This is a nationwide population-based research to systematically evaluate the mid-term outcomes of robotic-assisted mitral valve replacement in real-world setting. In this study, we demonstrated that RAMVR was associated with a significant survival benefit compared to CSMVR. Additionally, patients undergoing RAMVR experienced a significantly shorter hospital stay and ICU stay, which may contribute to improved postoperative recovery and reduced healthcare resource utilization.

Compared to mitral valve repair, RAMVR is considered more technically challenging due to the small surgical ports, which can complicate prosthesis implantation and suturing, leading to significantly longer aortic clamp and cardiopulmonary bypass durations [11, 12]. As a result, most studies have focused on robotic-assisted mitral valve repair rather than replacement, and the limited research available on RAMVR has primarily been conducted in highly specialized centers, where surgeries were performed by dedicated teams of experienced surgeons, anesthesiologists, and nurses, and comprehensive surgical details were kept [8, 11,12,13,14]. The three studies that compared the clinical outcomes of RAMVR and CSMVR reported similar rates of all-cause mortality and perioperative complications [8, 12, 14].

In contrast, our study, which included a broader patient population across multiple hospitals over six years, demonstrated superior survival rates for the RAMVR group in both unadjusted analyses and after adjustment using the inverse probability of treatment weighting method. Several factors may explain this discrepancy. First, most previous studies were conducted in high-volume institutions with strong surgical teams for both RAMVR and CSMVR, minimizing performance variability. In our nationwide cohort, RAMVR was almost exclusively performed in medical centers, while CSMVR was more frequently carried out in regional or district hospitals. This may reflect differences in institutional expertise and perioperative care, which could have influenced patient outcomes despite statistical adjustment.

Second, some confounding factors might not be fully captured by the Charlson comorbidity score, such as critical preoperative status, active endocarditis, significantly reduced left ventricular ejection fraction, and pulmonary hypertension. To further minimize selection bias, incorporating a well-validated risk assessment tool specifically designed for cardiac surgeries, such as the European System for Cardiac Operative Risk Evaluation II (EuroSCORE II) or Society of Thoracic Surgeons (STS) Score, would be more ideal [15,16,17]. Third, patient selection for RAMVR in Taiwan may inherently favor higher-performing institutions and better-coordinated multidisciplinary care, even though baseline clinical characteristics were similar after IPTW adjustment. Fourth, the survival benefit may be more apparent in our cohort because we focused specifically on mitral valve replacement rather than repair. MVR is typically reserved for patients with more advanced disease or unsuitable anatomy for repair, and the benefits of a less invasive approach—such as reduced surgical trauma, shorter ICU stays, and fewer complications—may translate more clearly into survival advantages in this higher-risk population. Fourth, prior studies often had limited follow-up periods and small RAMVR sample sizes, making them less likely to detect survival differences over time. Our study included follow-up up to 5–6 years, allowing the survival curves to separate more distinctly in the mid-term. Finally, advances in robotic technology, increased team experience, and improved perioperative management in recent years may have contributed to better outcomes in RAMVR patients compared to those reported in earlier studies.

Despite its potential benefits, RAMVR remained infrequent in our national database, accounting for only 2% of all mitral valve replacements during the study period. Several factors may have accounted for this low adoption rate. First, the high material and infrastructure costs for both hospitals and patients result in highly selective patient eligibility for robotic-assisted surgery. Second, the significantly longer operative time may reduce cost-effectiveness in operating room utilization. Third, RAMVR requires extensive training for surgeons to achieve consistent outcomes. A recent report from Germany found that material and infrastructure costs of RAMVR were significantly higher than the national benchmark for non-robotic surgeries by €1,384 vs. €985, respectively (p < 0.01), but the overall cost was not significantly higher, indicating the high cost of RAMVR could be partially offset by a shorter hospital stay [13]. Regarding operative time, studies from high-volume centers have shown that RAMVR typically took around 240 to 320 min to perform, which was about one to two hours longer than CSMVR [11,12,13]. However, with six months of structured team training or experience exceeding 50 cases, the operative time could be reduced to approximately 200 min [12, 13]. It is reasonable to deduce that cost-effectiveness of RAMVR could also be improved over time in high-volume centers [18, 19].

While our study demonstrates the advantages of RAMVR, it's important to contextualize these findings against other minimally invasive surgical approaches. Recent trials reported that non-robotic minimally invasive mitral valve procedures offers comparable safety and short-term functional outcomes to sternotomy, with benefits in cosmetic results [20,21,22]. However, robotic systems provide enhanced three-dimensional visualization, superior instrument dexterity, and tremor filtration—features that may improve precision in complex prosthesis implantation and suture management [23, 24]. Our findings reinforce these benefits of robotic-assisted surgery, and when performed in experienced centers, it may offer additional values over other minimally invasive surgical strategies [25].

In parallel, transcatheter-based therapies such as transcatheter mitral valve replacement (TMVR) and transcatheter edge-to-edge repair (TEER) are increasingly adopted, particularly in patients deemed high-risk or inoperable [26]. These catheter-based interventions provide a less invasive option without the need for cardiopulmonary bypass or general anesthesia. However, current limitations include anatomical constraints, higher procedural costs, and less long-term durability data compared to surgical replacement. Moreover, TMVR is generally reserved for select patients with suitable anatomy or failed prior bioprostheses [27, 28], while TEER may be less effective in cases of severe leaflet pathology or extensive calcification [29, 30]. Compared to these interventional therapies, RAMVR remains a more definitive and anatomically versatile option, particularly for patients who are suitable surgical candidates. As technology continues to evolve, head-to-head comparisons of RAMVR and transcatheter approaches in specific patient subgroups will be necessary to define optimal treatment strategies.

Our study has several limitations. First, although the use of data from the NHIRD provides the advantage of a large, nationally representative cohort, it lacks important clinical granularity. Key variables such as baseline left ventricular ejection fraction, presence and severity of pulmonary hypertension, and New York Heart Association functional class—factors known to influence surgical outcomes—were not available. Additionally, operative details including cardiopulmonary bypass duration, aortic cross-clamp time, and transfusion volume were not captured. These missing clinical variables may introduce residual confounding that could not be fully adjusted for using administrative data alone. Second, reliance on ICD-10 diagnosis and procedural codes may result in misclassification bias, coding errors, or inter-institutional variability that could affect data accuracy. Third, the maximum follow-up duration for patients who underwent RAMVR was five to six years, which may not be sufficient to assess long-term valve durability or survival outcomes comprehensively. Fourth, our cost analysis reflects only the portion reimbursed by Taiwan’s NHI and does not account for out-of-pocket payments related to robotic instruments, system usage, and advanced prosthetic valves. Based on hospital-reported estimates, the total patient-borne cost may range from approximately USD 6,000 to 9,000, depending on the institution and surgical setup. While these estimates are anecdotal and not standardized, they underscore the substantial expenses not captured in the claims database. As such, the apparent cost advantage of RAMVR observed in our study may not be generalizable to healthcare systems with different reimbursement models. Fifth, several of the complication outcomes in this study, such as stroke, dialysis, and re-operation, had very low or zero event counts in the RAMVR group. This limited number of events inherently reduces the statistical power to detect between-group differences, particularly for rare outcomes. Although IPTW was used to adjust for baseline covariates, the small sample size of the RAMVR group may still lead to unstable estimates or wide confidence intervals. Consequently, the findings related to these rare complications should be interpreted with caution. Future studies with larger cohorts or pooled datasets are warranted to more robustly assess these clinical endpoints. Finally, temporal improvements in surgical technique, perioperative management, and overall healthcare quality during the study period may have contributed to observed differences in outcomes, independent of surgical approach.

Future prospective studies with detailed clinical data—including echocardiographic parameters, operative variables, and validated surgical risk scores such as EuroSCORE II or STS risk models —are warranted to validate our findings and more precisely assess the comparative effectiveness of robotic-assisted versus conventional mitral valve replacement.