Pain Medicine 2018; 19: 485–490
doi: 10.1093/pm/pnx103
Brief Research Report
Opioid Use in Patients with Congestive Heart
Failure
Nancy L. Dawson, MD,* Victoria Roth,†
David O. Hodge, MS,‡ Emily R. Vargas, MPH, MS,§
and M. Caroline Burton, MD*
*Division of Hospital Internal Medicine, ‡Biostatistics
Unit, and §Department of Health Sciences Research,
Mayo Clinic, Jacksonville, Florida; †Mayo School of
Health Sciences, Mayo Clinic College of Medicine,
Jacksonville, Florida, USA
Correspondence to: Nancy L. Dawson, MD, Division of
Hospital Internal Medicine, Mayo Clinic, 4500 San Pablo
Rd, Jacksonville, FL 32224, USA. Tel. 904-956-0081;
Fax 904-953-2848; E-mail: dawson.nancy11@mayo.edu.
Disclosure and conflicts of interest: None.
Abstract
Objective. To understand the relationship between
opioid use in patients with congestive heart failure
and outcomes, we compared length of stay (LOS),
30-day readmission rates, and 30- and 90-day mor-
tality in patients discharged with a primary diagno-
sis of congestive heart failure (CHF) who were
taking opioids.
Design. Retrospective study design.
Setting. Patients were seen at a 320-bed academic
hospital.
Subjects. All patients not awaiting transplant who
were discharged with a primary diagnosis of heart fail-
ure from January 1, 2011, through December 31, 2014.
Methods. Records were reviewed for demographic
data, comorbidities, and opioid status at admission
or discharge. The association of opioid use (at ad-
mission and discharge) with LOS, 30-day readmis-
sion, and 30- and 90-day mortality was examined.
Results. Six hundred eighty-two patients with a
principle diagnosis of heart failure were admitted
C 2017 American Academy of Pain Medicine. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
V 485
Downloaded from https://academic.oup.com/painmedicine/article/19/3/485/3778222 by guest on 05 March 2021
during the study period, with 168 (24.6%) taking opi-
oids at admission. Opioid use at admission was not
significantly associated with 30-day readmission
(odds ratio [OR] 5 1.24, 95% confidence interval
[CI] 5 0.80–1.93), 30-day mortality (hazard ratio
[HR] 5 0.91, 95% CI 5 0.47–1.78), 90-day mortality
(HR 5 0.95, 95% CI 5 0.58–1.54), or LOS (parameter
estimate 5 20.21, 95% CI 5 20.91 to 0.48). One hun-
dred ninety-three patients (28.3%) were prescribed
opioids at discharge. No significant differences
were observed between those who were and were
not taking opioids at discharge for 30-day readmis-
sion (OR 5 1.10, 95% CI 5 0.72–1.69) or for 30- or
90-day mortality (HR 5 0.51, 95% CI 5 0.24–1.06, and
HR 5 0.67, 95% CI 5 0.41–1.10, respectively). LOS
was slightly shorter for patients not using opioids
at discharge than for those who were (mean 5 3.8
vs 4.6 days, respectively).
Conclusions. Opioid use at admission or discharge
in patients with CHF did not appear to affect
outcomes.
Key Words. Length of Stay; Mortality; Opioid;
Readmission
Introduction
Approximately 5.7 million adults in the United States
have a diagnosis of congestive heart failure (CHF), and
1 million are admitted to the hospital each year [1].
Improving outcomes such as length of stay (LOS),
short-term mortality rates, and 30-day readmission rates
in these patients is important to hospitals and patients
alike, and such outcomes are used to measure quality
of care. In addition, reimbursement by insurers is linked
to these outcome measures. In recent years, policy-
makers have increased emphasis on outcomes of pa-
tients with CHF; these outcomes are considered metrics
of care, and as such, they can be used to hold hospi-
tals accountable for the care that they provide [2]. To
improve care, it is important to understand the factors
that influence these outcomes.
Dawson et al.
Improving outcomes in patients with CHF is a high pri-
ority. National data show that more than 20% of pa-
tients with CHF are readmitted within 30 days, and
approximately 12% of these patients die during their
hospitalization [3]. Identifying risk factors associated with
poor outcomes is a crucial first step in developing pro-
grams that could reduce these rates [4]. Both CHF and
opioid use are known to be associated with higher rates
of readmission, but their combined risk has not been
studied [5]. Because opioids are often used to treat the
long-term pain associated with terminal illnesses such
as CHF, an understanding of how opioids affect patients
with CHF and influence outcomes is needed to reduce
costs and provide better care [6].
The physiologic effect of opioids on patients with CHF is
unclear. Opioids such as morphine can reduce some
symptoms of late-stage CHF, including dyspnea, partic-
ularly during exercise [7,8]. Morphine likewise appears
to relieve ischemia in patients with cardiovascular risk
factors [9,10]. These findings, along with evidence that
the pharmacologic mechanism of morphine produces
certain beneficial peptides, has led some researchers to
conclude that morphine may have some cardioprotec-
tive effects in patients with CHF [11,12]. However, the
research is contradictory; one study showed that pa-
tients with CHF who were prescribed morphine (along
with nitroglycerin and/or furosemide) had reduced mor-
tality rates [13], yet another study showed that opioids,
including morphine, have few effects on the cardiovas-
cular system at rest [9]. Moreover, the combination of
CHF and opioid use appears to increase the likelihood
of cardiovascular events for patients receiving long-term
opioid therapy [14]. Opioids generally have been associ-
ated with increased rates of readmission [15]. Similarly,
opioid users undergoing coronary artery bypass surgery
were more likely to be readmitted, although the mecha-
nism was unclear [16].
To better understand the relationship between opioid
use and outcomes of patients with CHF, we compared
LOS, 30-day readmission rates, and 30- and 90-day
mortality rates in patients discharged with a primary di-
agnosis of CHF who were receiving opioid medications
at admission, discharge, or both.
Methods
The Mayo Clinic Institutional Review Board approved
this study.
Setting and Design
This retrospective study was conducted at a 320-bed
academic hospital. All patients discharged with a pri-
mary diagnosis of heart failure (International
Classification of Diseases, Ninth Revision [ICD-9] codes
428.0–428.9) from January 1, 2011, through December
31, 2014, were included in the study. Patients who
were admitted to await heart transplant were excluded
because these patients are hospitalized for the purpose
486
of inotropic support until transplant is available and gen-
erally receive a transplant before discharge. The event
of heart transplantation can change outcomes and con-
found the results. All data were collected from the elec-
tronic health record; demographic data included age,
weight, sex, and race. Clinical data collected included
cardiac ejection fraction, a history of diabetes mellitus,
hypertension, coronary artery disease, cerebrovascular
disease, dementia, peripheral arterial disease, chronic
kidney disease, cancer, and lung and liver disease,
Medical comorbidity was measured by using the
Downloaded from https://academic.oup.com/painmedicine/article/19/3/485/3778222 by guest on 05 March 2021
Charlson Comorbidity Index (CCI) [17]. The CCI was
scored electronically using diagnoses in the institution’s
medical index database. Death record databases were
accessed for the 30- and 90-day mortality data. To de-
termine the number of patients readmitted within 30
days, data were collected and reviewed from the Mayo
Clinic electronic health records and the Medicare data-
base of readmissions to all other institutions obtained
from the Center for Medicare and Medicaid Services
QNet portal for the Hospital Readmissions Reduction
Program, accessed April 1, 2016. The readmission rate
for our hospital for CHF patients ranges between 20%
and 24% annually, considered no different from the na-
tional average (21.9%).
At our institution, all patients have a medication list gen-
erated at admission and confirmed by an admission
nurse with the patient or caregiver. At discharge, the
medication list is updated (medication reconciliation)
and becomes a permanent part of the medical record
on the date of discharge. These admission and dis-
charge medication lists were reviewed for all opioid
medications, both scheduled and as needed, including
morphine, oxymorphone, hydromorphone, codeine, oxy-
codone, hydrocodone, methadone, fentanyl, or
meperidine.
Statistical Analysis
Patient demographic characteristics and clinical data
were described after stratifying by opioid use at admis-
sion and discharge by using the sample median and
standard deviation for continuous variables and fre-
quency and percent for categorical variables. The com-
parison of individual variables between groups with and
without opioid use was completed with either a
Wilcoxon rank-sum test for continuous variables or a v2
test for independence for categorical variables. The as-
sociation of opioid use at admission and discharge with
30-day readmission, 30- and 90-day mortality, and LOS
was examined using logistic regression, Cox propor-
tional hazards regression, and linear regression models,
respectively. We chose to evaluate LOS as a variable
with opioids at discharge because of the assumption
that patients who are prescribed opioids at discharge
likely are also given opioids during hospitalization, and it
may therefore affect the LOS. Significant univariate as-
sociations were then tested. For logistic regression
models, odds ratios (ORs) and 95% confidence intervals
(CIs) were estimated. For Cox proportional hazards
 Opioids in CHF

Table 1 Patient characteristics, stratified by opioid use

Opioid use at admission Opioid use at discharge

Characteristic Yes (N ¼ 168)* No (N ¼ 514) P Yes (N ¼ 193)* No (N ¼ 489) P

Male sex, No. (%) 96 (57.1) 306 (59.5) 0.58 112 (58.0) 290 (59.3) 0.76
Race, No. (%) 0.22 0.26
Asian 2 (1.2) 6 (1.2) 2 (1.0) 6 (1.2)
Black 20 (11.9) 57 (11.1) 23 (11.9) 54 (11.0)
Not disclosed 4 (2.4) 3 (0.6) 4 (2.1) 3 (0.6)
Other 0 (0.0) 8 (1.6) 0 (0.0) 8 (1.6)

Downloaded from https://academic.oup.com/painmedicine/article/19/3/485/3778222 by guest on 05 March 2021

Unknown 0 (0.0) 1 (0.2) 0 (0.0) 1 (0.2)
White 142 (84.5) 439 (85.4) 164 (85.0) 417 (85.3)
Age, median (range), y 75 (29-96) 77 (19-101) 0.20 73 (28-101) 78 (19-100) 0.005
Body mass index, median 29 (16.2-51.8) 27.4 (11.2-70.5) 0.02 28.7 (11.2-65.5) 27.3 (0.0-70.5) 0.06
(range), kg/m2 (N ¼ 166) (N ¼ 493) (N ¼ 188) (N ¼ 472)
Ejection fraction, No. (%) 0.01 0.29
Preserved 91 (54.8) 214 (43.3) 94 (49.5) 211 (44.9)
Missing data 2 (1.2) 20 (3.9) 3 (1.6) 19 (3.9)
CHF, No. (%)
Diastolic 77 (45.8) 179 (34.8) 0.01 74 (38.3) 182 (37.2) 0.79
Systolic 57 (33.9) 204 (39.7) 0.18 78 (40.4) 183 (37.4) 0.47
Both systolic and diastolic 34 (20.2) 123 (23.9) 0.32 40 (20.7) 117 (23.9) 0.37
CHF NOS, No. (%) 0 (0.0) 8 (1.6) 0.10 1 (0.5) 7 (1.4) 0.32

CHF ¼ congestive heart failure; NOS ¼ not otherwise specified.

*Includes patients who also were receiving opioids at admission; 34 patients receiving opioids at admission did not have opioids
prescribed at discharge, and 59 patients were prescribed opioids at discharge but were not receiving them at admission.

regression models, hazard ratios (HRs) and 95% CIs
were estimated. For linear regression models, regression
coefficients (parameter estimates) and 95% CIs were
estimated. All statistical analyses were performed using
SAS software (version 9.4, SAS Institute, Inc, Cary, NC,
USA). Patient factors that have been seen to have an
effect on the outcomes of interest (LOS, readmission,
and mortality) and were shown to have significance in
the univariate analysis were included in the multivariate
analysis. These factors included age, body mass index,
history of solid tumor, ejection fraction (due to CHF be-
cause of the population studied), and CCI.
Results
Six hundred eighty-two patients with a principle diagno-
sis of heart failure were admitted during the study pe-
riod. Patients were grouped by their use of opioids at
admission and at discharge (Table 1). The groups were
demographically similar in that they were both primarily
white (overall, N ¼ 581 [85.2%]) and male (overall,
N ¼ 402 [58.9%]). The mean body mass index in each
group fell within the overweight category. The age range
of the entire study cohort was 19 to 101 years.
Opioids at Admission
not taking opioids at admission, these patients had a
mean LOS of 3.9 (SD ¼ 3.1) vs 4.1 (SD ¼ 4.2) days, 30-
day readmissions of 34 (20.7%) vs 88 (17.1%) patients,
30-day mortality of 12 (7.1%) vs 40 (7.8%) patients, and
90-day mortality of 25 (14.9%) vs 80 (15.6%) patients.
We did not observe any significant univariate associa-
tions between opioid use at admission and LOS, 30-day
readmissions, or 30- or 90-day mortality (Table 2).
Opioids at Discharge
One hundred ninety-three patients (28.3%) were pre-
scribed opioids at discharge. Compared with patients
who were not prescribed opioids at discharge, these
patients had a mean (SD) LOS of 4.6 (4.9) days vs 3.8
(3.6) days, a 30-day readmission rate of 37 (19.2%) vs
85 (17.4%), a 30-day mortality rate of 9 (4.7%) vs 4.3
(8.8%) patients, and a 90-day mortality rate of 23
(11.9%) vs 82 (16.7%) patients. In univariate analysis,
opioid use on discharge was not associated with
30-day readmissions (OR ¼ 1.10, 95% CI ¼ 0.72–1.69,
P ¼ 0.65), 30-day mortality (OR ¼ 0.51, 95% CI ¼ 0.24–
1.06, P ¼ 0.07), or 90-day mortality (OR ¼ 0.67, 95%
CI ¼ 0.41–1.10, P ¼ 0.12). Opioid use at discharge was
associated with LOS (parameter estimate ¼ 0.78 days,
95% CI ¼ 0.12–1.45 days, P ¼ 0.02) (Table 2).

Of the 682 patients, 168 (24.6%) reported taking opioids Multivariate logistic model analysis was then performed
at the time they were admitted. Compared with patients using the dependent variable LOS and the variables

487
Dawson et al.

Table 2 Opioid use at admission and discharge and outcomes

Outcome Value P

Opioid use at admission (N ¼ 168)

Length of stay, parameter estimate (95% CI)* 0.21 (0.91 to 0.48) 0.55
30-day readmission, odds ratio (95% CI)† 1.24 (0.80–1.93) 0.34
30-day mortality, hazard ratio (95% CI)‡ 0.91 (0.47–1.78) 0.79
90-day mortality, hazard ratio (95% CI)‡ 0.95 (0.58–1.54) 0.83
Opioid use at discharge (N ¼ 193)
Length of stay, parameter estimate (95% CI)* 0.78 (0.12–1.45) 0.02

Downloaded from https://academic.oup.com/painmedicine/article/19/3/485/3778222 by guest on 05 March 2021

30-day readmission, odds ratio (95% CI)† 1.10 (0.72–1.69) 0.65
30-day mortality, hazard ratio (95% CI)‡ 0.51 (0.24–1.06) 0.07
90-day mortality, hazard ratio (95% CI)‡ 0.67 (0.41–1.10) 0.12

CI ¼ confidence interval.
*Linear regression.
†
Logistic regression.
‡
Cox proportional hazards regression.

Table 3 Multivariate linear regression model for length of stay

Variable Parameter estimate (95% CI) P

Opioids at discharge 0.54 (0.11 to 1.2) 0.11

Age, y 0.11 (0.13 to 0.08) <0.001
Body mass index 0 (0.04 to 0.05) 0.92
Solid tumor 0.67 (1.42 to 0.09) 0.08
Charlson Comorbidity Index 0.16 (0.05–0.26) 0.003
Ejection fraction group
35–50% vs  50% 0.05 (0.83 to 0.72) 0.89
<35% vs  50% 0.64 (0.07 to 1.35) 0.08

CI ¼ confidence interval.

opioid use at discharge, age, body mass index, solid tu-
mor presence, CCI score (using comorbidities included in
this index), and ejection fraction. The final model showed
that LOS was associated with younger age (P < 0.001)
and higher CCI score (P ¼ 0.003) but was not associated
with opioid use on discharge (P ¼ 0.11) (Table 3).
Multivariate modeling was not done for the 30-day read-
missions or 30- or 90-day mortality because in the univari-
ate analysis opioids at admission and at discharge were
not significant for these variables (Table 2).
Discussion
The objective of this study was to examine the relation-
ship between opioid medication use in patients with
CHF and hospital outcomes. We found that nearly 25%
and 30% of patients with CHF were receiving opioids at
admission and at discharge, respectively. In our cohort,
opioid use at admission had no effect on LOS, 30-day
readmissions, or 30- or 90-day mortality rates. Similarly,
opioid use at discharge was not associated with 30-day
readmissions, LOS, or 30- or 90-day mortality rates, but
LOS was associated with younger age and higher CCI
score. A patient with a higher CCI score would be ex-
pected to have a longer LOS because of multiple co-
morbid conditions, yet the reason for the association
between younger age and increased LOS is less obvi-
ous. One possible explanation could be that younger
patients were better candidates for cardiac intervention
(e.g., a coronary catheterization procedure), which could
result in increased LOS when compared with medica-
tion adjustment in an older patient. We also observed
that patients with opioids prescribed at discharge were
significantly younger; the reason for this difference is
unclear. It is possible that providers are more comfort-
able prescribing opioids at discharge to younger pa-
tients, and many patients (N ¼ 59) who were not
receiving opioids at admission were prescribed these
medications at discharge.
Chronic opioid use is associated with health care utiliza-
tion and specifically with hospital readmissions in the
general population [16]. We did not observe an associa-
tion between opioid use and 30-day readmission in

488

patients with CHF. This finding suggests that the pri-
mary driver of readmissions for our patients was unre-
lated to acute or chronic pain control or to the
problematic adverse effects of opioid medication such
as delirium, constipation, or addiction. Although others
have reported that the combination of CHF and opioid
use increases the likelihood of cardiovascular events
[14], we did not find an association with worse out-
comes. The effect on LOS merits further investigation
with respect to in-hospital adverse events and indica-
tions for use.
A prior study showed that patients prescribed opioids
for noncancer pain have a higher risk of all-cause mor-
tality (excluding overdose), specifically cardiovascular
mortality [18]. We did not find this to be the case in pa-
tients with CHF; other cardiovascular risks may increase
with opioid use. Given the frequency of opioid therapy,
studying outcomes in larger populations of opioid users
may be beneficial to help delineate which comorbidities
confer higher risk for adverse outcomes. We also found
that a relatively high percentage of patients with CHF
were taking opioids (24.6%), similar to that seen in gen-
eral hospitalized patients [16].
This study has several limitations. It was a retrospective
study that depended on the accuracy of the health re-
cord. This single-site study had an ethnically homoge-
neous patient population; the findings should be
validated in a multicenter population. The indications for
opioid medications were not obtained from the health
record, and uncovering the indications would be chal-
lenging in a retrospective study. We also did not differ-
entiate between scheduled or as-needed dosing; there
could have been differences, based on the level of opi-
oid usage. Additionally, the study sample may have
been too small to show a statistically significant differ-
ence between patients who were and were not receiv-
ing opioid medication. We were not able to
retrospectively assess functional status; patients receiv-
ing opioids may have worse functional status, which
could affect outcomes and confound the results.
However, we did not identify worse outcomes in pa-
tients receiving opioids; even if their functional status
was diminished, it did not appear to affect our out-
comes of interest. We also did not assess other pre-
scription medications, for example, those prescribed
specifically for heart failure. These medications could af-
fect outcomes, but because many medications now are
used for heart failure, with varying efficacy, the effect of
other medications would be difficult to analyze with re-
spect to opioid use. However, we did use ejection frac-
tion in our multivariate analysis for LOS.
Although we did not observe significantly different out-
comes with opioids, their use in this patient population
was substantial. Nearly 25% of patients were receiving
opioids at admission, and nearly 30% of patients were
prescribed them at discharge. Patients with CHF are
commonly older and have comorbidities that increase
their risk for adverse drug events and poor outcomes.
Opioids in CHF
Future studies should examine in-hospital adverse
events and the indications for opioid medication use in
patients with CHF.
Authors’ Contributions
Study concept and design, acquisition of data, or analy-
sis and interpretation of data: NLD, VR, DOH, ERV,
MCB. Drafting/revising the manuscript for important in-
tellectual content: NLD, VR, MCB. Approval of the final
version to be published: NLD.
Downloaded from https://academic.oup.com/painmedicine/article/19/3/485/3778222 by guest on 05 March 2021
References
1 Hall MJ, Levant S, DeFrances CJ. Hospitalization for
congestive heart failure: United States, 2000–2010.
NCHS Data Brief 2012;(108):1–8.
2 Centers for Medicare and Medicaid Services (CMS),
HHS. Medicare program; hospital inpatient prospec-
tive payment systems for acute care hospitals and
the long-term care hospital prospective payment
system and fiscal year 2015 rates; quality reporting
requirements for specific providers; reasonable com-
pensation equivalents for physician services in ex-
cluded hospitals and certain teaching hospitals;
provider administrative appeals and judicial review;
enforcement provisions for organ transplant centers;
and electronic health record (EHR) incentive pro-
gram: final rule. Fed Regist 2014;79
(163):49853–50536.
3 Centers for Medicare and Medicaid Services.
Hospital compare. Available at: www.medicare.gov/
hospitalcompare (accessed October 26, 2016).
4 Hain DJ, Tappen R, Diaz S, Ouslander JG.
Characteristics of older adults rehospitalized within 7
and 30 days of discharge: Implications for nursing
practice. J Gerontol Nurs 2012;38(8):32–44.
5 Allaudeen N, Vidyarthi A, Maselli J, Auerbach A.
Redefining readmission risk factors for general medi-
cine patients. J Hosp Med 2011;6(2):54–60.
6 Johnson MJ, McDonagh TA, Harkness A, McKay
SE, Dargie HJ. Morphine for the relief of breathless-
ness in patients with chronic heart failure: A pilot
study. Eur J Heart Fail 2002;4(6):753–6.
7 Williams SG, Wright DJ, Marshall P, et al. Safety
and potential benefits of low dose diamorphine dur-
ing exercise in patients with chronic heart failure.
Heart 2003;89(9):1085–6.
8 Darrouj J, Karma L, Arora R. Cardiovascular mani-
festations of sedatives and analgesics in the critical
care unit. Am J Ther 2009;16(4):339–53.
489
Dawson et al.
9 Beattie WS, Buckley DN, Forrest JB. Epidural mor-
phine reduces the risk of postoperative myocardial
ischaemia in patients with cardiac risk factors. Can
J Anaesth 1993;40(6):532–41.
10 Wang TL, Hung CR. Enhanced endothelin-1 degra-
dation by intravenous morphine in patients with con-
gestive heart failure: Role of neutral endopeptidase
24.11. Heart 2003;89(2):211–2.
11 Ogutman C, Ozben T, Sadan G, Trakya A, Isbir M.
Morphine increases plasma immunoreactive atrial
natriuretic peptide levels in humans. Ann Clin
Biochem 1990;27(pt 1):21–4.
12 Wuerz RC, Meador SA. Effects of prehospital medi-
cations on mortality and length of stay in congestive
heart failure. Ann Emerg Med 1992;21(6):669–74.
13 LoCasale R, Kern DM, Chevalier P, et al.
Description of cardiovascular event rates in patients
initiating chronic opioid therapy for noncancer pain
in observational cohort studies in the US, UK, and
Germany. Adv Ther 2014;31(7):708–23.
490
14 Dispennette R, Elliott D, Nguyen L, Richmond R.
Drug burden index score and anticholinergic risk
scale as predictors of readmission to the hospital.
Consult Pharm 2014;29(3):158–68.
15 Safaii N, Kazemi B. Effect of opium use on short-
term outcome in patients undergoing coronary artery
bypass surgery. Gen Thorac Cardiovasc Surg 2010;
58(2):62–7.
16 Mosher HJ, Jiang L, Vaughan Sarrazin MS, et al.
Downloaded from https://academic.oup.com/painmedicine/article/19/3/485/3778222 by guest on 05 March 2021
Prevalence and characteristics of hospitalized adults
on chronic opioid therapy. J Hosp Med 2014;9
(2):82–7.
17 Charlson ME, Pompei P, Ales KL, MacKenzie CR. A
new method of classifying prognostic comorbidity in
longitudinal studies: Development and validation. J
Chronic Dis 1987;40(5):373–83.
18 Ray WA, Chung CP, Murray KT, Hall K, Stein CM.
Prescription of long-acting opioids and mortality in
patients with chronic noncancer pain. JAMA 2016;
315(22):2415–23.