Persistent Opioid Use After Open Aortic Surgery: Risk Factors, Costs, and Consequences

Abstract

Background.

The incidence and financial impact of persistent opioid use (POU) after open aortic surgery is undefined.

Methods.

Insurance claim data from opioid-naïve patients who underwent aortic root replacement, ascending aortic replacement, or transverse arch replacement from 2011 to 2017 were evaluated. POU was defined as filling an opioid prescription in the perioperative period and between 90 and 180 days postoperatively. Postoperative opioid prescriptions, emergency department visits, readmissions, and health care costs were quantified. Multivariable logistic regression identified risk factors for POU, and quantile regression quantified the impact of POU on postoperative health care costs.

Results.

Among 3240 opioid-naïve patients undergoing open aortic surgery, 169 patients (5.2%) had POU. In the univariate analysis, patients with POU were prescribed more perioperative opioids (375 vs 225 morphine milligram equivalents, P < .001), had more emergency department visits (45.6% vs 25.4%, P < .001), and had significantly higher health care payments in the 6 months postoperatively ($10,947 vs $7223, P < .001). Independent risk factors for POU in the multivariable logistic regression included preoperative nicotine use and more opioids in the first perioperative prescription (all P < .05). After risk adjustment, POU was associated with a $2439 increase in total health care costs in the 6 months postoperatively.

Conclusions.

POU is a challenge after open aortic operations and can have longer-term impacts on health care payments and emergency department visits in the 6 months after surgery. Strategies to reduce outpatient opioid use after aortic surgery should be encouraged when feasible.

More than 46,800 people died from an opioid overdose in 2018 in the United States. However, the medical community’s new awareness of the opioid epidemic is starting to make an impact as prescription opioid-involved death rates decreased by 13.5% from 2017 to 2018.¹ In a multicenter study of 3412 patients undergoing 25 different elective operations, 61.5% of opioid pills were not used, and 31.4% of patients did not use any opioids at home.² Given concerns for opioid addiction, there is increased interest in developing opioid prescribing guidelines to standardize postoperative opioid prescriptions and minimize the risk of persistent opioid use (POU) after elective operations.^2,3

POU occurs when an opioid-naïve patient undergoes an operation, fills an opioid prescription in the perioperative time frame (30 days before surgery to 14 days after surgery) and continues to use opioids 90 to 180 days after the initial operation.⁴ In an analysis of 36,177 patients, POU occurred in 5.9% to 6.5% of patients undergoing major and minor surgical procedures.⁴ While many studies have documented the incidence of POU after elective surgical procedures, whether POU impacts patient outcomes is unclear. In cardiothoracic surgery, POU has been documented after lobectomy (14.0%–16.0%), coronary artery bypass grafting (CABG; 8.1%–13.0%), and valve procedures (5.5%–11.0%) and is associated with increased health care costs and emergency department visits up to 6 months after surgery.^5–8 However, it is unknown whether this equally impacts patients undergoing repair of the ascending or descending thoracic aorta to treat aneurysms and dissections. Furthermore, little is known about the outpatient opioid requirements of these patients when they are discharged home.

The purpose of this study is 3-fold. First, we will determine the amount of outpatient opioid prescriptions filled by patients who are discharged home after open thoracic aortic surgery. Second, we will calculate the frequency and determine the risk factors for POU. Third, we will determine whether POU has an association with emergency department visits, readmissions, and health care costs in the 6 months after surgery. This analysis will help determine whether POU after open aortic surgery is contributing to the opioid epidemic, and we will propose patient-centered methods to curtail opioid use while maintaining sufficient pain control and patient satisfaction.

Patients and Methods

Patient Cohort

The University of Texas Health Science Center Institutional Review Board determined that this study qualified for exempt status, and a waiver of informed consent was granted. Patients who underwent open thoracic aortic surgery from January 2012 to December 2017 were identified in the IBM MarketScan Commercial and Medicare Supplemental Databases (IBM, Armonk, NY).⁹ These databases include deidentified health care claim information from employers, health plans, and Medicare programs and span across the continuum of care (eg, inpatient, outpatient, and outpatient pharmacy). These databases do not include Medicaid or uninsured patients.

The following open aortic operations were identified using Current Procedural Terminology (American Medical Association, Chicago, IL) codes and were included in the study: aortic root replacement (33863), ascending aortic replacement (33860), and transverse arch graft (33870). Isolated endovascular aortic repairs were excluded in this analysis. A flowchart summarizing the patient cohort is included in Figure 1.

Figure 1.

Flowchart of the cohort analyzed in this study. Patients were excluded to facilitate accurate assessment of outcomes and calculation of outpatient opioid prescriptions. Of the 3240 patients analyzed in the cohort, persistent opioid use developed in 5.2% (n = 1690).

Patients were excluded if they died in the hospital or were discharged to hospice. Patients were only included if they were enrolled in MarketScan from 1 year before surgery to 6 months after surgery, but gaps in enrollment of less than 30 days were permitted. Patients also were required to have prescription drug coverage from 1 year before surgery to 6 months after surgery to determine their outpatient prescription opioid use. Because this study focuses on opioid-naïve patients, patients who filled an opioid prescription 365 to 31 days before surgery were excluded.

Basic demographic information was extracted from MarketScan. The Agency for Healthcare Research and Quality Clinical Classification Software was used to identify preoperative comorbidities by grouping International Classification of Diseases (ICD) codes 9 and 10 into clinically relevant diagnoses.¹⁰ The Charlson Co-morbidity Index was used to provide a global score for a patient’s preoperative risk and predicts in-hospital and 1-year mortality.¹¹ Perioperative nerve blocks and postoperative complications were identified using ICD-9 and ICD-10 codes. Readmissions and emergency department visits were also tabulated for each patient in the 6 months after discharge.

Opioid and Cost Data

Consistent with previous studies, the perioperative time frame spanned from 30 days before surgery to 14 days after surgery.^4,6 POU occurs when an opioid-naïve patient fills an opioid prescription in the perioperative time period and continues to fill at least 1 opioid prescription 90 to 180 days after surgery. These definitions were used to be consistent with prior work in the field and to permit comparisons between studies.^4,6,7,12 Opioid prescriptions were converted to morphine milligram equivalents (MMEs).

Health care payments for the index admission, outpatient care, and pharmacy claims were tabulated directly from MarketScan for each patient. These cost data are actual payments for services rendered and include payments made by insurance companies and by the patient.

Statistics

Stata software (StataCorp, College Station, TX) was used for the statistical analysis.¹³ Categorical variables are shown as the count of patients (percentage) and were compared with the Pearson χ² test. Continuous variables are shown as medians (interquartile range) and were compared with the Mann-Whitney U test. Variables were included in the multivariable logistic regression dependent on statistical significance (P < .10) in the univariate analysis and clinical significance. Quantile regression was used to model the impact of POU and perioperative variables on median total health care costs from after the index admission to 6 months postoperatively.

Results

Among 3240 patients, 169 (5.2%) had POU. In the univariate analysis, patients with POU were younger (56 vs 58 years, P = .04), earned less annual income ($46,438 vs $48,964, P = .012), and were more likely to use nicotine (17.8% vs 8.5%, P < .001) compared with non-POU patients (Supplemental Table 1). Furthermore, there were differences in region of residence in the univariate analysis, but this was not statistically significant in the multivariable regression. Surprisingly, prior illicit drug use was not found to be risk factor for POU (P = .85). In the perioperative phase, POU and non-POU patients had no significant differences in receiving perioperative regional anesthesia, postoperative complications, length of stay, or discharge location (all P > .05) (Table 1). However, POU patients were more likely to visit the emergency department at least once (45.6% vs 25.4%, P < .001) and to have multiple emergency department visits (24.9% vs 12.9%, P < .001) compared with non-POU patients in the 6 months after surgery (Table 1). The incidence of POU from 2012 to 2017 decreased over time, but this was not statistically significant (Supplemental Figure 1).

Table 1.

Perioperative Information for Patients Undergoing Aortic Surgery

Variable	All Patients (N = 3240)	Persistent Opioid Use (n = 169)	No Persistent Opioid Use (n = 3071)	P Value
Perioperative epidural or regional nerve block	0 (0)	0 (0)	0 (0)	>.99
Postoperative complications
Pneumonia	85 (2.6)	5 (2.9)	80 (2.6)	.78
Acute kidney injury	189 (5.8)	11 (6.5)	178 (5.8)	.70
Stroke	80 (2.5)	4 (2.4)	76 (2.5)	.93
Surgical site infection	6 (0.2)	1 (0.6)	5 (0.2)	.21
Myocardial infarction	119 (3.7)	8 (4.7)	111 (3.6)	.45
Return to operating room for sternal débridement	1 (.03)	0 (0)	1 (0.03)	.82
Discharge metrics
Length of surgical admission, d	6 (5–9)	6 (5–9)	6 (5–9)	.42
Discharge status				.77
Home	1909 (61.3)	94 (58.0)	1815 (61.5)
Home with home health	903 (29.0)	56 (34.6)	847 (28.7)
Skilled nursing facility	138 (4.4)	4 (2.5)	134 (4.5)
Inpatient rehabilitation	90 (2.9)	7 (4.3)	83 (2.8)
Other	200 (6.2)	8 (4.7)	192 (6.3)
Emergency department visits and readmissions ≤6 months after surgery
Any emergency department visit after surgery	857 (26.5)	77 (45.6)	780 (25.4)	<.001
Number of emergency department visits
0	2241 (69.2)	90 (53.3)	2151 (70.0)	<.001
1	575 (17.8)	40 (22.5)	538 (17.5)
2	253 (7.8)	24 (14.2)	229 (7.5)
3	171 (5.3)	18 (10.7)	153 (4.9)
Readmission	487 (15.0)	30 (17.8)	457 (14.9)	.31

Categorical data are presented as n (%) and continuous data as the median (interquartile range).

Patients with POU also filled opioid prescriptions for more MMEs in the perioperative period (375 vs 225 MMEs, P < .001) and continued to do so in the 90 to 180 days after surgery (225 vs 0 MMEs, P < .001), respectively (Supplemental Table 2). Patients with POU were more likely to fill an opioid prescription than patients without POU during all time points examined up to a year after discharge (all P < .001). In the year after surgery, 11.8% of patients filled at least 1 opioid prescription, and POU patients were almost 3-fold more likely to fill opioids than non-POU patients (33.7% vs 13.9%, P < .001). The relationship between the frequency of POU and the amount of opioids prescribed in the first perioperative prescription is shown in Supplemental Figure 2.

In the multivariable logistic regression, nicotine use (odds ratio, 2.09; 95% confidence interval [CI], 1.35–3.25; P = .001) and more opioids prescribed in the initial opioid prescription (odds ratio, 1.008; 95% CI, 1.003–1.013; P = .002) were independent predictors of POU (Table 2). While the odds ratio for the initial opioid prescription is small, this reflects a 10-MME increase in opioids prescribed and is equivalent to 2 tablets of 5 mg hydrocodone.

Table 2.

Multivariable Logistic Regression of Predictors of Persistent Opioid Use After Aortic Surgery

		95% CI
Variable	Odds Ratio	Lower	Upper	P Value
Age (by 10-year increments)	0.87	0.74	1.03	.11
Male gender	1.37	0.97	1.94	.074
Annual income (by $10,000 increments)	0.82	0.61	1.10	.18
Surgery type (aortic root replacement is reference)
Ascending aortic graft	1.14	0.80	1.62	.45
Transverse arch graft	0.33	0.77	1.39	.13
Multiple concomitant open aortic procedures	0.52	0.25	1.09	.08
Region
North Central	1.19	0.70	2.02	.51
South	1.61	0.96	2.68	.069
West	1.13	0.63	2.04	.68
Medicare	1.22	0.73	2.06	.44
Nicotine use	2.09	1.35	3.25	.001
Illicit drug use	0.78	0.10	5.97	.91
Concomitant aortic surgery and valve replacement/repair	0.77	0.55	1.07	.11
Opioids prescribed in initial prescription (by 10 MMEs)	1.008	1.003	1.013	.002

Bolded variables have a P value of <.05.

CI, confidence interval; MME, morphine milligram equivalents.

Patients with and without POU had no significant differences in health care payments in the 6 months before surgery and for the index admission (all P > .05) (Supplemental Table 3). However, POU patients had significantly more outpatient, pharmacy, and total health care payments in the 6 months after surgery (all P < .05), with a notable $3274 increase in total health care payments ($10,947 vs $7,233, P < .001) compared with non-POU patients, respectively. After risk adjustment in the quantile regression, POU was independently associated with a $2439 increase in health care payments ($2439; 95% CI $805-$4073; P = .003) (Supplemental Table 4). The other variables associated with a statistically significant impact on health care payments are provided in Supplemental Table 4 and shown graphically in Figure 2. These results are summarized in our graphical abstract.

Figure 2.

Quantile regression to model total health care payments after the index admission to 6 months postoperatively. Only variables with a P value of <.05 were included in the figure, but the full results in table format are available in Supplemental Table 4. (U.S., United States.)

Comment

In summary, POU use occurs in 5.2% of open thoracic aortic surgery patients. Patients with POU receive substantially more opioids in the perioperative period and continue to fill more opioid prescriptions up to 1 year after their operation compared with patients without POU. Furthermore, patients with POU have significantly more emergency department visits and postoperative health care payments in the 6 months after their initial operation. We also identified the following risk factors for POU, which include use of nicotine products and more opioids in the first perioperative prescription. Daily and intermittent smokers are 3-times more likely to report nonmedical prescription opioid use than nonsmokers,¹⁴ and smoking is a risk factor for POU in this work and our previous study in patients after CABG.⁶

Although multiple studies have reported the frequency of POU after CABG, valve procedures, and thoracic surgery, we present a large analysis of patients who underwent open thoracic aortic surgery. Brescia and colleagues¹² analyzed POU in 24,549 Medicare patients who underwent CABG, valve surgery, and thoracic surgery and determined that 12.9% had POU across all the surgical subtypes.¹² However, their analysis did not include aortic surgery patients and had a significantly older patient cohort, with a median age of 71 years compared with 58 years in our analysis.

In a larger analysis of 35,817 patients in the Optum database (Optum, Eden Prairie, MN), 10.2% of CABG and 8.1% of valve patients had POU, but the standard definition of POU established by Brummett and colleagues⁴ was not used in this study.⁸ Dewan and colleagues¹⁵ compared outcomes between patients with and without an opioid use disorder and included aortic surgery patients, but only 1.3% of their cohort underwent open aortic surgery. Thus, this present study from a large population of aortic surgery patients across different insurance types uniquely contributes to this rapidly growing field of knowledge.

Surprisingly, the present study did not show a significant increase in POU in the 83% of patients in the cohort presenting with an aortic dissection. Whereas operations for acute aortic dissections are associated with high rates of postoperative depression and posttraumatic stress disorder,^16,17 there was no significant difference in the incidence of POU in patients who underwent operations for an aortic dissection (P = .23).

While no patients in our cohort received regional nerve blocks, regional anesthesia techniques, or intercostal nerve cryoablation, nonopioid adjuncts effectively treat pain, reduce inpatient opioid use, and may be a solution to reduce POU after aortic surgery.^18–21 In the vascular surgery literature, combined epidural-general anesthesia for elective aortic aneurysm repair improves pain control with less in-hospital opioid use, reduces postoperative complications, and improves 5-year survival.^22–25 A recent single-center study of thoracic epidural anesthesia showed improved pain control with no increased complications after thoracoabdominal aortic aneurysm repair, but the study did not evaluate 30-day mortality.¹⁸

Alternatively, intercostal nerve cryoablation may be a better adjunct for pain control because it lasts for several weeks and minimizes the procedural and hemodynamic complications of epidural catheters.¹⁹ In a study of 117 patients undergoing open thoracic or thoracoabdominal aortic aneurysm repair, the intercostal nerve cryoablation group used significantly less daily opioids (28 daily MMEs, P = .005), had equivalent pain scores, and had no significant differences in complications compared with the noncryoablation group.²⁰

Oral adjuncts like pregabalin may also play a role, and a recent double-blind placebo-controlled study in the United Kingdom demonstrated that 14 days of postoperative pregabalin significantly reduced 24-hour opioid use, 24-hour pain scores, and chronic postoperative pain at 3 and 6 months compared with standard postoperative pain control.²¹

Enhanced recovery after cardiac surgery programs in the United States have effectively used multimodal adjuncts to reduce intraoperative and inpatient opioid use after surgery, but it is unclear whether these protocols have an impact on outpatient opioid use and the incidence of POU.^26,27 After reviewing the literature, we recommend using intercostal nerve cryoablation for thoracotomy and scheduled acetaminophen, gabapentin, and opioid-sparing analgesia to optimize patient pain control in the perioperative period and reduce outpatient opioid use.

The next challenge is how to tailor the opioid prescriptions on discharge and minimize the downstream effects of POU. Hill and colleagues,^28,29 from Dartmouth-Hitchcock Medical Center, have described the wide variety and excess of opioid prescribing in general surgery and demonstrated that a simple educational intervention for providers reduced the amount of outpatient opioids prescribed by 53%. Furthermore, their team has provided opioid prescribing guidelines for elective general surgery procedures and demonstrated that prescribing fewer opioids does not negatively affect patient satisfaction scores.³⁰ “Ultrarestrictive” opioid prescribing guidelines have also been implemented in other surgical fields with no significant effect on postoperative complications, calls for refills, and pain scores in clinic.³¹

However, a more personalized, patient-centered way to prescribe opioids is likely the best option. A patient’s opioid use in the 24 hours before discharge is the best predictor of a patient’s opioid requirement at home, and we recommend the following guidelines proposed by Hill and colleagues.³ Patients who needed no opioids on the day before discharge go home without an opioid prescription. Patients who need 1 to 3 opioid pills or more than 4 pills on the day before discharge should be discharged with 15 or 30 opioid pills, respectively. Routine telemedicine follow-up for these patients should also include questions regarding opioid refills to help identify these patients at high risk for POU.

This study has certain limitations that may affect our results. MarketScan is not a cardiothoracic-specific data set such as The Society of Thoracic Surgeons Cardiac Surgery Database, and thus, data regarding the acuity or chronicity of the aortic dissection, cannulation strategy, use of hypothermic circulatory arrest, and surgical technique are not available. Data regarding certain postoperative complications, such as prolonged ventilation, deep sternal wound infection, and 30-day mortality, cannot be reliably obtained with the MarketScan data, and we cannot accurately quantify the amount of inpatient opioids used during the index admission. Because uninsured and Medicaid patients are not included in the MarketScan databases, our analysis may exclude groups with lower socioeconomic status, and these groups may have a higher risk of POU.

Furthermore, the quantity of opioids prescribed is not equivalent to the amount taken by the patient at home, and this study cannot account for opioids given to family members, friends, or discarded at home after the prescription is filled. We also excluded patients with preoperative opioid use, but whether this group needs special attention and monitoring for the development of POU after surgery is unclear.

MarketScan also does not include provider-specific data, and thus we are unable to link a subset of providers to higher opioid prescribing.

In conclusion, persistent opioid use after open thoracic aortic operations is a clinically relevant problem that affects emergency department visits and health care payments up to 6 months after surgery. A prospective study to determine how many opioid pills are taken by cardiothoracic surgery patients is needed to help providers better understand how to best prescribe these medications and minimize POU. Until these prospective studies are completed, we recommend using multimodal analgesia and using the amount of opioids needed on the day before discharge to guide opioid prescribing for discharging patients home. Given the $2439 increase in direct health care payments and increased emergency department visits linked to POU, the negative consequences of POU should incentivize hospitals, insurance companies, and providers to develop constructive solutions for this problem.