This retrospective analysis identified non-O blood type, elevated preoperative D-dimer levels, and intraoperative hypothermia (core temperature < 35.5 °C, a threshold selected based on internal institutional data review and the research results of Daniel et al. [13], though it differs from the conventional < 36 °C) as independent risk factors for DVT following RARP. These findings provide crucial insights for identifying high-risk patients and reveal potential targets for intervention.

The association between blood type and DVT may stem from the influence of the ABO system on coagulation. Previous studies [14,15,16,17,18] indicate that non-O blood type patients (A, B, AB) carry higher levels of von Willebrand factor (vWF) and coagulation factor VIII (FVIII), significantly increasing their venous thrombosis risk compared to type O patients. Our results show a significant negative relationship (Regression Coefficient = −1.119, z = −2.024, p = 0.043). The Odds Ratio (OR) of 0.327 indicates that for each unit increase in blood type (from Non-O = 0 to O = 1), the odds of DVT decrease by a factor of 0.327. Conversely, non-O blood type patients have a 3.058-fold (1/0.327) higher risk than type O patients. This risk estimate falls between the 1.53-fold risk reported by Spiezia et al. [19] and the 4.337-fold risk reported by Zheng et al. [16]. Furthermore, studies [20] suggest non-O blood type DVT patients have higher residual thrombus burden, recurrence rates, and lower recanalization rates, emphasizing the need for enhanced thromboprophylaxis in these patients perioperatively. Strategies include prophylactic anticoagulants (e.g., Rivaroxaban [16, 21]), perioperative mechanical prophylaxis using intermittent pneumatic compression (IPC) devices [22,23,24,25], and structured early mobilization [26]. Future research should explore the relationship between blood type and anticoagulant efficacy to optimize personalized treatment. The potential reasons why RARP surgery might particularly increase DVT risk in non-O blood type patients could be related to the specific surgical factors, such as prolonged pneumoperitoneum and the steep Trendelenburg position, which may exacerbate venous stasis and endothelial injury, interacting with the inherent prothrombotic state associated with non-O blood types.

Elevated preoperative D-dimer was identified as a strong independent predictor (Regression coefficient = 2.620, z = 4.132, p < 0.001, OR = 13.729), aligning with findings from studies on DVT after various cancer surgeries [27,28,29]. However, the wide confidence interval underscores the need for caution in interpreting the precise effect size. D-dimer, a specific degradation product of cross-linked fibrin [30], reflects a hypercoagulable state or increased fibrinolysis [31]. Surgical trauma induces a hypercoagulable stress response, compounded by postoperative immobility, causing venous stasis, creating an ideal environment for thrombus formation [29, 32]. Therefore, routine preoperative D-dimer testing, combined with Caprini scoring, aids in early identification of high-risk patients for timely pharmacological or physical prophylaxis.

Intraoperative hypothermia was another significant risk factor (OR = 3.497, z = 2.424, p = 0.015). Patients experiencing hypothermia had a 3.497-fold higher risk of DVT. Potential mechanisms include: (1) Hypothermia-induced fluid shift from intravascular to interstitial spaces, leading to hemoconcentration and reduced blood flow velocity [33]; (2) Hypothermia-triggered stress response activating platelets, inhibiting fibrinolysis, and increasing blood viscosity [34, 35]. Enhanced Recovery After Surgery (ERAS) guidelines [36, 37] strongly emphasize maintaining normothermia (core temperature ≥ 36 °C) using forced-air warming blankets, warmed fluids, and appropriate monitoring (e.g., nasopharyngeal probes [38], Temple Touch Pro system [39]). Our findings reinforce this emphasis, highlighting the critical role of the anesthesia team in intraoperative temperature monitoring and intervention.

Notably, while both operation duration and anesthesia duration were significant in univariate analysis, they were excluded from the multivariate model due to collinearity. This suggests prolonged surgery may indirectly increase DVT risk primarily by inducing hypothermia [40], rather than acting as an independent factor. Prospective studies are needed to elucidate the causal pathways.

The developed multivariate logistic regression model achieved an AUC of 0.777, indicating moderate discriminative ability for predicting DVT after RARP, comparable to or slightly better than the traditional Caprini score (typical AUC range 0.70–0.75) [9]. The core variables (blood type, preoperative D-dimer, intraoperative hypothermia) are readily available clinically and can be integrated into preoperative risk assessment. For instance, patients identified as high-risk (non-O blood type or elevated D-dimer) could receive personalized prophylaxis: early preoperative initiation of pharmacologic/physical measures, intraoperative IPC application and active warming, postoperative early mobilization protocols, and potentially extended anticoagulation. Furthermore, the combined prediction model can aid in identifying high-risk patients for optimized resource allocation.

Several important limitations must be acknowledged: (1) Retrospective design potentially introducing selection bias; unmeasured confounders like hereditary thrombophilia (e.g., Factor V Leiden mutation) were not included; (2) Single-center study with a relatively small sample size (n = 199) and limited DVT cases (n = 32), potentially affecting statistical power and generalizability. The number of events is just above the minimum suggested for the number of predictors, which may limit model stability; (3) Missing data on postoperative early mobilization specifics and anticoagulant dosing may influence accuracy. Future research should involve multicenter, large-sample prospective studies incorporating genomic analysis to refine the prediction model.