Telemonitoring NEJM

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

The n e w e ng l a n d j o u r na l of m e dic i n e

original article

Telemonitoring in Patients
with Heart Failure
Sarwat I. Chaudhry, M.D., Jennifer A. Mattera, M.P.H., Jeptha P. Curtis, M.D.,
John A. Spertus, M.D., M.P.H., Jeph Herrin, Ph.D., Zhenqiu Lin, Ph.D.,
Christopher O. Phillips, M.D., M.P.H., Beth V. Hodshon, M.P.H., J.D., R.N.,
Lawton S. Cooper, M.D., M.P.H., and Harlan M. Krumholz, M.D.

A bs t r ac t

Background
Small studies suggest that telemonitoring may improve heart-failure outcomes, but From the Sections of General Medicine
its effect in a large trial has not been established. (S.I.C.) and Cardiovascular Medicine
(J.P.C., J.H., B.V.H., H.M.K.) and the Rob-
ert Wood Johnson Clinical Scholars Pro-
Methods gram (H.M.K.), Department of Medicine,
We randomly assigned 1653 patients who had recently been hospitalized for heart and the Section of Health Policy and Ad-
ministration (J.A.M., H.M.K.), School of
failure to undergo either telemonitoring (826 patients) or usual care (827 patients). Public Health, Yale University School of
Telemonitoring was accomplished by means of a telephone-based interactive voice- Medicine; and the Center for Outcomes
response system that collected daily information about symptoms and weight that Research and Evaluation, Yale–New Ha-
ven Hospital (J.A.M., J.P.C., Z.L., H.M.K.)
was reviewed by the patients’ clinicians. The primary end point was readmission for — both in New Haven, CT; the University
any reason or death from any cause within 180 days after enrollment. Secondary end of Missouri at Kansas City School of
points included hospitalization for heart failure, number of days in the hospital, Medicine, and the Mid America Heart
and Vascular Institute, Saint Luke’s Health
and number of hospitalizations. System — both in Kansas City, MO
(J.A.S.); the Department of Medicine,
Results Morehouse School of Medicine, Atlanta
(C.O.P.); and the Division of Cardiovascu-
The median age of the patients was 61 years; 42.0% were female, and 39.0% were lar Sciences, National Heart, Lung, and
black. The telemonitoring group and the usual-care group did not differ significantly Blood Institute, Bethesda, MD (L.S.C.).
with respect to the primary end point, which occurred in 52.3% and 51.5% of pa- Address reprint requests to Dr. Krumholz
at Yale University School of Medicine,
tients, respectively (difference, 0.8 percentage points; 95% confidence interval [CI], 1 Church St., Suite 200, New Haven, CT
−4.0 to 5.6; P = 0.75 by the chi-square test). Readmission for any reason occurred in 06510, or at harlan.krumholz@yale.edu.
49.3% of patients in the telemonitoring group and 47.4% of patients in the usual-
This article (10.1056/NEJMoa1010029) was
care group (difference, 1.9 percentage points; 95% CI, −3.0 to 6.7; P = 0.45 by the published on November 16, 2010, and up-
chi-square test). Death occurred in 11.1% of the telemonitoring group and 11.4% of dated on February 2, 2011, at NEJM.org.
the usual care group (difference, −0.2 percentage points; 95% CI, −3.3 to 2.8; P = 0.88
N Engl J Med 2010;363:2301-9.
by the chi-square test). There were no significant differences between the two groups Copyright © 2010 Massachusetts Medical Society.
with respect to the secondary end points or the time to the primary end point or its
components. No adverse events were reported.

Conclusions
Among patients recently hospitalized for heart failure, telemonitoring did not
improve outcomes. The results indicate the importance of a thorough, indepen-
dent evaluation of disease-management strategies before their adoption. (Funded
by the National Heart, Lung, and Blood Institute; ClinicalTrials.gov number,
NCT00303212.)

n engl j med 363;24 nejm.org december 9, 2010 2301


The New England Journal of Medicine
Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
The n e w e ng l a n d j o u r na l of m e dic i n e

D
espite advances in the care of pa- the telemonitoring system, was paid for its ser-
tients with heart failure, outcomes after vices and had no role in the study design, the col-
hospitalization are not improving.1 De- lection or analysis of the data, or the writing of
veloping strategies to reduce readmissions of pa- the manuscript. The study was designed by one
tients with heart failure is a national priority, as of the authors, data were gathered by site coor-
indicated by national improvement initiatives2; fed- dinators and the Yale Coordinating Center, and the
eral, publicly reported performance measures3,4; data were analyzed by two of the authors. Two
and government incentives in the Patient Protec- of the authors drafted the manuscript; made the
tion and Affordable Care Act.5 decision to submit it for publication, with all the
Telemonitoring is a promising strategy for im- authors contributing to subsequent revisions; and
proving heart-failure outcomes by making it pos- vouch for the accuracy and completeness of the
sible to monitor patients remotely so that clini- data and analyses.
cians can intervene early if there is evidence of
clinical deterioration. A recent Cochrane review Eligibility and Enrollment of Patients
concluded that telemonitoring of patients with Patients were enrolled from 2006 through 2009 at
heart failure reduced the rate of death from any 33 cardiology practices across the United States
cause by 44% and the rate of heart-failure–related (see the Supplementary Appendix, available at
hospitalizations by 21%.6 However, the quality of NEJM.org). Practices were selected on the basis of
the methods used in the studies was variable, and their organizational capability and enthusiasm for
many of the studies were small. participating in the study. At each practice, a site
We conducted a multicenter, randomized, con- coordinator was assigned responsibility for over-
trolled trial, Telemonitoring to Improve Heart seeing implementation of and adherence to the
Failure Outcomes (Tele-HF), to determine whether protocol.
telemonitoring would reduce the combined end Site coordinators identified patients hospital-
point of readmission or death from any cause ized for heart failure in the previous 30 days. Ex-
among patients recently hospitalized for heart clusion criteria were residence in a long-term nurs-
failure. ing home; inability to participate in the study
protocol for any reason, including a low expected
Me thods probability of survival for the next 6 months owing
to conditions other than heart failure; inability to
Study Design and Oversight stand on a scale; severe cognitive impairment 8;
A description of the study design was published and a planned hospitalization for a procedure.
previously.7 The trial was conducted in accordance Patients were randomly assigned to receive
with the protocol (available with the full text of usual care or undergo telemonitoring, according to
this article at NEJM.org), which was approved by a sequence of computer-generated random num-
the institutional review boards at Yale University bers, with stratification on the basis of the study
School of Medicine and each participating site. site. Clinicians were instructed to treat patients
We appointed an independent data and safety in accordance with national guidelines for the
monitoring board to monitor adherence to the management of heart failure.9
protocol and to assess the recruitment and reten-
tion of patients and the quality of the data, as well Usual Care and Telemonitoring Protocols
as the safety and efficacy of the intervention. All At the time of enrollment, all patients received
study investigators and personnel (except for mem- educational materials developed by the Heart Fail-
bers of the data and safety monitoring board) were ure Society of America, and if needed, a scale. We
unaware of the treatment-group results until end- also provided patients in the telemonitoring group
point data had been finalized for all the patients. with detailed instructions and a demonstration
All patients provided written informed consent by site coordinators of how to use the telemoni-
before randomization. toring system, as well as a touch-tone telephone,
The study was funded by the National Heart, if needed.
Lung, and Blood Institute of the National Insti- Telemonitoring was performed with the use
tutes of Health. Pharos Innovations, the vendor of of a commercial system, Tel-Assurance (Pharos In-

2302 n engl j med 363;24 nejm.org december 9, 2010

The New England Journal of Medicine


Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Telemonitoring in Heart Failure

novations). We selected the system on the basis of who were unaware of the treatment assignments.
its technical quality and established use in many During these telephone interviews, information
clinical practices. was collected about the quality of life, satisfaction
The telemonitoring group was instructed to with care, and use of medications.
make daily, toll-free calls to the system. During At 6 months, site coordinators at each clinical-
each call, patients heard a series of questions practice site reviewed the office and hospital
about general health and heart-failure symptoms, medical records to ascertain readmissions. They
and they entered responses using the telephone also contacted patients and their primary care
keypad. Every 30 days, validated screening ques- providers to inquire about readmissions. Dis-
tions for symptoms of depression were also in- charge summaries, other chart documentation, or
cluded.10 Information from the telemonitoring both were obtained for all readmissions.
system was downloaded daily to a secure Inter- During the follow-up interviews at 6 months,
net site and was reviewed every weekday (except staff at the coordinating center also asked patients
on holidays) by site coordinators. All questions had whether they had been rehospitalized during the
predetermined responses that triggered “varianc- study period. The Social Security Death Index,
es” to flag clinicians’ attention (see the Supple- contact with family members, interviews with cli-
mentary Appendix). The protocol required the sites nicians, and searches for obituaries were used to
to contact any patient whose response generated ascertain vital status.
variances and document their management of the
variances. End Points
To maximize adherence to telemonitoring, pa- The primary end point was a composite of re­
tients were told that their information would be admission for any reason or death from any cause
reviewed by the clinicians responsible for man- within 180 days after enrollment. Prespecified
aging their heart failure. If patients did not use secondary end points included hospitalization for
the system for two consecutive days, they received any reason or death from any cause, hospitaliza-
a system-generated reminder call; after that, they tion for heart failure, number of days in the hos-
were contacted by site staff to encourage partici- pital, number of hospitalizations for any cause,
pation. and times to the primary end point and its com-
Sites were made aware of the importance of ponents. A committee of physicians, all of whom
timely review of telemonitoring information. Ev- were unaware of the treatment-group assignments,
ery 2 to 3 weeks, staff from the Yale Coordinat- adjudicated each potential readmission to ensure
ing Center reviewed responses to variances and that the event qualified as a readmission (and not
contacted the sites if there was no documenta- another clinical encounter such as an emergency
tion of how the variance was managed, to ensure department visit) and to determine the primary
that the information had been reviewed. cause of the readmission.
The primary anticipated adverse event associ-
ated with telemonitoring was a delay in seeking Statistical Analysis
care for urgent or emergency situations because We summarized patients’ baseline characteristics
of a belief that the telemonitoring data would according to treatment group. Adherence in the
immediately alert clinicians. To minimize this telemonitoring group was defined as placement
risk, patients were told to contact their clinicians of at least three calls a week to the telemonitor-
directly with any urgent concerns. ing system (a cutoff point representing approxi-
mately half the expected usage). We summarized
Data Collection the end points on the basis of treatment group
At baseline, site coordinators collected the medi- and tested the primary hypothesis using a chi-
cal history and physical-examination data by means square test of independence. In secondary analy-
of direct interview, examination, and medical- ses, we calculated the Kaplan–Meier time-to-event
record review. Patients were also interviewed with- function11 for readmission or death from any
in 2 weeks after enrollment (which was consid- cause. For each end point, we also estimated the
ered to be the baseline), and 3 and 6 months after corresponding hazard ratio and 95% confidence
enrollment, by staff at the coordinating center interval, using a Cox proportional-hazards mod-

n engl j med 363;24 nejm.org december 9, 2010 2303


The New England Journal of Medicine
Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
The n e w e ng l a n d j o u r na l of m e dic i n e

use of Stata software, version 11.1, and SAS soft-


ware, version 9.1.
5069 Patients were assessed for eligibility

R e sult s
3416 Were excluded
2442 Were not eligible Patients
1228 Could not participate
in the protocol A total of 5069 potential participants were screened
677 Had cardiac pro-
cedures planned
(Fig. 1). Of these, 2442 were ineligible and 974
315 Were nursing home declined to participate. The remaining 1653 pa-
residents
222 Had other reasons
tients were enrolled; 826 were randomly assigned
974 Declined to participate to undergo telemonitoring and 827 to receive usu-
al care. A total of 79% of the study patients com-
pleted the 6-month interview, with no between-
1653 Underwent randomization
group difference in the rate of completion.
Medical-record review for readmissions and vi-
tal-status verification were completed for 100%
of patients.
Baseline characteristics of the patients were
826 Were assigned to telemonitoring 827 Were assigned to usual care
707 Activated the system 827 Received usual care similar between the two groups (Table 1). The
119 Did not activate the system median age was 61 years; 42.0% of the patients
were female, 39.0% were black, and 70.6% had a
depressed left ventricular ejection fraction (<40%).
Diabetes mellitus, hypertension, and coronary ar-
826 Were included in primary analysis 827 Were included in primary analysis
tery disease were the most common coexisting
conditions, and 46.0% of patients had chronic
Figure 1. Screening, Randomization, and Follow-up of the Study Patients. kidney disease (defined as a glomerular filtration
AUTHOR: Krumholtz RETAKE: 1st
2nd
rate of <60 ml per minute).13 At enrollment, 66.9%
FIGURE: 1 of 3 3rd of patients had received a prescription for either
Revised
ARTIST:el.
12
ts The time to each event was calculated start- an angiotensin-converting–enzyme inhibitor or
SIZE
ing at the day of enrollment. We 4 col compared the an angiotensin-receptor blocker, and 79.2% had
TYPE: Line Combo 4-C H/T
numbers of readmissions and 22p3 hospital days be- received a prescription for a beta-blocker. Of the
AUTHOR, PLEASE NOTE:
tween the two groups using the
Figure has been redrawn and type has been reset.
Wilcoxon rank- patients with a depressed ejection fraction who
sum test. Please check carefully. did not have chronic kidney disease, 82.9% had
JOB: 36324
Prespecified subgroup analyses included as-
ISSUE: 12-09-10
received a prescription for an angiotensin-convert-
sessments of the primary end point within sub- ing–enzyme inhibitor or an angiotensin-receptor
groups based on age (<65 years vs. ≥65 years), sex, blocker. Of all the patients with a depressed ejec-
race, left ventricular ejection fraction (<40% vs. tion fraction, 82.9% had received a prescription
≥40%), and New York Heart Association class (I or for a beta-blocker.
II vs. III or IV), as well as the number of patients
enrolled at each site (<100 vs. ≥100). We also tested Primary End Point
for interactions between each pair of subgroups A total of 85.6% of patients in the telemonitoring
and the main treatment effect. group made at least one call; among these pa-
Our study was designed to have a power of 80% tients, adherence to the intervention was highest,
to detect a 15% relative reduction in the primary 90.2%, during the first week of the study period
end point in the telemonitoring group as com- and decreased to 55.1% by week 26. A total of
pared with the usual-care group. We calculated 29,163 variances were generated during the study
that 820 patients per treatment group would need period, with a median of 21 (interquartile range,
to be enrolled, assuming a primary-end-point rate 5 to 54) per patient.
of 50% in the usual-care group, 10% loss to fol- No significant difference was seen between
low-up in the study population, use of an inten- the two groups in the rate of the primary end
tion-to-treat analysis, and an alpha value of 0.05. point, which occurred in 432 patients (52.3%) in
All statistical analyses were performed with the the telemonitoring group and in 426 patients

2304 n engl j med 363;24 nejm.org december 9, 2010

The New England Journal of Medicine


Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Telemonitoring in Heart Failure

Table 1. Baseline Characteristics of the Study Population, According to Treatment Group.*

Telemonitoring Usual Care Total


Characteristic (N = 826) (N = 827) (N = 1653)
Age — yr
Median 61 61 61
Interquartile range 51–73 51–73 51–73
Female sex — no. (%) 359 (43.5) 336 (40.6) 695 (42.0)
Race — no. (%)†
White 414 (50.1) 401 (48.5) 815 (49.3)
Black 314 (38.0) 330 (39.9) 644 (39.0)
Other 98 (11.9) 96 (11.6) 194 (11.7)
Hispanic or Latino ethnic group — no. (%)† 22 (2.7) 23 (2.8) 45 (2.7)
NYHA class — no. (%)
I 48 (5.8) 52 (6.3) 100 (6.0)
II 301 (36.4) 306 (37.0) 607 (36.7)
III 416 (50.4) 423 (51.1) 839 (50.8)
IV 61 (7.4) 46 (5.6) 107 (6.5)
Left ventricular ejection fraction <40% — no./total no. (%) 572/806 (71.0) 563/802 (70.2) 1135/1608 (70.6)
Chronic kidney disease — no./total no. (%)‡ 370/814 (45.5) 378/813 (46.5) 748/1627 (46.0)
COPD — no. (%) 169 (20.5) 177 (21.4) 346 (20.9)
Diabetes mellitus — no. (%) 394 (47.7) 378 (45.7) 772 (46.7)
Hypertension — no. (%) 632 (76.5) 639 (77.3) 1271 (76.9)
Coronary artery disease — no. (%) 432 (52.3) 403 (48.7) 835 (50.5)
Blood pressure — mm Hg
Systolic 121.5±22.8 120.3±21.6 120.9±22.2
Diastolic 71.1±13.9 70.6±13.7 70.8±13.8
Serum potassium — mmol/liter 4.1±0.6 4.1±0.6 4.1±0.6
Blood urea nitrogen — mg/dl 27.8±18.0 26.8±17.2 27.3±17.6
Serum creatinine — mg/dl 1.5±0.8 1.4±0.7 1.5±0.7
Medications — no. (%)
ACE inhibitor or ARB 549 (66.5) 557 (67.4) 1106 (66.9)
Beta-blocker 668 (80.9) 641 (77.5) 1309 (79.2)
Loop diuretic 646 (78.2) 646 (78.1) 1292 (78.2)
Digoxin 214 (25.9) 198 (23.9) 412 (24.9)
Aldosterone-receptor antagonist 266 (32.2) 277 (33.5) 543 (32.8)
Did not graduate from high school — no./total no. (%) 167/701 (23.8) 171/711 (24.1) 338/1412 (23.9)
Annual household income <$10,000 — no./total no. (%) 168/571 (29.4) 152/585 (26.0) 320/1156 (27.7)

* All comparisons between the two groups were nonsignificant (P>0.05). To convert values for blood urea nitrogen to
millimoles per liter, multiply by 0.357. To convert values for creatinine to micromoles per liter, multiply by 88.4. ACE
denotes angiotensin-converting enzyme, ARB angiotensin-receptor blocker, COPD chronic obstructive pulmonary dis-
ease, and NYHA New York Heart Association.
† Race and Hispanic or Latino ethnic group (which was considered a subcategory of white race) were self-reported.
‡ Chronic kidney disease was defined by a glomerular filtration rate of less than 60 ml per minute.13

(51.5%) in the usual-care group (difference, 0.8 ble 2). The hazard ratio for the primary end point
percentage points; 95% confidence interval [CI], with telemonitoring versus usual care was 1.04
−4.0 to 5.6; P = 0.75 by the chi-square test) (Ta­ (95% CI, 0.91 to 1.19).

n engl j med 363;24 nejm.org december 9, 2010 2305


The New England Journal of Medicine
Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
The n e w e ng l a n d j o u r na l of m e dic i n e

cluding age (<65 years vs. ≥65 years), sex, race, left
Table 2. Clinical End Points, According to Treatment Group.*
ventricular ejection fraction (<40% vs. ≥40%), and
Telemonitoring Usual Care New York Heart Association class (I or II vs. III or
End Point (N = 826) (N = 827) P Value IV) — identified a group in which telemonitoring
Primary end point: death or readmission 432 (52.3) 426 (51.5) 0.75 was effective. The intervention was no more ef-
— no. (%)
fective at sites with greater experience using the
Secondary end points telemonitoring system (i.e., ≥100 patients enrolled
Death — no. (%) 92 (11.1) 94 (11.4) 0.88 in the study). The only significant interaction
Readmission — no. (%) 407 (49.3) 392 (47.4) 0.45 found was for sex, but there was no indication
Readmission for heart failure 227 (27.5) 223 (27.0) 0.81 that the intervention was beneficial in either men
— no. (%) or women.
No. of days in hospital 7.2±14.6 7.0±14.9 0.27
No. of readmissions — no. (%) 0.20
Adverse Events
No adverse events were reported during the study
0 419 (50.7) 435 (52.6)
period.
1 199 (24.1) 212 (25.6)
2 97 (11.7) 88 (10.6)
Discussion
3 53 (6.4) 52 (6.3)
4 33 (4.0) 20 (2.4) In our multicenter, randomized, controlled trial in-
≥5 25 (3.0) 20 (2.4) volving patients recently hospitalized with heart
failure, we found no reduction in the risk of re-
* Plus–minus values are means ±SD. admission or death from any cause with telemon-
itoring as compared with usual care. Moreover,
there were no reductions in the risk of hospital-
Secondary End Points ization for heart failure, the number of days in the
No significant differences were seen between the hospital, or the time to readmission or death. Sub-
two groups with respect to the secondary end group analyses failed to identify a group for which
points (Table 2). Readmission for any cause oc- the intervention was effective, despite efforts to
curred in 407 patients (49.3%) in the telemonitor- include sites and patients who demonstrated en-
ing group and 392 patients (47.4%) in the usual- thusiasm for participation in a daily telemonitor-
care group (difference, 1.9 percentage points; ing program and screening of the patients for
95% CI, −3.0 to 6.7; P = 0.45 by the chi-square test). their ability to follow the protocol. These results
The hazard ratio for readmission for any cause contrast with the findings of a recent Cochrane
with telemonitoring was 1.06 (95% CI, 0.93 to review of telemonitoring for patients with heart
1.22). A total of 92 patients (11.1%) in the tele- failure6; however, our study was of higher meth-
monitoring group and 94 patients (11.4%) in the odologic quality and was larger than most of the
usual-care group died during the 180-day study studies included in the review.
period (difference, −0.2 percentage points; 95% CI, The efficacy of complex interventions such as
−3.3 to 2.8; P = 0.88 by the chi-square test). The telemonitoring depends on the context in which
hazard ratio for death was 0.97 (95% CI, 0.73 to they are applied.14,15 For our study, we chose a
1.30). Readmissions for heart failure, the number telemonitoring system that is currently in use,
of days in the hospital, and the number of re­ collected standardized clinical information, en-
admissions were also similar in the two groups sured that site personnel were thoroughly trained
(Table 2). Kaplan–Meier time-to-event curves for by the system vendor, paid sites for participation,
the composite end point of readmission or death and provided patients with ongoing support (in-
from any cause, as well as for each component cluding the provision of scales and telephones, as
separately, did not reveal a significant difference needed) in using the telemonitoring system. Only
between the two groups (Fig. 2). sites that were willing to make a strong commit-
ment to integrating this system into their prac-
Subgroup Analyses tices were included in the study, and we worked
Subgroup analyses (Fig. 3) showed that none of closely with them to ensure sustained adherence
the baseline characteristics of the patients — in- to the protocol. Moreover, each site was required

2306 n engl j med 363;24 nejm.org december 9, 2010

The New England Journal of Medicine


Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Telemonitoring in Heart Failure

Figure 2. Kaplan–Meier Time-to-Event Estimates A Readmission for Any Reason or Death from Any Cause
for the Primary End Point — Readmission for Any
1.00
Reason or Death from Any Cause — and Each Compo-
nent Separately, According to Treatment Group.

Probability of Freedom from


0.75

Readmission or Death
to document its response to each variance, ensur-
ing that an explicit clinical decision was made Usual care

about each variance. These design features created 0.50


Telemonitoring
a reasonable setting in which to determine the ef-
fect of telemonitoring, were it to be broadly used 0.25
in an effort to reduce hospitalizations. Hazard ratio for readmission or death with tele-
monitoring, 1.04 (95% CI, 0.91–1.19)
In interpreting our results, it is important P=0.58
to recognize that automated telemonitoring rep- 0.00
0 60 120 180
resents a single, focused approach to disease
Days since Enrollment
management. The American Heart Association
No. at Risk
developed a classification system for disease man- Usual care 827 587 468 402
agement, and several of the components of this Telemonitoring 826 564 454 395
system merit consideration.16 The intervention in
our study consisted of providing physicians with B Readmission for Any Reason
increased information about their patients’ clin- 1.00

ical status; it is possible that including formal


education, medication management, or peer sup- Probability of Freedom
0.75
port would have enhanced the effectiveness of
from Readmission
Usual care
the intervention. The method of communication
used was an interactive voice-response system, 0.50 Telemonitoring
with contact by clinicians driven by their assess-
ment of patients’ status; an intervention with more 0.25
contact between clinicians and patients might have Hazard ratio for readmission with telemonitoring,
1.06 (95% CI, 0.93–1.22)
been more effective. P=0.39
In a previous, small, single-site trial of remote 0.00
0 60 120 180
monitoring of patients, our group found a 44-per-
Days since Enrollment
centage-point reduction in the rate of readmis-
sion, which was associated with significant cost No. at Risk
Usual care 827 587 468 402
savings.17 However, we were concerned that, in Telemonitoring 826 564 454 395
that trial, reliance on a single, highly skilled and
motivated nurse case-manager who deployed an C Death from Any Cause
intervention developed by the investigative team 1.00
Telemonitoring
limited the generalizability and scalability of the Usual care
findings. Moreover, the results prompted the ques-
Probability of Freedom

0.75
tion of whether an automated monitoring system
from Death

with transmission of information to the clinicians


responsible for the patients’ care, precluding the 0.50

need for one-to-one telephone calls with a clini-


cian, could provide a similar benefit with the po- 0.25
tential for widespread dissemination. Hazard ratio for death with telemonitoring,
0.97 (95% CI, 0.73–1.30)
Our results underscore the need for rigorous, P=0.86
independent evaluation of disease-management 0.00
0 60 120 180
systems before their adoption. In an environment
Days since Enrollment
in which vendors promote their products to health
No. at Risk
systems that are under increasing pressure to re- Usual care 827 789 756 733
duce readmission rates, the knowledge that tele- Telemonitoring 826 792 763 735
monitoring is ineffective suggests the need to
AUTHOR: Krumholtz RETAKE: 1st
2nd
n engl j med 363;24 nejm.org FIGURE:
december 2 of 3
9, 2010 3rd 2307
Revised
The New England Journal of ARTIST:
Medicinets
SIZE
Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. ForTYPE:personalLine
use only. No other 4 col
Combo 4-CusesH/T
without permission.
22p3
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
AUTHOR, PLEASE NOTE:
The n e w e ng l a n d j o u r na l of m e dic i n e

Telemoni- Usual Telemoni- Usual P Value for


Subgroup toring Care toring Care Hazard Ratio (95% CI) P Value Interaction
no. of patients no. of events
All patients 826 827 432 426 1.04 (0.91–1.19) 0.58
Age 0.21
<65 yr 496 521 261 259 1.11 (0.93–1.32) 0.24
≥65 yr 330 306 171 167 0.93 (0.75–1.15) 0.51
Sex 0.02
Male 467 491 264 249 1.18 (0.99–1.41) 0.06
Female 359 336 168 177 0.87 (0.70–1.07) 0.18
Race 0.97
White 414 401 227 215 1.05 (0.87–1.27) 0.59
Black 314 330 160 167 1.03 (0.83–1.28) 0.81
Other 98 96 45 44 1.00 (0.66–1.51) 0.99
LVEF 0.94
≥40% 234 239 125 122 1.07 (0.83–1.37) 0.60
<40% 572 563 298 286 1.05 (0.90–1.24) 0.52
NYHA class 0.05
I or II 349 358 153 174 0.88 (0.71–1.09) 0.25
III or IV 477 469 279 252 1.15 (0.97–1.37) 0.10
Per-site enrollment 0.26
<100 patients 306 311 161 174 0.94 (0.76–1.17) 0.60
≥100 patients 520 516 271 252 1.10 (0.93–1.31) 0.26
0.5 1.0 1.5

Usual Care Telemonitoring


Worse Worse

Figure 3. Subgroup Analyses of Readmission or Death


AUTHOR: from Any Cause.
Krumholtz RETAKE: 1st
The size of the squares representing the hazard ratios reflects the relative number of events.2nd
Data on left ventricular ejection fraction
FIGURE: 3 of 3 3rd
(LVEF) were missing for 45 patients. NYHA denotes New York Heart Association. Revised
ARTIST: ts
SIZE
7 col
TYPE: Line Combo 4-C H/T 36p6

consider alternative approaches toAUTHOR, PLEASE


improving NOTE: their ability to participate and their enthusiasm.
care.
Figure has been redrawn and type has been reset.
Our findings also raise questions Please aboutcheck value Although sites were required to document their
thecarefully.
of findings that areJOB:based
36324
on a systematic review responses to variances, clinicians did not record
ISSUE: 12-09-10
that includes many low-quality studies. these data in a systematic manner. We do know
An important consideration is that 14% of our that variances were reviewed and purposeful deci-
patients who were randomly assigned to undergo sions were made by the cardiologists responsible
telemonitoring never used the system. Moreover, for clinical management of patients’ heart fail-
by the final week of the study period, only 55% ure, including medication adjustments, education
of the patients were still using the system at least about dietary indiscretions, and referrals for of-
three times per week. This finding is important, fice visits.
given that considerable resources, which would be A total of 21% of the study patients did not
difficult to leverage outside a clinical trial, were complete the final telephone interview at 6 months.
directed toward optimizing patients’ engagement This rate is not surprising, given the severity of
with the system. Thus, the adherence rates in this illness in the study population. Missing data for
trial most likely represent the best-case scenario these patients should have had minimal influence
and are in fact similar to previously documented on our assessment of hospitalization and vital sta-
rates of medication adherence.18 tus, which were verified through medical-record
It is also possible that the telemonitoring strat- review and electronic databases.
egy would be more effective were it embedded in In summary, a telemonitoring strategy failed
cardiology practices with greater organizational to provide a benefit over usual care in a setting
capacity to implement it. However, the diverse optimized for its use. Previous claims of success
sites in our study were selected on the basis of of similar strategies, based on studies with small

2308 n engl j med 363;24 nejm.org december 9, 2010

The New England Journal of Medicine


Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Telemonitoring in Heart Failure

populations of patients and methodologic weak- ough, independent evaluation of disease-manage-


nesses, are not supported by the results of our ment strategies before their widespread adoption.
large, multicenter trial. There remains a need for Supported by a grant (5 R01 HL080228) from the National
Heart, Lung, and Blood Institute.
strategies to improve heart-failure outcomes, and Disclosure forms provided by the authors are available with
our findings indicate the importance of a thor- the full text of this article at NEJM.org.

References
1. Kosiborod M, Lichtman JH, Heiden- Failure Outcomes (Tele-HF): study de- A new equation to estimate glomerular
reich PA, et al. National trends in outcomes sign. J Card Fail 2007;13:709-14. filtration rate. Ann Intern Med 2009;150:
among elderly patients with heart failure. 8. Folstein MF, Folstein SE, McHugh PR. 604-12.
Am J Med 2006;119(7):616.e1-616.e7. “Mini–mental state”: a practical method 14. Craig P, Dieppe P, Macintyre S, Michie
2. H2H: Hospital to Home home page. for grading the cognitive state of patients S, Nazareth I, Petticrew M. Developing
(http://www.h2hquality.org.) for the clinician. J Psychiatr Res 1975;12: and evaluating complex interventions: the
3. Keenan PS, Normand SL, Lin Z, et al. 189-98. new Medical Research Council guidance.
An administrative claims measure suitable 9. Hunt SA, Abraham WT, Chin MH, et al. BMJ 2008;337:a1655.
for profiling hospital performance on the ACC/AHA 2005 guideline update for the 15. Clark AM, Savard LA, Thompson DR.
basis of 30-day all-cause readmission rates diagnosis and management of chronic What is the strength of evidence for heart
among patients with heart failure. Circ heart failure in the adult: a report of the failure disease-management programs?
Cardiovasc Qual Outcomes 2008;1:29-37. American College of Cardiology/American J Am Coll Cardiol 2009;54:397-401.
4. Krumholz HM, Merrill AR, Schone Heart Association Task Force on Practice 16. Krumholz HM, Currie PM, Riegel B,
EM, et al. Patterns of hospital perfor- Guidelines (Writing Committee to Update et al. A taxonomy for disease management:
mance in acute myocardial infarction and the 2001 Guidelines for the Evaluation and a scientific statement from the American
heart failure 30-day mortality and read- Management of Heart Failure): developed Heart Association Disease Management
mission. Circ Cardiovasc Qual Outcomes in collaboration with the American College Taxonomy Writing Group. Circulation
2009;2:407-13. of Chest Physicians and the International 2006;114:1432-45.
5. Pub. L. 111-148: Patient Protection Society for Heart and Lung Transplanta- 17. Krumholz HM, Amatruda J, Smith
and Affordable Care Act. Washington, DC: tion: endorsed by the Heart Rhythm Soci- GL, et al. Randomized trial of an educa-
Government Printing Office. (http://www ety. Circulation 2005;112(12):e154-e235. tion and support intervention to prevent
.gpo.gov/fdsys/pkg/PLAW-111publ148/ 10. Kroenke K, Spitzer RL, Williams JB. readmission of patients with heart failure.
content-detail.html.) The Patient Health Questionnaire-2: valid- J Am Coll Cardiol 2002;39:83-9.
6. Inglis SC, Clark RA, McAlister FA, et al. ity of a two-item depression screener. 18. Cutler DM, Everett W. Thinking out-
Structured telephone support or telemon- Med Care 2003;41:1284-92. side the pillbox — medication adherence
itoring programmes for patients with 11. Kaplan EL, Meier P. Nonparametric as a priority for health care reform. N Engl
chronic heart failure. Cochrane Database estimation from incomplete observations. J Med 2010;362:1553-5.
Syst Rev 2010;8:CD007228. J Am Stat Assoc 1958;53:457-81. Copyright © 2010 Massachusetts Medical Society.
7. Chaudhry SI, Barton B, Mattera JA, 12. Cox DR, Oakes D. Analysis of survival
Spertus J, Krumholz HM. Randomized data. London: Chapman & Hall, 1984.
trial of Telemonitoring to Improve Heart 13. Levey AS, Stevens LA, Schmid CH, et al.

personal archives in the journal online


Individual subscribers can store articles and searches using a feature
on the Journal’s Web site (NEJM.org) called “Personal Archive.”
Each article and search result links to this feature. Users can create
personal folders and move articles into them for convenient retrieval later.

n engl j med 363;24 nejm.org december 9, 2010 2309


The New England Journal of Medicine
Downloaded from nejm.org by ANTHONY VIERA on October 10, 2011. For personal use only. No other uses without permission.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.

You might also like