Abstract
Study design
Systematic Review and Meta-Analysis.
Objectives
Navigation-guided screw placement improves screw accuracy and may reduce hardware-related revision, yet its effect on overall unexpected return to the operating room (URTO) remains uncertain. This study compares image-guided (navigation and robotics) with freehand screw placement for hardware-related revision, non-hardware reoperation, and URTO following spinal fixation.
Methods
A PRISMA-guided systematic review identified comparative studies evaluating image-guided versus freehand screw fixation. Two reviewers performed screening, data extraction, and risk-of-bias assessment. Odds ratios (ORs) were pooled using a random-effects Mantel-Haenszel meta-analysis. Subgroup analyses compared navigated, robotic, and freehand techniques.
Results
Twenty-one studies were included, comprising 91,285 patients (8,743 image-guided; 82,442 freehand). Compared with freehand, image guidance was associated with lower odds of hardware-related revision (OR 0.55, 95% CI 0.40–0.75) and lower odds of URTO (OR 0.69, 95% CI 0.50–0.94), with no difference in non-hardware reoperation (OR 1.03, 95% CI 0.67–1.57). Navigation was associated with lower odds of hardware-related revision (OR 0.51, 95% CI 0.35–0.77) and URTO (OR 0.63, 95% CI 0.51–0.77) versus freehand, whereas robotics showed no significant differences across outcomes. Indirect comparison between navigation and robotics was not statistically significant. Risk of bias was overall high because of retrospective study design, confounding, and inconsistent outcome reporting.
Conclusions
Navigation was associated with lower odds of hardware-related revision and lower odds of URTO, without a difference in non-hardware reoperation. Robotics showed no statistically significant advantage over freehand fixation, but its underrepresentation and differing case mix limit comparative conclusions.
Keywords: navigation, reoperation, pedicle, screw, spine
Keywords: III
1. Introduction
Lateral mass and pedicle screws are widely used in spine surgery to provide stability, fixation, and alignment across diverse pathologies, including instability, deformity, trauma, and degenerative disease. Acting as anchors for fusion constructs, these implants help maintain alignment during osseous healing and have been shown to be safe and effective, making internal screw fixation a cornerstone of modern spinal surgery. 1
Pedicle screw fixation was originally developed using freehand techniques based on anatomic landmarks, with early descriptions emerging from European orthopaedic surgeons in the mid-20th century, including the foundational work of Boucher, 2 Roy-Camille, 3 and Magerl. 4 Later, frameless stereotactic systems were introduced to augment screw placement accuracy, as described by Kalfas et al in 1995. 5 With the advancement in intraoperative imaging, CT-based navigation entered routine practice in the early 2000s, giving rise to navigated screw fixation. 6 In parallel, the adoption of minimally invasive and percutaneous approaches expanded, motivated by the potential to reduce surgical morbidity. 7
An unexpected return to the operating room (URTO), whether for revision of mispositioned hardware or for reoperation due to other complications, remains a costly and burdensome outcome for patients and health systems. 8 Prior reviews have reported that navigation reduces hardware-related revision, whereas others find no difference between navigated and freehand techniques, leaving uncertainty about clinical impact.9,10 Despite the potential benefits of navigation, there are some studies that have raised concern for non-hardware-related complications and caution that these risks must be interpreted alongside any possible benefits.11,12
Given the growing clinical adoption of navigation and robotics and the importance of URTO as a patient-centered outcome, a focused evaluation is warranted. The objective of this study was to compare image-guided screw placement—including navigation and robotics—with freehand fixation with respect to hardware-related revision, non-hardware reoperation, and URTO, with subgroup analyses evaluating navigation and robotics separately.
2. Materials and methods
This systematic review was conducted in accordance with the 2020 PRISMA guidelines. 13 We included human studies comparing image-guided (navigation or robotic) versus conventional freehand/fluoroscopy-guided lateral mass or pedicle screw fixation across the cervical, thoracic, lumbar, or sacral spine. Primary outcomes were unexpected return to the operating room (URTO), hardware-related revision, and non-hardware reoperation.
2.1. Search Strategy and Study Selection
MEDLINE (Ovid), Embase (Ovid), and PubMed were searched from inception to March 31st, 2025 using predefined terms related to spine surgery, navigation, robotics, and freehand screw fixation. Citations were screened independently by two reviewers using Covidence, with disagreements resolved by consensus or a third reviewer.
2.2. Eligibility Criteria
Eligible studies included randomized controlled trials, prospective and retrospective cohort studies, and case-control studies (Level I–III evidence) reporting navigated or robotic-assisted versus freehand screw fixation and at least one outcome of interest. Exclusions included non-human studies, case reports/series, reviews, conference abstracts without full text, non-English studies, and follow-up <1 month.
2.3. Data Extraction and Quality Assessment
Data extraction was performed independently in duplicate, including study characteristics, patient demographics, surgical details (region, indication, guidance method, system used, primary vs. revision, open vs. percutaneous), and outcomes (hardware-related revision, non-hardware reoperation, and URTO). Risk of bias was assessed using RoB 2 for randomized trials and ROBINS-I for non-randomized studies.
2.4. Statistical analysis
Binary outcomes were pooled as odds ratios (ORs) with 95% confidence intervals using random-effects models. Heterogeneity was assessed with I2. Preplanned subgroup analyses compared navigation versus freehand fixation and robotics versus freehand fixation. An indirect Bucher comparison between navigation and robotics was performed using freehand fixation as the common comparator. Because navigation and robotics were evaluated in substantially different patient populations and surgical indications, this indirect comparison was considered exploratory and susceptible to between-study confounding. Analyses were performed in RStudio using the meta package, with p < 0.05 considered statistically significant.
3. Results
3.1. Study Characteristics
Of 7601 identified articles, 21 studies12,14-33 met inclusion criteria after screening (Figure 1), comprising 19 retrospective cohorts and two randomized trials (Supplemental Table S1). There was considerable agreement during the title and abstract (k = 0.752) and full text (k = 0.750) screening stages.
Figure 1.
PRISMA flow diagram of study selection
3.2. Patient and Procedure Characteristics
Across studies, 91,285 patients were included: 8,743 (9.6%) underwent image-guided fixation (navigated or robotic) and 82,442 (90.4%) freehand fixation. Guided patients were older (50.9 vs 38.5 years), more often male (45.5% vs 40.8%), and had significantly shorter follow-up (6.36 vs 7.01 months) (all p<0.001), while BMI differences were small (29.8 vs 30.2 kg/m2, p=0.016) (Table 1). Navigation accounted for 93.7% and robotics for 6.3% of guided cases. This imbalance limited statistical power for robotics-specific analyses. Image guidance was used more frequently in deformity and trauma, in cervical and thoracic regions, and in multilevel constructs. Freehand fixation was predominantly used for lumbar degenerative disease. Revision procedures were more common in image-guided cohorts (1.2% vs 0.4%, p<0.001). Most surgeries in both groups were performed using open approaches (Table 2).
Table 1.
Summary of Patient Characteristics
| General Characteristics | Total | Image-guided | Freehand | p-value |
|---|---|---|---|---|
| Total number of patients | 91,285 | 8,743 | 82,442 | |
| Mean age in years ± SD | 40.54 n = 36,219 | 50.94 ± 12.12 n = 5,273 | 38.46 ± 10.98 n = 30,284 | <0.0001 |
| Sex (%) | ||||
| Male | 35,749 (41.2) | 2,956 (45.5) | 32,746 (40.8) | <0.00001 |
| Female | 51,052 (58.8) | 3,542 (54.5) | 47,456 (59.2) | |
| Mean BMI in kg/m2 ± SD | 30.12 n = 18,302 | 29.80 ± 6.14 n = 1,876 | 30.17 ± 6.34 n = 16,326 | 0.0163 |
| Mean follow-up time in months ± SD | 6.95 ± 0.12 n = 90,496 | 6.36 ± 0.74 n = 8,391 | 7.01 ± 0.05 n = 82,005 | <0.0001 |
BMI: body mass index, F: female, kg: kilogram, M: male, m: meter, SD: standard deviation.
Table 2.
Summary of Procedure Characteristics
| General Characteristics | Total | Image-guided | Freehand | p-value |
|---|---|---|---|---|
| Guidance (%) | NA | |||
| Navigation | 7,815 (93.7) | 7,815 (93.7) | NA | |
| Robot | 522 (6.3) n = 8,337 | 522 (6.3) n = 8,337 | NA | |
| Navigation System (%) | NA | |||
| O-arm-based navigation (Medtronic) | 747 (39.6) | 747 (39.6) | NA | |
| StealthStation (Medtronic) | 337 (17.8) | 337 (17.8) | NA | |
| Brainlab navigation systems | 347 (18.4) | 347 (18.4) | NA | |
| Stereotactic computer-assisted systems | 456 (24.2) n = 1,887 | 456 (24.2) n = 1,887 | NA | |
| Robotic System (%) | NA | |||
| Renaissance (Mazor/Medtronic) | 146 (22.9) | 146 (22.9) | NA | |
| Globus Medical systems | 460 (72.2) | 460 (72.2) | NA | |
| Chinese robotic systems | 31 (4.9) n = 637 | 31 (4.9) n = 637 | NA | |
| Primary Operation (%) | 15,482 (99.4) | 3,355 (98.8) | 12,127 (99.6) | <0.00001 |
| Revision Surgery (%) | 86 (0.6) n = 15,568 | 41 (1.2) n = 3,396 | 45 (0.4) n = 12,172 | |
| Region of surgery (nb of patients) (%) | ||||
| Thoracic | 339 (0.5) | 174 (2.8) | 165 (0.2) | <0.00001 |
| Lumbar | 74,468 (99.5) | 6,016 (96.9) | 68,452 (99.8) | |
| Lumbosacral | 34 (<0.1) n = 74,841 | 19 (0.3) n = 6,209 | 15 (<0.1) n = 68,632 | |
| Region of surgery (nb of screws) (%) | ||||
| Cervical | 291 (21.5) | 124 (22.1) | 169 (21.3) | <0.00001 |
| Thoracic | 747 (55.1) | 343 (61.0) | 402 (50.7) | |
| Lumbar | 317 (23.4) n = 215 | 95 (16.9) n = 112 | 222 (28.0) n = 103 | |
| Indication for surgery (%) | <0.00001 | |||
| Deformity | 14,537 (24.4) | 2,072 (39.5) | 12,460 (23.0) | |
| Degeneration | 5,380 (9.0) | 2,811 (53.6) | 2,534 (4.7) | |
| Infection | 10 (<0.1) | 3 (<0.1) | 4 (<0.1) | |
| Trauma | 163 (0.3) | 73 (1.4) | 82 (0.1) | |
| Tumor | 39,476 (66.3) n = 59,566* | 289 (5.5) n = 5,248 | 39,178 (72.2) n = 54,258 | |
| Mean number of screws | 16,477 n = 2,565 | 5,624 n = 802 | 10,853 n = 1,763 | 0.3442 |
| Mean number of screws per patient (SD) | 5.4 (1.9, n = 1,792) | 5.5 (1.8, n = 409) | 5.4 (1.9, n = 1,383) | |
| Method (%) | 0.765846 | |||
| Open | 3,113 (96.5) | 1,213 (96.3) | 1,900 (96.5) | |
| Percutaneous | 114 (3.5) n = 3,227 | 46 (3.7) n = 1,259 | 68 (3.5) n = 1,968 | |
| Levels (%) | ||||
| Single level | 31,348 (64.8) | 2,083 (44.3) | 29,265 (67.0) | <0.0001 |
| ≥2 levels | 17,045 (35.2) n = 48,393 | 2,617 (55.7) n = 4,700 | 14,428 (33.0) n = 43,693 | |
| Mean number of levels per patient (SD) | 2.5 (1.5, n = 2,762) | 3.7 (2.0, n = 1,023) | 1.9 (1.3, n = 1,739) | <0.0001 |
n: number of patients, NA: not applicable, nb: number, NR: not reported, SD: standard deviation; *Group-specific indication totals do not equal the overall total because one included study reported 60 patients without providing group-level allocation.
3.3. Pooled Outcomes (Image-guided vs Freehand)
Hardware-related revision rates were low and similar between groups (guided 1.42% vs freehand 1.35%, p=0.640). Non-hardware reoperations were less frequent with image guidance (2.44% vs 3.48%, p=0.005). URTO occurred at comparable rates (2.41% vs 2.10%, p=0.051) (Table 3).
Table 3.
Summary of Surgical Outcomes
| Outcome | Total | Image-guided | Freehand | p-value |
|---|---|---|---|---|
| n | n | n | ||
| Revision of hardware (%) | ||||
| Yes | 707 (1.36) | 88 (1.42) | 619 (1.35) | 0.639632 |
| No | 51,359 | 6,098 | 45,261 | |
| Reoperation other than revision of hardware (%) | ||||
| Yes | 321 (3.14) | 81 (2.44) | 240 (3.48) | 0.005098 |
| No | 9,893 | 3,232 | 6,661 | |
| Unexpected return to operating room (%) | ||||
| Yes | 1,940 (2.13) | 211 (2.41) | 1,729 (2.10) | 0.051444 |
| No | 89,345 | 8,532 | 80,713 | |
n: number of patients.
3.4. Meta-analyses (Image-guided vs Freehand)
Random-effects meta-analysis demonstrated lower odds of hardware-related revision with image guidance (OR 0.55, 95% CI 0.40–0.75; p<0.001; I2 = 24%). No significant difference was observed for non-hardware reoperations (OR 1.03, 95% CI 0.67–1.57; p=0.899; I2 = 37%). Image guidance was associated with lower odds of URTO (OR 0.69, 95% CI 0.50–0.94; p = 0.018; I2 = 51%) (Figures 2-4: Table 4).
Figure 3.
Non-hardware reoperation: image-guided versus freehand fixation (random-effects Mantel-Haenszel odds ratio)
Figure 2.
Hardware-related revision: image-guided versus freehand fixation (random-effects Mantel-Haenszel odds ratio)
Figure 4.
Unexpected return to the operating room: image-guided versus freehand fixation (random-effects Mantel-Haenszel odds ratio)
Table 4.
Overall Results of Meta-analyses for Surgical Outcomes
| Outcome | Nb of Studies | Total Sample (Guided/Freehand) | Pooled OR (95% CI) | p (overall effect) | I2 (%) | p (heterogeneity) |
|---|---|---|---|---|---|---|
| Revision of hardware | 16 | 6,186/45,880 | 0.55 (0.40-0.75) | 0.0002 | 24.1 | 0.2067 |
| Reoperation other than revision of hardware | 8 | 3,313/6,901 | 1.03 (0.67-1.57) | 0.8986 | 36.7 | 0.1487 |
| Unexpected return to operating room | 21 | 8,743/82,442 | 0.69 (0.50-0.94) | 0.0182 | 50.8 | 0.0072 |
CI: confidence interval, nb: number, OR: odd ratio, p: p value, I2: inconsistency statistic.
3.5. Subgroup Characteristics
Navigated and robotic patients were older (49.5 and 61.8 years vs 38.5 years) and more often male than freehand patients (all p<0.001). BMI differences were small, and follow-up was longest in robotic cohorts. Navigation was used more often for deformity and trauma and in multilevel constructs (58% ≥2 levels). Robotics was used almost exclusively for lumbar degenerative disease and single-level constructs (Table 5).
Table 5.
Summary of Patient and Procedure Characteristics for Subgroup Analysis
| General Characteristics | Navigation | Robotics | Freehand | p* | p** |
|---|---|---|---|---|---|
| Total number of patients | 7,785 | 552 | 82,442 | | |
| Mean age in years ± SD | 49.52 ± 11.86 n = 4,590 | 61.81 ± 12.24 n = 522 | 38.46 ± 10.98 n = 30,284 | <0.0001 | <0.0001 |
| Sex (%) | |||||
| Male | 2,544 (45.7) | 265 (50.8) | 32,746 (40.8) | <0.02561 | <0.00001 |
| Female | 3,026 (54.3) | 257 (49.2) | 47,456 (59.2) | ||
| Mean BMI in kg/m2 ± SD | 30.32 ± 6.32 n = 1,354 | 28.46 ± 5.66 n = 522 | 30.17 ± 6.34 n = 16,326 | <0.0001 | <0.0001 |
| Mean follow-up time in months ± SD | 5.97 ± 0.76 n = 7,548 | 11.96 ± 1.23 n = 437 | 7.01 ± 0.05 n = 82,005 | <0.0001 | <0.0001 |
| Primary Operation (%) | 2,871 (98.8) | 484 (98.6) | 12,127 (99.6) | 0.631925 | <0.00001 |
| Revision Surgery (%) | 34 (1.2) n = 2,905 | 7 (1.4) n = 491 | 45 (0.4) n = 12.172 | ||
| Region of surgery (nb of patients) (%) | 174 (3.0) | 0 (0.0) | 165 (0.2) | <0.00001 | <0.00001 |
| Thoracic | 5,596 (96.9) | 175 (94.1) | 68,452 (99.8) | ||
| Lumbar | 8 (0.1) | 11 (5.9) | 15 (<0.1) | ||
| Lumbosacral | n = 5,778 | n = 186 | n = 68,632 | ||
| Region of surgery (nb of screws) (%) | NR | 291 (21.5) | 169 (21.3) | NA | <0.04820 |
| Cervical | NR | 747 (55.1) | 402 (50.7) | ||
| Thoracic | NR | 317 (23.4) | 222 (28.0) | ||
| Lumbar | | n = 1,355 | n = 103 | ||
| Indication for surgery (%) | 1,911 (40.5) | 0 (0.0) | 12,460 (23.0) | <0.00001 | <0.00001 |
| Deformity | 2,691 (57.0) | 120 (100.0) | 2,534 (4.7) | ||
| Degeneration | 3 (0.1) | 0 (0.0) | 4 (<0.1) | ||
| Infection | 73 (1.5) | 0 (0.0) | 82 (0.1) | ||
| Trauma | 44 (0.9) | 0 (0.0) | 39,178 (72.2) | ||
| Tumor | n = 4,722 | n = 120 | = 54,711 | ||
| Mean number of screws | 3,244 | 2,380 | 10,853 | NA | 0.3442 |
| Mean number of screws per patient (SD) | N = 365 | N = 437 | N = 1,763 | ||
| NR | 5.4 (1.8, n = 409) | 5.5 (1.9, n = 1,383) | |||
| Method (%) | |||||
| Open | | | 1,900 (96.5) | <0.00001 | <0.00001 |
| Percutaneous | | | 68 (3.5) n = 1,968 | ||
| Levels (%) | |||||
| Single level | 1,819 (42.0) | 97 (79.5) | 29,265 (67.0) | <0.00001 | <0.00001 |
| ≥2 levels | 2,514 (58.0) n = 4,333 | 25 (20.5) n = 122 | 14,428 (33.0) n = 43,693 | ||
| Mean number of levels per patient (SD) | 4.8 (2.7, n = 614) | 1.9 (0.9, n = 409) | 1.9 (1.3, n = 1,739) | <0.0001 | <0.0001 |
BMI: body mass index, F: female, kg: kilogram, M: male, m: meter, n: number of patients, NA: not applicable, nb: number, SD: standard deviation.
*p-value for the difference between the Robotics and Navigation groups.
**p-value for the difference between the Robotics, Navigation, and Freehand groups.
3.6. Subgroup Outcomes
Crude hardware-related revision rates were 1.50%, 1.15%, and 1.35% in the navigation, robotics, and freehand groups, respectively (p = 0.620). URTO rates were 2.32%, 2.36%, and 2.10%, respectively (p = 0.381). Non-hardware reoperation rates were 2.11%, 1.72%, and 3.48%, respectively (p < 0.001 across the three groups), with no significant difference between navigation and robotics (p = 0.608) (Table 6).
Table 6.
Summary of Surgical Outcomes for Subgroup Analysis
| Outcome | Navigation | Robotics | Freehand | p* | p** |
|---|---|---|---|---|---|
| n | n | n | |||
| Revision of hardware (%) | |||||
| Yes | 81 (1.50) | 6 (1.15) | 619 (1.35) | 0.523902 | 0.620062 |
| No | 5,336 | 518 | 45,261 | | |
| Reoperation other than revision of hardware (%) | |||||
| Yes | 58 (2.11) | 7 (1.72) | 240 (3.48) | 0.607601 | 0.000576 |
| No | 2,688 | 399 | 6,661 | | |
| Unexpected return to operating room (%) | |||||
| Yes | 181 (2.32) | 13 (2.36) | 1,729 (2.10) | 0.963859 | 0.381054 |
| No | 7,604 | 539 | 80,713 | | |
n: number of patients.
*p-value for the difference between Robotics and Navigation groups.
**p-value for the difference between Robotics, Navigation, and Freehand groups.
3.7. Subgroup Meta-analyses
Navigation was associated with lower odds of hardware-related revision (OR 0.51, 95% CI 0.35–0.77; p = 0.001; I2 = 42%) and URTO (OR 0.63, 95% CI 0.51–0.77; p < 0.001; I2 = 31%) compared with freehand fixation, without a significant difference in non-hardware reoperation (OR 0.79, 95% CI 0.55–1.15; p = 0.220; I2 = 0%). Robotics showed no significant differences versus freehand fixation for hardware-related revision (OR 0.83, 95% CI 0.35–2.00; p = 0.682; I2 = 0%), non-hardware reoperation (OR 1.10, 95% CI 0.46–2.64; p = 0.827; I2 = 0%), or URTO (OR 0.96, 95% CI 0.51–1.79; p = 0.891; I2 = 0%). Exploratory indirect Bucher comparisons did not identify statistically significant differences between navigation and robotics for hardware-related revision (OR 0.62, 95% CI 0.24–1.62; p = 0.328), non-hardware reoperation (OR 0.72, 95% CI 0.28–1.86; p = 0.499), or URTO (OR 0.66, 95% CI 0.34–1.27; p = 0.209); these comparisons should be interpreted cautiously because the technologies were evaluated in substantially different patient populations and procedural contexts (Figures 5-10; Tables 7-9).
Figure 6.
Hardware-related revision: robotics versus freehand fixation (random-effects subgroup meta-analysis)
Figure 7.
Non-hardware reoperation: navigation versus freehand fixation (random-effects subgroup meta-analysis)
Figure 8.
Non-hardware reoperation: robotics versus freehand fixation (random-effects subgroup meta-analysis)
Figure 9.
Unexpected return to the operating room: navigation versus freehand fixation (random-effects subgroup meta-analysis)
Table 8.
Overall Results of Meta-analyses for Surgical Outcomes (Robotics vs. Freehand)
| Outcome | Nb of Studies | Total Sample (Robotics/Freehand) | Pooled OR (95% CI) | p (overall effect) | I2 (%) | p (heterogeneity) |
|---|---|---|---|---|---|---|
| Revision of hardware | 5 | 524/1,585 | 0.83 (0.35-2.00) | 0.6821 | 0.0 | 0.5091 |
| Reoperation other than revision of hardware | 3 | 406/1,380 | 1.10 (0.46-2.64) | 0.8272 | 0.0 | 0.9433 |
| Unexpected return to operating room | 6 | 552/1,613 | 0.96 (0.51-1.79) | 0.8911 | 0.0 | 0.6733 |
CI: confidence interval, nb: number, OR: odd ratio, p: p value, I2: inconsistency statistic.
Figure 5.
Hardware-related revision: navigation versus freehand fixation (random-effects subgroup meta-analysis)
Figure 10.
Unexpected return to the operating room: robotics versus freehand fixation (random-effects subgroup meta-analysis)
Table 7.
Overall Results of Meta-analyses for Surgical Outcomes (Navigation vs. Freehand)
| Outcome | No of Studies | Total Sample (Navigation/Freehand) | Pooled OR (95% CI) | p (overall effect) | I2 (%) | p (heterogeneity) |
|---|---|---|---|---|---|---|
| Revision of hardware | 10 | 5,417/5,153 | 0.51 (0.35-0.77) | 0.0011 | 41.9 | 0.0989 |
| Reoperation other than revision of hardware | 4 | 2,746/2,492 | 0.79 (0.55-1.15) | 0.2197 | 0.0 | 0.6011 |
| Unexpected return to operating room | 13 | 7,785/38,658 | 0.63 (0.51-0.77) | <0.0001 | 31.4 | 0.1399 |
CI: confidence interval, nb: number, OR: odd ratio, p: p value, I2: inconsistency statistic.
Table 9.
Overall Results of Meta-analyses for Surgical Outcomes (Navigation vs. Robotics)
| Outcome | No of Studies (Navigation/Robotics) | Total Sample (Navigation/Robotics) | Pooled OR (95% CI) | p-value |
|---|---|---|---|---|
| Revision of hardware | 10/5 | 5,417/524 | 0.618 (0.236-1.620) | 0.3280 |
| Reoperation other than revision of hardware | 4/3 | 2,746/406 | 0.721 (0.279-1.861) | 0.4986 |
| Unexpected return to operating room | 13/6 | 7,785/552 | 0.657 (0.341-1.266) | 0.2090 |
CI: confidence interval, nb: number, OR: odd ratio.
3.8. Risk of Bias
Overall risk of bias was high, driven by retrospective study designs, confounding, and incomplete outcome reporting, which limited causal inference. The two randomized trials were judged to have some concerns or a high risk of bias, and most nonrandomized studies were judged to be at serious risk of bias using ROBINS-I (Supplemental Table S2). For the URTO outcome, Egger’s test was not significant (t = -1.30, df = 16, p=0.21), with no clear funnel plot asymmetry observed (Figure 11).
Figure 11.
Funnel plot assessing publication bias for unexpected return to the operating room
4. Discussion
This systematic review and meta-analysis evaluated navigation, robotics, and freehand screw fixation across more than 91,000 patients, with a focus on URTO as a patient-centered outcome rather than screw-placement accuracy alone. The crude aggregate event rates in Tables 3 and 6 did not consistently favour image guidance, whereas the random-effects meta-analysis showed lower odds of hardware-related revision and URTO. This apparent discrepancy reflects the different analytic approaches: crude rates pool all patients without study-level weighting, whereas the meta-analysis combines study-specific odds ratios using random-effects weighting and incorporates between-study heterogeneity. The pooled estimates nevertheless remain susceptible to residual confounding because most included studies were retrospective and the cohorts differed substantially in baseline characteristics and case complexity. Robotics showed no statistically significant differences compared with freehand fixation, but its underrepresentation and distinct case mix preclude conclusions of equivalence.
URTO represents a valuable patient-centered endpoint that captures the net clinical effect of navigation, integrating both hardware-related failures and broader postoperative complications. Prior work has demonstrated that navigation improves screw placement and reduces revision for malpositioned hardware.9,10,34 In the present meta-analysis, navigation was associated with lower odds of hardware-related revision and URTO, with no significant difference in non-hardware reoperation. This pattern is consistent with the mechanism of navigation: it improves technical accuracy but does not directly influence complications such as infection, hematoma, pseudarthrosis, or medical events.
Demographic and procedural differences across groups help explain these results. Navigation was more frequently used in older patients, deformity and trauma indications, thoracic and cervical regions, and multilevel constructs—situations with intrinsically higher risk. Freehand fixation was most common for lumbar degenerative disease and malignancy, often involving shorter constructs. Robotics appeared to be deployed in a narrower clinical context, largely degenerative lumbar disease and single-level instrumentation. These case-mix differences likely contributed to the higher unadjusted complication rates observed in image-guided cohorts. Although the study-level meta-analysis favoured navigation for hardware-related revision and URTO, it could not fully adjust for patient-level differences in age, surgical indication, construct length, or procedural complexity. Residual confounding and selection bias therefore remain, and the observed associations should not be interpreted as definitive evidence of a causal protective effect.
The potential benefits of navigation must be interpreted alongside its trade-offs. Navigation requires increased operating room traffic, more equipment, and workflow-related setup which raises the concern for sterile field contamination. Additionally, Katz et al reported higher unadjusted reoperation rates in adult spinal deformity navigated cases (9.9% vs 5.2%, p=0.011), likely reflecting longer operative time in navigated cohorts (405 min vs 320 min, p<0.001), increased transfusion requirements in navigated cohorts (52.2% vs 41.8%, p=0.01), and greater procedural complexity rather than a direct causal effect of navigation itself. 12 Similarly, concerns have been raised regarding infection risk. Berman et al observed higher infection-related return to the operating room in patients undergoing intraoperative 3D navigation (4.9% vs 0.6%, p=0.025), although this association did not reach statistical significance after adjustment (adjusted OR 6.49, p=0.073). 11 In contrast, Shuman et al found that navigation was associated with decreased odds of return to the operating room for wound-related procedures despite longer operative times, highlighting the inconsistency in current literature. 35 Taken together, these findings provide no clear signal of increased non-hardware reoperation with navigation. However, heterogeneity and residual confounding preclude definitive conclusions regarding infection or other non-hardware complications.
Economic considerations may also influence the value of navigation. 21 Although navigation entails substantial acquisition and operating costs, avoidance of hardware-related revision could offset some of these expenses; however, formal cost-effectiveness analysis was beyond the scope of this review.
This study has several important limitations. The evidence base was predominantly retrospective, with inherent risks of selection bias, residual confounding, and inconsistent outcome reporting, limiting causal inference. Heterogeneity in study design, outcome definitions, and reporting formats—including per-patient versus per-screw analyses—further reduced comparability across studies. Mean follow-up was relatively short in both groups—6.36 months for image-guided fixation and 7.01 months for freehand fixation—which may have undercaptured reoperations occurring beyond the early postoperative period. Robotics was substantially underrepresented, limiting statistical power and restricting conclusions regarding its comparative effectiveness. Moreover, navigation and robotics were evaluated in markedly different patient populations and surgical indications, and the absence of direct head-to-head comparisons required exploratory indirect analyses that remain susceptible to confounding. Future prospective, multicenter studies with standardized outcome definitions, adequate follow-up, and direct platform comparisons are needed to clarify the role of image guidance in reducing URTO. 36
5. Conclusion
Navigation was associated with lower odds of hardware-related revision and a modest reduction in URTO, without a difference in non-hardware reoperation. These associations should be interpreted cautiously because the evidence was predominantly retrospective and remained subject to residual confounding. Robotics showed no statistically significant advantage over freehand fixation, but its underrepresentation and differing case mix preclude definitive conclusions regarding comparative effectiveness. Prospective, multicenter studies with standardized outcome definitions, adequate follow-up, and direct platform comparisons are needed.
Supplemental Material
Supplemental material for The Impact of Image-Guided Screw Placement on Unexpected Return to the Operating Room: A Systematic Review and Meta-Analysis of 91,285 Patients by Jaskaran Singh, Marion Sylvain, Mikko Vorster, Bona Baissa, Aazad Abbas, and Michael Goytan in Global Spine Journal.
Supplemental material for The Impact of Image-Guided Screw Placement on Unexpected Return to the Operating Room: A Systematic Review and Meta-Analysis of 91,285 Patients by Jaskaran Singh, Marion Sylvain, Mikko Vorster, Bona Baissa, Aazad Abbas, and Michael Goytan in Global Spine Journal.
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.
Supplemental Material: Supplemental material for this article is available online.
ORCID iDs
Jaskaran Singh https://orcid.org/0009-0007-3462-091X
Mikko Vorster https://orcid.org/0009-0002-6467-0495
References
- 1.Gaines RW. The use of pedicle-screw internal fixation for the operative treatment of spinal disorders. J Bone Joint Surg Am. 2000;82(10):1458-1476. doi: 10.2106/00004623-200010000-00013. [DOI] [PubMed] [Google Scholar]
- 2.Boucher HH. A method of spinal fusion. J Bone Joint Surg Br. 1959;41-B(2):248-259. doi: 10.1302/0301-620X.41B2.248. [DOI] [PubMed] [Google Scholar]
- 3.Roy-Camille R, Saillant G, Mazel C. Internal fixation of the lumbar spine with pedicle screw plating. Clin Orthop Relat Res. 1986(203):7-17. [PubMed] [Google Scholar]
- 4.Magerl FP. Stabilization of the lower thoracic and lumbar spine with external skeletal fixation. Clin Orthop Relat Res. 1984;189:125-141. [PubMed] [Google Scholar]
- 5.Kalfas IH, Kormos DW, Murphy MA, et al. Application of frameless stereotaxy to pedicle screw fixation of the spine. J Neurosurg. 1995;83(4):641-647. doi: 10.3171/jns.1995.83.4.0641. [DOI] [PubMed] [Google Scholar]
- 6.Hott JS, Papadopoulos SM, Theodore N, Dickman CA, Sonntag VK. Intraoperative Iso-C C-arm navigation in cervical spinal surgery: review of the first 52 cases. Spine (Phila Pa 1976). 2004;29(24):2856-2860. doi: 10.1097/01.brs.0000147742.20637.49. [DOI] [PubMed] [Google Scholar]
- 7.Mobbs RJ, Sivabalan P, Li J. Technique, challenges and indications for percutaneous pedicle screw fixation. J Clin Neurosci. 2011;18(6):741-749. doi: 10.1016/j.jocn.2010.09.019. [DOI] [PubMed] [Google Scholar]
- 8.Adogwa O, Parker SL, Shau D, et al. Cost per quality-adjusted life year gained of revision fusion for lumbar pseudoarthrosis: defining the value of surgery. J Spinal Disord Tech. 2015;28(3):101-105. doi: 10.1097/BSD.0b013e318269cc4a. [DOI] [PubMed] [Google Scholar]
- 9.Papalia GF, Vadalà G, Russo F, et al. Higher Accuracy and Better Clinical Outcomes in Navigated Thoraco-Lumbar Pedicle Screw Fixation Versus Conventional Techniques: A Systematic Review and Meta-Analysis. Spine (Phila Pa 1976). 2024;49(19):1370-1380. doi: 10.1097/BRS.0000000000005105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.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]
- 11.Berman D, Eleswarapu A, Krystal J, Hoang H. Is the Use of Intraoperative 3D Navigation for Thoracolumbar Spine Surgery a Risk Factor for Post-Operative Infection? J Clin Med. 2022;11(8):2108. doi: 10.3390/jcm11082108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Katz AD, Galina J, Song J, et al. Impact of Navigation on 30-Day Outcomes for Adult Spinal Deformity Surgery. Global Spine J. 2023;13(7):1728-1736. doi: 10.1177/21925682211047551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Page MJ, Moher D, McKenzie JE. Introduction to preferred reporting items for systematic reviews and meta-analyses 2020 and implications for research synthesis methodologists. Res Synth Methods. 2022;13(2):156-163. doi: 10.1002/jrsm.1535. [DOI] [PubMed] [Google Scholar]
- 14.Ansari D, DesLaurier JT, Almadidy Z, Chaudhry NS, Mehta AI. A Retrospective Comparative Analysis of Perioperative Complications in Navigated versus Conventional Thoracolumbar Fusion in the Setting of Adult Spinal Deformity. World Neurosurg. 2022;162:e616-e625. doi: 10.1016/j.wneu.2022.03.085. [DOI] [PubMed] [Google Scholar]
- 15.Asada T, Simon CZ, Lu AZ, et al. Robot-navigated pedicle screw insertion can reduce intraoperative blood loss and length of hospital stay: analysis of 1,633 patients utilizing propensity score matching. Spine J. 2024;24(1):118-124. doi: 10.1016/j.spinee.2023.09.004. [DOI] [PubMed] [Google Scholar]
- 16.Baky FJ, Milbrandt T, Echternacht S, Stans AA, Shaughnessy WJ, Larson AN. Intraoperative Computed Tomography-Guided Navigation for Pediatric Spine Patients Reduced Return to Operating Room for Screw Malposition Compared With Freehand/Fluoroscopic Techniques. Spine Deform. 2019;7(4):577-581. doi: 10.1016/j.jspd.2018.11.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Bovonratwet P, Nelson SJ, Ondeck NT, Geddes BJ, Grauer JN. Comparison of 30-Day Complications Between Navigated and Conventional Single-level Instrumented Posterior Lumbar Fusion: A Propensity Score Matched Analysis. Spine (Phila Pa 1976). 2018;43(6):447-453. doi: 10.1097/BRS.0000000000002327. [DOI] [PubMed] [Google Scholar]
- 18.Bovonratwet P, Gu A, Chen AZ, et al. Computer-Assisted Navigation Is Associated With Decreased Rates of Hardware-Related Revision After Instrumented Posterior Lumbar Fusion. Global Spine J. 2023;13(4):1104-1111. doi: 10.1177/21925682211019696. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Chen X, Feng F, Yu X, et al. Robot-assisted orthopedic surgery in the treatment of adult degenerative scoliosis: a preliminary clinical report. J Orthop Surg Res. 2020;15(1):282. doi: 10.1186/s13018-020-01796-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Crawford BD, Nchako CM, Rebehn KA, Israel H, Place HM. Transpedicular Screw Placement Accuracy Using the O-Arm Versus Freehand Technique at a Single Institution. Global Spine J. 2022;12(3):447-451. doi: 10.1177/2192568220956979. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Dea N, Fisher CG, Batke J, et al. Economic evaluation comparing intraoperative cone beam CT-based navigation and conventional fluoroscopy for the placement of spinal pedicle screws: a patient-level data cost-effectiveness analysis. Spine J. 2016;16(1):23-31. doi: 10.1016/j.spinee.2015.09.062. [DOI] [PubMed] [Google Scholar]
- 22.Jiang B, Pennington Z, Azad T, et al. Robot-Assisted versus Freehand Instrumentation in Short-Segment Lumbar Fusion: Experience with Real-Time Image-Guided Spinal Robot. World Neurosurg. 2020;136:e635-e645. doi: 10.1016/j.wneu.2020.01.119. [DOI] [PubMed] [Google Scholar]
- 23.Karamian BA, DiMaria SL, Sawires AN, et al. Clinical Outcomes of Robotic Versus Freehand Pedicle Screw Placement After One-to Three-Level Lumbar Fusion. Global Spine J. 2023;13(7):1871-1877. doi: 10.1177/21925682211057491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kaur J, Koltsov JCB, Kwong JW, Cheng I, Vorhies JS. Does Navigation Make Spinal Fusion for Adolescent Idiopathic Scoliosis Safer? Insights From a National Database. Spine (Phila Pa 1976). 2021;46(19):E1049-E1057. doi: 10.1097/BRS.0000000000004037. [DOI] [PubMed] [Google Scholar]
- 25.Lee GY, Massicotte EM, Rampersaud YR. Clinical accuracy of cervicothoracic pedicle screw placement: a comparison of the “open” lamino-foraminotomy and computer-assisted techniques. J Spinal Disord Tech. 2007;20(1):25-32. doi: 10.1097/01.bsd.0000211239.21835.ad. [DOI] [PubMed] [Google Scholar]
- 26.Malik AT, Drain JP, Jones J, et al. Robotic-Assisted Versus Conventional Posterior Lumbar Fusion-An Analysis of 90-Day Complications and Readmissions. World Neurosurg. 2021;152:e168-e174. doi: 10.1016/j.wneu.2021.05.072. [DOI] [PubMed] [Google Scholar]
- 27.Nooh A, Aoude A, Fortin M, et al. Use of Computer Assistance in Lumbar Fusion Surgery: Analysis of 15 222 Patients in the ACS-NSQIP Database. Global Spine J. 2017;7(7):617-623. doi: 10.1177/2192568217699193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Park SM, Kim HJ, Lee SY, Chang BS, Lee CK, Yeom JS. Radiographic and Clinical Outcomes of Robot-Assisted Posterior Pedicle Screw Fixation: Two-Year Results from a Randomized Controlled Trial. Yonsei Med J. 2018;59(3):438-444. doi: 10.3349/ymj.2018.59.3.438. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Pendharkar AV, Rezaii PG, Ho AL, et al. Conventional Versus Stereotactic Image-guided Pedicle Screw Placement During Posterior Lumbar Fusions: A Retrospective Propensity Score-matched Study of a National Longitudinal Database. Spine (Phila Pa 1976). 2019;44(21):E1272-E1280. doi: 10.1097/BRS.0000000000003130. [DOI] [PubMed] [Google Scholar]
- 30.Rezaii PG, Pendharkar AV, Ho AL, et al. Conventional versus stereotactic image guided pedicle screw placement during spinal deformity correction: a retrospective propensity score-matched study of a national longitudinal database. Int J Neurosci. 2021;131(10):953-961. doi: 10.1080/00207454.2020.1763343. [DOI] [PubMed] [Google Scholar]
- 31.Wang C, Zhang H, Zhang L, et al. Accuracy and deviation analysis of robot-assisted spinal implants: A retrospective overview of 105 cases and preliminary comparison to open freehand surgery in lumbar spondylolisthesis. Int J Med Robot. 2021;17(4):e2273. doi: 10.1002/rcs.2273. [DOI] [PubMed] [Google Scholar]
- 32.Wang Y, Hu Y, Liu H, Li C, Li H, Yi X. Navigation Makes Transforaminal Lumbar Interbody Fusion Less Invasive. Orthopedics. 2016;39(5):e857-e862. doi: 10.3928/01477447-20160517-01. [DOI] [PubMed] [Google Scholar]
- 33.Xiao R, Miller JA, Sabharwal NC, et al. Clinical outcomes following spinal fusion using an intraoperative computed tomographic 3D imaging system. J Neurosurg Spine. 2017;26(5):628-637. doi: 10.3171/2016.10.SPINE16373. [DOI] [PubMed] [Google Scholar]
- 34.De la Garza Ramos R, Echt M, Benton JA, et al. Accuracy of Freehand versus Navigated Thoracolumbar Pedicle Screw Placement in Patients with Metastatic Tumors of the Spine. J Korean Neurosurg Soc. 2020;63(6):777-783. doi: 10.3340/jkns.2020.0001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Shuman WH, Valliani AA, Chapman EK, et al. Intraoperative Navigation in Spine Surgery: Effects on Complications and Reoperations. World Neurosurg. 2022;160:e404-e411. doi: 10.1016/j.wneu.2022.01.035. [DOI] [PubMed] [Google Scholar]
- 36.Kao FC, Chang YC, Chen TS, Liu PH, Tu YK. Risk factors for unplanned return to the operating room within 24 hours: A 9-year single-center observational study. Medicine (Baltimore). 2021;100(49):e28053. doi: 10.1097/MD.0000000000028053. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental material for The Impact of Image-Guided Screw Placement on Unexpected Return to the Operating Room: A Systematic Review and Meta-Analysis of 91,285 Patients by Jaskaran Singh, Marion Sylvain, Mikko Vorster, Bona Baissa, Aazad Abbas, and Michael Goytan in Global Spine Journal.
Supplemental material for The Impact of Image-Guided Screw Placement on Unexpected Return to the Operating Room: A Systematic Review and Meta-Analysis of 91,285 Patients by Jaskaran Singh, Marion Sylvain, Mikko Vorster, Bona Baissa, Aazad Abbas, and Michael Goytan in Global Spine Journal.