Perioperative Factors Associated With Chronic Opioid Use After Spine Surgery

Abstract

Study Design:

Retrospective case control.

Objectives:

The purpose of the current study is to determine risk factors associated with chronic opioid use after spine surgery.

Methods:

In our single institution retrospective study, 1,299 patients undergoing elective spine surgery at a tertiary academic medical center between January 2010 and August 2017 were enrolled into a prospectively collected registry. Patients were dichotomized based on renewal of, or active opioid prescription at 3-mo and 12-mo postoperatively. The primary outcome measures were risk factors for opioid renewal 3-months and 12-months postoperatively. These primarily included demographic characteristics, operative variables, and in-hospital opioid consumption via morphine milligram equivalence (MME). At the 3-month and 12-month periods, we analyzed the aforementioned covariates with multivariate followed by bivariate regression analyses.

Results:

Multivariate and bivariate analyses revealed that script renewal at 3 months was associated with black race (P = 0.001), preoperative narcotic (P < 0.001) or anxiety/depression medication use (P = 0.002), and intraoperative long lumbar (P < 0.001) or thoracic spine surgery (P < 0.001). Lower patient income was also a risk factor for script renewal (P = 0.01). Script renewal at 12 months was associated with younger age (P = 0.006), preoperative narcotics use (P = 0.001), and ≥4 levels of lumbar fusion (P < 0.001). Renewals at 3-mo and 12-mo had no association with MME given during the hospital stay or with the usage of PCA (P > 0.05).

Conclusion:

The current study describes multiple patient-level factors associated with chronic opioid use. Notably, no metric of perioperative opioid utilization was directly associated with chronic opioid use after multivariate analysis.

Introduction

Patients undergoing spinal surgery frequently require opioid pain medications perioperatively to control acute pain. However, the efficacy of long-term opioid treatment for pain management is controversial and often ineffective.^1-3 A subset of patients remain on these medications for extended periods after surgery, leading to chronic opioid dependence, addiction, and/or hyperalgesia.^4-7 In particular, patients who undergo orthopedic or neurological surgery are at higher risk of prolonged opioid use. ⁸ The relative abundance of prescription opioids has played a significant role in the opioid epidemic in the United States. In 2018, prescription opioids accounted for nearly one-third of all opioid overdose-related deaths. ⁹ Although opioids are effective in reducing acute postoperative pain, a comprehensive understanding of risk factors for long-term opioid use is vital to lessen the role of prescription opioids in driving the epidemic. Early identification of at-risk patients and intervention with multimodal pain programs for these patients have been shown to mitigate long-term opioid consumption.^10-12

Several risk factors for chronic opioid use after spinal surgery have been identified. Factors that have been consistently associated with prolonged postoperative opioid use after spine surgery include sustained preoperative opioid use, age, increased invasiveness of surgery, comorbid psychiatric disorders, and perioperative pain scores.^6,7,13-26 Studies have also found differences in chronic postoperative opioid use patterns between race, gender, different socioeconomic classes and geographic areas in the United States.^27-29 However, most literature on prolonged opioid use after spine surgery utilizes national or insurance databases, limiting granularity of the data.^19-23,27-30 There is also scant data on the association between in-hospital postoperative opioid consumption and protracted opioid use.

The aim of our study is to determine predictors of extended postoperative opioid use in a large single-institution sample of patients undergoing spinal surgery. Given the design and size of our study, we may have greater ability to detect new risk factors and confirm ones found through national databases. Our findings may help guide risk stratification for spine surgeons deciding which patient are at risk for chronic opioid use.

Methods

Patient Selection

We conducted a single-institution retrospective case-control study of all consecutively treated patients between January 2010 and August 2017. All patients included in our study were 18 or older and underwent cervical, thoracic, or lumbar spinal surgery including discectomy, laminectomy, fusion, foraminotomy, disc replacement, deformity correction, hardware replacement, corpectomy or combinations thereof. The approach toward treating postoperative pain at our institution is well standardized as part of the ERAS pathways and involves a multidisciplinary team that follows the patient before and after surgery. This includes Physical Therapy, Pain Management (including postoperative trigger point injections), as well as psychiatry. This cooperative effort is geared toward minimizing the patient need for oral narcotics medication. Demographic variables, operative details, in-hospital opioid consumption data, and outcome measures were collected from a prospectively maintained database. Our study was approved by the institutional ethical review board (IRB #2019-0947) and was exempt from having to obtain informed consent.

Definition of Variables

Enrolled patients were dichotomized based on the presence or renewal of any active opioid script at 3-months and 12-months postoperatively. Demographic, operative, and in-hospital data was collected to analyze association with opioid use at the 3-mo and 12-mo time frames.

Demographic factors included age, gender, average income of zip code, preoperative use of opioids, psychiatric medication use, use of benzodiazepines, race, alcohol use, tobacco use, and drug use. IRS 2018 income tax statistics were used to infer average income of each patient through their most current zip code. ²⁷ Preoperative use of opioids was defined as any active opioid script within 90 days of planned surgery date. Opioid medications included buprenorphine, codeine, fentanyl, hydrocodone, hydromorphone, meperidine, methadone, morphine, oxycodone, pentazocine, tapentadol, and tramadol. Psychiatric medications were any active medications to treat depression or anxiety.

Operative factors included surgical approach, presence of spinal fusion, and epidural for pain control. Surgical approaches were divided into anterior cervical, posterior cervical, isolated thoracic, lumbar short, and lumbar long. Lumbar long procedures were defined as thoracolumbar fusion constructs of ≥4 vertebral levels.

In-hospital data included length of stay (LOS), total morphine milligram equivalents (MME) of opioid medications throughout the hospital stay, MME per hospital day, use of patient-controlled analgesia (PCA), and MME given through PCA only. All opioid medications administered through any route including PCA throughout the hospital stay were combined and converted to MME. MME given through PCA was considered separately in addition. MME was divided by LOS to calculate MME per day. MME were calculated based on current CDC guidelines for opioid prescription practices. ⁹

Statistical Analysis

Descriptive statistics were summarized for all variables as mean and standard deviations (SD) for continuous variables and percentages for categorical variables. Primary outcome variables were opioid script renewal at 3- and 12-months post-surgery. Data were compared at the 3-mo and 12-mo time points between those who renewed an opioid script and those who did not (control). Multivariate logistic regression was completed using an ordinary least squares model. Unstandardized beta coefficients were calculated with 95% confidence intervals for each covariate. Inferential statistics for bivariate analyses were completed using T-test, Wilcoxon Rank Sum test, and Chi Square test as appropriate for variable type. A confidence level of 95% was used throughout the study. All statistical analyses were completed using Python Scipy. ³¹

Results

A total of 1299 patients were included in the analysis. Multivariate logistic regression for script renewal at 3 months post-surgery (Table 1) was associated with black race (P < 0.001) and inversely with income (P = 0.013). Notable variables that did not meet statistical significance were pre-operative usage of benzodiazepines (P = 0.071), ethanol (P = 0.076), tobacco (P = 0.093), and anxiety or depression medications (P = 0.122), as well as usage of PCA (P = 0.107) and long lumbar surgery type (P = 0.106). There was no association with gender, substance usage, white or other race, and length of stay. During the duration of the hospital stay, total MME, PCA, and epidural were not associated with script renewal at 3 months, as well as multiple surgery types including fusion, anterior cervical, posterior cervical, thoracic, and short lumbar.

Table 1.

Multivariate Analysis of Opioid Script Renewal at 3 Months Postoperatively.^a

Variable	Renewed (n = 339)	Not renewed (n = 856)	Coefficient	95% CI	P-value
Gender (female)	51.6%(n = 175)	55.1%(n = 472)	−0.0153	−0.067-0.036	0.562
Age (Years)	59.7 (13.9)	60.4 (14.4)	0.0002	−0.002-0.002	0.856
Income	$76,951 ($47,733)	$91,584 ($72,669)	−4.878e-7	−8.72e-7 to −1.04e-7	0.013
Length of stay (weeks)	3.36 (4.78)	2.11 (3.12)	0.0040	−0.005-0.13	0.96
Total MME	302.9 (839.6)	145.5 (337.8)	2.149e-5	−3.54e-5 to 7.84e-5	0.459
PCA MME	155.7 (162.4)	97.2 (120.8)	0.0002	−4.74e-5 to 0.00	0.107
PCA (% yes)	59.9% (n = 203)	54.8% (n = 469)	−0.0304	−0.094-0.033	0.344
Epidural (% yes)	1.8% (n = 6)	0.7% (n = 6)	0.1132	−0.138-0.364	0.376
Fusion (% yes)	59.9% (n = 203)	51.1% (n = 437)	0.0169	−0.050-0.084	0.622
Surgery: Anterior cervical	13.1% (n = 44)	16.7% (n = 141)	−0.0925	−0.324-0.139	0.434
Surgery: Posterior cervical	12.8% (n = 43)	7.6% (n = 64)	0.0467	−0.200-0.293	0.711
Surgery: Thoracic	6.0% (n = 20)	2.4% (n = 20)	0.116	−0.167-0.399	0.422
Surgery: Lumbar (short)	56.8% (n = 191)	70.5% (n = 596)	−0.0821	−0.321-0.157	0.500
Surgery: Lumbar (long)	11.3% (n = 28)	2.8% (n = 24)	0.2231	−0.047-0.493	0.106
EtOH	51.3% (n = 174)	53.5% (n = 458)	0.0256	−0.025-0.076	0.076
Tobacco (current)	8.8% (n = 30)	7.8% (n = 67)	0.0014	−0.090-0.093	0.093
Recreational Drugs (current)	1.2% (n = 4)	0.8% (n = 7)	−0.0817	−0.342-0.179	0.179
Anxiety/Depression medication	54.0% (n = 162)	42.2% (n = 363)	0.0581	−0.006-0.122	0.122
Benzodiazepines	22.4% (n = 76)	16.5% (n = 141)	−0.0034	−0.078-0.071	0.071
Race (Black)	11.8% (n = 40)	5.8% (n = 50)	0.1769	0.081-0.273	0.0001
Race (White)	75.2% (n = 255)	81.4% (n = 697)	0.0030	−0.077-0.083	0.942
Race (Other)	13% (n = 44)	12.7% (n = 109)	0.0327	−0.051-0.117	0.444

^aContinuous variables reported as mean (standard deviation). Unstandardized beta coefficients listed in the coefficient column.

Script renewal at 12 months post-op (Table 2) was associated with preoperative age (P = 0.005) and long lumbar surgery (P = 0.021). Usage of PCA was greater in the renewal group but did not meet statistical significance. Other race was inversely associated with renewal (P = 0.018). Script renewal at 12 months did not associate with black or white race, gender, income, length of stay, or the remainder of the surgical types characterized. It was also not associated with preoperative tobacco, substance, anxiety/depression medication or benzodiazepine usage, total MME, receipt of PCA or MME via PCA, or epidural.

Table 2.

Multivariate Analysis of Script Renewal at 12 Months Postoperatively.^a

Variable	Renewed (n = 122)	Not renewed (n = 1073)	Coefficient	95% CI	P-value
Gender (female)	50.0% (n = 61)	54.6% (n = 586)	0.0054	−0.030-0.041	0.767
Age (Years(SD))	64.3 (13.1)	59.8 (14.3)	0.0018	0.001-0.003	0.005
Income	$88,453 ($58,348)	$87,155 ($67,590)	5.05e-8	−2.15e-7 to 3.16e-7	0.709
Length of stay	2.78 (2.76)	2.43 (3.80)	−0.0017	−0.008-0.005	0.593
Total MME	225.3 (316.1)	186.1 (554.5)	−7.83e-6	−4.72e-5 to 3.16e-5	0.697
PCA MME	150.2 (157.4)	110.5 (134.0)	0.0001	−5.73e-5 to 0.00	0.176
PCA (% yes)	56.6% (n = 69)	56.2% (n = 603)	−0.0203	−0.064-0.023	0.362
Epidural	1.6% (n = 2)	0.9% (n = 10)	−0.0026	−0.138-0.364	0.376
Fusion	54.9% (n = 67)	53.4% (n = 573)	0.0019	−0.048-0.045	0.937
Surgery: Anterior cervical	9.1% (n = 11)	16.4% (n = 174)	−0.0096	−0.170-0.151	0.906
Surgery: Posterior cervical	8.3% (n = 10)	9.2% (n = 97)	0.0365	−0.134-0.207	0.675
Surgery: Thoracic	2.5% (n = 3)	3.5% (n = 3.7)	0.0110	−0.185-0.207	0.912
Surgery: Lumbar (short)	64.5% (n = 78)	66.9% (n = 709)	0.0126	−0.153-0.178	0.881
Surgery: Lumbar (long)	15.7% (n = 19)	4.1% (n = 43)	0.2200	0.033-0.407	0.021
EtOH	53.3% (n = 65)	52.8% (n = 567)	0.0081	−0.027-0.043	0.652
Tobacco (current)	6.6% (n = 8)	8.3% (n = 89)	−0.0161	−0.079-0.047	0.618
Recreational drugs (current)	0.8% (n = 1)	0.9% (n = 10)	−0.0410	−0.221-0.139	0.656
Anxiety/Depression medication	54.9% (n = 67)	44.5% (n = 477)	0.0156	−0.029-0.060	0.488
Benzodiazepines	25.4% (n = 31)	17.3% (n = 186)	0.0310	−0.021-0.083	0.240
Race (Black)	11.5% (n = 14)	7.1% (n = 76)	0.0438	−0.023-0.110	0.197
Race (White)	83.6% (n = 102)	79.2% (n = 850)	−0.0208	−0.076-0.035	0.460
Race (Other)	4.9% (n = 6)	13.8% (n = 147)	−0.0702	−0.128 to −0.012	0.018

^aContinuous variables reported as mean (standard deviation). Unstandardized beta coefficients listed in the coefficient column.

Bivariate logistic regression revealed associations for script renewal at 3 months (Table 3) with income (P = 0.010) and Black race (P = 0.001), as well as preoperative narcotics usage (P < 0.001) and anxiety or depression diagnosis (P = 0.002). Surgery type was also associated with renewal at 3 months, specifically long lumbar (P < 0.001), posterior cervical (P < 0.001) or thoracic (P < 0.001). Notably, 3-month script renewal was associated with greater total MME after bivariate analysis but not after multivariate analysis, likely due to outlying data (total MME range: 0-14 357). Bivariate analysis for script renewal at 12 months (Table 4) yielded significant associations with age (P = 0.006), non-black or white race (P = 0.021), preoperative narcotics usage (P = 0.001), and long lumbar surgery (P < 0.001).

Table 3.

Bivariate Analysis of Covariates Associated With Script Renewal at 3 Months.^a

Variable	Coefficient	95% CI	P-value
Income	−4.96e-7	−8.74e-7 to −1.18e-7	0.010b
Race (Black)	0.1548	0.062-0.248	0.001b
Preop Narcotics	0.1507	0.101-0.200	0.0001b
Surgery: Lumbar (long)	0.3692	0.031-0.129	0.0001b
Surgery: Thoracic	0.2599	0.124-0.396	0.0001b
Surgery: Posterior cervical	0.1654	0.081-0.250	0.0001b
Anxiety/Depression medication	0.800	0.031-0.129	0.002b

^a Coefficients are unstandardized beta coefficients.

^b =P < 0.05.

Table 4.

Bivariate Analysis of Covariates Associated With Script Renewal at 12 Months.

Variable	Coefficient	95% CI	P-value
Age	0.0017	0.000-0.003	0.006b
Race (Other)	−0.0597	−0.110 to −0.009	0.021b
Preop Narcotics	0.0582	0.024-0.092	0.001b
Surgery: Lumbar (long)	0.2092	0.133-0.285	0.0001b

^a Coefficients are unstandardized beta coefficients.

^b =P < 0.05.

Discussion

The current opioid epidemic has been one of the worst health calamities in recent history. The Center for Disease Control states that opioid overdoses alone accounted for nearly 50,000 deaths in 2018, with prescription opioids accounting for one-third of such deaths. ⁹ Given the utility of opioids in managing acute perioperative pain associated with surgeries, surgeons must find ways to mitigate postoperative opioid abuse. Our study analyzes perioperative factors associated with protracted opioid use after spine surgery, providing granularity to patient-specific variables. Beyond one’s history of opioid usage, a detailed understanding of the patient’s risk profile will be vital to determine who is most likely to suffer from chronic opioid use and who may benefit from analgesic interventions.

The current study found multiple factors associated with prolonged opioid use after surgery. Controlling for baseline variables through multivariate regression, lower income, race (black), preoperative opioid usage, preoperative use of anxiety or depression medication, and multiple types of surgical procedures were associated with opioid script renewal at 3 months post-surgery. At 12 months, script renewal was associated with older age, race (non-black or white), preoperative opioid usage, or more invasive lumbar surgery (≥4 levels of fusion). Although our study did not track visual analog pain scores, most patients requiring renewal of their narcotics prescription had a pain rating of greater than 7. Many of these findings corroborate previously noted associations for chronic opioid use in spine surgery, including preoperative opioid use and surgical invasiveness, but a few observations of note arose.^6,7

First, our study uncovered a subtle but important delineation in the effects of pre- and perioperative opioid use on the risk of long-term maladaptive opioid use after spine surgery. Confirming previous literature, we found that preoperative opioid use was a significant predictor of chronic opioid use postoperatively. However, our data signify that opioid consumption during the inpatient perioperative period is not associated with higher risk chronic opioid use. Although some metrics of opioid consumption, including MME consumed per day, total MME, or usage of an epidural or PCA, yielded significant associations after univariate analysis, none remained significant after controlling for other patient-specific variables through multivariate analysis (Figure 1). This is an important distinction that few studies have captured previously in the spine surgery literature.

Figure 1.

Perioperative opioid consumption compared between groups with and without chronic opioid use. *=P < 0.05.

In a single institution study, David et al found that patients undergoing transforaminal lumbar interbody fusion surgery who received greater than 500 total inpatient MME were at higher risk for chronic opioid use than those who received less than 250 MME. ³² However, the authors did not quantify the daily rate of opioid consumption. One recent study on daily inpatient opioid use by Sanford et al demonstrated a positive correlation between the daily opioids received during the hospital stay with discharge opioid prescription dosage. ³³ However, the authors did not analyze the risk for follow-up chronic opioid use. Comparatively, our study finds that, after controlling for the baseline variables, none of the inpatient opioid consumption metrics maintain a significant relationship with chronic opioid use. This finding suggests that spine surgeons can treat acute postoperative pain without marked concern that in-hospital usage of opioids will increase long-term risk for chronic opioid use.

Patient income was inversely associated with renewal at 3 months while older age was associated with renewal at 12 months. The association with income highlights the impact of living in resource-poor areas on chronic opioid use, particularly in the near-term post-surgery. Although older age may be an intuitive correlate to reduced ability to recover and thus greater dependence on opioids, Clarke et al found that younger age is at greater risk of chronic opioid use. ⁷ It should be noted, however, that their study is in an age-restricted cohort, where the young group is age 66-75. African Americans were found to be at greater risk for renewal at 3 months, whereas other races (non-black or white) were at decreased risk for renewal at 12 months. Given the study location in a region with a substantial Hispanic population, the current findings are particularly relevant for informing physician practice in the area.

Prior studies have attempted to capture the pre- and postoperative factors associated with prolonged opioid use in spine surgery. In a study of 8,377 lumbar fusion patients drawn from a commercial insurance database, preop opioid usage, re-fusion surgery, and preop depression were associated with chronic opioid use. ²² Fatemi et al analyzed 25,506 surgically-treated low back pain patients from a commercial database and found that patients with high-frequency prescriptions were more likely to continue opioid use than those with low frequency. ²⁵ Reviewing 13,257 insurance database patients, Adogwa et al found that, after controlling for baseline variables, female sex (95% CI: 1.058-1.237), obesity (95% CI: 1.004-1.212), and preoperative narcotic usage (95% CI 3.179-3.708) were directly associated with chronic opioid use. The most consistent variable associated with chronic postop opioid use is preoperative narcotics use, maintaining significance at the meta-analytic level of evidence.^13,34-36

A commonality of the methodology with many previous studies is that their samples are from insurance databases, thus lacking patient-level granularity. A few studies at the single- or oligo-center level have confirmed the above findings while also revealing more nuanced associations.^6,34,37-39 In a study of 583 consecutive spine surgery patients at a single center, Armaghani et al found preoperative opioid use, along with surgical invasiveness, anxiety, and revision surgery, to be directly associated with chronic postop opioid use. ⁶ Wright et al reviewed 819 patients undergoing discectomy or laminectomy at a single institution and found preoperative depression, preoperative opioid and benzodiazepine use, and greater discharge morphine equivalent dosage to be associated with chronic opioid use. ³⁹ In a series of oligo-center studies, Karhade et al found instrumented spinal fusion, preoperative opioid use duration, and depression to have the strongest association with postoperative chronic opioid use, ¹⁷ with preoperative use of benzodiazepines and gabapentin also associated in patients that were preoperatively opioid naive. ³⁷

The Karhade et al group validated a comprehensive risk model of chronic opioid use called the Stopping Opioid after Surgery (SOS) score in spine patients. ¹⁵ While developing such models is critical to rapid and evidence-based risk assessment, missing from the factors included in the model are income, surgery type and race, factors which the current study found to be independently associated with chronic opioid use. The challenge of incorporating race in a risk model for chronic opioid use may be due to its heterogeneous sociocultural interactions, but due to such qualitative differences, race certainly plays an informative role in determining the risk of chronic opioid use. In one study, Davison et al found Hispanics and blacks to be the highest consumers of opioids, while Asians had the lowest risk of chronic use. ²⁹ Although the interplay of race and opioid use may be regionally dependent, surgeons should continue to characterize and understand how race may inform opioid use in their patient population. Surgery type, or invasiveness, is a factor that multiple previously noted studies, including the current study, have found to be associated with opioid risk. The SOS model only includes major or minor procedures, but there is likely a greater hierarchy of surgical impact that can be delineated. Similarly, the SOS model bases socioeconomic status on Medicaid insurance status, which greatly simplifies the stratification of risk for chronic opioid use based on income.

The current study is not without its limitations. As a retrospective observational study at a large academic center, our research methodology may be burdened by recall bias and sampling bias, which would affect results. Second, the location of the study at a single center limits the size and scope of the study population, and thus its external validity. The relationship between patient-intrinsic variables, such as race, and opioid usage is likely intertwined with local sociocultural factors that may vary greatly between study populations.^28,29 Such variables provide a challenge to inter-study comparisons, but understanding their role is necessary to provide a comprehensive profile of a patient’s opioid abuse risk. Lastly, although standardized to morphine equivalents, the methodology of the study was not able to fully capture physician-specific differences in opioid prescription patterns.

The current study describes an array of variables and their association with opioid script renewal postoperatively at 3 and 12 months. Many of the current findings corroborate previous data, and some provide a greater depth of detail to our understanding of chronic opioid risk. The finding that inpatient opioid consumption did not associate with chronic opioid use helps provide clarity to understanding the role of perioperative opioids in the battle to reduce chronic usage. Prescription-opioid related deaths are slowly beginning to decline, but the opioid epidemic rages on. Surgeons must continue to seek a deeper understanding of the role patient-intrinsic factors play in determining the patient’s risk for opioid abuse.