Secondary Logo

Journal Logo

Featured Articles: Original Clinical Research Report

Opioid-Sparing Cardiac Anesthesia: Secondary Analysis of an Enhanced Recovery Program for Cardiac Surgery

Grant, Michael C. MD, MSE*,†; Isada, Tetsuro MD*; Ruzankin, Pavel PhD‡,§; Gottschalk, Allan MD, PhD*,‖; Whitman, Glenn MD; Lawton, Jennifer S. MD; Dodd-o, Jeffrey MD*; Barodka, Viachaslau MD*

Author Information
doi: 10.1213/ANE.0000000000005152

Abstract

See Article, p 1850

KEY POINTS

  • Question: Are nonopioid interventions, incorporated as part of an Enhanced Recovery Program (ERP) for cardiac surgery, associated with reduced intraoperative opioid administration?
  • Findings: Nonopioid interventions are associated with a marked reduction in intraoperative opioid administration in cardiac surgery, and no differences were detected in selected postoperative outcomes between patients who received high (>50 morphine sulfate equivalent [MSE]) or low (<50 MSE) intraoperative opioids.
  • Meaning: Although an ERP for cardiac surgery may facilitate low dose (<50 MSE) opioid administration, additional prospective trials specifically designed to evaluate the impact of low opioid anesthetics on postoperative outcomes are necessary.

In 1969, Lowenstein et al1 published their landmark article detailing the use of high dose morphine (1 mg/kg) as the sole anesthetic for aortic valve disease requiring open-heart surgery. The anesthetic, noteworthy for its hemodynamic stability, became increasingly popularized in the setting of cardiac surgery over the interceding decades.2 Although a variety of adjuncts have been adopted during that time, including amnestic, maintenance, and paralytic medications,3,4 present-day cardiac anesthesia continues to rely heavily on opioids. At the turn of the century, fast-track cardiac surgery sought to lower opioid administration,5 and although the overall amount of opioid has been reduced since the era of Lowenstein and Stoelting, standard fentanyl use for cardiac surgery stands at 10–15 µg/kg (~70–105 morphine sulfate equivalents [MSE]) for a 70 kg patient.5,6

There are several reasons to reconsider the role of opioids in the setting of cardiac surgery. First, the US opioid crisis has led to the reevaluation of periprocedural anesthesia and pain management. Evidence suggests an association between intraoperative opioid administration and subsequent postoperative opioid use7 as well as the incidence of postoperative complications, including readmission.8 A number of procedures have been implicated as significant contributors to new persistent opioid use, defined as opioids prescribed greater than 90 days following surgery in otherwise naive patients.9 Cardiac surgery, a potentially underappreciated contributor, likely plays a more central role than expected with a 6%–12% proportion of persistent opioid use compared to more modest estimates in general surgical populations.10,11 These figures, coupled with ever more knowledgeable patients and expanding state opioid “opt-out” legislation,12,13 which allows patients to refuse opioid-based analgesia as part of their inpatient hospital stay, suggest that cardiac service lines must develop strategies to markedly reduce the use of perioperative opioids. The use of multimodal analgesia and opioid reduction strategies was included in recommendations recently put forth by the Enhanced Recovery After Surgery Cardiac Society.14

Our group has implemented an Enhanced Recovery Program (ERP) for cardiac surgery, which employs a series of nonopioid interventions, that may reduce the reliance on and potentially eliminate the use of intraoperative opioids. Although other results of the program have been previously reported,15 here, we hypothesized that the use of nonopioid interventions introduced as part of an ERP for cardiac surgery was associated with reduced intraoperative opioid administration. In addition, we hypothesized that intraoperative opioid administration would not be associated with differences in postoperative outcomes.

METHODS

Institutional Cardiac Surgery Study Overview

This study represents a secondary analysis of data originally obtained as part of a parent ERP for the cardiac surgery program, the results of which have been published previously.15 As reported, the ERP for cardiac surgery is an extension of an existing pathway instituted by the cardiac surgery Clinical Communities Committee, a multidisciplinary collaboration between surgery, anesthesiology, and perioperative nursing, among others.15 Further details of the Clinical Communities initiatives have also been outlined previously.15,16 Following approval, the requirement for written informed consent was waived by the institutional review board. Data were retrospectively collected for the period of September 2017 to May 2018 on all patients undergoing coronary artery bypass grafting (CABG), aortic, mitral, tricuspid valve and combination CABG + valve procedures at the Johns Hopkins East Baltimore campus. Patients undergoing aortic, ventricular assist, or transplant procedures were excluded from the analysis. All surgeries were performed via median sternotomy incision by 1 of the 6 primary board-certified cardiac surgeons.

Standard Cardiac Anesthetic

The anesthetic was guided by a formal institutional ERP for cardiac surgery Preanesthetic and Intraoperative Guidelines.15 The selection of specific induction or maintenance anesthetics were left to the discretion of individual providers. Interventions adopted as part of the program were developed from a combination of literature review and institutional experience, agreed on by provider consensus, introduced at program inception, and reinforced as part of anesthetic guidelines over a 9-month timeframe. The 5 interventions that were instituted as part of the program included

  1. Acetaminophen (ie, 1000 mg; either administered orally in preanesthesia area or via the intravenous route in the operating room after separation from bypass and before exiting the operating room)
  2. Gabapentin (ie, administered orally in the preanesthesia area; 600 mg standard dose, reduce to 300 mg if age >70 years or renal clearance <60 mL/min).
  3. Ketamine (ie, intraoperative infusion; 0.2–0.3 mg/kg/h)
  4. Dexmedetomidine (ie, intraoperative infusion; 0.2–1.5 µg/kg/h; administered at the time of cardiopulmonary bypass and throughout transport to the intensive care units [ICU] and titrated based on hemodynamics and sedation goal)
  5. Regional nerve block (ie, Serratus Anterior Plane block; administered at the end of surgery under ultrasound guidance via “single-shot” technique; bupivacaine 0.375%–0.5%, 20–30 mL bilaterally)

It was intended that patients receive all interventions unless contraindicated due to existing medical comorbidities. The decision to withhold or alter a specific intervention was guided by both the anesthetic outline (ie, recommended medication dose modification based on glomerular filtration rate) as well as individual anesthesiologist discretion. Intraoperative opioid administration was not withheld as part of this program but rather utilized as a rescue medication when providers noted evidence of noxious stimulation (ie, tachycardiac and/or hypertension associated with a surgical incision) after the administration of nonopioid analgesics, which is outlined in the supplement to our prior publication.15 The amount or interval of opioid use was left to the discretion of the individual anesthesiologist.

Outcome Variables and Data Collection

The primary objective was to assess the association between the number of interventions provided and intraoperative opioid administration, with the primary outcome defined as the amount of opioid administered intraoperatively. To investigate the primary objective, data, including the medication, dose, and attending anesthesia provider, were collected. Other relevant patient-specific variables, such as age, sex, procedure, Society of Thoracic Surgeons (STS) predicted risk of mortality (PROM), left ventricular ejection fraction, laboratory data, and cardiopulmonary bypass time, were also collected. A predefined secondary objective was to assess the association between individual interventions and intraoperative opioid administration. This utilized a similar approach to the primary objective, but with separate interventions, rather than a composite one.

An additional predefined secondary objective was to assess the association between intraoperative opioid administration and postoperative outcomes. Clinical outcome data, including 12- and 24-hour postoperative opioid administration, ICU patient-reported pain scores, time to extubation, and postoperative ICU, floor and hospital length of stay (LOS), the incidence of atrial fibrillation, delirium, acute kidney injury (AKI), chest tube output, reintubation, and mortality were thus collected. Pain scores, delirium, atrial fibrillation, and chest tube output were derived from nursing, vital signs, and fluid balance flowsheets. Delirium assessment was based on per shift Confusion Assessment Method–ICU evaluation. Postoperative pain scores were based on the 11-point numeric pain rating scale (0–10), with 0 = no pain and 10 = the worst pain imaginable. AKI was diagnosed utilizing the Kidney Disease Improving Global Outcomes criteria and based on plasma creatinine (0 corresponding no AKI, and 1, 2, 3 corresponding to AKI stage 1, 2, or 3). Data were obtained by querying the existing institutional information system (Epic Systems Corporation, Verona, WI) and combined into a single patient database.

Statistical Methods

To assess the primary objective, a linear mixed-effects regression model was constructed with total opioid dose as the outcome. The secondary objective concerning the association between intraoperative opioid administration with individual interventions was assessed with linear mixed-effects regression as well. The random effect for the intercept in each case was the first cardiac anesthesiologist conducting the anesthetic. To control for potential confounding, we used the following additional covariates in these regression models: age, sex, ICU arrival Vasoactive-Inotropic Score (VIS),17 STS PROM, ejection fraction, preoperative hemoglobin, creatinine, lactate, height, weight, cardiopulmonary bypass time, and procedure type (CABG, aortic valve, mitral valve, and CABG/aortic valve procedures were considered).

Secondary analysis via propensity score matching was performed to assess the association between intraoperative opioid administration and postoperative outcomes. First, patients were categorized into high (>50 MSE) and low (≤50 MSE) opioid administration groups to assess if the overall intraoperative opioid dose was associated with postoperative outcomes. This cutoff was selected as it represents the median per patient amount of intraoperative opioid administration over the course of the study. High opioid patients were 1-to-1 propensity matched to low opioid patients and the score was estimated with logistic regression based on the following patient-level variables: sex, age, ICU arrival VIS, STS PROM, weight, height, left ventricular EF, CBT, preoperative creatinine, hemoglobin, and lactate, with the outcome being the probability of low opioid administration. The matching was exact for the procedure type. The matching was performed with a greedy algorithm without replacement (Algorithm A18) with caliper width 0.0332 for propensity score. The caliper width was chosen by the following procedure. All of the calipers from 0 to 0.1 in increments of 0.0001 were examined. Of these, we selected the calipers corresponding to matchings with absolute standardized differences19 <0.1 and variance ratios20 <2 for all variables. Of those calipers, the ones that correspond to the maximal number of matched pairs were selected, and the minimal caliper is thus selected. Covariates’ balance was assessed in R with Cobalt package v. 3.9.0 (N.Greifer, R, Vienna, Austria).Standardized differences were calculated with pooled variances. All covariates were appropriately balanced (Supplemental Digital Content, Figure 1, https://links.lww.com/AA/D178). Overall results are expressed in unmatched and matched forms.

Postoperative outcomes for the matched observations were analyzed via the following statistical tests. For continuous variables, data are expressed as mean value ± standard deviation. For binary variables, data are expressed as frequency (percentage). Comparisons for unmatched continuous variables were analyzed by the Wilcoxon rank sum test. For unmatched binary variables, comparisons were analyzed by the χ2 test if all the expected values were >5 and the Fisher exact test otherwise. For matched variables, comparisons were analyzed by the Wilcoxon signed rank test or McNemar test for paired proportions. Median time (ie, time to extubation) was expressed as an estimate with 95% confidence interval (CI) based on the cumulative hazard. The null hypotheses for all the tests were that there is no association between the comparison group (ie, high or low opioids) and the outcome.

To further assess the association between opioid administration and postoperative outcomes, a post hoc analysis was performed where ultralow (≤25 MSE) opioid patients were identified and propensity matched (1-to-3) to high (>50 MSE) opioid patients in an identical fashion to that described above (Supplemental Digital Content, Figure 2, https://links.lww.com/AA/D178), except that the matching algorithm was greedy nearest neighbor matching without replacement followed by optimal rematching.18 The caliper width was 0.066 for propensity score, and the comparisons for binary variables were analyzed with univariate conditional logistic regression. The caliper width was chosen by a procedure similar to that in the previous matching. The matching algorithm was changed, since Algorithm A,18 though known to decrease maximal within-pair score distance (for a given number of matched pairs) in one-to-one matching compared to greedy nearest neighbor matching, does not perform well for matching with multiple controls. On the other hand, greedy nearest neighbor matching with optimal rematching18 is known to decrease average within-pair score distance compared to greedy nearest neighbor matching both in one-to-one and one-to-many matching.

Results of each propensity matching analysis are reported. For matched binary variables, data are expressed as weighted count (percentage) (unweighted count [percentage]). When 2 or 3 high opioid patients were matched to 1 ultralow opioid patient, the high opioid patients were assigned weights 1/2 or 1/3, respectively. All other cases had a weight of 1. For all analyses, P < .05 was considered statistically significant. Data were processed and analyzed with the software programs Excel, v. 14.0 (Microsoft Inc, Redmond, WA), and R v. 3.5.2 (R Core Team, Vienna, Austria) statistical packages.

RESULTS

Interventions and Intraoperative Opioid Administration

In total, 451 consecutive patients were included in our cohort. Figure 1 depicts the study time course, including overall per-patient interventions and temporal opioid administration. As shown, as the number of interventions increased over the duration of the study, there was an associated decrease in intraoperative opioid administration. Figure 2 depicts the stepwise reduction in opioids associated with the total number of interventions employed on a per-patient basis. A total of 99 patients received no interventions, 116 patients received 1 intervention, 78 patients received 2 interventions, 87 patients received 3 interventions, 58 patients received 4 interventions, and 13 patients received all 5 interventions. Overall, gabapentin was provided to 33.9%, acetaminophen to 47.2%, ketamine infusion to 57.0%, dexmedetomidine to 17.5%, and regional analgesia to 28.4% of patients. Results of the linear mixed-effects regression model for the primary objective (Table 1) revealed that intraoperative opioid administration was inversely related to the number of interventions provided (estimated −7.96 MSE per 1 intervention, 95% CI, −9.82 to −6.10, P < .001). Assessment of the relationship between individual interventions and intraoperative opioid administration is also shown in Table 1. The individual interventions independently associated with low opioids were dexmedetomidine (estimated −25.80 MSE, 95% CI, −32.45 to −19.17, P < .001), regional analgesia (estimated −12.04 MSE, 95% CI, −17.70 to −6.38, P < .001), gabapentin (estimated −9.39 MSE, 95% CI, −16.64 to −2.13, P = .01).

Table 1. - Results of the 2 Linear Mixed-Effects Regression Models for Intraoperative Opioid Administration
Variable Estimate (95% CI) P
Model 1
 Number of interventions per patient −7.96 (−9.82 to −6.10) <.001
Model 2
 Ketamine infusion −4.53 (−9.78 to 0.73) .099
 Dexmedetomidine infusion −25.80 (−32.45 to −19.17) <.001
 Acetaminophen 3.23 (−3.38 to 9.83) .35
 Regional analgesia −12.04 (−17.70 to −6.38) <.001
 Gabapentin −9.39 (−16.64 to −2.13) .014
The table depicts the results of the 2 linear mixed-effects regression models for total intraoperative opioid administration in MSE as the outcome. The random effect of intercept was the first anesthesiologist. Other covariates included were age, sex, ICU arrival VIS, STS PROM, ejection fraction, preoperative hemoglobin, creatinine, lactate, height, weight, cardiopulmonary bypass time, procedure type (CABG, aortic valve, mitral valve, and CABG/aortic valve procedures were considered). The first model (model 1) assesses the association between the total number of interventions per patient and the intraoperative opioid administration. The second model (model 2) assesses the association of individual interventions with intraoperative opioid administration. The estimated value represents the average increase in opioids per intervention.
Abbreviations: CABG, coronary artery bypass grafting; CI, confidence interval; ICU, intensive care unit; MSE, morphine sulfate equivalents; STS PROM, Society of Thoracic Surgeons Predicted Risk of Mortality; VIS, Vasoactive-Inotropic Score.

Figure 1.
Figure 1.:
Intraoperative opioid administration and coincident intervention compliance as a function of time. An increased composite number of interventions is inversely associated with intraoperative opioid use. MSE indicates morphine sulfate equivalents.
Figure 2.
Figure 2.:
Mean intraoperative opioid administration as a function of total number of nonopioid interventions employed per-patient. The boxes depict mean ± standard deviation. The whiskers depict minimal and maximal values. MSE indicates morphine sulfate equivalents.

Supplemental Digital Content, Figure 3, https://links.lww.com/AA/D178, compares the individual anesthesiologist median intervention compliance during the early (September 2017–January 2018) and late (February 2018–May 2018) stages of the study period. As shown, individual provider compliance was nearly universally increased over the duration of the program (17 of 19 providers). Supplemental Digital Content, Figure 4, https://links.lww.com/AA/D178, compares the individual anesthesiologist median intraoperative opioid administration during the early and late stages of the study period. Intraoperative opioid administration was reduced across all cardiac anesthesiologists.

High Versus Low Opioids and Clinical Outcomes

Table 2. - Demographics and Procedure Characteristics of High and Low Propensity-Matched Cohorts
Unmatched Matched
>50 MSE ≤50 MSE ASD >50 MSE ≤50 MSE ASD
Cases, no. 214 (100.0%) 237 (100.0%) 136 (100.0%) 136 (100.0%)
Sex, male 166 (77.6%) 166 (70.0%) 0.052 109 (80.1%) 106 (77.9%) 0.022
Age, y 61.4 (±12.0) 66.0 (±11.3) 0.481 64.2 (±10.1) 65.2 (±9.2) 0.085
Vasoactive-Inotropic score, units 4.85 (±4.22) 6.76 (±5.57) 0.330 5.52 (±4.70) 5.20 (±3.76) 0.068
STS PROM, % 1.93 (±3.52) 2.47 (±3.43) 0.170 1.79 (±3.54) 1.78 (±2.67) 0.002
Ejection fraction, % 53.27 (±11.83) 53.33 (±12.41) < 0.001 51.88 (±12.14) 52.28 (±12.05) 0.033
Preoperative hemoglobin, g/dL 13.06 (±2.10) 12.88 (±2.21) 0.088 13.17 (±1.92) 13.07 (±2.00) 0.050
Creatinine, mg/dL 1.11 (±0.76) 1.31 (±1.32) 0.193 1.08 (±0.33) 1.09 (±0.28) 0.008
Lactate, mmol/L 1.00 (±0.61) 0.95 (±0.36) 0.069 0.97 (±0.50) 0.97 (±0.38) 0.018
Weight, kg 88.49 (±21.75) 85.06 (±19.10) 0.162 88.22 (±22.11) 89.83 (±19.22) 0.079
Height, m 1.72 (±0.10) 1.72 (±0.10) 0.020 1.72 (±0.10) 1.73 (±0.09) 0.061
Cardiopulmonary bypass time, min 116.8 (±54.9) 116.3 (±52.9) 0.056 113.7 (±53.0) 109.1 (±50.7) 0.087
CABG 155 (72.4%) 148 (62.4%) 0.074 114 (83.8%) 114 (83.8%) 0
AVR 22 (10.3%) 20 (8.4%) 0.015 7 (5.1%) 7 (5.1%) 0
MVR 13 (6.1%) 28 (11.8%) 0.066 5 (3.7%) 5 (3.7%) 0
CABG + AVR 15 (7.0%) 21 (8.9%) 0.024 10 (7.4%) 10 (7.4%) 0
CABG + MVR 5 (2.3%) 4 (1.7%) 0 0
AVR + TVR 1 (0.5%) 0 0 0
AVR + MVR + TVR 1 (0.5%) 1 (0.4%) 0 0
AVR + MVR 1 (0.5%) 5 (2.1%) 0 0
TVR 1 (0.5%) 6 (2.5%) 0 0
AVR+MVR + CABG 0 2 (0.8%) 0 0
MVR + TVR 0 2 (0.8%) 0 0
Abbreviations: ASD, absolute standardized difference; AVR, aortic valve replacement/repair; CABG, coronary artery bypass grafting; ICU, intensive care unit; MSE, morphine sulfate equivalents; MVR, mitral valve replacement/repair; STS PROM, Society of Thoracic Surgeons Predicted Risk of Mortality; TVR, tricuspid valve replacement/repair.

Table 3. - Clinical Parameters and Outcomes Based on High and Low Propensity Matching
Unmatched Matched
All Cases >50 MSE ≤50 MSE P All Cases >50 MSE ≤50 MSE P 95% CI
Cases, no. 451 (100.0%) 214 (100.0%) 237 (100.0%) 272 (100.0%) 136 (100.0%) 136 (100.0%)
Intraoperative lactate (peak) 2.8 (±1.8) 2.9 (±2.0) 2.8 (±1.6) .89a 2.7 (±1.5) 2.8 (±1.7) 2.6 (±1.3) .68c −0.40 to 0.25d
Intraoperative hemoglobin (low) 8.5 (±1.6) 8.6 (±1.6) 8.4 (±1.6) .070a 8.6 (±1.5) 8.7 (±1.6) 8.5 (±1.5) .28c −0.60 to 0.20d
Pain score 0–12 h (NRS) 3.9 (±2.9) 4.2 (±2.9) 3.6 (±2.9) .039a 3.9 (±2.8) 4.0 (±2.8) 3.9 (±2.8) .91c −0.77 to 0.67d
Pain score 12–24 h (NRS) 4.0 (±2.6) 4.4 (±2.6) 3.7 (±2.5) .001a 4.1 (±2.5) 4.2 (±2.6) 4.0 (±2.4) .44c −0.91 to 0.42d
Creatinine (Δ) 0.1 (±0.7) 0.1 (±0.7) 0.1 (±0.8) .65a 0.1 (±0.6) 0.1 (±0.6) 0.1 (±0.6) .27c −0.15 to 0.05d
No AKI 360 (79.8%) 171 (79.9%) 189 (79.7%) .87b 219 (80.5%) 107 (78.7%) 112 (82.4%) .62c −1.0 to 0.5d
AKI1 66 (14.6%) 34 (15.9%) 32 (13.5%) 41 (15.1%) 24 (17.6%) 17 (12.5%)
AKI2 16 (3.5%) 5 (2.3%) 11 (4.6%) 7 (2.6%) 1 (0.7%) 6 (4.4%)
AKI3 9 (2.0%) 4 (1.9%) 5 (2.1%) 5 (1.8%) 4 (2.9%) 1 (0.7%)
ICU hemoglobin (low) 8.3 (±1.6) 8.5 (±1.5) 8.2 (±1.6) .058a 8.4 (±1.6) 8.4 (±1.5) 8.5 (±1.6) .75c −0.25 to 0.45d
ICU lactate highest 5.4 (±3.5) 5.4 (±3.7) 5.4 (±3.4) .80a 5.4 (±3.4) 5.6 (±3.4) 5.3 (±3.5) .39c −1.15 to 0.45d
Reintubation 24 (5.3%) 8 (3.7%) 16 (6.8%) .23e 14 (5.1%) 5 (3.7%) 9 (6.6%) .42f 0.62 to 6.18g
ICU atrial fibrillation 81 (18.0%) 24 (11.2%) 57 (24.1%) .001e 41 (15.1%) 15 (11.0%) 26 (19.1%) .10f 0.97 to 3.86g
ICU chest tube output (24 h) 1378.8 (±1150.7) 1305.5 (±989.9) 1445.0 (±1277.2) .20a 1341.5 (±1125.9) 1330.6 (±915.8) 1352.4 (±1306.1) .45c −165.00 to 85.00d
ICU delirium 20 (4.4%) 10 (4.7%) 10 (4.2%) 1.00e 11 (4.0%) 8 (5.9%) 3 (2.2%) .13f 0.08 to 1.28g
In hospital mortality 25 (5.5%) 13 (6.1%) 12 (5.1%) .79e 8 (2.9%) 6 (4.4%) 2 (1.5%) .29f 0.05 to 1.43g
Median intubation time, h 5 (4.5, 5.5) 5 (4.5, 5.5)h 5 (4.5, 5.5)h .64i 4.5 (4, 5) 5 (4, 5.5)h 4 (3.5, 5)h .17i
Median ICU time, h 41.5 (35.5, 45) 36.75 (27, 44)h 44.5 (38, 48.5)h .010i 39.5 (28, 44) 40.25 (28, 45.5)h 37 (26, 44.5)h .76i
Median floor time, h 119 (116, 122) 120.25 (116, 124)h 117.5 (102.5, 122.5)h .13d 117.25 (101, 121.5) 119 (101, 122.5)h 116 (99.5, 122.5)h .37i
Median hospital time, h 165.5 (160, 170.5) 164 (147, 169.5)h 168.5 (160.5, 188.5)h .028i 160.5 (146, 167) 164 (145, 170.5)h 160 (143, 170)h .65i
For matched binary variables, the values are reported as weighted count (percentage) (unweighted count [percentage]).
Abbreviations: AKI (1, 2, 3), acute kidney injury (Stage 1, 2, 3); CI, confidence interval; ICU, intensive care unit; MSE, morphine sulfate equivalents; NRS, numeric rating score.
aWilcoxon rank sum test.
bFisher exact test.
cWilcoxon signed rank test.
dHodges–Lehmann CI for pseudomedian of matched differences (≤50 MSE observation) − (>50 MSE observation).
eχ2 test.
fMcNemar test for paired proportions with continuity correction.
gLogistic regression confidence interval for odds ratios (odds for ≤50 MSE) versus (odds for >50 MSE).
hCI based on the cumulative hazard.
iLog rank test.

Patient characteristics are represented in Table 2. Among unmatched patients, high opioid (>50 MSE, n = 214) participants were younger (average 61.4 vs 66.0 years), utilized less vasopressor (average 4.85 vs 6.76 units), had lower STS PROM (average 1.93% vs 2.47%), had lower creatinine (average 1.11 vs 1.31 mg/dL), and were heavier (average 88.5 vs 85.1 kg) compared to the low (≤50 MSE, n = 237) opioid counterparts. Following propensity score matching, a total of 272 patients (n = 136 in each group) were identified for analysis. As shown in Table 2, matching was appropriate, all the covariates being balanced with a maximal absolute standardized difference of 0.087 and a maximal variance ratio of 1.75. The Love plot depicting the covariates’ balance is shown in Supplemental Digital Content, Figure 1, https://links.lww.com/AA/D178. After propensity matching, no differences were detected between groups in postoperative outcomes (Table 3), including postoperative pain scores within the first 12 and 24 hours, respectively. Of note, both groups experienced a >60% rate of extubation within the first 6 postoperative hours.

High Versus Ultralow Opioids and Clinical Outcomes

Supplemental Digital Content, Table 1, https://links.lww.com/AA/D178, provides baseline comparison demographics, intraoperative variables, and procedural information for ultralow (n = 91) and high (n = 214) opioid participants. High opioid patients were younger (average 61.4 vs 66.0 years), required less vasopressors (average 4.85 vs 7.20 units), had lower STS PROM (average 1.93 vs 2.21), had lower ejection fraction (average 53.3% vs 54.7%), were less anemic (average 13.06 vs 12.51 g/dL), had lower creatinine level (average 1.11 vs 1.23 mg/dL), had higher lactate level (average 1.00 vs 0.93 mmol/L), and were heavier (average 88.5 vs 84.9 kg) than ultralow opioid patients. A total of 8 patients (1.8% of the entire cohort, 3.4% of unmatched low opioid patients) received no intraoperative opioids during their surgery. Following propensity score matching, a total of 195 patients (n = 63 in the ultralow opioid group and n = 132 in the high opioid group) were identified for analysis. The matching was appropriate as all covariates were balanced, with a maximal absolute standardized difference of 0.067 and a maximal variance ratio of 1.46. The Love plot depicting the covariates’ balance is shown in Supplemental Digital Content, Figure 2, https://links.lww.com/AA/D178. No differences were detected in postoperative outcomes between groups (Supplemental Digital Content, Table 2, https://links.lww.com/AA/D178).

DISCUSSION

The results of our institutional ERP for cardiac surgery program show that deployment of nonopioid interventions was associated with a reduction in the mean intraoperative opioid administration from 80 MSE (800 µg of fentanyl) at program inception to <20 MSE (200 µg of fentanyl) at the conclusion. This corresponds to 2–3 µg/kg of fentanyl use (for a 70 kg patient). The reduction in intraoperative opioid use coincided with the increased use of the 5 prescribed interventions. Multivariable regression revealed that the number of interventions utilized was associated with low opioid administration. Finally, no significant differences were detected in postoperative outcomes between low or ultralow and high opioid administration.

The most important finding in our study is that the use of an evidence-based multimodal analgesic strategy in the setting of cardiac surgery is associated with a reduction in the reliance on opioid-based anesthesia. The number of interventions employed—and, in particular, the use of preoperative gabapentin, intraoperative dexmedetomidine, and regional analgesia—was independently associated with low opioid administration. The 5 interventions were selected based on evidence of their efficacy in the general perioperative and/or cardiac surgery setting as well as their ability to be successfully incorporated in our local institutional service line. Preoperative and intraoperative acetaminophen (ie, oral and intravenous forms) has been studied extensively and more recently validated as an agent to facilitate opioid reduction, pain management, and reduce delirium in the setting of cardiac surgery.21,22 Preoperative gabapentin has been investigated to a similar extent and found to be efficacious in cardiac surgery as an opioid-sparing agent,23,24 despite some caution recommended based on a recent update that revealed an association between its use and respiratory depression.25 Ketamine, evaluated in major surgery for its potential role in the prevention of postoperative delirium,26 has also been shown to be useful in reducing intraoperative and postoperative opioids, particularly, when deployed as a “subhypnotic” infusion in our program.27,28 Dexmedetomidine, increasingly utilized in cardiac surgery for its ability to facilitate rapid postoperative recovery, early extubation, and potential delirium prevention has been investigated for its opioid-sparing characteristics as well.29,30 Finally, use of a “single-shot” thoracic regional nerve block has been described for numerous chest wall procedures, including minimally invasive thoracic, thoracotomy, and sternotomy incisions.31,32 In this case, the nerve block was typically applied at the end of the surgery and, therefore, after surgical incision and closure.

Based on our results, no significant differences were noted between low (or ultralow) and high (ie, traditional) opioid patients in postoperative outcomes. This suggests that low opioid administration was an insufficient predictor of early extubation. A prior publication from our group revealed that perioperative interventions were associated with increased rates of early extubation15—though the effect does not appear to be facilitated through opioid reduction alone. Although this stands in contrast to the findings of prior opioid reduction cardiac anesthetics,5,6 it may be that selecting short-acting, titratable, nonopioid strategies is as important an aspect of patient recovery as opioid reduction itself. One other compelling result of our study is the ability to achieve ultralow and even opioid-free cardiac anesthesia. While the approach has been described in several surgical subtypes,33,34 to our knowledge it has not been published in the setting of cardiac surgery aside from limited individual case reports.35,36 Regardless, the role for withholding opioid analgesia is generally unstudied and cautioned against, particularly if done so at the expense of pain control or in patients with chronic opioid use or opioid tolerance.37 Our study did not identify optimal candidates for ultralow or opioid-free cardiac anesthetics and additional prospective studies are necessary in this regard. As a result, our findings should not be interpreted as an endorsement of ultralow or opioid-free anesthesia, but rather a recognition of its feasibility.

Despite efforts, overall compliance with selected interventions was modest. There are many potential explanations for this finding, which include the fact that the protocol was issued as a guideline for practice rather than a mandate. In addition, it is possible that providers withheld medications based on personal preference or patient contraindication. Given the electronic medical record was utilized for data query, certain interventions may have been incorrectly or insufficiently documented. Regardless, we saw a steady increase in compliance throughout the study period. Despite the overall modest compliance, intraoperative opioid administration was reduced, a compelling finding, nonetheless. While our results support a strong association between composite interventions and intraoperative opioids, we cannot exclude the potential for either bias or confounding. Importantly, our study was not blinded, and providers were aware that the program was being monitored. This introduces the potential for the Hawthorne effect, which is a limitation of our results.

There are other important limitations to this study. Data presented are specific to a single tertiary academic institution and the existence of unmeasured factors, such as the organizational workforce, unit protocols, nursing, and provider workflows, may have impacted our findings. Regardless, it is likely that foundational principles such as the establishment of a multidisciplinary team, the use of evidence-based interventions, guideline development, and system auditing are transferrable. Whereas certain confounders may be addressed in part by propensity matching and multivariable regression analysis, we anticipate that results were potentially influenced by unmeasured provider bias. While the use of propensity score matching for statistical analysis of secondary outcomes likely produced a more robust model than an alternative approach (ie, inverse probability of treatment weighting38,39), this robustness may have been achieved at the cost of a reduction in sample size.40 We cannot exclude the potential for type II error. For example, we report an increased proportion of atrial fibrillation in both the low and ultralow groups, and although this difference did not reach statistical significance, we cannot exclude the impact of patient selection or insufficient power on this finding. In addition, given that this represents a secondary analysis, we are unable to determine which patients utilized opioids before surgery (ie, chronic opioids), which may have impacted the results. While we have targeted the use of several well-established nonopioid analgesics, we are unable to comment on the potential efficacy of other candidate analgesics (ie, intravenous lidocaine, esmolol, and/or other regional techniques). Our study was not designed to evaluate the entire postoperative stay or assess the proportion of persistent opioid use in our patient population. These likely represent opportunities for further evaluation. Finally, and perhaps most importantly, the results of this study should be interpreted with caution, as it is principally designed to establish the feasibility of opioid minimization in cardiac anesthesia. We recommend these results be viewed as hypothesis generating and recognize the importance of the extensive further investigation to determine the impact of opioid reduction on postoperative outcomes.

In conclusion, an institutional ERP for cardiac surgery program coupled with the use of nonopioid interventions resulted in the marked reduction in intraoperative opioid administration over the course of a 9-month study period without negatively impacting key postoperative outcomes. A further study designed to assess the impact of opioid minimization and opioid-free cardiac anesthetics on postoperative outcomes is key to understanding the intermediate or long-term implications of this practice.

ACKNOWLEDGMENTS

The authors acknowledge Dr Colleen Koch, Chair and Clinical Director of Johns Hopkins Anesthesiology and Critical Care Medicine, Dr James Abernathy, Director of Johns Hopkins Division of Cardiac Anesthesiology, and Dr Robert Higgins, Chair and Surgeon-in-Chief of Johns Hopkins Department of Surgery for their direction and support in program development and implementation.

DISCLOSURES

Name: Michael C. Grant, MD, MSE.

Contribution: This author helped conceive the study, assist in data collection, analyze the data, draft the manuscript, revise the manuscript, and approve the final manuscript.

Conflicts of Interest: M. C. Grant is on the Executive Board of the Enhanced Recovery After Surgery Cardiac Society.

Name: Tetsuro Isada, MD.

Contribution: This author helped in collect the data, draft the manuscript, revise the manuscript, and approve the final manuscript.

Conflicts of Interest: None.

Name: Pavel Ruzankin, PhD.

Contribution: This author helped analyze the data, draft the manuscript, revise the manuscript, and approve the final manuscript.

Conflicts of Interest: None.

Name: Allan Gottschalk, MD, PhD.

Contribution: This author helped analyze the data, draft the manuscript, revise the manuscript, and approve the final manuscript.

Conflicts of Interest: None.

Name: Glenn Whitman, MD.

Contribution: This author helped conceive the study, draft the manuscript, revise the manuscript, and approve the final manuscript.

Conflicts of Interest: None.

Name: Jennifer S. Lawton, MD.

Contribution: This author helped conceive the study, draft the manuscript, revise the manuscript, and approve the final manuscript.

Conflicts of Interest: None.

Name: Jeffrey Dodd-o, MD.

Contribution: This author helped conceive the study, analyze the data, draft the manuscript, revise the manuscript, and approve the final manuscript.

Conflicts of Interest: None.

Name: Viachaslau Barodka, MD.

Contribution: This author helped conceive the study, collect the data, analyze the data, draft the manuscript, revise the manuscript, and approve the final manuscript.

Conflicts of Interest: None.

This manuscript was handled by: Nikolaos J. Skubas, MD, DSc, FACC, FASE.

    REFERENCES

    1. Lowenstein E, Hallowell P, Levine FH, Daggett WM, Austen WG, Laver MB. Cardiovascular response to large doses of intravenous morphine in man. N Engl J Med. 1969;281:1389–1393.
    2. Hasbrouck JD. Morphine anesthesia for open-heart surgery. Ann Thorac Surg. 1970;10:364–369.
    3. Stoelting RK, Gibbs PS. Hemodynamic effects of morphine and morphine-nitrous oxide in valvular heart disease and coronary-artery disease. Anesthesiology. 1973;38:45–52.
    4. Stanley TH, Webster LR. Anesthetic requirements and cardiovascular effects of fentanyl-oxygen and fentanyl-diazepam-oxygen anesthesia in man. Anesth Analg. 1978;57:411–416.
    5. Myles PS, Daly DJ, Djaiani G, Lee A, Cheng DC. A systematic review of the safety and effectiveness of fast-track cardiac anesthesia. Anesthesiology. 2003;99:982–987.
    6. Silbert BS, Scott0 DA, Evered LA, et al. A comparison of the effect of high- and low-dose fentanyl on the incidence of postoperative cognitive dysfunction after coronary artery bypass surgery in the elderly. Anesthesiology. 2006;104:1137–1145.
    7. Angst MS, Clark JD. Opioid-induced hyperalgesia: a qualitative systematic review. Anesthesiology. 2006;104:570–587.
    8. Long DR, Lihn AL, Friedrich S, et al. Association between intraoperative opioid administration and 30-day readmission: a pre-specified analysis of registry data from a healthcare network in New England. Br J Anaesth. 2018;120:1090–1102.
    9. Brummett CM, Waljee JF, Goesling J, et al. New persistent opioid use after minor and major surgical procedures in US adults. JAMA Surg. 2017;152:e170504.
    10. Clement KC, Canner KJ, Lawton JS, et al. Predictors of new persistent opioid use after coronary artery bypass grafting. J Thorac Cardiovasc Surg. 2019 October 10 [Epub ahead of print].
    11. Brescia AA, Waljee JF, Hu HM, et al. Impact of prescribing on new persistent opioid use after cardiothoracic surgery. Ann Thorac Surg. 2019;108:1107–1113.
    12. STEPAct. Available at: https://www.mass.gov/files/documents/2017/01/xd/non-opioid-directive_0.pdf. Accessed May 28, 2019.
    13. Voluntary Nonopioid Directive. Available at: https://www.health.pa.gov/topics/Documents/Opioids/Pennsylvania%20Non-Opioid%20Directive.pdf. Accessed May 28, 2019.
    14. Engelman DT, Ben Ali W, Williams JB, et al. Guidelines for perioperative care in cardiac surgery: enhanced recovery after surgery society recommendations. JAMA Surg. 2019;154:755–766.
    15. Grant MC, Isada T, Ruzankin P. Johns Hopkins Enhanced Recovery Program for the Cardiac Surgery Working Group. Results from an enhanced recovery program for cardiac surgery. J Thorac Cardiovasc Surg. 2020;159:1393–1402.e7.
    16. Wu CL, Benson AR, Hobson DB, et al. Initiating an enhanced recovery pathway program: an anesthesiology department’s perspective. Jt Comm J Qual Patient Saf. 2015;41:447–456.
    17. Koponen T, Karttunen J, Musialowicz T, Pietiläinen L, Uusaro A, Lahtinen P. Vasoactive-inotropic score and the prediction of morbidity and mortality after cardiac surgery. Br J Anaesth. 2019;122:428–436.
    18. Ruzankin PS. A fast algorithm for maximal propensity score matching. Methodol Comput Appl Prob. 2020;22:477–495.
    19. Zhu Y, Coffman DL, Ghosh D. A boosting algorithm for estimating generalized propensity scores with continuous treatments. J Causal Inference. 2015;3:25–40.
    20. Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25:1–21.
    21. Jelacic S, Bollag L, Bowdle A, Rivat C, Cain KC, Richebe P. Intravenous acetaminophen as an adjunct0 analgesic in cardiac surgery reduces opioid consumption but not opioid-related adverse effects: a randomized controlled trial. J Cardiothorac Vasc Anesth. 2016;30:997–1004.
    22. Subramaniam B, Shankar P, Shaefi S, et al. Effect of intravenous acetaminophen vs placebo combined with propofol or dexmedetomidine on postoperative delirium among older patients following cardiac surgery: the DEXACET randomized clinical trial. JAMA. 2019;321:686–696.
    23. Ucak A, Onan B, Sen H, Selcuk I, Turan A, Yilmaz AT. The effects of gabapentin on acute and chronic postoperative pain after coronary artery bypass graft surgery. J Cardiothorac Vasc Anesth. 2011;25:824–829.
    24. Menda F, Köner O, Sayin M, Ergenoğlu M, Küçükaksu S, Aykaç B. Effects of single-dose gabapentin on postoperative pain and morphine consumption after cardiac surgery. J Cardiothorac Vasc Anesth. 2010;24:808–813.
    25. FDA. Drug Safety Communications. Available at: https://www.fda.gov/media/133681/download. Accessed April 27, 2020.
    26. Avidan MS, Maybrier HR, Abdallah AB. PODCAST Research Group. Intraoperative ketamine for prevention of postoperative delirium or pain after major surgery in older adults: an international, multicentre, double-blind, randomised clinical trial. Lancet. 2017;390:267–275.
    27. Lahtinen P, Kokki H, Hakala T, Hynynen M. S(+)-ketamine as an analgesic adjunct reduces opioid consumption after cardiac surgery. Anesth Analg. 2004;99:1295–1301.
    28. Mazzeffi M, Johnson K, Paciullo C. Ketamine in adult cardiac surgery and the cardiac surgery intensive care unit: an evidence-based clinical review. Ann Card Anaesth. 2015;18:202–209.
    29. Djaiani G, Silverton N, Fedorko L, et al. Dexmedetomidine versus propofol sedation reduces delirium after cardiac surgery: a randomized controlled trial. Anesthesiology. 2016;124:362–368.
    30. Habibi V, Kiabi FH, Sharifi H. The effect of dexmedetomidine on the acute pain after cardiothoracic surgeries: a systematic review. Braz J Cardiovasc Surg. 2018;33:404–417.
    31. Mittnacht AJC, Shariat A, Weiner MM, et al. Regional techniques for cardiac and cardiac-related procedures. J Cardiothorac Vasc Anesth. 2019;33:532–546.
    32. Kumar KN, Kalyane RN, Singh NG, et al. Efficacy of bilateral pectoralis nerve block for ultrafast tracking and postoperative pain management in cardiac surgery. Ann Card Anaesth. 2018;21:333–338.
    33. Frauenknecht J, Kirkham KR, Jacot-Guillarmod A, Albrecht E. Analgesic impact of intra-operative opioids vs. opioid-free anaesthesia: a systematic review and meta-analysis. Anaesthesia. 2019;74:651–662.
    34. Soffin EM, Wetmore DS, Beckman JD, et al. Opioid-free anesthesia within an enhanced recovery after surgery pathway for minimally invasive lumbar spine surgery: a retrospective matched cohort study. Neurosurg Focus. 2019;46:E8.
    35. Landry E, Burns S, Pelletier MP, Muehlschlegel JD. A successful opioid-free anesthetic in a patient undergoing cardiac surgery. J Cardiothorac Vasc Anesth. 2019;33:2517–2520.
    36. Chanowski EJP, Horn JL, Boyd JH, Tsui BCH, Brodt JL. Opioid-free ultra-fast-track on-pump coronary artery bypass grafting using erector spinae plane catheters. J Cardiothorac Vasc Anesth. 2019;33:1988–1990.
    37. Brandal D, Keller MS, Lee C, et al. Impact of enhanced recovery after surgery and opioid-free anesthesia on opioid prescriptions at discharge from the hospital: a historical-prospective study. Anesth Analg. 2017;125:1784–1792.
    38. Rubin DB. On principles for modeling propensity scores in medical research. Pharmacoepidemiol Drug Saf. 2004;13:855–857.
    39. Austin PC, Mamdani MM. A comparison of propensity score methods: a case-study estimating the effectiveness of post-AMI statin use. Stat Med. 2006;25:2084–2106.
    40. Schulte PJ, Mascha EJ. Propensity score methods: theory and practice for anesthesia research. Anesth Analg. 2018;127:1074–1084.

    Supplemental Digital Content

    Copyright © 2020 International Anesthesia Research Society