Journal of Cardiothoracic and Vascular Anesthesia 35 (2021) 17041711

Contents lists available at ScienceDirect

Journal of Cardiothoracic and Vascular Anesthesia

journal homepage: www.jcvaonline.com

Original Article
Introduction of an Analgesia Prescription Guideline
Can Reduce Unused Opioids After Cardiac Surgery:
A Before and After Cohort Study
1
Joseph J. Pena, MD*, , Chung Jen Chen, MD*, Hugo Clifford, MD*
, Zenghui Xue, PhD, MSy, Shuang Wang, PhDy,
Michael Argenziano, MD, FACSz, Ruth Landau, MD*,
Marie-Louise Meng, MDx
*
Columbia University Irving Medical Center, Department of Anesthesiology, New York, NY
y
Mailman School of Public Health, Department of Biostatistics, Columbia University, New York, NY
z
Columbia University Irving Medical Center, Department of Cardiothoracic Surgery, New York, NY
x
Duke University Medical Center, Department of Anesthesiology, Durham, NC

Objective(s): The authors aimed to assess whether the introduction of a tailored Analgesia Prescription Guideline would decrease the amount of
unused opioid following discharge from cardiac surgery.
Design: Prospective, observational, before and after study.
Setting: Quaternary care university hospital.
Participants: A total of 191 participants who underwent cardiac surgery requiring midline sternotomy and cardiopulmonary bypass. There were
99 participants in the before cohort (prior to introduction of the Analgesia Prescription Guideline), and 92 participants in the after cohort (after
introduction of the Analgesia Prescription Guideline).
Interventions: Using prospectively collected observational data on participant opioid consumption in the before cohort, a tailored Analgesia Pre-
scription Guideline was developed. This guideline then was introduced to all opioid-prescribing providers in the cardiothoracic surgery depart-
ment. Prospective data then were collected in the after cohort of participants. Opioid prescription practices and opioid consumption between the
two groups then were compared.
Measurements and Main Results: Opioid prescriptions were given to 62/99 participants (63%) in the before cohort, and 48/92 (52%) in the after
cohort (rate difference 0.1, CI 95% -0.26, 0.046). In the before cohort, the mean (§ standard deviation) number of opioid tablets prescribed,
used, and leftover was 26 (§10), 11 (§10), and 15 (§12), respectively. In the after cohort, the mean number of opioid tablets prescribed, used,
and leftover was 18 (mean difference -8, CI 95% -12, -5), 10 (mean difference -1, CI 95% -5, 3), and 8 (mean difference -7, CI 95% -11, -3),
respectively. There were 110/191 (58%) participants using no opioids following discharge, and 10/191 (5%) still using opioids two weeks after
discharge. There were no differences between groups with regard to demographics, opioid-related side effects, pain scores, satisfaction, opioid
storage. and disposal practices.
Conclusions: The development and implementation of a tailored Analgesia Prescription Guideline decreased the amount of opioids prescribed
after cardiac surgery and resulted in lower numbers of unused leftover opioid tablets in the community. Patient comfort and satisfaction scores
remained high.
Ó 2020 Elsevier Inc. All rights reserved.

Key Words: opioid prescription; postoperative analgesia; cardiac surgery

THE OPIOID epidemic in the United States remains a pub-

1
Address correspondence to Joseph Pena, 622 W. 168th St., PH5-133, New lic health emergency.1 The number of deaths related to pre-
York, NY 10032. scription opioid use have increased dramatically during the
E-mail address: joseph.pena.j@gmail.com (J.J. Pena).

https://doi.org/10.1053/j.jvca.2020.12.021
1053-0770/Ó 2020 Elsevier Inc. All rights reserved.
 J.J. Pena et al. / Journal of Cardiothoracic and Vascular Anesthesia 35 (2021) 17041711 1705

past two decades.2 Prescription patterns are partly responsible room schedule and electronic medical record. Enrollment criteria
for this crisis, as the amount of opioids prescribed per capita included patients who had a coronary artery bypass graft
has tripled since 19993 and more than 40% of opioid overdose (CABG) procedure, single left-heart valve repair or replacement,
deaths involve a prescription opioid.4 aortic root replacement, or any combination of these procedures.
Although current Centers for Disease Control and Preven- Exclusion criteria were limited English proficiency (defined as
tion guidelines emphasize the judicious use of opioids in the requiring interpreter services), lack of capacity to provide
treatment of chronic noncancer pain,5 dosing guidelines in the informed consent, postoperative mechanical circulatory device
acute postsurgical period are scarce.6 Multimodal stepwise support, including extracorporeal membrane oxygenator cannula-
opioid-sparing analgesia prescriptions allow optimal pain tion, ventricular assist device implantation, or aortic balloon
management and reduce the risk of opioid-related side effects. pump placement, or a prolonged postoperative length of hospital-
In addition, reducing the duration of opioid exposure should ization defined as greater than 21 days.
reduce the risk for persistent opioid use.7 Upward of 6.5% of Consent was obtained in accordance with the approved insti-
opioid-naive patients develop new persistent opioid use in the tutional review board protocol. Patients were approached for
postoperative period. participation postoperatively after the intensive care unit
Opioids frequently are overprescribed following procedures (ICU) or the cardiac surgery inpatient service, usually on the
across a number of surgical specialties, including general sur- second-to-fourth days after surgery. Study staff first
gery, obstetrics, gynecology, thoracic, urology, and dental sur- approached the patient’s nursing staff to assess the patient’s
gery.8-10 A majority of the opioid tablets prescribed following ability to provide consent; patients who had postoperative
these procedures are unused, undisposed, and are stored in delirium, excessive sedation, or who otherwise lacked deci-
unsecure locations in patients’ homes.8-11 Undisposed opioids sion-making capacity as determined by the nursing staff were
create a reservoir in the community for further misuse and rescreened on a subsequent day. Written consent was obtained
abuse. In one recent national survey, approximately 20% of prior to hospital discharge.
respondents reported sharing prescribed opioid medication Participants enrolled in the study were contacted by phone
with another person.11 ten-to-14 days after hospital discharge. Those who did not
There currently is limited information regarding the patterns answer the first phone call attempt were called two-to-four
of opioid prescription and postdischarge opioid use after car- times per week for three consecutive weeks and were deemed
diac surgery. This information is essential to guide providers lost to follow-up if no contact was made during that period.
in prescribing opioid medications in a way that effectively pro- Participants who were still using opioids at the time of the
vides analgesia while minimizing excess, unused medication. phone survey were called again every week until they stopped
Furthermore, although patients’ in-hospital opioid use prior to using opioids, or up to eight weeks of continued opioid use, at
discharge has been suggested as a surrogate for opioid require- which point the patient was considered a new persistent opioid
ment after discharge, there are very few studies that assess the user if not taking any opioids before this surgery. If a patient
efficacy of individualized opioid prescription guidelines that had a history of chronic opioid use prior to the cardiac proce-
are centered on postsurgical opioid use, written as closely as dure, follow-up was discontinued once the baseline amount of
possible to discharge when patients’ pattern of postoperative opioid intake was resumed.
opioid intake is available to guide the prescription. To the Two members of the study team (J.P. and H.C.) conducted
authors’ knowledge, no study has focused on opioid require- all the phone interviews. The phone survey was a modified
ment after discharge in the cardiac surgery population.6,12-14 version of the structured interview used by Bateman et al. that
The authors hypothesized that the implementation of a tai- was adapted to fit the cardiac surgery population, and the ele-
lored Analgesia Prescription Guideline to guide the predis- ments of the survey can be found in Appendix 1.10 The inter-
charge prescription of opioids after cardiac surgery would lead view consisted of: maximal pain scores on an 11-point
to a reduction in unused leftover opioids. As a first step, they numeric rating scale at five time points (on the day of their pro-
assessed opioid use after cardiac surgery in a before-interven- cedure, immediately after discharge, the first week after dis-
tion cohort. Using this information, they designed an Analge- charge, the second week after discharge, and at the time of
sia Prescription Guideline. The impact of the Analgesia the interview), additional medical care obtained after dis-
Prescription Guideline was assessed in a subsequent after- charge, use of opioids and other analgesics, medication-
cohort of patients, and opioid prescription patterns, opioid tab- related side effects at any time point (including drowsiness,
lets used, and the number of unused leftover opioid tablets nausea/vomiting, abdominal discomfort, constipation, dizzi-
were assessed. ness, confusion, insomnia/sleeping issues, itching, difficult
urination, mood swings), and a Likert-scale, which assessed
Methods overall satisfaction with pain management (rated as very
satisfied, slightly satisfied, satisfied, slightly dissatisfied,
This before and after study was performed at a large quater- dissatisfied, or very dissatisfied).
nary academic medical center from January 2, 2018 to November If the participant filled the opioid prescription, the type of
30, 2018. Institutional review board approval was obtained. All opioid dispensed, the dose, and the number of tablets were
patients with planned or unplanned cardiac surgery requiring a defined by asking the participant to read out the label. This
midline sternotomy incision were screened using the operating information was confirmed using the discharge prescription in
1706 J.J. Pena et al. / Journal of Cardiothoracic and Vascular Anesthesia 35 (2021) 17041711

the participant’s medical record. If the participant received an use in the 48-hour period prior to discharge was included in
opioid prescription at discharge but did not fill the prescription, the individualized Analgesia Prescription Guideline as a factor
the participant was not called again. If the bottle was not avail- to be considered in the opioid prescription. Second, the overall
able, this information was abstracted from the participant’s number of opioid tablets that participants consumed after dis-
medical record. For two participants, information about the charge in the BEFORE cohort (median = ten, interquartile
discharge prescription was missing from the medical record range [IQR]one18) was used to determine an estimated appro-
and could not be estimated by the participants; these two priate range of opioid tablets in the Analgesia Prescription
patients were excluded from analysis. Participants also were Guideline. Third, to avoid underprescribing, a range of opioid
asked whether they requested a refill of the opioid medication. tablets was used in the Analgesia Prescription Guideline that
If a participant was not given an opioid prescription, he or she assumed an overestimation of the number of pills a patient
was asked about whether he or she obtained an opioid prescrip- likely would require.
tion after discharge. The intervention consisted of presenting this guideline in a
The number of leftover tablets then was defined. If the par- lecture to all opioid-prescribing providers on the cardiac sur-
ticipant reported taking all the tablets that were dispensed in gery inpatient service and providing physical copies of the
the initial prescription, the number of leftover tablets was guideline for providers’ reference. The lecture was given on
defined as zero. If there was leftover medication and the bottle multiple occasions July 2 to 13, 2018. An attendance sheet
was still available, the number of tablets left in the bottle were was used to ensure that all providers on the service were
counted. If the bottle was not available, the participant was exposed to the intervention. The team of prescribers consisted
asked to estimate the number of leftover tablets. The number of physician assistants (n = 19) who care for patients under the
of used tablets was defined by subtracting the number of left- supervision of cardiac surgical physicians. The after cohort
over tablets from the number of tablets dispensed as written on was evaluated from July 2 to November 30, 2018 (n = 149
the bottle. screened participants) to assess the efficacy of this intervention
Chart review was performed to gather patient demographics, in reducing the amount of unused opioid.
surgery performed, anesthetic management, cardiopulmonary
bypass duration, length of stay, postoperative analgesic regi- Statistical Analysis
men, smoking status, and home medications. Collected survey
data were transcribed into the Research Electronic Data Cap- This data represent a convenience sample with the number
ture database (REDCap). The code book for the data elements of participants determined by comparison to similar descrip-
collected in the survey is available in Appendix 1. tive studies performed in other surgical populations.8 Descrip-
With use of data from a total of 147 enrolled participants tive statistics are presented as means (standard deviations
between January 2 and April 30 (the before cohort), the Anal- [SD]), medians (IQR), or frequency counts (%). For continu-
gesia Prescription Guideline (Fig 1) was developed. A linear ous variables, normality was assessed using Q-Q plot and his-
regression model was applied to determine the relationship togram, and the Mann-Whitney U or independent two-sample
among participant demographics, postoperative data, known t tests were used to compare medians or means of the two
risk factors for pain after sternotomy15 and opioid use after dis- groups. For categorical variables, a chi-square or Fisher exact
charge. Among all variables tested, opioid use in the 24- and tests were used to test for associations between the categorical
48-hour period prior to discharge had the strongest association variables and group status. Confidence intervals of difference
with opioid use after discharge. Therefore, participants’ opioid of proportions were calculated using the chi-square test with

Fig. 1. This Analgesia Prescription Guideline was presented to all prescribing providers in the cardiothoracic surgery inpatient service and included both inpatient
analgesia and discharge opioid prescription recommendations. Inpatient recommendations emphasized the routine use of nonopioid analgesics prior to prescribing
opioids, and discharge opioid prescription recommendations advised prescribing a varying number of opioid tablets based on the amount of opioid tablets patients
used in the 48-hour period prior to discharge as well as shared decision-making with the patient.
 J.J. Pena et al. / Journal of Cardiothoracic and Vascular Anesthesia 35 (2021) 17041711 1707

Yates continuity correction, and confidence intervals of differ- Control and Prevention and Centers for Medicare and Medic-
ence of medians were calculated using bias-corrected and aid Services were used: oxycodone to morphine 1:1.5 mg; tra-
accelerated bootstrap. madol to morphine 1:0.1 mg; and hydromorphone to morphine
A univariate test was performed to examine the association 1:4 mg.16,17
of each variable and the use of opioids after discharge. Partici-
pant demographic and opioid consumption data from the
before and after cohorts were combined and a simple logistic Results
regression model was built for each variable with the binary
use of opioids after discharge as outcome and each variable of There were 353 eligible patients for inclusion in the study
interest as covariate. The odds ratio (OR) of using opioids after (before n = 178, after n = 175). Participants were not enrolled
discharge and its 95% CI, which were derived from the coeffi- with refusal to participate, or excluded for loss to follow-up,
cient of the model, together with the p value, were determined readmission after discharge, or refusal to participate at the
for each variable. All analyses were conducted using SPSS time of contact. An additional two participants were removed
(IBM SPSS version 25) or R statistical software (RStudio, ver- following chart review and prior to data analysis due to an
sion 3.5.1; R Foundation for Statistical Computing, Vienna, inability to obtain discharge opioid prescription information.
Austria). Figures were made using Microsoft Excel, SPSS, or Two participants were removed from the authors’ analysis
R statistical software. This manuscript adheres to the applica- after intervention because the prescribing provider was not
ble EQUATOR (STROBE) guidelines. exposed to the authors’ intervention. A total of 191 partici-
To standardize opioid prescriptions across different opioid pants were included in the study analysis (before n = 99, after
formulations, conversion factors from the Centers for Disease n = 92; Fig 2).

Fig. 2. Participant flow chart. Abbreviation: LOS, length of stay.

1708 J.J. Pena et al. / Journal of Cardiothoracic and Vascular Anesthesia 35 (2021) 17041711

Table 1
Participant Demographics and Risk Factors for Pain after Sternotomy

Participant demographics Before Cohort (n = 99) After Cohort (n = 92) p Value Mean/Rate Difference (95% CI)

Female (%) 25 (25) 21 (23) 0.737 -0.02 (-0.16, 0.11)

Age (SD) 65 (11) 67 (10) 0.248 2 (-1, 5)
BMI (SD) 28.9 (4.8) 28.7 (5.0) 0.730 -0.2 (-2, 1)
Participant risk factors for poststernotomy pain
History of diabetes (%) 25 (25) 25 (27) 0.869 0.02 (-0.12, 0.15)
History of myocardial infarction (%) 16 (16) 15 (16) >0.999 0.001 (-0.105, 0.107)
History of anxiety/depression (%) 13 (13) 6 (7) 0.151 -0.07 (-0.16, 0.03)
History of substance abuse
Tobacco (%) 33 (33) 31 (34) 0.908 0.004 (-0.134, 0.141)
Other (alcohol, marijuana, and cocaine) (%) 5 (5) 6 (7) 0.761 0.015 (-0.062, 0.092)
History of chronic opioid use (%) 4 (4) 1 (1) 0.370 -0.03 (-0.08, 0.02)
Surgical information
Nonelective surgery (%) 21 (21) 24 (26) 0.496 0.05 (-0.08, 0.18)
Repeat sternotomy (%) 10 (10) 5 (5) 0.287 -0.05 (-0.13, 0.04)
Harvested internal mammary artery (%) 50 (51) 51 (55) 0.562 0.05 (-0.10, 0.20)
Bypass time, min (SD) 109 (41) 113 (46) 0.521 4 (-8, 16)
Total intraoperative fentanyl, mg (SD) 2,400 (753) 2,470 (875) 0.550 71 (-161, 303)
Postoperative length of stay, d (SD) 7 (3) 7 (2) 0.445 -0.3 (-1.0, 0.4)
Opioid taken in last 24 hours prior to discharge (SD) 2 (3) 1 (2) 0.413 -0.27 (-0.9, 0.4)
Opioid taken in last 48 hours prior to discharge (SD) 3 (5) 3 (4) 0.510 -0.44 (-1.8, 0.9)

Abbreviation: SD, standard deviation.

Demographic data, medical comorbidities, and surgical oxycodone 325 mg/5 mg (n = 3, 5%), or hydromorphone 2 mg
information for both cohorts are presented in Table 1. There (n = 1, 2%). Four participants received an opioid prescription
were no differences in demographic characteristics or type of but did not fill it because they did not want or need opioids
procedure in mean intraoperative fentanyl administered or (n = two), and/or experienced side effects during prior expo-
mean postoperative length of stay between the before and after sure (n = one), or already had opioids at home (n = one)
cohorts. There was no difference between groups in the mean (Table 2).
amount of opioid used in the 24- and 48-hour period prior to In the after cohort (N = 92), 48 participants were given an
discharge (24-hour: mean difference -0.3 tablets, CI 95% -0.9, opioid prescription (rate difference -0.1, CI 95% -0.3, 0.05).
0.4, 48-hour use: mean difference -0.4 tablets, CI 95% -1.8, Prescriptions were written for either oxycodone 5 mg (n = 41,
0.9). Additionally, there were no differences among partici- 85%), tramadol 50 mg (n = 5, 10%), acetaminophen/oxyco-
pants who completed the survey and those who were lost to done 325 mg/5 mg (n = 1, 2%), or hydromorphone 2 mg
follow-up with regard to the variables in Table 1. (n = 1, 2%). Four participants received a prescription but did
There were 19 different providers writing a discharge pre- not fill it (rate difference 0.02, CI 95% -0.1, 0.14), because
scription for 62 participants in the before cohort (n = 99). Pre- they did not want or need opioids (n = three) and/or they had
scriptions were written for either oxycodone 5 mg (n = 43, experienced side effects during prior exposure to opioids
69%), tramadol 50 mg (n = 15, 24%), acetaminophen/ (n = two).

Table 2
Opioid Prescription, Consumption, and Storage Outcomes

Before Cohort (n = 99) After Cohort (n = 92) p Value Mean/Rate Difference (95% CI)

Number of participants prescribed opioids (%) 62 (63) 48 (52) 0.187 -0.10 (-0.26, 0.046)
Number of participants who did not fill prescription (%) 4 (6) 4 (8) 0.727 0.02 (-0.10, 0.14)
Mean opioid tablets prescribed (SD) 26 (10) 18 (8) <0.001 -8 (-12, -5)
Mean opioid tablets consumed (SD) 11 (10) 10 (9) 0.589 -1 (-5, 3)
Mean opioid tablets leftover (SD) 15 (12) 8 (8) <0.001 -7 (-11, -3)
Stored opioids in a secure location (%) 4 (7) 6 (14) 0.321 0.07 (-0.07, 0.21)
Disposed of unused opioids (%) 5 (10) 3 (9) >0.999 -0.01 (-0.14, 0.12)
Refill/new opioid prescription obtained (%) 3 (5) 4 (8) 0.697 0.03 (-0.08, 0.15)
Median pain score at time of follow-up (interquartile range) 1 (0, 2) 1 (1, 2) 0.733 0 (-1, 0)
Experienced adverse opioid side effect (%) 18 (31) 21 (48) 0.102 0.16 (-0.04, 0.38)
Satisfied with pain management (%) 83 (95) 83 (93) 0.747 -0.02 (-0.10, 0.06)

Abbreviations: CI, confidence interval; SD, standard deviation.

 J.J. Pena et al. / Journal of Cardiothoracic and Vascular Anesthesia 35 (2021) 17041711 1709

There were no differences in secure storage (rate difference

0.07, CI 95% -0.07, 0.2) or opioid disposal (rate difference
-0.01, CI 95% -0.14, 0.12) between groups.
A total of 39/102 (38%) participants experienced adverse
side effects related to the opioid medication. The most com-
mon side effects reported were excessive sedation/drowsiness
(n = 21), gastrointestinal disturbances including nausea/vomit-
ing/constipation (n = 19), and confusion (n = 12). Other
reported side effects included visual disturbances (n = three),
mood swings (n = two), itching (n = one), and urinary retention
(n = one). There was no statistically significant difference
between overall incidence of side effects between the before
and after cohorts (rate difference 0.16, CI 95% -0.04, 0.38, p
value 0.102).
Fig. 3. Number of days that participants used opioids after discharge from car- With regard to participant satisfaction, 83/87 (95%) partici-
diac surgery.
pants in the before group responded that they were “satisfied”
or “very satisfied” with the perioperative pain management,
Prescribed opioid tablets, reported as mean oxycodone 5- compared with 83/87 (95%) participants in the after group
mg equivalents, used and leftover, are listed in Table 2. (rate difference -0.02, CI 95% -0.10, 0.06).
There was a mean difference of -8 tablets prescribed (CI The results of a univariate logistic regression model for the
95% -12, -4), -1 tablets consumed (CI 95% -5, 3), and -7 use of opioids after discharge using combined data from the
tablets leftover (CI 95% -11, -3) between the before and before and after cohorts are in Table 3. Factors that were asso-
after cohorts. ciated with opioid use after discharge are the number of pre-
Overall, there were 58% of participants who did not use any scribed tablets (OR 1.02, CI 95% 1.01, 1.02), the number of
opioids (n = 110/191) and 5% (N = 10/191) opioids for more tablets used 24 hours prior to discharge (OR 1.31 CI 95% 1.20,
than two weeks after discharge (Fig 3). Among participants 1.42), and age (OR 0.92, CI 95% 0.89, 0.95).
who received an opioid prescription (N = 102), 90% reported
storing the opioid medication in an unlocked or unsecure loca- Discussion
tion. Three participants reported losing track of the opioid pills
and were unable to find the prescribed pills at the time of the The major finding of this study was that the implementation
interview. Among all participants with leftover opioid tablets of a tailored Analgesia Prescription Guideline for opioid pre-
(N = 87), 93% had not disposed of the excess medication. scribers reduced 1.5-fold the number of prescribed opioid

Table 3
Univariate Logistic Regression for Opioid Use after Discharge

Variable Reference Value p Value Odds Ratio 95% CI

Amount of opioid prescribed at discharge (morphine milligram equivalent) <0.001 1.02 1.01, 1.02
Amount of opioid received 24 h prior to discharge (morphine milligram equivalent) <0.001 1.31 1.2, 1.42
Amount of opioid received 48 h prior to discharge (morphine milligram equivalent) <0.001 1.10 1.07, 1.13
Age, y <0.001 0.92 0.89, 0.95
Sex (female/male) Female 0.099 1.81 0.89, 3.65
On home medication for neuropathic pain (yes/no) No 0.152 0.46 0.16, 1.34
History of myocardial infarction (yes/no) No 0.201 0.58 0.26, 1.33
Postoperative length of stay, d 0.306 0.94 0.84, 1.06
History of smoking (nonsmoker/smoker) Nonsmoker 0.413 1.30 0.70, 2.41
Redo sternotomy (yes/no) No 0.451 1.50 0.52, 4.33
History of alcohol or substance abuse (yes/no) No 0.471 1.57 0.46, 5.33
History of chronic opioid use (yes/no) No 0.473 1.94 0.32, 11.91
Total intraoperative fentanyl received, mg 0.569 1.00 1.00, 1.00
History of diabetes (yes/no) No 0.656 1.16 0.60, 2.26
Bypass time, min 0.697 1.00 0.99, 1.01
Length of surgery, min 0.786 1.00 1.00, 1.00
History of depression/anxiety (yes/no) No 0.859 0.92 0.35, 2.40
BMI 0.912 1.00 0.95, 1.06
Internal mammary artery harvested (yes/no) No 0.929 1.03 0.57, 1.84
Emergent surgery (scheduled/emergent) Scheduled 48 1.02 0.52, 2.01

Abbreviations: BMI, body mass index; CI, confidence interval.

1710 J.J. Pena et al. / Journal of Cardiothoracic and Vascular Anesthesia 35 (2021) 17041711

tablets (from 26 to 18) and reduced two-fold the number of related adverse events.19 In the authors’ study, opioid-related
leftover tablets (from 15 to 8) after cardiac surgery. There side effects, such as increased sedation, confusion, and gastro-
were no reductions in the in-hospital or the postdischarge opi- intestinal complications, were reported in 38% of patients
oid use after implementation of the guideline. The authors’ using opioids.
data suggest that the significant decrease in leftover opioids The authors acknowledge several limitations. They con-
observed was due to fewer opioid tablets being prescribed, ducted a single-center observational study of English-speaking
rather than a change in opioid consumption behavior or post- participants undergoing cardiac surgery at a large, academic
operative pain. medical center, generally expected to have uncomplicated
Prior to the implementation of the Analgesia Prescription postoperative recoveries (“fast-track” patients); as such, their
Guideline, there was wide variation in the amount of opioids results may not be representative of or generalizable to all car-
prescribed from zero to 56 tablets, with a mean number of tab- diac surgery patients. Notably, right-heart valve, multiple
lets (oxycodone 5-mg equivalent) of 26; 63% of participants valve, and aortic arch interventions, as well as mechanical cir-
received an opioid prescription at discharge. The majority of culatory device insertions, were not included in this study as
patients used less than half of the amount prescribed. This was these procedures were anticipated to have more complex intra-
consistent with similar studies across other surgical popula- operative and postoperative courses. With regard to the inter-
tions,8 which have estimated that 70% to 80% of patients vention, the authors did not audit prescribers to see whether
either do not fill or do not finish their discharge opioid pre- they were using the Analgesia Prescription Guideline, nor did
scription after surgery. they evaluate their satisfaction with the tool. Although the
The before cohort was used to assess patterns of opioid use guideline was implemented as a recommendation, the authors
in the 48 hours before hospital discharge and subsequent opi- cannot be certain that all prescribers actually followed the rec-
oid use after discharge; these data were used to develop the ommendations and used it on a case-by-case basis. The Anal-
Analgesia Prescription Guideline. The goal of the Analgesia gesia Prescription Guideline did not incorporate shared
Prescription Guideline was to reduce both in-hospital opioid decision-making, which may have further allowed to individu-
use and discharge opioid prescriptions, using multimodal opi- alize the prescription. Approximately one-third of participants
oid-sparing analgesia. Postoperative analgesic prescriptions were lost to follow-up, and it is possible that these participants
included standard orders for nonopioid analgesics before esca- had different opioid use and pain recovery trajectories. For
lating to opioids, which were prescribed for breakthrough pain participants unable to retrieve their prescription bottle at the
only on an as-needed basis. Recommendations for the dis- time of the phone survey but who reported still having remain-
charge opioid prescription were based on two important factors ing pills (n = nine), the authors relied on recall of their pre-
identified in the initial survey, one general and one specific to scribed amount to calculate used and leftover opioid tablets,
each patient. First, the median number of opioid tablets con- which may result in inaccuracy in their estimates. In addition,
sumed across all surgeries was ten tablets (interquartile range the authors did not include a control group simultaneous to
one-18). Second, patients’ opioid use during the 24- and 48- the study period to specifically evaluate the effect of the
hour period prior to discharge was correlated with the number Analgesia Prescription Guideline, and, therefore, they can-
of opioid tablets they used at home. The authors used these not exclude the impact of the lay and scientific press cov-
two factors to propose ranges of opioid tablets to be pre- erage of this topic, although the two cohorts were studied
scribed, and this resulted in a recommendation from no opioid within the same year.
prescription to a maximum of 30 tablets. The authors decided Finally, although the authors’ Analgesia Prescription Guide-
for a relatively high upper limit of tablets to reduce the risk of line appears to have had a positive effect in reducing the
insufficient prescription and need for refill, and this amount amount of prescribed and unused opioid, it is not clear if this
was higher than the zero to- 20-tablet range later proposed by intervention will have a long-lasting effect on opioid prescrib-
expert consensus.6 Notably, of the 15 patients in the after ing practice, and additional follow-up is needed to demonstrate
cohort who received prescriptions for more than 20 tablets, a the lasting efficacy of this intervention.
median 16 tablets per participant were left over, suggesting Although the authors’ educational intervention was effec-
that the authors’ recommendation could be restricted further tive in changing prescription practices, it did not improve out-
without adverse consequences. patient safe storage nor disposal of opioid medications.
Using univariate logistic regression, several factors were Further interventions will be necessary to address the patient’s
associated with opioid use after discharge, including younger role in opioid stewardship.
age, higher number of prescribed tablets, and higher opioid The pragmatic guideline used in this study was a cost-free
use at 24 and 48 hours prior to discharge. Among these, the and effective intervention that reduced opioid prescriptions in
variable with the strongest association for opioid use after dis- this cardiac surgery service. Future iterations of this guideline
charge was opioid use during the 24-hour period prior to dis- will explore further reduction in the recommended amount of
charge. This finding was consistent with a study of opioid use opioids prescribed, as well as incorporate the use of additional
in obstetric and thoracic surgical populations.18 nonopioid analgesics, which may reduce chronic pain follow-
Individualized opioid prescription is especially important ing cardiac surgery.20 The implementation of the Analgesia
following cardiac surgery, as geriatric patients are more sensi- Prescription Guideline or similar guides that use multimodal
tive to opioid medications and are at increased risk for opioid- therapies for perioperative analgesia likely would result in a
 J.J. Pena et al. / Journal of Cardiothoracic and Vascular Anesthesia 35 (2021) 17041711 1711

significant reduction in the number of leftover opioid tablets in 4 Seth P, Scholl L, Rudd RA, et al. Overdose deaths involving opioids,
the cardiac surgery population without a reduction in patient cocaine, and psychostimulants—United States, 2015-2016. MMWR Morb
satisfaction and with improvement in postsurgical analgesia. Mortal Wkly Rep 2018;67:349–58.
5 Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids
J. Pena helped across all aspects of this study, including con- for chronic pain—United States, 2016. MMWR Recomm Rep 2016;65:1–
ceptualization, study design, participant recruitment, data col- 49.
lection, data analysis, figure preparation, and manuscript 6 Overton HN, Hanna MN, Bruhn WE, et al. Opioid-prescribing guidelines
preparation. C.J. Chen helped in participant recruitment, data for common surgical procedures: An expert panel consensus. J Am Coll
collection, and figure preparation and helped review and edit Surg 2018;227:411–8.
7 Farley KX, Anastasio AT, Kumar BA, et al. Association between quantity
the manuscript. H. Clifford helped in participant recruitment, of opioids prescribed after surgery or preoperative opioid use education
data collection, and figure preparation and helped review and with opioid consumption. JAMA 2019;321:2465–7.
edit the manuscript. Z. Xue along with S. Wang performed the 8 Brummett CM, Waljee JF, Goesling J, et al. New persistent opioid use after
majority of statistical analyses for the manuscript and assisted minor and major surgical procedures in US Adults. JAMA Surg 2017;152:
in figure preparation as well as manuscript editing. S. Wang e170504.
9 Bicket MC, Long JJ, Pronovost PJ, et al. Prescription opioid analgesics
along with Z. Xue performed the majority of statistical analy- commonly unused after surgery: A systematic review. JAMA Surg
ses for the manuscript and assisted in figure preparation as 2017;152:1066–71.
well as manuscript editing. M. Argenziano assisted in study 10 Bateman BT, Cole NM, Maeda A, et al. Patterns of opioid prescription and
design as well as manuscript editing. R. Landau assisted in use after cesarean delivery. Obstet Gynecol 2017;130:29–35.
11 Kennedy-Hendricks A, Gielen A, McDonald E, et al. Medication sharing,
study design and data analysis, as well as manuscript prepara-
storage, and disposal practices for opioid medications among US adults.
tion and editing. M.-L. Meng acted as the principal investiga- JAMA Intern Med 2016;176:1027–9.
tor for the study and coordinated all aspects of the study 12 Flanagan CD, Wysong EF, Ramey JS, et al. Understanding the opioid epi-
including conceptualization, study design, institutional review demic: Factors predictive of inpatient and postdischarge prescription opi-
board submission, participant recruitment, data analysis, and oid use after orthopaedic trauma. J Orthop Trauma 2018;32:e408–14.
manuscript and figure preparation, as well as editing. 13 Mark J, Argentieri DM, Gutierrez CA, et al. Ultrarestrictive opioid pre-
scription protocol for pain management after gynecologic and abdominal
surgery. JAMA Netw Open 2018;1:e185452.
Declaration of Competing Interest 14 Chen EY, Marcantonio A, Tornetta P. Correlation between 24-hour predis-
charge opioid use and amount of opioids prescribed at hospital discharge.
None. JAMA Surg 2018;153:e174859.
15 Veal FC, Bereznicki LRE, Thompson AJ, et al. Pain and functionality fol-
lowing sternotomy: a prospective 12-month observational study. Pain Med
Supplementary materials 2016;17:1155–62.
16 Centers for Disease Control and Prevention (CDC). Calculating total daily dos-
Supplementary material associated with this article can be age of opioids for safer dosage. Available at: https://www.cdc.gov/drugover-
found in the online version at doi:10.1053/j.jvca.2020.12.021. dose/pdf/calculating_total_daily_dose-a.pdf. Accessed 2 January 2021.
17 Centers for Medicare and Medicaid Services (CMS). Opioid morphine
equivalent conversion factors. Available at: https://www.cms.gov/Medi-
References care/Prescription-Drug-Coverage/PrescriptionDrugCovContra/Downloads/
OpioidMorphine-EQ-Conversion-Factors-March-2015.pdf. Accessed 2
1 Zhu W, Chernew ME, Sherry TB, et al. Initial opioid prescriptions among January 2021.
U.S. commercially insured patients, 2012-2017. N Engl J Med 18 Bartels K, Mayes LM, Dingmann C, et al. Opioid use and storage patterns
2019;380:1043–52. by patients after hospital discharge following surgery. PLoS One 2016;11:
2 Rudd RA, Seth P, David F, et al. Increases in drug and opioid-involved e0147972.
overdose deaths—United States, 2010-2015. MMWR Morb Mortal Wkly 19 Chau DL, Walker V, Pai L, et al. Opiates and elderly: Use and side effects.
Rep 2016;65:1445–52. Clin Interv Aging 2008;3:273–8.
3 Guy GP Jr, Zhang K, Bohm MK, et al. Vital signs: Changes in opioid pre- 20 Anwar S, Cooper J, Rahman J, et al. Prolonged perioperative use of prega-
scribing in the United States, 2006-2015. MMWR Morb Mortal Wkly Rep balin and ketamine to prevent persistent pain after cardiac surgery. Anes-
2017;66:697–704. thesiology 2019;131:119–31.