Robotic-Assistance in Total Hip Arthroplasty Is Associated With Decreased Dislocation Rates

Abstract

Background

As the use of robotics in total hip arthroplasty (THA) continues to gain popularity, differences in clinical outcomes when compared to manual techniques have remained unclear. This study aimed to compare postoperative complications between patients undergoing robotic-assisted techniques and manual THA for primary osteoarthritis at 90 days, 1 year, and 2 years.

Methods

Using an all-payer national database, we identified 405,048 patients who underwent either robotic-assisted or manual THA for primary osteoarthritis. A propensity match was performed for age, sex, a comorbidity index, chronic kidney disease, obesity, and diabetes, resulting in 7652 patients in each cohort. We assessed postoperative outcomes, including surgical site infections, pulmonary emboli, venous thromboemboli, wound complications, dislocations, aseptic revisions, periprosthetic joint infections, and periprosthetic fractures. We completed bivariate analyses via chi-square tests to assess categorical variables. We utilized student’s t-tests to compare continuous variables, including ages and comorbidities. Odds ratios (ORs) were calculated for complications using 95% confidence intervals (CIs).

Results

The robotic-assisted cohort had lower rates of dislocation at 90 days (0.93 vs 1.41%, OR 0.65, 95% CI 0.48-0.88, P = .007), 1 year (1.32 vs 1.92%, OR 0.68, 95% CI 0.53-0.88, P = .004), and 2 years (1.66 vs 2.1%, OR 0.79, 95% CI 0.62-0.99, P = .049). Total surgical complications were significantly lower in the robotic-assisted cohort at 1 year (5.29 vs 6.16%, OR 0.85, 95% CI 0.74-0.98, P = .0205), but were similar at 90 days and 2 years. At 90 days, the rates of medical complications, including surgical site infections, pulmonary emboli, venous thromboemboli, and wound complications, were similar (all P > .05). The rates of periprosthetic joint infections, aseptic revision, periprosthetic fractures, and aseptic loosening were similar at all time points (all P > .05).

Conclusions

Patients who underwent robotic-assisted THA had lower rates of dislocation at 90 days, 1 year, and 2 years. This finding supports the use of robotic assistance in THA, though further research is needed to confirm and strengthen these findings.

Introduction

Instability is the second most common complication following total hip arthroplasty (THA) and is the most common indication for early revision, with reported dislocation rates ranging from 0.5% to 10%. [[1], [2], [3], [4]] Several studies have identified incorrect positioning of THA components as a major contributor to this complication, though the risk factors are multifactorial. [[5], [6], [7]] Historically, the Lewinnek Safe Zone has been utilized to determine acetabular cup placement, though recent literature suggests that it is not an accurate predictor of dislocation risk. [[8], [9], [10], [11]] Understanding hip-spine syndrome and its effect on stability may also be an important factor in determining the proper cup placement for maximal stability. [[12], [13], [14]] The recognized need for precise and patient-specific cup positioning has led many to suggest that the navigation techniques utilized for robotic-assisted cup and stem placement may solve the instability problem in THA.

Available literature consistently reports increased precision associated with acetabular cup placement in robotic-assisted THA. A meta-analysis by Wang et al. compared radiographic findings between manual and robotic THA, revealing that acetabular cups in the robotic-assisted THA group were more likely to be placed in the planned safe zones. [15] Similarly, a prospective review by Elmallah et al. of 224 patients who underwent robotic-assisted THA found that 99% of patients had acetabular cup placement within their preoperatively determined safe zone. [16] Though these zones may not be accurate in predicting dislocation rates, these studies demonstrate the precise component placement associated with robotic hip arthroplasty. Fontalis et al. investigated the outcomes of 30 patients who underwent robotic THA with preoperatively determined functional cup placement based on spino-pelvic alignment. Of these 30 patients, none had experienced a dislocation at 1-year follow-up. [17]

While the literature has demonstrated improved component placement in robotic THA, there is limited and conflicting evidence assessing clinical outcomes. A retrospective review by Shaw et al. demonstrated a significantly lower dislocation rate in patients undergoing robotic-assisted THA (0.6%) compared to manual THA (2.5%) (P < .46). [18] A large study by Bendich et al. investigated the outcomes of over 13,000 patients undergoing THA through a posterior approach at a single institution. They found that patients who underwent THA with robotic-assisted techniques were significantly less likely to undergo reoperation for dislocation within 1 year of the procedure (P = .046). [19] Similarly, when investigating the clinical outcomes associated with 300 THA procedures, 100 of which used robotic assistance, Illgen et al. found a significantly lower dislocation rate (P < .001) when robotic-assisted techniques were used. [20] However, there have also been several studies that have demonstrated no significant difference in dislocation rates between robotic and manual THA. [[21], [22], [23], [24], [25]]

As the popularity of robotic-assisted THA rises, the demand for comprehensive clinical outcome data becomes imperative. To our knowledge, no recent large-cohort national database studies have compared the clinical outcomes of robotic vs manual THA, particularly with the use of a propensity-score match to account for the impact of patient comorbidities. This study aims to compare manual and robotic THA through a large retrospective review to assess: (1) the risk of dislocation and aseptic revision when controlling for comorbidities including lumbar spinal fusion, alcohol abuse, obesity, and diabetes; (2) the risk of additional complications such as periprosthetic joint infection (PJI), aseptic loosening, and periprosthetic fracture between the robotic-assisted and manual cohorts at 90 days, 1 year, and 2 years.

Material and methods

Database

We queried a national, all-payer database (PearlDiver, Colorado Springs, Colorado), containing deidentified Health Insurance Portability and Accountability-compliant records of over 122 million patients across the United States. Patients in the database include commercial, Medicare, Medicaid, government, and cash payers. Data were collected through International Classification of Diseases, 10th revision (ICD-10) codes, Current Procedural Terminology (CPT) codes, and patient demographics. Institutional review board approval was waived for this study due to the use of deidentified information.

Patient selection

All patients who underwent a primary THA for osteoarthritis from January 1, 2010, to January 1, 2022, were identified. (n = 405,048). Exclusion criteria included patients who had hip fractures, prior diagnoses of rheumatoid arthritis, and malignancies. CPT codes and ICD-10 procedure codes were utilized to split patients into 2 groups: manual THA (n = 397,396) and robotic-assisted THA (n = 7652). We compared the incidences of dislocations and aseptic revisions, in addition to several other complications, including PJIs, surgical site infections, pulmonary emboli, venous thromboses, wound complications, periprosthetic fractures (PPFx), and aseptic loosenings. All patients had follow-up data available at 90 days, 1 year, and 2 years postoperatively.

Demographics and matching

Patient demographics included age, sex, and the Elixhauser Comorbidity Index (ECI). [26] The ECI is calculated based on the presence of 38 pre-existing comorbidities. Patient comorbidities included alcohol abuse, chronic kidney disease, hypothyroidism, obesity, tobacco use, anemia, diabetes, and chronic pulmonary disease. There were significant differences (P < .001) in several variables between groups (Table 1), including age, ECI, chronic kidney disease, obesity, and diabetes. Therefore, a 1:1 propensity-score match was conducted based on these variables. After the match, manual and robotic-assisted THA groups each included 7652 patients. Propensity matching was conducted using the R software provided by PearlDiver. Following the propensity match, we verified that rates of lumbar fusion (prior and subsequent) were similar between the groups (P > .05) (Table 1). Additionally, we stratified the matched groups by year of surgery to confirm that they were similar (Fig. 1).

Table 1.

Demographics and baseline characteristics.

Variables	Robotic-assisted THA	Manual THA	P-value
	n = 7652	n = 397,396
Mean age in years	64 ± 11	65 ± 10	<.001
Mean ECI	4.62 ± 3.17	4.4 ± 3.16	<.001
Sex, n (%)			.002
Women	4220 (0.55)	226,223 (0.57)
Men	3432 (0.45)	171,173 (0.43)
Alcohol abuse	630 (0.08)	32,466 (0.08)	.857
CKD	1089 (0.14)	68,667 (0.17)	<.001
Hypothyroidism	2192 (0.29)	110,322 (0.28)	.089
Obesity	3987 (0.52)	190,718 (0.48)	<.001
Tobacco use	3418 (0.45)	171,608 (0.43)	.010
Anemia	1359 (0.18)	68,464 (0.17)	.228
Diabetes	1229 (0.16)	69,458 (0.17)	.001
Chronic pulmonary disease	2508 (0.33)	132,288 (0.33)	.352

ECI, Elixhauser Comorbidity Index; CKD, chronic kidney disease.

Figure 1.

THAs included in cohorts by year of procedure.

Data analyses

We performed a propensity score match with a 1:1 ratio to ensure comparable demographics and comorbidities between patient groups and minimize potential confounding bias. We completed bivariate analyses via chi-square tests to assess categorical variables, including demographics, comorbidities, and complications. Continuous variables, such as age and ECI, were compared using student’s t-tests. Odds ratios (ORs) were also calculated using 95% confidence intervals (CIs). All analyses were performed using R Studio (Statistics Department of the University of Auckland, New Zealand), with significance defined as P < .05.

Results

Ninety-day outcomes

At 90 days, the robotic-assisted cohort had a significantly lower rate of dislocation (0.93 vs 1.41%, OR 0.65, 95% CI 0.48-0.88, P = .007). The robotic-assisted cohort had similar rates of PJI (0.88 vs 1.15%, P = .106), aseptic revision (0.94 vs 0.81%, P = .435), PPFx (0.91 vs 0.71%, P = .176), aseptic loosening (0.18 vs 0.2%, P = 1), and total complications (3.84 vs 4.27%, P = .178). The robotic-assisted cohort also had similar rates of medical complications, including surgical site infections (0.91 vs 1.15%, P = .174), pulmonary emboli (0.16 vs 0.21%, P = .570), venous thromboses (0.59 vs 0.68%, P = .541), and wound complication (1.22 vs 0.95%, P = .138).

One-year outcomes

At 1 year, the rate of dislocation remained significantly lower for the robotic-assisted cohort (1.32 vs 1.92%, OR 0.68, 95% CI 0.53-0.88, P = .004). Additionally, the total complication rate was lower in the robotic-assisted cohort (5.29 vs 6.16%, OR 0.85, 95% CI 0.74-0.98, P = .021). The rates of PJI, aseptic revision, PPFx, and aseptic loosening remained similar at 1 year (all P > .05).

Two-year outcomes

At 2 years, the rate of dislocation remained significantly lower for the robotic-assisted cohort (1.66 vs 2.1%, OR 0.79, 95% CI 0.62-0.99, P = .049). The rates of PJI, aseptic revision, PPFx, aseptic loosening, and total complications remained similar at 2 years (all P > .05) (Tables 2 and 3).

Table 2.

Incidence of 90-d, 1-y, and 2-y complications.

Outcomes	Robotic-assisted THA	Manual THA	P-value
	n = 7652 (%)	n = 7652 (%)
90-D complications
PJI	67 (0.88)	88 (1.15)	.106
Aseptic revision	72 (0.94)	62 (0.81)	.435
SSI	70 (0.91)	88 (1.15)	.174
PE	12 (0.16)	16 (0.21)	.570
VT	45 (0.59)	52 (0.68)	.541
WC	93 (1.22)	73 (0.95)	.138
Dislocation	71 (0.93)	108 (1.41)	.007
Periprosthetic fracture	70 (0.91)	54 (0.71)	.176
Aseptic loosening	14 (0.18)	15 (0.2)	1
Total complications	294 (3.84)	327 (4.27)	.178
1-Y complications
PJI	87 (1.14)	112 (1.46)	.087
Aseptic revision	100 (1.31)	108 (1.41)	.625
Dislocation	101 (1.32)	147 (1.92)	.004
Periprosthetic fracture	80 (1.05)	65 (0.85)	.243
Aseptic loosening	37 (0.48)	39 (0.51)	.909
Total complications	405 (5.29)	471 (6.16)	.021
2-Y complications
PJI	101 (1.32)	118 (1.54)	.278
Aseptic revision	122 (1.59)	130 (1.7)	.657
Dislocation	127 (1.66)	161 (2.1)	.049
Periprosthetic fracture	92 (1.2)	70 (0.91)	.097
Aseptic loosening	62 (0.81)	53 (0.69)	.454
Total complications	504 (6.59)	532 (6.95)	.375

PJI, periprosthetic joint infection; SSI, surgical site infection; PE, pulmonary embolism; VT, venous thrombosis; WC, wound complication.

Table 3.

Odds ratios of 90-d, 1-y, and 2-y complications.

Outcomes	Robotic-assisted THA
	Odds ratio	95% CI
90-D complications
PJI	0.76	0.55-1.04
Aseptic revision	1.16	0.83-1.64
SSI	0.79	0.58-1.09
PE	0.75	0.35-1.59
VT	0.86	0.58-1.29
WC	1.28	0.94-1.74
Dislocation	0.65	0.48-0.88
Periprosthetic fracture	1.3	0.91-1.86
Aseptic loosening	0.93	0.45-1.93
Total complications	0.90	0.76-1.05
1-Y complications
PJI	0.77	0.58-1.03
Aseptic revision	0.92	0.70-1.22
Dislocation	0.68	0.53-0.88
Periprosthetic fracture	1.23	0.89-1.71
Aseptic loosening	0.95	0.60-1.49
Total complications	0.85	0.74-0.98
2-Y complications
PJI	0.85	0.65-1.12
Aseptic revision	0.94	0.73-1.20
Dislocation	0.79	0.62-0.99
Periprosthetic fracture	1.32	0.96-1.80
Aseptic loosening	1.17	0.81-1.69
Total complications	0.94	0.83-1.07
Lumbar fusion
Preceding fusion	0.935	0.78-1.12
Subsequent fusion	1.13	0.83-1.52

CI, confidence interval; PJI, periprosthetic joint infection; SSI, surgical site infection; PE, pulmonary embolism; VT, venous thrombosis; WC, wound complication.

Discussion

Emara et al. project that by 2030, robotic assistance will be utilized in 65.8% of THAs. [27] As technological advancements continue to influence the practice of orthopaedic surgery, it is crucial that we understand how these changes actually impact patient care and outcomes. The present study utilized a national database to compare clinical outcomes at 90 days, 1 year, and 2 years following manual or robotic-assisted THA. We found that there were significantly fewer dislocations in the robotic-assisted THA group at 90 days, 1 year, and 2 years. Additionally, total surgical complications were significantly lower in the robotic-assisted group at 1 year. Other outcomes, such as PJI and aseptic revision, were similar between groups at all time points.

The precision of component placement is widely reported in the literature, though this is not always correlated with clinical outcomes. A meta-analysis by Wang et al. identified 18 studies with 2845 hips comparing the radiographic outcomes of robotic-assisted and manual THA. Robotic-assisted THAs were significantly more likely to have cups placed in the Lewinnek and Callanan safe zones (P < .00001) as planned preoperatively. Surprisingly, this study found that robotic assistance was associated with higher rates of dislocation (P < .05). However, it is important to note that 3 of the 7 studies were performed prior to 2010, when the robotic technology that is currently used in THA became available. [15]

In terms of clinical outcomes, the existing literature demonstrates varying findings regarding dislocation rates between manual and robotic-assisted THA, with some studies demonstrating decreased dislocation rates in robotic-assisted THA, similar to the present one. A retrospective review of 300 patients by Illgen et al. investigated the outcomes of patients undergoing robotic-assisted THA. Patients were split into 3 groups, each consisting of 100 patients. Patients in group 1 underwent manual THA in 2000; patients in group 2 underwent manual THA in 2010; and patients in group 3 underwent robotic-assisted THA in 2012. At a 2-year follow-up, they observed a significantly lower dislocation rate in the robotic group (0%) when compared to the early manual (5%) and late manual (3%) groups (P < .001). [20] A retrospective review by Shaw et al. compared outcomes from 2247 total hip arthroplasties performed by a cohort of 3 surgeons. Of these surgeries, 523 were performed with robotic assistance, and 1724 were performed manually. They observed a significantly lower dislocation rate in the robotic group (0.6%) compared to the manual group (2.5%) (P < .46). [18] A review of 24 studies by Chen et al. found that 23 out of 24 studies supported the use of robotic technology in THA, based on observed improvements in dislocation rates and clinical outcomes. [28]

While several studies demonstrate reduced dislocation rates with the use of robotic assistance in THA, there is also literature demonstrating no difference in this complication. Remily et al. compared the outcomes of manual and robotic THAs performed from 2010 to 2018, finding no difference in dislocation rates between groups (P > .498). [24] A systematic review by Sweet et al. in 2021 examined 7 studies comparing robotic and manual THA. While 6 studies indicated that robotic THA was associated with more accurate component placement, all but one demonstrated no difference in dislocation rates. [29]

Though other studies have demonstrated no difference in dislocation rates, the main difference in the present study is the inclusion of data from 2018 to 2022, which correlates with a rise in popularity of robotic THA and encompasses a larger population of THA patients. Figure 1 demonstrates that within the matched cohorts, the majority of the robotic cases occurred after 2018. Additionally, exclusion criteria, which may have an impact on the dislocation rate, were utilized in this current study. Furthermore, propensity-score matching was completed to further minimize confounders.

Current understanding is that stability of the hip is multifactorial and that many patients may require cup placement outside of the “safe zone,” which does not account for concurrent spine disease. Sultan et al. performed a review of 14 studies with 1167 patients to further investigate this relationship. They found that patients who have concurrent spinal deformities and THA are at increased risk for dislocations and revisions. Additionally, they demonstrated that correction of spinal deformity resulted in a reduction in acetabular anteversion (mean reduction of 8.08 degrees, P < .001) and tilt (mean reduction of 7 degrees, P < .001). This finding supports the idea that optimal cup placement in patients who have spinal deformities may differ from those who do not. [30] Further research should focus on the synergistic use of functional cup placement and robotic assistance to confer stability. In conjunction with functional cup placement in the setting of hip-spine syndrome, robotic-assisted THA appears to be promising. Fontalis et al. investigated this relationship in a study of 30 patients who underwent robotic-assisted THA. All 30 patients had concomitant spine stiffness and underwent preoperative dynamic spine imaging. Functional cup placement was determined based on these images, and robotic assistance was used for precise component positioning. At 1-year follow-up, none of these patients had experienced a dislocation. [17]

Our study acknowledges certain potential limitations related to the utilization of a large-scale insurance database. As the PearlDiver database relies on the entry of procedural coding and billing information, errors in data entry may lead to inaccuracies in the available data. However, these risks are mitigated through routine third-party audits. Additional concerns with PearlDiver include the limited availability of patient follow-up data due to de-identified information, as well as sampling bias due to insurance plan participation in the database. We were able to utilize ICD-10 and CPT codes to identify complications within the 2-year follow-up period. Another limitation of this study is the lack of available data on the surgical approach used in each cohort, as the surgical approach is known to influence dislocation rates following THA. The identified outcomes may also not apply in patients undergoing THA in the setting of inflammatory arthritis, fracture, or malignancy. Despite these limitations, our findings are strengthened by the large sample size and propensity-score matching to minimize confounders.

Conclusions

This study demonstrated decreased dislocation rates in patients undergoing robotic-assisted THA in a large database with up to 2-year outcomes, supporting the use of this technology. However, additional research is necessary to strengthen these conclusions.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Conflicts of interest

J. Nace is a paid consultant for Microport Orthopedics; receives research support from Microport Orthopedics, Stryker, and United Orthopedic Corporation; is an editorial/governing board member of the Journal of Arthroplasty, the Journal of the American Osteopathic Medicine Association, Orthopaedic Knowledge, and the Journal of Knee Surgery, Knee; and is a board/committee member of Arthritis Foundation. M. A. Mont receives royalties from Medicus Works LLC and Stryker; is a paid consultant for 3M, Exactech, Inc, Johnson & Johnson, Kolon TissueGene, Next Science, Pacira, Smith & Nephew, and Stryker; has stock options in CERAS Health, MirrorAR, Peerwell, and USMI; receives research support from Johnson & Johnson, Medtronic, National Institutes of Health (NIAMS & NICHD), Organogenesis, Orthofix, Inc, Patient-Centered Outcomes Research Institute (PCORI), and Stryker; receives royalties or financial support from Medicus Works LLC, Up-to Date, Wolters Kluwer Health - Lippincott Williams & Wilkins; is an editorial/governing board member of the Journal of Arthroplasty, the Journal of Knee Surgery, Surgical Techniques International; and is a board/committee member of the American Association of Hip and Knee Surgeons, Hip Society, and Knee Society. R. E. Delanois receives research support from Johnson & Johnson, Biocomposites, CyMedica Orthopedics, Depuy Synthes Product, Flexion Therapeutics, Microport, Orthopedics, Orthofix, Patient-Centered Outcomes Research Institute (PCORI), Smith & Nephew, Stryker, Tissue Gene, and United Orthopedic Corporation; is an editorial board member of the Journal of Knee Surgery; and is a board/committee member of Baltimore City Medical Society. All other authors declare no potential conflicts of interest.

For full disclosure statements refer to https://doi.org/10.1016/j.artd.2024.101473.

CRediT authorship contribution statement

Gabrielle N. Swartz: Conceptualization, Data curation, Writing – original draft. Sandeep S. Bains: Data curation, Software, Writing – review & editing. Jeremy A. Dubin: Methodology, Writing – review & editing. Reza Katanbaf: Data curation, Methodology, Writing – review & editing. Hunter Hayes: Writing – review & editing, Supervision. James Nace: Visualization, Writing – review & editing, Supervision. Michael A. Mont: Supervision, Visualization, Writing – review & editing. Ronald E. Delanois: Investigation, Methodology, Supervision, Visualization, Writing – review & editing.