Noninstrumented Posterolateral Lumbar Fusions Continue to Fade Relative to Instrumented Posterolateral Approaches for Lumbar Degenerative Spondylolisthesis Over the Past Decade

Abstract

Introduction:

Although posterolateral lumbar fusions (PLFs) were historically routinely performed without instrumentation, instrumentation has become ubiquitous in this setting. Recent assessments of noninstrumented PLFs practices are lacking.

Methods:

Lumbar degenerative spondylolisthesis patients undergoing noninstrumented or instrumented single-level PLFs (with or without interbody) were identified from 2012 to 2022 M170 Ortho PearlDiver. Univariable analysis identified differences in patient characteristics. Yearly utilization trends were tracked and analyzed using simple linear regression/overall F-tests. Ninety-day postoperative adverse outcomes were compared using multivariable logistic regression with 4:1 matching for patient age, sex, and Elixhauser comorbidity index. Five-year lumbar revision surgery rates were assessed using Kaplan-Meier survival analyses and log-rank tests.

Results:

A total of 117,796 (95.8%) instrumented and 5147 (4.2%) noninstrumented PLF lumbar degenerative spondylolisthesis patients were identified. Noninstrumented PLFs declined from 5.52% to 3.49% from 2012 to 2022 (P < 0.001). These were more common in older, male, osteoporotic patients, varied by insurance and region, and were more often performed by orthopaedic surgeons. After matching, no notable differences were found between instrumented versus noninstrumented cases in 90-day aggregated adverse events or 5-year lumbar revision surgery rates.

Discussion:

For lumbar degenerative spondylolisthesis, noninstrumented fusion represents a small and declining percentage of PLFs over the past decade, with usage varied by clinical and nonclinical patient characteristics. No differences were found in 90-day aggregated postoperative adverse events and 5-year lumbar revision surgery rates, suggesting that instrumentation can be safely performed and that both approaches had similarly durable results. These findings support considering noninstrumented PLFs in select cases when the safety, utility, or cost-benefit of instrumentation is questioned.

Over the past several decades, there has been an upward trend in the number of posterolateral lumbar fusions (PLFs) being performed,¹ for which lumbar degenerative lumbar spondylolisthesis is the most common indication.² Although PLFs were historically routinely performed without instrumentation,^3-5 instrumentation has become ubiquitous in this setting.^6-8 Nonetheless, although most studies agree that instrumentation improves fusion rates,^9-15 there is disagreement as to whether this improves clinical outcomes and revision surgery rates.^9,11,15-22

In 2008, the North American Spine Society developed evidence-based clinical guidelines for the management of degenerative lumbar spondylolisthesis.^23,24 The 2008 guidelines recommended the addition of instrumentation for improving fusion rates but not for improving clinical outcomes for degenerative lumbar spondylolis thesis.²⁴ The current guidelines were updated in 2014 and no longer recommend but rather suggest that instrumentation improves fusion rates but not clinical outcomes.^23,25

Informing such clinical guidelines, numerous studies have examined the comparative outcomes of instrumented versus noninstrumented PLFs.^{12-14,16,21,22,26-28} One of the most widely cited studies is a prospective randomized study by Fischgrund et al¹³ that followed 67 degenerative spondylolisthesis patients and reported higher rates of fusion with instrumentation but no differences in clinical outcomes. Furthermore, Abdu et al¹⁵ compared noninstrumented versus instrumented PLF procedures through a subgroup analysis of the Spine Patient Outcomes Research Trial, comprising 239 instrumented PLF patients and 80 noninstrumented PLF patients. This study found that instrumented PLFs had higher rates of fusion, higher rates of postoperative complications, lower rates of revision surgery at 4 years, and, in total, suggested no notable advantages of one over the other.

Although the above-noted studies have helped inform current understandings of noninstrumented PLF efficacy, existing studies suffer from being based on older data sets, potentially nonrepresentative patient populations, and small sample sizes, indicating a need for more extensive exploration. Furthermore, very few studies have looked into utilization trends of noninstrumented PLFs. Kepler et al²⁹ queried the American Board of Orthopedic Surgery database from 1999 to 2011 and identified 5639 instrumented and noninstrumented PLFs. This study reported that noninstrumented PLF rates markedly decreased from 36% to 8% from 1999 to 2011. This study focused on surgeries of early career surgeons and lacked representation of neurosurgeons.

This study aimed to refresh the evaluation of noninstrumented versus instrumented PLF trends in a recent, large, and nationally representative database. Furthermore, the study aimed to assess the relative rates of perioperative adverse events and subsequent surgery between these cohorts.

Methods

Database and Study Cohort

This study queried data from 2012 to 2022 in the PearlDiver M170 Ortho data set, a large national health administrative database. The M170 Ortho data set contains aggregated, deidentified, and Health Insurance Portability and Accountability Act compliant health information for nearly 170 million orthopedic patients in the United States. The use of this database is well-established in spine literature.^30-35 Our institutional review board determined studies using the PearlDiver data set exempt from review.

All patients who underwent a single-level PLF, with or without interbody fusion, were identified using the Current Procedural Terminology codes CPT-22612, CPT-22630, or CPT-22633. They were further characterized by having the International Classification of Diseases (ICD) diagnostic codes for degenerative spondylolisthesis of ICD-9-D-7384, ICD-10-D-M43.15, ICD-10-D-M43.16, or ICD-10-D-M43.17. Patients were excluded from the study if they were younger than 18 years, had concurrent neoplasm, infection, or trauma, or had less than 90 days of follow-up. They were also excluded if there were additional segment fusion codes or anterior fusion codes.

The instrumented PLF cohort was comprised of patients receiving concurrent coding for posterior nonsegmental instrumentation (CPT-22840) with any single-level PLF with or without interbody. The noninstrumented PLF cohort was the balance of the single-level PLF patients.

Patient variables were then abstracted. These included age, sex, Elixhauser comorbidity index (ECI, a standardized patient comorbidity index constructed using ICD codes),³⁶ presence of osteoporosis, insurance type (commercial, Medicaid, or Medicare), geographic region of the procedure (Northeast, Midwest, South, or West), surgical specialty of the performing physician (orthopaedic vs. neurological surgery), and year of surgery. Patients who underwent instrumented PLFs were matched in 4:1 ratio to the noninstrumented PLF cohort on the basis of age, sex, and ECI for outcome analysis.

Trends in Procedure Utilization

The yearly incidence of noninstrumented and instrumented PLFs was tabulated from 2012 to 2022 and normalized to 100,000 total patients in the M170 Ortho database. The percentage of PLFs performed by noninstrumented approaches was computed yearly and tracked longitudinally. The yearly proportional utilization of noninstrumented PLFs was further broken down by surgical specialty—orthopaedic and neurological surgery—and assessed over the study period.

Ninety-Day Adverse Events and Five-Year Revision Surgeries

The incidence and multivariable odds of 90-day postoperative adverse events following instrumented and noninstrumented PLFs were determined for the matched patient cohorts. Postoperative adverse events were identified in both independent and aggregated forms using ICD codes for specific complications. Aggregated groups included serious adverse events (surgical site infection, sepsis, deep vein thrombosis, pulmonary embolism, myocardial infarction, cardiac arrest), minor adverse events (wound dehiscence, hematoma, acute kidney injury, transfusion, urinary tract infection, pneumonia), or any adverse events (severe or minor adverse event).

Kaplan-Meier survival analyses compared the five-year survival to subsequent lumbar operations in the noninstrumented and instrumented PLF cohorts. Survival time was measured from the index PLF date to the first subsequent lumbar spinal procedure—determined through Current Procedural Terminology coding for any surgery to the lumbar spine within the next five years. Only patients with five years of follow-up were included in the survival curve determination.

Data Analysis

Univariable analyses comparing patient demographics before and after matching were done through Welch t-tests and Pearson χ² tests. Welch t-tests identified notable cohort differences in continuous variables (age and ECI), whereas χ² tests determined notable differences in categorical variables (sex, insurance plan, and country region where procedure was performed).

For utilization trendlines, simple linear regression models were fitted. Overall, F-tests were computed to determine whether slopes of the simple linear regression models were nonzero, indicating notable differences in incidence or proportion.

Multivariable analysis was done to determine if the lack of instrumentation during PLFs was an independent risk factor for any individual or aggregated adverse event. Multivariable logistic regression, controlling for age, sex, and ECI, was done to determine notable differences in the incidence of adverse events. Odds ratios and 95% confidence intervals were similarly determined. For univariable and multivariate analyses, significance was corrected for multiple analyses using Bonferroni correction. The adjusted P value was P < 0.0036 for significance in both analyses.

Notable differences in revision surgery curves reported in Kaplan-Meier survival analysis were computed using log-rank (Mantel-Cox) tests. Significance was defined at P < 0.05.

Data organization was done in Microsoft Excel (Microsoft Corporation). Statistical analyses were done using the PearlDiver system and Prism 10 (GraphPad Software). Figures were created in Prism 10.

Results

Study Cohort and Patient Demographics

From 2012 to 2022, 117,796 (95.8%) instrumented PLFs and 5147 (4.2%) noninstrumented PLFs were identified in the data set. Before matching, patients who underwent noninstrumented PLF procedures were more likely to be older (66.7 vs. 62.4, P < 0.001), more likely to be male (41.3% vs. 36.6%, P < 0.001), have a higher comorbidity level (4.4 vs. 4.2, P < 0.001), be more osteoporotic (2.9% vs. 1.7%, P < 0.001), more likely to be on Medicare (35.6% vs. 28.8%, P < 0.001), and be operated on more by orthopaedic surgeons (46.2% vs. 38.4%, P < 0.001; Table 1).

Table 1.

Demographics and Descriptive Characteristics of Patients Undergoing Instrumented and Noninstrumented Posterior Lumbar Fusions

Factor or Variable	Unmatched			4:1 Matched
	Instrumented Fusion	Noninstrumented Fusion	P Value	Instrumented Fusion	Noninstrumented Fusion	P Value
Total	117,796	5147		19,374	4850
Age ± SD	62.4 ± 12.0	66.7 ± 11.0	<0.001	66.8 ± 10.2	66.8 ± 10.3	0.975
Sex, n (%)			<0.001			0.885
Female	30,873 (63.4)	3022 (58.7)		11,116 (59.2)	2788 (59.0)
Male	17,787 (36.6)	2125 (41.3)		7674 (40.8)	1935 (41.0)
ECI ± SD	4.2 ± 3.4	4.4 ± 3.4	<0.001	4.2 ± 3.1	4.2 ± 3.2	0.660
Osteoporosis, n (%)	1981 (1.7)	150 (2.9)	<0.001	368 (1.9)	132 (2.7)	<0.001
Insurance, n (%)			<0.001			0.009
Commercial	32,019 (67.8)	3129 (62.2)		11,073 (60.7)	2885 (62.5)
Medicaid	1620 (3.4)	107 (2.1)		388 (2.1)	97 (2.1)
Medicare	13,586 (28.8)	1787 (35.6)		6795 (37.2)	1632 (35.4)
Region, n (%)			<0.001			<0.001
Northeast	9081 (18.7)	1250 (22.6)		3619 (19.3)	1171 (25.0)
Midwest	13,818 (28.5)	1154 (24.5)		5243 (28.0)	1063 (22.7)
South	18,512 (38.2)	2174 (42.5)		7136 (38.1)	1979 (42.2)
West	7032 (14.5)	533 (10.4)		2711 (14.5)	477 (10.2)
Surgical specialty, n (%)			<0.001			<0.001
Neurological	39,485 (33.5)	1312 (25.5)		6401 (33.0)	1232 (25.4)
Orthopaedic	45,176 (38.4)	2380 (46.2)		7514 (38.8)	2258 (46.6)

ECI = Elixhauser comorbidity index

The cohort was matched 4:1 controlling for age, sex, and ECI. Bolded values indicate statistical significance at p < 0.05.

After matching for age, sex, and ECI, 19,374 patients (80.0%) in the instrumented cohort and 4850 patients (20.0%) in the noninstrumented cohort were there. These matched cohorts no longer had notable differences in age, sex, and ECI.

Noninstrumented Posterolateral Lumbar Fusion Utilization Trends

In 2012, noninstrumented PLF procedures had an incidence of 0.34 per 100,000 patients compared with 5.82 per 100,000 patients for instrumented approaches. By 2022, noninstrumented PLFs had an incidence of 0.37 per 100,000 patients compared with 10.20 per 100,000 patients for the instrumented cohort.

Over this study period, the proportional utilization of noninstrumented compared with instrumented decreased from 5.52% to 3.49% (Figure 1). This 36.8% decline in proportion was statistically significant (P < 0.001). Notably, the incidence of noninstrumented PLFs did not change over the years (P = 0.381), although the instrumented PLFs demonstrated a 78.9% increase in utilization (P < 0.001).

Figure 1.

Graph showing notable decline in proportional yearly incidence of noninstrumented versus instrumented posterolateral fusions for lumbar spondylolisthesis (P < 0.001).

The proportional utilization of noninstrumented PLFs broken down by surgical specialty is shown in Figure 2. Although orthopaedic surgeons have consistently performed most noninstrumented PLFs, there was a 29.4% decline in percentage usage (P < 0.001). However, neurological surgeons demonstrated no notable changes in proportional utilization over the years (P = 0.126).

Figure 2.

Graph of yearly noninstrumented posterolateral fusion incidence by specialty showing decline in orthopaedic use (P < 0.001) and no change in neurosurgical use (P = 0.126).

Ninety-Day Adverse Events and Five-Year Revision Surgeries

The incidence and multivariable odds of 90-day adverse events between the noninstrumented and instrumented PLF cohorts are shown in Table 2. No notable differences were found in the incidence and multivariable odds of aggregated adverse events (any, severe, or minor). For independent events, higher incidence (2.8% vs. 1.9%, P < 0.001) and multivariable odds (OR: 1.48, P < 0.001) of surgical site infection were noted.

Table 2.

Univariable and Multivariable Analyses of 90-Day Complications for Instrumented and Noninstrumented Matched Cohorts (Controlled for Age, Sex, and Elixhauser Comorbidity Index)

Factor or Variable	Univariate Analysis			Multivariate Analysis
	Instrumented Fusion (n = 19,374)	NonInstrumented Fusion (n = 4850)	P Valuea	Odds Ratio (95% CI)	P Valuea
Any adverse events	2127 (11.0%)	579 (11.9%)	0.061	1.10 (1.00-1.21)	0.056
Severe adverse events	938 (4.8%)	278 (5.7%)	0.012	1.20 (1.04-1.38)	0.010
Cardiac arrest	11 (0.1%)	<10 (0.2%)	1	1.09 (0.30-3.92)	0.892
Myocardial infarction	122 (0.6%)	28 (0.6%)	0.754	0.92 (0.61-1.39)	0.686
Pulmonary embolism	159 (0.8%)	39 (0.8%)	0.980	0.98 (0.69-1.40)	0.920
Sepsis	196 (1.0%)	58 (1.2%)	0.295	1.19 (0.88 - 1.59)	0.254
Deep vein thrombosis	271 (1.4%)	73 (1.5%)	0.623	1.08 (0.83-1.40)	0.563
Surgical site infection	372 (1.9%)	136 (2.8%)	<0.001	1.48 (1.21-1.80)	<0.001
Minor adverse events	1601 (8.3%)	418 (8.6%)	0.441	1.05 (0.94-1.17)	0.423
Pneumonia	264 (1.4%)	70 (1.4%)	0.717	1.06 (0.82-1.39)	0.653
Urinary tract infection	846 (4.4%)	213 (4.4%)	0.970	1.00 (0.86-1.17)	0.963
Acute kidney injury	366 (1.9%)	105 (2.2%)	0.236	1.14 (0.92-1.42)	0.240
Wound dehiscence	229 (1.2%)	67 (1.4%)	0.290	1.17 (0.89-1.54)	0.251
Hematoma	120 (0.6%)	26 (0.5%)	0.571	0.87 (0.57-1.33)	0.510
Transfusion	99 (0.5%)	34 (0.7%)	0.135	1.38 (0.93-2.04)	0.108

CI = confidence interval

^aBonferroni correction was used to account for multiple comparisons. The adjusted significance threshold is P ≤ 0.003, and bolded values indicate statistical significance at this level.

Multivariable odds ratio computed with instrumented group as reference group.

The 5-year rates of survival to subsequent lumbar surgeries for instrumented and noninstrumented PLF cohorts are demonstrated on the Kaplan-Meier survival curves in Figure 3. The 5-year survival to revision surgery was 91.1% for noninstrumented and 90.4% for instrumented approaches, corresponding to revision surgery rates of 8.9% and 9.6%, respectively. A log-rank test failed to show a statistically significant difference in the 5-year survival curves between the two groups.

Figure 3.

Kaplan-Meier survival curve for 5-year revision surgeries after instrumented versus noninstrumented posterolateral fusions showing no differences at 5 years (P = 0.3).

Discussion

Over the past decade, there has been a notable decrease in the proportion of posterolateral fusions being performed without instrumentation. The proportional rates and trends of noninstrumented PLFs are consistent with findings from Kepler et al²⁹ who reported a decline from 36% in 1999 to 8% in 2011. This study picked up in 2012 reporting a percent usage of nearly 6% and ended in 2022 just shy of 3.5%. Of interest, the declining proportion of noninstrumented PLFs was driven by a notable increase in the utilization of instrumented cases in the setting of no notable changes in the utilization of noninstrumented cases. Side by side, the two studies give us a picture of the relative utilization of noninstrumented PLFs from 1999 to 2022.

Patients undergoing noninstrumented fusions were different than those undergoing instrumented fusions by clinical and nonclinical metrics. Noninstrumented fusions were more likely to be performed for older, male, and osteoporotic patients. These clinical characteristics align with populations for whom surgeons were likely more worried about the toleration of instrumentation.^37-39 In addition, however, differences were noted in nonclinical measures, where noninstrumented fusions were more likely to be on Medicare patients, of differing geography, and more likely to be performed by orthopaedic surgeons. These differences suggest inconsistent practice patterns that are influenced by various factors, such as training, local practices, or costs.^40-42

This study found comparable perioperative outcomes between noninstrumented and instrumented PLF approaches. When considering aggregated postoperative adverse events, patients undergoing noninstrumented PLFs did not have differing incidence or odds of any, severe, or minor adverse events. When examining individual postoperative adverse events, a higher rate and odds of surgical site infection were observed in the noninstrumented cohort. This is largely inconsistent with the existing literature that has typically reported higher rates of infection in instrumented spinal surgeries^43-46 but may reflect differences in patient selection that were not fully addressed with matching/controlling. Overall, these findings suggest that instrumentation was safely pursued in those for whom it was selected.

In addition, no significant differences were found in revision surgeries over a five-year period between the noninstrumented and instrumented cohorts. Although other studies have conflictingly reported both higher revision surgery rates in instrumented PLFs and no differences in revision surgery rates, this study's findings support the latter. This study found overall revision surgery rates of 8.9% and 9.6% for noninstrumented and instrumented PLFs, respectively. These rates correspond to the lower end of rates reported in literature, which ranged from 5.7% to 64%.^{9,11,13,15,19,22,28}

There are several limitations to this study. For one, the study design was a retrospective cohort using administrative data set, wherein the accuracy of the data is limited to the accuracy of the coding and thoroughness of inclusion and exclusion criteria. Furthermore, patient-reported outcomes were not able to be assessed, and perioperative medical events and revision surgeries were used as surrogates for surgical success.

Overall, for lumbar degenerative spondylolisthesis, noninstrumented fusion represents a small and declining percentage of PLFs over the past decade, with usage that varied with clinical and nonclinical patient characteristics. No differences were found for 90-day aggregated postoperative adverse events, suggesting that instrumentation can be safely performed. Nonetheless, this study also found no difference in 5-year subsequent lumbar surgeries, suggesting that both approaches had similarly durable results. These findings support considering noninstrumented PLFs in select cases when the safety, utility, or cost-benefit of instrumentation is questioned.