Abstract
Study Design
Retrospective Serial Cross-Sectional Study.
Objectives
Robotic assistance in spinal fusion surgery has been associated with improved perioperative outcomes. Given the potential benefits and resource requirements of robotic platforms, it is important to identify adoption trends to inform efforts in surgical resource allocation. This study analyzes trends by age group, race, urban-rural location, primary payer, and ZIP-code income quartile among patients undergoing robot-assisted spinal fusion relative to concurrent trends in free-hand fusion.
Methods
A retrospective serial cross-sectional analysis was performed using the 2010-2021 National Inpatient Sample (NIS). This study analyzed 34,911 robot-assisted and 5.19 million free-hand spinal fusion cases. Survey-weighted Poisson regression estimated subgroup-specific prevalence ratios across sequential three-year eras, and difference-in-differences rate ratios compared changes in the robotic cohort with concurrent changes in the free-hand cohort. An ICD-10-era sensitivity analysis was performed.
Results
Compared with free-hand, robotic representation declined in central-metropolitan counties (DiD RR 0.55, 95% CI 0.46-0.66, p<0.0001) and increased in small-metropolitan counties (RR 2.81, 1.06-7.43, p=0.037) and among Medicaid-insured patients (RR 6.49, 1.02-41.50, p=0.048). Pediatric representation declined (RR 0.23, 0.13-0.42, p<0.0001). In ICD-10-era sensitivity analyses, the central-metropolitan decline and Medicaid finding persisted, whereas the small-metropolitan and pediatric findings were not reproduced.
Conclusions
Robot-assisted spinal fusion demonstrated geographic and payer-related shifts relative to free-hand fusion. Decreased representation in central-metropolitan counties was the most consistent finding. Medicaid representation increased among robotic cases relative to free-hand fusion, but this finding should be interpreted cautiously because the baseline Medicaid count in the robotic cohort was small.
Keywords: robotic surgery, spinal fusion, medicaid, race, income, urban, age, payer
Introduction
Pedicle screw placement remains technically demanding with risks such as malposition, neurovascular injury, and substantial radiation exposure to both the patient and surgical team.1,2 Robotic-assisted spinal surgery arose in the early 2000s to mitigate these complications, with early systems such as SpineAssist and Renaissance demonstrating feasibility for pedicle screw trajectory planning.3,4 Subsequent generations of robotic platforms, such as MazorX, ROSA, and Excelsius, demonstrated increased pedicle screw placement accuracy, better workflow integrations, lower breach rates, reduced neurological complications, and improved reproducibility of results. 5 In accordance with these advancements in robotic spinal surgery, its utilization has increased significantly in the United States since its initial implementation. 6
A substantial body of evidence now supports the effectiveness of these robotic spinal surgery platforms. A meta-analysis of 12 randomized trials by Matur et al demonstrated that robotic-guided screws achieved significantly higher placement accuracy than free-hand or fluoroscopy-guided placements, with no differences in nerve root injuries or revision operations. Robotic surgical platforms were also associated with fewer facet joint violations and other serious perioperative complications. 7 A recent study performed by Quiceno et al demonstrated perfect accuracy in 1126/1156 (97.2%) robotic-assisted pedicle screws as measured by the Gertzbein-Robbins classification, as well as low fluoroscopy time and complication rates. 8 Another systematic review detailed by Pennington et al demonstrated a learning curve in robotic spinal surgery, demonstrating noticeable decreases in operative time after 20-30 cases, with accuracy remaining high even in early adoption. 9 These findings suggest that although a learning curve exists for robotic spinal platforms, this learning curve is surmountable and results in positive changes in surgical accuracy, operative time, complication rates, and radiation exposure.
There exists a historical precedent for disparities in robotic surgery. For example, an analysis performed by Mao et al regarding robotic prostatectomy demonstrated that Hispanic and Black patients were significantly more likely to be treated by surgeons with low robotic volumes or at centers lacking robotic surgical systems. These patients were also more likely to experience postoperative complications, likely as a result of decreased access to advances in surgical technologies. 10 These findings have been supported by several other studies demonstrating decreased diffusion of robotic surgical platforms to underserved populations, with resultant disparities in outcomes.11,12 Recent analyses suggest that similar patterns exist in robotic spine surgery; a 2025 study of 91 robotic spine surgeons found that access was highly clustered, with “hotspots” in the Northeast and Southeast and extensive “coldspots” in rural and Western regions. These hotspots were highly influenced by county-level patient demographics, such as insurance status, race, and median income. 13 Minorities and patients of low socioeconomic status are concurrently less likely to undergo surgery at high-volume technologically advanced centers, which may decrease access to robotic spinal surgeries as diffusion to small centers occurs over time. 14
Considering the increased implementation of separate robotic surgical systems over time, 4 a temporal analysis of robotic spinal fusion diffusion and access patterns is necessary. This study seeks to evaluate how the demographic, payer, and geographic composition of patients undergoing robotic spinal fusion has changed in the United States relative to the concurrent trends in free-hand fusion.
Methods
A retrospective serial cross-sectional analysis was performed using the 2010-2021 National Inpatient Sample (NIS). Two cohorts were defined using ICD-9-CM and ICD-10-PCS codes. The free-hand cohort consisted of patients with free-hand spinal fusion surgeries. This cohort was defined by patients with ICD codes pertaining to spinal fusion procedures and excluded any patients with ICD codes pertaining to robotic procedures or computer-assisted procedures. The robotic cohort consisted of patients with robotic spinal fusion surgeries. This cohort was defined as patients with ICD codes pertaining to spinal fusion procedures and ICD codes pertaining to robotic procedures. Computer-assisted or navigation codes were not used as exclusions from the robotic cohort, as these codes may be used in conjunction with robotic surgery and may be coded alongside robotic assistance.
Each patient sample was grouped into one of four eras: 2010-2012, 2013-2015, 2016-2018, and 2019-2021. The primary analysis of interest was 2019-2021 versus 2010-2012. Furthermore, each patient sample was classified into categories for the following variables: urbanicity, age group, race, primary payer, and ZIP-code income quartile. These categories are based on the NIS provided categories. ICD-10 coding was implemented in October 2015 and introduced more granular robotic-assistance coding; therefore, a sensitivity analysis restricted to the ICD-10 era (2016-2021) was also performed, comparing 2019-2021 with 2016-2018. This sensitivity analysis was intended to evaluate whether key findings were consistent when robotic case identification relied on a single coding system.
Furthermore, NIS discharge weights were used as frequency weights to scale to the entire United States population. To allow for realistic standard errors and to avoid inflating precision, weights were normalized to an average of 1. In lieu of excluding, values that were missing for race, income quartile, and other categorical covariates were categorized as “missing.” For comparing categorical variables between cohorts and across eras, chi-square tests were used to provide unadjusted p-values for differences in proportions.
For each category-variable pair within the free-hand and robotic cohorts, Poisson regression with a log link was used to estimate prevalence ratios. For each subgroup-specific model, a binary indicator of membership in the category of interest was used as the dependent variable. Each model included era and the remaining study covariates other than the index variable, thereby producing cohort-specific adjusted prevalence ratios (aPRs) for the later versus earlier era. When overdispersion was present, negative binomial generalized linear models were used to account for the additional variability. HC2 robust standard errors were used to compute p-values and confidence intervals.
To calculate the difference in change over time between the robotic and the free-hand cohort, the following difference-in-differences (DiD) rate ratio was used:
RRDiD represents the ratio of the aPR from the robotic and free-hand cohorts and indicates whether the change over time differed between cohorts. An RRDiD greater than 1 represents a larger relative increase, or smaller relative decrease, in the robotic cohort than in the free-hand cohort. An RRDiD less than 1 represents a smaller relative increase, or larger relative decrease, in the robotic cohort than in the free-hand cohort. These DiD estimates account for changes occurring during the same period in the free-hand cohort, but they should not be interpreted as results from a single pooled robotic-versus-free-hand interaction model.
It is possible for a change across eras to be significant, but the size of this change may be so small that it is not meaningful. Cohen’s h was used to determine if the effect size is meaningful. Cohen’s h was calculated as:
where p1 and p2 represent the proportions in the earlier and later eras, respectively. The further a value is from 0, the more meaningful the effect size is. 15 A threshold of |h| ≥ 0.2, corresponding to Cohen’s conventional threshold for a small effect size, was used to identify changes of at least small magnitude, allowing statistically significant but negligible changes to be distinguished from changes with a small or greater effect size. Furthermore, the direction of change is inferred from the sign of Cohen’s h.
All analyses were conducted with Python (version 3.13) using the statsmodels package for generalized linear models. This study used de-identified, publicly available NIS data and was deemed exempt from IRB review by the Indiana University Human Research Protection Program (IU HRPP) (protocol #30394).
Results
Free-Hand Spinal Fusion Patients
It was found that 5,191,437 patients underwent free-hand spinal fusion surgeries between 2010 and 2021 (Table 1). These patients experienced statistically significant changes between 2010–2012 and 2019–2021 in most category–variable pairs (Table 2; Figure 1). However, only patients aged 65 and over (aPR 1.19, 95% CI 1.18–1.20, p<0.0001), patients aged 35–49 (aPR 0.64, 0.63–0.65, p<0.0001), those with private insurance as the primary payer (aPR 0.92, 0.91–0.93, p<0.0001), and those with Medicare as the primary payer (aPR 1.01, 1.00–1.02, p=0.0037) experienced shifts that were both statistically significant and of at least small magnitude (Table 2; Figure 1). For patients aged 65 and over, prevalence increased from 29.8% to 42.5%, while patients aged 35–49 decreased from 24.1% to 13.9%; the share of privately insured patients decreased from 46.3% to 36.3%, whereas Medicare representation increased from 35.1% to 45.1% (Table 1; Table 2).
Table 1.
Subgroups of Patients Undergoing Free-hand Spinal Fusion Surgery
| Characteristic | 2010 to 2012 | 2013 to 2015 | 2016 to 2018 | 2019 to 2021 |
|---|---|---|---|---|
| Total Admissions, n (%) | 1,372,942 (100.0%) | 1,350,811 (100.0%) | 1,346,374 (100.0%) | 1,121,310 (100.0%) |
| Age Groups, n (%) | ||||
| — Under 18 | 35,452 (2.6%) | 38,555 (2.9%) | 39,975 (3.0%) | 37,080 (3.3%) |
| — 18 - 34 | 82,334 (6.0%) | 67,810 (5.0%) | 54,525 (4.0%) | 43,330 (3.9%) |
| — 35 - 49 | 330,946 (24.1%) | 268,980 (19.9%) | 222,890 (16.6%) | 156,040 (13.9%) |
| — 50 - 64 | 514,612 (37.5%) | 513,570 (38.0%) | 507,970 (37.7%) | 408,000 (36.4%) |
| — 65+ | 409,282 (29.8%) | 461,835 (34.2%) | 520,970 (38.7%) | 476,845 (42.5%) |
| — Missing | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) |
| Race, n (%) | ||||
| — White | 1,019,238 (74.2%) | 1,022,201 (75.7%) | 1,019,890 (75.8%) | 848,035 (75.6%) |
| — Black | 108,147 (7.9%) | 115,430 (8.5%) | 124,260 (9.2%) | 114,720 (10.2%) |
| — Hispanic | 75,672 (5.5%) | 79,205 (5.9%) | 89,835 (6.7%) | 78,455 (7.0%) |
| — Asian or Pacific Islander | 14,846 (1.1%) | 17,350 (1.3%) | 18,955 (1.4%) | 18,960 (1.7%) |
| — Native American | 5,538 (0.4%) | 6,230 (0.5%) | 6,805 (0.5%) | 6,195 (0.6%) |
| — Other | 31,862 (2.3%) | 32,625 (2.4%) | 33,645 (2.5%) | 28,065 (2.5%) |
| — Missing | 117,640 (8.6%) | 77,770 (5.8%) | 52,985 (3.9%) | 26,880 (2.4%) |
| Patient Location: NCHS Urban-Rural Code, n (%) | ||||
| — “Central” metro counties >=1 million population | 318,448 (23.2%) | 316,380 (23.4%) | 328,565 (24.4%) | 279,020 (24.9%) |
| — “Fringe” metro counties >=1 million population | 336,556 (24.5%) | 333,900 (24.7%) | 341,785 (25.4%) | 294,940 (26.3%) |
| — Metro counties of 250,000-999,999 population | 277,425 (20.2%) | 289,840 (21.5%) | 289,535 (21.5%) | 237,005 (21.1%) |
| — Metro counties of 50,000-249,999 population | 140,367 (10.2%) | 147,100 (10.9%) | 138,970 (10.3%) | 110,095 (9.8%) |
| — Micropolitan counties | 170,703 (12.4%) | 151,470 (11.2%) | 140,230 (10.4%) | 114,320 (10.2%) |
| — Not metropolitan or micropolitan counties | 113,008 (8.2%) | 107,550 (8.0%) | 103,990 (7.7%) | 82,705 (7.4%) |
| — Missing | 16,436 (1.2%) | 4,570 (0.3%) | 3,300 (0.2%) | 3,225 (0.3%) |
| Primary Payer, n (%) | ||||
| — Medicare | 481,263 (35.1%) | 533,485 (39.5%) | 581,055 (43.2%) | 505,915 (45.1%) |
| — Medicaid | 93,430 (6.8%) | 110,760 (8.2%) | 128,495 (9.5%) | 110,845 (9.9%) |
| — Private insurance | 636,398 (46.4%) | 569,925 (42.2%) | 518,165 (38.5%) | 407,145 (36.3%) |
| — Self-pay | 20,772 (1.5%) | 17,890 (1.3%) | 17,315 (1.3%) | 17,475 (1.6%) |
| — No charge | 2,339 (0.2%) | 1,930 (0.1%) | 1,295 (0.1%) | 1,235 (0.1%) |
| — Other | 135,042 (9.8%) | 114,785 (8.5%) | 98,180 (7.3%) | 77,425 (6.9%) |
| — Missing | 3,699 (0.3%) | 2,035 (0.2%) | 1,870 (0.1%) | 1,270 (0.1%) |
| Median household income for patient’s ZIP Code, n (%) | ||||
| — 0 to 25th percentile | 321,152 (23.4%) | 319,190 (23.6%) | 321,640 (23.9%) | 271,710 (24.2%) |
| — 26th to 50th percentile | 347,029 (25.3%) | 352,725 (26.1%) | 352,965 (26.2%) | 286,280 (25.5%) |
| — 51st to 75th percentile | 357,497 (26.0%) | 348,575 (25.8%) | 345,850 (25.7%) | 285,395 (25.5%) |
| — 76th to 100th percentile | 319,458 (23.3%) | 303,765 (22.5%) | 303,335 (22.5%) | 261,335 (23.3%) |
| — Missing | 27,806 (2.0%) | 26,555 (2.0%) | 22,585 (1.7%) | 16,590 (1.5%) |
Table 1 presents subgroups of patients undergoing free-hand spinal fusion surgery between 2010 and 2021
Table 2.
Analysis of Trends for Patients Undergoing Free-hand Spinal Fusion Surgery
| Characteristic | aPR | 95% CI | P Value | Prevalence 2010 – 2012 (%) | Prevalence 2019 – 2021 (%) | Absolute change (pp) | Relative change (%) | Cohen's h |
|---|---|---|---|---|---|---|---|---|
| Total Admissions, n (%) | ||||||||
| Age Groups, n (%) | ||||||||
| — Under 18 | 1.37 | [1.32, 1.42] | <0.0001 | 2.6% | 3.3% | +0.7 pp | 28.1% | 0.040 |
| — 18 - 34 | 0.71 | [0.69, 0.73] | <0.0001 | 6.0% | 3.9% | -2.1 pp | -35.6% | 0.100 |
| — 35 - 49 | 0.64 | [0.63, 0.65] | <0.0001 | 24.1% | 13.9% | -10.2 pp | -42.3% | 0.260 |
| — 50 - 64 | 1.07 | [1.06, 1.08] | <0.0001 | 37.5% | 36.4% | -1.1 pp | -2.9% | 0.020 |
| — 65+ | 1.19 | [1.18, 1.20] | <0.0001 | 29.8% | 42.5% | +12.7 pp | 42.7% | 0.270 |
| Race, n (%) | ||||||||
| — White | 1.02 | [1.01, 1.03] | <0.0001 | 74.2% | 75.6% | +1.4 pp | 1.9% | 0.030 |
| — Black | 1.24 | [1.22, 1.27] | <0.0001 | 7.9% | 10.2% | +2.4 pp | 29.9% | 0.080 |
| — Hispanic | 1.27 | [1.24, 1.30] | <0.0001 | 5.5% | 7.0% | +1.5 pp | 27.0% | 0.060 |
| — Asian or Pacific Islander | 1.45 | [1.38, 1.52] | <0.0001 | 1.1% | 1.7% | +0.6 pp | 56.4% | 0.050 |
| — Native American | 1.48 | [1.36, 1.61] | <0.0001 | 0.4% | 0.6% | +0.1 pp | 37.0% | 0.020 |
| — Other | 1.11 | [1.07, 1.15] | <0.0001 | 2.3% | 2.5% | +0.2 pp | 7.9% | 0.010 |
| Patient Location: NCHS Urban-Rural Code, n (%) | ||||||||
| — “Central” metro counties >=1 million population | 1.01 | [1.00, 1.02] | 0.1885 | 23.2% | 24.9% | +1.7 pp | 7.3% | 0.040 |
| — “Fringe” metro counties >=1 million population | 1.08 | [1.07, 1.09] | <0.0001 | 24.5% | 26.3% | +1.8 pp | 7.3% | 0.040 |
| — Metro counties of 250,000-999,999 population | 1.05 | [1.04, 1.06] | <0.0001 | 20.2% | 21.1% | +0.9 pp | 4.6% | 0.020 |
| — Metro counties of 50,000-249,999 population | 0.98 | [0.96, 1.00] | 0.0375 | 10.2% | 9.8% | -0.4 pp | -3.9% | 0.010 |
| — Micropolitan counties | 0.85 | [0.84, 0.87] | <0.0001 | 12.4% | 10.2% | -2.2 pp | -18.0% | 0.070 |
| — Not metropolitan or micropolitan counties | 0.96 | [0.94, 0.98] | <0.0001 | 8.2% | 7.4% | -0.9 pp | -10.4% | 0.030 |
| Primary Payer, n (%) | ||||||||
| — Medicare | 1.01 | [1.00, 1.02] | 0.0037 | 35.1% | 45.1% | +10.1 pp | 28.7% | 0.210 |
| — Medicaid | 1.72 | [1.69, 1.76] | <0.0001 | 6.8% | 9.9% | +3.1 pp | 45.3% | 0.110 |
| — Private insurance | 0.92 | [0.91, 0.93] | <0.0001 | 46.3% | 36.3% | -10.0 pp | -21.6% | 0.200 |
| — Self-pay | 1.22 | [1.17, 1.28] | <0.0001 | 1.5% | 1.6% | +0.0 pp | 3.0% | 0.000 |
| — No charge | 0.74 | [0.63, 0.86] | 0.0001 | 0.2% | 0.1% | -0.1 pp | -35.2% | 0.020 |
| — Other | 0.82 | [0.81, 0.84] | <0.0001 | 9.8% | 6.9% | -2.9 pp | -29.8% | 0.110 |
| Median household income for patient’s ZIP Code, n (%) | ||||||||
| — 0 to 25th percentile | 1.02 | [1.00, 1.03] | 0.0111 | 23.4% | 24.2% | +0.8 pp | 3.6% | 0.020 |
| — 26th to 50th percentile | 1.03 | [1.02, 1.05] | <0.0001 | 25.3% | 25.5% | +0.3 pp | 1.0% | 0.010 |
| — 51st to 75th percentile | 0.98 | [0.97, 0.99] | <0.0001 | 26.0% | 25.5% | -0.6 pp | -2.2% | 0.010 |
| — 76th to 100th percentile | 0.98 | [0.97, 1.00] | 0.0081 | 23.3% | 23.3% | +0.0 pp | 0.2% | 0.000 |
Table 2 provides statistical analysis of subgroups of patients undergoing free-hand spinal fusion surgery. The aPR denotes the adjusted prevalence ratio for each subgroup, with the Confidence Interval giving the corresponding 95% CI and P value from the regression model. Prevalence 2010–2012 (%) and Prevalence 2019–2021 (%) report the percentage of all free-hand spinal fusions occurring in each subgroup during the early and late eras, respectively. Absolute Change (pp) is the difference in prevalence between eras expressed in percentage points, and Relative Change (%) is the percent change relative to 2010–2012 prevalence. Cohen’s h provides the standardized effect size for the change in proportions between eras.
Figure 1.
Longitudinal Trends in Subgroups of Patients Undergoing Free-hand Spinal Fusion. Forest plot of adjusted prevalence ratios (aPRs) for changes in free-hand spinal fusion utilization by patient subgroup, comparing 2019–2021 with 2010–2012. Each diamond represents the aPR for a given category, with horizontal lines indicating 95% confidence intervals. The vertical dashed line at 1.0 denotes no change in relative utilization between eras. Blue highlighting indicates a subgroup with a statistically significant change between eras (P < 0.05).
Robotic Spinal Fusion Patients
The robotic cohort had 34,911 patients between 2010 and 2021 (Table 3). Multiple categories experienced meaningful and statistically significant shifts, including patients aged less than 18 years (aPR 0.32, 0.18–0.58, p=0.0002), those with private insurance as the primary payer (aPR 0.86, 0.75–0.98, p=0.0233), and those with Medicaid as the primary payer (aPR 11.18, 1.75–71.49, p=0.0107), as well as patients residing in small metropolitan counties with populations between 50,000 and 249,999 (aPR 2.76, 1.04–7.29, p=0.0409) and those in central metropolitan counties with populations ≥1 million (aPR 0.55, 0.46–0.67, p<0.0001) (Table 4; Figure 2). Correspondingly, the prevalence of pediatric robotic cases decreased from 7.4% to 2.1%, private insurance decreased from 53.4% to 39.0%, Medicaid increased from 0.8% to 6.0%, and the share of cases in small metropolitan counties increased from 3.2% to 8.7% (Table 3; Table 4).
Table 3.
Subgroups of Patients Undergoing Robotic Spinal Fusion Surgery
| Characteristic | 2010 to 2012 | 2013 to 2015 | 2016 to 2018 | 2019 to 2021 |
|---|---|---|---|---|
| Total Admissions, n (%) | 621 (100.0%) | 2,625 (100.0%) | 8,600 (100.0%) | 23,065 (100.0%) |
| Age Groups, n (%) | ||||
| — Under 18 | 46 (7.4%) | 140 (5.3%) | 185 (2.2%) | 485 (2.1%) |
| — 18 - 34 | 32 (5.1%) | 125 (4.8%) | 365 (4.2%) | 785 (3.4%) |
| — 35 - 49 | 96 (15.4%) | 310 (11.8%) | 1,175 (13.7%) | 2,965 (12.9%) |
| — 50 - 64 | 224 (36.0%) | 750 (28.6%) | 3,015 (35.1%) | 8,030 (34.8%) |
| — 65+ | 224 (36.1%) | 1,300 (49.5%) | 3,860 (44.9%) | 10,800 (46.8%) |
| — Missing | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) |
| Race, n (%) | ||||
| — White | 475 (76.5%) | 2,050 (78.1%) | 7,125 (82.8%) | 18,240 (79.1%) |
| — Black | 30 (4.8%) | 170 (6.5%) | 595 (6.9%) | 1,835 (8.0%) |
| — Hispanic | 62 (9.9%) | 165 (6.3%) | 430 (5.0%) | 1,490 (6.5%) |
| — Asian or Pacific Islander | 16 (2.5%) | 40 (1.5%) | 75 (0.9%) | 380 (1.6%) |
| — Native American | 0 (0.0%) | 0 (0.0%) | 25 (0.3%) | 50 (0.2%) |
| — Other | 15 (2.4%) | 70 (2.7%) | 180 (2.1%) | 595 (2.6%) |
| — Missing | 24 (3.9%) | 130 (5.0%) | 170 (2.0%) | 475 (2.1%) |
| Patient Location: NCHS Urban-Rural Code, n (%) | ||||
| — “Central” metro counties >=1 million population | 291 (46.9%) | 1,140 (43.4%) | 2,700 (31.4%) | 5,925 (25.7%) |
| — “Fringe” metro counties >=1 million population | 124 (20.0%) | 505 (19.2%) | 2,205 (25.6%) | 6,470 (28.1%) |
| — Metro counties of 250,000-999,999 population | 131 (21.1%) | 465 (17.7%) | 1,495 (17.4%) | 4,955 (21.5%) |
| — Metro counties of 50,000-249,999 population | 20 (3.2%) | 210 (8.0%) | 820 (9.5%) | 2,015 (8.7%) |
| — Micropolitan counties | 40 (6.4%) | 180 (6.9%) | 880 (10.2%) | 2,350 (10.2%) |
| — Not metropolitan or micropolitan counties | 15 (2.4%) | 110 (4.2%) | 480 (5.6%) | 1,285 (5.6%) |
| — Missing | 0 (0.0%) | 15 (0.6%) | 20 (0.2%) | 65 (0.3%) |
| Primary Payer, n (%) | ||||
| — Medicare | 215 (34.6%) | 1,340 (51.0%) | 4,040 (47.0%) | 10,810 (46.9%) |
| — Medicaid | 5 (0.8%) | 105 (4.0%) | 405 (4.7%) | 1,380 (6.0%) |
| — Private insurance | 332 (53.4%) | 1,050 (40.0%) | 3,435 (39.9%) | 8,990 (39.0%) |
| — Self-pay | 4 (0.7%) | 10 (0.4%) | 35 (0.4%) | 165 (0.7%) |
| — No charge | 0 (0.0%) | 5 (0.2%) | 25 (0.3%) | 15 (0.1%) |
| — Other | 55 (8.9%) | 115 (4.4%) | 655 (7.6%) | 1,665 (7.2%) |
| — Missing | 10 (1.6%) | 0 (0.0%) | 5 (0.1%) | 40 (0.2%) |
| Median household income for patient’s ZIP Code, n (%) | ||||
| — 0 to 25th percentile | 76 (12.2%) | 450 (17.1%) | 1,860 (21.6%) | 4,755 (20.6%) |
| — 26th to 50th percentile | 145 (23.4%) | 645 (24.6%) | 2,060 (24.0%) | 5,470 (23.7%) |
| — 51st to 75th percentile | 189 (30.4%) | 835 (31.8%) | 2,580 (30.0%) | 6,150 (26.7%) |
| — 76th to 100th percentile | 211 (34.0%) | 635 (24.2%) | 1,905 (22.2%) | 6,370 (27.6%) |
| — Missing | 0 (0.0%) | 60 (2.3%) | 195 (2.3%) | 320 (1.4%) |
Table 3 presents subgroups of patients undergoing robotic spinal fusion surgery between 2010 and 2021.
Table 4.
Analysis of Trends for Patients Undergoing Robotic Spinal Fusion Surgery
| Characteristic | aPR | 95% CI | P Value | Prevalence 2010 – 2012 (%) | Prevalence 2019 – 2021 (%) | Absolute change (pp) | Relative change (%) | Cohen's h |
|---|---|---|---|---|---|---|---|---|
| Total Admissions, n (%) | ||||||||
| Age Groups, n (%) | ||||||||
| — Under 18 | 0.32 | [0.18, 0.58] | 0.0002 | 7.4% | 2.1% | -5.3 pp | -71.6% | 0.260 |
| — 18 - 34 | 0.77 | [0.36, 1.66] | 0.5104 | 5.1% | 3.4% | -1.7 pp | -33.0% | 0.080 |
| — 35 - 49 | 0.91 | [0.61, 1.37] | 0.6515 | 15.4% | 12.9% | -2.6 pp | -16.8% | 0.070 |
| — 50 - 64 | 1.09 | [0.88, 1.36] | 0.4261 | 36.0% | 34.8% | -1.2 pp | -3.2% | 0.020 |
| — 65+ | 1.11 | [0.96, 1.29] | 0.1558 | 36.1% | 46.8% | +10.7 pp | 29.8% | 0.220 |
| Race, n (%) | ||||||||
| — White | 0.98 | [0.90, 1.07] | 0.6541 | 76.5% | 79.1% | +2.6 pp | 3.4% | 0.060 |
| — Black | 1.83 | [0.85, 3.91] | 0.1200 | 4.8% | 8.0% | +3.1 pp | 64.7% | 0.130 |
| — Hispanic | 0.82 | [0.49, 1.39] | 0.4654 | 9.9% | 6.5% | -3.4 pp | -34.8% | 0.130 |
| — Asian or Pacific Islander | 1.07 | [0.35, 3.24] | 0.9079 | 2.5% | 1.6% | -0.9 pp | -34.0% | 0.060 |
| — Native American | NE | NE | NE | NE | NE | NE | NE | NE |
| — Other | 1.51 | [0.50, 4.52] | 0.4616 | 2.4% | 2.6% | +0.2 pp | 8.2% | 0.010 |
| Patient Location: NCHS Urban-Rural Code, n (%) | ||||||||
| — “Central” metro counties >=1 million population | 0.55 | [0.46, 0.67] | <0.0001 | 46.9% | 25.7% | -21.2 pp | -45.2% | 0.440 |
| — “Fringe” metro counties >=1 million population | 1.54 | [1.08, 2.20] | 0.0173 | 20.0% | 28.1% | +8.1 pp | 40.4% | 0.190 |
| — Metro counties of 250,000-999,999 population | 1.00 | [0.71, 1.40] | 0.9965 | 21.1% | 21.5% | +0.4 pp | 1.8% | 0.010 |
| — Metro counties of 50,000-249,999 population | 2.76 | [1.04, 7.29] | 0.0409 | 3.2% | 8.7% | +5.5 pp | 171.3% | 0.240 |
| — Micropolitan counties | 1.19 | [0.62, 2.26] | 0.6012 | 6.4% | 10.2% | +3.8 pp | 58.6% | 0.140 |
| — Not metropolitan or micropolitan counties | 1.71 | [0.58, 5.05] | 0.3325 | 2.4% | 5.6% | +3.2 pp | 130.7% | 0.160 |
| Primary Payer, n (%) | ||||||||
| — Medicare | 1.09 | [0.94, 1.28] | 0.2577 | 34.6% | 46.9% | +12.3 pp | 35.6% | 0.250 |
| — Medicaid | 11.18 | [1.75, 71.49] | 0.0107 | 0.8% | 6.0% | +5.2 pp | 643.3% | 0.310 |
| — Private insurance | 0.86 | [0.75, 0.98] | 0.0233 | 53.4% | 39.0% | -14.4 pp | -27.0% | 0.290 |
| — Self-pay | 1.15 | [0.13, 10.22] | 0.9027 | 0.7% | 0.7% | +0.0 pp | 1.3% | 0.000 |
| — No charge | NE | NE | NE | NE | NE | NE | NE | NE |
| — Other | 0.88 | [0.51, 1.52] | 0.6465 | 8.9% | 7.2% | -1.7 pp | -19.0% | 0.060 |
| Median household income for patient’s ZIP Code, n (%) | ||||||||
| — 0 to 25th percentile | 1.30 | [0.83, 2.03] | 0.2575 | 12.2% | 20.6% | +8.5 pp | 69.6% | 0.230 |
| — 26th to 50th percentile | 0.95 | [0.70, 1.30] | 0.7636 | 23.4% | 23.7% | +0.3 pp | 1.4% | 0.010 |
| — 51st to 75th percentile | 0.95 | [0.73, 1.23] | 0.6919 | 30.4% | 26.7% | -3.7 pp | -12.3% | 0.080 |
| — 76th to 100th percentile | 0.96 | [0.76, 1.21] | 0.7065 | 34.0% | 27.6% | -6.4 pp | -18.9% | 0.140 |
Table 4 provides statistical analysis of subgroups of patients undergoing robotic spinal fusion surgery. The aPR denotes the adjusted prevalence ratio for each subgroup, with the Confidence Interval giving the corresponding 95% CI and P value from the regression model. Prevalence 2010–2012 (%) and Prevalence 2019–2021 (%) report the percentage of all robotic spinal fusions occurring in each subgroup during the early and late eras, respectively. Absolute Change (pp) is the difference in prevalence between eras expressed in percentage points, and Relative Change (%) is the percent change relative to 2010–2012 prevalence. Cohen’s h provides the standardized effect size for the change in proportions between eras. NE indicates values that were not estimable due to zero counts in the 2010–2012 reference period.
Figure 2.
Longitudinal Trends in Subgroups of Patients Undergoing Robotic Spinal Fusion. Forest plot of adjusted prevalence ratios (aPRs) for changes in robotic spinal fusion utilization by patient subgroup, comparing 2019–2021 with 2010–2012. Each diamond represents the aPR for a given category, with horizontal lines indicating 95% confidence intervals. The vertical dashed line at 1.0 denotes no change in relative utilization between eras. Blue highlighting indicates a subgroup with a statistically significant change between eras (P < 0.05).
Difference in Differences
After accounting for the changes that occurred within the free-hand cohort, one of the previously meaningful and significant changes was no longer statistically significant (Table 5; Figure 3). Robotic patients with private insurance as the primary payer had an aPR of 0.86 (95% CI 0.75–0.98; p=0.0233) prior to accounting for the free-hand cohort (Table 4; Figure 2). After accounting for the free-hand cohort, this category had a non-significant DiD rate ratio of 0.93 across eras (0.82–1.06; p=0.2967) (Table 5; Figure 3). After accounting for changes occurring during the same study period in the free-hand cohort, the main statistically significant DiD findings included Medicaid primary payer, small metropolitan counties, central metropolitan counties, and patients aged less than 18 years. Patients with Medicaid as the primary payer had a DiD rate ratio of 6.49 (1.02–41.50; p=0.0481). Patients in small metropolitan counties had a DiD rate ratio of 2.81 (1.06–7.43; p=0.0373). Those located within central metropolitan counties had a DiD rate ratio of 0.55 (0.46–0.66; p<0.0001). Pediatric patients had a DiD rate ratio of 0.23 (0.13–0.42; p<0.0001) (Table 5; Figure 3). The primary analysis did not show significant differences in the change over time by race or ZIP-code income quartile between robotic and free-hand fusion.
Table 5.
Difference-in-Differences Analysis of Robotic Versus Free-Hand Spinal Fusion Surgery
| Primary analysis | ICD-10-Era sensitivity analysis | |||||
|---|---|---|---|---|---|---|
| Characteristic | Rate ratio | 95% CI | P Value | Rate ratio | 95% CI | P Value |
| Total Admissions, n (%) | ||||||
| Age Groups, n (%) | ||||||
| — Under 18 | 0.23 | [0.13, 0.42] | <0.0001 | 0.81 | [0.58, 1.13] | 0.2120 |
| — 18 - 34 | 1.09 | [0.51, 2.33] | 0.8326 | | | |
| — 35 - 49 | 1.41 | [0.94, 2.13] | 0.0956 | | | |
| — 50 - 64 | 1.02 | [0.82, 1.27] | 0.8273 | | | |
| — 65+ | 0.94 | [0.81, 1.09] | 0.3886 | | | |
| Race, n (%) | ||||||
| — White | 0.96 | [0.88, 1.05] | 0.3606 | | | |
| — Black | 1.47 | [0.69, 3.15] | 0.3201 | | | |
| — Hispanic | 0.65 | [0.38, 1.10] | 0.1072 | | | |
| — Asian or Pacific Islander | 0.74 | [0.24, 2.24] | 0.5927 | | | |
| — Native American | NE | NE | NE | | | |
| — Other | 1.36 | [0.45, 4.08] | 0.5809 | | | |
| Patient Location: NCHS Urban-Rural Code, n (%) | ||||||
| — “Central” metro counties >=1 million population | 0.55 | [0.46, 0.66] | <0.0001 | 0.78 | [0.71, 0.84] | <0.0001 |
| — “Fringe” metro counties >=1 million population | 1.43 | [1.00, 2.04] | 0.0498 | | | |
| — Metro counties of 250,000-999,999 population | 0.95 | [0.68, 1.33] | 0.7771 | | | |
| — Metro counties of 50,000-249,999 population | 2.81 | [1.06, 7.43] | 0.0373 | 1.03 | [0.87, 1.22] | 0.7193 |
| — Micropolitan counties | 1.39 | [0.73, 2.64] | 0.3168 | | | |
| — Not metropolitan or micropolitan counties | 1.78 | [0.60, 5.27] | 0.2966 | | | |
| Primary Payer, n (%) | ||||||
| — Medicare | 1.08 | [0.92, 1.26] | 0.3340 | | | |
| — Medicaid | 6.49 | [1.02, 41.50] | 0.0481 | 1.31 | [1.05, 1.65] | 0.0184 |
| — Private insurance | 0.93 | [0.82, 1.06] | 0.2967 | | | |
| — Self-pay | 0.94 | [0.11, 8.37] | 0.9541 | | | |
| — No charge | NE | NE | NE | | | |
| — Other | 1.07 | [0.62, 1.85] | 0.8126 | | | |
| Median household income for patient’s ZIP Code, n (%) | ||||||
| — 0 to 25th percentile | 1.28 | [0.81, 2.00] | 0.2860 | | | |
| — 26th to 50th percentile | 0.92 | [0.68, 1.26] | 0.6078 | | | |
| — 51st to 75th percentile | 0.97 | [0.74, 1.26] | 0.8172 | | | |
| — 76th to 100th percentile | 0.97 | [0.77, 1.23] | 0.8068 | | | |
Table 5 provides difference-in-differences rate ratios for robotic versus free-hand spinal fusion utilization by patient subgroup. The primary analysis compares 2019–2021 with 2010–2012. The ICD-10-era sensitivity analysis compares 2019–2021 with 2016–2018 and is shown for selected key findings to assess whether results were consistent when robotic case identification relied on a single coding system. The Rate Ratio represents the ratio of prevalence ratios for robotic vs free-hand fusion. A Rate Ratio greater than 1 indicates that the subgroup increased more, or decreased less, in the robotic cohort than in the free-hand cohort. A Rate Ratio less than 1 indicates that the subgroup increased less, or decreased more, in the robotic cohort than in the free-hand cohort. Confidence Interval gives the corresponding 95% confidence interval for the Rate Ratio. The P Value is derived from the difference-in-differences regression model. Blank cells indicate that ICD-10-era sensitivity estimates were not displayed for that subgroup. NE indicates values that were not estimable due to zero counts in the 2010–2012 reference period.
Figure 3.
Difference-in-Differences Ratios for Robotic vs Free-hand Spinal Fusion Utilization Across Patient Subgroups. Difference-in-differences forest plot of robotic versus free-hand spinal fusion utilization by patient subgroup. Each diamond represents the DiD rate ratio (ratio of prevalence ratios for robotic vs free-hand fusion between 2019–2021 and 2010–2012), with horizontal lines indicating 95% confidence intervals. The vertical dashed line at 1.0 denotes no differential change in utilization between robotic and free-hand cohorts across eras. Values to the right of 1.0 indicate that the subgroup increased more, or decreased less, in the robotic cohort than in the free-hand cohort. Values to the left of 1.0 indicate that the subgroup increased less, or decreased more, in the robotic cohort than in the free-hand cohort. Blue highlighting indicates a subgroup with a statistically significant difference-in-differences rate ratio (P < 0.05).
Sensitivity Analysis: ICD-10 Era
To evaluate whether the primary findings were sensitive to the ICD-9 to ICD-10 coding transition, a sensitivity analysis restricted to the ICD-10 era was performed. In this analysis, the decline in central metropolitan robotic representation persisted (DiD RR 0.78, CI 0.71–0.84; p<0.0001), and Medicaid representation remained increased in the robotic cohort relative to free-hand fusion, although the magnitude was less pronounced (DiD RR 1.31, CI 1.05–1.65; p=0.0184). In contrast, the relative increase observed in small metropolitan counties and the relative decline observed among pediatric patients in the primary analysis were not reproduced in ICD-10-era analyses (small metropolitan counties: DiD RR 1.03, CI 0.87–1.22; p=0.7192; pediatric patients: DiD RR 0.81, CI 0.58–1.13; p=0.2120).
Discussion
The analysis of NIS data from 2010-2021 demonstrated that the United States robot-assisted spinal fusion population has undergone geographic-, payer-, and age-related shifts. The most consistent finding was a relative decline in robotic utilization within central metropolitan counties compared with free-hand fusion. In the primary analysis, robotic utilization also showed relative increases in small metropolitan counties and among Medicaid-insured patients, as well as a relative decline among pediatric patients. However, sensitivity analyses restricted to the ICD-10 era demonstrated that the central metropolitan decline and the Medicaid finding persisted, whereas the small metropolitan and pediatric findings were not reproduced. These findings suggest that the central metropolitan decline was the most consistent geographic change, rather than showing a shift to one specific non-central metropolitan category. The Medicaid finding also persisted in ICD-10-era analyses, but payer-related findings should still be interpreted with greater caution because of the small baseline Medicaid count in the robotic cohort. The primary analysis did not show significant differences in change over time by race or ZIP-code income quartile between robotic and free-hand fusion.
Geographic Diffusion
The decline in robot-assisted fusion patients in central metropolitan areas (DiD rate ratio of 0.55) and the increase in robot-assisted fusion patients in small metropolitan counties in the primary analysis (DiD rate ratio of 2.81) are unsurprising. Large academic surgery centers, which are predominantly located in central metropolitan areas, have access to the large initial capital investment required for robotic surgery through revenue that has been both substantial and sustainable across decades. 16 The approximately $1 to $2.5 million capital outlay for a single unit, the need for at least 100 to 150 cases per year to reach the break-even point, and the comparatively lower profit margin of robotic procedures may make adoption a poor financial risk for lower-revenue centers in smaller metropolitan areas.17-19 Despite this, past research has shown that local robotic competition drives the growth of robotic technology.20,21 This robotic technology growth in competitive markets, along with the declining share of robotic fusion cases occurring in central metropolitan counties, may be the result of hospitals attempting to retain market share.
Importantly, the decline in central metropolitan representation persisted in ICD-10-era analyses (DiD rate ratio 0.78), supporting that this geographic finding is not solely explained by the ICD-9 to ICD-10 coding transition. In contrast, the relative increase in small metropolitan counties was not reproduced in ICD-10-era analyses. Instead, ICD-10-era analyses showed a relative increase in metropolitan counties with populations of 250,000 to 999,999. Taken together, these findings support diffusion away from the most central metropolitan settings, but they do not establish that the primary destination of diffusion was specifically the smallest metropolitan tier or another specific non-central metropolitan category.
Medicaid Expansion
The 2014 implementation of Affordable Care Act coverage expansion policies included the option for states to expand Medicaid eligibility to all adults with incomes up to 138% of the federal poverty level. Prior to 2014, Medicaid eligibility for adults generally required meeting categorical criteria – such as being pregnant, a child, a parent of a dependent child, disabled, or aged 65 years or older – in addition to income requirements.22,23 As expected, following the expansion of eligible individuals, Medicaid utilization increased in both the free-hand and robotic cohorts. However, the increase was disproportionately greater among patients undergoing robot-assisted spinal fusion compared with free-hand fusion (DiD rate ratio 6.49). Although robotic surgery use in central metropolitan counties declined relative to free-hand fusion, robotic spinal fusion remained highly concentrated in large metropolitan areas, with over half (53.8%) of cases performed in counties located within metropolitan areas with populations exceeding 1 million. Academic health centers—many of which are located in these large metropolitan areas—provide over 40% of inpatient care for Medicaid beneficiaries. 24 The combination of Medicaid expansion and the predilection for Medicaid patients to be treated at academic health centers may have contributed to the greater relative increase in Medicaid patients undergoing robotic fusion compared with free-hand fusion.
This payer finding should be interpreted cautiously. In the primary analysis, the Medicaid DiD estimate was based on a small Medicaid baseline count in the robotic cohort, which produced a wide confidence interval (1.02–41.50). Nevertheless, the direction of the Medicaid finding persisted in ICD-10-era analyses, where the estimate was less pronounced but more precise (DiD rate ratio 1.31, 95% CI 1.05–1.65). This pattern suggests that increased Medicaid representation in robotic fusion may be present, but the magnitude of the increase in the full-period model should not be overinterpreted given the small baseline Medicaid count and wide confidence interval.
Interestingly, Medicaid utilization, which is available to low-income populations, increased more in the robotic cohort than in the free-hand cohort, while patients in the lowest ZIP-code income quartile did not show an increase. One of the reasons believed to cause this is the high percentage of patients 50 and older that make up the demographic that undergo spinal fusion (78.9% in the free-hand cohort) and the lower income of seniors. Seniors have higher net worth but lower income than younger, pre-retirement adults. 25 This phenomenon may play a large role in why Medicaid expansion occurred in the robotic cohort without a relative increase in patients in the lowest income quartile. Studies in the USA, Canada, Europe, Japan, and elsewhere have previously shown income to be a strong correlate of hospital utilization in working-class adults, but a weak correlate in seniors, which is the predominant population for spinal fusion. 26
Pediatric Decline
Pediatric patients undergoing robotic spinal fusion surgery showed a large, significant relative decline in comparison to the free-hand cohort in the primary analysis (DiD rate ratio 0.23). One possible explanation is that early robotic adoption occurred in a limited number of centers whose case mix included a somewhat larger pediatric proportion than that seen later, after robotic diffusion broadened. Another possibility is that pediatric robotic use is more sensitive to the economic constraints of platform acquisition and maintenance than adult robotic use. As previously noted, the high fixed costs of robotic platforms require substantial annual case volume to offset. With pediatric cases only comprising 3.3% of free-hand spinal fusion cases, pediatric hospitals may face difficulty generating sufficient case volume to justify these costs. Furthermore, the low utilization rate of robotics in pediatric spinal fusion may reduce competitive pressure to adopt robotic platforms. However, this pediatric finding did not persist in ICD-10-era analyses. Therefore, the apparent pediatric decline should be interpreted as specific to the full-period analysis rather than a consistent finding across analyses.
Race and Socioeconomic Disparities
Extensive documentation of racial disparities exists in spinal surgery.27-30 However, no significant changes were observed in the racial composition between the robotic and free-hand cohorts in the primary analysis. Therefore, this study does not provide evidence that robotic diffusion meaningfully worsened or meaningfully reduced pre-existing disparities. This is consistent with previous findings that the lumbar spinal fusion surgery method used is generally similar between racial groups. 31
ZIP-code income quartile also did not display consistent changes across analyses. This likely reflects heterogeneity within individual ZIP codes—for example, adjacent luxury and affordable housing, a high proportion of retirees with low reported income but higher net worth, or PO-box addresses used by individuals without a fixed residence (such as those experiencing homelessness or working as migrant laborers). 32 These issues may limit the ability of ZIP-code income quartile to capture individual-level socioeconomic differences in robotic fusion access.
Contemporary Relevance
This analysis ends in 2021, and the current distribution of robotic fusion may differ from the patterns observed here as platform adoption has continued. For spine surgeons and health systems, the present findings highlight geographic shifts in robotic diffusion and suggest that capital feasibility may be relevant when interpreting adoption patterns.
Limitations
This study has several limitations. The findings depend on the accuracy and completeness of the ICD coding used within the NIS database. Any miscoding, undercoding, or changes in coding practices may affect results. The transition from ICD-9 to ICD-10 may also have altered the detectability of robotic cases over time. Although sensitivity analyses were performed to address this concern, coding-related bias cannot be completely excluded. Furthermore, the NIS database is an inpatient data sample. Any robotic spine procedures performed in an ambulatory surgery or outpatient setting are not represented in the findings. If some spinal fusion procedures migrated from inpatient hospitals to ambulatory or outpatient settings during the study period, the later inpatient eras may capture a different patient population than the overall population undergoing spinal fusion. This could affect observed inpatient payer and demographic trends because patients captured in the inpatient setting may differ from patients treated in outpatient or ambulatory settings. The NIS dataset is a serial cross-sectional dataset and does not allow the tracking of individual patients, surgeons, or hospitals over time. Additional factors not provided include surgeon volume, local robotic availability, and brand of robotic system – factors that may influence geographic and payer shifts. Prior studies have reported increased use of computer-assisted navigation in spinal fusion over time,33,34 and excluding coded navigation-only cases may have changed the composition of the non-robotic comparison group across eras. Socioeconomic analysis was limited by the use of ZIP-code income quartiles, which are weakly associated with individual income. Missing categorical data were retained as “missing” categories, but missingness in NIS variables may reflect hospital- or state-level reporting practices and should be considered when interpreting subgroup findings. Finally, unknown changes in referral patterns, reimbursement, and case selection may influence the findings.
Conclusions
This retrospective serial cross-sectional analysis of 34,911 robot-assisted spinal fusion patients and 5,191,437 free-hand spinal fusion patients demonstrates that the most consistent change in robot-assisted spinal fusion utilization was a decline in representation within central metropolitan counties relative to free-hand fusion. In the primary analysis, robotic utilization also showed greater relative representation in small metropolitan counties and among Medicaid-insured patients, as well as lower relative representation among pediatric patients. However, in ICD-10-era sensitivity analyses, the central metropolitan decline and Medicaid finding persisted, whereas the small metropolitan and pediatric findings were not reproduced. No consistent differences in change were observed by race or ZIP-code income quartile. Overall, these results suggest that the most consistent geographic change in robot-assisted spinal fusion was a shift away from central metropolitan counties, while payer-related findings should be interpreted cautiously given the small baseline Medicaid count in early robotic cases.
Footnotes
Funding: The authors received no financial support for the research, authorship, and/or publication of this article.
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
ORCID iDs
Connor Burke https://orcid.org/0009-0004-7930-6199
Anoop S. Chinthala https://orcid.org/0000-0002-6065-1177
Ethical Considerations
This study used de-identified, publicly available National Inpatient Sample (NIS) data and was deemed exempt from IRB review by the Indiana University Human Research Protection Program (IU HRPP) (protocol #30394).
Consent to Participate
Not applicable. Informed consent is not required because this study analyzed de-identified, publicly available NIS data.
Data Availability Statement
The data used in this study are from the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS), Agency for Healthcare Research and Quality. NIS data are available to qualified researchers through purchase and completion of the HCUP Data Use Agreement and, therefore, cannot be publicly shared by the authors. Researchers may obtain the data directly from HCUP.*
References
- 1.Noriega DC, Rodriguez-Morata A, Ardura F, Hernández-Rama JA, Sánchez-Lite I, Toribio B. Risk-benefit analysis of navigation techniques for vertebral pedicle screw placement in thoracolumbar spine surgery: a prospective, randomized, observational study. Spine J. 2017;17(1):76-83. doi: 10.1016/j.spinee.2016.08.012.27497889 [DOI] [Google Scholar]
- 2.Rivkin MA, Yocom SS. Thoracolumbar instrumentation with CT-guided navigation (O-arm) in 270 consecutive patients: accuracy rates and lessons learned. Neurosurg Focus. 2014;36(3):E7. doi: 10.3171/2014.1.FOCUS13540. [DOI] [PubMed] [Google Scholar]
- 3.Ghasem A, Sharma A, Greif DN, Alam M, Al MM. The arrival of robotics in spine surgery: a review of the literature. Spine (Phila Pa 1976). 2018;43(23):1670-1677. doi: 10.1097/BRS.0000000000002695. [DOI] [PubMed] [Google Scholar]
- 4.Perfetti DC, Kisinde S, Rogers-LaVanne MP, Satin AM, Lieberman IH. Robotic spine surgery: past, present, and future. Spine (Phila Pa 1976). 2022;47(13):909-921. doi: 10.1097/BRS.0000000000004357. [DOI] [PubMed] [Google Scholar]
- 5.Koucheki R, Bonello JP, Abbas A, et al. A comparison of key performance metrics of major robotic platforms in spine surgery: a network meta-analysis of 14,462 screws. Eur Spine J. 2025;34(9):4088-4101. doi: 10.1007/s00586-025-08990-y. [DOI] [PubMed] [Google Scholar]
- 6.Pando A, Hanna G, Goldstein I. Robotic assistance in lumbar fusion surgery: trends and patterns from 2016-2019. Eur Spine J. 2023;32(6):1966-1972. doi: 10.1007/s00586-023-07663-y. [DOI] [PubMed] [Google Scholar]
- 7.Matur AV, Palmisciano P, Duah HO, Chilakapati SS, Cheng JS, Adogwa O. Robotic and navigated pedicle screws are safer and more accurate than fluoroscopic freehand screws: a systematic review and meta-analysis. Spine J. 2023;23(2):197-208. doi: 10.1016/j.spinee.2022.10.006. [DOI] [PubMed] [Google Scholar]
- 8.Quiceno E, Soliman MAR, Khan A, Mullin JP, Pollina J. How do robotics and navigation facilitate minimally invasive spine surgery? A case series and narrative review. Neurosurgery. 2025;96(3S):S84-S93. doi: 10.1227/NEU.0000000000003352. [DOI] [PubMed] [Google Scholar]
- 9.Pennington Z, Judy BF, Zakaria HM, et al. Learning curves in robot-assisted spine surgery: a systematic review and proposal of application to residency curricula. Neurosurg Focus. 2022;52(1):E3. doi: 10.3171/2021.10.FOCUS21496. [DOI] [PubMed] [Google Scholar]
- 10.Mao J, Artiguenave F, Corkum MT, et al. Robot-assisted surgery and racial and ethnic disparities in postoperative outcomes after radical prostatectomy. Ann Surg Oncol. 2024;31(1):604-614. doi: 10.1245/s10434-023-14362-3. [DOI] [PubMed] [Google Scholar]
- 11.Kim J, Hu JC, Nguyen P, Pashos CL, Keating NL. Disparities in the receipt of robot-assisted radical prostatectomy: evidence from California. BMJ Open. 2015;5(4):e007409. doi: 10.1136/bmjopen-2014-007409. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kim SP, Boorjian SA, Shah ND, et al. Disparities in access to hospitals with robotic surgery for patients with prostate cancer undergoing radical prostatectomy. J Urol. 2013;189(2):514-520. doi: 10.1016/j.juro.2012.09.033. [DOI] [PubMed] [Google Scholar]
- 13.Mastrokostas PG, Mastrokostas LE, Emara AK, et al. Geographic and socioeconomic disparities in robotic spine surgery access in the continental United States: a cross-sectional ecological analysis. Global Spine J. 2025;16:577-587. doi: 10.1177/21925682251356218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Cole T, Li NY, Gil JA, Daniels AH. Racial and ethnic disparities in the utilization of high-volume hospitals for anterior cervical discectomy and fusion. J Neurosurg Spine. 2022;36(3):363-370. doi: 10.3171/2021.5.SPINE21466. [DOI] [Google Scholar]
- 15.Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Routledge; 1988. [Google Scholar]
- 16.Foley BL, Sheridan EM, Dougherty CE, Are C, Mammen J, Santamaria-Barria JA. The landscape of U.S. academic surgery departments in the corporate medicine era: federal and industry funding, academic output, and corporate medicine revenue analysis. J Surg Res. 2025;315:950-958. doi: 10.1016/j.jss.2025.10.013. [DOI] [PubMed] [Google Scholar]
- 17.Turchetti G, Palla I, Pierotti F, Coudin J, Poley MJ. Economic evaluation of da Vinci-assisted robotic surgery: a systematic review. Surg Endosc. 2012;26(3):598-606. doi: 10.1007/s00464-011-1921-6. [DOI] [PubMed] [Google Scholar]
- 18.Jeong IG, Khandwala YS, Kim JH, et al. Association of robotic-assisted vs laparoscopic radical nephrectomy with perioperative outcomes and health care costs, 2003 to 2015. JAMA. 2017;318(16):1561-1568. doi: 10.1001/jama.2017.15849. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nabi J, Friedlander DF, Chen X, et al. Assessment of out-of-pocket costs for robotic cancer surgery in US adults. JAMA Netw Open. 2020;3(1):e1919185. doi: 10.1001/jamanetworkopen.2019.19185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Barbash GI, Friedman B, Glied SA, Steiner CA. Factors associated with adoption of robotic surgical technology in U.S. hospitals and relationship to radical prostatectomy procedure volume. Ann Surg. 2014;259(1):1-6. doi: 10.1097/SLA.0b013e3182a5c8b8. [DOI] [PubMed] [Google Scholar]
- 21.Wright JD, Sohn MW, Kuo TM, et al. Effect of hospital competition and hospital use of robotic-assisted surgery. JAMA Surg. 2016;151(11):1023-1031. doi: 10.1001/jamasurg.2016.2491. [DOI] [PubMed] [Google Scholar]
- 22.Gaffney A, McCormick D. The Affordable Care Act: implications for health-care equity. Lancet. 2017;389(10077):1442-1452. doi: 10.1016/S0140-6736(17)30786-9. [DOI] [PubMed] [Google Scholar]
- 23.Altman D, Frist WH. Medicare and Medicaid at 50 years: perspectives of beneficiaries, health care professionals and institutions, and policy makers. JAMA. 2015;314(4):384-395. doi: 10.1001/jama.2015.7811. [DOI] [PubMed] [Google Scholar]
- 24.Retchin SM, Xu WY. Medicaid innovations and the role of academic health centers. Acad Med. 2018;93(10):1454-1456. doi: 10.1097/ACM.0000000000002296. [DOI] [PubMed] [Google Scholar]
- 25.Ozawa MN, Kim J, Joo M. Income class and the accumulation of net worth in the United States. Soc Work Res. 2006;30(4):211-222. doi: 10.1093/swr/30.4.211. [DOI] [Google Scholar]
- 26.Cooper RA, Cooper MA, McGinley EL, Fan X, Rosenthal JT. Poverty, wealth, and health care utilization: a geographic assessment. J Urban Health. 2012;89(5):828-847. doi: 10.1007/s11524-012-9689-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Khan IS, Huang E, Maeder-York W, et al. Racial disparities in outcomes after spine surgery: a systematic review and meta-analysis. World Neurosurg. 2022;157:e232-e244. doi: 10.1016/j.wneu.2021.09.140. [DOI] [PubMed] [Google Scholar]
- 28.Aladdin DEH, Tangel V, Lui B, Pryor KO, Witkin LR, White RS. Black race as a social determinant of health and outcomes after lumbar spinal fusion surgery: a multistate analysis, 2007 to 2014. Spine (Phila Pa 1976). 2020;45(10):701-711. doi: 10.1097/BRS.0000000000003367. [DOI] [PubMed] [Google Scholar]
- 29.Seicean A, Seicean S, Neuhauser D, Benzel EC, Weil RJ. The influence of race on short-term outcomes after laminectomy and/or fusion spine surgery. Spine (Phila Pa 1976). 2017;42(1):34-41. doi: 10.1097/BRS.0000000000001657. [DOI] [PubMed] [Google Scholar]
- 30.Elsamadicy AA, Serrato P, Ghanekar SD, et al. Racial/ethnic disparities on resource utilization and outcomes for patients undergoing spine surgery: a NSQIP analysis of 402,765 patients. Clin Neurol Neurosurg. 2025;254:108931. doi: 10.1016/j.clineuro.2025.108931. [DOI] [PubMed] [Google Scholar]
- 31.Reyes AM, Katz JN, Schoenfeld AJ, Losina E, Chang Y. National utilization and inpatient safety measures of lumbar spinal fusion methods by race/ethnicity. Spine J. 2021;21(5):785-794. doi: 10.1016/j.spinee.2020.11.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Stephens CQ, Souza JM, Clary R, et al. Indices for social determinants of health in pediatric surgical patients. JAMA Netw Open. 2020;3(7):e2087624. doi: 10.1001/jamanetworkopen.2020.287624. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Sharma M, Uddin SA, Hanna G, et al. Trends and long-term health care utilization of computer-assisted neuronavigation in spine fusions: an exact matched analysis of national administrative database. World Neurosurg. 2022;166:e850-e858. doi: 10.1016/j.wneu.2022.07.116. [DOI] [PubMed] [Google Scholar]
- 34.Kelley BV, Hsiue PP, Upfill-Brown AM, et al. Utilization trends and outcomes of computer-assisted navigation in spine fusion in the United States. Spine J. 2021;21(8):1246-1255. doi: 10.1016/j.spinee.2021.03.029. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data used in this study are from the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS), Agency for Healthcare Research and Quality. NIS data are available to qualified researchers through purchase and completion of the HCUP Data Use Agreement and, therefore, cannot be publicly shared by the authors. Researchers may obtain the data directly from HCUP.*