Postoperative Delerium Associated with Autologous vs Allogeneic Blood Transfusion in Elderly Hip Arthroplasty Patients Undergoing Combined Spinal-Epidural Anesthesia: A Single-Center Retrospective Study

Introduction

Postoperative delirium (POD) is a severe complication characterized by acute confusion, inattention, and fluctuating cognition, typically occurring within the first 5 days after surgery, usually within 24–48 hours in elderly patients [1,2]. With an incidence of 10–30% in major orthopedic surgeries, POD is associated with prolonged hospitalization, delayed functional recovery, increased healthcare costs, long-term cognitive decline, and even higher 1-year mortality [3,4]. As the global population ages and hip arthroplasty (HA) becomes more prevalent, the proportion of elderly patients undergoing HA for conditions such as hip fractures, osteoarthritis, and femoral head necrosis has risen significantly [5]. However, HA often involves substantial blood loss and transfusion requirements, making perioperative blood management a critical factor in reducing POD risk and improving outcomes in elderly patients.

Although the pathophysiology of POD remains incompletely understood, advanced age, preoperative cognitive impairment, and intraoperative blood transfusion have been identified as key risk factors [6,7]. Notably, patients with HA frequently require blood transfusions due to extensive surgical trauma, hemostatic challenges, and poor bone quality. These factors collectively make transfusion strategy a pivotal aspect of perioperative care [8]. The impact of transfusion on POD remains debated. Allogeneic transfusions can increase POD risk by triggering immune responses, microemboli formation, and neuroinflammation [9,10]. In contrast, autologous transfusion – which minimizes immunomodulation and transfusion-related complications – has garnered increasing attention, with growing evidence suggesting its potential to reduce POD [11]. Nevertheless, systematic comparisons between autologous and allogeneic transfusion strategies in HA patients, particularly regarding their differential effects on POD risk, remain scarce. This study retrospectively evaluated the association between intraoperative autologous/allogeneic transfusion and POD incidence in HA patients, aiming to optimize evidence-based transfusion strategies and identify effective interventions to reduce POD in elderly HA populations.

Material and Methods

STUDY POPULATION:

This retrospective cohort study was approved by the Ethics Committee of Taihe Hospital (Shiyan, China) with a waiver of informed consent (Reference Number: WZ2025036). We screened all elderly patients (≥65 years) who underwent primary or revision HA at our orthopedic center between January 2018 and April 2025, using the hospital’s HT-EMRS electronic medical record system. Two independent researchers extracted data using standardized case report forms (SCRFs), with discrepancies resolved by a third investigator. Inclusion criteria were: 1) Age ≥65 years; 2) Scheduled unilateral HA; 3) Complete transfusion records and POD assessment data. Exclusion criteria were: 1) Preoperative delirium or psychiatric disorders; 2) Preoperative allogeneic blood transfusion or coagulation dysfunction; 3) Emergency surgeries or missing key variables (incomplete collection of height, weight, smoking, and drinking history in emergency patients; missing records of intraoperative blood loss; incomplete data acquisition for certain special cases); and 4) General anesthesia. In our center, combined spinal-epidural anesthesia (CSEA) is the preferred anesthesia method for hip replacement surgery, with general anesthesia reserved for patients with contraindications to regional anesthesia. Due to the low proportion of patients receiving general anesthesia (5.6%), the sample size was insufficient for separate analysis. Therefore, patients receiving general anesthesia were excluded to maintain methodological consistency.

Patients were categorized into 2 groups based on the transfusion strategy: the autologous blood group, which received intraoperative cell salvage transfusion, and the allogeneic blood group, which received leukocyte-depleted allogeneic packed red blood cell transfusion. The selection of transfusion type is a dynamic decision that incorporates objective indicators such as hemoglobin and hematocrit levels, along with subjective clinical factors, including symptoms and comorbidities. The decision to use intraoperative autologous blood recovery is influenced both by the clinician’s assessment of the patient’s transfusion needs and by the patient’s suitability for autologous transfusion. Suitability criteria include the absence of active infection, coagulation disorders, and foreign body contamination, as well as the patient’s provision of informed consent.

STUDY METHODS:

Baseline data included age, sex, height, weight, body mass index (BMI), and American society of anesthesiologists (ASA) physical status. Due to the small number of cases in Class I and Class IV, they could not be analyzed separately. Therefore, ASA Class I and Class IV were classified as Class I-II and Class III–IV, respectively. Smoking and alcohol consumption history were also recorded. Preoperative laboratory values (assessed within 24 h before surgery) comprised hemoglobin (Hb, g/L), white blood cell count (WBC, ×10⁹/L), and platelet count (PLT, ×10⁹/L). Comorbidities were defined as: hypertension (SBP ≥140 mmHg or DBP ≥90 mmHg), diabetes (fasting glucose ≥7.0 mmol/L or HbA1c ≥6.5%), coronary artery diseases (CAD), chronic obstructive pulmonary disease (COPD), history of cerebral infarction (CI), heart failure (HF), or myocardial infarction (MI). Intraoperative parameters included surgical duration, blood loss volume, transfusion type (autologous/allogeneic), and blood transfusion volume (mL).

STUDY OUTCOMES:

The primary outcome was POD, diagnosed with daily review of nursing records in the electronic medical records (EMR) to obtain the results of the Confusion Assessment Method (CAM) and keywords related to delirium (eg, “altered mental status”, “confusion”, “disorientation”, “agitation”, “delirium”, “inappropriate behavior”, “inattention”, and “hallucinations”). Two blinded researchers independently evaluated all cases, with discrepancies resolved through multidisciplinary consensus. Secondary outcomes were: 1) Length of stay (admission to discharge date); and 2) Postoperative complications (deep vein thrombosis, pulmonary infection), confirmed by clinical and radiographic criteria. The diagnosis of deep vein thrombosis was based on clinical symptoms such as limb swelling, pain, warmth, or tenderness, combined with confirmation by duplex ultrasonography [12]. The diagnosis of pulmonary infection was established based on clinical manifestations (eg, fever, cough, tachypnea, and auscultatory findings such as crackles) supported by imaging evidence from chest X-rays or CT scans showing pulmonary infiltrates, consolidation, or other characteristic features of infection [13].

STATISTICAL ANALYSIS:

To address baseline imbalance and reduce confounding bias due to baseline differences, we used inverse probability of treatment weighting (IPTW) based on propensity scores to balance the autotransfusion and allogeneic transfusion groups in terms of demographic characteristics, comorbidities, and preoperative conditions. The covariates included age, sex, BMI, ASA classification, smoking history and alcohol history, Hb, WBC and PLT, hypertension, diabetes, CAD, COPD, history of CI, HF, or MI, surgical duration, blood loss volume, transfusion type (autologous/allogeneic), and blood transfusion volume. The average treatment effect (ATE) was estimated using the gradient boosting machine algorithm in the WeightIt package to construct a propensity score-weighted model. After weighting, covariate balance was assessed via hypothesis testing. Continuous variables are expressed as median and interquartile range (IQR), and between-group comparisons were performed using the Mann-Whitney U test. Categorical variables are expressed as frequency and percentage, and group comparisons were conducted using the chi-square test or Fisher’s exact test.

Based on the weighted samples, a weighted survey design object was constructed using the survey package. Weighted generalized linear models were applied to analyze clinical outcomes. Binary outcomes – including postoperative delirium, pneumonia, and deep vein thrombosis – were analyzed using a weighted binomial regression model, and the odds ratio (OR) with its 95% confidence interval (CI) was calculated. For continuous outcomes (eg, length of hospital stay), a weighted linear regression model was used to estimate the between-group mean difference and its 95% CI.

All statistical analyses and propensity score weighting were performed in the R environment (R software version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria). Two-sided tests were used with a significance level set at α=0.05.

Results

STUDY POPULATION:

A total of 544 patients were included in the analysis after initial screening. Figure 1 shows the flowchart of the study. Of these, 142 were in the autologous blood transfusion group and 402 were in the allogeneic blood transfusion group. In the preliminary screening, we excluded the following patients: 43 patients with a history of delirium or psychiatric illness before surgery; 167 patients who received allogeneic blood transfusion or had coagulation dysfunction before surgery; 187 patients with missing key variables in clinical data(mainly due to incomplete records of height, weight, smoking/alcohol history, intraoperative blood loss, or special case information); and 121 patients who received general anesthesia during surgery. In addition, we excluded 1031 patients who did not receive blood transfusion during surgery and 35 patients who received both autologous and allogeneic blood transfusion simultaneously.

COMPARISON OF BASELINE CHARACTERISTICS:

Baseline characteristics and perioperative data are summarized in Table 1. The autologous (n=142) and allogeneic (n=402) transfusion groups showed no significant differences in age, sex distribution, ASA classification, smoking/alcohol history, or hypertension. Compared with the allogeneic transfusion group, the autologous transfusion group had lower BMI, fewer preoperative histories of stroke, less blood loss, and lower transfusion volume. After adjustment with IPTW, the 2 groups were similar except for blood loss and transfusion volume (Table 1).

ASSOCIATION BETWEEN INTRAOPERATIVE BLOOD TRANSFUSION TYPES AND POSTOPERATIVE CLINICAL OUTCOMES:

The overall incidence of POD was 9.9%. The incidence of POD was 4.2% in the autologous transfusion group and 11.9% in the allogeneic transfusion group. IPTW-adjusted analysis revealed a significantly lower risk of delirium in the autologous transfusion group compared to the allogeneic transfusion group. Autologous transfusion was associated with a 65.7% lower POD risk (adjusted OR 0.32; 95% CI, 0.13–0.79; P=0.014). No statistically significant differences were observed between the autologous and allogeneic transfusion groups in the secondary outcomes, including pneumonia, deep vein thrombosis, or length of hospital stay (Table 2).

Discussion

This single-center retrospective cohort study preliminarily explored the association between autologous versus allogeneic blood transfusion and the risk of POD in elderly patients undergoing HA. The results indicated a lower incidence of POD in the autologous transfusion group compared to the allogeneic group. After adjustment for predefined confounding variables using IPTW, this association remained statistically significant. These findings suggest that autologous transfusion is associated with a lower risk of POD relative to allogeneic transfusion in this patient population.

POD is a common complication among elderly patients following major surgery [6,14]. Multiple studies have demonstrated a significant correlation between blood transfusion and the occurrence of POD. Behrends et al identified intraoperative transfusion during major non-cardiac surgery as an independent risk factor for POD in older adults [15]. A large retrospective study by Ouyang et al, published in 2023, demonstrated that allogeneic transfusion was an independent predictor of POD in elderly patients undergoing total hip or knee arthroplasty [11]. Our study suggests that, among elderly patients undergoing hip HA under CSEA, autologous transfusion is associated with an approximately 65.7% lower risk of POD compared to allogeneic transfusion. These findings indicate that prioritizing autologous transfusion in blood management strategies may help reduce the incidence of POD. However, given the inherent limitations of residual confounding and population heterogeneity in observational studies, well-designed multicenter randomized controlled trials are warranted to validate these results, establish robust evidence, and optimize perioperative transfusion practice.

Transfusion-related inflammatory responses are considered a key mechanism underlying the development of POD. Stored red blood cells and their metabolic byproducts, such as free hemoglobin and iron ions, along with pro-inflammatory cytokines, including TNF-α and IL-6, can activate both peripheral and central immune systems, trigger neuroinflammation, alter blood-brain barrier permeability, and damage neurons and neurotransmitter systems [10,16–18]. In contrast, autologous transfusion avoids exposure to allogeneic antigens and can reduce the risk of POD by attenuating the inflammatory cascade [19,20]. These pathophysiological pathways align with Soranoglou et al’s clinical observations in bilateral knee arthroplasty and Tan et al’s experimental evidence of stored erythrocyte neurotoxicity [21,22], suggesting a shared pathophysiological basis for allogeneic transfusion-associated POD in orthopedic surgery.

Although recent guidelines advocating blood conservation primarily emphasize resource utilization and prevention of blood-borne infections, they often pay limited attention to the potential increase in postoperative morbidities [23]. This shows the need for a more comprehensive evaluation of transfusion practices, especially in clinical settings with high intraoperative transfusion rates. This issue is particularly pronounced in our center, where the intraoperative transfusion rate exceeds that of many similar institutions. This discrepancy is largely due to a strong subjective influence in surgeons’ decision-making, reflecting persistent gaps in awareness regarding the risks of transfusion-related complications. In practice, transfusion decisions, although informed by objective indicators such as estimated blood loss, are often heavily influenced by subjective assessments of the patient’s cardiopulmonary reserve, comorbidities, and intraoperative hemodynamic status. Transfusion decision-making thus is a multifactorial balancing act, dependent not only on patient-specific factors but also on the surgical team’s preoperative evaluation of bleeding risk and transfusion indications. Although the transfusion rates in this study were relatively high, after rigorous adjustment for confounders, our data reveal that the incidence of POD was significantly lower in patients receiving autologous transfusion compared to those receiving allogeneic transfusion. These findings underscore the critical importance of optimizing intraoperative management and judiciously selecting autologous transfusion. By minimizing reliance on allogeneic transfusion, refined transfusion strategies may reduce transfusion-related complications and ultimately enhance postoperative safety and recovery quality in elderly patients.

This study has inherent limitations. First, its retrospective design precluded full control of certain potential confounding factors, such as postoperative analgesia methods and the frequency of hemoglobin monitoring, which may have influenced delirium assessment but were not included in the analysis. Second, POD diagnosis relied on clinical records rather than standardized assessment tools, potentially leading to underdiagnosis of hypoactive delirium and affecting incidence estimates. Third, the single-center data may limit the generalizability of the findings, especially since transfusion strategies are influenced by institutional clinical pathways. Lastly, the study included only patients receiving CSEA, excluding those under general anesthesia (accounting for 5.6%), and although the literature suggests there is a minimal impact of anesthesia type on POD incidence, this restriction may still influence the external validity of the results. Future studies employing multicenter, prospective designs with standardized cognitive assessment tools are warranted to comprehensively evaluate the effects of different transfusion protocols and postoperative management on POD.

Conclusions

Our study indicates that elderly patients undergoing hip arthroplasty (HA) who received autologous transfusion tended to have a lower incidence of postoperative delirium (POD) compared to those who received allogeneic transfusion. These findings show that autologous transfusion might offer potential benefits in reducing the risk of POD. However, the exact clinical advantages and underlying pathophysiological mechanisms remain to be elucidated and require further validation through large-scale, multicenter prospective studies. Such investigations are important to better inform personalized transfusion management and to enhance perioperative care strategies for the elderly surgical population.