Abstract
Background
Robotic-assisted unicompartmental knee arthroplasty (UKA) has gained popularity for its potential to improve implant precision and reduce surgical errors. However, comparative evidence on short-term outcomes versus conventional UKA is lacking. Thus, the purpose of this study was to compare the short-term outcomes of robotic-assisted versus conventional UKA using a nationally representative database.
Methods
The Nationwide Readmissions Database 2016–2020 was retrospectively examined to identify adult patients who received an elective UKA. After applying exclusion criteria and 1:2 propensity score matching (PSM), 8310 patients were included in the analysis. Outcomes included in-hospital complications, implant malposition or failure, perioperative fracture, length of hospital stay (LOS), hospital costs, and 30- and 90-day readmission rates. Multivariable regression analyses were performed to adjust for residual confounding factors.
Results
Robotic-assisted UKA was associated with significantly lower complication rates compared with conventional UKA (3.7% versus 13.2%, p < 0.001). Specifically, robotic-assisted procedures had reduced risks of implant malposition or failure (odds ratio [OR] = 0.08; 95% confidence interval [CI]: 0.05–0.13; p < 0.001) and perioperative fracture (OR = 0.18; 95% CI 0.04–0.76; p = 0.020). No significant differences were observed in LOS, total hospital costs, or readmission rates at 30 and 90 days.
Conclusions
Robotic-assisted UKA is associated with improved short-term surgical safety, including fewer complications, particularly, reduced implant malposition and perioperative fractures. However, broader hospital metrics such as LOS, cost, and readmissions were comparable between the two approaches. Further prospective studies are needed to validate these findings and assess long-term outcomes and cost-effectiveness.
Level of evidence
Level III.
Clinical trial registration number
Not applicable.
Supplementary Information
The online version contains supplementary material available at 10.1186/s10195-025-00896-1.
Keywords: National Readmission Database, Osteoarthritis, Robotic-assisted surgery, Short-term outcomes, Unicompartmental knee arthroplasty
Introduction
Knee osteoarthritis (OA) is a leading cause of chronic pain and disability worldwide, affecting approximately 10% of men and 13% of women ≥ 60 years old. As the global population ages and obesity rates rise, the prevalence of knee OA is expected to increase, placing a significant burden on healthcare systems [1]. The global prevalence of OA is over 7%, with higher rates in countries with established market economies [2]. Reported in 2019, the economic burden of OA ranged from 1% to 2.5% of gross national product in developed countries, with average annual costs per individual estimated between US $700 and US $15,600 [2]. Risk factors for OA include advancing age, female sex, obesity, joint injury, and genetic susceptibility [3]. In older adults, OA often coexists with other chronic conditions, potentially worsening their adverse long-term prognosis [4].
While conservative treatments such as physical therapy, intra-articular steroid injections, and pharmacologic therapy provide symptomatic relief, surgical interventions remain the definitive treatment for patients with advanced knee OA who fail nonoperative management. Of the different surgical methods, unicompartmental knee arthroplasty (UKA) is an effective option for patients with isolated compartmental knee OA, offering faster recovery and preservation of native knee structures as compared with total knee arthroplasty (TKA) [5, 6].
Furthermore, studies have demonstrated that robotic-assisted UKA leads to improved component alignment accuracy [7, 8]. While robotic procedures typically require longer operative time, they are associated with better short-term functional outcomes and lower reintervention rates [8, 9]. Patients undergoing robotic-assisted UKA have reported higher Oxford Knee Scores and lower pain scores postoperatively [7, 8, 10, 11]. Although robotic-assisted UKA is increasingly adopted, robust comparative evidence with conventional UKA remains limited. Concerns persist regarding longer operative times and higher healthcare costs [8, 12]. Larger prospective studies with extended follow-up in real-world, large-scale datasets are needed to validate the advantages of robotic-assisted UKA.
Therefore, this study aimed to comprehensively compare the short-term outcomes between robotic and conventional UKA using a nationally representative readmission database. We hypothesized that compared with conventional UKA, robotic-assisted UKA would be associated with more favorable short-term outcomes, including shorter length of hospital stay (LOS), fewer perioperative complications, and lower readmission rates.
Methods
Data source and study design
Data from the US Nationwide Readmissions Database (NRD) were retrospectively analyzed. The NRD is a publicly available all-payer database developed by the Agency for Healthcare Research and Quality as part of the Healthcare Cost and Utilization Project. The NRD is derived from the Healthcare Cost and Utilization State Inpatient Databases and provides a comprehensive representation of hospitalizations and readmissions across the USA, regardless of insurance provider.
With verified patient linkage numbers, the NRD enables tracking of individuals across hospitals within a given year while maintaining strict adherence to privacy regulations. It includes a full calendar year of data, with diagnoses and procedures coded using the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) and ICD-10 Procedure Coding System (ICD-10-PCS) starting from the 2016 data year. More details about the NRD can be found at: https://hcup-us.ahrq.gov/nrdoverview.jsp.
Study population
Hospitalized adults aged ≥ 18 years who underwent a UKA between 2016 and 2020 and were identified in the NRD database were eligible for inclusion in this study. Patients who received emergency surgery, and those with missing sex, outcome, and dataset sample weight were excluded. Patients undergoing emergency surgery were excluded because UKA is typically performed as an elective procedure. Emergency cases often involve different clinical contexts, which could confound the outcomes of interest and reduce comparability between groups. The included patients were then categorized into robotic-assisted UKA versus conventional UKA. Robotic-assisted UKA was identified using ICD-10-PCS codes that specifically capture robotic arm-assisted procedures (e.g., MAKO-like platforms), rather than computer-assisted navigation. Detailed ICD codes used for identifying the procedures can be found in Table 1.
Table 1.
ICD codes used to identify diagnoses and procedures
| ICD-10-CM/ICD-10-PCS | |
|---|---|
| Unicompartmental knee arthroplasty (UKA) | ICD-10-PCS: 0SRC0L, 0SRC0M, 0SRC0N, 0SRD0L, 0SRD0M, 0SRD0N |
| Robot-assisted surgery | ICD-10-PCS: 8E0Y0C, 8E0Y3C, 8E0Y4C, 8E0YXC |
| Acute myocardial infarction (AMI) | I21 |
| Cerebrovascular accident (CVA) | I60, I61, I63, I69 |
| Venous thromboembolism (VTE) | I26.0, I26.9, I80.0, I80.1, I80.2, I80.3, I80.8, I80.9, I81, I82 |
| Sepsis | R78.81, A41, R65.2, A42.7, A22.7, B37.7, A26.7, A28.2, A54.86, A32.7, A39.2, A20.7, A21.7, A48.3, A24.1 |
| Shock | R57, T81.1, T88.2, R65.21, A48.3 |
| Pneumonia | J13, J14, J15, J16, J17, J18 |
| Surgical site infection | T81.4 |
| Urinary tract infection (UTI) | N39.0 |
| Respiratory failure | J95.2, J95.3, J95.4, J95.5, J95.6, J95.7, J95.8, J96.00, J96.90, J80, J81.0 |
| Mechanical ventilation | Z99.12, ICD-10-PCS: 5A1935Z, 5A1945Z, 5A1955Z |
| Acute kidney injury (AKI) | N17 |
| Bleeding/hematoma | M96.81, M96.83 |
| Implant malposition or failure | T84 |
| Nervous system complication | S84, G97 |
| Perioperative fracture | M96.6, M97 |
| Delirium | F05 |
| Diabetes mellitus (DM) | E10, E11, E12, E13 |
| Hypertension | I10, O10.0, O10.9, I16, I67.4 |
| Coronary heart disease | I25 |
| Congestive heart failure | I09.9, I11.0, I13.0, I13.2, I25.5, I42.0, I42.5, I42.6, I42.7, I42.8, I42.9, I43, I50, P29.0 |
| Connective tissue disorders | M30, M31, M32, M33, M34, M35, M36 |
| Hypothyroidism | E00, E01, E02, E03 |
| Hyperlipidemia | E78 |
| Chronic kidney disease | N18 |
| Chronic pulmonary disease | I27.8, I27.9, J40-J47, J60-J67, J68.4, J70.1, J70.3 |
| Obesity | E66 |
| Severe liver disease | I85.0, I86.4, K70.4, K71.1, K72.1, K72.9, K76.5, K76.6, K76.7 |
| Osteoporosis | M80, M81 |
| Depression | F32, F33 |
| Dementia | F01, F02, F03 |
| Systemic connective tissue disorders | M30, M31, M32, M33, M34, M35, M36 |
ICD, International Classification of Diseases; CM, Clinical Modification; PCS, Procedure Coding System
Ethics statement
This study used precollected, anonymized data, where no patients were directly involved. The lack of personally identifiable information aligns with ethical research standards. The study protocol was submitted to our institution’s institutional review board, which confirmed that no additional approval was necessary.
Outcomes
The outcomes assessed included complications, LOS, total hospital costs, and 30-day readmission and 90-day readmission rates. Complications were identified using corresponding ICD-9 and ICD-10 diagnostic codes and included in-hospital mortality, acute myocardial infarction (AMI), cerebrovascular disease, venous thromboembolism (VTE), sepsis or shock, pneumonia, urinary tract infection, respiratory failure or need for mechanical ventilation, acute kidney injury (AKI), bleeding or hematoma, implant malposition or failure, nervous system complications, perioperative fracture, and delirium, as summarized in Table 1. Implant malposition or failure was identified using ICD-10 codes under T84, which encompass postoperative mechanical complications of orthopedic implants.
Covariates
Patient characteristics included in the analysis were age, categorized by age group (18–39, 40–49, 50–59, 60–69, 70–79, and ≥ 80 years), sex, household income, insurance status/primary payer, and admission type (emergent, elective). Major comorbidities were identified using the ICD coding system, and included diabetes mellitus, hypertension, coronary heart disease, congestive heart failure, connective tissue disorders, hypothyroidism, hyperlipidemia, chronic kidney disease (CKD), chronic pulmonary disease, obesity, severe liver disease, osteoporosis, depression, dementia, and systemic connective tissue disorders. The Charlson Comorbidity Index (CCI) was calculated for each patient to represent the overall comorbidity burden. The ICD codes used for identifying the comorbid conditions are summarized in Table 1. These diagnoses were made by US clinicians on the basis of their clinical judgment and relevant guidelines, and were recorded in the dataset accordingly.
Statistical analysis
Descriptive statistics were used to summarize patient demographic and clinical characteristics. Categorical variables were presented as count and weighted percentage, while continuous variables were reported as mean with standard error. Group comparisons for categorical variables were conducted using the Rao–Scott chi-squared test, while weighted mean differences for continuous variables were analyzed using survey methods that account for stratification, clustering, and sampling weights, ensuring valid and robust statistical inferences within the context of complex survey designs. To minimize confounding, propensity score matching (PSM) was performed using a 1:2 ratio, with household income, insurance status/primary payer, diabetes mellitus, hypertension, congestive heart failure, CKD, depression, and CCI as the matching variables. Age and sex were not used as matching variables because they were balanced between groups after weighting, and body mass index is not available in the NRD. The matching process followed a one-to-many approach [13]. The method prioritizes “best” matches first, and then proceeds with “next-best” matches until no more can be made. Logistic regression analysis was used to calculate odds ratios (ORs) and 95% confidence intervals (CIs), and linear regression analysis was used to calculate beta and 95% CI for outcomes, adjusting for covariates with significant differences between the matched groups. A two-sided p-value of < 0.05 was considered statistically significant. All analyses accounted for the NRD’s complex survey design to ensure accurate national estimates. Statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).
Results
Patient selection
The selection process for the study population is shown in Fig. 1. A total of 18,108 adults received a UKA during the period from 2016 to 2020. Patients who had missing information on total hospital costs or underwent emergent surgery were excluded (n = 1161), leaving a total of 16,753 patients. After 1:2 PSM, 8310 patients remained for subsequent analyses. After applying the NRD-provided discharge weights, this sample corresponds to an estimated 14,823 patients nationwide (Fig. 1).
Fig. 1.
Flow diagram of patient selection
Patient characteristics
The characteristics of all patients before PSM are summarized in Additional File 1: Supplementary Table S1. The mean age of the study population was 64 years, and 53% were female. Significant between-group differences were observed in household income, insurance status/primary payer, diabetes mellitus, hypertension, congestive heart failure, CKD, depression, and CCI.
Patient characteristics after PSM are summarized in Table 2. A significant difference was still present in the proportion of patients with coronary heart disease between the robot-assisted and conventional UKA groups (Table 2).
Table 2.
Patient characteristics
| Characteristics | All patients (N = 8310) | After propensity score matching | p-Value | |
|---|---|---|---|---|
| Robot-assisted (n = 2770) | Conventional (n = 5540) | |||
| Outcomes | ||||
| Any complications | 847 (10.0) | 115 (3.7) | 732 (13.2) | < 0.001 |
| In-hospital mortality | 1 (0.01) | 1 (0.02) | 0 (0.0) | – |
| AMI | 7 (0.1) | 2 (0.05) | 5 (0.1) | 0.315 |
| CVA | 26 (0.3) | 11 (0.4) | 15 (0.3) | 0.452 |
| VTE | 13 (0.1) | 5 (0.1) | 8 (0.2) | 0.891 |
| Sepsis/shock | 7 (0.1) | 2 (0.05) | 5 (0.1) | 0.496 |
| Pneumonia | 10 (0.1) | 4 (0.1) | 6 (0.1) | 0.573 |
| Surgical site infection | 1 (0.01) | 0 (0.0) | 1 (0.01) | – |
| UTI | 44 (0.5) | 15 (0.5) | 29 (0.5) | 0.822 |
| Respiratory failure/mechanical ventilation | 11 (0.1) | 3 (0.1) | 8 (0.1) | 0.339 |
| AKI | 100 (1.2) | 41 (1.3) | 59 (1.1) | 0.316 |
| Bleeding/hematoma | 0 (0.0) | 0 (0.0) | 0 (0.0) | – |
| Implant malposition or failure | 651 (7.8) | 32 (1.0) | 619 (11.1) | < 0.001 |
| Nervous system complication | 3 (0.03) | 2 (0.1) | 1 (0.01) | 0.120 |
| Perioperative fracture | 25 (0.3) | 2 (0.1) | 23 (0.4) | 0.008 |
| Delirium | 10 (0.1) | 4 (0.1) | 6 (0.1) | 0.769 |
| LOSa | 1.6 ± 0.04 | 1.6 ± 0.09 | 1.6 ± 0.03 | 0.799 |
| Total hospital charge, 1000 USD | 61.3 ± 1.2 | 63.5 ± 2.3 | 60.2 ± 1.1 | 0.195 |
| 30-day readmissiona | 173 (2.1) | 56 (1.9) | 117 (2.1) | 0.600 |
| 90-day readmissiona | 441 (5.4) | 136 (4.8) | 305 (5.7) | 0.107 |
| Demography | ||||
| Age, years | 63.4 ± 0.2 | 63.5 ± 0.3 | 63.4 ± 0.2 | 0.923 |
| 18–39 | 118 (1.5) | 38 (1.4) | 80 (1.6) | 0.685 |
| 40–49 | 705 (8.7) | 237 (8.7) | 468 (8.6) | |
| 50–59 | 2091 (25.7) | 717 (26.6) | 1374 (25.2) | |
| 60–69 | 2878 (34.5) | 923 (33.4) | 1955 (35.0) | |
| 70–79 | 1908 (22.7) | 639 (22.6) | 1269 (22.8) | |
| 80+ | 610 (6.9) | 216 (7.2) | 394 (6.8) | |
| Sex | 0.709 | |||
| Male | 3936 (47.7) | 1286 (47.4) | 2650 (47.9) | |
| Female | 4374 (52.3) | 1484 (52.6) | 2890 (52.1) | |
| Household income | 0.625 | |||
| Q1 | 1112 (14.2) | 375 (13.3) | 737 (14.6) | |
| Q2 | 1962 (26.0) | 650 (26.5) | 1312 (25.7) | |
| Q3 | 2265 (27.8) | 759 (28.7) | 1506 (27.4) | |
| Q4 | 2846 (32.0) | 947 (31.5) | 1899 (32.3) | |
| Missing | 125 | 39 | 86 | |
| Insurance status/primary payer | 0.830 | |||
| Medicare/Medicaid | 4116 (48.6) | 1365 (48.0) | 2751 (48.9) | |
| Private including HMO | 3720 (45.9) | 1248 (46.5) | 2472 (45.6) | |
| Self-pay/no charge/other | 474 (5.5) | 157 (5.4) | 317 (5.5) | |
| Hospital characteristics | ||||
| Hospital bed capacity | 0.549 | |||
| Small (< 250) | 2544 (30.9) | 886 (30.8) | 1658 (31.0) | |
| Medium (250–450) | 2419 (30.3) | 813 (32.4) | 1606 (29.2) | |
| Large (> 450) | 3347 (38.8) | 1071 (36.8) | 2276 (39.7) | |
| Location/teaching status | 0.229 | |||
| Metropolitan nonteaching | 2194 (24.6) | 699 (24.3) | 1495 (24.8) | |
| Metropolitan teaching | 5481 (65.3) | 1870 (63.4) | 3611 (66.2) | |
| Nonmetropolitan hospital | 635 (10.1) | 201 (12.3) | 434 (9.0) | |
| Weekend admission | 0.337 | |||
| No | 8236 (99.2) | 2740 (99.1) | 5496 (99.3) | |
| Yes | 74 (0.8) | 30 (0.9) | 44 (0.7) | |
| Major comorbidities | ||||
| Diabetes mellitus | 1253 (14.8) | 409 (14.2) | 844 (15.0) | 0.339 |
| Hypertension | 4132 (49.3) | 1379 (48.2) | 2753 (49.9) | 0.171 |
| Coronary heart disease | 768 (9.1) | 284 (10.1) | 484 (8.7) | 0.034 |
| Congestive heart failure | 181 (2.2) | 60 (2.0) | 121 (2.3) | 0.364 |
| Connective tissue disorders | 61 (0.8) | 25 (0.9) | 36 (0.8) | 0.438 |
| Hypothyroidism | 1089 (13.2) | 375 (13.2) | 714 (13.2) | 0.970 |
| Hyperlipidemia | 3511 (42.2) | 1194 (41.8) | 2317 (42.3) | 0.681 |
| Chronic kidney disease | 364 (4.5) | 120 (4.2) | 244 (4.6) | 0.415 |
| Chronic pulmonary disease | 1179 (14.1) | 406 (14.1) | 773 (14.1) | 0.974 |
| Obesity | 2056 (24.4) | 726 (25.1) | 1330 (24.0) | 0.521 |
| Severe liver disease | 6 (0.1) | 2 (0.1) | 4 (0.1) | 0.986 |
| Osteoporosis | 252 (2.9) | 90 (2.9) | 162 (2.9) | 0.895 |
| Depression | 939 (11.4) | 319 (11.5) | 620 (11.3) | 0.776 |
| Dementia | 30 (0.4) | 10 (0.4) | 20 (0.4) | 0.882 |
| Systemic connective tissue disorders | 61 (0.8) | 25 (0.9) | 36 (0.8) | 0.438 |
| CCI | 0.621 | |||
| 0 | 5447 (66.0) | 1812 (66.8) | 3635 (65.7) | |
| 1 | 1912 (22.5) | 629 (21.6) | 1283 (22.9) | |
| 2 | 506 (6.2) | 175 (6.3) | 331 (6.2) | |
| 3 | 220 (2.7) | 79 (2.9) | 141 (2.6) | |
| 4+ | 225 (2.7) | 75 (2.5) | 150 (2.7) | |
Continuous variables are presented as mean ± standard error (SE); categorical variables are presented as unweighted counts (weighted percentage)
aExcluding patients who died in the hospital
p-Values < 0.05 are shown in bold
AMI, acute myocardial infarction; CVA, cerebrovascular accident; VTE, venous thromboembolism; UTI, urinary tract infection; AKI, acute kidney injury; LOS, length of stay; CCI, Charlson Comorbidity Index; UKA, unicompartmental knee arthroplasty; USD, US dollars
Patient outcomes
After PSM, patients who underwent robot-assisted UKA had a lower complication rate (3.7% versus 13.2%, p < 0.001) than those who underwent conventional UKA. Similarly, compared with patients who underwent conventional UKA, those who underwent robot-assisted UKA had a lower frequency of implant malposition or failure (1% versus 11%, p < 0.001) and perioperative fracture (0.1% versus 0.4%, p = 0.008) (Table 3).
Table 3.
Outcomes between robot-assisted versus conventional UKA, after PSM
| Outcomes | All patients (N = 8310) | After propensity score matching | p-Value | |
|---|---|---|---|---|
| Robot-assisted (n = 2770) | Conventional (n = 5540) | |||
| Complications, any | 847 (10.0) | 115 (3.7) | 732 (13.2) | < 0.001 |
| In-hospital mortality | 1 (0.01) | 1 (0.02) | 0 (0.0) | – |
| AMI | 7 (0.1) | 2 (0.05) | 5 (0.1) | 0.315 |
| CVA | 26 (0.3) | 11 (0.4) | 15 (0.3) | 0.452 |
| VTE | 13 (0.1) | 5 (0.1) | 8 (0.2) | 0.891 |
| Sepsis/shock | 7 (0.1) | 2 (0.05) | 5 (0.1) | 0.496 |
| Pneumonia | 10 (0.1) | 4 (0.1) | 6 (0.1) | 0.573 |
| Surgical site infection | 1 (0.01) | 0 (0.0) | 1 (0.01) | – |
| UTI | 44 (0.5) | 15 (0.5) | 29 (0.5) | 0.822 |
| Respiratory failure/mechanical ventilation | 11 (0.1) | 3 (0.1) | 8 (0.1) | 0.339 |
| AKI | 100 (1.2) | 41 (1.3) | 59 (1.1) | 0.316 |
| Bleeding/hematoma | 0 (0.0) | 0 (0.0) | 0 (0.0) | – |
| Implant malposition or failure | 651 (7.8) | 32 (1.0) | 619 (11.1) | < 0.001 |
| Nervous system complication | 3 (0.03) | 2 (0.1) | 1 (0.01) | 0.120 |
| Perioperative fracture | 25 (0.3) | 2 (0.1) | 23 (0.4) | 0.008 |
| Delirium | 10 (0.1) | 4 (0.1) | 6 (0.1) | 0.769 |
| LOSa | 1.6 ± 0.04 | 1.6 ± 0.09 | 1.6 ± 0.03 | 0.799 |
| Total hospital charge, 1000 USD | 61.3 ± 1.2 | 63.5 ± 2.3 | 60.2 ± 1.1 | 0.195 |
| 30-day readmissiona | 173 (2.1) | 56 (1.9) | 117 (2.1) | 0.600 |
| 90-day readmissiona | 441 (5.4) | 136 (4.8) | 305 (5.7) | 0.107 |
Continuous variables are presented as mean ± SE; categorical variables are presented as unweighted counts (weighted percentage)
aExcluding patients who died in the hospital
p-Values < 0.05 are shown in bold
AMI, acute myocardial infarction; CVA, cerebrovascular accident; VTE, venous thromboembolism; UTI, urinary tract infection; AKI, acute kidney injury; LOS, length of stay; CCI, Charlson Comorbidity Index; UKA, unicompartmental knee arthroplasty
Logistic regression analyses results are summarized in Table 4. After adjustment in the multivariable analysis, patients who underwent robot-assisted UKA had a lower risk of overall complications (odds ratio [OR] = 0.25, 95% CI 0.19–0.32, p < 0.001) as compared with those that received a conventional UKA. Furthermore, patients who underwent robot-assisted UKA had a lower risk of implant malposition or failure (OR = 0.08, 95% CI 0.05–0.13, p < 0.001) and perioperative fracture (OR = 0.18, 95% CI 0.04–0.76, p = 0.020) compared with those who received a conventional UKA. The risks of prolonged LOS, total hospital costs, and 30-day and 90-day readmissions were not significantly different between the two groups (Table 4).
Table 4.
Outcomes of robot-assisted versus conventional unicompartmental knee arthroplasty
| Outcomes | OR/betab,c (95% CI) | p-Value |
|---|---|---|
| Complications, any | 0.25 (0.19, 0.32) | <0.001 |
| AMI | 0.39 (0.07, 2.08) | 0.271 |
| CVA | 1.33 (0.60, 2.98) | 0.484 |
| VTE | 0.92 (0.29, 2.94) | 0.893 |
| Sepsis/shock | 0.56 (0.11, 2.85) | 0.481 |
| Pneumonia | 1.40 (0.39, 5.00) | 0.609 |
| UTI | 0.91 (0.44, 1.92) | 0.810 |
| Respiratory failure/mechanical ventilation | 0.56 (0.16, 1.96) | 0.359 |
| AKI | 1.21 (0.79, 1.85) | 0.374 |
| Implant malposition or failure | 0.08 (0.05, 0.13) | <0.001 |
| Nervous system complication | 5.21 (0.49, 55.83) | 0.172 |
| Perioperative fracture | 0.18 (0.04, 0.76) | 0.020 |
| Delirium | 1.19 (0.34, 4.20) | 0.785 |
| LOSa | −0.03 (−0.21, 0.15) | 0.753 |
| Total hospital costs, 1000 USD | 3.26 (−1.71, 8.23) | 0.198 |
| 30-day readmissiona | 0.90 (0.63, 1.27) | 0.538 |
| 90-day readmissiona | 0.82 (0.66, 1.03) | 0.082 |
aExcluding patients who died in the hospital
bBeta was used to compare LOS and total hospital costs
cAdjusted for the significant variables in Table 2, including coronary heart disease
p-Values < 0.05 are shown in bold
OR, odds ratio; CI, confidence interval; AMI, acute myocardial infarction; CVA, cerebrovascular accident; VTE, venous thromboembolism; UTI, urinary tract infection; AKI, acute kidney injury; LOS, length of stay
Discussion
In this study using a large, nationally representative database, the results showed several notable differences in short-term outcomes between patients undergoing robotic-assisted UKA versus conventional UKA. After PSM and adjustments in the multivariable analysis, robotic-assisted UKA was associated with a significantly lower risk of overall complications compared with the conventional approach. Specifically, robotic-assisted UKA was associated with significantly reduced rates of implant malposition or failure, and perioperative fractures. The odds of implant malposition or failure were 92% lower in the robotic-assisted group, and the risk of perioperative fracture was reduced by 82% in the robotic-assisted group. However, there were no significant differences between the two groups in terms of LOS, total hospital costs, or 30- and 90-day readmission rates.
Technology continually advances, and this is true for every medical field. For surgical specialties, the most recent technological advancement is robotic-assisted surgery, and advanced guidance systems. Robotic-assisted surgery is being explored in all surgical specialties, including spine surgery [14], TKA in an outpatient setting [15], colorectal cancer [16], urological cancers [17], and microsurgery and supermicrosurgery [18]. Overall, studies have suggested that the use of robotics results in generally improved surgical outcomes. UKA, also known as partial knee replacement, is a surgical procedure where only the damaged portion of the knee is replaced with an implant; the undamaged parts of the knee are left intact. The benefits of UKA include its minimally invasive nature and preservation of healthy cartilage, ligaments, and bone in the unaffected compartments of the knee. This is in contrast to TKA where the whole knee joint is replaced; a procedure that requires major postoperative rehabilitation. UKA, however, is very technically demanding, and outcomes are associated with the surgeon’s experience. Thus, it follows that robotic-assisted surgery that allows for very fine tissue dissection would be an advantage for performing UKA.
Our results showed that robotic-assisted UKA is associated with a lower risk of implant malposition and failure and perioperative fractures. Robotic-assisted procedures are known for their advantages in precision, and thus, less surgical trauma, which has been shown to translate to fewer complications in many smaller studies [19–22]. For example, Yeung et al. [19] reported that compared with conventional UKA, robotic-assisted UKA was associated with better component alignment and radiological accuracy, and comparable clinical outcomes. Another study of UKA reported that femoral component positioning within 2 mm from the joint line was attained in 64% of patients who underwent robotic-assisted UKA compared with only 18% who received conventional surgery [20]. The authors concluded that robotic assistance increases the precision of implant placement and also provides excellent native joint line restoration and balancing.
Foissey et al. [21] reported the long-term follow-up results of 356 knees treated with UKA: 159 underwent conventional UKA and 197 robotic-assisted UKA. At 9 years of follow-up, implant survival in the robotic-assisted group was 96% compared with 87% in the conventional group (p = 0.004). Wu et al. [22] also reported that compared with conventional UKA, robotic-assisted UKA was associated with higher accuracy in bone anatomical alignment; however, the surgical time was longer, and the blood loss was greater.
Although some findings of our study are consistent with those of previous studies, other results differ from those observed in our national sample. There are many confounding factors that can affect results, with one of the greatest being the surgeon’s experience with robotic-assisted surgery, a factor that was not available for analysis in our study. We found that robotic-assisted UKA was associated with significantly reduced rates of implant malposition or failure, and perioperative fractures. Other studies have reported that robotic-assisted UKA is associated with improved, immediate perioperative outcomes. Negrin et al. [8] reported that robotic-assisted UKA achieved more precision in radiographic parameters, and also was associated with a trend toward better Oxford Knee Score and lower numeric rating scale (NRS) pain score. Similarly, Wignadasan et al. [23] reported that robotic-assisted UKA was associated with improvements in Oxford Knee Score and range of motion; however, the study did not include a conventional UKA comparison group.
While our results did not show that robotic-assisted UKA was associated with a shorter LOS, other studies have indicated that robotic assistance reduces the LOS [24–26]. Cheung et al. [24] reported that robotic-assisted UKA was associated with a shorter LOS but stressed the importance of distinguishing between different robotic platforms. Another study also showed that robotic assistance was associated with a shorter length of stay in patients undergoing primary UKA, as well as patients undergoing primary TKA [25]. Shearman et al. [26] reported that robotic-assisted UKA was associated with earlier discharge from physiotherapy and reduced LOS compared with procedures performed with conventional navigated techniques.
While our results showed that robotic assistance did not affect 30- or 90-day readmission rates, most studies suggest that robotic assistance reduces the risk of revision and readmission in short-term or long-term follow-up [9, 27–29]. Two recent systematic reviews and meta-analyses examining different aspects of UKA outcomes concluded that robotic assistance is associated with improved outcomes. Avram et al. [27] reported that robotic-assisted UKA was associated with better forgotten Joint Score-12 (FJS-12) results compared with conventional UKA, as well as a lower all-cause revision rate at 36 months and a lower aseptic loosening rate at 60 months. The other meta-analysis by Bensa et al. [28] showed that robotic-assisted UKA was associated with a significant improvement in functional outcomes compared with conventional UKA; had similar radiological outcomes and longer operating time; and had reduced complication and revision rates. The 5-year clinical outcomes of a randomized controlled trial comparing robotic-assisted and conventional UKA showed that there were no differences in any of the patient reported outcomes, and a lower reintervention rate in the robotic-assisted group: 0% in the robotic group and 9% in the conventional group (p < 0.001) [9]. A very recently published study reported that compared with conventional UKA, robotic-assisted UKA was associated with a lower revision rate and a lower aseptic loosening rate at 2-year follow-up but a higher total cost for the initial procedure [29].
Surgeon experience and the learning curve are the key determinants of UKA outcomes, given the procedure’s sensitivity to millimetric bone resections, alignment, and soft-tissue balancing. Although robotic guidance can enhance the reproducibility of component positioning, inaccuracies in registration or intraoperative decision-making may still occur and are operator-dependent. Because the NRD does not capture surgeon case or experience, residual confounding related to difference of experience cannot be excluded. This limitation should be considered when interpreting the lower risks of implant malposition/failure and perioperative fractures observed in the robotic group.
Cost is always a concern with medical procedures, and this is especially true for procedures that rely on very advanced technologies. While some studies suggest robotic-assisted surgery is associated with a lower cost [30, 31], others indicate it has a higher cost compared with conventional surgery [29]. This is an important area for future research, as the definition of “cost” may apply to only the index surgery, or included complications seen during follow-up (which may be reduced using robotic assistance). Moreover, as the adoption of new technologies increases, their cost decreases.
Our findings should also be interpreted in light of potential coding biases inherent to administrative databases. For example, ICD-10 code T84 encompasses a spectrum of mechanical complications, and coding practices may vary across hospitals, which could inflate or misclassify the incidence of implant malposition or failure. In addition, the NRD does not capture specific robotic platforms, such as MAKO versus other systems, which may differ in precision, adoption, and learning curves. These unmeasured confounders could partly account for the observed differences and highlight the need for prospective studies with more granular operative data.
Strengths and limitations
The strength of using national hospitalization and readmission data is the high external validity of the results. We directly compared complications, LOS, total hospital costs, and readmission rates between the two surgical methods. The use of PSM improved the precision of comparisons by minimizing baseline differences between the groups. Furthermore, we analyzed a broad range of short-term clinical outcomes, including specific complications such as implant malposition and postoperative fractures.
However, this study has several limitations. First, as a retrospective observational study, the ability to infer causality is limited, and the findings may be subject to residual confounding from unmeasured variables. Second, the analysis was restricted to outcomes within 90 days postoperatively and did not capture mid- to long-term outcomes, such as joint function durability, revision rates, and implant survival, which are critical for UKA evaluation but could not be assessed using the NRD. In addition, a limitation of our analysis is that ICD-10 code T84, as implemented in our dataset, did not allow reliable separation of malposition from mechanical failure. Future studies with more granular implant-specific adjudication should analyze these outcomes separately to better delineate mechanisms.
Third, the reliance on administrative data and ICD coding may introduce inaccuracies in diagnosis and outcome classification. Moreover, hospitalization costs were estimated on the basis of medical charges, which may not reflect the actual resource utilization. A more accurate approach would be to use the Healthcare Cost and Utilization Project (HCUP) Cost-to-Charge Ratio (CCR), which allows conversion of charges into more accurate estimated costs. Another limitation is that the NRD does not provide hospital- or surgeon-level case volumes, which may influence perioperative outcomes and cost efficiency. The brand- or platform-specific details of robotic systems (e.g., MAKO versus others) and facility identifiers are deidentified in NRD, constraining external linkage and limiting hospital-level inferences to distributional counts.
Finally, although PSM was used to adjust for potential confounders, residual confounding cannot be excluded, as unmeasured variables were not captured. In particular, surgeon experience, hospital case volume, the specific robotic platform, and whether the UKA was medial or lateral were unavailable in the NRD and could have influenced outcomes, especially implant malposition or failure. Moreover, the NRD does not include patient-reported outcome measures (PROMs), limiting assessment of functional recovery. We also did not conduct formal sensitivity analyses, such as excluding certain complications or stratifying by study years, which could further test the robustness of our findings. The low incidence of some complications also limits the statistical power to detect significant differences.
In summary, while our findings indicate that robotic-assisted UKA may offer advantages in reducing certain perioperative complications, the inherent limitations of the study design and data source warrant cautious interpretation. Prospective studies with longer follow-up are needed to confirm these results and better understand the long-term benefits and cost-effectiveness of robotic-assisted UKA.
Conclusions
In this nationally representative analysis, robotic-assisted UKA was associated with significant reductions in specific surgical complications, including implant malposition, failure, and perioperative fractures. However, its use did not confer advantages in broader hospital metrics, such as length of stay, total costs, or readmission rates. These findings highlight the short-term safety benefits of robotic UKA and underscore the need for further research to fully assess its long-term clinical and economic value.
Supplementary Information
Acknowledgments
The authors gratefully acknowledge the assistance of the Department of Computer Science, Tunghai University, for their support in data analysis during this study.
Abbreviations
- AKI
Acute kidney injury
- AMI
Acute myocardial infarction
- CCI
Charlson Comorbidity Index
- CI
Confidence interval
- CKD
Chronic kidney disease
- ICD
International Classification of Diseases
- LOS
Length of hospital stay
- NRD
US Nationwide Readmissions Database
- OA
Osteoarthritis
- OR
Odds ratio
- PSM
Propensity score matching
- TKA
Total knee arthroplasty
- UKA
Unicompartmental knee arthroplasty
- VTE
Venous thromboembolism
Author contributions
Conceptualization: Cheng-Min Shih; data curation: Kun-Hui Chen; formal analysis: Kun-Hui Chen; funding acquisition: Cheng-Min Shih; investigation: Shun-Ping Wang and Cheng-Min Shih; methodology: Shun-Ping Wang; project administration: Cheng-Chi Wang; resources: Cheng-Min Shih; software: Fuu-Cheng Jiang; supervision: Fuu-Cheng Jiang; validation: Cheng-En Hsu; visualization: Cheng-En Hsu; writing—original draft: Cheng-Min Shih; writing—review and editing: Cheng-Min Shih.
Funding
This research was funded by Taichung Veterans General Hospital, Taiwan (TCVGH-1141703B).
Availability of data and materials
All data generated or analyzed during this study are included in this published article.
Declarations
Ethics approval and consent to participate
This study used precollected, anonymized data, where no patients were directly involved. The lack of personally identifiable information aligns with ethical research standards. The study protocol was submitted to our institution’s institutional review board, which confirmed that no additional approval was necessary.
Consent for publication
Not applicable.
Competing interests
The authors declare no conflicts of interest.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Cui A, Li H, Wang D, Zhong J, Chen Y, Lu H (2020) Global, regional prevalence, incidence and risk factors of knee osteoarthritis in population-based studies. EClinicalMedicine 29–30:100587 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Leifer VP, Katz JN, Losina E (2022) The burden of OA-health services and economics. Osteoarthritis Cartilage 30(1):10–16 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Allen KD, Thoma LM, Golightly YM (2022) Epidemiology of osteoarthritis. Osteoarthrit Cartil 30(2):184–195 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hawker GA, King LK (2022) The burden of osteoarthritis in older adults. Clin Geriatr Med 38(2):181–192 [DOI] [PubMed] [Google Scholar]
- 5.Crawford DA, Berend KR, Thienpont E (2020) Unicompartmental knee arthroplasty: US and global perspectives. Orthop Clin North Am 51(2):147–159 [DOI] [PubMed] [Google Scholar]
- 6.Wilson HA, Middleton R, Abram SGF, Smith S, Alvand A, Jackson WF et al (2019) Patient relevant outcomes of unicompartmental versus total knee replacement: systematic review and meta-analysis. BMJ (Clinical research ed) 364:l352 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chin BZ, Tan SSH, Chua KCX, Budiono GR, Syn NL, O’Neill GK (2021) Robot-assisted versus conventional total and unicompartmental knee arthroplasty: a meta-analysis of radiological and functional outcomes. J Knee Surg 34(10):1064–1075 [DOI] [PubMed] [Google Scholar]
- 8.Negrín R, Duboy J, Iñiguez M, Reyes NO, Barahona M, Ferrer G et al (2021) Robotic-assisted vs conventional surgery in medial unicompartmental knee arthroplasty: a clinical and radiological study. Knee Surg Relat Res 33(1):5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Banger M, Doonan J, Rowe P, Jones B, MacLean A, Blyth MJB (2021) Robotic arm-assisted versus conventional medial unicompartmental knee arthroplasty: five-year clinical outcomes of a randomized controlled trial. Bone Joint J 103-b(6):1088–1095 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zhang P, Xu K, Zhang J, Chen P, Fang Y, Wang J (2021) Comparison of robotic-assisted versus conventional unicompartmental knee arthroplasty for the treatment of single compartment knee osteoarthritis: a meta-analysis. Int J Med Robot Comput Assist Surg 17(1):1–11 [DOI] [PubMed] [Google Scholar]
- 11.Ghazal AH, Fozo ZA, Matar SG, Kamal I, Gamal MH, Ragab KM (2023) Robotic versus conventional unicompartmental knee surgery: a comprehensive systematic review and meta-analysis. Cureus 15(10):e46681 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gilmour A, MacLean AD, Rowe PJ, Banger MS, Donnelly I, Jones BG et al (2018) Robotic-arm-assisted vs conventional unicompartmental knee arthroplasty. The 2-year clinical outcomes of a randomized controlled trial. J Arthroplasty 33(7s):S109–S115 [DOI] [PubMed] [Google Scholar]
- 13.Parsons LS, ed. Performing a 1: N case-control match on propensity score. Proceedings of the 29th Annual SAS users group international conference; 2004: SAS Institute Montreal, Canada.
- 14.Davidar AD, Jiang K, Weber-Levine C, Bhimreddy M, Theodore N (2024) Advancements in robotic-assisted spine surgery. Neurosurg Clin N Am 35(2):263–272 [DOI] [PubMed] [Google Scholar]
- 15.Eason T, Mihalko W, Toy PC (2023) Robotic-assisted total knee arthroplasty is safe in the ambulatory surgery center setting. Orthop Clin North Am 54(2):153–159 [DOI] [PubMed] [Google Scholar]
- 16.Yao Q, Sun QN, Ren J, Wang LH, Wang DR (2023) Comparison of robotic-assisted versus conventional laparoscopic surgery for mid-low rectal cancer: a systematic review and meta-analysis. J Cancer Res Clin Oncol 149(16):15207–15217 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Bignante G, Orsini A, Lasorsa F, Lambertini L, Pacini M, Amparore D et al (2024) Robotic-assisted surgery for the treatment of urologic cancers: recent advances. Expert Rev Med Devices 21(12):1165–1177 [DOI] [PubMed] [Google Scholar]
- 18.Burns HR, McLennan A, Xue EY, Yu JZ, Selber JC (2023) Robotics in microsurgery and supermicrosurgery. Semin Plast Surg 37(3):206–216 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Yeung MHY, Fu H, Cheung A, Kwan VCW, Cheung MH, Chan PK et al (2023) Robotic arm-assisted unicondylar knee arthroplasty resulted in superior radiological accuracy: a propensity score-matched analysis. Arthroplasty 5(1):55 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ashok Kumar PS, Pawar S, Kanniyan K, Pichai S, Bose VC, Patil S (2024) Does robotic-assisted unicompartmental knee arthroplasty restore native joint line more accurately than with conventional instruments? J Robot Surg 18(1):49 [DOI] [PubMed] [Google Scholar]
- 21.Foissey C, Batailler C, Vahabi A, Fontalis A, Servien E, Lustig S (2023) Better accuracy and implant survival in medial imageless robotic-assisted unicompartmental knee arthroplasty compared to conventional unicompartmental knee arthroplasty: two- to eleven-year follow-up of three hundred fifty-six consecutive knees. Int Orthop 47(2):533–541 [DOI] [PubMed] [Google Scholar]
- 22.Wu C, Fukui N, Lin YK, Lee CY, Chou SH, Huang TJ et al (2021) Comparison of robotic and conventional unicompartmental knee arthroplasty outcomes in patients with osteoarthritis: a retrospective cohort study. J Clin Med. 10.3390/jcm11010220 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wignadasan W, Chang J, Fontalis A, Plastow R, Haddad FS (2023) Short term outcomes following robotic arm-assisted lateral unicompartmental knee arthroplasty. Front Surg 10:1215280 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Cheung KSC, Chan KCA, Cheung A, Chan PK, Luk MH, Chiu KY et al (2025) Current trends of unicompartmental knee arthroplasty (UKA): choosing between robotic-assisted and conventional surgeries and timing of procedures. Arthroplasty (London, England) 7(1):6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Fontalis A, Raj RD, Haddad IC, Donovan C, Plastow R, Oussedik S et al (2023) Length of stay and discharge dispositions following robotic arm-assisted total knee arthroplasty and unicompartmental knee arthroplasty versus conventional technique and predictors of delayed discharge. Bone Joint Open 4(10):791–800 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Shearman AD, Sephton BM, Wilson J, Nathwani DK (2021) Robotic-assisted unicompartmental knee arthroplasty is associated with earlier discharge from physiotherapy and reduced length-of-stay compared to conventional navigated techniques. Arch Orthop Trauma Surg 141(12):2147–2153 [DOI] [PubMed] [Google Scholar]
- 27.Avram GM, Tomescu H, Dennis C, Rusu V, Mengis N, Ammann E et al (2024) Robotic-assisted medial unicompartmental knee arthroplasty provides better FJS-12 score and lower mid-term complication rates compared to conventional implantation: a systematic review and meta-analysis. J Pers Med. 10.3390/jpm14121137 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bensa A, Sangiorgio A, Deabate L, Illuminati A, Pompa B, Filardo G (2024) Robotic-assisted unicompartmental knee arthroplasty improves functional outcomes, complications, and revisions. Bone Joint Open 5(5):374–384 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Guild G, Schwab J, Ross BJ, McConnell MJ, Najafi F, Bradbury TL (2025) Is robotic-assisted unicompartmental knee arthroplasty compared to manual unicompartmental knee arthroplasty associated with decreased revision rates? An updated matched cohort analysis. Arthroplasty Today 32:101652 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Blyth MJG, Clement ND, Choo XY, Doonan J, MacLean A, Jones BG (2025) Robotic arm-assisted medial compartment knee arthroplasty is a cost-effective intervention at ten-year follow-up. Bone Joint J 107-b(1):72–80 [DOI] [PubMed] [Google Scholar]
- 31.Goh GS, Haffar A, Tarabichi S, Courtney PM, Krueger CA, Lonner JH (2022) Robotic-assisted versus manual unicompartmental knee arthroplasty: a time-driven activity-based cost analysis. J Arthroplasty 37(6):1023–1028 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
All data generated or analyzed during this study are included in this published article.