Impact of Perioperative Buyang Huanwu-Based Traditional Chinese Medicine Intervention on Postoperative Venous Thromboembolism, Coagulation Markers and Systemic Inflammation: A Propensity Score-Weighted Retrospective Cohort Study

Abstract

Objective

This study aimed to (1) evaluate, in a propensity-score-weighted retrospective cohort, the effectiveness and safety of a standardised perioperative Buyang Huanwu-based traditional Chinese medicine (TCM) decoction combined with standard low-molecular-weight heparin (LMWH) prophylaxis on the 30-day incidence of symptomatic and asymptomatic venous thromboembolism (VTE), (2) characterise the longitudinal trajectories of coagulation (D-dimer, fibrinogen) and inflammatory (C-reactive protein [CRP]) biomarkers and (3) quantify the net clinical benefit of this integrative approach.

Methods

We conducted a retrospective cohort study of 1847 adult patients undergoing major general-surgical or gynaecological procedures between January 2020 and December 2024 at a tertiary integrative-medicine centre. Patients receiving a standardised perioperative TCM decoction based on the modified Buyang Huanwu Tang in addition to standard LMWH prophylaxis (n = 892) were compared with patients receiving standard prophylaxis alone (n = 955). The TCM decoction comprised seven herbal components dispensed at fixed daily doses (Huangqi, 60 g; Danggui, 9 g; Chishao, 6 g; Chuanxiong, 6 g; Honghua, 3 g; Taoren, 6 g; Dilong, 9 g) supplemented with 15 g of Danshen; granules were prepared as a single-batch good manufacturing practice-certified concentrated formulation, with the principal bioactive markers astragaloside IV, ferulic acid, paeoniflorin and salvianolic acid B quantified by high-performance liquid chromatography and required to fall within ±10% of the label content to ensure dose standardisation. Inverse probability of treatment weighting (IPTW) was used to balance baseline characteristics. The primary outcome was 30-day VTE confirmed by duplex ultrasonography or computed tomography pulmonary angiography, with symptomatic and asymptomatic events recorded separately. Secondary outcomes included major bleeding, length of stay, patient-reported pain and 90-day all-cause mortality. Linear mixed-effects models were used to analyse longitudinal trajectories of D-dimer, fibrinogen and CRP measured at baseline and on postoperative days 1, 3 and 7.

Results

After IPTW adjustment, TCM intervention was associated with significantly reduced 30-day VTE incidence (hazard ratio [HR]: 0.58, 95% confidence interval [CI]: 0.41–0.82, p = 0.002), with consistent protection observed for both symptomatic VTE (HR: 0.55, 95% CI: 0.34–0.89) and ultrasound-detected asymptomatic distal deep-vein thrombosis (HR: 0.61, 95% CI: 0.39–0.94). Landmark analysis revealed sustained protection during both hospitalisation (HR: 0.52, 95% CI: 0.33–0.81) and the post-discharge period (HR: 0.67, 95% CI: 0.42–0.93). Linear mixed-effects modelling demonstrated significant group-by-time interactions for D-dimer (p < 0.001) and CRP (p = 0.003), indicating accelerated biomarker clearance in the TCM group: 42% faster D-dimer resolution and 38% faster CRP normalisation than in controls. No significant difference was observed for major bleeding (2.1% vs 2.4%, p = 0.68) or 90-day all-cause mortality (1.6% vs 2.0%, p = 0.51). Decision-curve analysis confirmed superior net clinical benefit across the clinically relevant threshold-probability range.

Conclusion

Perioperative Buyang Huanwu-based TCM combined with standard LMWH prophylaxis significantly reduces postoperative VTE risk – for both symptomatic and ultrasound-detected asymptomatic events – through accelerated resolution of coagulation and inflammatory cascades, without increasing bleeding or 90-day all-cause mortality. These findings support the integration of standardised TCM into enhanced-recovery-after-surgery protocols, particularly for moderate- to high-risk surgical patients in whom anticoagulation intensification may be contraindicated.

Introduction

Venous thromboembolism (VTE), comprising deep vein thrombosis (DVT) and pulmonary embolism (PE), constitutes one of the most prevalent and potentially life-threatening complications following surgical intervention. The pathophysiological triad originally described by Virchow – encompassing vascular endothelial injury, venous stasis and hypercoagulability – remains the cornerstone of our understanding of perioperative thrombogenesis.¹ Epidemiological data indicate that without prophylaxis, the incidence of VTE following major abdominal and pelvic surgeries ranges from 15% to 40%.² The consequences of VTE extend beyond the acute event, with survivors facing substantial risks of post-thrombotic syndrome, chronic thromboembolic pulmonary hypertension and recurrent thromboembolism, collectively imposing considerable burdens on healthcare systems and patient quality of life.³ Despite major advances in pharmacological and mechanical prophylaxis over the past three decades, VTE remains a leading cause of preventable in-hospital mortality, underscoring the imperative for continued refinement of preventive strategies.⁴

Contemporary standard prophylaxis protocols incorporate pharmacological agents – predominantly low-molecular-weight heparin (LMWH), unfractionated heparin and novel oral anticoagulants – frequently combined with mechanical interventions, including graduated compression stockings and intermittent pneumatic compression devices.⁵ These multimodal approaches have demonstrably reduced VTE incidence across diverse surgical populations; however, their implementation is not without limitations. Pharmacological prophylaxis carries an inherent bleeding risk, with major bleeding complications reported in 1%–3% of patients receiving therapeutic anticoagulation; this concern is particularly relevant in procedures with high baseline haemorrhagic risk, such as neurosurgery and hepatic resection.⁶ Furthermore, patient adherence to extended prophylaxis regimens following hospital discharge remains suboptimal, with studies documenting compliance rates <60% for home-based anticoagulation protocols.⁷ These challenges have stimulated interest in complementary therapeutic modalities that may enhance thromboprophylaxis efficacy while maintaining acceptable safety profiles, particularly in populations where conventional anticoagulation intensification is precluded by bleeding concerns.⁸

Among complementary strategies under investigation, several traditional Chinese medicine (TCM) formulations have undergone rigorous pharmacological characterisation and demonstrated mechanisms potentially relevant to thromboprophylaxis. The classical formula Buyang Huanwu Tang contains a defined set of bioactive constituents, including astragaloside IV from Astragalus membranaceus, ferulic acid and ligustilide from Angelica sinensis and Ligusticum chuanxiong, paeoniflorin from Paeonia lactiflora, hydroxysafflor yellow A from Carthamus tinctorius, amygdalin from Prunus persica seed and lumbrokinase from Pheretima.^9,10 The single herb Danshen (Salvia miltiorrhiza) contains tanshinones and salvianolic acid B, which have been pharmacologically characterised in relation to microcirculation and vascular protection.¹¹ Phytochemical and pharmacological studies have shown that these compounds collectively exert antiplatelet effects, enhancement of tissue-plasminogen-activator expression with reciprocal suppression of plasminogen-activator-inhibitor-1, improvement of haemorheological parameters and suppression of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB)-mediated cytokine release. Because the concentrations of these active principles can vary substantially between batches of raw herbs, our institution dispenses the formulation as a single-batch good manufacturing practice (GMP)-certified concentrated decoction granule with high-performance liquid chromatography (HPLC)-quantified marker compounds to ensure dose standardisation across all treated patients.¹²

Pharmacological studies of Danshen-related formulations support vascular and anti-inflammatory mechanisms relevant to thromboprophylaxis.¹³ Importantly, the inflammation–coagulation nexus provides a coherent mechanistic rationale for combining TCM with conventional anticoagulants rather than viewing them as alternative strategies. Surgical trauma initiates an acute inflammatory response in which interleukin-6 and tumour-necrosis-factor-α upregulate tissue-factor expression, downregulate protein-C and protein-S signalling and inhibit fibrinolysis through increased plasminogen-activator-inhibitor-1 production; thrombin generated by the activated coagulation cascade itself functions as a pro-inflammatory mediator via protease-activated-receptor signalling, perpetuating a self-amplifying loop. Low-molecular-weight heparin acts predominantly downstream of this loop by accelerating antithrombin-mediated inactivation of factor Xa and thrombin but without attenuating the upstream inflammatory drive.^14,15 Buyang Huanwu-based TCM, by contrast, targets the upstream NF-κB/cytokine arm and the fibrinolytic/haemorheological compartment.¹³ The two interventions therefore act on complementary rather than redundant nodes of the inflammation–haemostasis network, providing the pharmacological basis for the synergism evaluated in the present study.

Despite the theoretical rationale and growing clinical use of TCM in perioperative medicine, the evidence base for VTE prevention remains insufficient. Most prior reports relied on single-timepoint outcome assessments and did not characterise the dynamic evolution of coagulation and inflammatory parameters across the postoperative recovery trajectory. They also presented efficacy and safety outcomes in isolation, without quantifying net clinical benefit, and rarely used propensity-score methodology to mitigate confounding by indication in observational data. In addition, almost all existing studies pooled symptomatic and ultrasound-detected asymptomatic VTE without distinction, which limits clinical interpretability because the prognostic weight of these two endpoints differs.¹⁶

This study was designed to address these gaps through three complementary objectives. First, we evaluated whether perioperative Buyang Huanwu-based TCM, administered as an adjunct to standard LMWH-based prophylaxis, reduces the 30-day incidence of imaging-confirmed VTE – analysed separately for symptomatic events and for ultrasound-detected asymptomatic distal DVT – compared with standard prophylaxis alone, using inverse probability of treatment weighting (IPTW) for causal inference in a retrospective cohort. Second, through serial assessment of coagulation (D-dimer, fibrinogen) and inflammatory (C-reactive protein [CRP]) biomarkers at multiple postoperative timepoints, we characterised differential trajectories between treatment groups to elucidate the mechanisms underlying the observed clinical effects. Third, we quantified the net clinical benefit of TCM integration by balancing thrombotic event reduction against bleeding complications using decision-curve analysis and additionally assessed 90-day all-cause mortality to capture longer-term safety. By addressing these objectives with methodological rigour, the investigation generates clinically actionable evidence on the role of standardised TCM within contemporary enhanced-recovery-after-surgery (ERAS) frameworks.

Methods

Study Design and Setting

This retrospective cohort study was conducted at a tertiary academic medical centre (Mianyang Traditional Chinese Medicine Hospital) with integrated traditional and Western medical services, spanning the period from 1 January 2020 to 31 December 2024. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for cohort studies, and the completed checklist is provided as a supplementary file.¹⁷ The institutional electronic medical record system was utilised to identify eligible patients and extract relevant clinical data. The study protocol received approval from the Institutional Review Board, with waiver of individual informed consent granted given the retrospective nature of the investigation and implementation of appropriate data de-identification procedures.

Participants

The inclusion criteria encompassed the following: (1) adult patients aged ≥18 years; (2) undergoing elective or urgent surgical procedures in general surgical or gynaecological departments requiring postoperative thromboprophylaxis; (3) completion of standardised preoperative VTE risk assessment using the Caprini Risk Assessment Model; (4) availability of complete perioperative imaging studies (duplex ultrasonography and/or computed tomography pulmonary angiography) for VTE diagnosis; and (5) availability of serial laboratory data, including D-dimer, fibrinogen and CRP measurements at baseline and on postoperative days 1, 3 and 7. The exclusion criteria included the following: (1) documented contraindication to pharmacological anticoagulation (active bleeding, severe thrombocytopenia with platelet count <50,000/μL or recent haemorrhagic stroke within 3 months); (2) pre-existing VTE diagnosed within 6 months prior to surgery; (3) therapeutic anticoagulation for other indications (atrial fibrillation, mechanical heart valves); (4) pregnancy or immediate postpartum period; (5) documented allergy or intolerance to study TCM formulations; (6) incomplete medical records precluding accurate outcome ascertainment; or (7) concurrent enrolment in interventional clinical trials affecting anticoagulation management.

Sample size estimation was performed using G*Power software (version 3.1.9.7) based on anticipated effect sizes derived from preliminary institutional data and the published literature. Assuming a baseline VTE incidence of 8% in the control group, a clinically meaningful relative risk reduction of 40% (to 4.8% in the intervention group), two-sided alpha of 0.05 and statistical power of 80%, the minimum required sample size was calculated as 752 patients per group. Anticipating approximately 15% data incompleteness or loss to follow-up, an enrolment of at least 865 patients per group was targeted. The final analytical cohort of 1,847 patients exceeded this threshold, providing adequate statistical power for primary and exploratory analyses.

Exposure Definition

The exposure of interest was a standardised perioperative Buyang Huanwu-based TCM decoction administered as an adjunct to conventional VTE prophylaxis. The expected pharmacological contribution of the TCM arm was prespecified as adjunctive upstream modulation of inflammatory activation, fibrinolysis and microcirculatory flow; the control arm received the downstream anticoagulant effect of LMWH plus mechanical prophylaxis only. The decoction was prepared as a single-batch GMP-certified concentrated-granule formulation manufactured by the institutional TCM pharmacy, and the daily dose contained 60 g of Huangqi (As. membranaceus radix), 9 g of Danggui (An. sinensis radix), 6 g of Chishao (Pa. lactiflora radix rubra), 6 g of Chuanxiong (L. chuanxiong rhizoma), 3 g of Honghua (C. tinctorius flos), 6 g of Taoren (Pr. persica semen) and 9 g of Dilong (Pheretima), with 15 g of Danshen (S. miltiorrhiza radix) added for its documented effects on microcirculation and fibrinolysis. Quality control was performed using HPLC quantification of the principal bioactive markers astragaloside IV (≥0.40 mg/g), ferulic acid (≥0.10 mg/g), paeoniflorin (≥1.20 mg/g) and salvianolic acid B (≥3.00 mg/g); each released batch was required to fall within ±10% of the labelled marker content.¹² The decoction was reconstituted to 200 mL of warm water and administered as 100 mL twice daily; treatment was initiated within 24 h of surgical completion after resumption of enteral intake, continued throughout hospitalisation and for 14 days following discharge. Standard prophylaxis in both groups consisted of 40 mg of subcutaneous enoxaparin once daily (weight-adjusted in patients weighing <50 or >100 kg), supplemented by mechanical prophylaxis with graduated compression stockings and intermittent pneumatic compression devices during periods of immobility. Pharmacological prophylaxis was maintained throughout hospitalisation and discontinued at discharge or upon full ambulation, in accordance with departmental protocol. The decision to add TCM was made by the attending surgeon in consultation with the institutional integrative-medicine service; criteria for offering TCM included a Caprini score ≥5, planned major abdominal-pelvic or gynaecological surgery and absence of pre-specified contraindications (severe hepatic impairment, documented herbal allergy or active malignant bowel obstruction precluding enteral intake). Patients who declined the additional decoction or who had logistical barriers to post-discharge decoction supply formed the control group; this non-randomised assignment created the confounding by indication that the IPTW analysis was specifically designed to address.

Data Collection and Variables

Demographic and clinical data were extracted from electronic medical records using a structured data collection instrument. Variables included age, sex, body mass index, smoking status, comorbidities (diabetes mellitus, hypertension, coronary artery disease, chronic kidney disease, malignancy), surgical speciality, specific procedure type, surgical duration, anaesthesia modality (general, regional or combined), American Society of Anesthesiologists physical status classification, preoperative Caprini VTE risk score and baseline laboratory parameters, including complete blood count, coagulation studies (prothrombin time, activated partial thromboplastin time) and renal function. Serial biomarker data collected at preoperative baseline and on postoperative days 1, 3 and 7 included D-dimer (quantitative immunoturbidimetric assay, μg/mL fibrinogen equivalent units), fibrinogen (Clauss method, g/L) and high-sensitivity CRP (immunonephelometry, mg/L). All laboratory analyses were performed in the institutional central laboratory with appropriate quality control procedures.

Outcome Measures

The primary outcome was the occurrence of objectively confirmed VTE within 30 days of the index surgical procedure. Venous thromboembolism was defined as the composite of DVT (confirmed by compression duplex ultrasonography demonstrating non-compressibility of deep venous segments) and PE (confirmed by computed tomography pulmonary angiography demonstrating intraluminal filling defects). Each VTE event was further classified at adjudication as symptomatic – defined as new-onset unilateral limb swelling, calf pain, dyspnoea, pleuritic chest pain, syncope or unexplained tachycardia/hypoxaemia prompting imaging – or as asymptomatic, when the event was detected by protocol-mandated surveillance duplex ultrasonography in the absence of clinical symptoms. Routine surveillance ultrasonography was performed on postoperative day 7 and at the 30-day follow-up visit; additional imaging was obtained whenever symptoms suggestive of VTE arose. All imaging studies were independently re-read by two vascular radiologists blinded to treatment allocation, with disagreements resolved by a third radiologist. Secondary outcomes included (1) major bleeding events defined according to International Society on Thrombosis and Haemostasis criteria (fatal bleeding, bleeding into a critical organ, bleeding causing a haemoglobin decrease ≥2 g/dL or bleeding requiring transfusion of ≥2 units of packed red blood cells); (2) length of hospital stay; (3) intensive-care-unit admission and duration; (4) patient-reported pain intensity on a visual-analogue scale (0–100 mm); and (5) 90-day all-cause mortality, ascertained from inpatient records and from structured telephone follow-up on days 60 and 90 by research nurses blinded to treatment allocation. Exploratory outcomes included biomarker clearance rates calculated as the slope of log-transformed D-dimer and CRP values across postoperative days 0–7.

Statistical Analysis

All statistical analyses were performed using R software (version 4.2.3; R Foundation for Statistical Computing, Vienna, Austria) with significance set at two-tailed p < 0.05. Baseline characteristics were compared between groups using independent t-tests or Mann–Whitney U-tests for continuous variables and chi-square or Fisher’s exact tests for categorical variables. To address confounding by indication inherent in observational studies, the previously defined inverse probability of treatment weighting framework was employed as the primary analytical approach. Propensity scores representing the probability of receiving TCM intervention were estimated using multivariable logistic regression incorporating all clinically relevant baseline covariates, including demographics, comorbidities, surgical parameters and Caprini scores. Stabilised weights were calculated to reduce variance while maintaining balance, with extreme weights truncated at the 1st and 99th percentiles. Covariate balance after weighting was assessed using standardised mean differences (SMDs), with SMD < 0.10 indicating adequate balance. The primary outcome was analysed using weighted Cox proportional hazards regression, with hazard ratios (HRs) and 95% confidence intervals (CIs) reported. The proportional hazards assumption was evaluated using Schoenfeld residuals. Landmark analysis stratified the observation period into hospitalisation (days 0–7) and post-discharge (days 8–30) phases to characterise the temporal pattern of TCM effectiveness. Sensitivity analyses included traditional propensity score matching (1:1 nearest-neighbour matching with a calliper of 0.2), multivariable Cox regression adjusting for propensity score quintiles and doubly robust estimation combining IPTW with outcome regression. Evidence (E)-values were calculated to quantify the minimum strength of association with unmeasured confounders required to explain away observed effects. Longitudinal biomarker trajectories were analysed using linear mixed-effects models with fixed effects for group, time and group-by-time interaction and random intercepts and slopes for individual participants. Subgroup analyses examined effect modification by Caprini risk category, surgical speciality, age and sex, with interaction p-values reported. Net clinical benefit was assessed using decision curve analysis comparing the strategies of TCM plus standard prophylaxis versus standard prophylaxis alone across a range of threshold probabilities, with net benefit calculated as (true positive rate) − (false positive rate × weight), where weight reflects the harm-to-benefit ratio.

Results

Cohort Characteristics and Propensity Score Weighting

Of the 2,463 patients initially screened, 1,847 met the inclusion criteria and were included in the final analytical cohort (Figure 1). The TCM intervention group comprised 892 patients (48.3%), and the standard-prophylaxis-alone (control) group comprised 955 patients (51.7%). The most common reasons for exclusion were incomplete biomarker data (n = 287), contraindications to anticoagulation (n = 156), pre-existing VTE within 6 months (n = 89) and concurrent therapeutic anticoagulation (n = 84). Before weighting, the TCM group exhibited a higher prevalence of malignancy (18.7% vs 14.2%, p = 0.008), longer mean operative duration (142 ± 56 vs 128 ± 48 minutes, p < 0.001) and higher Caprini scores (6.8 ± 2.1 vs 6.2 ± 1.9, p < 0.001), consistent with confounding by indication in which TCM was preferentially administered to higher-risk patients. Of the 105 VTE events observed during follow-up, 46 (43.8%) were symptomatic at the time of detection, and 59 (56.2%) were detected solely through protocol-mandated surveillance ultrasonography, justifying the pre-specified separate analysis of these two presentations. Table 1 presents baseline characteristics before and after IPTW adjustment.

Figure 1.

Flowchart depicting participant selection, exclusion criteria application and final cohort composition for the TCM intervention and control groups.

Abbreviations: TCM, Traditional Chinese Medicine; VTE, venous thromboembolism.

Table 1.

Baseline Characteristics of Study Cohort Before and After IPTW Adjustment

Variable	TCM (n=892)	Control (n=955)	SMD	TCM (Weighted)	Control (Weighted)	SMD
Age, years, mean±SD	58.4±14.2	56.8±13.8	0.11	57.6±14.0	57.4±13.9	0.01
Female, n (%)	498 (55.8)	512 (53.6)	0.04	54.2%	54.8%	0.01
BMI, kg/m2, mean±SD	26.4±4.8	25.9±4.5	0.11	26.1±4.6	26.0±4.5	0.02
Malignancy, n (%)	167 (18.7)	136 (14.2)	0.12	16.1%	16.4%	0.01
Caprini score, mean±SD	6.8±2.1	6.2±1.9	0.30	6.5±2.0	6.4±1.9	0.05
Surgical duration, min	142±56	128±48	0.27	134±52	133±50	0.02

Abbreviations: BMI, body mass index; IPTW, inverse probability of treatment weighting; SD, standard deviation; SMD, standardized mean difference; TCM, Traditional Chinese Medicine.

Following IPTW adjustment, excellent covariate balance was achieved across all measured confounders, with all standardised mean differences reduced to <0.10 (Table 1). The propensity score distributions demonstrated substantial overlap between groups, confirming the positivity assumption (Figure 2). The effective sample sizes after weighting were 847 for the TCM group and 912 for the control group.

Figure 2.

Covariate balance assessment. (A) Mirrored histogram showing propensity-score distributions for the TCM and control groups before and after IPTW, demonstrating substantial overlap. (B) Love plot displaying standardized mean differences for all covariates before weighting (red circles) and after IPTW (blue squares). Dashed vertical lines indicate the prespecified adequate-balance threshold of |SMD| < 0.10.

Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index; IPTW, inverse probability of treatment weighting; SMD, standardized mean difference; TCM, Traditional Chinese Medicine.

Primary Outcome: Venous Thromboembolism

During the 30-day follow-up period, VTE events occurred in 37 patients (4.1%) in the TCM group compared with 68 patients (7.1%) in the control group (absolute risk difference: −3.0%, 95% CI: −5.1% to −0.9%). In the IPTW-weighted analysis, TCM intervention was associated with significantly reduced VTE risk (HR: 0.58, 95% CI: 0.41–0.82, p = 0.002). This protective effect remained robust across all sensitivity analyses: propensity score-matched cohort (HR: 0.61, 95% CI: 0.42–0.88), multivariable-adjusted model (HR: 0.55, 95% CI: 0.38–0.79) and doubly robust estimation (HR: 0.57, 95% CI: 0.40–0.81). The calculated E-value of 2.86 indicated that unmeasured confounding would need to be associated with both treatment assignment and VTE outcome by a risk ratio of ≥2.86 to fully explain away the observed effect, suggesting reasonable robustness to residual confounding (Table 2).

Table 2.

Primary and Secondary Outcomes Analysis Across Multiple Statistical Models

Outcome	Crude HR (95% CI)	Adjusted HR (95% CI)	PSM HR (95% CI)	IPTW HR (95% CI)
VTE (composite)	0.57 (0.38–0.85)	0.55 (0.38–0.79)	0.61 (0.42–0.88)	0.58 (0.41–0.82)*
Symptomatic VTE	0.54 (0.34–0.86)	0.52 (0.32–0.85)	0.58 (0.36–0.92)	0.55 (0.34–0.89)*
Asymptomatic distal DVT	0.65 (0.42–1.01)	0.62 (0.40–0.97)	0.66 (0.43–1.03)	0.61 (0.39–0.94)*
Proximal DVT	0.48 (0.27–0.85)	0.46 (0.26–0.82)	0.52 (0.30–0.90)	0.49 (0.29–0.83)*
Distal DVT	0.63 (0.40–0.99)	0.61 (0.39–0.96)	0.66 (0.42–1.03)	0.64 (0.41–0.99)*
Pulmonary embolism	0.60 (0.30–1.20)	0.58 (0.29–1.16)	0.64 (0.32–1.28)	0.62 (0.32–1.21)
Major bleeding	0.88 (0.48–1.61)	0.91 (0.50–1.66)	0.94 (0.51–1.73)	0.89 (0.49–1.62)
90-day all-cause mortality	0.80 (0.41–1.56)	0.76 (0.38–1.53)	0.82 (0.41–1.62)	0.78 (0.39–1.55)

Notes: *P<0.05.

Abbreviations: CI, confidence interval; DVT, deep vein thrombosis; HR, hazard ratio; IPTW, inverse probability of treatment weighting; PSM, propensity score matching; VTE, venous thromboembolism.

Landmark analysis revealed that the protective effect of TCM was evident during both the hospitalisation phase (days 0–7 – HR: 0.52, 95% CI: 0.33–0.81, p = 0.004) and the post-discharge phase (days 8–30 – HR: 0.67, 95% CI: 0.42–0.93, p = 0.021), indicating sustained therapeutic benefit beyond the inpatient period (Figure 3). This pattern has particular clinical relevance given the documented decline in post-discharge prophylaxis adherence. When VTE events were analysed by clinical presentation, TCM was associated with reduced risk of both symptomatic VTE (15 events [1.7%] vs 31 events [3.2%]; IPTW HR: 0.55, 95% CI: 0.34–0.89) and ultrasound-detected asymptomatic distal DVT (22 events [2.5%] vs 37 events [3.9%]; IPTW HR: 0.61, 95% CI: 0.39–0.94), with the effect on symptomatic events being numerically larger but statistically similar (interaction p = 0.62). When analysed by anatomical site, TCM reduced the risk of both proximal DVT (HR: 0.49, 95% CI: 0.29–0.83) and distal DVT (HR: 0.64, 95% CI: 0.41–0.99); the effect on PE approached but did not reach statistical significance (HR: 0.62, 95% CI: 0.32–1.21, p = 0.16), reflecting the limited statistical power for this comparatively uncommon endpoint.

Figure 3.

Kaplan-Meier curves for VTE-free survival over 30 days comparing Buyang Huanwu-based TCM plus standard prophylaxis with standard prophylaxis alone. Shaded areas represent 95% confidence intervals; the number-at-risk table is shown below the plot.

Abbreviations: CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighting; TCM, Traditional Chinese Medicine; VTE, venous thromboembolism.

Biomarker Trajectories: Coagulation and Inflammation

Linear mixed-effects modelling revealed significant differences in postoperative biomarker trajectories between groups (Figure 4A–C). Both groups demonstrated expected postoperative elevations in D-dimer, fibrinogen and CRP, with peak values typically observed on postoperative day 1. However, the TCM group exhibited significantly accelerated biomarker normalisation compared with controls. For D-dimer, the group-by-time interaction was highly significant (p < 0.001), with the TCM group demonstrating a clearance rate 42% faster than controls (slope −0.18 vs −0.11 log μg/mL/day). Similarly, CRP showed a significant interaction effect (p = 0.003), with 38% faster normalisation in the TCM group. Fibrinogen trajectories showed a trend towards faster resolution in the TCM group, though the interaction term did not achieve statistical significance (p = 0.078) (Table 3).

Figure 4.

Longitudinal trajectories of coagulation and inflammatory biomarkers from baseline through postoperative day 7. (A) D-dimer, (B) fibrinogen and (C) C-reactive protein trajectories are shown for TCM plus standard prophylaxis versus standard prophylaxis alone; points represent observed means and error bars represent 95% confidence intervals.

Abbreviations: CRP, C-reactive protein; POD, postoperative day; TCM, Traditional Chinese Medicine.

Table 3.

Longitudinal Analysis of Coagulation and Inflammatory Biomarkers Using Linear Mixed-Effects Models

Biomarker	Time Effect (P-Value)	Group Effect (P-Value)	Time×Group Interaction (P-Value)
D-dimer (log μg/mL)	<0.001	0.042	<0.001*
Fibrinogen (g/L)	<0.001	0.156	0.078
C-reactive protein (log mg/L)	<0.001	0.018	0.003*

Notes: *Statistically significant interaction indicating differential biomarker trajectories between groups.

Subgroup Analyses and Effect Modification

Forest plot visualisation of subgroup analyses revealed consistent protective effects across most pre-specified subgroups (Figure 5). Notably, the protective effect appeared more pronounced among patients with Caprini scores ≥10 (HR: 0.42, 95% CI: 0.24–0.73) than those with scores 5–9 (HR: 0.68, 95% CI: 0.45–1.02), with a significant interaction p-value of 0.032. This finding suggests enhanced benefit in the highest-risk patients. No significant effect modification was observed by age, sex, surgical speciality or anaesthesia modality.

Figure 5.

Forest plot displaying hazard ratios for VTE across pre-specified subgroups with 95% confidence intervals. The vertical dashed line at HR = 1.0 represents no effect; values <1 favor TCM plus standard prophylaxis and values >1 favor standard prophylaxis alone. P values for interaction are shown in the rightmost column.

Abbreviations: CI, confidence interval; HR, hazard ratio; TCM, Traditional Chinese Medicine.

Safety Outcomes and Net Clinical Benefit

Major bleeding events occurred in 19 patients (2.1%) in the TCM group and 23 patients (2.4%) in the control group, with no significant difference (IPTW HR: 0.89, 95% CI: 0.49–1.62, p = 0.68). Clinically relevant non-major bleeding was similarly balanced between groups (5.8% vs 6.2%, p = 0.74). Ninety-day all-cause mortality was 14 deaths (1.6%) in the TCM group and 19 deaths (2.0%) in the control group (IPTW HR: 0.78, 95% CI: 0.39–1.55, p = 0.48), and no death was adjudicated as related to either the TCM decoction or bleeding from anticoagulation. Decision-curve analysis demonstrated that TCM combined with standard prophylaxis provided superior net clinical benefit to standard prophylaxis alone across threshold probabilities ranging from 2% to 15%, encompassing the clinically relevant range for VTE prevention decision-making (Figure 6). The net-benefit advantage was most pronounced at threshold probabilities between 4% and 10%, corresponding to moderate- to high-risk surgical patient populations (Table 4).

Figure 6.

Decision-curve analysis comparing net clinical benefit of TCM plus standard prophylaxis versus standard prophylaxis alone across threshold probabilities. The grey line represents the treat-all strategy and the black line represents the treat-none strategy; higher net-benefit values indicate superior clinical utility.

Abbreviation: TCM, Traditional Chinese Medicine.

Table 4.

Secondary Outcomes: Postoperative Recovery and Healthcare Resource Utilization

Outcome	TCM Group (n=892)	Control Group (n=955)	P-Value
Length of stay, days, median (IQR)	6 (4–9)	7 (5–10)	0.024
ICU admission, n (%)	67 (7.5)	82 (8.6)	0.38
Pain VAS at day 3, mean±SD	32.4±18.6	38.2±21.3	0.008
Total hospitalization cost, ¥, median (IQR)	42,650 (28,400–68,200)	45,820 (30,100–72,500)	0.112
90-day all-cause mortality, n (%)	14 (1.6)	19 (2.0)	0.51

Abbreviations: ICU, intensive care unit; IQR, interquartile range; SD, standard deviation; VAS, visual analog scale.

Discussion

This large retrospective cohort study demonstrates that perioperative Buyang Huanwu-based TCM intervention, when combined with standard pharmacological and mechanical prophylaxis, significantly reduces the incidence of postoperative VTE without increasing bleeding complications or 90-day mortality. The 42% relative risk reduction observed in the IPTW-weighted analysis translates to a clinically meaningful absolute risk reduction of 3.0%, with a number needed to treat of approximately 33 patients to prevent one VTE event. Importantly, the protective effect was demonstrated for both symptomatic VTE and ultrasound-detected asymptomatic distal DVT and was sustained beyond hospital discharge, addressing a critical vulnerability window when prophylaxis adherence typically declines. The mechanistic biomarker analyses provide pathophysiological support for these clinical findings by demonstrating accelerated resolution of both coagulation activation (D-dimer) and systemic inflammation (CRP) in TCM-treated patients. Taken together, these results support the integration of standardised TCM into multimodal VTE-prevention strategies, particularly within ERAS frameworks.

A mechanistic rationale for combining Buyang Huanwu-based TCM with LMWH is provided by the complementary, rather than overlapping, sites of pharmacological action of the two regimens. Low-molecular-weight heparin exerts its antithrombotic effect almost exclusively at the level of factor-Xa– and thrombin–antithrombin complex formation, with little impact on upstream tissue-factor expression, NF-κB-driven cytokine release or endogenous fibrinolytic capacity. By contrast, the bioactive constituents of the present decoction act predominantly upstream and laterally; astragaloside IV and salvianolic acid B suppress monocyte tissue-factor and NF-κB signalling, ferulic acid and hydroxysafflor yellow A inhibit platelet aggregation and improve endothelial function, lumbrokinase and tanshinone IIA enhance tissue-plasminogen-activator release while reducing plasminogen-activator-inhibitor-1 and paeoniflorin attenuates endothelial inflammatory injury. Because the inflammation–coagulation crosstalk operates as a self-amplifying loop, simultaneously dampening the upstream inflammatory drive (TCM) and the downstream thrombin generation (LMWH) is expected to produce greater than additive benefit. The accelerated clearance of both D-dimer and CRP – and the absence of an excess bleeding signal – observed in the present cohort is consistent with this synergistic, mechanistically rational combination rather than with redundant pharmacology.^13–15,18

The magnitude of VTE risk reduction observed in our study aligns with findings from several prior investigations of TCM in perioperative settings while extending the evidence base through more rigorous methodology and mechanistic exploration. Chen et al¹⁸ reported a similar risk reduction in DVT among patients receiving TCM combined with LMWH after surgery, though the available evidence was dominated by small randomized trials and did not incorporate propensity-score adjustment. Similarly, Yu et al¹⁹ demonstrated improved D-dimer profiles in patients undergoing gynaecological surgery receiving TCM but did not assess clinical VTE endpoints. Our findings confirm and extend these observations by demonstrating both clinical efficacy and mechanistic plausibility across multiple surgical specialities. The sustained protective effect during the post-discharge period observed in our landmark analysis carries particular clinical significance. Studies have consistently documented implementation and adherence challenges for extended postoperative LMWH prophylaxis after discharge.²⁰ The extended benefit of TCM may reflect either continued pharmacological effect from the prescribed post-discharge regimen or persistent modification of the coagulation–inflammation milieu established during hospitalisation, consistent with evidence supporting extended thromboprophylaxis after abdominopelvic surgery.²¹ Future studies incorporating biomarker sampling during the post-discharge period would help distinguish between these possibilities.

The observed acceleration of D-dimer and CRP clearance in TCM-treated patients provides mechanistic insight into the clinical efficacy findings. As a fibrin degradation product, D-dimer serves as a sensitive marker of fibrin turnover and has established prognostic value for VTE prediction.²² The 42% faster D-dimer resolution observed in our TCM cohort suggests enhanced fibrinolytic capacity, consistent with in vitro studies demonstrating that Danshen compounds upregulate tissue plasminogen activator expression and inhibit plasminogen activator inhibitor-1.²³ The parallel acceleration of CRP normalisation supports the hypothesis that TCM exerts concurrent anti-inflammatory effects, potentially interrupting the inflammation–coagulation crosstalk that amplifies perioperative thrombotic risk. Preclinical studies have demonstrated that salvianolic acid B, a principal bioactive component of Danshen, suppresses nuclear factor-kappa B signalling and reduces proinflammatory cytokine production by activated macrophages.²⁴ The significant group-by-time interaction for both D-dimer and CRP in our mixed-effects models suggests that these effects are not merely differences in baseline values but represent genuinely differential trajectories; that is, TCM modifies the temporal dynamics of postoperative pathophysiology rather than simply shifting absolute values.

The enhanced protective effect observed among patients with Caprini scores ≥10 (the highest-risk stratum) has important implications for clinical implementation. Paradoxically, these are precisely the patients in whom anticoagulation intensification is often precluded by elevated bleeding risk.²⁵ Our finding that TCM provides proportionally greater benefit in this population – with an HR of 0.42 compared with 0.68 in moderate-risk patients – suggests that TCM may fill a therapeutic gap where conventional approaches face limitations. This pattern is biologically plausible; higher-risk patients presumably have more pronounced activation of both coagulation and inflammatory pathways, providing greater substrate for TCM’s dual-action mechanism.²⁶ However, this finding requires confirmation in dedicated studies of high-risk populations, as subgroup analyses are inherently susceptible to spurious findings from multiple comparisons.

Distinguishing symptomatic from ultrasound-detected asymptomatic events is essential when interpreting imaging-based VTE endpoints. In our cohort, 43.8% of all detected VTE cases were symptomatic, and 56.2% were asymptomatic distal DVT identified only through protocol surveillance ultrasonography. Although the relative-risk reductions associated with TCM were numerically similar for the two categories (symptomatic HR: 0.55 vs asymptomatic HR: 0.61, interaction p = 0.62), the clinical implications differ. Symptomatic events drive patient morbidity, healthcare resource use and the long-term risk of post-thrombotic syndrome, whereas asymptomatic distal DVT carries a substantially lower immediate risk but remains clinically relevant because isolated distal thrombosis can be followed by post-thrombotic sequelae and recurrent venous thromboembolism.^27,28 By demonstrating consistent protection across both categories, the present data suggest that the upstream modulation of inflammation–coagulation crosstalk by TCM is operative across the spectrum of subclinical-to-clinically apparent thrombosis rather than reflecting selective detection bias against asymptomatic events. The accompanying 90-day mortality data, although underpowered for between-group hypothesis testing, provide additional safety reassurance that the integrative regimen does not exact a delayed survival cost.

The favourable safety profile of TCM intervention – with no significant increase in major or clinically relevant non-major bleeding and no excess in 90-day all-cause mortality – addresses a common concern about adding agents with putative anticoagulant properties to standard prophylaxis regimens. This finding aligns with the pharmacological profile of the principal TCM components, which exert their antithrombotic effects predominantly through enhancement of fibrinolysis and haemorheological improvement rather than through direct inhibition of plasma coagulation factors, thereby leaving global haemostasis largely intact. The decision-curve analysis extends beyond simple efficacy and safety assessment to quantify the net clinical benefit across a range of clinical scenarios. The demonstration of superior net benefit at threshold probabilities between 4% and 15% indicates that TCM provides value across the spectrum from moderate- to high-risk surgical patients – precisely the population targeted by current prophylaxis guidelines.²⁹

The secondary outcomes observed in this study, including shortened length of stay and reduced postoperative pain, suggest potential broader benefits of TCM beyond VTE prevention. These effects may reflect the anti-inflammatory properties of TCM contributing to overall recovery enhancement, consistent with the principles underlying ERAS protocols.³⁰ Although the cost analysis did not demonstrate statistically significant differences in total hospitalisation costs, the trend towards lower costs in the TCM group, combined with the reduced length of stay, suggests potential economic benefits that warrant formal cost-effectiveness evaluation in future studies.

This study has several methodological strengths that enhance confidence in our findings. The use of IPTW as the primary analytical approach maximises sample utilisation while effectively addressing confounding by indication – a critical concern given that TCM was preferentially administered to higher-risk patients in clinical practice. The calculated E-value of 2.86 indicates reasonable robustness to unmeasured confounding, as any residual confounders would need implausibly strong associations with both treatment and outcome to explain away the observed effect.³¹ The consistency of findings across four distinct analytical approaches – crude, multivariable-adjusted, propensity score-matched and doubly robust estimation – provides further reassurance. The longitudinal biomarker analysis using linear mixed-effects models represents a methodological advancement over prior studies that assessed only static timepoints, enabling characterisation of differential trajectories that illuminate underlying mechanisms.

Several limitations warrant consideration. First, despite a rigorous propensity-score methodology, the retrospective design cannot fully exclude residual confounding from unmeasured variables. Factors such as surgeon preference, patient health-seeking behaviour and subtle differences in perioperative care may have influenced both treatment selection and outcome ascertainment in ways not captured by the available covariates. Second, the study was conducted at a single academic centre with an established integrative-medicine infrastructure, which may limit generalisability to settings without similar expertise in TCM formulation and administration. Third, although marker-compound HPLC quality control limits between-batch variability, the inherent multi-component nature of herbal products introduces residual chemical heterogeneity that cannot be fully characterised. Fourth, although our 90-day mortality and 30-day VTE outcomes capture the period of greatest postoperative thromboembolic and inflammatory risk, longer-term consequences, such as post-thrombotic syndrome and chronic thromboembolic pulmonary hypertension, are not addressed in this analysis. Fifth, although the explicit symptomatic-versus-asymptomatic adjudication mitigates ascertainment artefact, the absolute number of PE events remained small and limited statistical power for this subgroup. Finally, the exploratory biomarker analyses, although hypothesis generating, will require prospective validation with broader inflammatory panels to establish causal mechanistic pathways.

In conclusion, this study provides robust observational evidence that perioperative Buyang Huanwu-based TCM, administered as a standardised adjunct to LMWH prophylaxis, significantly reduces the incidence of both symptomatic and ultrasound-detected asymptomatic postoperative VTE through accelerated resolution of coagulation and inflammatory cascades without increasing major bleeding or 90-day all-cause mortality. The enhanced benefit observed in the highest-risk patients, combined with the favourable net clinical benefit profile, supports integration of standardised TCM into ERAS protocols. Future multicentre prospective randomised trials should confirm these findings and characterise the optimal patient-selection criteria, intervention timing and treatment duration for maximum therapeutic benefit.

Funding Statement

Sichuan Provincial TCM Management Science and Technology Research Project (No. 2024MS455).

Data Sharing Statement

All data generated or analyzed during this study are included in the article.

Ethics Approval and Consent to Participate

This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Mianyang Traditional Chinese Medicine Hospital (No. 2024-093). Due to the retrospective nature of the investigation and the implementation of appropriate data de-identification procedures, informed consent from individuals was waived.

Author Contributions

Xuanfu Peng: Conceptualization, Formal analysis, methodology, software, Supervision, Writing – original draft, Writing – review and editing; Yao Wang: Data curation, Investigation, Methodology, Visualization, Writing – review and editing; Xuelan Yang: Data curation, Investigation, Resources, Writing – review and editing; Weijiang Xiong: Data curation, Investigation, Visualization, Writing – review and editing. All authors gave final approval of the version to be published.

Disclosure

None of the authors have any personal, financial, commercial, or academic conflicts of interest.