Study design and patients
In this retrospective cohort study, consecutive patients with endovascular interventions were identified by reviewing the databases from the large urban medical center, Nanjing Drum Tower Hospital, from January 1, 2019 to December 31, 2022. This study was conducted in accordance with the Declaration of Helsinki. Approval for the study was obtained from the Ethics Committee of Nanjing Drum Tower Hospital (2023-117-02), and the requirement for informed consent was waived due to the retrospective nature of the study. All methods performed in accordance with the relevant guidelines and regulations.
Specific endovascular interventions referred to the procedures of percutaneous puncture under the guidance of imaging equipment, in which instruments such as puncture needles, special catheters and guide wires were introduced into the blood vessels for therapeutic purposes. The procedures (cerebrovascular interventions, aortic aneurysm repair, peripheral artery disease interventions, splanchnic artery interventions, etc.) were classified according to the international classification of diseases (ICD-10), as detailed in Supplementary Table Ⅰ.
Exclusion criteria include: (1) Patients who discontinued antithrombotic medications before the procedures. (2) Patients receiving both antiplatelet and anticoagulant medications simultaneously. (3) Percutaneous coronary interventions, typically preceded by a loading dose of antiplatelet medications such as aspirin, clopidogrel and ticagrelor. (4) Endovascular interventions for angiography. (5) Endovascular interventions performed via venous access. (6) Open surgeries and endovascular interventions performed simultaneously. (7) Endovascular interventions involving the use of thrombolytic medications.
Patients who had been taking antithrombotic medications did not discontinued prior to the procedures. For the procedures, briefly, the puncture site was identified and an incision was made to expose the vessel. Systemic heparinization was performed with heparin and the activated clotting time (ACT) value was monitored to achieve about two times the normal value. Then the arterial sheaths were inserted by puncture in the target vessel.
The antithrombotic medications were carefully reviewed and documented. According to the preoperative antithrombotic medications, patients were classified into four groups: (1) those not receiving any antithrombotic therapy (Non-ATT). (2) those receiving single antiplatelet therapy (SAPT), such as aspirin, clopidogrel, etc.. (3) those receiving dual antiplatelet therapy (DAPT), such as aspirin and clopidogrel, aspirin and ticagrelor, etc.. (4) those receiving anticoagulant therapy (ACT), such as warfarin, rivaroxaban, dabigatran, etc..
Covariates
The demographic characteristics (age, gender, BMI), comorbidities (hypertension, diabetes, vascular disease, immune disease, congestive heart failure), vascular access, number of puncture sites, sheath size, anesthesia type, duration of procedure and laboratory indexes were used to describe the baseline characteristics of study population. Immune disease mainly referred to systemic vasculitis that involved the aorta and its branch arteries, which led to wall thickening, stenosis or aneurysm of involved arteries, such as Takayasu arteritis9. Vascular disease, a systemic condition with atherosclerosis as the common pathological feature, can manifest as coronary artery disease, cerebrovascular disease or peripheral artery disease, etc., or a combination of two or more vascular bed diseases. Renal function was estimated by calculating estimated glomerular filtration rate (eGFR) using Cockcroft-Gault formula, with eGFR categorized into five grades: ≥90 mL/min/1.73m2, 60-89 mL/min/1.73m2, 30-59 mL/min/1.73m2, 15-29 mL/min/1.73m2 and <15 mL/min/1.73m2. The vascular access for endovascular intervention included femoral artery, popliteal artery, tibiofibular artery, radial artery, brachial artery, axillary artery and carotid artery. The choice of vascular access depended on various factors, including the nature of the procedure, patient anatomy and condition, physician expertise and considerations for minimizing complications. The introducer sheath was a thin, flexible tube designed to facilitate the insertion of medical instruments or devices into the body, serving as a conduit through which other tools, such as guide wire or catheter, can be safely guided into the target area. The size of an introducer sheath was determined by its diameter (French size). Anesthesia types were categorized into local and general anesthesia. Local anesthesia involved numbing a specific area to block sensation and pain during a medical procedure. General anesthesia induced a state of unconsciousness and unresponsiveness to pain during surgery or other invasive procedures.
Study outcomes
The primary efficacy outcome was transfusion of whole blood or red blood cells. Indications for transfusion were determined by the attending physicians, with institutional guidelines recommended transfusion at a serum hemoglobin < 8 g/L10. The secondary efficacy outcomes consisted of postoperative hematoma and pseudoaneurysm at the puncture site. Hematoma, typically induced by trauma, results from the rupture of blood vessels and subsequent blood leakage into adjacent tissues, forming a coagulated mass whose size varies with the severity of the trauma and extent of hematoma. Pseudoaneurysm, in contrast to true aneurysms which involve permanent dilation of the vessel wall, arises from a breach in the vessel wall, causing blood to extravasate and form a sac-like structure composed of fibrous tissue and thrombosis. Clinical presentations may include pain, a pulsatile mass, or symptoms of compression on adjacent structures.
The clinical outcomes used to analyze the prognostic impact of blood transfusion were one-year all-cause mortality, one-year adverse cardiovascular events and three-month infectious events. All-cause mortality was defined as death from any cause. Adverse cardiovascular events encompassed a composite of myocardial infarction, ischemic stroke, or death from cardiovascular causes. Infectious events were classified as either: (1) microbiologically defined infections (patients with a clinical syndrome and compatible microbiology result for an infection); (2) clinically defined infections (patients with a clinical syndrome consistent with an infectious origin or etiology, but with no compatible microbiological results).
Statistical Analysis
In the cohort of 5743 patients, 802 (13.96%) patients had missing data in at least one of eGFR, platelet, hemoglobin and fibrinogen. Missing data was imputed using multiple imputation by chained equations in R statistical software, resulting in multiple imputed datasets with pooled odds ratios (OR) values. The outcomes between groups were analyzed using multiple logistic regression, with the Non-ATT group serving as the reference, which were reported as OR values with corresponding 95% confidence intervals (CI). Complete data was utilized for subsequent sensitivity analysis. P-value < 0.05 was considered significant.
The risk of blood transfusion in specified subgroups was defined by age (<70 years and ≥70 years), vascular disease, anesthesia type, numbers of puncture sites, sheath size (<7 F or ≥7 F) and renal function (eGFR ≥90 mL/min/1.73m2, eGFR 30-89 mL/min/1.73m2 and eGFR <30 mL/min/1.73m2). For subgroup analysis, multivariable Cox proportional hazards regression was used to evaluate the clinical outcomes. The significance of interaction between antithrombotic therapies and subgroups was defined as P-for-interaction < 0.05.
To analyze the impact of blood transfusion on one-year all-cause mortality, adverse cardiovascular events and three-month infectious events, the methods of multiple imputation and inverse probability weighting were performed. The propensity score method, which simulated the effect of a randomized clinical trial in observational cohort study was used to estimate the clinical outcomes between patients who received blood transfusion and those who did not. The propensity score was calculated using logistic regression to estimate the probability of receiving antithrombotic medications based on observed covariates. Then stabilized inverse probability of treatment weights (IPTW) derived from the propensity scores was used to adjust for measured covariates, thereby creating a pseudo dataset by preserving sample size. To assess the performance before and after stabilized IPTW, we compared the covariates using standardized mean difference (SMD), with differences > 10% regarded as imbalanced. Weighted incidence rate (IR) per 100 person-years (PY) was calculated. The risk of clinical outcomes was analyzed by survival analysis with Kaplan-Meier method and the log-rank test for univariate analysis or Cox proportional hazard regression models for multivariate analysis, expressed as hazard ratio (HR) and 95% CI.