Risk factors for postoperative ecchymosis after medial open-wedge high tibial osteotomy: an exploratory analysis of thromboelastography parameters

Abstract

Objective

This study aims to identify risk factors for ecchymosis and its associated coagulation characteristics following medial open-wedge high tibial osteotomy (OWHTO) using thromboelastography (TEG).

Methods

This single-center retrospective study included 220 consecutive patients undergoing unilateral medial OWHTO between January 2023 and September 2025. All patients received rivaroxaban 10 mg orally once daily for 14 days, commencing 12 h postoperatively. The primary endpoint was the incidence of peri-incisional ecchymosis, assessed daily during the hospital stay (postoperative days 1–4). Patients were categorized into ecchymosis (n = 74) and no-ecchymosis groups. Pre-operative and post-operative day-4 TEG parameters, blood loss volumes, and baseline characteristics were analyzed. Multivariable logistic regression (variables selected based on clinical relevance and univariate P < 0.10) determined independent factors, while group discrimination was assessed descriptively using AUC.

Results

Ecchymosis occurred in 33.6% (74/220) of patients. The ecchymosis group demonstrated higher BMI (mean difference 1.57 kg/m², 95% CI 0.81–2.33, P < 0.001), prolonged tourniquet time, greater hidden blood loss, and impaired TEG-CI recovery (ΔTEG-CI: −1.04, 95% CI − 1.48 to − 0.60, P < 0.001). Multivariable analysis identified BMI (OR 1.26, 95% CI 1.11–1.43), tourniquet time (OR 1.03, 95% CI 1.01–1.05), and ΔTEG-CI (OR 0.71, 95% CI 0.56–0.91) as independently associated with ecchymosis (AUC = 0.81, 95% CI 0.75–0.87). Increased hidden blood loss correlated with ecchymosis in affected patients, suggesting a shared bleeding process rather than an independent pathogenic factor.

Conclusions

This exploratory single-center retrospective analysis identified BMI, tourniquet time, and impaired perioperative recovery of the coagulation index as independently associated with ecchymosis following medial OWHTO. These findings detail a post-event coagulation status in patients developing ecchymosis under consistent perioperative anticoagulation; however, they do not support temporal or predictive inferences and remain hypothesis-generating. Prospective multicenter validation is necessary before any clinical utility can be inferred.

Introduction

The global prevalence of knee osteoarthritis (KOA) rose significantly from 256 million cases in 1990 to 607 million in 2021, with age-standardized prevalence rates escalating to 6,967 cases per 100,000 population. This burden is disproportionately borne by older adults, particularly within the 50–69 age group, where both prevalence and disability markedly increase after age 50. Consequently, KOA has emerged as a leading cause of lower-limb disability in adults, imposing considerable health and socioeconomic challenges worldwide [1, 2]. Medial open-wedge high tibial osteotomy (OWHTO) is increasingly recognized as an effective joint-preserving treatment for KOA [3]. However, the potential for perioperative complications has drawn heightened scrutiny, particularly concerning venous thromboembolism (VTE), bleeding-related issues such as wound hematoma, wound healing complications, and infections [4, 5]. Traditional postoperative anticoagulant therapy is employed to prevent VTE, a potentially life-threatening condition, with studies indicating an incidence rate of 13.1% following OWHTO. Nonetheless, this strategy may also elevate the risk of bleeding throughout the perioperative period. Previous research has focused less on the occurrence of wound complications following OWHTO surgery [6, 7]. Garfinkel et al.‘s study demonstrated that rivaroxaban usage is associated with an 18.7% increase in the risk of bleeding and wound complications in total knee arthroplasty [8].

In clinical practice, the direct oral anticoagulant rivaroxaban is commonly prescribed after OWHTO due to its convenient oral administration, rapid onset of action, and proven efficacy in major orthopedic surgeries. However, high-quality evidence specifically supporting the use of rivaroxaban in OWHTO is limited, and its application largely derives from institutional practices extrapolated from total knee arthroplasty guidelines [9]. As the usage of rivaroxaban becomes more prevalent, incidents of postoperative hemorrhage and incisional complications, including ecchymosis, have increased. Ecchymosis is characterized by bluish-purple or dark-red patches on the skin and is the most common bleeding-related complication following orthopedic surgery. When subcutaneous hemorrhage exceeds 3 mm in diameter, ecchymosis is typically accompanied by significant limb swelling and pain, exacerbating local inflammation and indicating an increased risk of additional blood loss [10, 11]. While ecchymosis often resolves on its own, it can have substantial negative psychological effects, increase the risk of surgical incision complications, prolong hospitalization and treatment duration, and delay early functional rehabilitation [12].

Previous studies have focused primarily on ecchymosis following total knee or hip arthroplasty [13]. Compared to total knee arthroplasty, OWHTO presents additional uncertainties regarding anticoagulation management. This procedure involves subperiosteal dissection of the medial collateral ligament, controlled osteotomy gap opening, and preservation of the lateral cortical hinge, resulting in unique bleeding characteristics that may respond differently to Xa factor inhibition. Traditionally, when bleeding complications such as ecchymosis occur, clinicians have relied on conventional coagulation assays to guide anticoagulant management. However, these tests may lack the sensitivity needed to detect subtle coagulation changes, particularly alterations in factor levels, following high tibial osteotomy (HTO). Thromboelastography (TEG) has been increasingly utilized in major orthopedic surgeries, including total knee arthroplasty, to comprehensively evaluate coagulation status and predict bleeding events [13, 14]. This study applied TEG to monitor perioperative hemostasis in OWHTO patients receiving rivaroxaban, hypothesizing that TEG-derived parameters might help identify individuals at heightened risk of postoperative ecchymosis and address the critical evidence gap regarding systemic coagulation status in OWHTO patients experiencing ecchymosis, thus guiding specific anticoagulation management. It was hypothesized that (1) BMI, hidden blood loss, and operative factors would serve as independent risk factors for ecchymosis, and (2) patients who developed ecchymosis would exhibit a hypocoagulable TEG profile compared to those without ecchymosis, suggesting that TEG monitoring may assist in identifying individuals at increased bleeding risk and enhance personalized anticoagulation management.

Materials and methods

Patient selection and study design

A retrospective review was conducted on consecutive patients diagnosed with KOA who underwent unilateral HTO between January 2023 and September 2025. A total of 220 patients meeting the inclusion criteria and possessing complete perioperative records were enrolled, with no eligible patients excluded due to missing data (Fig. 1). The inclusion criteria were as follows: (1) isolated medial compartment KOA confirmed by weight-bearing X-rays indicating significant narrowing of the medial joint space without notable patellofemoral involvement; (2) unilateral HTO performed at the Affiliated Hospital of Qingdao University during the study period, strictly defined as a one-sided procedure on a single lower extremity with the contralateral limb being unaffected; and (3) written informed consent obtained from all participants. Exclusion criteria included: (1) a history of hematological disorders, previous vascular surgery, VTE, or prior anticoagulant therapy; (2) pre-existing coagulopathy, contraindications to anticoagulation, or active anticoagulant usage that could not be discontinued; (3) simultaneous or staged bilateral HTO; (4) severe hepatic or renal insufficiency; (5) hypersensitivity to rivaroxaban or any excipient; (6) pregnancy or lactation; and (7) ongoing antiplatelet therapy, active inflammatory disease requiring systemic steroids, or revision HTO procedures. Patients with mild-to-moderate hepatic dysfunction or renal impairment were included but identified for sensitivity analysis. Baseline anemia was recorded and adjusted for in the multivariate analysis.

Fig. 1.

Flow diagram of patient selection

The study protocol received approval from the Ethics Committee of the Affiliated Hospital of Qingdao University (approval No. QYFYWZLL29739).

Surgical technique

All surgeries were performed by the same surgical team, consisting of two senior orthopedic surgeons, following a standardized medial open-wedge HTO technique. Under general anesthesia and in a supine position, a thigh tourniquet was inflated to 300 mmHg (1.5 × systolic blood pressure, with a minimum of 250 mmHg) after limb exsanguination. An 8-cm medial longitudinal incision was made over the proximal tibia, and the skin and subcutaneous tissues were dissected to expose the pes anserinus. Diagnostic knee arthroscopy was not routinely performed prior to the osteotomy; instead, preoperative magnetic resonance imaging was utilized to evaluate the knee joint. A sub-periosteal elevator was inserted beneath the superficial medial collateral ligament to elevate the soft tissues until the posterior tibial cortex was visualized. The proximal osteotomy cut was made parallel to the medial tibial plateau joint line, beginning 35–40 mm distal to the medial joint line while preserving a lateral hinge of 5–8 mm medial to the lateral cortex. Two Kirschner wires were inserted obliquely above the pes anserinus for guidance, and an oscillating saw was employed to perform a bi-cortical osteotomy. Following a gradual opening to achieve the preplanned correction (target mechanical axis of 3°–5° valgus, 62%–66% weight-bearing line), a TomoFix locking plate (DePuy Synthes, Zuchwil, Switzerland; standard titanium plate with a 4.5-mm locking screw system) was applied. The osteotomy gap was filled with synthetic β-tricalcium phosphate (ChronOS, DePuy Synthes, USA), and screw lengths were measured and verified radiographically. Drainage tubes were routinely placed after each surgery. The wound was thoroughly irrigated and closed in layers, followed by a compressive dressing. Tourniquet time and intra-operative blood loss were monitored by the anesthesiologist; no patient required intra-operative transfusion.

Peri-operative anticoagulation and coagulation monitoring protocol

All patients received standardized perioperative care under general anesthesia, with no intra-articular tranexamic acid (TXA) administered. Postoperatively, rivaroxaban 10 mg was given 12 h after surgery and continued once daily for 14 days. This regimen was maintained for all patients, including those who developed peri-incisional ecchymosis, as anticoagulation was not interrupted for this non-life-threatening bleeding manifestation. Mechanical thromboprophylaxis included graduated compression stockings (20–30 mmHg) applied immediately postoperatively and intermittent pneumatic compression devices used throughout the hospitalization period. Patients were encouraged to perform isometric quadriceps contractions and ankle pumps immediately after surgery, followed by the initiation of continuous passive motion of the knee starting on postoperative day 1. Early mobilization with non-weight-bearing crutch ambulation began on the day of surgery. Non-steroidal anti-inflammatory drugs (NSAIDs) were avoided for 48 h postoperatively to minimize bleeding risk, after which selective COX-2 inhibitors were allowed for analgesia. Systemic corticosteroids were not administered.

Ecchymosis was assessed daily during the hospital stay (postoperative days 1–4) by two independent, trained research nurses who were blinded to group allocation. Assessments were conducted at the bedside under standardized lighting conditions. Patients were classified into “ecchymosis” or “no-ecchymosis” groups based on the presence of visible subcutaneous blood extravasation as an isolated finding, without accompanying pain, swelling, or restricted knee movement that would indicate a wound hematoma or active bleeding complication. In the ecchymosis group, the maximum extent of discoloration was documented and quantified using the patient’s palm (approximately 1% total body surface area) as a reference. Severity was graded using a validated 6-point mottling score (0–5), assessing the extent of discoloration from the center of the knee to the proximal thigh: Grade 0 (none), Grade 1 (coin-sized, knee center), Grade 2 (superior patellar edge), Grade 3 (mid-thigh), Grade 4 (groin fold), and Grade 5 (beyond groin). Both assessors underwent standardized training in the palm-based quantification method and the mottling score prior to the study commencement. Inter-rater reliability was found to be excellent (Cohen’s kappa = 0.92) between the two independent assessors in a randomized subset of 30 patients. Standardized digital photographs were obtained using a fixed camera position and consistent lighting for documentation and retrospective verification; however, automated digital image analysis software was not utilized.

Routine blood tests, including conventional coagulation assays and TEG, were performed the day before surgery and on postoperative day 4. Postoperative TEG was conducted on day 4 to align with previous total knee arthroplasty studies employing TEG for postoperative coagulation assessment. However, this timing might miss the early phase when ecchymosis typically develops (usually on postoperative days 2–3), thereby serving as a post-event measurement rather than a pre-event predictor. TEG was executed using a TEG 5000 Thrombelastograph Hemostasis Analyzer (Haemonetics Corporation, Braintree, MA, USA) with kaolin-activated citrated whole blood samples. Blood samples were collected in 3.2% sodium citrate tubes (Becton Dickinson, USA) and processed within 30 min. The following TEG variables were recorded: R time (reaction time), K time (clot formation time), α-angle, MA (maximum amplitude), LY30 (lysis at 30 min), and the comprehensive coagulation index (CI). Total blood loss was estimated using the Nadler formula: Blood volume (mL) = k₁ × height(m)³ + k₂ × weight(kg) + k₃, with specific constants for men (k₁ = 0.3669, k₂ = 0.03219, k₃ = 0.6041) and women (k₁ = 0.3561, k₂ = 0.03308, k₃ = 0.1833). The total blood loss was calculated as blood volume multiplied by (pre-operative Hct − post-operative Hct), and hidden blood loss was derived by subtracting intra-operative visible loss from this total. All patients were followed for one month after surgery.

Statistical analysis

All analyses were conducted using R software (version 4.5.1). Normally distributed data are presented as mean ± SD and were compared using the independent-samples t-test, while non-normally distributed data are presented as median [IQR] and were compared using the Mann-Whitney U test. This study was designed as an exploratory analysis to investigate associations between perioperative coagulation profiles and the occurrence of ecchymosis, rather than to establish a formal prediction model.

It is important to acknowledge the conceptual overlap between hidden blood loss and ecchymosis, as both quantify the same underlying bleeding process. Consequently, hidden blood loss was excluded from the primary multivariable model and analyzed only in a sensitivity analysis, with results interpreted as correlated manifestations rather than independent determinants. The primary multivariable model included body mass index (BMI), tourniquet time, and perioperative changes in CI (ΔTEG-CI) as independent variables.

Multivariate logistic regression was employed to identify independent risk factors for ecchymosis, adjusting for age, sex, BMI, mild-to-moderate hepatic/renal dysfunction, and operative time. Variance inflation factors (VIF) were calculated for all covariates in the multivariable model to assess multicollinearity, with all VIF values remaining below 2.5, indicating no substantial multicollinearity. Sensitivity analyses were performed by excluding: (1) patients with mild-to-moderate organ dysfunction, and (2) patients with baseline anemia (hemoglobin < 120 g/L for women, < 130 g/L for men) to evaluate the robustness of the findings. Receiver operating characteristic (ROC) curves were constructed, and the area under the curve (AUC) was calculated to assess the discriminative ability of identified risk factors in distinguishing between patients with and without ecchymosis. This AUC reflects group-level discrimination within this single-center retrospective dataset and should not be interpreted as indicative of predictive performance, generalizability, or external validity. No formal sample size calculation was performed, as this retrospective study utilized all consecutive patients meeting the inclusion criteria during the study period. With 74 ecchymosis events and 5 covariates in the multivariate model, the events-per-variable ratio was approximately 15:1, surpassing the commonly recommended minimum of 10 events per variable.

Results

Patient characteristics

Between January 2023 and September 2025, 220 patients who underwent unilateral HTO met the inclusion criteria and were enrolled in the study. Among these patients, 74 presented with ecchymosis, exhibiting a median mottling score (a validated 6-point scale) of 2 (IQR 2–3, range 1–5), with classifications as follows: 18.9% in Grade 1, 35.1% in Grade 2, 31.1% in Grade 3, 10.8% in Grade 4, and 4.1% in Grade 5. Consequently, patients were categorized into an ecchymosis group (n = 74) and a no-ecchymosis group (n = 146). The only significant difference observed between the groups was in BMI, with a mean difference of 1.57 kg m⁻² (95% CI 0.81–2.33, P < 0.001) (Table 1).

Table 1.

Comparison of baseline characteristics between the ecchymosis and non-ecchymosis groups

Variables	No-ecchymosis group	Ecchymosis group	t/χ²	P
n	146	74
Age(year)	55.48 ± 4.99	55.82 ± 4.96	t=-0.49	0.628
Gender(male), n(%)	64 (43.84)	31 (41.89)	χ²=0.08	0.783
KL Grade, n(%)			χ²=0.07	0.787
II	12	6
III	90	47
IV	44	21
BMI (kg/m²)	25.00 ± 2.56	26.57 ± 2.75	t=-4.19	<0.001

Data are presented as mean ± standard deviation or number (percentage)

BMI Body Mass Index, KL Grade Kellgren-Lawrence Osteoarthritis Grade

Bold values indicate statistically significant differences

Pre-operative coagulation comparison

Pre-operative TEG and conventional coagulation tests revealed no significant differences between the two groups, indicating that baseline coagulation status was comparable among all patients before surgery (Table 2).

Table 2.

Comparison of preoperative TEG parameters and conventional coagulation tests

Variables	No-ecchymosis group	Ecchymosis group	t	P
Preoperative
R(min)	4.95 ± 1.11	4.83 ± 1.11	t = 0.75	0.451
K(min)	1.72 ± 0.39	1.69 ± 0.45	t = 0.52	0.606
Angle(°)	67.76 ± 3.74	68.28 ± 3.67	t=-0.98	0.329
MA(mm)	60.89 ± 3.34	60.87 ± 3.47	t = 0.05	0.962
TEG-CI	0.96 ± 0.93	1.19 ± 0.82	t=-1.77	0.078
LY30(%)	1.18 ± 0.78	1.08 ± 0.65	t = 0.88	0.381
EPL(%)	1.18 ± 0.78	1.08 ± 0.64	t = 0.94	0.350
G(dynes/cm²)	7930.36 ± 1043.76	7837.94 ± 971.91	t = 0.63	0.526
A (mm)	58.83 ± 3.08	58.53 ± 2.39	t = 0.80	0.427
PT	11.40 ± 0.88	11.50 ± 0.83	t=-0.81	0.416
APTT	30.79 ± 2.14	31.10 ± 2.00	t=-1.04	0.300
TT	18.50 ± 0.75	18.48 ± 0.56	t = 0.26	0.798

Data are presented as mean ± standard deviation

TEG Thromboelastography, R Reaction time, K Clot formation time, MA Maximum amplitude, CI Coagulation index, LY30 Lysis at 30 min, EPL Estimated percent lysis, G Shear elastic modulus, A Ampli

Post-operative coagulation comparison and peri-operative changes

TEG performed on postoperative day 4 demonstrated a pronounced hypocoagulable profile in the ecchymosis group. Analysis of perioperative changes (Δ = post-operative − pre-operative) supported these findings, showing that the CI increased less in the ecchymosis group (ΔTEG-CI: −1.04, 95% CI − 1.48 to − 0.60, P < 0.001) (Table 3, Table 4). Since TEG was conducted after the typical onset of ecchymosis (postoperative days 2–3), this observation reflects a post-event coagulation state rather than a pre-event predictor.

Table 3.

Postoperative TEG parameters and conventional coagulation tests

Variables	No-ecchymosis group	Ecchymosis group	t	P
Postoperative
Post R(min)	4.03 ± 0.67	4.28 ± 0.55	t=-2.76	0.006
Post K(min)	1.22 ± 0.32	1.33 ± 0.25	t=-0.71	0.479
Post Angle(°)	72.20 ± 3.92	71.30 ± 2.84	t = 1.94	0.054
Post MA(mm)	62.48 ± 4.48	62.66 ± 4.21	t=-0.30	0.767
Post TEG-CI	2.62 ± 1.29	1.81 ± 0.86	t = 3.99	< 0.001
Post LY30(%)	1.31 ± 1.17	1.57 ± 1.31	t=-1.51	0.133
Post EPL(%)	1.30 ± 1.18	1.58 ± 1.31	t=-1.61	0.110
Post G(dynes/cm²)	8508.71 ± 1557.57	8538.44 ± 1567.15	t=-0.13	0.894
Post A(mm)	59.90 ± 5.36	59.61 ± 5.49	t = 0.38	0.706
Post PT	12.51 ± 1.90	12.75 ± 1.87	t=-0.84	0.405
Post APTT	30.40 ± 6.60	33.05 ± 9.83	t=-0.23	0.818
Post TT	17.68 ± 0.82	17.52 ± 0.73	t = 1.40	0.163

Data are presented as mean ± standard deviation. TEG parameters and abbreviations as defined in Table 2

Bold values indicate statistically significant differences

Table 4.

Perioperative changes (Δ) in TEG parameters

Variables	No-ecchymosis group	Ecchymosis group	t	P
Value change
∆ R(min)	-0.92 ± 1.37	-0.55 ± 1.19	t=-1.98	0.049
∆ K(min)	-0.43 ± 0.43	-0.36 ± 0.35	t=-1.08	0.282
∆ Angle(°)	4.44 ± 5.65	3.02 ± 4.80	t = 1.85	0.066
∆TEG-Cl	1.66 ± 1.64	0.62 ± 1.15	t = 5.45	< 0.001

Data are presented as mean ± standard deviation. Δ represents the change from preoperative to postoperative day 4 (Postoperative value - Preoperative value).Bold values indicate statistically significant differences. TEG parameters and abbreviations as defined in Table 2

Surgical parameters and blood loss

Operative time and intraoperative blood loss did not differ significantly between the ecchymosis and no-ecchymosis groups. In contrast, key indices reflecting postoperative bleeding exhibited substantial differences. The ecchymosis group experienced an average hidden blood loss of 439.79 ± 152.90 mL, which was significantly higher than the 345.59 ± 129.43 mL observed in the no-ecchymosis group (P < 0.001). Total blood loss was also significantly greater in the ecchymosis group. Additionally, tourniquet time was longer for patients who developed ecchymosis (Table 5) (Fig. 1).

Table 5.

Comparison of operative parameters and blood loss

Variables	No-ecchymosis group	Ecchymosis group	t	P
Operative time(min)	62.86 ± 11.51	62.20 ± 11.01	t = 0.40	0.687
Intraop blood loss(ml)	43.65 ± 12.79	44.02 ± 17.27	t=-0.18	0.859
Hidden blood loss(ml)	345.59 ± 129.43	439.79 ± 152.90	t=-4.54	< 0.001
Tourniquet time(min)	43.12 ± 20.36	53.08 ± 12.26	t=-4.51	< 0.001
Total blood loss(ml)	370.73 ± 117.52	584.64 ± 139.63	t=-11.96	< 0.001

Intraop: Intraoperative. Data are presented as mean ± standard deviation

Bold values indicate statistically significant differences

Multivariable association analysis for post-HTO ecchymosis

Univariable and multivariable logistic regression analyses were employed to quantify the associations between baseline characteristics, surgical variables, coagulation indices, and postoperative ecchymosis. Univariable screening (Fig. 2A) identified eight variables significantly associated with ecchymosis (P < 0.05). Notable risk factors included higher BMI, greater total blood loss, increased intraoperative blood loss, and prolonged tourniquet time. Among coagulation indices, a longer postoperative R time correlated with a higher occurrence of ecchymosis, whereas a lower postoperative TEG-CI and a smaller perioperative increase in TEG-CI (ΔTEG-CI) were associated with a lower occurrence (OR 0.62, 95% CI 0.50–0.76).

Fig. 2.

Univariable screening and sensitivity analysis of factors associated with postoperative ecchymosis. A Univariable logistic regression analysis of baseline characteristics, surgical variables, and coagulation parameters. Odds ratios (ORs) with 95% confidence intervals (CIs) are presented for each variable. All continuous variables are scaled per unit increase. B Sensitivity analysis multivariable model. The primary multivariable model included body mass index (BMI), tourniquet time, and perioperative change in coagulation index (ΔTEG-CI) as independent variables. This panel presents a sensitivity analysis that additionally includes hidden blood loss; given the conceptual overlap between hidden blood loss and ecchymosis (both reflect the same underlying bleeding process), hidden blood loss is presented as a correlated bleeding parameter rather than an independent determinant. The odds ratio for hidden blood loss is scaled per 100 mL increase; all other ORs are scaled per unit increase for continuous variables (BMI: per 1 kg/m²; tourniquet time: per 1 min; ΔTEG-CI: per 1 unit)

In the multivariable model, after adjustments, significant independent factors associated with postoperative ecchymosis included BMI (OR 1.26, 95% CI 1.11–1.43, P < 0.001), tourniquet time (OR 1.03, 95% CI 1.01–1.05, P = 0.005), and ΔTEG-CI (OR 0.71, 95% CI 0.56–0.91). Although hidden blood loss was strongly correlated with ecchymosis occurrence (OR 1.10 per 100 mL increase, 95% CI 1.05–1.15, P < 0.001), it should be interpreted as a correlated manifestation rather than an independent determinant due to the conceptual overlap between these two variables, both reflecting the same underlying bleeding process. A sensitivity analysis, which excluded hidden blood loss from the model, confirmed that BMI, tourniquet time, and ΔTEG-CI remained significantly associated with ecchymosis, reinforcing the robustness of these independent associations. No substantial multicollinearity was detected among the independent variables (all VIF < 2.5) (Fig. 2B).

The combination of BMI, tourniquet time, and ΔTEG-CI demonstrated superior group discrimination (AUC = 0.81, 95% CI 0.75–0.87) compared to any single factor alone (ΔTEG-CI: AUC = 0.68; BMI: AUC = 0.64; tourniquet time: AUC = 0.63) in this exploratory analysis. This AUC reflects group-level discrimination within this retrospective dataset and should not be interpreted as indicative of predictive performance or clinical utility in an external population (Fig. 3).

Fig. 3.

Receiver-operating-characteristic curves showing exploratory group discrimination for postoperative ecchymosis by individual predictors and their combination; AUC values reflect within-dataset discrimination only and should not be interpreted as predictive performance

Discussion

In this study involving 220 patients who underwent unilateral HTO with rivaroxaban thromboprophylaxis, 74 patients (33.6%) developed peri-incisional ecchymosis within 2–3 days postoperatively. This cohort exhibited a hypocoagulable phenotype characterized by impaired recovery of CI on TEG, along with increased hidden blood loss, prolonged tourniquet application, and elevated BMI. As Enhanced Recovery After Surgery (ERAS) programs continue to expand, achieving an optimal perioperative “anticoagulation–hemostasis” balance has become a pivotal concern in orthopedic care [15]. Most mechanistic research on ecchymosis has focused narrowly on total knee arthroplasty [14, 16, 17], highlighting a significant gap in the development of a systematic coagulation assessment and risk-prediction tool tailored for HTO. This study utilized TEG to elucidate the postoperative coagulation status in patients who developed ecchymosis and identified factors independently associated with ecchymosis.

Traditionally, ecchymosis has been perceived as a trivial manifestation of incisional ooze or capillary fragility. Recent data, however, suggest that it may reflect systemic coagulation characteristics [18]. In the present cohort, hidden blood loss was 27% greater in ecchymosis patients, and postoperative TEG-CI was significantly lower, indicating that ecchymosis serves as a signal of systemic hypocoagulation rather than merely local seepage.

Conventional coagulation assays have limited sensitivity to factor Xa inhibitors and may fail to detect subtle alterations in global coagulation status [19]. TEG offers viscoelastic measurements that provide a comprehensive assessment of coagulation function, encompassing the interactions between procoagulant factors, platelet function, and fibrinolytic activity [20]. This exploratory analysis revealed a significant association between ΔTEG-CI and ecchymosis status (OR 0.71, 95% CI 0.56–0.91), with effect sizes comparable to established clinical factors such as BMI (OR 1.26) and tourniquet time (OR 1.03). However, because TEG was performed on POD4—after the typical onset of ecchymosis (POD2–3)—ΔTEG-CI reflects a post-event coagulation characteristic rather than a pre-event risk factor or predictor. Consequently, the incremental discriminative value of TEG beyond conventional clinical variables remains unclear. While the combination of all factors yielded an AUC of 0.81, TEG alone (ΔTEG-CI) exhibited modest discriminative ability (AUC 0.68). Determining whether TEG monitoring could inform future prospective studies necessitates serial perioperative measurements and rigorous methodological design.

Hidden blood loss, traditionally associated with anemia or extended hospitalization, was notably greater in patients with ecchymosis. Micro-arterial bleeding from the osteotomy surface and periosteal stripping led to continuous seepage into the fascial–subcutaneous plane, where blood not visible in the surgical field contributed to visible bruising. Conceptually, ecchymosis represents the observable manifestation of the same bleeding process quantified as hidden blood loss; thus, hidden blood loss reflects an underlying pathophysiology and should be interpreted as a correlated parameter rather than as an independent causal determinant.

The mean BMI in the ecchymosis group was 26.6 kg/m², compared to 25.0 kg/m² in the non-ecchymosis group. Although this difference was statistically significant, the modest magnitude (1.6 kg/m²) suggests that BMI may be one of several interrelated factors contributing to ecchymosis occurrence rather than a singular or dominant determinant. Additionally, longer tourniquet times were associated with ecchymosis; however, this relationship may reflect greater surgical complexity—such as more extensive soft-tissue dissection or increased osteotomy correction—rather than a direct causal mechanism whereby prolonged tourniquet duration directly promotes bruising. These observations collectively highlight the multifactorial nature of postoperative bleeding manifestations, where patient characteristics and operative factors may interact in intricate ways. Further research is necessary to investigate whether integrated perioperative management strategies can mitigate the risk of ecchymosis in clinical practice. Given the associative nature of this study, which included multiple exploratory comparisons without adjustments for multiplicity, the findings are susceptible to type I error and should be viewed as hypothesis-generating. Prospective validation is essential before clinical implementation.

Another methodological consideration pertains to the timing of TEG measurements. While rivaroxaban exposure was consistent across both groups on postoperative day 4, and all patients continued the standardized regimen regardless of ecchymosis status, the TEG conducted on POD 4 remains an inherently post-event measurement. Since ecchymosis typically manifests on POD 2–3, the observed TEG parameters represent the coagulation state after bruising has occurred rather than pre-event physiology that might have predicted it. Therefore, the lack of temporal precedence restricts any predictive inferences. Our findings primarily describe a post-event coagulation status without temporal implications, rather than defining an intrinsic pre-event risk profile.

The study has several limitations. It was a single-center, retrospective analysis involving patients on a single anticoagulant regimen, which limits generalizability. The retrospective design introduces potential selection bias in patient inclusion and recall bias in data documentation; for example, the degree of mottling may have influenced the completeness of documentation, and TEG testing timing could have been affected by clinical concerns rather than strict adherence to protocol. TEG sampling was limited to the day before surgery and postoperative day 4, which may have overlooked early dynamic changes in coagulation. Furthermore, the study did not account for unmeasured confounders such as variations in surgeon-specific techniques or postoperative activity levels that could affect bleeding outcomes. As an associative study, it identifies risk factors significantly linked to ecchymosis but does not establish a validated predictive tool. Prospective multicenter studies with more frequent TEG time-point assessments, drug-level monitoring, and standardized data collection are needed to validate these associations. The palm-based quantification (~ 1% BSA) and mottling score assessment should undergo further clinical validation, and future studies should incorporate objective quantification methods, such as digital image analysis.

Conclusion

This exploratory study identifies BMI, tourniquet time, and impaired perioperative recovery of CI as independently associated with ecchymosis following HTO. Patients who developed ecchymosis exhibited a hypocoagulable phenotype characterized by attenuated recovery of clotting function, representing a post-event coagulation characteristic rather than a pre-event predictor. These findings emphasize the multifactorial nature of postoperative bruising and may aid in generating hypotheses for future prospective studies. However, as a single-center retrospective analysis without adjustments for multiplicity or external validation, causal inferences are precluded, and the findings remain exploratory. Thus, prospective multicenter studies with serial coagulation monitoring, objective outcome quantification, and rigorous validation are warranted to confirm these associations and establish clinical utility.

Authors’ contributions

SWL and KSX conceived and designed the study; JY, LJY and LH drafted the manuscript; WTR and YN collected and analyzed the data; ZYZ and ZL revised the manuscript; ZX and CJL approved the final version.

Funding

This study received no funding from any public or commercial sources.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the medical ethics committee of the Affiliated Hospital of Qingdao University according to the Declaration of Helsinki, and informed consent was obtained from all individual participants included in the study. All methods were carried out in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.