Increased healthcare utilization associated with complete atrioventricular block in pacemaker patients
Journal of Interventional Cardiac Electrophysiology
Increased healthcare utilization associated with complete atrioventricular block in pacemaker patients
Suneet Mittal 0 1 2 3
Dan L. Musat 0 1 2 3
Michael H. Hoskins 0 1 2 3
Julie B. Prillinger 0 1 2 3
Gregory J. Roberts 0 1 2 3
Yelena Nabutovsky 0 1 2 3
Faisal M. Merchant 0 1 2 3
0 Emory University School of Medicine , Atlanta, GA , USA
1 New York , USA
2 The Snyder Center for Comprehensive Atrial Fibrillation, the Valley Health System , Ridgewood, NJ , USA
3 Abbott , Sylmar, CA , USA
Purpose The purpose of the current study is to characterize and quantify the impact of complete atrioventricular block (cAVB) on heart failure hospitalization (HFH) and healthcare utilization in pacemaker (PM) patients. Methods Patients ≥ 18 years implanted with a dual-chamber PM from April 2008 to March 2014 were selected from the MarketScan® Commercial and Medicare Supplemental claims databases. Patients with ≤ 1-year continuous MarketScan enrollment prior to and post-implant, and those with prior HF diagnosis were excluded. Patients were dichotomized into those with cAVB, defined as a 3rd degree AVB diagnosis or AV node ablation in the year prior to PM implant, versus those without any AVB (noAVB). Post-implant HFH and associated costs were compared based on inpatient claims. Results The study cohort included 21,202 patients, of which 14,208 had no AVB and 6994 had cAVB, followed for 2.39 and 2.27 years, respectively. Patients with cAVB were associated with a significantly increased risk of cumulative HFH (HR 1.59 [95% CI 1.35-1.86] p < 0.001) and significantly higher costs ($636 [609-697] vs $369 [353-405] per pt-year, p < 0.001) compared to those with no AVB. Conclusions Among dual-chamber PM patients without prior HF, cAVB is associated with a significantly increased risk of HFH and greater HF-related healthcare utilization. Identifying patients at high risk for HF in the setting of RV pacing, and potentially earlier use of biventricular or selective conduction system pacing, may reduce HF-related healthcare utilization.
Pacemakers; Heart failure; Atrioventricular block; Right ventricular pacing; Healthcare utilization
Pacemaker implantation is most commonly performed in
patients with symptomatic sinus node dysfunction or
atrioventricular block (AVB). It is now recognized that some patients
Research was supported by Abbott.
* Suneet Mittal
develop a pacing induced cardiomyopathy due to the
dyssynchrony induced by right ventricular pacing [1, 2].
Although pacing algorithms have been developed to minimize
ventricular pacing in patients with sinus node dysfunction,
patients with advanced heart block require ventricular pacing.
Most recent studies have focused on showing incidence of
new heart failure onset associated with AV block and
identifying predictors of pacing induced cardiomyopathy [1, 3–5].
However, the impact of the new heart failure onset on
healthcare utilization has not been studied. Heart failure
imposes an enormous burden on the healthcare system,
consuming more Medicare dollars than any other diagnosis .
Therefore, in this large retrospective study using real-world
data from a nationwide billing claims database, we sought to
quantify the impact of complete AVB at the time of pacemaker
implant on heart failure-related healthcare utilization.
Specifically, we sought to compare incidence of heart failure
hospitalizations and concomitant heart failure-related costs
between patients with and without complete AVB at the time
of dual-chamber pacemaker implantation. We have previously
published on the clinical experience of these pacemaker
patients and found an elevated risk of new HF development
associated with complete AVB . This work builds on the
prior analysis by evaluating the associated impact to the US
Data source Retrospective data for this study were derived
from the Truven Health MarketScan® Commercial Claims
and Medicare Supplemental databases, which capture paid
and adjudicated billing claims from inpatient hospital
encounters and outpatient physician office visits for privately insured
and Medicare Supplemental patients throughout the USA. The
nationally representative databases include records from > 170
million enrollees since 1995  and have supported
publications on outcomes of patients undergoing cardiac procedures
and receiving implantable electronic devices [1, 8–10].
Study population Patients implanted with a de novo dual
chamber pacemaker (Current Procedural Terminology
[CPT] code 33208 and/or Healthcare Common Procedure
Coding System [HCPCS] codes C1785, C2619) from any
manufacturer between April 1, 2008, and March 31, 2014,
were selected for study inclusion. De novo implants were
identified in the MarketScan® databases as pacemaker (PM)
patients without a prior device implant and without a remote
or in-office PM follow-up visit in the 1 year prior to implant.
Patients with a left ventricular lead placed (CPT codes 33224
or 33225) at the time of PM implant were excluded. All
included patients had at least 1 year of continuous MarketScan®
enrollment prior to and post-PM implant, as evidenced by a
monthly enrollment indicator in the MarketScan® database.
Finally, patients were required to be ≥ 18 and ≤ 100 years old
at the time of PM implant and without a primary or secondary
diagnosis of heart failure (HF) prior to PM implant.
To evaluate the impact of atrioventricular block (AVB) on
hospitalizations following PM implant, the study cohort was
dichotomized into patients with a diagnosis of complete AVB
(cAVB) versus those without a diagnosis of AVB (noAVB).
Patients with cAVB were identified by a diagnosis of third
degree AVB (International Classification of Diseases, Ninth
Revision [ICD-9] code 426.0) or an ablation of the
atrioventricular junction (AVJ) (CPT code 93650) in the 1 year prior to
PM implant. Patients with an AVJ ablation occurring > 1 year
prior to PM implant or at any time after PM implant were
excluded from the study. The noAVB cohort included patients
who were never diagnosed with any degree of AVB (ICD-9
codes 426.0–426.1) throughout the study period. Patients with
cAVB were presumed to have a high burden of right
ventricular (RV) pacing relative to noAVB patients, although the
actual percent of RV pacing is not available in the
Patient demographics were characterized using age, sex,
remote monitoring status, US region, year of PM implant,
and 20 baseline (≤ 1 year prior to implant) comorbidities based
on the Charlson comorbidity index. Patients were defined as
active on remote monitoring if they transmitted ≥ 1 remote
follow-up within 1 year following PM implant. US regions
included Northeast, North Central, South, and West. Claims
codes used for diagnoses and procedures were collected
across all available fields (up to 15) in the MarketScan®
inpatient and outpatient encounters, as shown in the Supplement
(Table S1) and validated previously [11, 12]. Propensity
scores for the diagnosis of cAVB were calculated for every
patient in the study cohort based on a multivariable logistic
regression model including all covariates used in the patient
Outcomes The primary outcomes included HF
hospitalizations (HFHs) and associated payments following
dualchamber PM implant. A HFH was identified in the
MarketScan® databases as any inpatient encounter for which
the primary diagnosis was HF-related, as outlined in the
Supplement (Table S1). The unadjusted rate of HFH (events
per 100 patient-years [pt-years]) was computed as the
cumulative number of HFH divided by the total duration of patient
follow-up for each group.
Both unadjusted (actual) and adjusted (predicted)
payments associated with HFH were computed for patients with
noAVB and cAVB. Due to differences in reimbursement rates
and patient demographics, unadjusted payments are reported
separately for patients covered by commercial insurance and
those with Medicare Supplemental plans. A two-part model
was utilized to predict the annual adjusted HFH payments in
noAVB and cAVB patients following PM implant. The
twopart model is a well-established econometric modeling
technique that accounts for samples with a large proportion of zero
measurements, common to healthcare data in which healthy
participants incur no medical costs. Further, the model enables
adjustment for patient characteristics. In part 1, a logistic
regression was used to model the likelihood of incurring
nonzero payments following PM implant, adjusting for AVB group,
follow-up duration, and the computed propensity score. Using
this model, the numeric probability of incurring nonzero
payments at 1 year post-implant was then estimated for each
patient. In part 2, using only those patients who had nonzero
hospitalization payments following PM implant, a linear
regression with gamma distribution and log link was used to
model the total hospitalization costs, adjusting again for
AVB group, follow-up duration, and the computed propensity
score. The total payments at 1 year post-implant were then
predicted for all patients using results from the linear
regression model. The final adjusted HFH payment for each patient
was computed as the product of the probability from part one
and the predicted payments from part 2.
Secondary outcomes included the length of stay (LOS) for
each hospitalization and rates of 30-day HF readmissions as
defined by the Centers for Medicare and Medicaid Services
(CMS). The LOS for each HFH was computed as the number
of days between hospital admission and discharge. A 30-day
HF readmission was identified in the MarketScan® databases
as any all-cause hospital admission occurring within 30 days
of discharge from a HF hospitalization.
Statistics Baseline characteristics were compared between
noAVB and cAVB patients meeting inclusion criteria.
Continuous variables, including follow-up duration and age,
were compared using a Student’s t test or Mann-Whitney test
if the distribution was not normal. Categorical variables, such
as sex and baseline comorbidities, were compared using a
chisquare (χ2) test.
The cumulative rate of HFH in the noAVB and cAVB
groups was compared using a Poisson regression. Inpatient
LOS and 30-day HF readmissions were compared using a
Student’s t test and chi-square (χ2) test, respectively. A
multivariable Cox proportional hazards model with Andersen-Gill
extension and propensity score adjustment was used to
evaluate HFH following PM implant. Patients were censored at
the time of upgrade to cardiac resynchronization therapy
(CRT) or at the end of MarketScan® enrollment. Billing codes
used to identify CRT upgrade are outlined in the Supplement
(Table S1). The proportional hazards assumption was tested
using Schoenfeld residuals and was met. For the outcome of
costs associated with HFH, a Mann-Whitney test was used to
compare unadjusted and adjusted payments between noAVB
and cAVB patients. Statistical significance was determined
using α = 0.05.
All analyses were performed on Revolution Analytics
Revolution R Enterprise with Open Source R version 3.1.1
or SAS version 9.3. Propensity scores were computed using
the LOGISTIC procedure in base SAS.
Study cohort The study cohort included 21,202 patients in the
MarketScan® databases, of which 14,208 had noAVB and
6994 had cAVB (Fig. 1). The mean age in the study cohort
was 74.0 ± 12.6 years and 54% of subjects were male.
Baseline characteristics are shown in Table 1. The majority
(93%) of patients in the cAVB cohort received a diagnosis of
third degree AVB or an AVJ ablation within 1 week of PM
implant, most of which (86%) occurred on the same day as
PM implant. Overall, 32 noAVB and 61 cAVB patients
underwent a CRT upgrade following initial PM implant,
accounting for < 1% of the study cohort.
HF hospitalizations following PM implant Over a median 2.35
[IQR 1.62, 3.39] years of follow-up, 459 noAVB (3.2%) and
320 cAVB (4.6%) patients were hospitalized for HF
(p < 0.001). The unadjusted rate of HFH was significantly
higher for patients with cAVB (2.28 [95% CI 2.06–2.51] per
100 pt-years) compared to those with noAVB (1.55 [95% CI
1.43–1.69] per 100 pt-years) (p < 0.001, Table 2). Patients
with cAVB were associated with a significantly increased risk
of cumulative HFH (adjusted HR 1.59 [95% CI 1.35–1.86],
p < 0.001, Fig. 2). However, the mean LOS (noAVB 5.1 ±
8.0 days; cAVB 4.6 ± 4.7 days; p = 0.181) and the rate of
30day HF readmissions (noAVB 4.9%; cAVB 5.3%; p = 0.914)
were not different between groups, indicating that the severity
of each HFH was similar for noAVB and cAVB patients
(Table 2). Interestingly, the subset of patients with commercial
insurance experienced 30-day HF readmission rates of 12.9%
overall, with no difference between noAVB and cAVB
patients (p = 1.000). The 30-day HF readmission rate for those
with Medicare Supplemental insurance was 4.2%, similarly
with no difference between AVB groups (p = 1.000). The
majority of hospitalized patients were hospitalized just one time
for HF, with a range of 0–6 total HFH over the duration of
follow-up (Fig. 3). Only three patients in the entire study
cohort (one noAVB and two cAVB) experienced greater than
four HFH following PM implant.
Payments associated with HF hospitalizations following PM
implant Patients with noAVB were associated with 42% lower
annual adjusted HFH payments compared to those with cAVB
(p < 0.001, Fig. 4). Similarly, the unadjusted mean payments
per pt-year were significantly reduced for patients with
noAVB (Table 2). Importantly, the payments per
hospitalization were not different between the two groups (Table 2),
indicating that the overall cost reduction was driven by the fewer
number of patients hospitalized in the noAVB group.
Data reported as count (%), median [interquartile range], and mean ± standard deviation. Continuous variables
were compared using a Student’s t test or Mann-Whitney test for normal and nonnormal distributions,
respectively. Categorical variables were compared using a chi-square (χ2 ) test
Patients in the study cohort that were enrolled in commercial
insurance plans were younger than those covered by Medicare
Supplemental insurance plans (56.1 ± 8.7 and 79.8 ± 7.0 years,
respectively; p < 0.001) and experienced lower rates of baseline
atrial fibrillation (28 and 42%; p < 0.001), coronary artery
disease (36 and 49%; p < 0.001), and hypertension (63 and 78%;
p < 0.001). Remote monitoring utilization was also higher for
commercially insured patients (42 versus 34%; p < 0.001).
While HFH was generally less common for patients with
commercial versus Medicare Supplemental insurance, patients
with cAVB were associated with higher rates of HFH and
associated payments compared to those with noAVB, regardless of
the type of insurance (Table 2). The median payment per HFH
was not different between patients with cAVB versus noAVB in
both the commercial (p = 0.242) and Medicare Supplemental
(p = 0.751) groups, although those hospitalizations covered by
commercial insurance were generally higher in cost compared to
hospitalizations covered by Medicare (Table 2).
2.39 years [1.63, 3.44]
2.27 years [1.58, 3.25]
N = 14,208
Top panel shows overall study cohort, middle panel only patients with commercial insurance, and bottom panel
only patients with Medicare Supplemental insurance plans. Values reported as count (%), median [interquartiles],
and mean ± standard deviation
cAVB complete atrioventricular block, CI confidence interval, HFH heart failure hospitalization, noAVB no
Using data from a large nationwide claims database, we find
that patients without an antecedent history of HF and with a
presumed high burden of RV pacing are associated with a
significantly heightened risk of HF hospitalization and
related healthcare costs. We have previously evaluated clinical
outcomes in this same population and have shown that
complete AVB is associated with increased risk of new onset HF,
which appears to develop quite soon after pacemaker
implantation . In the current study, we illustrate that over a
median of nearly 2½ years of follow-up, 4.6% of patients
with complete heart block were hospitalized with HF, which
was 44% greater than the 3.2% rate observed in patients
without complete AV block. Taken together, these analyses
suggest that the increase in new onset HF associated with
RV pacing in turn leads to a 59% increase in cumulative
hospitalizations and a 72% increase in heart failure
Hospitalization constitutes the major contributor to the
expense related to the care of HF patients . Higher costs can
occur if patients experience more severe decompensations of
HF and/or a greater number of hospitalizations during
followup. In our study, we show that although patients with cAVB
undergoing pacemaker implantation had more
hospitalizations, the length and cost per hospitalization did not vary
between patients with and without cAVB. This suggests that
costs are being driven by the number of hospitalizations and
not the severity of HF. In fact, it was the first hospitalization
for HF following pacemaker implantation that occurred more
commonly in cAVB patients.
The short- and long-term adverse impact of right
ventricular pacing is now well understood. There are changes in
electrical and mechanical activation, alterations in metabolism and
perfusion, adverse atrial and ventricular remodeling, changes
in hemodynamics, and changes in mechanical function .
Although in individual patients each of these changes has
been observed either alone or in combination, it has been more
difficult to ascertain the adverse impact of pacing in cohorts of
patients followed over time.
Prior studies (Table 3) Zhang et al. reported 304 patients who
underwent ventricular pacing for second or third-degree AV
block . Patients were excluded if their ejection fraction
was < 50% prior to pacemaker implantation, if they had an
existing diagnosis of HF, and if ventricular pacing occurred <
90% of the time during follow-up. After a median follow-up of
7.8 years, 26% of patients developed new-onset HF. Of note,
18% of the cohort underwent single-chamber ventricular
pacing and these patients were much more likely to develop HF.
Our study was limited to inpatient HFH for patients who
received a dual-chamber pacemaker, which likely explains our
lower observed incidence of HFH. Importantly, a previous
publication from our group on the same cohort analyzed in the
current study found that 28% of patients with cAVB received
a clinical diagnosis of HF in the inpatient or outpatient setting
during the 4 years following pacemaker implant, which aligns
well with the study by Zhang et al. . Ebert et al. enrolled 991
patients who underwent pacemaker implantation for either AV
block (n = 500) or sinus node disease (n = 491); the cohort
included patients with normal (> 55%, n = 791) or mildly
reduced (41–55%, n = 200) ejection fraction . Over a
followup period of 44 months, 17% of the cohort died and 6%
experienced a ≥ 2 LVEF category deterioration. The indication for
pacing and baseline ejection fraction had no impact on
outcome. Again, 14% of the cohort underwent only a
singlechamber device and heart failure was not a measured outcome
variable, unless it necessitated upgrade to a CRT device.
Two additional studies have examined the development of
a pacing induced cardiomyopathy (PICM) following right
ventricular pacing. The first study evaluated 1750 consecutive
patients who underwent pacemaker implantation; a study
cohort of 257 patients was identified who underwent single or
dual chamber pacemaker implantation, had normal LV
function at baseline, had ≥ 20% RV pacing, and had a repeat
echocardiogram ≥ 1 year following pacemaker implantation
. PICM was defined as ≥ 10% decline in LVEF, resulting in
a LVEF < 50%. During a mean follow-up of 3.3 years, ~ 20%
of patients developed a PICM. Whether this resulted in HF or
other adverse clinical events was not assessed. The second
study evaluated consecutive patients with complete heart
block and LVEF > 50% who underwent pacemaker
implantation . PICM was defined as CRT upgrade or a decline in
LVEF to ≤ 40%. During a mean of 4.3 years, 12% of the
cohort developed a PICM. Patients with ≥ 20% RV pacing
were at significantly greater risk for developing a PICM.
Our study is unique given the large sample size, inclusion of
Patients with AV block, no prior
history of HF, who were RV
paced > 90% of the time 
Patients with complete AV block
and a dual chamber pacemaker,
no prior history of HF 
Patients with baseline normal
(> 55%, n = 791) or mildly
reduced (41–55%, n = 200)
Follow-up who underwent PPM
implantation for AV block (n = 500)
or sinus node disease (n = 491). 
Patients with normal LVEF who were
RV paced > 20% of the time 
Consecutive patients with complete
heart block and LVEF > 50%
under-going PPM implantation 
Prevalence and clinical predictors
for development of HF
Clinical diagnosis of HF during an
inpatient or outpatient encounter,
as reflected by billing codes
All-cause mortality and deterioration
of LV function ≥ 2 LVEF
categories at last follow-up
Development of a PICM (≥ 10%
decrease in LVEF resulting in
LVEF < 50%)
Development of a PICM CRT
upgrade or LVEF ≤ 40%)
26% of patients developed HF,
which was associated with
28% of patients developed HF.
The incidence was higher in the
first 6 months post-implant.
Younger individuals and those
with a history of AF experienced
the highest risk of new HF
Death from any cause occurred in
17% and deterioration of LV
function ≥ 2 LVEF categories
in 6% patients.
There was no significant difference
in outcome between patients with
AV block and sinus node disease.
~ 20% likelihood of developing
12% likelihood of developing
AF atrial fibrillation, AV atrioventricular, CRT cardiac resynchronization therapy, HF heart failure, LVEF left ventricular ejection fraction, PICM
pacinginduced cardiomyopathy, PPM permanent pacemaker, RV right ventricular
only patients with a dual chamber pacemaker, and use of
claims data to identify all patients with hospitalizations related
to new-onset HF and healthcare costs associated with the care
of these patients.
Heart failure is the most common reason for hospitalization
among the elderly; although patients with HF represent only
14% of the overall Medicare population, they account for 43%
of Medicare spending . It has been estimated that a
diagnosis of HF is associated with annual costs of $8500 per
patient; three quarters of the total costs are associated with
hospital admissions, in-hospital treatment, and patient care in
nursing homes . Importantly, costs are high at the time
of initial diagnosis, likely reflecting that the initial diagnosis
is often made while patients are hospitalized . Thus, it is
important to identify patients who may be at risk for
developing HF. Our study shows that one at risk population is
comprised of patients who undergo dual-chamber pacemaker
implantation for management of complete heart block. These
patients were at significantly greater risk of being hospitalized,
which resulted in increased healthcare payments to manage
these patients. This suggests that strategies to prevent
development of heart failure in pacemaker patients may have
significant positive implications to the healthcare system. To
date, there is interest in determining whether selective
conduction system pacing or biventricular pacing may mitigate the
adverse clinical effects of right ventricular pacing in patients
with advanced heart block.
Limitations The major limitations of this study are that, given
the nature of claims data, we do not have information about
ejection fraction (either at baseline or over follow-up) and thus
cannot distinguish between heart failure with preserved or
reduced ejection fraction. Additionally, we presume the dual
chamber pacing in patients with complete heart block will
result in high burden of ventricular pacing; by nature of this
analysis, we lack information about actual percentage of
pacing delivered in these patients. As a result, we are also unable
to determine whether there is a threshold degree of pacing that
results in development of heart failure. Finally, mortality is not
available in the MarketScan® dataset used for this analysis, so
could not be evaluated.
In a large, national cohort of patients undergoing pacemaker
implantation, those with a diagnosis of complete heart block
(who likely have a high burden of RV pacing) experienced a
significantly increased risk of HF hospitalization. This was
associated with higher healthcare payments for the care of
these patients. Future efforts need to identify patients at
greatest risk and develop strategies to mitigate the need for
hospitalization in these patients, as this remains a potent driver
to overall healthcare costs.
Compliance with ethical standards
Disclosures SM: Consulting fees from Abbott.
MHH: Consulting fees from Abbott.
JBP: Stock Medtronic, Inc., stock and salary from Abbott.
GJR: Stock and salary from Abbott.
YN: Stock and salary from Abbott.
Open Access This article is distributed under the terms of the Creative
C o m m o n s A t t r i b u t i o n 4 . 0 I n t e r n a t i o n a l L i c e n s e ( h t t p : / /
creativecommons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a link
to the Creative Commons license, and indicate if changes were made.
Merchant FM , Hoskins MH , Musat DL , Prillinger JB , Roberts GJ , Nabutovsky Y , et al. Incidence and time course for developing heart failure with high-burden right ventricular pacing . Circ Cardiovasc Qual Outcomes . 2017 ; 10 ( 6 ):e003564. https://doi.org/10.1161/ CIRCOUTCOMES.117.003564.
Tops LF , Schalij MJ , Bax JJ . The effects of right ventricular apical pacing on ventricular function and dyssynchrony implications for therapy . J Am Coll Cardiol . 2009 ; 54 ( 9 ): 764 - 76 . https://doi.org/10.
1016/j.jacc. 2009 . 06 .006.
2016 ; 5(7):e003485 . https://doi.org/10.1161/JAHA.116.003485.
Khurshid S , Epstein AE , Verdino RJ , Lin D , Goldberg LR , Marchlinski FE , et al. Incidence and predictors of right ventricular 5 .
pacing-induced cardiomyopathy . Heart Rhythm . 2014 ; 11 ( 9 ): 1619 - 25 . https://doi.org/10.1016/j.hrthm. 2014 . 05 .040.
2016 ; 13 ( 12 ): 2272 - 8 . https://doi.org/10.1016/j.hrthm. 2016 . 09 .027.
Circ Cardiovasc Qual Outcomes . 2011 ; 4 ( 1 ): 68 - 75 . https://doi.org/ 10.1161/CIRCOUTCOMES.110.957225.
Hansen L , Chang S. White paper - Health research data for the real world: the MarketScan database . Ann Arbor: Truven Health Analytics; 2012 .
Lad SP , Petraglia FW 3rd, Kent AR , Cook S , Murphy KR , Dalal N , et al. Longer delay from chronic pain to spinal cord stimulation results in higher healthcare resource utilization . Neuromodulation.
2016 ; 19 ( 5 ): 469 - 76 . https://doi.org/10.1111/ner.12389.
Nazarian S , Reynolds MR , Ryan MP , Wolff SD , Mollenkopf SA , Turakhia MP . Utilization and likelihood of radiologic diagnostic imaging in patients with implantable cardiac defibrillators . J Magn Reson Imaging . 2016 ; 43 ( 1 ): 115 - 27 . https://doi.org/10.
Reynolds MR , Gunnarsson CL , Hunter TD , Ladapo JA , March JL , Zhang M , et al. Health outcomes with catheter ablation or antiarrhythmic drug therapy in atrial fibrillation: results of a propensitymatched analysis . Circ Cardiovasc Qual Outcomes . 2012 ; 5 ( 2 ): 171 - 81 . https://doi.org/10.1161/CIRCOUTCOMES.111.963108.
Birman-Deych E , Waterman AD , Yan Y , Nilasena DS , Radford MJ , Gage BF . Accuracy of ICD-9-CM codes for identifying cardiovascular and stroke risk factors . Med Care . 2005 ; 43 ( 5 ): 480 - 5 .
Quan H , Sundararajan V , Halfon P , Fong A , Burnand B , Luthi JC , et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data . Med Care . 2005 ; 43 ( 11 ): 1130 - 9 .
Zhang XH , Chen H , Siu CW , Yiu KH , Chan WS , Lee KL , et al. Newonset heart failure after permanent right ventricular apical pacing in patients with acquired high-grade atrioventricular block and normal left ventricular function . J Cardiovasc Electrophysiol . 2008 ; 19 ( 2 ): 136 - 41 . https://doi.org/10.1111/j.1540- 8167 . 2007 . 01014 .x.
Dall TM , Blanchard TD , Gallo PD , Semilla AP . The economic impact of Medicare part D on congestive heart failure . Am J Manag Care . 2013 ; 19 ( 6 Suppl) : s97 - 100 .
Braunschweig F , Cowie MR , Auricchio A . What are the costs of heart failure? Europace . 2011 ; 13 ( Suppl 2 ): ii13 - 7 . https://doi.org/ 10.1093/europace/eur081.