12). The average LOS was
9.7 per 100 patient days followed for TSM patients vs 4.4 in COM patients. Differences between
groups were not statistically significant for HF- or CVD-related hospital admission LOS.
31
Table 1. Baseline Characteristicsa
Total COM Group TSM Group
Participants (N) 104 58 46
Age: mean (SD, range) 59.9 (15.1, 19-93)
61.1 (15.0, 26-90)
58.4 (15.2, 19-93)
Gender: female n (%) 43 (41%) 23 (40%) 20 (43%)
Hispanic: n (%) 32 (31%) 17 (29%) 15 (33%)
Black: n (%) 72 (69%) 41 (71%) 31 (67%)
Female: n (%) 43 (41%) 23 (40%) 20 (44%)
Reduced ejection fraction (≤ 40%): n (%)
62 (61%) 36 (63%) 26 (58%)
Borderline ejection fraction (41%-49%): n (%)
10 (10%) 6 (11%) 4 (9%)
Preserved ejection fraction ( ≥ 50%): n (%)
30 (29%) 15 (26%) 15 (33%)
Index hospitalization LOS (days): n (SD)
5.97 (4.6) 5.88 (4.3) 6.09 (5.3)
NYHA class 3: n (%) 73 (70%) 40 (69%) 33 (72%)
NYHA class 2: n (%) 31 (30%) 18 (31%) 13 (28%)
Uninsured: n (%) 23 (23%) 11 (20%) 12 (27%)
Medicaid: n (%) 33 (33%) 21 (38%) 12 (27%)
Medicare: n (%) 21 (21%) 8 (15%) 13 (29%)
Dual eligible: n (%) 6 (6%) 5 (10%) 1 (2%)
Private insurance: n (%) 9 (9%) 4 (7%) 5 (11%)
Other insurance: n (%) 8 (8%) 6 (11%) 2 (4%)
Income: < $10 000/year: n (%) 74 (72%) 39 (68%) 35 (76%)
Education: graduated from high school: n (%) 53 (54%) 28 (53%) 25 (55%)
aNo significant between group differences were found.
32
Table 2. ED Utilization
Intention-to-Treat Analysis/90 Days
Post Hoc Analysis 1: No Home Care/ 90 Days
Post Hoc Analysis 2: Patients Hospitalized at Least Once/90 Days
Group TSM COM TSM COM TSM COM
Participants (n) 46 58 26 34 19 22
All-cause ED visits—binary, n (%) with ≥ 1 ED visit
15 (32.6%)
26 (44.8%)
7 (26.9%)
13 (38.2%)
14 (73.7%)
19 (86.4%)
Relative risk (RR), 95% CI
RR = 1.37 CI: [0.83-2.27]
RR = 1.42 CI: [0.66-3.05]
RR = 1.17 CI: [0.85-1.61]
P value p = 0.21 p = 0.36 p = 0.44
All-cause ED visits: mean, median (SD)
0.63, 0 (SD = 1.18)
0.69, 0 (SD = 0.99)
0.42, 0 (SD = 0.81)
0.59, 0 (SD = 1.05)
1.4, 1 (SD = 1.5)
1.3, 1 (SD = 0.84)
P value p = 0.73 p = 0.75 p = 0.44 All-cause ED visits per 100 days of follow-up (adjusted for exposure time)
0.86 0.79 0.59 0.68 1.9a 1.6a
HF-related ED visits—binary (% with ≥ 1 ED visit)
10.8% 12.1% 3.9% 11.7% 26.3% 31.8%
RR, 95% CI
RR = 1.11 CI: [0.38-3.27]
RR = 3.06 CI: [0.36-25.76]
RR = 1.21 CI: [0.46-3.19]
P value p = 0.85 p = 0.38 p = 0.70
HF-related ED visits: mean, median, (SD)
0.13, 0 (SD = 0.4)
0.14, 0 (SD = 0.4)
0.04, 0 (SD = 0.20)
0.12, 0 (SD = 0.33)
0.32, 0 (SD = 0.58)
0.36, 0 (SD = 0.58)
P value p = 0.83 p = 0.42 p = 0.98
33
Intention-to-Treat Analysis/90 Days
Post Hoc Analysis 1: No Home Care/ 90 Days
Post Hoc Analysis 2: Patients Hospitalized at Least Once/90 Days
Group TSM COM TSM COM TSM COM
HF-related ED visits per 100 days of follow-up (adjusted for exposure time)
0.18 0.16 0.05 0.13 0.43a 0.43a
Intention-to-Treat Analysis/90 Days
Post Hoc Analysis 1: No Home Care/ 90 Days
Post Hoc Analysis 2: Patients Hospitalized at Least Once/90 Days
Group TSM COM TSM COM TSM COM
CVD-related ED visits—binary (% with ≥ ED visit)
13.0% 15.5% 7.7% 17.7% 31.6% 36.4%
RR, 95% CI interval
RR = 1.19 CI: [0.46-3.1]
RR = 2.29 CI: [0.50-10.45]
RR = 1.15 CI: [0.49-2.73]
P value p = 0.72 p = 0.45 p = 0.75
CVD-related ED visits: mean, median, (SD)
0.17, 0 (SD = 0.49)
0.17, 0 (SD = 0.42)
0.12, 0 (SD = 0.43)
0.18, 0 (SD = 0.39)
0.42, 0 (SD = 0.69)
0.41, 0 (SD = 0.59)
P value p = 0.70 p = 0.76 p = 0.74
CVD-related ED visits per 100 days of follow-up (adjusted for exposure time)
0.24 0.20 0.16 0.20 0.57a 0.48a
aNumber of ED visits beyond the first hospitalization.
34
Table 3. Hospital Utilization
Intention-to-Treat Analysis/90 Days
Post Hoc Analysis 1: No Home Care/90 Days
Post Hoc Analysis 2: Patients Hospitalized at Least Once/90 Days
Group TSM COM TSM COM TSM COM Participants (n) 46 58 26 34 19 22 All-cause hospitalizations —binary, n (%) with ≥ 1 hospitalization
19 (41.3%)
22 (37.9%)
10 (38.5%)
10 (29.4%)
19 (100%)
22 (100%)
Relative risk (RR), 95% CI
RR = 0.92 CI: [0.57-1.48]
RR = 0.76 CI: [0.38-1.56] N/A
P value p = 0.73 p = 0.46 N/A All-cause hospitalizations: mean, median, (SD)
0.78, 0 (SD = 1.3)
0.55, 0 (SD = 0.9)
0.58, 0 (SD = 1.0)
0.38, 0 (SD = 0.7)
1.9, 1 (SD = 1.4)
1.5, 1 (SD = 0.8)
P value p = 0.03 p = 0.10 p = 0.10 Number of all-cause hospitalizations per 100 days of follow-up (adjusted for exposure time)
1.07 0.63 0.82 0.44 2.6 1.7
All-cause hospital LOS
5.2, 0 (SD = 8.1)
3.6, 0 (SD = 6.9)
5.2, 0 (SD = 8.6)
3.5, 0 (SD = 7.7)
12.5, 11 (SD = 8.1)
9.4, 5 (SD = 8.4)
P value p = 0.12 p = 0.16 p = 0.02 All-cause hospital LOS per 100 days of follow-up (adjusted for exposure time)
9.7 4.4 12.4 4.0 22.0 11.5
HF-related hospitalizations —binary (% with ≥ 1 hospitalization)
10.9% 13.8% 3.9% 14.7% 26.3% 36.4%
RR, 95% CI
RR = 1.27 CI: [0.44-3.6]
RR = 3.82 CI: [0.48-30.77]
RR = 1.38 CI: [0.54-3.52]
35
Intention-to-Treat Analysis/90 Days
Post Hoc Analysis 1: No Home Care/90 Days
Post Hoc Analysis 2: Patients Hospitalized at Least Once/90 Days
Group TSM COM TSM COM TSM COM
P value p = 0.65 p = 0.22 p = 0.49
HF-related hospitalizations: mean, median (SD)
0.15, 0 (SD = 0.47)
0.16, 0 (SD = 0.41)
0.04, 0 (SD = 0.20)
0.15, 0 (SD = 0.36)
0.37, 0 (SD = 0.68)
0.41, 0 (SD = 0.59)
P value p = 0.76 p = 0.30 p = 0.96
HF-related hospitalizations per 100 days of follow-up (adjusted for exposure time)
0.20 0.18 0.05 0.17 0.50 0.5
Intention-to-Treat Analysis/90 Days
Post Hoc Analysis 1: No Home Care/90 Days
Post Hoc Analysis 2: Patients Hospitalized at Least Once/90 Days
Group TSM COM TSM COM TSM COM
HF-related hospital LOS: mean, median (SD)
0.54, 0 (SD = 1.7)
0.91, 0 (SD = 3.0)
0.23, 0 (SD = 1.2)
0.82, 0 (SD = 2.6)
1.3, 0 (SD = 2.5)
2.4, 0 (SD = 4.5)
P value p = 0.60 p = 0.30 p = 0.47 HF-related hospital LOS per 100 days of follow-up (adjusted for exposure time)
0.72 1.2 0.3 1.2 1.7a 3.0a
CVD-related hospitalizations —binary (% with ≥ ED visit)
13.0% 17.2% 7.7% 17.7% 31.6% 45.5%
RR, 95% CI
RR = 1.32 CI: [0.52-3.4]
RR = 2.29 CI: [0.50-10.45]
RR = 1.44 CI: [0.64-3.22]
P value p = 0.56 p = 0.45 p = 0.36
36
Intention-to-Treat Analysis/90 Days
Post Hoc Analysis 1: No Home Care/90 Days
Post Hoc Analysis 2: Patients Hospitalized at Least Once/90 Days
Group TSM COM TSM COM TSM COM CVD-related hospitalizations; mean, median (SD)
0.20, 0 (SD = 0.54)
0.19, 0 (SD = 0.44)
0.12, 0 (SD = 0.43)
0.18, 0 (SD = 0.39)
0.5, 0 (SD = 0.77)
0.5, 0 (SD = 0.60)
P value p = 0.65 p = 0.76 p = 0.86 Number of CVD-related hospitalizations per 100 days of follow-up (adjusted for exposure time)
0.27 0.22 0.16 0.20 0.60a 0.60a
CVD-related hospital LOS: mean, median (SD)
1.0, 0 (SD = 3.7)
1.6, 0 (SD = 5.2)
1.1, 0 (SD = 4.6)
2.3, 0 (SD =6.7)
2.4, 0 (SD = 5.6)
4.1, 0 (SD = 8.0)
P value p = 0.67 p = 0.60 p = 0.54 Number of CVD-related hospital LOS per 100 days of follow-up (adjusted for exposure time)
1.3 1.9 1.4 2.8 3.1a 4.9a
Days out of hospital: mean, median, (SD)
0.90, 1 (SD = 0.08)
0.96, 1 (SD = 0.20)
0.87, 1 (SD = 0.25)
0.96, 1 (SD = 0.09)
0.76, 0.83, (SD = 0.25)
0.88, 0.94, (SD = 0.10)
P value p = 0.39 p = 0.28 p = 0.06 aNumber of hospitalization days beyond the first hospitalization.
37
Table 4. Quality of Life, Anxiety and Depression
38
Days out of hospital. As can be seen in Table 3, there was no significant difference in
the proportion of days out of hospital between the TSM and COM groups: 0.90 (SD = 0.08) vs
0.96 (SD = 0.20), respectively (p = 0.39).
Quality of Life (QoL). Overall, QoL did not differ between groups over time (Table 4).
MLHFQ QoL was 62.7 for TSM at enrollment vs 59.9 for COM and 36.3 for TSM vs 27.8 for COM
at 90 days (p = 0.50). The physical and emotional subscale results were also similar for both
groups (p = 0.30 and p = 0.82).
Screening scale for anxiety and depression (PHQ-4). Scores for overall psychological
distress did not significantly differ over time (5.0 for TSM vs 5.0 for COM at enrollment and 2.8
for TSM vs 2.0 for COM at 90 days; p = 0.43). Differences over time for the PHQ-4 anxiety
subscale between enrollment and 90 days were statistically significant, indicating that while
anxiety symptoms improved for both groups, the improvement was greater for COM patients
than for TSM patients (50% of TSM vs 57% of COM screened positive at enrollment, and 28%
TSM vs 13% COM screened positive at 90 days; p = 0.05). We found no significant differences
over time between the proportions of patients who screened on the PHQ-4 depression subscale
between enrollment and 90 days.
Post Hoc Analysis 1: Patients Not Receiving Home Care (Tables 2-4)
It is important to look at the receipt of home care for underserved patients (eg,
uninsured), as they often do not receive home care as part of usual care. In our study, most
patients (60 of 104) did not receive home care services (56.5% of TSM and 58.6% of COM
patients).
ED visits. Table 2 presents ED utilization by group; 26.9% of patients not receiving home
care randomized to the TSM group were hospitalized at least once during the 90-day study
period compared with 38.2% of COM patients. We observed no statistically significant
differences between groups for ED visits (RR = 1.42; 95% CI, 0.66-3.05; p = 0.36).
39
Inpatient utilization. As can be seen in Table 3, 38.5% of patients not receiving home
care randomized to the TSM group were hospitalized for any reason at least once, compared
with 29.4% of COM patients. This difference was not statistically significant (RR = 0.76; 95% CI,
0.38-1.56; p = 0.46). The average number of all-cause hospitalizations during the 100 days of
follow-up for patients not receiving home care was not statistically significantly different
between groups 0.58 (SD = 1.0) and 0.38 (SD = 0.7) for TSM and COM patients, respectively (p =
0.10). Differences between groups for HF- and CVD-related hospitalization were also not
statistically significant.
LOS. The mean all-cause LOS for TSM patients not receiving home care was 5.2 days (SD
= 8.6) vs 3.5 days for COM patients (SD = 7.7), (p = 0.16). The average LOS was 12.4 per 100
patient days followed for TSM patients vs 4.0 in COM patients. Differences between groups for
LOS for HF- and CVD-related admissions were not significant.
Post Hoc Analysis 2: Patients Hospitalized at Least Once (Tables 2-4)
It is important to compare utilization between groups specifically for patients who were
hospitalized, since most patients (64 of 104 or 61.5%) were not hospitalized at any time (58.7%
of TSM and 62.1% of COM patients) during the 90-day follow-up period.
ED visits. Table 2 presents ED utilization by group; 73.7% of hospitalized TSM patients
utilized the ED at least once during the 90-day study period compared with 86.4% of COM
patients. This difference was not statistically significant (RR = 1.17; 95% CI, 0.85-1.61; p = 0.44).
Similarly, differences between groups for HF- and CVD-related ED utilization were not
statistically significant.
Inpatient utilization. As shown in Table 3, the average number of all-cause
hospitalizations during 90 days of follow-up for patients with at least 1 hospitalization was not
statistically significant between groups (1.9 [SD = 1.4] and 1.5 [SD = 0.8] for TSM and COM
40
patients, respectively (p = 0.10). Differences between groups for HF- and CVD-related
hospitalization were also not statistically significant.
LOS. The mean all-cause LOS for TSM patients was 12.5 days (SD = 8.1) vs 9.4 days (SD =
8.4) for COM patients. Patients with at least 1 hospitalization in the TSM group had a
significantly longer LOS than patients in the COM group during the 90 days (12.5 [SD = 8.1] vs
9.4 [SD = 8.4); p = 0.02). We found no significant differences between groups for HF- and CVD-
related hospital LOS.
Post Hoc Analysis 3: ED and Hospitalization by Heart Failure Class (Table 5)
ED visits. As shown in Table 5, 7.7% of NYHA class 2 patients in the TSM group had at
least 1 ED visit compared with 44.4% of COM patients, while 42.4% of TSM Class 3 patients had
at least 1 ED visit vs 45.0% for the COM group (p = 0.08). We found no significant differences
between groups by HF class for HF- or CVD-related ED visits.
Inpatient utilization. As can be seen in Table 5, 15.4% of TSM class 2 patients were
admitted at least once compared to 27.8% of COM patients, while 51.5% of TSM class 3
patients had at least 1 hospitalization vs 42.5% for the COM group. Differences between groups
were not statistically significant (p = 0.29). Similarly, we observed no significant differences
between groups for HF- or CVD-related hospitalizations.
41
Table 5. ED and Hospitalization by Heart Class
LOS. As can be seen in Table 5, we observed no significant differences between groups
for NYHA class 2 or 3 patients for all-cause, HF-, and CVD-related LOS. The average LOS for
42
NYHA class 2 patients followed during the 90 days was 1.0 (SD = 2.8) for TSM patients vs 1.1 (SD
= 2.3) for COM patients, while the average LOS for NYHA class 3 patients was 6.8 (SD = 8.9) for
TSM patients vs 4.7 (SD = 7.9) for COM patients (p = 0.31).
Post Hoc Analysis 4: Utilization by Subpopulations (Black/Hispanic; Table 6)
Given the low number of events (and relatively small subgroup sample size) when
analyzing outcomes by race/ethnicity, we present only all-cause utilization outcomes. For the
black cohort, 32.3% of TSM patients had at least 1 ED visit during the 90 days vs 48.8% of COM
patients (RR = 1.50; 95% CI, 0.83-2.70; p = 0.16). Similarly, 38.7% of black TSM patients were
hospitalized at least once vs 43.9% of COM patients (RR = 1.13; 95% CI, 0.65-1.96; p = 0.66). For
the Hispanic cohort, 33.3% of TSM patients had at least 1 ED visit during the 90 days vs 35.3% of
COM patients (RR = 1.04; 95% CI, 0.48-2.30; p = 0.91), whereas 46.7% of Hispanic TSM patients
were hospitalized at least once vs 23.5% of Hispanic COM patients (RR = 0.60; 95% CI, 0.30-
1.21; p = 0.17).
43
Table 6. Hospital Utilization by Race/Ethnicity
Black Subjects/90 Days Hispanic Subjects/90 Days Group TSM COM TSM COM Participants (n) 31 41 15 17 All-cause ED visits—binary (% with ≥ 1 ED visit): n (%)
10 (32.3%)
20 (48.8%)
5 (33.3%)
6 (35.3%)
Relative risk (RR), 95% CI
RR = 1.50 CI: [0.83-2.70]
RR = 1.04 CI: [0.48-2.30]
P value p = 0.16 p = 0.91 All-cause hospitalizations—binary (% with ≥ 1 hospitalization): n (%)
12 (38.7%)
18 (43.9%)
7 (46.7%)
4 (23.5%)
Relative risk (RR), 95% CI
RR = 1.13 CI: [0.65-1.96]
RR = 0.60 CI: [0.30-1.21]
P value p = 0.66 p = 0.17
Post Hoc Analysis 5: Adherence and Utilization of Study Intervention (TSM Group Only; Table 7)
Table 7 presents utilization data by adherence level for patients in the TSM group only.
For this study, we define high TSM adherence as at least 10 vital signs transmittals during the
90-day study period. Although patients were asked to transmit daily vital signs and attend a
weekly televisit with the RNC, 23 (50%) of TSM patients had provided fewer than 10
transmittals during the 90-day period. Patients in the high adherence group versus the low
adherence group had similar rates of having 1 or more ED visits (30.4% vs 34.8%; p = 0.75) and
1 or more hospitalizations (30.4% vs 52.2%; p = 0.13). Patients with higher adherence had a
significantly lower mean all-cause LOS (2.6; SD = 5.0) vs low adherence patients (7.7; SD = 9.7);
p = 0.01. All-cause LOS per 100 days of follow-up was also lower for the high adherence group
(2.9 days vs 16.7 days). We observed no significant differences between adherence groups for
HF-related LOS, but patients with higher adherence had a significantly lower mean CVD-related
LOS (0.13 vs 1.87; p = 0.02).
44
Table 7. Overall TSM Adherence and Utilization
Low Adherence (n = 23) (< 10 transmittals)
High Adherence (n = 23) (10 or more transmittals )
All-cause ED visits—binary (% with ≥ 1 ED visit) 34.8% 30.4%
Relative risk (RR); 95% CI
RR = 1.14 CI: [0.5-2.6]
P value p = 0.75 HF-related ED visits—binary (% with ≥ 1 ED visit) 13.0% 8.7%
RR; 95% CI
RR = 1.5 CI: [0.3-8.2]
P value p = 1.0 CVD-related ED visits—binary (% with ≥ 1 ED visit) 17.4% 8.7%
RR; 95% CI
RR = 2.0 CI: [0.4-9.9]
P value p = 0.67 All-cause hospitalizations—binary (% with ≥ 1 hospitalization)
52.2% 30.4%
RR; 95% CI
RR = 1.7 CI: [0.8-3.6]
P value p = 0.13 HF-related hospitalizations—binary (% with ≥ 1 hospitalization)
13.0% 8.7%
Relative risk (RR); 95% CI
RR = 1.5 CI: [0.3-8.2]
p value p = 1.0 CVD-related hospitalizations—binary (% with ≥ 1 hospitalization)
17.4% 8.7%
45
Table 7. Overall TSM Adherence and Utilization
Low Adherence (n = 23) (< 10 transmittals)
High Adherence (n = 23) (10 or more transmittals )
RR; 95% CI
RR = 2.0 CI: [0.4-9.9]
P value p = 0.67 All-cause LOS (days); (SD); Median 7.74 (9.74); 3 2.61 (4.97); 0
P value p = 0.01a All-cause LOS per 100 days of follow-up (adjusted for exposure time)
16.7 2.9
HF-related LOS (days); (SD); median 0.70 (2.05); 0 0.39 (1.37); 0
P value p = 0.53a HF-related LOS per 100 days of follow-up (adjusted for exposure time)
1.0 0.4
CVD-related LOS (days); (SD); median 1.87 (5.15); 0 0.13 (0.46); 0
P value p = 0.02a CVD-related LOS per 100 days of follow-up (adjusted for exposure time)
2.5 0.15
aNumber of hospitalization days beyond the first hospitalization.
46
DISCUSSION
Decisional Context
To our knowledge, this RCT was the first to use a telehealth self-management
intervention to monitor daily vital signs and symptoms exclusively in an underserved population
of Hispanics and blacks recently discharged from the hospital for HF-related symptoms. Patients
with HF and other stakeholders can use the study results as detailed in this report in deciding
whether to participate in a telemonitoring program.
Study Results in Context
For the 3 main outcomes—all-cause ED utilization, hospital utilization, and LOS—TSM
did not result in a significant improvement over COM. In fact, the mean number of all-cause
hospitalizations was significantly lower for the COM group during the 90 days. Additionally,
COM patients reported a significantly greater reduction of anxiety symptoms at 90 days than
the TSM group.
As for QoL, we observed that patients in both arms reported improved QoL scores
during the 90-day observation period on the basis of the MLHFQ. Similar to the study by
Konstam et al,37 which evaluated an automated home monitoring intervention to a more
comprehensive disease management intervention, our study did not reveal any effect of TSM
on HF-related QoL.
COM patients reported a significantly greater reduction of anxiety symptoms at 90 days
than the TSM group. This is consistent with the findings of Huygens et al, which demonstrated
increased anxiety for cardiovascular patients monitoring vital signs.38 Regarding the present
study, we cannot determine from our data the reasons for this greater reduction of anxiety in
the COM group; it may be due to patient anxiety from taking the measurements themselves,
discomfort with out-of-range data, or unease due to frequent reminders of their chronic
condition. These findings are exploratory and require validation by other multicentered studies.
47
Despite the community-based participatory research process utilized to adapt the TSM
intervention for our target population,39 only half of TSM patients actively participated in the
intervention (transmitted their vital signs 10 times or more during the 90 days), despite our
original target of daily transmissions, which would have been 90 per patient (1/day). These
findings are similar to the Tele-HF study22 and to the BeAT-HF study (Better Effectiveness After
Transition-Heart Failure),21 in which approximately half of the patients were adherent to
telemonitoring and telephone calls. In contrast, in the TIM-HF study (Telemedical Interventional
Monitoring in Heart Failure), utilization did not differ between groups (telemonitoring vs usual
care), despite high adherence to telemonitoring intervention (> 70% of daily transmissions).
Still, it is conceivable that low adherence rates to the TSM intervention may have affected our
results. Future studies are needed to address methods to improve adherence, such as
incremental incentives.
The observed rate of hospitalization was half what we had projected, based on
preliminary (unpublished) data, for which our average age at enrollment was 82. This lower-
than-expected observed hospitalization rate may be a function of the much younger age group
enrolled in the current study (mean age = 60). It is also notable that our 30-day readmission
rate was 15.5% compared with the 25% national rate for Medicare patients (mean age =
81.8),40 also likely a function of the younger patient cohort enrolled in this study.
Implementation of Study Results
We conducted this study at a safety net hospital that operates an HF clinic, with patients
reporting very low incomes (72% reported incomes of less than $10 000 per year). Although
TSM has shown benefit in HF management of other populations,20 our study did not find
significant benefit for TSM patients in terms of all-cause ED utilization, hospital utilization, LOS,
and QoL. Similar to Chaudhry’s22 telephone-based interactive voice response system in patients
recently hospitalized for HF (Tele-HF study), and Ong’s combined health coaching telephone
48
calls and telemonitoring,21 our TSM and COM groups did not differ significantly for our 3 main
outcome variables.
Generalizability
Because this was a single-center study at a safety net hospital with an HF clinic, our
results may not be generalizable to other underserved populations in different regions and
settings. It should be noted that our COM patients received a relatively high level of care. Most
usual care, especially for underserved populations, is provided in outpatient settings without an
HF dedicated clinic. This HF clinic closely monitors patients by phone weekly for 30 days after
discharge. This weekly contact was likely to be more robust than usual care in other safety net
hospitals.
Similar to the BEAT-HF study, our study recruited patients from hospital inpatient units,
which may have been composed of less adherent patients than if we had recruited in the
outpatient setting. Another important factor to consider is patient age: at an average age of 60
(BEAT-HF average age at enrollment was 73), many patients in our study were still employed
and may have therefore been less likely to attend weekly video visits during the 9-to-5 workday
of the RNC.21 Ease of use is also a factor to consider: about one-third of TSM patients reported
the equipment was not easy to use, despite the adaptation process described herein. Finally,
connectivity is a concern in certain geographic areas: 25% of TSM patients reported the
equipment did not always provide a reliable connection to the RNC.
Subpopulation Considerations (Post Hoc Analyses)
Interestingly, TSM seemed to have a stronger effect on NYHA Class 2 patients than on
NYHA Class 3 patients. Although these findings—based on small subgroup sample sizes—were
not significant, the effect of heart class on TSM interventions requires further investigation.
49
In terms of racial/ethnic subgroups, although TSM did not show significant benefit in
either group, Hispanics appeared to benefit less than blacks in their rates of ED visits and
inpatient utilization.
When we analyzed the subgroup of patients that did not receive home care services
postdischarge, we did not observe a decrease in the number of ED visits and all-cause
hospitalizations for TSM as expected.
Some of our post hoc analyses suggest that in underserved populations not receiving
home care after hospitalization there may be an effect of telemonitoring on HF-related
utilization. Although this trend was based on small subgroups and was not significant, the effect
of telemonitoring on HF-related utilization in underserved populations not receiving home care
requires further investigation. Although no previous studies have examined the effect of TSM
on HF-related utilization for underserved populations, the TIM-HF study41 (which did not
present patient socioeconomic demographics) found no effect of TSM on the rate of HF-related
hospitalization.
Study Limitations
This study was single-centered with a relatively small sample size operating at a safety
net hospital located in the New York Metropolitan area. At the start of our study, most
telehealth equipment utilized a relatively large unit installed in the patient’s home. Now, many
telehealth companies are offering products that are smaller and wireless, effectively eliminating
the need for installation, and decreasing the 7-day delay between discharge and monitoring.
For this population especially, this new smaller, smarter technology might offer a more viable
option—especially since many of our patients were ineligible or experienced low adherence
due to unstable living arrangements (eg, living with relatives, living in nursing homes or
rehabilitation facilities, being homeless).
50
Future Research
Effective translation of TSM into real-world practice with consistent positive outcomes
resulting from successful interventions will require large-scale pragmatic studies. Multicenter
studies should be employed to enroll many more subjects, which is challenging given
equipment costs in relation to budget limitations. However, equipment costs will undoubtedly
decrease as technology evolves.
Future research is also needed to identify methods to improve adherence. Similar to
BEAT-HF,21 we observed that high adherence TSM patients tended to have lower hospital
utilization compared with low adherence TSM patients. Nonetheless, our adherence numbers
were extremely low and interpretation would have been difficult if we had not chosen a very
low minimally acceptable level of adherence (< 10 transmittals/90 days = low adherence). It is
important to note that our TSM adherence criterion was not based on other published studies
(such as the BEAT-HF study21) in which the authors measured adherence in both groups; our
study did not measure and compare adherence for the COM group.
It should also be noted that most HF management is delivered in the cardiology
outpatient practice. Future studies might want to consider targeting outpatients for
enrollment; however, event rates would likely be lower in outpatients, and larger, longer
studies would be needed to achieve sufficient power. Because telehealth is usually initiated at
hospital discharge, this study sought to mimic real-world processes as much as possible.
51
CONCLUSION This is the first study to describe the formative process of a community-based
participatory research study aimed at optimizing telehealth utilization among black and
Hispanic patients from underserved communities. The adaptation—based on the analysis of
data from 3 focus groups, theater testing, and a small pilot study—resulted in an intervention
that was considered to be acceptable and feasible for black and Hispanic HF patients residing in
underserved communities. However, despite the efforts of the focus groups, the CAB, and the
study team to tailor the intervention for this population, participation was surprisingly low.
Our study showed that TSM did not reduce all-cause ED or hospital utilization for
patients with HF from underserved communities during the 90-day study period. Furthermore,
we found all-cause hospitalization to be significantly higher for the TSM patient cohort. TSM
also did not show significant benefit on QoL for underserved patients with HF. Interestingly,
while anxiety symptoms improved for both groups over time, this improvement was greater for
COM patients compared with TSM patients. Although reasons for these group differences were
not captured, they may range from discomfort related to more intense monitoring (daily
reminders of their health condition) to increased anxiety over having the technology removed
at the end of the monitoring period. Furthermore, lack of night and weekend support for the
TSM intervention, given the relatively young age (mean = 60 years) of the population (many of
whom may have been working) may have contributed to a reduced effectiveness of TSM.
In order to improve patient outcomes, future studies of telemonitoring in this patient
population should employ large, multicenter RCT designs enrolling many more subjects,
emphasizing the health care needs of the individual patient rather than focusing on the
management of a single disease. At a minimum, telemonitoring interventions should be tailored
based on severity of disease and the patient's capacity for self-management, particularly when
employed with underserved patient populations with an extraordinarily high burden of disease.
52
REFERENCES
1. Franciosa JA, Ferdinand KC, Yancy CW. Treatment of heart failure in African Americans: a consensus statement. Congest Heart Fail. 2010;16(1):27-38.
2. Bibbins-Domingo K, Pletcher MJ, Lin F, et al. Racial differences in incident heart failure among young adults. N Engl J Med. 2009;360(12):1179-1190.
3. Albert NM, Barnason S, Deswal A, et al. Transitions of care in heart failure: a scientific statement from the American Heart Association. Circ Heart Fail. 2015;8(2):384-409.
4. Choi D, Nemi E, Fernando C, Gupta M, Mjoe GW. Differences in the clinical characteristics of ethnic minority groups with heart failure managed in specialized heart failure clinics. JACC Heart Fail. 2014;2(4):392-399.
5. Wong MD, Shapiro MF, Boscardin WJ, Ettner SL. Contribution of major diseases to disparities in mortality. N Engl J Med. 2002;347(20):1585-1592.
6. Mensah GA, Mokdad AH, Ford ES, Greenlund KJ, Croft JB. State of disparities in cardiovascular health in the United States. Circulation. 2005;111(10):1233-1241.
7. Vivo RP, Krim SR, Cevik C, Witteles RM. Heart failure in Hispanics. J Am Coll Cardiol. 2009;53(14):1167-1175.
8. Daviglus ML, Talavera GA, Aviles-Santa ML, et al. Prevalence of major cardiovascular risk factors and cardiovascular diseases among Hispanic/Latino individuals of diverse backgrounds in the United States. JAMA. 2012;308(17):1775-1784.
9. Yancy CW. Heart failure in African Americans. Am J Cardiol. 2005;96(7B):3i-12i.
10. Yancy CW, Strong M. The natural history, epidemiology, and prognosis of heart failure in African Americans. Congest Heart Fail. 2004;10(1):15-18, quiz 21-12.
11. Tuckson RV, Edmunds M, Hodgkins ML. Telehealth. N Engl J Med. 2017;377:1585-1592.
12. Bashi N, Karunanithi M, Fatehi F, Ding H, Walters D. Remote monitoring of patients with heart failure: an overview of systematic reviews. J Med Internet Res. 2017;19(1):e18. doi: 10.2196/jmir.6571
13. Kobb R, Hoffman N, Lodge R, Kline S. Enhancing elder chronic care through technology and care coordination: report from a pilot. Telemed J E Health. 2003;9(2):189-195.
53
14. Kotb A, Cameron C, Hsieh S, Wells G. Comparative effectiveness of different forms of telemedicine for individuals with heart failure (HF): a systematic review and network meta-analysis. PLoS One. 2015;10(2):e0118681. doi: 10.1371/journal.pone.0118681
15. Clark RA, Inglis SC, McAlister FA, Cleland JG, Stewart S. Telemonitoring or structured telephone support programmes for patients with chronic heart failure: systematic review and meta-analysis. BMJ. 2007;334(7600):942.
16. Schofield RS, Kline SE, Schmalfuss CM, et al. Early outcomes of a care coordination-enhanced telehome care program for elderly veterans with chronic heart failure. Telemed J E Health. 2005;11(1):20-27.
17. Finkelstein SM, Speedie SM, Potthoff S. Home telehealth improves clinical outcomes at lower cost for home healthcare. Telemed J E Health. 2006;12(2):128-136.
18. Hebert MA, Korabek B, Scott RE. Moving research into practice: a decision framework for integrating home telehealth into chronic illness care. Int J Med Inform. 2006;75(12):786-794.
19. Cooper R, Cutler J, Desvigne-Nickens P, et al. Trends and disparities in coronary heart disease, stroke, and other cardiovascular diseases in the United States: findings of the national conference on cardiovascular disease prevention. Circulation. 2000;102(25):3137-3147.
20. Inglis SC, Clark RA, Dierckx R, Prieto-Merino D, Cleland JG. Structured telephone support or non-invasive telemonitoring for patients with heart failure. Cochrane Database Syst Rev. 2015(10):CD007228. doi: 10.1002/14651858.CD007228.pub3
21. Ong MK, Romano PS, Edgington S, et al. Effectiveness of remote patient monitoring after discharge of hospitalized patients with heart failure: the better effectiveness after transition—heart failure (BEAT-HF) randomized clinical trial. JAMA Intern Med. 2016;176(3):310-318.
22. Chaudhry SI, Mattera JA, Curtis JP, et al. Telemonitoring in patients with heart failure. N Engl J Med. 2010;363(24):2301-2309.
23. Trief PM, Izquierdo R, Eimicke JP, et al. Adherence to diabetes self care for white, African-American, and Hispanic American telemedicine participants: 5-year results from the IDEATel project. Ethn Health. 2013;18:83-96.
24. Artinian NT, Washington OG, Templin TN. Effects of home telemonitoring and community-based monitoring on blood pressure control in urban African Americans: a pilot study. Heart Lung. 2001;30(3):191-199.
54
25. Riegel B, Carlson B, Glaser D, Romero T. Randomized controlled trial of telephone case management in Hispanics of Mexican origin with heart failure. J Card Fail. 2006;12(3):211-219.
26. Wingood GM, DiClemente RJ. The ADAPT-ITT model: a novel method of adapting evidence-based HIV Interventions. J Acquir Immune Defic Syndr. 2008;47 (suppl 1):S40-S46.
27. Pekmezaris R, Schwartz RM, Taylor TN, et al. A qualitative analysis to optimize a telemonitoring intervention for heart failure patients from disparity communities. BMC Med Inform Decis Mak. 2016;16:75.
28. MacQueen K. Codebook development for team-based qualitative analysis. Cultural Anthropology Methods. 1998;10(2):31-36.
29. Wallerstein NB, Duran B. Using community-based participatory research to address health disparities. Health Promot Pract. 2006;7(3):312-323.
30. Yancy CW, Jessup M, Bozkurt B, et al. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2013;62(16):e147-e239. doi: 10.1016/j.jacc.2013.05.019
31. Yancy CW, Jessup M, Bozkurt B, et al. 2013 ACCF/AHA guideline for the management of heart failure: executive summary: a report of the American College of Cardiology Foundation/American Heart Association Task Force on practice guidelines. Circulation. 2013;128(16):1810-1852.
32. Rector TS, Cohn JN. Assessment of patient outcome with the Minnesota Living with Heart Failure questionnaire: reliability and validity during a randomized, double-blind, placebo-controlled trial of pimobendan. Pimobendan Multicenter Research Group. Am Heart J. 1992;124(4):1017-1025.
33. Kroenke K, Spitzer RL, Williams JB, Lowe B. An ultra-brief screening scale for anxiety and depression: the PHQ-4. Psychosomatics. 2009;50(6):613-621.
34. Pekmezaris R, Mitzner I, Pecinka KR, et al. The impact of remote patient monitoring (telehealth) upon Medicare beneficiaries with heart failure. Telemed J E Health. 2012;18(2):101-108.
55
35. Austin P, Rothwell M, Tu J. A comparison of statistical modeling strategies for analyzing length of stay after CABG surgery. Health Serv Outcomes Res Methodol. 2002;3(2):107-133.
36. Feltner C, Jones CD, Cene CW, et al. Transitional care interventions to prevent readmissions for persons with heart failure: a systematic review and meta-analysis. Ann Intern Med. 2014;160(11):774-784.
37. Konstam V, Gregory D, Chen J, et al. Health-related quality of life in a multicenter randomized controlled comparison of telephonic disease management and automated home monitoring in patients recently hospitalized with heart failure: SPAN-CHF II trial. J Card Fail. 2011;17(2):151-157.
38. Huygens MW, Vermeulen J, Swinkels IC, Friele RD, van Schayck OC, de Witte LP. Expectations and needs of patients with a chronic disease toward self-management and eHealth for self-management purposes. BMC Health Serv Res. 2016;16:232.
39. Israel BA, Schulz AJ, Parker EA, Becker AB. Community-Campus Partnerships for Health. Community-based participatory research: policy recommendations for promoting a partnership approach in health research. Educ Health. 2001;14(2):182-197.
40. Kilgore M, Patel HK, Kielhorn A, Maya JF, Sharma P. Economic burden of hospitalizations of Medicare beneficiaries with heart failure. Risk Manag Healthc Policy. 2017;10:63-70.
41. Koehler F, Winkler S, Schieber M, et al. Impact of remote telemedical management on mortality and hospitalizations in ambulatory patients with chronic heart failure: the telemedical interventional monitoring in heart failure study. Circulation. 2011;123(17):1873-1880.
56
ACKNOWLEDGMENTS We wish to thank the members of the PCORI HF Community Advisory Board for their
enormous contributions to this study and the patients who participated in the study. We also
wish to acknowledge the contributions of Jill Cotroneo; Stephanie Morahan, RN; Myia Williams,
MA; Meng Zhang, PhD; and Tito Orona for administrative, nursing, research compliance
support, statistical support, and technical support, respectively.
57
PUBLICATIONS
Pekmezaris R, Schwartz RM, Taylor T, et al. A qualitative analysis to optimize a telemonitoring intervention for heart failure patients from disparity communities. BMC Med Inform Decis Mak. 2016;16:75. doi: 10.1186/s12911-016-0300-9
Television Programming and YouTube links:
Telehealth for Heart Patients: https://www.youtube.com/watch?v=U5XamiMYyX0
Virtual Medicine: https://www.youtube.com/watch?v=_ZZ1U36FrWo
58
Copyright© 2019. Feinstein Institute for Medical Research. All Rights Reserved.
Disclaimer:
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
Acknowledgement:
Research reported in this report was [partially] funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award (#AD-1304-6294) Further information available at: https://www.pcori.org/research-results/2013/comparing-two-methods-caring-black-and-hispanic-adults-heart-failure-after