Opioid Use Increases the Rate of Pseudarthrosis and Revision Surgery in Patients Undergoing Anterior Cervical Discectomy and Fusion

Abstract

Study Design

Retrospective Cohort

Objectives

To (1) quantify the risk opioids impart on pseudarthrosis development, (2) analyze the effect of pseudarthrosis on clinical outcomes, and (3) identify if the amount of opioids prescribed are predictive of pseudarthrosis revision.

Methods

Patients who underwent ACDF at a single institution between 2017-2019 were retrospectively identified. Postoperative dynamic cervical spine radiographs were reviewed to assess fusion status. Logistic regression models measured the effect of morphine milligram equivalents (MME) prescribed on the likelihood of pseudarthrosis development. Receiver operating characteristic (ROC) curves were generated to predict the probability of surgical revision based on MME prescribed.

Results

Of 298 included patients, an average of 2.01 ± 0.82 levels were included in the construct and 121 (40.9%) patients were diagnosed with a pseudarthrosis. However, only 14 (4.7%) required a pseudarthrosis revision. Patients requiring pseudarthrosis revision had worse one-year postoperative Δ PCS-12 (−1.70 vs. 7.58, P = 0.004), Δ NDI (3.33 vs. −15.26, P = 0.002), and Δ VAS Arm (2.33 vs. −2.48, P = .047). Multivariate logistic regression analyses found the three-month postoperative (OR=1.00, P = .010), one-year postoperative (OR=1.001, P = 0.025), and combined pre- and postoperative MME (OR=1.000, P = .035) increased the risk of pseudarthrosis. ROC analysis identified cutoff values to predict pseudarthrosis revision at 90.00 (area under the curve (AUC): 0.693, confidence interval (CI): 0.554-0.832), 132.86 (0.710, CI: 0.589-0.840), 224.76 (0.687, CI: 0.558-0.817) and 285.00 (0.711, CI: 0.585-0.837) MME in the preoperative, three-month postoperative, one-year postoperative, and combined pre-and postoperative period.

Conclusion

Increased prescription of opioid medications following ACDF procedures may increase the risk of pseudarthrosis development and revision surgery.

Level of Evidence

Therapeutic Level III

Introduction

Anterior cervical discectomy and fusions (ACDF) are increasing at an annual rate of approximately 5.7% in the United States. ¹ This is partly due to high patient satisfaction scores and excellent long-term clinical outcomes.^2-4 While ACDFs have a favorable complication profile, symptomatic pseudarthrosis is a common complication and can cause recurrent radicular or axial neck pain leading to poor clinical outcomes and potentially revision surgery.^5-7 The etiology of a pseudarthrosis is multifactorial, often caused by a combination of patient-specific and surgeon-specific risk factors. ⁷ Therefore, further characterization of surgeon-controlled risk factors contributing to pseudarthrosis development is paramount to mitigate the healthcare burden associated with revision surgery. ⁸

In 2017, the Department of Health and Human Services (HHS) declared a public health emergency related to the ongoing opioid epidemic in the United States.^9,10 From 1999 to 2018, over 400,000 Americans died of an opioid overdose.^11,12 Alarmingly, between April 2020 to April 2021 there were another 100,000 opioid-related deaths in the United States, corresponding to a 28.5% increase in opioid-related deaths compared to the previous year. ¹³ Since 52% of patients fill opioid prescriptions prior to an ACDF procedure, exploring the effects of opioid prescriptions on ACDF outcomes is of significant interest. ¹⁴

Basic science research has previously demonstrated opioids inhibit osteoblast activity in vitro. ¹⁵ Translational research in animal models suggest opioids delay bone maturation and remodeling in spine fusions. ¹⁶ Given recent clinical evidence indicates opioids increase the risk of pseudarthrosis following long bone fractures, an investigation into the contribution of opioids to pseudarthrosis development after ACDF is warranted, especially considering the current opioid epidemic.^14,17-20

Therefore, the objectives of this study are: (1) to investigate if opioid use increases the pseudarthrosis rate after ACDFs as analyzed on postoperative dynamic cervical spine radiographs; (2) to determine if symptomatic and asymptomatic pseudarthrosis affects patient-reported outcome measures (PROMs); and (3) identify if the quantity of opioids prescribed predicts symptomatic pseudarthrosis requiring revision based on morphine milligram equivalents (MME) prescribed.

Materials and Methods

Inclusion Criteria

This study was approved by our Institutional Review Board with exempt status from obtaining patient informed consent due to its retrospective design and minimal risk to subjects. All patients ≥ 18 years of age with one-year postoperative dynamic cervical spine radiographs who underwent primary one- to four-level ACDFs by one of five fellowship-trained spine surgeons at a single academic medical center from 2017-2019 were retrospectively identified. Patients were excluded if the index ACDF was performed as a revision procedure, utilized a combined anterior/posterior approach, included a cervical corpectomy, if no patient data existed in our state’s Prescription Drug Monitoring Program (PDMP) database -- implemented on August 25, 2016, or if the surgical indication was for trauma, infection, or neoplasm (Appendix A). Each ACDF was performed in a similar manner with a standard Smith-Robinson approach, subperiosteal elevation of the longus colli, skeletonization of the pedicles, and sacrifice of the posterior longitudinal ligament at the level of compression. An anterior cervical plate with locking screws was utilized to improve stability and demineralized bone matrix or allograft chips were typically utilized to augment the fusion. For the purpose of this manuscript, only patients with a revision due to a symptomatic pseudarthrosis were classified as “revisions.” A symptomatic pseudarthrosis was defined as either recurrent axial neck pain or radiculopathy with evidence of >1 mm of interspinous motion on dynamic radiography. In the absence of recurrent pain, patients were not offered revision surgery, even if a pseudarthrosis was detected on radiographic imaging. Each revision procedure included posterior cervical decompression and fusion and anterior plate removal if indicated due to screw and/or plate loosening. Smoking status was documented as current, never, and former. The former smokers had quit smoking at least 6 months prior to surgery.

Data Extraction

Patient demographics and surgical characteristics were collected through a Structured Query Language (SQL) search and manual chart review of electronic medical records. The MME corresponding to each filled opioid prescription was recorded through our state’s PDMP database one-year preoperatively and one-year postoperatively. In addition, each opioid prescriber’s specialty was recorded through internet search of individual national provider identification numbers. PROMs were retrospectively collected through our institution’s prospectively collected database (OBERD, Columbia, MO) and were included at the preoperative, three-month postoperative, and one-year postoperative periods. Preoperative diagnoses obtained were categorized as radiculopathy, myelopathy, and myeloradiculopathy. PROMs extracted included the Visual Analogue Scale for neck pain (VAS Neck) and arm pain (VAS Arm), the Neck Disability Index (NDI), the modified Japanese Orthopaedic Association scale (mJOA), and the Mental and Physical Component Scores of the Short-Form 12 (SF-12) Health Survey (MCS-12 and PCS-12, respectively).

Radiographic Evaluation

Dynamic cervical spine radiographs were reviewed through our institution’s Picture Archiving and Communication System (PACS; Sectra AB, Linköping, Sweden). The distance (in millimeters) between the superior and inferior spinous process at each level of the ACDF was measured on flexion and extension radiographs. Radiographic fusion was defined as < 1 mm of interspinous motion between each instrumented level and ≥ 4 mm of motion an any adjacent unfused level in accordance with the guidelines published by the Cervical Spine Research Society (CSRS) Special Project Committee. ²¹ This method of pseudarthrosis evaluation has been demonstrated to have an interobserver reliability of 0.825 (95% confidence interval (CI): 0.746 to 0.888) with an average interobserver measurement difference of 0.1 mm. ²²

Statistical Analysis

Descriptive statistics including mean and standard deviation were used to record patient demographics, surgical characteristics, and surgical outcomes. A Shapiro-Wilk test was used to analyze the normality of each continuous variable, and parametric data was analyzed with independent t-tests, while non-parametric data was analyzed with Mann-Whitney U tests. Categorical variables were analyzed with Pearson’s chi-square tests or Fisher’s exact tests. Multivariate logistic regression models, accounting for age, sex, body mass index (BMI), smoking status (current, former, or non-smoker), Elixhauser Comorbidity Index (ECI), preoperative diagnosis (radiculopathy, myelopathy, and myeloradiculopathy), osteoporosis, interbody composition (machined allograft, polyetheretherketone (PEEK), or titanium), and construct length were developed to analyze the relationship between radiographic pseudarthrosis and both MME prescribed and opioid use. ROC curves were generated to predict the probability of surgical revision for pseudarthrosis based on the total MME prescribed. R software, version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria) was used for all data analysis. Statistical significance was set at P < 0.05.

Source of Funding

There was no external funding for this study.

Results

A total of 298 patients met the inclusion criteria. Three main groups were compared: (1) patients with a complete arthrodesis versus those with a pseudarthrosis, (2) patients with a pseudarthrosis not requiring revision versus patients with a pseudarthrosis requiring revision, and (3) patients with a pseudarthrosis requiring revision versus patients without a revision procedures.

Pseudarthrosis vs. Fusion Cohort

Patient Demographics

Of the 298 included patients, an average of 2.01 ± 0.82 levels were included in the construct and 121 patients (40.6%) were diagnosed with a pseudarthrosis. Rates of pseudarthrosis were 28.7%, 41.7%, 50.0%, and 63.6%, for constructs of 1, 2, 3, or 4-levels, respectively, averaging a pseudarthrosis rate of 20.5% per level. No significant differences in baseline demographics including age (P = .157), sex (P = .977), BMI (P = .078), smoking status (P = .218), ECI (P = .205), or osteoporosis (P = .118) existed between patients who had a pseudarthrosis compared to patients who had a complete arthrodesis. A significantly higher proportion of patients who had a pseudarthrosis underwent three- and four-level ACDFs (P = .017) (Table 1).

Table 1.

Radiographic Fusion – Demographics and Opioid Use.

	Construct Not Fused	Construct Fused	P-Value
	N = 121	N = 177
Age (years)	54.0 (11.2)	52.2 (11.0)	0.157
Sex			0.977
Female	68 (56.2%)	101 (57.1%)
Male	53 (43.8%)	76 (42.9%)
BMI (kg/m2)	28.5 (5.57)	29.5 (5.72)	0.078
Smoking status			0.218
Never	65 (53.7%)	109 (61.6%)
Current	4 (3.31%)	2 (1.13%)
Former	52 (43.0%)	66 (37.3%)
Elixhauser comorbidity index	1.40 (1.46)	1.60 (1.45)	0.205
Osteoporosis			0.118
No	114 (94.2%)	156 (88.1%)
Yes	7 (5.79%)	21 (11.9%)
Opioid prescription (MME)
Preoperative	296 (474)	191 (440)	0.101
3 Month Postoperative	160 (156)	120 (145)	0.001*
1 Year Postoperative	303 (495)	195 (379)	0.010*
Total Preoperative to 1 Year Postoperative	494 (867)	305 (711)	0.010*

^*Indicates statistical significance (P < .05).

Surgical Characteristics and Outcomes

The time to the final dynamic radiographs was not significantly different between patients with a pseudarthrosis and those with a complete construct arthrodesis (15.8 ± 9.81 vs. 17.4 ± 11.2 months, P = 0.136). There was also no significant difference in the 90-day all cause readmission rate (0.83% vs. 0.56%, P = 1.000) between patients with and without a pseudarthrosis (Appendix B).

Objective 1 - Opioid Comparison

Opioid use was not significantly different in the preoperative period (296 vs. 191 MME, P = .101) for patients who were diagnosed with a pseudarthrosis compared to patients who were fused. However, patients who were diagnosed with a pseudarthrosis had significantly greater three-month postoperative (160 vs. 120, P = .001), one-year postoperative (303 vs. 195, P = .010), and combined pre- and postoperative MME use (494 vs. 305, P = .010) (Table 1). Multivariate logistic regression analyses showed that when accounting for patient demographics, preoperative diagnosis, and construct length, three-month postoperative (OR = 1.003, P = .010), one-year postoperative (OR=1.001, P = .025), and combined pre- and postoperative MME (OR = 1.000, P = .035) increased the likelihood of pseudarthrosis development (Table 2).

Table 2.

Multivariate Logistic Regression of Radiographic Pseudarthrosis at Follow Up – Opioid Use Periods.

Period of Opioid Prescription	Predictor	Estimate (β) 1	P-Value	Odds Ratio	95% CI
Preoperative	MME (per day)	0.001	0.113	1.001	1.000 – 1.001
	Levels fused	0.503	0.030	1.654	1.057 – 2.640
	Smoking status
	Non-Smoker	Ref
	Current Smoker	1.900	0.126	6.685	0.722 – 150.620
	Former Smoker	0.011	0.976	1.011	0.507 – 2.002
3 Month postoperative	MME (per day)	0.003	0.010*	1.003	1.001 – 1.005
	Levels fused	0.584	0.002*	1.792	1.257 – 2.597
	Smoking status
	Non-Smoker	Ref
	Current Smoker	1.886	0.076	6.590	0.948 – 70.820
	Former Smoker	0.037	0.897	1.037	0.593 – 1.082
1 Year postoperative	MME (per day)	0.001	0.025*	1.001	1.000 – 1.002
	Levels fused	0.532	0.002 b	1.704	1.223– 2.403
	Smoking status
	Non-Smoker	Ref
	Current Smoker	1.850	0.067	6.344	0.978 – 59.539
	Former Smoker	0.051	0.849	1.052	0.623 – 1.769
Total preoperative + postoperative	MME (per day)	0.0004	0.035*	1.000	1.000 – 1.001
	Levels fused	0.528	0.002*	1.696	1.218 – 2.390
	Smoking status
	Non-Smoker	Ref
	Current Smoker	1.848	0.066	6.351	0.984 – 59.134
	Former Smoker	0.061	0.819	1.063	0.630 – 1.786

¹Accounting for age, sex, body mass index, Elixhauser Comorbidity Index, osteoporosis, interbody type (allograft, polyetheretherketone, or titanium), and preoperative diagnosis.

MME: milligram of morphine equivalent.

^*Indicates statistical significance (P < .05).

Objective 2(a) – Patient-Reported Outcomes

There were no significant differences in three-month or one-year postoperative PROM scores between patients who were diagnosed with a pseudarthrosis compared to patients who were fused (all, P > .05), with the exception of patients with a pseudarthrosis having significantly lower Δ MCS-12 at the three-month postoperative period (−2.02 vs. 2.47, P = .045) (Appendix C).

Pseudarthrosis with Revision vs. Pseudarthrosis without Revision

Of the 122 patients diagnosed with a pseudarthrosis (one patient was diagnosed with a pseudarthrosis intraoperatively without evidence of a pseudarthrosis on dynamic radiographs), 108 patients had a pseudarthrosis not requiring revision and 14 patients had a pseudarthrosis requiring revision. A significant difference existed in preoperative (256 vs. 504, P = .042), 3-month postoperative (146 vs. 258, P = .048), and combined preoperative and postoperative MME (416 vs. 1,070, P = .02) and a nonsignificant difference in 1-year postoperative MME (257 vs. 638, P = .065) (Table 3).

Table 3.

Patients with a Pseudarthrosis Revision Compared to Patients with a Pseudarthrosis Not Requiring a Revision.

	Pseudarthrosis No Revision	Pseudarthrosis Revision	P-Value 1
	N = 108	N = 14
Opioid Prescription (MME):
Preoperative	256 (429)	504 (647)	0.042*
3-Month Postoperative	146 (138)	258 (238)	0.048*
1-Year Postoperative	257 (390)	638 (935)	0.065
Combined Preoperative and 1-Year Postoperative	416 (712)	1070 (1541)	0.020*

MME: Milligram of morphine equivalent.

¹Mann-Whitney U test

^*Indicates statistical significance (P < .05)

Pseudarthrosis Revision vs. Non-Revision Cohort

Patient Demographics

Of the 298 included patients, 14 patients (4.7%) underwent revision due to symptomatic pseudarthrosis. There were no significant differences in baseline demographics including age (P = .713), sex (P = 1.000), BMI (P = .608), smoking status (P = .450), ECI (P = .913), or osteoporosis (P = .000) (Table 4).

Table 4.

Revision for Symptomatic Pseudarthrosis – Demographics and Opioid Use.

	No Revision	Revision	P-Value
	N = 284	N = 14
Age (years):	53.0 (11.3)	52.2 (7.26)	0.713
Sex:			1.000
Female	161 (56.7%)	8 (57.1%)
Male	123 (43.3%)	6 (42.9%)
BMI (kg/m2):	29.0 (5.59)	30.3 (7.38)	0.608
Smoking status:			0.450
Never	168 (59.2%)	6 (42.9%)
Current	6 (2.11%)	0 (0.00%)
Former	110 (38.7%)	8 (57.1%)
Elixhauser comorbidity index:	1.52 (1.45)	1.57 (1.50)	0.913
Opioid prescription (MME):
Preoperative	217 (437)	504 (647)	0.013*
3 Month Postoperative	130 (143)	258 (238)	0.011*
1 Year Postoperative	219 (384)	638 (935)	0.018*
Total Preoperative to 1 Year Postoperative	348 (713)	1070 (1541)	0.004*

^*Indicates statistical significance (P < 0.05).

Surgical Characteristics and Outcomes

Construct length (P = .542) and the length of follow-up (P = .962) were not significantly different between the two groups. There was also no significant difference in the 90-day all-cause readmission rate (0.00% vs. 0.70%, P = 1.000) between the two groups (Appendix D).

Objective 2(b) – Patient-Reported Outcomes

Pseudarthrosis revision patients had significantly lower one-year postoperative PCS-12 (30.0 vs. 40.4, P = .034), and greater NDI (49.6 vs. 25.0, P = .028). These differences corresponded to significantly lower Δ PCS-12 (−1.70 vs. 7.68, P = .004) and greater Δ NDI (3.33 vs. −15.26, P = .002) at the one-year postoperative period. Additionally, pseudarthrosis revision patients had significantly lower preoperative index procedure VAS Arm (2.39 vs. 5.42, P = .010), but a significantly worse magnitude of improvement in the one-year postoperative Δ VAS Arm scores (2.33 vs. −2.48, P = .047) (Table 5).

Table 5.

Revision for Symptomatic Pseudarthrosis – Patient-Reported Outcomes.

	No Revision	Revision	P-Value
	N = 284	N = 14
MCS-12
Preoperative	47.3 (12.0)	44.5 (12.2)	0.495
3 Months Postoperative	48.7 (11.7)	26.2 (8.46)	0.028*
Δ 3 Months	1.04 (9.63)	−9.44 (0.37)	<0.001*
1 Year Postoperative	48.1 (11.6)	40.8 (17.6)	0.319
Δ 1 Year	−0.49 (12.4)	−1.93 (9.13)	0.814
PCS-12
Preoperative	34.0 (8.48)	35.0 (11.9)	0.871
3 Months Postoperative	37.3 (10.2)	33.9 (13.5)	0.650
Δ 3 Months	4.01 (8.43)	−4.59 (5.01)	0.232
1 Year Postoperative	40.4 (11.3)	30.0 (6.99)	0.034*
Δ 1 Year	7.58 (9.83)	−1.70 (2.46)	0.004*
NDI
Preoperative	40.5 (19.3)	44.2 (19.1)	0.628
3 Months Postoperative	29.1 (17.5)	39.0 (21.2)	0.416
Δ 3 Months	−11.24 (15.8)	−15.00 (12.7)	0.748
1 Year Postoperative	25.0 (18.5)	49.6 (26.3)	0.028*
Δ 1 Year	−15.26 (19.6)	3.33 (4.62)	0.002*
VAS neck
Preoperative	5.93 (2.92)	6.27 (1.76)	0.961
3 Months Postoperative	3.24 (2.33)	9.00 (1.41)	0.017*
Δ 3 Months	−2.97 (2.69)	1.00 (0.00)	0.030*
1 Year Postoperative	2.79 (2.50)	5.25 (3.77)	0.132
Δ 1 Year	−2.77 (2.94)	−0.33 (2.08)	0.207
VAS arm
Preoperative	5.42 (3.21)	2.39 (2.27)	0.010*
3 Months Postoperative	2.64 (2.78)	1.50 (2.12)	0.611
Δ 3 Months	−2.94 (3.53)	1.00 (1.41)	0.072
1 Year Postoperative	2.51 (2.79)	3.25 (2.75)	0.497
Δ 1 Year	−2.48 (3.11)	2.33 (2.08)	0.047*
mJOA
Preoperative	14.9 (3.00)	15.2 (2.32)	0.952
3 Months Postoperative	16.0 (2.81)	18.0 (0.00)	0.100
Δ 3 Months	1.16 (3.15)	1.00 (1.41)	0.925
1 Year Postoperative	15.9 (2.46)	16.2 (0.84)	0.583
Δ 1 Year	1.18 (2.92)	0.50 (2.12)	0.737

^*Indicates statistical significance (P < .05).

Objective 3 - Opioid Comparison and ROC Curve Analysis

MME prescriptions were greater in the preoperative (504 vs. 217, P = .013), three-month postoperative (258 vs. 130, P = .011), one-year postoperative (638 vs. 219, P = .018), and the combined pre- and postoperative period (1070 vs. 348, P = .004) for patients who had a revision procedure (Table 4). ROC analysis identified optimal cutoff values to predict the probability of surgical revision based on the total MME. The corresponding AUCs were 90.00 (AUC: 0.693, CI: 0.554-0.832), 132.86 (AUC: 0.710, CI: 0.589-0.840) (Figure 1), 224.76 (AUC: 0.687, CI: 0.558-0.817), and 285 MME (AUC: 0.711, CI: 0.585-0.837), respectively (Table 6).

Figure 1.

ROC Curve – 3 Month Postoperative MME as a Predictor of Pseudarthrosis Revision.

Table 6.

MME Per Day Cutoffs for ACDF Revision for Symptomatic Pseudarthrosis.

Variable	Cutoff	AUC (95% CI)	Sensitivity	Specificity
Preoperative MME/day	90.00	0.693 (0.554, 0.832)	0.824	0.574
3 Month postoperative MME/day	132.86	0.710 (0.589, 0.840)	0.722	0.737
1 Year postoperative MME/day	224.76	0.687 (0.558, 0.817)	0.650	0.786
Preoperative to 1 Year postoperative MME/day	285.00	0.711 (0.585, 0.837)	0.700	0.753

Opioid Prescribers

Primary care physicians were responsible for 35.9% of opioid prescriptions in our patients while spine surgeons accounted for 15.7% of the total opioid prescriptions. Providers in other specialties (mostly Acute Care, Emergency Medicine, Neurology, Psychiatry, and General Surgery) accounted for 30.2% of opioid prescriptions. Non-operative spine physicians (Anesthesia, Physical Medicine and Rehabilitation, Pain Management) and non-spine orthopedic surgeons accounted for a modest 8.8% and 9.5% of prescriptions, respectively (Table 7).

Table 7.

Opioid providers within the 1-year preoperative and 1-year postoperative window.

Prescriber Group 1	Number of Prescribers
Spine Orthopedic providers	299 (15.7%)
Non-spine Orthopedic providers	181 (9.5%)
Pain providers 2	196 (8.8%)
Primary care providers 3	685 (35.9%)
Other providers 4	576 (30.2%)

¹MD, DO, DMD, DDS, NP, and PA.

²Anesthesia, Physical Medicine and Rehabilitation, Pain Management.

³Family Medicine and Internal Medicine.

⁴Including mostly Acute Care, Emergency Medicine, Neurology, General Surgery, and Psychiatry.

Single-level ACDF in Former Smokers vs. Non-Smokers and in opioid users vs. non-users

When comparing former smokers versus non-smokers, no significant differences were observed in pseudarthrosis rates (P = 1.000), 90-day readmissions (P = 1.00), or revision surgery due to pseudarthrosis (P = 1.000), or adjacent level surgery (P = .679) (Appendix E). When evaluating opioid users versus non- users, our one year postoperative data found that opioid use was a significant positive predictor of pseudarthrosis (OR=1.562, P = .012) (Appendix F).

Discussion

ACDF is an effective procedure with high overall patient satisfaction rates and low revision rates.^4,23-25 Since improved pain relief is one of the primary drivers of high patient satisfaction following ACDF, this may be a contributor to the historically high rates of opioid prescriptions in postoperative pain control regimens.^26-28 Given that opioids adversely affect bone healing and remodeling, identifying the effect of opioids on pseudarthrosis is a clinical necessity.^16,17,29-33 Our study suggests ACDF construct length and increased MME prescriptions may be critical factors affecting pseudarthrosis rates. To our knowledge, this is the first clinical study demonstrating a correlation between increased opioid use and greater rates of cervical spine pseudarthrosis, supporting the aforementioned basic and translation research findings.^15,16

One potential mechanism for the adverse effect of opioids on bone remodeling is through immunomodulation of the inflammatory environment.^34,35 This reportedly occurs through T-cell and monocyte/macrophage signaling modification which adversely affects bone healing.^34,36,37 Another proposed hypothesis involves osteoblastic inhibition, whereby treatment of osteoblast-like cells with morphine leads to a decrease in osteocalcin synthesis (a commonly used biomarker for osteoblastic activity). ¹⁵ Future well-designed translational research is necessary to confirm the exact mechanism whereby opioids inhibit bone remodeling.

Previous literature has suggested chronic preoperative opioid use increases the risk of postoperative complications following ACDF. ³⁸ Our ROC analysis identified cutoff values of 90.00, 132.86, 224.76, and 285.00 MME in the preoperative, three-month postoperative, one-year postoperative, and combined pre- and one-year postoperative periods as potential risk factors for pseudarthrosis, which increased the risk of revision surgery by 69.3%, 71.0%, 68.7%, and 71.1%, respectively. The Centers for Disease Control and Prevention (CDC) calculated MME conversions for hydrocodone and oxycodone (1.0 and 1.5, respectively). ³⁹ A patient who takes an entire prescription of eighteen 5 mg oxycodone pills (135 MME) within the 3-month postoperative window would have a 71% increased risk of pseudarthrosis revision. To put the increased revision rates into perspective, previous literature has indicated revision rates due to an ACDF pseudarthrosis is approximately 6%; therefore, reaching the above MME thresholds would increase a patients pseudarthrosis revision risk to approximately 10%. ³⁸ This data may be used as a tool to increase communication between surgeons and patients about potential risks of prolonged or excessive opioid use pre- and postoperatively.^28,40-42

For multilevel ACDFs, previous literature has found rates of pseudarthrosis, defined as <1 mm of interspinous motion between segments, to be 22.2%, 34.8%, 49.7%, and 75.0% for 1, 2, 3, and 4-level constructs. ⁴³ However, most patients who develop a pseudarthrosis are asymptomatic with only a minority of patients developing recurrent symptomatic axial neck pain or radiculopathy requiring revision surgery.^44,45 Since bony fusion rates may continue to increase up to two years postoperatively, some patients who were defined as having a pseudarthrosis at one-year follow up may have ultimately progressed to bony union.^44,46

Our multivariable analysis suggests greater MME prescriptions during the three-month, one-year, and total pre- to one-year postoperative periods significantly increases the risk of patients developing pseudarthrosis. Further, increasing opioid prescriptions in the preoperative and 3-month postoperative periods appears to have a direct effect on pseudarthrosis revision rates. However, it is difficult to conclude if greater MME prescriptions at one-year contribute to increased symptomatic pseudarthrosis rates, or if the symptoms from the pseudarthrosis are the cause of increased MME prescriptions. Prospective trials comparing opioids to alternative pain management medications are needed to investigate the impact of opioids on ACDF revision rates.

A discrepancy exists between opioids modest effect on pseudarthrosis rates and their substantial effect on pseudarthrosis revision rates. One potential explanation for this finding is that opioids affect fusion mass quality to a greater extent than fusion mass quantity, which has previously been demonstrated in translational research. ³⁸ Additionally, our data suggests patients who progress to pseudarthrosis revision are prescribed approximately twofold the MME of patients who have an asymptomatic pseudarthrosis. This supports our hypothesis that poor fusion mass quality imparted by excessive opioid use may lead to elevated rates of pseudarthrosis revision. Well-designed prospective trials with detailed documentation of intraoperative findings may help confirm this hypothesis.

The impact of pseudarthrosis on PROMs has been understudied in the spine literature.^30,44 Our results suggest that, in general, patients with an asymptomatic pseudarthrosis at the one-year postoperative point do not have a significant difference in PROMs compared to patients with a complete arthrodesis. This finding mirrors those of Crawford et al., who reported no significant differences in PROMs between patients with radiographic evidence of pseudarthrosis and patients with radiographic evidence of fusion at one-, two, and five years postoperatively. ³⁰ On the other hand, patients who underwent revision surgery for symptomatic pseudarthrosis had significantly worse one-year postoperative PCS-12 and NDI, along with significantly less improvement in PCS-12, NDI, and VAS Arm as indicated by lower one-year postoperative Δ PCS-12 scores and higher one-year postoperative Δ NDI and Δ VAS Arm scores, respectively.

Multiple limitations were present in our study. First, only patients with one-year postoperative dynamic radiographs and a history of an opioid prescription in our state’s PDMP database were included. It is not institutional policy to obtain dynamic cervical radiographs during ACDF follow up and the majority of patients without any residual arm or neck pain do not attend the one-year clinic appointment. This may ultimately artificially inflate our pseudarthrosis rate. Therefore, the low proportion of patients included in the study may have imparted significant follow up bias given that approximately 85% of our patients who underwent an ACDF were excluded. Second, it is unclear whether patients with a pseudarthrosis had greater opioid consumption 1-year postoperatively due to pain caused by the pseudarthrosis or if the greater number of opioids prescribed resulted in a pseudarthrosis. Prospective studies may increase our understanding of this relationship. Finally, our cohort had an average time interval of approximately 17 months between surgery and the final postoperative dynamic radiographs. Therefore, we are unable to comment on long-term radiographic fusion status since there is an increase in osseous fusion between one- and two-years postoperatively. ⁴⁶ While data collection on opioid use from our state’s PDMP registry is a strength of the study, we are unable to ascertain whether patients consumed the entire opioid prescription they filled. Our state’s PDMP registry utilizes interstate data sharing, but it does not capture records from all states, therefore, opioid prescriptions filled by each patient may be underestimated if they visited states without opioid data sharing. ⁴⁷

Conclusion

Increased prescription of opioid medications at the three-month, one-year, and combined preoperative to one-year postoperative periods may increase the risk of pseudarthrosis following ACDF. Patients with greater opioid prescriptions were also more likely to have a symptomatic pseudarthrosis indicated by increased one-year neck disability and less magnitude of improvement in radicular arm pain. ROC analysis identified 90.00, 132.86, and 224.76 MME in the preoperative, three-month postoperative, and one-year postoperative periods (corresponding to 12, 18, and 30 pills of 5 mg oxycodone) as optimal ROC cutoffs, which predicted a 69.3%, 71.0%, and 68.7% increased risk of pseudarthrosis revision.

ORCID iDs

Mark J. Lambrechts https://orcid.org/0000-0002-9106-2228

Nicholas D. D’Antonio https://orcid.org/0000-0001-8484-0207

Gregory R. Toci https://orcid.org/0000-0003-4770-3507

Brian A. Karamian https://orcid.org/0000-0003-0512-6019

Matthew Sherman https://orcid.org/0000-0002-3258-3568

Jose A. Canseco https://orcid.org/0000-0002-2152-5725

Alan S. Hilibrand https://orcid.org/0000-0001-8811-9687