The Cost of Robot-assisted Total Hip Arthroplasty

Comparing Safety and Hospital Charges to Conventional Total Hip Arthroplasty

Gregory J. Kirchner, MD, MPH; Alexander M. Lieber, MD; Brett Haislup, MD; Yehuda E. Kerbel, MD; Vincent M. Moretti, MD


J Am Acad Orthop Surg. 2021;29(14):609-615. 

In This Article


Previous research has demonstrated that intraoperative robot assistance during THA improves implant accuracy compared with the conventional technique.[11,12,15–20] We have examined whether RA-THA also improve patient safety by examining the risk of perioperative complications and death. Utilization of a large national database facilitated the creation of a large head-to-head comparison of rigorously matched patients. The cohort-matched design enabled minimization of confounding variables that often contaminate retrospective studies of procedures performed on diverse populations. In our comparison, we demonstrate that RA-THA has a similar perioperative safety profile to C-THA, although RA-THA incurs higher hospital charges despite a shorter LOS.

Owing to the high cost of RA-THA, there has been notable scrutiny as to whether the robotic arm is accurate and efficacious in THA. The Mako robotic arm (Stryker) is estimated to cost $730,000 United States Dollars, which does not include annual maintenance costs.[21] Although our study found a statistically significant difference in hospital charges between RA-THA and C-THA, the difference was only $1,788 on average. This difference is relatively minimal, considering the significant cost of robotic navigation technologies. However, the extent to which the cost of intraoperative navigation systems is represented in hospital charges to payers or consumers is unclear. Whether the benefit of RA-THA outweighs its additional costs to payers and hospitals is also uncertain. The current literature is limited, with a paucity of cost effectiveness data and inconsistent findings regarding the outcomes using robot assistance. Some studies have demonstrated improved intraoperative accuracy of the robotic arm, whereas others do not find improved success compared with C-THA.[19,20,22–24]

At least one explanation regarding the increased costs associated with RA-THA involves increased surgical time when using new technology. Heng et al[25] found C-THA surgical time averaged 84.9 minutes, whereas RA-THA averaged 96.7 minutes. The study noted that although a difference between surgical times was observed, RA-THA surgical time did decrease by a minute per operation after the initial operation. This trend is most likely because of increased experience with the robot in each subsequent operation. Because our data represents patients operated on in the early stages of developing RA-THA technology, differences in surgical time likely contributed largely to the differences seen in hospital charges.

Our findings that LOS was shorter in RA-THA patients compared with C-THA patients has been previously substantiated in the literature. Heng et al[25] noted LOS for C-THA to be 5.93 days compared with 4.22 days for RA-THA. In addition, Banchetti et al and Chen et al compared C-THA with RA-THA and found that RA-THA had a shorter LOS than C-THA.[12,26] LOS can be attributed to many variables including complications, pain scores, and functional outcomes. However, our study did not show any notable differences in the risk of perioperative complications. Other studies have shown RA-THA to be associated with equal or lower rates of complication, which may account for shortened LOS in most literature.[12,25,26] Furthermore, shorter LOS among RA-THA patients might also be attributed to surgeon or patient bias. For example, surgeons may be selecting less complicated cases for RA-THA than they otherwise would for C-THA, and patients might have an implicit motivation to perform better postoperatively because their surgery used the use of robotic assistance. Further research is needed to investigate the effects of biases such as those on the early outcomes of RA-THA.

The cohort-matched design represents a major strength of our study. Statistically matched cohorts have been used previously with NIS data to study complications in orthopaedic ankle and spine procedures.[24,27] Although confounding factors are seldom eliminated in observational studies such as this, the statistical matching enables us to minimize selection bias in our retrospective analysis, where random allocation is not available. The addition of a multivariate logistic regression model minimized confounding variables further still. Using the NIS enabled us to create a side-by-side comparison of each surgical technique. In addition, our study analyzes clinically relevant postoperative complications of both techniques. Most comparisons of RA-THA and C-THA have focused on joint mechanics, such as angulation or inclination of various parts of the prosthesis or leg-length discrepancies.

Despite these strengths, there are inherent limitations of using an ICD-9-CM coded database such as the NIS that introduce confounding variables beyond the control of our methods. For example, there may be variability in coding practices between hospitals and physicians, which might result in underreporting of intraoperative monitoring or navigation technology, patient comorbidities, and perioperative complications. In a comparison of the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) and the NIS regarding surgical stabilization of hip fractures, Bohl et al[28] found variable reporting of some comorbidities and adverse events between the two databases.[28,29] Furthermore, the same research estimated that the NIS does not capture approximately 50% of surgical site infections and deaths compared with the ACS NSQIP because NIS data collection ends at discharge, whereas the ACS NSQIP captures all events for 30 days postoperatively. Although the external validity of the NIS has been critiqued, researchers continue to support the use of the NIS for comparing competing procedures because of evidence that it is internally valid.[28] Therefore, the NIS remains a useful source of information for examining trends and outcomes in orthopaedic surgery.

The ICD-9-CM coding of the NIS further limited our ability to examine characteristics specific to orthopaedic hip surgery. The NIS did not enable us to differentiate between the various RA-THA devices used for hip arthroplasty. Patients cannot be stratified as a factor of their hip pathology, and the NIS does not collect preoperative or postoperative radiographic data. Similarly, the NIS does not capture events that occur after patient discharge, such as readmission or return to operating room. Furthermore, pain scores and functional outcomes are not recorded by the NIS. Another important limitation to consider is that the NIS only captures costs associated with a patient's singular inpatient admission. Thus, the NIS fails to capture costs associated with the additional preoperative assessments that are necessary to use robot assistance intraoperatively.

Our cohort-matched study design using a large, nationwide database suggests that RA-THA has a similar perioperative safety profile compared with the conventional technique. However, we demonstrated that RA-THA accrues greater inpatient hospital charges. Although the large upfront costs of RA-THA systems remains a concern for many hospital systems, our research suggests that additional procedure costs should be considered as well.