Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Effect of α₁-adrenergic antagonist use on clinical outcomes in patients with heart failure
(USC Thesis Other)
Effect of α₁-adrenergic antagonist use on clinical outcomes in patients with heart failure
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Ghaznavi
|
1
EFFECT OF α1-ADRENERGIC ANTAGONIST USE ON CLINICAL OUTCOMES IN
PATIENTS WITH HEART FAILURE
Master of Science
Applied Biostatistics and Epidemiology
University of Southern California
Zunera Ghaznavi
December 2015
Ghaznavi
|
2
TABLE OF CONTENTS
ABSTRACT 3
BACKGROUND 4-9
INTRODUCTION 4
DEVELOPMENT OF HF 4
SUMMARY OF α –BLOCKER USE 5
SUMMARY OF PREVIOUS STUDIES WITH α –BLOCKER USE 5
STUDY HYPOTHESES 8
METHODS 9-13
STUDY DESIGN AND DATA SOURCE 9
POPULATION 9
EXPOSURE 10
OUTCOME 10
DATABASE DESCRIPTION 11
STATISTICAL ANALYSIS 11
RESULTS 13-18
DISCUSSION 18-21
STRENGTHS AND LIMITATIONS 20
TABLES AND FIGURES 22-26
TABLE 1. DEMOGRAPHIC CHARACTERISTICS 22
FIGURE 1. KAPLAN-MEIR CURVES 24
TABLE 2-5. RESULTS 25
APPENDICES 27-36
APPENDIX A. MODEL SELECTION 27
APPENDIX B. UNADJUSTED AND ADJUSTED SURVIVAL CURVES 29
APPENDIX C.SURVIVAL CURVES BY ALPHA-BLOCKER TYPE 32
APPENDIX D. SURVIVAL CURVES BY ALPHA- AND BETA- BLOCKER USE 33
APPENDIX E. PROPORTIONAL HAZARD ASSUMPTION 35
APPENDIX F. SUMMARY OF ALPHA-BLCOKER USE (DOSE) 36
THESIS ACKNOWLEDGEMENTS 37
REFERENCES 38
Ghaznavi
|
3
ABSTRACT
Cardiovascular disease is the leading cause of death in the United States. Evidence
from
previous
clinical
and
epidemiological
studies
suggests
α
1
–adrenergic receptor antagonist (α-
blocker) use in patients with essential hypertension may be involved in increased risk of developing
heart failure. These effects may be particularly important in patients with pre-existing heart failure
(HF) who are already at high risk for HF hospitalizations and cardiovascular mortality. This
thesis
documents
the
conduction
of
a
retrospective
cohort
study
at
the
Veterans Affairs Greater Los
Angeles Healthcare System (VAGLAHS) to examine the association between α-blocker use in
patients with heart failure and clinical outcomes including 2-year HF re-hospitalization, all-cause
mortality, and a composite HF re-hospitalization and mortality outcome; we sought to determine if
different α-blockers differ in their associations with these outcomes and further elucidate any
modification of effect by use of concomitant β-blocker therapy.
We hypothesized a stronger positive association of α-blocker and 2-year HF re-hospitalization
and all-cause mortality amongst users of non-selective α-blocker’s including terazosin, doxazosin,
and alfuzosin due to their equal affinity for all α
1
-receptor subtypes compared to users of selective α-
blocker’s including tamsulosin.
Multivariate Cox proportional hazards regression models were used to examine the cohort of
720 patients discharged with a primary diagnosis of heart failure between 2002 and 2014
Use of any α-blocker was associated with significant increase in risk of HF re-hospitalization
(aHR 1.28, 95%CI: 1.01-1.63, p=0.04) and significant decrease in risk of all-cause mortality (aHR
0.66, 95%CI: 0.49-0.88, p=0.006). It was not significantly associated with risk of composite heart
failure and all-cause mortality (aHR 1.05, 95%CI: 0.86, 1.29, p=0.61). There was no difference in
effect of use of selective α-blocker versus non-selective α-blockers on the association with HF re-
hospitalization or all-cause mortality.
Overall, these findings suggest there is an increased risk of HF hospitalizations but no
increased risk of mortality with use of any α-blocker (selective or non-selective) in patients with pre-
existing HF.
Ghaznavi
|
4
BACKGROUND
Cardiovascular disease is the leading cause of death in the United States. Heart failure (HF)
affects nearly 5.7 million Americans older than 20 years of age;
1
870,000 new cases are diagnosed
annually. HF is a major cause of both morbidity and mortality. Although survival rates after
diagnoses are continually improving,
2
deaths attributable to HF remain high. Five-year survival
following an HF diagnosis is reportedly 50% in the US.
3
4
In 2013, HF was mentioned as a
contributing cause in one out of every nine deaths.
5
HF is a global burden that will continue to grow
as our population ages.
Identification of both risk factors for development of HF and determinants of outcome
following diagnosis remains a pressing priority. Recognized risk factors that may be targeted for
prevention of HF include hypertension, obesity, diabetes, and cigarette smoking.
6
In both the
Original Framingham Heart Study and the Framingham Offspring Study cohorts, hypertension was
the most common risk factor for congestive heart failure (CHF), preceding development of HF in
91% of cases.
7
Treatment regimens to manage HF have evolved over recent decades to address changes in
the understanding of the development of the disease.
8
Previously, diuretic therapies based on a
hemodynamic model focused mainly on controlling fluid volume due to excess salt and water
retention. Though diuretics have been helpful in management of symptoms, they have not been
shown to improve survival rates.
9
10
More recently, HF therapies have evolved to target the
neuroendocrine systems, presumed to effect what is referred to as the “cardiac remodeling process”
11
as described below; unlike diuretics, these therapies reduce mortality rates by up to 40%.
11
In general, heart failure develops as a response to an insult to the heart resulting in
diminished pumping capacity (and subsequent reduction in cardiac output). The initial insult can be
acute (myocardial infarction), chronic (hypertension), or even genetic (genetic cardiomyopathy).
Ghaznavi
|
5
After the initial decline in pumping capacity, the sympathetic nervous system (SNS) responds by
activating the adrenergic nervous system, the renin-angiotensin system, and the cytokine system. The
SNS stimulation accelerates heart rate, increases cardiac contractility and stroke volume, and
constricts peripheral vessels.
12
Although compensatory at first, if sustained over longer periods of
time, this level of activation can lead to damage within the ventricle (changes are often referred to as
“left ventricular remodeling”) and subsequent progression from asymptomatic to symptomatic HF.
Therapies that target this adverse ventricular remodeling, namely angiotensin-converting
enzyme inhibitors (“ACE-inhibitors”), angiotensin receptor blockers (“ARBs”), β-adrenergic
blocking agents (“β-blockers”), and aldosterone receptor antagonists have been shown in multiple
clinical trials to significantly reduce HF related morbidity and total mortality
11
. The success of these
therapies has shifted the paradigm of HF treatment from hemodynamics to left ventricular
remodeling. According to the latter, HF progresses as a result of overexpression of neurohormones
(i.e. norepinephrine, angiotensin II, endothelin, aldosterone, etc.) due to excessive stimulation of the
SNS system (referred to as a “neurohormonal mechanism”). Accordingly, the 2013 American
College of Cardiology Foundation/American Heart Association guidelines for management of HF
specified use of ACE-inhibitors and β-blockers in all patients with a reduced ejection fraction (EF) to
prevent symptomatic HF.
13
As HF is often seen in the elder population, it is not uncommon for HF patients to have a wide
range of comorbidities that may contribute to the progression of their HF and/or affect their response
to treatment. Common comorbidities include respiratory conditions, renal dysfunction, hyperkalemia,
anemia, depression, arthritis, hypertension, and benign prostrate hypertrophy (BPH).
α
1
-adrenergic receptor antagonists (“α-blockers”) are one of the classes of medications used in
the treatment of BPH. Though sometimes used as an “add-on” antihypertensive therapy in the
medication regimen of patients with drug-resistant hypertension, more commonly α-blockers are used
for treating lower urinary tract symptoms (LUTS) secondary to BPH.
14
LUTS include urinary
Ghaznavi
|
6
hesitancy, decreased force of stream, and polyuria resulting from the inability to empty the bladder
completely; α-blockers are the most effective, least costly, and best tolerated of the drugs available
for relieving these symptoms.
15
α
1
-blockers relax smooth muscle and in-turn, improve urinary flow. There are two identified
classes of α–adrenergic receptors (α
1
and α
2
), each with three subtypes. The α
1
-AR blockers function
by mediating constriction (they are vasoconstrictors): the α
1A
subtype regulates smooth muscle tone
in the prostate and bladder neck; the α
1B
subtype regulates blood pressure (BP); the α
1D
subtype is
believed to act through bladder muscle contraction and sacral spinal cord innervation.
16
The non-
selective α
1
-adrenergic receptor antagonists terazosin, doxazosin, and alfuzosin show equal affinity
for all α
1
-receptor subtypes. By contrast, tamsulosin is selective for the α
1A
- and α
1D
-adrenergic
receptors and has less affinity for the α
1B
subtype.
17
18
20
Treatment-related adverse events including
hypotension, dizziness, and syncope are more likely to occur during use of some α-AR blockers
compared with others; these differences can be especially pronounced in elderly patients. The
Vasodilator Heart Failure Trial I (V-HeFT I), designed to evaluate the effect of treatment with
hydralazine plus isosorbide dinitrate on reduction of mortality compared with either prazosin or
placebo, raised initial concerns of increased risk of HF with use of α-blockers. Investigators noted
27% of patients (50 of 183) randomized to an α-blocker, prazosin, discontinued therapy early
compared to 22% of patients (41 of 186 and 60 of 273, respectively) randomized to either
hydralazine-nitrate or placebo. The most common reason cited, in 15 patients (30% of patients that
discontinued prazosin treatment) was increased CHF.
19
Following the V-HeFT trial, the
Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) in patients
with essential hypertension found that patients randomized to the α-blocker doxazosin were at twice
the risk of developing HF (RR 2.04, 1.79-2.32, p<0.001) compared to patients randomized to a
diuretic chlorthalidone. This increased risk led to the early discontinuation of the α-blocker treatment
Ghaznavi
|
7
arm.
20
Excess risk of developing HF in hypertensive patients treated with α -blockers was confirmed
in the Cardiovascular Health Survey
21
, an observational cohort study.
The findings of these large clinical trials and cohort studies led to renewed interest in the
safety and efficacy of doxazosin and other α-blockers as add-on antihypertensive medications in
patients with HF. In a clinical trial of 223 hypertensive patients, Matsui et al. found that doxazosin
appears to significantly lower blood pressure, but could also increase left ventricular diameter, which
increases the risk of CHF.
22
In light of results of the V-HeFT and ALLHAT trials and subsequent observational studies,
additional research was desired to assess potential s effects of α-blockers in patients with pre-existing
HF, who are already at high risk for HF hospitalizations and cardiovascular mortality.
In 2009, Dhaliwal and colleagues investigated this question in a small retrospective cohort
study conducted at a Houston VA Medical Center. They estimated the association between α
1
-AR
antagonist use and clinical outcomes, including HF hospitalization and total mortality, among 388
patients hospitalized with decompensated HF from 2002 to 2004. Comparing patients who used α-
blockers to nonusers, they observed no significant difference in frequency of HF re-hospitalization
(HR 1.20, 95%CI: 0.85-1.7, p=0.31) or all-cause mortality (HR 1.10 95%CI: 0.78-1.56, p=0.57).
However, within the subset of patients not treated with β-blockers, HF re-hospitalization was
significantly more common among α-blocker users (HR 1.94, 95%CI: 1.14, 3.32, p=0.015); in
contrast, among those receiving background β-blockers, risk of HF hospitalization was no different in
α-blocker users compared to non-users (HR 0.92, 95%CI: 0.57, 1.47, p=0.72).
23
As a possible
explanation for this result, the authors proposed that among patients taking α-blockers, β-blockers
may have the cardioprotective effect of opposing an increase in plasma renin activity that would
otherwise be caused by unopposed α-adrenergic antagonism.
The study population of Dhaliwal et al. was not sufficiently large to support analyses
addressing separate effects of α-blocker based on type, selective versus non-selective. Since α-
Ghaznavi
|
8
adrenergic receptors are located throughout the vascular system, it is plausible that blocking these
receptors may have an effect on the cardiovascular system.
24
The V-HeFT and ALLHAT trials
provide evidence of this deleterious effect. However, α-blockers with increased α
1A
-receptor
specificity may potentially result in less adverse cardiovascular events associated with antagonizing
α
1B
-receptors. In particular, due to tamsulosin’s selective properties compared to the other
nonselective α-blockers terazosin, prazosin, and doxazosin, tamsulosin may have fewer effects on the
cardiovascular system if used in a population with pre-existing HF. We hypothesized a stronger
positive association of α-blocker use and 2-year HF re-hospitalization and all-cause mortality
amongst users of non-selective α-blocker’s including terazosin, doxazosin, and alfuzosin due to their
equal affinity for all α
1
-receptor subtypes compared to users of selective α-blocker’s including
tamsulosin.
The results of Dhaliwal et al. are consistent with a cardio-protective effect of β-blockers in
patients who use α-blockers. Some β-blockers are also more selective than others, and non-selective
β-blockers, like carvedilol, have additional a
1
-AR blocking properties. It therefore stands to reason
that selective and non-selective β-blockers may have differing effects when used in conjunction with
α-blockers. Specifically, the potential cardioprotective effect of β-blockers may differ by type of α-
blocker and/or type of β-blocker. Unfortunately, Dhaliwal et al. were not able to address this
question.
The 2013 American College of Cardiology Foundation/American Heart Association
(ACCF/AHA) guidelines specified that data addressing the possibility that HF patients using α-
blockers experience elevated risk of developing HF or worsening HF remain inconclusive.
1
Although
the Houston VA Medical Center study was informative, safety of α -blocker usage in HF patients
remains unclear at present.
To address the ongoing need to clarify safety of α -blocker use in HF patients – accounting in
particular for differing modes of action of selective and non-selective α-blockers and their potential
Ghaznavi
|
9
for interaction with background β-blocker therapy -- we studied a cohort of 720 patients following
hospital admission for HF. We estimated associations of α-blocker use with 2-year heart failure re-
hospitalization, all-cause mortality, and a composite outcome of heart failure re-hospitalization and
all-cause mortality. Analyses distinguished between selective and non-selective α-blockers and
accounted for background use of selective and non-selective β-blockers. We anticipated that these
data would provide a comparatively large contribution to existing data addressing safety of α-
blockers for use in HF patients, potentially expanding understanding of the mechanism by which α –
blockers function.
METHODS
Study Design and Data Source
We conducted a retrospective cohort study at the Veterans Affairs Greater Los Angeles
Healthcare System (VAGLAHS) with the approval of VAGLAHS’s Institutional Review Board.
Patient data were extracted from the VAGLAHS’s Computerized Patient Records System (CPRS) by
five data abstractors.
Population
Patients 18 years of age or older discharged from VAGLAHS with a primary diagnosis of
heart failure (ICD-9 428.x) between January 1, 2002 and April 1, 2014 were identified by the
Veterans Health Information System and Technology Architecture administrative database as
potential participants. The index date was defined as the date of discharge. Information on identified
patients included name, SSN, date of birth, and if the patient had been prescribed an α-blocker
(tamsulosin, terazosin, doxasozin, or alfulozin) as part of their treatment regimen specified in their
outpatient medications recorded in their electronic medical record.
Each patient was assigned a number and a sample of patients were selected from the base
population for participation in this study by random selection via random number generation. 58% of
patients prescribed α-blockers (402 of 692) were sampled, while 24% of the patients not on an α-
Ghaznavi
|
10
blockers (420 of 1730) were sampled. Patients on α-blocker were oversampled to allow for stratified
analysis by type of α-blocker. Approximately half of the resulting study cohort members were
patients on α-blocker.
Individuals were excluded if they had one or more of the following: a competing diagnosis
that was as likely or more likely than HF (e.g. COPD, STEMI, NSTEMI, pneumonia, etc.), no
medication records at discharge (e.g. patients who left against medical advice or died before
discharge), or incomplete data (e.g. no medical record post-indexed discharge or if this was the first
and only visit to the VAGLA).
Exposure
Patients were categorized by use of α–blocker (use and no use, as well as a breakdown by specific
type: tamsulosin, terazosin, doxasozin, alfulozin, or other) at indexed discharge date.
Covariates
Patients were also categorized by use of β–blocker (use and no use, as well as a breakdown by
specific type: carvedilol, metropolol succinate, metropolol tartrate, atenolol, or other) at indexed
discharge date.
Outcomes
Primary and secondary outcomes of interest included HF re-hospitalization, all-cause
mortality, and a composite outcome combining HF readmission and all-cause mortality. Per protocol,
HF re-hospitalization was defined as “patients who had worsening symptoms of heart failure who
were admitted to the hospital after indexed admission up to a maximum of 2 years follow-up.”
Though a maximum follow-up on mortality was not imposed, analytic follow-up time was truncated
to 2 years for all outcomes for comparability of results to the Dhaliwal et al. paper. Patients with no
outcome events were right-censored 2 years from their indexed discharge date.
Outcomes were noted as time to first event; time was calculated as the difference in number
of days between first event and indexed discharge date. Time to first event (HF re-hospitalization or
Ghaznavi
|
11
death) was used for composite outcome. Due to limitations in electronic medical record access, HF
re-hospitalization was only noted as an event if it occurred within the VAGLA system. Date of
mortality was determined by date of “death note” in the electronic medical record.
Database Description
Baseline measurements at index discharge that were extracted into the database included
demographic variables (age, sex, ethnicity), existing comorbidities (including hypertension, diabetes
mellitus, chronic obstructive pulmonary disease, pulmonary embolism, pulmonary infection,
coronary artery disease, cancer, hyperlipidemia, liver disease, history of alcohol abuse, anemia,
chronic kidney disease, acute kidney injury, atrial fibrillation, valvular disease, history of myocardial
infarction, and benign prostrate hypertrophy), medications at discharge, non-compliance with salt
restriction and/or HF medications as described by notes from electronic medical record, vital signs
(heart rate and blood pressure), lab values (serum creatinine, sodium, potassium, blood urea nitrogen
(BUN), B-type natriuretic peptide (BNP), and serum hemoglobin (Hgb)), and previous HF
hospitalization. Heart function (left ventricular ejection fraction) was categorized as <40% vs ≥ 40%
and New York Heart Association (NYHA) functional class was categorized as I/II vs III/IV
determined by explicit notation in the electronic medical record or by the subject’s self-reported
exercise capacity at time of HF hospitalization. HF-related medications at discharge included
ACEI/ARBs, BBs, diuretics, α-blockers, digoxin, hydralazine, nitrates, spironolactone/eplerenone,
amiodarone, and statin.
Statistical Analysis
Patients’ baseline characteristics were summarized as proportions (n(%)) for categorical
variables and means (mean ± standard deviation) for continuous variables. Pearson’s chi-square tests
were used for comparing categorical variables between the α-blocker user group and non α-blocker
user group. Student t-tests were used for continuous variables. Corresponding nonparametric
measures, Fisher’s exact test and Mann-Whitney-U test, were used for non-normally distributed data
Ghaznavi
|
12
(as determined by a small sample size, histogram plot, Shapiro/Wilk test, or Skewness-Kurtosis
tests).
Subjects free of incident outcome event, were right-censored 2 years from their indexed HF
discharge date. Outcomes were analyzed by use of α-blocker at indexed HF discharge. Kaplan-Meir
curves were constructed for cumulative event free rates for the occurrence of HF re-hospitalization,
all-cause death, and the composite endpoint of HF re-hospitalization and all-cause death. The
resulting survival curves were compared by log-rank test.
Relative risk (hazard ratios) with 95% confidence intervals and Wald’s 2-sided pvalues were
calculated by multivariate Cox proportional hazards regression to determine differences in risk of
each outcome; non α-blocker users served as the referent group in each analysis. Potential
confounders were those considered clinically relevant by the Cardiovascular Pharmacy Research
team at VAGLAHS
*
. For the HF re-hospitalization outcome, potential confounders included age,
HFrEF, BUN, serum creatinine, and SBP; for the all-cause mortality outcome, potential confounders
included age, SBP, BUN, hemoglobin, sodium, history of COPD, history of cancer, history of liver
disease, history of BPH, previous HF hospitalization, NYHA class, heart rate, and use of background
β-blockers. For the composite outcome, potential confounders included age, BUN, SBP, HFrEF,
history of cancer, and hemoglobin levels. Potential confounders were added one at a time to the base
models. Confounding was evaluated by percent change in the hazard ratio estimate for α-blocker use;
evidence for confounding was defined as change in the estimate by at least 10%. Variables of clinical
relevance** for a given outcome were also forced into the multivariate model, regardless of their
statistical significance or confounding.
Interactions tested included any of the terms from the main effects model. Likelihood ratio
tests were used to assess statistical significance of interactions between all combinations of main
*
Cardiovascular Pharmacy Research team at the Greater Los Angeles VA includes Cardiologist Alberta Warner, MD and Pharmacist
Cynthia Jackevicius, PharmD with 23 and 12 years of experience, respectively, in clinical research.
** clinical relevance was determined by consensus of the Cardiovascular Pharmacy Research team at VAGLAHS
Ghaznavi
|
13
effects. An interaction term was considered statistically significant and retained in the model if
p≤0.10. An outline of the procedure for inclusion of a variable in the model is provided in Appendix
A.
The Cox proportional hazards assumption was verified by graphing model estimates from the
final multivariate model with observed Kaplan-Meir survival. The two curves not overlaying one
another would be considered a violation of the Cox proportional hazards assumption.
Secondary analyses included comparison of the magnitude of effects of type of α-blocker
(tamsulosin, terazosin, doxasozin, alfulozin, other) and evaluation of modification of effect of α-
blocker by β-blocker use via sub-group analysis of type of background β-blockade therapy. Non α-
and β-blocker users served as the referent group.
All analyses were performed using STATA, version 13.0 (StataCorp, College Station, Texas).
Except for interaction effects, two-sided p<0.05 was considered statistically significant.
RESULTS
Of 2,422 patients who met inclusion criteria, 692 (29%) were prescribed alpha-blockers at HF
discharge. 822 patients were randomly selected to be included in this study including 402 patients on
α-blockers and 420 patients not on α-blockers. The following patients were excluded from analysis:
52 patients who were not actually hospitalized—often this was the result of a documented admission
into the emergency department and subsequent discharge from emergency department with no
hospitalization, 26 patients who had competing discharge diagnoses, 6 patients who left against
medical advice, 5 patients who died during hospitalization, 6 patients who were transferred to other
hospitals, and 7 patients who were considered lost to follow up, as they had no medical record other
than the indexed date of discharge. A total of 720 patients were included in the analysis consisting of
387 patients on α-blockers and 333 patients not on α-blockers.
Ghaznavi
|
14
Patients’ baseline characteristics at indexed heart failure hospitalization are presented in Table
1. The cohort consisted primarily of males, of whom about half were White, most had severe HF
(class III/IV), and about half had previously been admitted for HF-related symptoms. Common
prevalent comorbidities in this cohort included preexisting coronary artery disease, diabetes,
hyperlipidemia, and hypertension.
As shown in Table 1, among the 720 participants, 387 (54%) were taking an α1-adrenergeic
antagonist. Tamsulosin 199 (51%) and terazosin 188 (49%) were the most commonly prescribed α-
blockers with mean ± SD doses 0.5 ± 0.4 mg/day (range 0.4 to 0.8) and 5.1 ± 3.8 mg/day (range 1 to
20), respectively, among patients prescribed these medications. 78% of all patients were prescribed
β–blockers, and 82% were prescribed ACE-I’s or ARBs. The most common β–blocker was
metropolol (61%), and the most common ACE-I’s were lisinopril (39%) and benazepril (35%). Mean
± SD dose of metropolol was 104±79 mg/day, mean ± SD dose of lisinopril was 23 ±16 mg/day, and
mean ± SD dose of benazepril was 27±19 mg/day, among patients taking each of these drugs. No
statistically significant difference was observed in frequency of use of concomitant medications in
patients on α-blockers compared to patients not on α-blocker’s at the indexed discharge date,
including background HF therapy at discharge (β-blockade, use of ACE-I or ARBs, aldosterone
antagonists, diuretics, or statins).
Patients on α-blockers tended to be older (mean age 75 vs 67, p<0.0001), have higher ejection
fractions (61% with EF≥ 40 vs. 40%, p<0.0001), higher BUN (p=0.02), higher BNP (p=0.02), and to
have more comorbidities including greater prevalence of atrial fibrillation, CAD, hypertension, CKD,
anemia, liver disease, BPH, and cancer, compared to those not taking α-blockers. Patients on α-
blockers had higher systolic blood pressures (128 vs. 125 mmHg, p=0.03) but lower diastolic blood
pressures (72 vs. 73 mmHg, p=0.05) and heart rates (73 vs. 76 beats/min, p=0.02) during
hospitalization.
Ghaznavi
|
15
Follow up times were similar for the two exposure groups, with mean follow up of 6.6 ± 0.2
years for patients on α-blockers and 6.9 ± 0.2 years for the patients not on α-blockers (ttest p=0.96)
before censoring. Constrained to the 2-year composite outcome or censoring follow-up time, a total
of 194 deaths (27%) occurred during the 2-year follow up, 103 (27%) amongst patients on α-blockers
at discharge versus 91 (27%) among the others. Cumulative events within 2 years included 315 heart
failure hospitalizations (44%), with 182 (47%) amongst patients on α-blockers versus 132 (40%)
among non α-blockers users. Figure 1 shows unadjusted Kaplan-Meir 2-year survival curves for all
outcomes according to α-blocker use; risk tables for each outcome indicate number of patients at risk.
Table 2 reports estimates of the association between α-blocker use and 2-year heart failure
hospitalization, all-cause mortality, and composite outcome of heart failure hospitalization and all-
cause mortality.
Multivariate Regression Model
Analyses using the Cox proportional-hazard regression revealed no variable to be
significantly associated with the primary outcome of 2-year heart failure re-hospitalization; however,
due to clinical relevance determined by the Cardiovascular Pharmacy Research team, age, HFrEF (EF
< 40%), and BUN were kept in the multivariate model. No statistically significant interaction effects
were identified. Previous hospitalization(s) for HF was considered a potential confounder by
indication (it is the single greatest predictor of future HF hospitalizations and could influence the
decision to treat with α-blockers). However, risk of HF hospitalization due to α-blocker use did not
differ in patients with no previous HF hospitalization(s) (crude HR 1.35, 95%CI: 0.96-1.89, p=0.08)
or in patients with previous HF hospitalization(s) (crude HR 1.12, 95%CI: 0.12-1.51, p=0.46). Since
the stratified crude risks did not differ, as determined by the overlap of the crude 95% confidence
intervals, previous HF hospitalization was not added to the model of HF re-hospitalization or
combined outcome. The final model with main effects of age, HFrEF, and BUN was checked and
confirmed to not violate assumptions of proportionality.
Ghaznavi
|
16
For the primary outcome of 2-year all-cause mortality, age and serum hemoglobin were
considered potential confounders and kept in the multivariate model. There were no statistically
significant interaction effects. Graphing the Cox estimates from the final model with age and
hemoglobin vs. observed Kaplan Meir curves showed no violation of the assumption of
proportionality and Cox proportional hazards modeling was considered appropriate.
For the outcome of 2-year composite combined heart failure hospitalization and all-cause
mortality, no variables other than age were considered potential confounders. Nonetheless, BUN,
HFrEF, and serum hemoglobin levels were kept in the model for clinical relevance and comparability
to the independent heart failure hospitalization and all-cause mortality outcomes. There were no
statistically significant interaction effects and the final model with age, BUN, HFrEF, and
hemoglobin was checked and confirmed to not violate assumptions of proportionality.
Primary and Secondary Outcomes
Use of any α-blocker was statistically significantly associated with an increase in risk of
heart failure re-hospitalization (aHR 1.28, 95%CI: 1.01-1.63, p=0.04), statistically significant
decrease in risk of all-cause mortality (aHR 0.66, 95%CI: 0.49-0.88, p=0.006), and not statistically
significantly associated with risk of composite heart failure and all-cause mortality (aHR 1.05,
95%CI: 0.86, 1.29, p=0.61).
For the outcome of HF re-hospitalization, there was no difference in association with use of
selective α-blocker (tamsulosin) (aHR 1.32, 95%CI:0.99-1.76, p=0.06) versus use of non-selective α-
blockers (terazosin, prazosin, or alfulozin) (aHR 1.26, 95%CI: 0.95-1.68, p=0.11). The association of
increased risk of heart failure was not attenuated by use of background β-blocker treatment (aHR
1.27, 95%CI 0.97-1.65, p=0.08).
All-Cause Mortality Outcome-in Depth
Contrary to our hypothesis, we did not observe an increase in risk of mortality in the α-
blocker group. There was no difference in association with of α-blocker with all-caue mortality by
Ghaznavi
|
17
type of α-blocker; reduced mortality risk was seen in those prescribed both selective (tamsulosin)
(aHR 0.59, 95%CI: 0.41-0.85, p=0.005) and non-selective α-blockers (terazosin, alfulozin, prazosin)
(aHR 0.71, 95%CI: 0.49-1.01, p=0.06).
Results of analyses stratified on use of β -blockers suggested that risk of mortality may be
modified by β –blocker use. Among patients receiving no β-blocker therapy (n=157), those on α-
blockers experienced appreciably less all-cause mortality than those not using α-blockers (aHR 0.38,
95%CI: 0.20-0.71, p=0.003). Among patients on background β-blocker therapy (n=563), all cause
mortality was more common in patients on α-blockers that were also on carvedilol (n=169)
(aHR1.34, 95%CI: 0.70-2.55, p=0.37) but lower for those on metropolol (n=114) (aHR 0.65
95%CI:0.31-1.36, p=0.25). Stratification by other specific β –blocker therapies was not possible due
to smaller group sizes.
Sensitivity analysis addressed the association of α-blocker therapy with all-cause mortality,
by use of models that included several variables that had been excluded from the final analytic model
but were nonetheless considered potentially relevant. The three alternative models considered:
included as covariates: (Model 1) age, HFrEF, and BUN; (Model 2) age, HFrEF, BUN, serum
hemoglobin, baseline systolic blood pressure, and history of cancer; and (Model 3) age, BUN, SBP,
Na (<136 mEq/L), history of cancer, history of chronic obstructive pulmonary disease, and history of
cirrhosis. Estimated associations of α-blocker use with all-cause mortality did not appreciably differ
in any of these models compared to the final model used.
A second set of sensitivity analyses addressed potential bias in mortality outcomes by date of
indexed hospitalization. VAGLAHS implemented a new Heart-Failure Post-Discharge clinic in late
2009 to follow patients closely within 30 days of a discharge with an HF diagnosis. The percent of
patients on α-blocker therapies differed significantly by date of indexed discharge; after 10/01/2009,
about 61% of the heart failure patients were on α-blocker therapy compared to 48% before
10/01/2009 (Pearson’s χ
2
=11.5, df=1, p=0.001). As this could potentially influence mortality
Ghaznavi
|
18
outcomes, discharge before/after 10/01/2009 was added to the all-cause mortality regression model.
The effect of α-blocker therapy on all-cause mortality did not differ whether the indexed
hospitalization occurred before/after 10/01/2009.
DISCUSSION
Overall, among heart failure patients, use of α
1
-adrenergic antagonist treatment was
associated with an increased risk of heart failure re-hospitalization but decreased risk of
mortality.The finding of lower mortality was contrary to our initial hypothesis. Moreover, effect of α-
blocker on all-cause mortality was not attenuated by α-blocker type (selective vs. non-selective).
However, there was evidence of modification of effect of α- blocker and all-cause mortality
association by use of β -blocker therapies.
Overall mortality rates were lower in this cohort compared to the Houston VA reported in
2009; the cohort in this study experienced 194 (27%) deaths within 2 years of indexed HF
hospitalization compared to 184 deaths (47%) in the Houston VA. This could potentially be the result
of the comparatively small size of this cohort, of members having less severe HF (mean ± SD EF 40
± 16 compared to 26 ± 13 in the Houston cohort), or of our providers following ACC/AHA 2013 HF
medication guidelines more closely (overall 78% of patients on β –blocker’s and 80% on
ACEI/ARB’s compared to 63% and 78%, respectively, in the Houston cohort), implementation of the
HF post-discharge clinics that followed patients relatively closely, or some combination of these
differences.
In light of other epidemiological studies including the one conducted by Dhaliwal and
colleagues, the finding of reduced mortality in patients on α-blocker’s was unexpected. However, it
is not un-corroborated: before ALLHAT, doxazosin had been shown to be an effective treatment for
CHF.
25
Also of note, 90% of patients in the ALLHAT trial were already on antihypertensive
medications before the trial began. These therapies were stopped and study medications started with
Ghaznavi
|
19
no wash out period in between. With no information on specific prior antihypertensive therapies, it is
reasonable to presume they were on either diuretics and/or ACE-I’s that may have treated underlying
CHF. A potential explanation of the increased CHF risk in α-blocker users in ALLHAT is that when
these therapies were stopped and replaced by doxasozin, underlying HF may have then become
symptomatic. This risk was not seen in the clorthalidone arm because of influence of the continued
diuretic treatment. Thus, it is possible the use of α-blocker did not cause the development of HF, but
rather, just did not prevent it.
26
27
This hypothesis warrants further examination.
It is possible that the difference in mortality following use of selective versus non-selective
β–blockade therapy is the result of carvedilol’s additional α
1
-andernergic receptor blocking
properties; having both a mild and a potent α-blocker may result in excessive α-blockade leading to
worse outcomes. This association could potentially be examined by taking into account dosing of
both the β–blocker therapy (carvedilol) and α-blocker therapy. Possibly there is a threshold at which
α
1
-andernergic antagonism becomes harmful to HF patients.
It should also be noted that although a different result was achieved compared to Dhaliwal et
al. for the association between α-blocker use and 2-year all-cause mortality, these findings are not
contrary to those of Dhaliwal et al. In particular, due to the smaller sample size of the Houston VA
cohort, though the point estimate for the association with mortality is in the direction of increased risk
of mortality (HR 1.10, 95%CI: 0.78-1.56, p=0.57), the confidence intervals are wide enough such
that is possible the direction of the association could be that of decreased risk. With a larger study
population and smaller confidence intervals, this retrospective study may indicate the true direction
of the association between use of α-blocker therapy and mortality. However, this remains to be seen
and due to the somewhat unexpected findings of this study, further investigation of the safety and
efficacy of α-blocker use in the HF population is warranted. Reproducing these findings in different
settings, for example including using in-patients as well as out-patients or including more HF patients
that have less severe HF (class I/II), will be relevant.
Ghaznavi
|
20
Strengths and Limitations
This study has several limitations including use of data from electronic medical records that
may be incomplete or contain inaccuracies. Mortality outcomes could also be missed if the patient
chose non-VA providers for care post indexed HF hospitalization at VAGLAHS. Any
misclassification that may have occurred as result of this practice may potentially be differential; a
larger percentage of patients were prescribed α-blocker therapies in a more recent calendar year
period and would be more likely to have their records not up to date. Though it is not possible to
determine the extent to which this could have influenced our results, we cannot rule out the
possibility that resulting bias may be strong enough to lead to the unanticipated findings with respect
to mortality. Additionally, mortality outcomes were not verified by cross-reference to the National
Death Index. Though this may in part explain the lower mortality observed in this study compared to
that in the Houston VA cohort, Dhaliwal et al. do not note verifying via NDI either.
It’s also possible among the outcomes collected, HF re-hospitalization events were missed if
the admitting clinician did not accurately diagnose HF; however, we would expect this
misclassification to be non-differential, since it is unlikely that the patient’s exposure status (α-
blocker use versus no α-blocker use) would influence their diagnosis. In general, this form of
misclassification would lead to bias towards the null; thus, it is unlikely the findings we see could be
a result of missing patients with HF diagnoses.
It is also possible there are confounders of the α-blocker and clinical outcomes associations
that were unmeasured by this study; for example, smoking data were not collected. Though it is
possible that unknown and uncollected confounders could influence the association between α-
blocker use and HF re-hospitalization or all-cause mortality, the direction of influence is difficult to
determine and it is unlikely that the results we see are driven by uncontrolled confounding alone.
Despite these limitations, this study adds to the existing knowledge regarding the safety of α-
blocker use in the HF population. The possibility of differing α-blocker types differing in their
Ghaznavi
|
21
associations with the HF re-hospitalization and mortality outcomes had been previously unaddressed;
the findings of this study suggest there is no difference in risk of any of the specified 2 year outcome
events by type of α-blocker used, selective versus nonselective.
This study’s strengths include having a well-defined based population, large enough cohort to
address stratified analysis, and access to detailed exposure and modifier variables. Additionally, due
to the nature of the integrated VA healthcare system, the patient population is more likely to be
followed up closely within the same health care system lending to more complete medical records
compared to different healthcare systems where patients are seen by multiple providers in multiple
systems. As a result, the data variables collected including exposure and modifier variables have high
value and are generally complete; it is unlikely that there was enough misclassification of the
exposure variable of α-blocker use or of the modifier variable of β -blocker use to have spuriously
led to the observed associations.
CONCLUSIONS
Overall, our findings suggest additional risk of HF hospitalizations associated with use of any
α-blocker in patients with pre-existing HF. However, there was no indication that elevated risk of
mortality was associated with use of any α-blocker. Selective and non-selective α-blockers did not
differ in their associations with HF re-hospitalization or mortality. No cardioprotective effect with
respect to either HF re-hospitalization or all-cause mortality was apparent; in particular, patients on
carvedilol and any α-blocker had the greatest risk of mortality compared to any other group, perhaps
due to the dual α1-blockade.
To better understand apparent discrepancies in findings of studies addressing safety of α-
blockers following HF, future research should take into account indication, duration, and dosage of
α1-adrnergic blockade and focus on the possibility of a threshold at which α
1
-andernergic antagonism
becomes harmful to HF patients where an increase in risk of mortality is seen rather than a decrease.
Ghaznavi
|
22
TABLES AND FIGURES
Table 1. Baseline Cohort Characteristics
Table&1.&Demographic&characteristics
No
(n=333)
Yes
(n=387)
Age&(years) 71&±&12 67.4&±&12 74.7&±&11 <0.0001
Male& 705&(98%) 323&(97%) 382&(99%) 0.109
Ethnicity/Race 0.37
&&&&&Asian&or&Pacific&Islander 23&(3%) 11&(3%) 12&(3%)
&&&&&Black&or&African&American 244&(34%) 102&(31%) 142&(37%)
&&&&&Caucasian/White 360&(50%) 169&(51%) 191&(49%)
&&&&&Hispanic&or&Latino 37&(5%) 18&(5%) 19&(5%)
&&&&&Other 56&(8%) 33&(10%) 23&(6%)
SBP,&mmHg 127&±&21.5 124.5&±&22 128&±&21 0.03
DBP,&mmHg 72&±&13 73&±&14 71.5&±&12 0.05
HR&(beats/min) 74&±&14 75.5&±&15 73&±&14 0.02
Salt&Noncompliance,&y 99&(14%) 44&(13%) 55&(14%) 0.15
Med&Noncompliance,&y 155&(22%) 81&(24%0 74&(19%) 0.09
Ejection&Fraction&(%) 40&±&16.6 36&±&16.5 43&±&16 <0.0001
HFrEF,&y 352&(49%) 201&(60%) 151&(39%) <0.0001
NYHA&classification 0.17
&&&&&(I)&No&limitation&of&physical&activity 13&(2%) 3&(1%) 10&(2.5%)
&&&&&(II)&Slight&limitation&of&physical&activity 65&(9%) 30&(9%) 35&(9%)
&&&&&(III)&Marked&limitation&of&physical&activity 542&(75%) 260&(78%) 282&(73%)
&&&&&(IV)&Unable&to&carry&out&physical&activity 100&(14%) 40&(12%) 60&(16%)
Etiology&of&HF 0.014
&&&&&Ischemic 219&(30%) 85&(26%) 134&(35%)
&&&&&Non]Ischemic 54&(8%) 31&(9%) 23&(6%)
&&&&&Unknown/Undocumented 447&(62%) 217&(65%) 230&(59%)
Previous&HF&hospitalization,&y 338&(47%) 145&(44%) 193&(50%) 0.09
Serum&Lab&Values
&&&&&Creatinine&(mg/dL) 1.6&±&0.04 1.5&±&0.05 1.7&±&0.06 0.006
&&&&&Sodium&(mEq/L) 138.3&±&3.7 138.2&±&3.6 138.3&±&3.7 0.81
&&&&&Potassium&(mEq/L) 4.3&±&0.02 4.3&±&0.03 4.2&±&0.03 0.49
&&&&&Blood&Urea&Nitrogen&(mg/dL) 27.7&±&0.7 26.3&±&0.9 29&±&0.9 0.024
£
&&&&&B]type&Natriuretic&Peptide&(pg/mL) 1023&±&1186 1012&±&992 1032&±&1327 0.023
£
&&&&&Hemoglobin&(g/dL) 12&±&2.1 12.5&±&2.1 11.6&±2 <0.0001
Pre]existing&Comorbidities,&y
&&&&&Atrial&Fibrillation 237&(33%) 97&(29%) 140&(36%) 0.045
&&&&&Myocardial&Infarction 157&(22%) 68&(20%) 89&(23%) 0.4
&&&&&Valvular&Disease 95&(13%) 30&(9%) 65&(17%) 0.002
&&&&&Coronary&Artery&Disease 379&(53%) 162&(49%) 217&(56%) 0.047
&&&&&Diabetes&Mellitus 350&(49%) 161&(48%) 189&(49%) 0.9
&&&&&Hyperlipidemia 432&(60%) 188&(56%) 244&(63%) 0.07
&&&&&Hypertension 601&(84%) 265&(80%) 336&(87%) 0.01
Characteristics> All>Subjects
(n=720)
Use>of>Alpha>Antagonist pvalue*
Ghaznavi
|
23
!!!!!Chronic!obstructive!
!!!!!!!pulmonary!disease
199!(28%0 87!(26%) 112!(29%) 0.4
!!!!!Pulmonary!Embolism 10!(1%) 7!(2%) 3!(1%)
0.13
¥
!!!!!Pulmonary!Infection 49!(7%) 22!(7%) 27!(7%) 0.84
!!!!!Acute!Kidney!Failure 72!(10%) 26!(8%) 46!(12%) 0.07
!!!!!Chronic!Kidney!Disease 229!(32%) 80!(24%) 149!(29%) <0.0001
!!!!!Liver!Disease 63!(9%) 37!(11%) 26!(7%) 0.04
!!!!!Alcohol!Abuse 105!(15%) 60!(18%) 45!(12%) 0.02
!!!!!Sleep!Apnea 91!(13%) 40!(12%) 51!(13%) 0.64
!!!!!Anemia 164!(23%) 54!(16%) 110!(28%) <0.0001
!!!!!Benign!Prostrate!Hypertrophy 253!(35%) 34!(10%) 219!(57%) <0.0001
!!!!!Cancer 123!(17%) 40!(12%) 83!(21%) 0.001
Concomitant!Medications!
!!!!!!!(at!discharge),!y
'''''β)blocker 563!(78%) 264!(79%) 299!(77%) 0.152
¥
!!!!!!!Carvedilol 169!(30%) 90!(34%) 79!(26%)
!!!!!!!Metroprolol!succinate! 114!(20%) 57!(22%) 57!(19%)
!!!!!!!Metroprolol!tartrate 233!(41%) 99!(38) 134!(45%)
!!!!!!!Others! 47!(8%) 18!(7%) 29!(10%)
'''''ACE'inhibitor' 460!(64%) 218!(65%) 243!(63%) 0.34
!!!!!!!Lisinopril 176!(39%) 80!(38%) 96!(42%)
!!!!!!!Benazepril 155!(35%) 76!(36%0 79!(35%)
!!!!!!!Fosinopril 101!(23%) 53!(25%) 48!(21%)
!!!!!!!Others 5!(1%) 4!(2%) 1!(0.5%)
'''''ARB 118!(16%) 54!(16%) 64!(17) 0.55
!!!!!!!Losartan 49!(46%) 18!(38%) 31!(52%)
!!!!!!!Valsartan 35!(33%) 18!(38%0 17!(29%)
!!!!!!!Irbesartan 22!(21%) 11!(23%) 11!(19%)
!!!!!Digoxin 177!(25%) 85!(25%) 92!(24%) 0.59
!!!!!Loop!Diuretic 611!(85%) 286!(86%) 325!(84%) 0.48
!!!!!Thiazide 47!(7%) 21!(6%) 26!(7%) 0.41
!!!!!Aldosterone!Antagonist 125!(17%) 64!(19%) 61!(16%)! 0.22
!!!!!Hydralazine 89!(12%) 33!(10%)! 56!(14%) 0.06
!!!!!Nitrates/Nitrites 242!(34%) 106!(32%) 136!(35%) 0.16
!!!!!Any!Statin 467!(65%) 212!(64%) 255!(66%) 0.53
!!!!!Amiodarone 55!(8%) 28!(8%) 27!(7%) 0.47
!!!!!Anticoagulant 164!(23%) 87!(26%)! 77!(20%) 0.05
Continuous!variables!are!presented!as!mean!!±!sd!while!categorical!variables!are!presented!as!n(%)
*!pvalues!evaluated!by!t]tests!in!continuous!variables!and!Pearson's!chi]2!tests!in!categorical!variables!
£!non]parameteric!determination!(using!Mann]Whitney]U!tests)
Ghaznavi
|
24
Figure 1. 2-year Adjusted Kaplan-Meir curves for primary and secondary outcomes by alpha-blocker use at
discharge. Tables list numbers of patients at risk. Outcomes include (A) Heart failure re-hospitalization (B) All-cause
mortality and (C) Composite heart failure re-hospitalization and all-cause mortality. The dotted red line represents
patients on an α-blocker while the solid blue line represents patients not on an α-blocker.
A
Figure 1A. Kaplan-Meir curve for Cumulative HF-hospitalization free survival adjusted for age, reduced ejection
fraction (HFrEF), and serum blood urea nitrogen (BUN).
B
Figure 1B. Kaplan-Meir curve for cumulative survival adjusted for age and serum hemoglobin.
aHR 1.28 (95%CI 1.01-1.63; p=0.04)
aHR 0.66 (95%CI: 0.49-0.88; p<0.01)
Ghaznavi
|
25
C
Figure 1C. Kaplan-Meir curve for composite outcome of HF-rehospitalization and all-cause mortality adjusted
for age, serum blood urea nitrogen (BUN), and reduced ejection fraction (HFrEF).
Table 2. 2-year Heart Failure re-Hospitalization outcome
Association of α1-adrenergic antagonist use on clinical outcome of heart failure hospitalization within 2 years from
indexed discharge date. Outcome is presented for overall cohort and stratified by type of α-blocker and background β-
blocker therapy.
HR for heart failure hospitalization adjusted for age, HFrEF, and BUN. *pvalues determined by Wald’s test with Ho:
HR=1 for HR estimated by multivariate Cox proportional hazard regression
HR=Hazard Ratio, CI=Confidence Interval, HFrEF= Ejection fraction <40%, BUN=blood urea nitrogen, Hgb=serum
hemoglobin
HF#re&hospitalization
Variable no#of#pt#
%#with#
outcome no#of#pt
%#with#
outcome crude#HR#(95%#CI) adjusted#HR*#(95%CI) pvalue*
Overall#Cohort 387 47% 333 40% 1.26+(1.00-1.58) 1.28+(1.01,+1.63) 0.04
By#alpha&blocker
Selective+(tamsulosin) 199 48% 333 40% 1.31+(1.00,+1.70) 1.32+(0.99,+1.76) 0.06
Non-Selective+
(Terazosin/Prazosin/Alfulosin/Ot
her) 188 46% 333 40% 1.21+(0.92,+1.59) 1.26+(0.95,+1.68) 0.11
By#Background#Beta&blocker
No+B-Blocker+use 88 43% 68 32% 1.32+(0.78-2.23) 1.29+(0.73-2.29) 0.86
Any+B-Blocker+Use 299 48% 264 42% 1.25+(0.98,+1.61) 1.27+(0.97,+1.65) 0.08
Selective+B-Blocker+use+
(carvedilol) 79 43% 90 38% 1.22+(0.76,+1.97) 1.69+(1.00,+2.87) 0.05
Non-Selective+B-Blocker+use+
(metropolol+succinate,+
metropolol+tartrate,+atenolol) 173 43% 217 50% 1.57+(0.92,+2.66) 1.32+(0.74,+2.35) 0.35
Use#of#alpha&blocker
n=387
No#use#of#
alpha&blocker#n=333 Hazard#Ratio
aHR 1.05 (95%CI 0.86-1.29; p=0.61)
Ghaznavi
|
26
Table 3. Mortality outcome
Association of α1-adrenergic antagonist use on all-cause mortality within 2 years from indexed discharge date. Outcome
is presented for overall cohort and stratified by type of α-blocker and background β-blocker therapy.
HR for all cause mortality adjusted for age and hemoglobin. *pvalues determined by Wald’s test with Ho: HR=1 for HR
estimated by multivariate Cox proportional hazard regression
HR=Hazard Ratio, CI=Confidence Interval, Hgb=serum hemoglobin
Table 4. 2-year Composite Outcome
Association of α1-adrenergic antagonist use on clinical composite outcome of heart failure hospitalization and all cause
mortality within 2 years from indexed discharge date. Outcome is presented for overall cohort and stratified by type of α-
blocker and background β-blocker therapy.
HR for composite outcome adjusted for age, BUN, HFrEF, and hemoglobin. *pvalues determined by Wald’s test with
Ho: HR=1 for HR estimated by multivariate Cox proportional hazard regression
HR=Hazard Ratio, CI=Confidence Interval, Hgb=serum hemoglobin
All#Cause)Mortality
Variable no)of)pt)
%)with)
outcome no)of)pt
%)with)
outcome crude)HR)(95%)CI) adjusted)HR*)(95%CI) pvalue*
Overall)Cohort 387 27% 333 27% 0.97)(0.73,)1.29) 0.66)(0.49,)0.88) 0.006
By)alpha#blocker
Selective)(tamsulosin) 199 26% 333 27% 0.96)(0.68,)1.35) 0.59)(0.41,)0.85) 0.005
Non?Selective)
(Terazosin/Prazosin/Alfulosin/Ot
her) 188 27% 333 27% 0.98)(0.70,)1.39) 0.71)(0.49,)1.01) 0.06
By)Background)Beta#blocker
No)B?Blocker)use 88 26% 69 36% 0.69)(0.39,)1.21) 0.38)(0.20,)0.71) 0.003
Any)B?Blocker)Use 299 27% 264 25% 1.07)(0.78,)1.50) 0.77)(0.55,)1.08) 0.003
Selective)B?Blocker)use)
(carvedilol) 79 33% 90 21% 1.67)(0.93,)3.03) 1.34)(0.70,)2.55) 0.37
Non?Selective)B?Blocker)use)
(metropolol)succinate,)
metropolol)tartrate,)atenolol) 173 27% 217 30% 1.57)(0.92,)2.66) 0.65)(0.31,)1.36) 0.25
Use)of)alpha#blocker
n=387
No)use)of)
alpha#blocker)n=333 Hazard)Ratio
Composite)HF)hospitalization)and)All3Cause)Mortality
Variable no)of)pt)
%)with)
outcome no)of)pt
%)with)
outcome crude)HR)(95%)CI) adjusted)HR*)(95%CI) pvalue*
Overall)Cohort 387 66% 333 60% 1.2*(1.00,*1.44) 1.05*(0.86,*1.29) 0.61
By)alpha3blocker
Selective*(tamsulosin) 199 63% 333 60% 1.3*(1.04,*1.61) 1.09*(0.86,*1.38) 0.49
Non?Selective*
(Terazosin/Prazosin/Alfulosin/Ot
her) 188 63% 333 60% 1.1*(0.88,*1.38) 1*(0.78,*1.28) 1
By)Background)Beta3blocker
No*B?Blocker*use 88 63% 69 58% 1.11*(0.74,*1.66) 0.9*(0.56,*1.45) 0.67
Any*B?Blocker*Use 299 67% 264 60% 1.23*(1.00,*1.51) 1.06*(0.85,*1.33) 0.58
Selective*B?Blocker*use*
(carvedilol) 79 72% 90 58% 1.46*(1.00,*2.13) 1.55*(1.01,*2.36) 0.04
Non?Selective*B?Blocker*use*
(metropolol*succinate,*
metropolol*tartrate,*atenolol) 173 65% 217 61% 1.15*(0.83,*1.58) 1*(0.71,*1.41) 0.99
Use)of)alpha3blocker
n=387
No)use)of)
alpha3blocker)n=333 Hazard)Ratio
Ghaznavi
|
27
APENDICES
Appendix A. Model Selection
Table 5. HF re-hospitalization multivariate model selection
Main effects model for HF re-hospitalization outcome included the following variables: age, HFrEF, and BUN.
Likelihood ratio tests for interactions compares deviance from a model with the given interaction term and main effects to
the base model of age, HFrEF, and BUN.
Table 6. All cause mortality multivariate model selection
Main effects model for all-cause mortality outcome included the following variables: age and hemoglobin. Likelihood
ratio tests for interactions compares deviance from a model with the given interaction term and main effects to the base
model of age and hemoglobin.
HF#re&hospitalization
variable#adjusted#for#
univariately# HR
%change#in#
HR#estimate Kept#in#model?
Base,&no&adjustments 1.259
age 1.273 1.09 yes,&clinically&relevant
hfref 1.291 2.56 yes,&clinically&relevant
bun 1.232 A2.15 yes,&clinically&relevant
scr 1.245 A1.12 no
sbp 1.265 0.48 no
Interactions LR#Chi&2 pvalue Kept#in#model?
age*hfref 0.720 0.39 no
age*bun 2.190 0.14 no
bun*hfref 0.05 0.82 no
Mortality
variable,adjusted,for,
univariately, HR
%change,in,
HR,estimate Kept,in,model?
base,&no&adjustments 0.970
age 0.726 525.20 yes,&clinically&relevant&&confounder
sbp 1.006 3.65 no
bun 0.914 55.80 no
hgb 0.832 514.22 yes,&counfounder
na2 0.967 50.33 no
nx_COPD 0.964 50.63 no
hx_cancer 0.894 57.90 no
hx_liver&disease 0.982 1.18 no
prev_hf 0.945 52.64 no
nyha&class&(3/4&vs&1/2) 0.975 0.46 no
hr 0.957 51.34 no
hx_bph 0.899 57.36 no
BB_any 0.964 50.66 no
Interactions LR,Chi@2 pvalue Kept,in,model?
age*hgb 0.160 0.69 no
Ghaznavi
|
28
Table 7. Composite HF re-hospitalization and all-cause mortality multivariate model selection
Main effects model for composite outcome included the following variables: age, BUN, HFrEF, and hemoglobin.
Likelihood ratio tests for interactions compares deviance from a model with the given interaction term and main effects to
the base model of age, BUN, HFrEF, and hemoglobin.
Composite)HF)re-hospitalization)+)All)cause)mortality
variable)adjusted)for)
univariately) HR
%change)in)
HR)estimate Kept)in)model?
base 1.20
age 1.07 +10.60 yes,/confounder/+/clinical/relevance
bun 1.16 +3.20 yes,/clinically/relevant
sbp_bl 1.21 1.20 no
hfref 1.22 1.98 yes,/clinically/relevant
hx_cancer 1.17 +2.34 no
hgb 1.11 +7.59 yes,/clinically/relevant
Interactions LR)Chi-2 pvalue Kept)in)model?
age*bun 1.160 0.28 no
age*hgb 0.540 0.46 no
age*hfref 0.000 0.96 no
bun*hgb 2.190 0.14 no
bun*hfref 1.200 0.25 no
hgb*hfref 1.68 0.2 no
Ghaznavi
|
29
Appendix B
Unadjusted and Adjusted Survival Curves for all Outcomes
Figure 2A. Unadjusted Kaplan Meir Curve for association of α-blocker use and HF re-Hospitalization.
Figure 2B. Kaplan Meir Curve for association of α-blocker use and HF re-Hospitalization adjusted for age, HFrEF, and
BUN.
Ghaznavi
|
30
Figure 3A. Unadjusted Kaplan Meir Curve for association of α-blocker use and all-cause mortality.
Figure 3B. Kaplan Meir Curve for association of α-blocker use and all-cause mortality adjusted for age and hemoglobin.
Ghaznavi
|
31
Figure 4A. Unadjusted Kaplan Meir Curve for association of α-blocker use and composite outcome.
Figure 4B. Kaplan Meir Curve for association of α-blocker use and composite outcome adjusted for age, BUN,
hemoglobin, and HFrEF.
Ghaznavi
|
32
Appendix C
Survival curves by alpha-blocker type
A
Figure 5A. Kaplan Meir Curve for association of α-blocker use and HF re-Hospitalization adjusted by α-blocker
type after adjusting for age, HFrEF, and BUN.
B
Figure 5B. Kaplan Meir Curve for association of α-blocker use and all-cause mortality by α-blocker after
adjusting for age and serum hemoglobin.
C
Figure 5B. Kaplan Meir Curve for association of α-blocker use and composite HF re-hospitalization and all-
cause mortality by α-blocker after adjusting for age, HFrEF, BUN, and serum hemoglobin.
Ghaznavi
|
33
Appendix D
Survival curves by alpha- and beta-blocker usage
A
Figure 6A. Kaplan Meir Curve for association of α-blocker use and HF re-Hospitalization adjusted by α-and
β-blocker type after adjusting for age, HFrEF, and BUN.
B
Figure 6B. Kaplan Meir Curve for association of α-blocker use and all-cause mortality by α- and β-blocker
type after adjusting for age and serum hemoglobin.
Ghaznavi
|
34
C
Figure 6C. Kaplan Meir Curve for association of α-blocker use and composite HF re-hospitalization and
all-cause mortality by α- and β-blocker type after adjusting for age, HFrEF, BUN, and serum hemoglobin.
Ghaznavi
|
35
Appendix E.
Checking proportionality assumption
Figure 7A. HF re-hospitalization. Graph of HF re-hospitalization free survival estimates from final Cox-proportional
hazards model adjusted for age, HFrEF, and BUN compared to observed Kaplan-Meir survival shows no evidence of
violation of the Cox-proportionality assumption.
Figure 7B. All-cause mortality. Graph of overall survival estimates from final Cox-proportional hazards model adjusted
for age and hemoglobin compared to observed Kaplan-Meir survival shows no evidence of violation of the Cox-
proportionality assumption.
Figure 7C. Composite Outcome. Graph of composite HF re-hospitalization and all-cause mortality survival estimates
from final Cox-proportional hazards model adjusted for age, HFrEF, BUN, and hemoglobin compared to observed
Kaplan-Meir survival shows no evidence of violation of the Cox-proportionality assumption.
Ghaznavi
|
36
Appendix F.
Summary of alpha-blocker use (dose)
. tabulate alpha_blocker_code, summarize(alpha_blocker_dose)
Alpha | Summary of Total daily dose of
Blocker | alpha-blocker(mg)
Name | Mean Std. Dev. Freq.
------------+------------------------------------
Doxazosin | 4.8 2.2803509 5
Prazosin | 5.6666667 4.8733972 9
Tamsulosi | .49547739 .35608654 199
Terazosin | 5.1289157 3.7822562 166
Alfuzosin | 10 0 8
------------+------------------------------------
Total | 2.8552972 3.6209725 387
Ghaznavi
|
37
THESIS ACKNOWLEDGEMENTS
Foremost I would like to express my most sincere gratitude to Dr. Victoria Cortessis for her support,
patience, motivation, enthusiasm, and immense knowledge. Her insightful comments and questions
have helped me in not only writing this thesis, but also in providing great insight into navigating
research and epidemiological studies. I could not have asked for a more caring and helpful advisor.
I would also like to thank my advisors for this thesis, Dr. Wendy Mack and Dr. Howard Hodis. Dr.
Mack provided invaluable feedback and evaluation.
Besides my advisors, I would also like to thank Dr. Alberta Warner, attending cardiologist and
principal investigator of this study, Dr. Cynthia Jackevicius, pharmacist, and Dr. Lingyun Lu,
pharmacist, for providing the resources to be able to conduct this study. Dr. Warner has provided a
great deal of support and inspiration throughout all of my time with her.
I would also like to sincerely thank Dr. Jim Hagar and Dr. Golnaz Vahdani for their time, feedback,
and stimulating discussions. I also thank Jeesun Cho, Tina Lee, Natasha Gonzalez, and Huong-giang
Vu for helping with the electronic medical record abstractions making the data available in a timely
fashion.
Last but certainly not least, I would like to thank my parents: Mohammad Tariq Ghaznavi and
Ghazala Ghaznavi. They have provided support for my whole life and without them, none of this
work would have been possible.
Ghaznavi
|
38
REFERENCES
1
Heart Disease and Stroke Statistics--2015 Update: A Report From the American Heart Association. Circulation, 2014.
2
Levy D, Kenchaiah S, Larson MG, et al. Long-term trends in the incidence of and survival with heart failure . N Engl J Med .
2002;347:1397–1402 .
3
National Center for Health Statistics. Mortality Multiple Cause Micro-data Files, 2011.
4
Centers for Disease Control and Prevention, National Center for Health Statistics. Compressed Mortality File 1999–2010 . Series 20
No . 2P . CDC WONDER Online Database. Released January 2013 .
5
Go AS, Mozaffarian D, Roger VL, et al. Heart disease and stroke statistics—2013 update: a report from the American Heart
Association. Circulation. 2013;127:e6–e245.
6
Bibbins-Domingo K, Pletcher MJ, Lin F, et al. Racial differences in incident heart failure among young adults. N Engl J Med .
2009;360:1179–1190 .
7
Levy D, Larson MG, Vasan RS, et al. The progression from hypertension to congestive heart failure. JAMA. 1996; 275: 1557-1562
8
Mann, D. L. Mechanisms and Models in Heart Failure: The Biomechanical Model and Beyond. Circulation. 2005; 111: 2837–49.
9
Swedberg K, Cleland J, Dargie H . Guidelines for the diagnosis and treatment of chronic heart failure: executive summary. Eur Heart
J 2005; 26: 1115–1140.
10
Hunt SA, Abraham WT, Chin MH . ACC/AHA 2005 Guideline Update for the Diagnosis and Management of Chronic Heart Failure
in the Adult. Circulation 2005; 112: 1825–1852
11
Lechat, Philippe. The Evolution of Heart Failure Management over Recent Decades: From CONSENSUS to CIBIS.” European
Heart Journal Supplements. 2006; 8: C5–C12.
12
Filippos T, Karayannis G, Giamouzis G, et al. The Sympathetic Nervous System in Heart Failure. JACC. 2009; 54(19):1747–62.
13
Yancy C, 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 Guideline. Circulation 2013; 128: e 240-e327.
14
McVary KT, Roehrborn CG, Avins AL, et al. Update on AUA (American Urological Association) guideline on the management of
benign prostatic hyperplasia. J Urol 2011; 185(5):1793-803.
15
Lepor, Herbert. “α Blockers for the Treatment of Benign Prostatic Hyperplasia.” Reviews in Urology 9, no. 4 (2007): 181–90.
16
Fine, Shari R., and Phillip Ginsberg. “α -adrenergic Receptor Antagonists in Older Patients with Benign Prostatic Hyperplasia: Issues
and Potential Complications.” JAOA: Journal of the American Osteopathic Association 108, no. 7 (2008): 333–37.
17
Beduschi MC, Beduschi R, Oesterling JE. α -blockade therapy for benign prostatic hyperplasia: from a nonselective to a more
selective α
1A
-adrenergic antagonist [review]. Urol. 1998;51:861-872.
18
Schwinn DA, Michelotti GA. α
1
-Adrenergic receptors in the lower urinary tract and vascular bed: potential role for the α
1D
-subtype
in filling symptoms and effects of aging on vascular expression [review]. BJU Int. 2000;85(suppl 2):6-11.
19
Cohn JN, et al. "Effect of vasodilator therapy on mortality in chronic congestive heart failure".The New England Journal of
Medicine. 1986. 314(24):1547-52.
20
ALLHAT Collaborative Research Group. Major cardiovascular events in hypertensive patients randomized to doxazosin vs
chlorthalidone: the antihypertensive and lipid-lowering treatment to prevent heart attack trial (ALLHAT). JAMA 2000; 283(15):1967-
75.
21
Bryson CL, Smith NL, Kuller LH, et al. Risk of Congestive Heart Failure in an Elderly Population Treated with Peripheral αlpha-1
Antagonists. J Am Geriatr Soc. 2004 Oct;52(10):1648-54.
22
Matsui Y, Equchi k, Shibasaki S, et al. Effect of doxazosin on the left ventricular structure and function in morning hypertensive
patients: the Japan Morning Surge 1 study. J Hypertens 2008; 26(7):1463-71.
23
Dhaliwal A, Habib G, Deswal A, et al. Impact of α 1-adrenergic antagonist use of the benign prostatic hypertrophy on outcomes in
patients with heart failure. Am J Cardiol 2009; 104(2): 270-275.
24
Kaplan
SA.
Factors
in
predicting
failure
with
medical
therapy
for
BPH.
Rev
Urol.
2005;
7(Supppl
7):
S34-‐S30.
25
DiBianco R, Parker JO, Chakko S, et al. Doxazosin for the treatment of chronic congestive heart failure: results of a randomized
double-blind and placebo-controlled study. Am Heart J. 1991;121:372–380.
26
Sica DA, Deedwania PC. Doxazosin and congestive heart failure. Pharmacotherpay in CHF. 2002: May/June 178-184.
27
Beevers DG and Lip GYH. Do alpha blockers cause heart failure and stroke? Observations from ALLHAT. J Human HTN. 2000; 14:
287-289
Abstract (if available)
Abstract
Cardiovascular disease is the leading cause of death in the United States. Evidence from previous clinical and epidemiological studies suggests α₁–adrenergic receptor antagonist (α-blocker) use in patients with essential hypertension may be involved in increased risk of developing heart failure. These effects may be particularly important in patients with pre-existing heart failure (HF) who are already at high risk for HF hospitalizations and cardiovascular mortality. This thesis documents the conduction of a retrospective cohort study at the Veterans Affairs Greater Los Angeles Healthcare System (VAGLAHS) to examine the association between α-blocker use in patients with heart failure and clinical outcomes including 2-year HF re-hospitalization, all-cause mortality, and a composite HF re-hospitalization and mortality outcome
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Obesity paradox in acute heart failure decompensation
PDF
Associations between isoflavone soy protein (ISP) supplementation and breast cancer in postmenopausal women in the Women’s Isoflavone Soy Health (WISH) clinical trial
PDF
Contemporary outcomes for adult congenital heart surgery in an adult tertiary care hospital
PDF
Association of subclinical atherosclerosis with plasma B-vitamin, cysteine, homocysteine, and cysteinyl glycine in a cardiovascular disease-free population
PDF
Association of oophorectomy with progression of carotid atherosclerosis
PDF
The effects of hormone therapy on carotid artery intima-media thickness and serum lipids by ApoE4 genotype
PDF
An analysis of disease-free survival and overall survival in inflammatory breast cancer
PDF
Effects of post-menopausal hormone therapy on arterial stiffness in the ELITE trial
PDF
Cross-sectional association of blood pressure, antihypertensive medications, MRI volumetric measures and cognitive function scores in an aging population
PDF
Visual acuity outcomes after cataract extraction in Chinese Americans: the Chinese American Eye Study (CHES)
PDF
The effect of renal function on toxicity of E7389 (eribulin) among patients with bladder cancer
PDF
A randomized controlled trial of hormone therapy on postmenopausal women’s quality of life
PDF
Assessment of the mortality burden associated with ambient air pollution in rural and urban areas of India
PDF
Survival of children and adolescents with low-risk non-rhabdomyosarcoma soft tissue sarcomas (NRSTS) treated with surgery only: an analysis of 234 patients from the Children’s Oncology Group stud...
PDF
Analysis of SNP differential expression and allele-specific expression in gestational trophoblastic disease using RNA-seq data
PDF
Hormone therapy timing hypothesis and atherosclerosis
PDF
An assessment of necrosis grading in childhood osteosarcoma: the effect of initial treatment on prognostic significance
PDF
Origins of the gender disparity in bladder cancer risk: a SEER analysis
PDF
Development and validation of survey instrument designed for cervical cancer screening in Malawi, and other low resource settings
PDF
Red and processed meat consumption and colorectal cancer risk: meta-analysis of case-control studies
Asset Metadata
Creator
Ghaznavi, Zunera
(author)
Core Title
Effect of α₁-adrenergic antagonist use on clinical outcomes in patients with heart failure
School
Keck School of Medicine
Degree
Master of Science
Degree Program
Applied Biostatistics and Epidemiology
Publication Date
12/02/2016
Defense Date
09/17/2015
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
BPH,heart failure,HF,OAI-PMH Harvest,α₁,α₁-adrenergic antagonist,α-blocker
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Cortessis, Victoria (
committee chair
), Hodis, Howard N. (
committee member
), Mack, Wendy (
committee member
)
Creator Email
zghaznav@usc.edu,zunerag.89@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-201570
Unique identifier
UC11277405
Identifier
etd-GhaznaviZu-4061.pdf (filename),usctheses-c40-201570 (legacy record id)
Legacy Identifier
etd-GhaznaviZu-4061-1.pdf
Dmrecord
201570
Document Type
Thesis
Format
application/pdf (imt)
Rights
Ghaznavi, Zunera
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
BPH
heart failure
HF
α₁
α₁-adrenergic antagonist
α-blocker