This single-center, retrospective cohort study conducted at Zhongshan Hospital, Xiamen University, which obtained ethics approval (2024–113) and adhered to the Declaration of Helsinki, enrolled patients with VHD undergoing elective valve surgery within ERAS protocol.

The selected cases obtained verbal informed consent during the follow-up calls. Data were extracted from the institutional Hospital Information System-electronic medical record (HIS-EMR). Inclusion criteria were age 18–75 years, New York Heart Association (NYHA) class I–III, absence of significant extra-cardiac organ dysfunction or hepatorenal disease history, and no conditions like pulmonary edema that could interfere with the intervention. Exclusion criteria included NYHA class IV, major organ dysfunction, age ≥75 years, or significant pulmonary/limb edema. After screening, 48 eligible patients were included in this retrospective analysis. The sample size was determined by the available cohort during the study period. The assignment to the quadruple therapy (QT, n = 34) or triple therapy (TT, n = 14) group was determined by the treating physician based on prevailing heart failure guidelines and individual patient clinical status at admission. Patients in the QT group received a SGLT2 inhibitor (e.g., dapagliflozin 10 mg/day or empagliflozin 10 mg/day) as part of their regimen. Medication administration followed institutional protocols, typically continuing until the day before surgery, continue to take the medication when regaining the ability to swallow after anesthesia recovery.

It is important to note that this was not a randomized allocation. Consequently, significant baseline differences were observed, such as lower diastolic and systolic blood pressure in the QT group, which may reflect differences in underlying cardiac function and represent potential confounding factors for postoperative recovery outcomes. All patients were managed according to the institutional ERAS path-way, and baseline demographics, clinical characteristics, VHD classifications, and surgical types were recorded (Supplementary Table S1). The primary outcome was mobility, assessed using the Barthel Index for Activities of Daily Living (ADL). Secondary outcomes included the Braden scale for stress ulcer risk and the Modified Thomas scale for fall risk. These outcomes were measured at five timepoints: hospital admission, post-anesthesia care unit (PACU) discharge, ICU discharge, transition to secondary care, and final hospital discharge.

The study implemented a three-phase ERAS protocol for elective valve surgery (15, 16). The protocol included preoperative respiratory and walking exercises, intraoperative goal-directed fluid therapy and myocardial protection, and postoperative early mobilization and analgesia. Patient mobility and recovery metrics were assessed perioperatively using standardized scales.(Barthel Index for activities of daily living (17), the Braden scale for stress ulcer risk (18), and the Modified Thomas Scale for fall risk (19). (Supplementary Table S2), and outcomes were evaluated based on operative times, ICU stay, mobilization milestones, and cardiac function.

Preoperatively phase began upon admission and continued until surgery, encompassing respiratory training (diaphragmatic breathing: slow 3–5 s inhalation with abdominal expansion, 1-second breath-hold, controlled 3–5 s exhalation with abdominal retraction, followed by a 1-second pause, performed 20 times daily) and walking exercises (15–30 min, 1–2 times daily). Prophylactic antibiotics, patient education, and psychological support were also provided to mitigate surgical risks.

Intraoperatively, goal-directed fluid therapy, guided by transesophageal echocardiography, was employed for stroke volume optimization. Key techniques included cardiopulmonary bypass (CPB), antegrade cold-blooded cardioplegia, modified ultrafiltration, retrograde warm blood perfusion, and near-infrared spectroscopy monitoring to ensure cerebral oxygen saturation, plasma colloid osmotic pressure, and myocardial protection.

Postoperatively, patients received analgesia, anti-inflammatory agents, and early mobilization protocols, including bed-based limb exercises, early catheter removal, and supervised ambulation. Criteria for Intensive Care Unit (ICU) discharge included hemodynamic stability (urine output >1 mL/kg/h, independence from vasoactive drugs) and oxygenation adequacy (SpO₂ > 90%). Ward-based rehabilitation advanced from seated exercises to independent walking (150–200 meters). Patients were discharged upon achieving stable cardiac function, independence in daily activities, and echocardiographic confirmation of valve integrity. This comprehensive protocol demonstrably reduced postoperative complications and expedited functional recovery.

Statistical analyses were executed utilizing IBM SPSS Statistics 25.0 (SPSS Inc., Chicago, IL, USA). Before conducting hypothesis tests, data distribution was systematically assessed via the Shapiro–Wilk test for normality and Levene's test for homogeneity of variances. For comparisons between groups, parametric datasets were analyzed using an independent samples t-test with Welch's correction. Non-normally distributed data, however, were subjected to the Mann–Whitney U-test, with exact significance computation. Within-group comparisons involving repeated measures were conducted using the Friedman test, followed by Bonferroni-adjusted post-hoc analyses. The level of statistical significance was predefined at α = 0.05 (two-tailed). Given the observed baseline differences (e.g., blood pressure), their potential influence was considered in the interpretation of correlations. Analyses were performed using complete cases only; no imputation was conducted for missing data. The strength of these correlations was interpreted based on the following established benchmarks: ∣r∣ > 0.70 (very strong), 0.50–0.69 (strong), 0.30–0.49 (moderate), 0.10–0.29 (weak), and <0.10 (negligible).

As shown in Table 1. When benchmarked against ERAS protocol targets, the QT group exhibited a statistically significant difference in duration of ICU duration compared to the TT group (median: 49 vs. 43 h; p = 0.003). Similarly, the time to first off-bed mobilization was longer in the QT group (median: 67 vs. 53 h; p = 0.042). No significant intergroup differences were observed in other intraoperative or postoperative ERAS elements.

Both groups exhibited significant mobility improvements following ERAS-guided early mobilization interventions (Figures 1a–c). Activities of Daily Living (ADL) scores reached their nadir immediately post-surgery but progressively recovered, returning to preoperative levels by discharge (QT group: χ² = 160.11, p < 0.01; TT group: χ² = 61.99, p < 0.01). Similarly, Braden scales—indicating peak stress ulcer risk—were lowest postoperatively, with risk markedly reduced through mobilization (QT group: χ² = 158.19, p < 0.01; TT group: χ² = 58.80, p < 0.01). Modified Thomas fall risk assessment scale shows that there were significant changes in the intra-group comparison of risk of falling in both groups of patients. (QT group: χ² = 14.245, p < 0.05; TT group: χ² = 58.80, p < 0.01) While triple therapy patients demonstrated declining fall risk post-intervention, the QT group paradoxically showed escalating fall risk during initial ERAS implementation, peaking at secondary care transition before decreasing at discharge.

Intergroup analysis demonstrated differential ADL recovery trajectories between groups. During the post-anesthesia recovery phase, the QT group exhibited notably lower ADL scores compared to the TT group (median: 10 vs. 15; p = 0.007). Conversely, this pattern reversed during secondary care, with the QT group achieving significantly higher ADL functionality (median: 95 vs. 85; p = 0.017), suggesting accelerated late-phase rehabilitation efficacy in this group. Braden scale assessments revealed distinct stress ulcer risk profiles between groups. The QT group demonstrated significantly lower Braden scales (indicating higher ulcer risk) than the TT group during post-anesthesia recovery (median: 13 vs. 16; p = 0.025) and at discharge from ICU (median: 16 vs. 17; p = 0.006). The risk of stress ulcers was higher in the QT group. However, this trend reversed during the secondary care phase, (median: 22 vs. 21; p = 0.043) where the QT group demonstrated a significantly lower risk of stress ulcers. Inter-group analysis demonstrated no statistically significant intergroup differences in Modified Thomas fall risk assessment scale across all five assessment time points (all p > 0.05). The results of functional mobility assessement intra and intergroup comparison presented in Supplementary Table S4.

No significant differences were observed in cardiac structure/function or laboratory parameters between the QT and TT groups during postoperative and discharge assessments (Mann–Whitney U-test, all p > 0.05) (Supplementary Table S5). The comparable parameters included left atrial diameter, ascending aortic diameter, right atrial diameter, main pulmonary artery diameter, interventricular septal thickness, left ventricular end-diastolic diameter, ejection fraction, fractional shortening, white blood cell count, red blood cell count, hemoglobin, platelets, epidermal growth factor receptor (eGFR), and serum albumin levels.

To identify determinants of early mobilization efficacy, a multivariate correlation analysis was performed within the QT group. This analysis assessed the relationships between mobility indicators, ERAS protocol compliance parameters, cardiac functional indices, and laboratory biomarkers in medication-adherent patients.

Analysis of perioperative parameters revealed significant correlations with functional outcomes (Figure 1d). Mechanical ventilation duration positively correlated with ADL at ICU discharge (r = 0.581, p < 0.001) but negatively correlated with ADL at admission (r = −0.355, p = 0.039) and during secondary care (r = −0.382, p = 0.026). Similarly, ICU stay duration showed a positive correlation with ADL at ICU discharge (r = 0.533, p = 0.001) and a negative correlation with the Braden scale at ICU discharge (r = −0.412, p = 0.015). Earlier in-bed mobilization was associated with better functional outcomes at ICU discharge (r = 0.414, p = 0.017). A longer postoperative hospitalization was associated with a higher stress ulcer risk (Braden scale at ICU discharge: r = 0.405, p = 0.017) and an elevated fall risk (Modified Thomas scale).

Analysis of laboratory parameters revealed significant correlations with functional outcomes (Figure 1e). Platelet count (PLT) at discharge demonstrated protective effects, showing strong negative correlations with fall risk (TRAFS) at awakening (r = −0.370, p = 0.034), ICU discharge (r = −0.520, p = 0.002), and final discharge (r = −0.589, p < 0.001). Conversely, elevated NT-proBNP at discharge was associated with impaired functional recovery, correlating negatively with ADL during secondary care (r = −0.480, p = 0.005) and the Braden scale at awakening (r = −0.585, p < 0.001). Preoperative albumin levels appeared beneficial, showing positive correlations with ADL in secondary care (r = 0.368, p = 0.032) and the Braden scale at admission (r = 0.442, p = 0.009). A paradoxical association was observed for postoperative eGFR, which correlated negatively with ADL at ICU discharge (r = −0.395, p = 0.023) but positively with the Braden scale at awakening (r = 0.547, p = 0.001). Additionally, admission RBC count negatively correlated with postoperative ADL (r = −0.398, p = 0.020), while postoperative PLT positively correlated with ADL at ICU discharge (r = 0.393, p = 0.022).

Cardiovascular Parameter Correlations with Functional Outcomes (Figure 1f): Aortic Root (AAO): Preoperative diameter positively correlated with fall risk (TRAFS) at ICU discharge (r = 0.406, p = 0.021) and final discharge (r = 0.408, p = 0.020). Main Pulmonary Artery (MPA): Preoperative diameter positively correlated with stress ulcer risk (Braden scale) at ICU discharge (r = 0.526, p = 0.002). Interventricular Septum (IVS): Preoperative thickness showed the strongest positive correlation with Braden scale at discharge (r = 0.625, p < 0.001). Left Ventricular Diameter (LVD): Preoperative and postoperative diameters were negatively correlated with ADL scores at discharge (preoperative: r = −0.404, p = 0.022; postoperative: r = −0.522, p = 0.006). Ejection Fraction (EF) & Fractional Shortening (FS): Postoperative EF and FS were positively correlated with multiple functional recovery metrics, including ADL at discharge (EF: r = 0.462; FS: r = 0.484) and Braden scale at awakening (FS: r = 0.609). Correlation data can be found in Supplementary Tables S6–S8.

Comparative analysis shows that patients receiving quadruple therapy (QT) containing SGLT-2i have significant differences in enhanced recovery after surgery (ERAS) reporting elements compared with those receiving triple therapy (TT), manifested as prolonged ICU stay and delayed first ambulation time, thereby increasing the risk of postoperative pulmonary and thromboembolic complications and the consumption of medical resources (20, 21). Early getting out of bed and moving around is the cornerstone of postoperative rehabilitation after cardiac surgery. However, the impaired functional recovery of patients in the QT group indicates that their neuromuscular rehabilitation is inhibited. As a component of quadruple therapy, our finding of delayed early mobilization in the QT group suggests that SGLT-2i may potentially exacerbate motor dysfunction, possibly through mechanisms promoting catabolism and potentially having a muscle-reducing effect (22–24).

The perioperative activities of daily living (ADL) score was significantly correlated with laboratory indicators of inflammation, nutrition and cardiac function, revealing the multi-level physiological mechanisms of functional recovery. The ADL score is positively correlated with hemoglobin, platelet count and serum albumin, emphasizing the role of oxygen delivery, coagulation capacity and nutritional status in functional recovery (25). Organizational Doppler analysis linked impaired functional activity to left ventricular dilation and systolic dysfunction, while maintaining good cardiac output was associated with better activity. Although SGLT-2i can improve cardiac function, the potential negative impact on muscle mass observed cannot be ignored.

However, the subsequent functional recovery trajectory analysis of the QT group revealed an important clinical phenomenon: despite the challenges in the early stage of anesthesia recovery, patients in this group demonstrated accelerated functional benefits during postoperative hospitalization with the support of the structured ERAS protocol. This bidirectional recovery model emphasizes two fundamental surgical principles: First, there is a dynamic interaction between drug-induced neural regulation and the intensity of functional rehabilitation; Second, there is a negative correlation between activity limitations and the survival ability of an organization. The reduction in the risk of stress ulcers in the QT group also indicates from the side that rehabilitation exercise has a positive effect on reducing the side effects of SGLT-2i drugs.

The subsequent functional recovery of the QT group highlighted the crucial role of the ERAS protocol. The standardized and proactive activity program inherent in the ERAS protocol may serve as an effective anabolic countermeasure. By enforcing early and progressive physical activity, ERAS is likely to promote mitochondrial biosynthesis and metabolic flexibility, thereby offsetting the catabolic stress jointly induced by surgery and SGLT-2i. This is consistent with the research finding that “earlier bedside activities are associated with better functional outcomes when transferred out of the ICU” (26–30).

Activity intervention, as a core component of rehabilitation physiology, especially a crucial part of the ERAS protocol, can effectively counteract the catabolic state that SGLT-2i may induce. As a key stimulus, exercise activates mitochondrial biogenesis through signaling pathways such as AMPK/PGC-1α and optimizes the regulation of mitochondrial dynamics. The metabolic state similar to starvation induced by SGLT-2i treatment (manifested as hypoglycemia and high ketone bodies) may exacerbate the tendency towards catabolism. However, the ERAS strategy, which combines early postoperative activity with exercise intervention, enhances the utilization efficiency of glucose and fatty acids by muscles and reduces the surgical stress response (such as lowering cortisol levels), thereby synergistically improving energy metabolism balance and optimizing patient prognosis (31–33).

These findings collectively advocate that for patients receiving metabolic regulators (such as SGLT-2i), a stratified rehabilitation model should be adopted, emphasizing targeted activities in the early stage to counteract drug-induced sarcopenia, while maintaining vigilance against delayed functional compensation during the rehabilitation phase. Ultimately, the ERAS strategy that combines early postoperative intervention with exercise provides a crucial biological basis for the accelerated recovery of patients undergoing cardiovascular surgery by enhancing mitochondrial functional volume and metabolic adaptability.

At baseline, the QT group exhibited significantly lower diastolic and systolic blood pressure than the TT group. In heart failure, such hypotension often reflects impaired cardiac output and inadequate tissue perfusion, indicating severe cardiac dysfunction (34). The primary mechanism involves a decline in pumping capacity, reducing effective circulatory volume (35). Notably, excessive blood pressure reduction may compromise organ perfusion, potentially exacerbating renal impairment and increasing mortality risk. Valvular disorders or cardiac tamponade can further impair ventricular function and reduce coronary perfusion. As a compensatory response, sympathetic activation induces peripheral vasoconstriction to maintain coronary perfusion pressure, albeit at the cost of increased cardiac afterload (36). Consequently, managing hypotension in HF requires a delicate balance between maintaining vital organ perfusion and avoiding increased cardiac workload. Therefore, blood pressure serves as a significant indicator for risk prediction and assessment within the ERAS framework.

Several limitations should be considered. First, the non-randomized design and baseline differences introduce potential for confounding. Second, the lack of objective muscle mass measurements limits direct assessment of sarcopenia. Third, the findings are limited to the perioperative period, and long-term functional outcomes were not assessed. While we infer SGLT2i's sarcopenic effects from functional scores, future studies should incorporate these direct measurements to quantitatively validate the drug's impact on skeletal muscle and its interaction with rehabilitation. Future multicenter studies should validate these findings, develop individualized rehabilitation strategies for VHD patients, and enhance perioperative monitoring of limb circumference.