Survival and Cost Analysis of Surgical Mitral Valve Replacement with Different Prostheses: A Nationwide Cohort Study in Taiwan

Introduction

Mitral valve disease, including mitral regurgitation and mitral stenosis, affects up to 2% of the general population and contributes significantly to cardiovascular morbidity and mortality [1]. Current guidelines recommend surgical intervention for severe primary mitral regurgitation in symptomatic patients who are operable, and in asymptomatic patients with left ventricular dysfunction, secondary atrial fibrillation, or pulmonary hypertension [2]. For chronic secondary mitral regurgitation, surgery is typically reserved for patients who remain symptomatic despite optimal medical therapy and who are undergoing concurrent cardiac surgery [2]. In cases of clinically significant mitral stenosis, particularly in rheumatic heart disease, surgical valve replacement has been shown to be one of the most effective treatments for symptomatic patients [2,3].

For patients requiring mitral valve replacement, both mechanical and bioprosthetic valves (porcine and bovine) are viable options [4]. Mechanical valves require lifelong anticoagulation to prevent thromboembolism, which increases the risk of bleeding [5]. Bioprosthetic valves offer the advantage of avoiding long-term anticoagulation but have limited durability due to structural valve degeneration, particularly in younger patients [6–8].

A 2022 metanalysis comparing 35 903 patients undergoing mitral valve replacement found that the mechanical valve group exhibited a 16% lower risk of long-term mortality, a 66% lower risk of mitral re-operations, and a 21% greater risk of major bleeding compared to the bioprosthetic valve group [9]. However, newer research suggested that long-term survival (up to 15 years) is similar between mechanical and bioprosthetic valves in patients aged 50 to 70, although bioprosthetic valves carry a higher risk of repeat mitral valve replacement, challenging the conventional preference for mechanical valves in younger patients [10]. Consequently, valve selection should be individualized based on patient-specific risk factors and preferences [2,4].

Recent advancements in bioprosthetic valve technology, including calcium-blocking tissue treatments, flexible stent designs to absorb cardiac cycle stress, and bioengineered construction, have led to the development of durable bioprosthetic valves, which can extend valve longevity [11–14]. In Taiwan, such valves were since categorized as durable, under the same National Health Insurance (NHI) medical device code (FHVD1), apart from the older-generation bioprosthesis (coded as FHV02) [15], and were partially reimbursed, with the remaining cost borne by the patient. Despite their growing use, real-world data comparing clinical outcomes and costs between mechanical valves (MV), older porcine valves (PV), and newer durable valves (DV) remain limited.

This study used a large, nationwide administrative database to evaluate survival, hospitalization, and cost outcomes associated with MV, PV, and DV in patients undergoing surgical mitral valve replacement in Taiwan.

Material and Methods

RESEARCH ETHICS APPROVAL:

The study protocol was approved by the MacKay Memorial Hospital Institutional Review Board Taiwan R.O.C. (Approval Number: 19MMHIS083e). Given the study’s retrospective nature, the requirement for informed consent was waived. Data from NHIRD were provided in encrypted form, with all personal identification removed.

Results

SAMPLE DESCRIPTION:

A total of 10 406 patients who underwent surgical mitral valve replacement (SMVR) were included in the study: 5301 (50.9%) received mechanical valves (MV), 4300 (41.3%) received porcine bioprosthetic valves (PV), and 805 (7.7%) received durable valves (DV). Baseline characteristics are summarized in Table 2. The MV group had the youngest mean age (53.0±13.6 years), followed by DV (61.4±12.1) and PV (66.2±12.6) (P<0.01). The PV group had higher proportions of comorbidities such as diabetes, cerebrovascular disease, renal disease, and cancer (all P<0.01), while the MV group had the lowest comorbidity burden. After propensity-score matching (PSM), we identified 678 matched pairs between MV and DV, and 711 pairs between PV and DV. Baseline characteristics were well balanced after matching.

SURVIVAL ANALYSIS:

Kaplan-Meier survival analysis revealed that the PV group had the highest cumulative all-cause mortality (log-rank P<0.01). Figure 2A shows this trend in the overall cohort. When stratified by age, DV consistently showed lower mortality compared to PV in all age subgroups: ≤50, 51–70, and ≥71 years (Figure 2B–2D; all P<0.01). Among patients ≤70 years, MV also demonstrated a survival advantage over PV (p<0.01), but this difference was smaller in the ≥71 age group.

In the matched cohorts (Figure 3), DV maintained a survival advantage over PV and MV. In DV vs PV pairs, the durable valve group showed significantly better 5-year survival (hazard ratio [HR]: 0.61; 95% CI: 0.50–0.74; P<0.001). Similarly, in DV vs MV matched patients, DV was associated with reduced mortality risk (HR: 0.72; 95% CI: 0.60–0.88; P=0.002).

MULTIVARIATE REGRESSION ANALYSIS FOR PATIENTS RECEIVING SMVR:

Multivariable Cox regression analysis (Table 3) revealed that DV use was independently associated with lower mortality (HR: 0.63; 95% CI: 0.52–0.76), as was female sex (HR: 0.82; 95% CI: 0.77–0.87). Factors linked to increased mortality included PV use (HR: 1.21; 95% CI: 1.13–1.29), older age (per year increase: HR: 1.04; 95% CI: 1.037–1.042), renal disease (HR: 2.00; 95% CI: 1.78–2.25), diabetes with complications (HR: 1.51; 95% CI: 1.28–1.78), myocardial infarction (HR: 1.55; 95% CI: 1.37–1.75), and moderate/severe liver disease (HR: 2.01; 95% CI: 1.24–3.26).

SECONDARY OUTCOMES:

Average costs based on NHI reimbursement records of the index hospitalization are presented in Figure 4. The PV group had the highest average index hospitalization cost (USD 17 796.2±1720.4), significantly greater than both the MV group (USD 15 192.7±1294.4) and the DV group (USD 17 487.0±1058.4; P<0.01). Similarly, length of hospital stay was longest in the PV group (26.7±1.1 days), followed by DV (23.8±1.6 days) and MV (22.5±1.1 days; P<0.01). These cost and length-of-stay differences persisted after PSM.

We also examined re-operation rates within 3 years of valve replacement, acknowledging the limited follow-up for the DV group. During this period, the DV group had the lowest observed re-operation rate (0.75%), compared to MV (1.96%) and PV (3.21%), with significantly higher odds of re-operation in the MV group (OR: 2.66, 95% CI: 1.17–6.09; P=0.0201) and the PV group (OR: 4.42, 95% CI: 1.94–10.03; p=0.0004) relative to DV. These findings suggest a lower short-term re-operation rate for DVs; however, longer-term data are needed to determine whether this advantage persists.

Discussion

This nationwide, population-based cohort study provides real-world evidence on the comparative outcomes and costs of surgical mitral valve replacement (SMVR) using mechanical valves (MV), older-generation porcine bioprostheses (PV), and newer durable bioprostheses (DV) in Taiwan. Our findings indicate that DV was associated with the best early survival outcomes and a moderate cost profile, while PV demonstrated the highest mortality and hospitalization burden.

The American Heart Association recommended a mechanical valve for patients under 65 years old who had no contraindication for anticoagulation, while advising a bioprosthesis for those aged 65 and older [4]. In our study cohort, we found that over 90% of the patients in the MV group were aged ≤70 years, while 75% of the patients in the PV group were older than 60 years old, indicating that the selection of valve for SMVR in Taiwan was aligned with current guidelines [2,4].

Previous research comparing the clinical outcomes of SMVR with different prosthetic valves had inconsistent conclusions. A retrospective cohort study of 3433 patients in New York City suggested that among the non-elderly patients (aged 50 to 69 years) undergoing mitral valve replacement between 1997 and 2007, there was no significant survival difference between those receiving mechanical and bioprosthetic valves at 15 years in patients matched by propensity score [7]. Another analysis comparing outcomes of tissue versus mechanical valve replacement in patients aged 50 to 70 between 2004 and 2018 also showed a similar survival up to 15 years of follow-up [10]. In our cohort, patients in the PV group had higher all-cause mortality, especially in the ≤50 and 51–70 age groups, than those who received mechanical valves. Our findings were consistent with several other retrospective analyses involving patients younger than 65 years who underwent SMVR, which showed that bioprostheses were associated with reduced survival and higher re-operation rates [6,8,17]. It remains unclear whether the lower long-term survival is solely attributable to the type of valve. We observed significantly more comorbidities in the group that received porcine valves (Table 1), as in a recently published propensity-score matched analysis [10] in which the difference in long-term all-cause mortality between tissue and mechanical valves became insignificant after PSM. They therefore suggested that survival benefit associated with mechanical valves in non-elderly people might be due to bioprostheses being implanted in sicker patients.

In addition, most previous studies evaluating the outcomes of bioprosthesis were conducted before the newer durable valves became widely available, or without addressing the information regarding to which tissue valve was used [6–8,17]. Although manufacturers published data showing better long-term durability for the newer tissue valves treated with newer anti-calcification technology and designed with better hemodynamic profiles [18,19], real-world studies comparing the currently available durable valves with older-generation porcine and mechanical valves are lacking. The present report could give us an insight on the subject, using the nationwide database to directly compare the overall survival for patients receiving MVs, PVs and DVs. The all-cause mortality was lowest in patients receiving DVs in all age groups (≤50, 51–70 and ≥71 years) (Figure 2), indicating DV could be superior to PV or MV. However, the observed superiority warrants cautious interpretation. The follow-up period for the DV group was shorter, given their introduction into the Taiwanese market in 2014, which limits our ability to evaluate long-term durability and outcomes. Moreover, DV recipients had fewer comorbidities and underwent surgery in more recent years, potentially benefiting from improved perioperative care and surgical techniques. While propensity-score matching helped balance key clinical variables, we were unable to adjust for the year of surgery due to the limited DV sample size and study design constraints.

Pressure on health care providers to reduce health-related expenses has increased over time. There were limited data available regarding to the cost and hospital stay of mitral valve surgery, and most of them did not separate the different types of bioprosthetic valves, did not calculate in-hospital cost year by year, or were not performed from the perspective of an Asian country [20–22]. Our study demonstrated the cost of the index hospitalization for patients receiving SMVR with MVs, PVs, and DVs using the national data in Taiwan, which enrolled patients with all surgical risks. In the United States, Ferket et al reported their in-hospital costs associated with SMVR was 78 216 USD [22]. In Europe, Trochu et al reported the total in-hospital cost associated with SMVR per patient was 29 732 EUR (approximately 35 678 USD; exchange rate=1.2: 1) [21]. The costs in Taiwan were much lower, although the cost has also increased over time (Figure 4). The length of hospital stay in Taiwan was longer than in France (22.5–26.7 days from 2000 to 2017 compared with 17.0±14.7 days) [21]. The differences in length of hospital stay could be ascribed to the lower co-payment system of Taiwan. Interestingly, the length of hospital stay for SMVR in China (19.5±9.1 days) was similar to that in Taiwan [23]. The reason for this requires further investigation.

In the present study, we also examined re-operation rates within 3 years of valve replacement, given the limited follow-up for the DV group. The DV group showed the lowest short-term re-operation rate compared to MV and PV. This may reflect improvements in valve design and tissue treatment, healthier patient profiles, and later surgical era. In contrast, higher early re-operation rates in the PV and MV groups may have been influenced by the older patient populations, greater comorbidities, and earlier surgical techniques, and, for mechanical valves, complications from anticoagulation. Importantly, as structural valve degeneration typically occurs beyond 5 years, the lower re-operation rate observed for DVs likely reflects short-term advantages rather than definitive evidence of superior long-term durability. These findings should therefore be interpreted cautiously and confirmed in studies with extended follow-up.

Our study has several important limitations. First, as a retrospective observational study without randomization, measured and unmeasured confounders may have influenced the results despite propensity-score matching. Uncaptured factors such as frailty, socioeconomic status, or echocardiographic severity may have contributed to differences between groups. Second, patients receiving durable valves underwent surgery exclusively during the later study years (2014–2017), potentially benefiting from advances in perioperative care and surgical techniques (era effect). This temporal disparity, combined with the smaller size of the DV group, limits the generalizability of our findings. Third, the durable valve group included heterogeneous models from different manufacturers (Table 1), each with unique designs and tissue treatments, so the observed outcomes may not apply equally to all models. Fourth, the short follow-up for the DV group precludes assessment of long-term durability, re-operation rates beyond 3 years, and sustained survival benefits. Finally, the NHIRD lacks detailed clinical data (eg, echocardiographic findings, laboratory values, and disease severity), and our cost analysis was confined to in-hospital expenses, excluding patient copayments and self-paid medications, potentially underestimating the total economic burden.

Despite these limitations, our study offers valuable insights into the evolving landscape of mitral valve replacement in a public healthcare system. The higher hospitalization costs and longer stays associated with PV may reflect both increased complication rates and poorer baseline health in this group. These findings support the need for further comparative research and may guide policymakers considering reimbursement strategies and device approvals for SMVR.

Conclusions

In this nationwide, population-based analysis, the use of newer durable bioprosthetic valves in surgical mitral valve replacement was associated with favorable early survival outcomes, lower short-term re-operation rates, and moderate hospitalization costs compared with mechanical and porcine bioprostheses. However, given the retrospective design, potential confounding factors, heterogeneity of valve models, and limited follow-up of the durable valve group, these findings should be interpreted with caution. Longer-term, prospective studies are warranted to assess the durability, re-operation risk, and long-term cost-effectiveness of newer bioprosthetic technologies.