Appropriateness and determinants of packed red blood cell transfusion in knee arthroplasty: a retrospective tertiary hospital study

Abstract

Inappropriate packed red blood cell (PRC) transfusion remains a concern in surgical patients, including those undergoing knee arthroplasty. This study aimed to evaluate the appropriateness of PRC transfusion and identify clinical and procedural factors associated with inappropriate transfusion practices. A retrospective cohort study was conducted in patients who underwent knee arthroplasty at a tertiary academic hospital. Transfusion episodes were assessed for appropriateness using predefined clinical criteria. Multivariable logistic regression was used to identify independent predictors of inappropriate transfusion. Of 2,983 patients who underwent knee arthroplasty between August 2015 and December 2018, 275 (9.22%) received PRC transfusion in 333 episodes. Among these, 77.5% were appropriate and 22.5% were inappropriate. Most transfusions occurred postoperatively, with a single-unit transfusion being the most common practice. Inappropriate transfusions were frequently administered to patients with hemoglobin ≥ 10 g/dL or hematocrit ≥ 30%, without documented clinical indications. The absence of clinical risk factors for ischemia and a preoperative hemoglobin level < 10 g/dL were independent predictors of inappropriate transfusion with an adjusted risk ratio 10.85 (95% CI, 5.43–21.70; p < 0.001) and 1.41 (95% CI, 1.01–1.96; p = 0.042), respectively. Although most PRC transfusions in knee arthroplasty patients were appropriate, a substantial proportion did not align with established clinical guidelines. The absence of clinical risk factors for ischemia was strongly associated with inappropriate transfusion. These findings underscore the need for enhanced adherence to evidence-based transfusion criteria and improved clinical documentation.

Introduction

Knee arthroplasty is frequently associated with perioperative blood loss, and reported rates of postoperative allogeneic transfusion range from 18% to 67% of patients^{1, 2, 3–4}. While transfusion may be life-saving in select circumstances, allogeneic red cell transfusions carry well-documented risks, including transfusion reactions, immunomodulation, infectious disease transmission, periprosthetic joint infection, and increased 30-day mortality^{5, 6, 7–8}. Moreover, accumulating evidence indicates that liberal transfusion strategies may increase the risk of adverse outcomes, including pulmonary, infectious, and immunological complications^{9, 10–11}.

Despite the widespread implementation of perioperative patient blood management (PBM) programs and international guidelines endorsing restrictive transfusion thresholds, inappropriate red blood cell transfusions continue to occur in surgical settings. Recent studies indicate that 25% to 46% of perioperative transfusions do not meet evidence-based appropriateness criteria^{12, 13, 14, 15–16}. A multicenter study across 42 hospitals in Tanzanian reported that orthopedic procedure significantly increased the likelihood of unnecessary blood transfusion (risk ratio 3.8; 95% CI, 2.2–6.7; p < 0.001)¹³. The study also identified factors predicting blood overutilization, including the absence of documented clinical indications prior to a transfusion¹³. Furthermore, recent meta-analysis suggests that routine preoperative crossmatching for all patients undergoing primary total knee arthroplasty (TKA) may be unnecessary and resource-inefficient, particularly in the context of improved surgical hemostasis and adoption of blood conservation strategies¹⁷.

Given these concerns, we conducted a retrospective study to evaluate the appropriateness of packed red blood cell (PRC) transfusion among patients undergoing knee arthroplasty at a tertiary academic medical center. This investigation represents a collaboration effort between the Department of Anesthesiology and the Adult Reconstructive Surgery Unit, Department of Orthopedic Surgery, Faculty of Medicine Siriraj Hospital, Mahidol University. The primary objectives were to determine the proportion of appropriate versus inappropriate transfusions, and to identify clinical and procedural factors associated with inappropriate transfusion episodes.

Methods

Study design and population

This retrospective observational study was conducted following approval by the Siriraj Institutional Review Board approved (Si 726/2019) and was registered with ClinicalTrials.gov (Registration No.‎ NCT04303910)‎.‎ The study included patients who underwent knee arthroplasty, including primary unilateral or bilateral total knee arthroplasty (TKA) as well as unicompartmental knee arthroplasty (UKA), and received at least one unit of red blood cell components, either allogenic or autologous blood, during the perioperative period. Patients who underwent revision knee arthroplasty and concurrent surgical procedures were excluded from the analysis.

Perioperative surgical and anesthetic practices

In our institute, the use of a pneumatic tourniquet is standard practice during knee arthroplasty and was employed in all cases during the surgical procedure. The tourniquet is typically inflated to 300 mmHg prior to incision and maintained throughout the procedure until wound closure was completed.

Tranexamic acid (TXA) was administered in all cases according to surgeon preference using one of two institutional protocols during the study period: (1) intravenous TXA, consisting of 750 mg administered before tourniquet inflation, followed by an additional 750 mg administered three hours after surgery, or (2) intra-articular TXA, consisting of 2 g for unilateral total knee arthroplasty or 3 g total (1.5 g per side) for bilateral procedures, administered via intra-articular injection after capsular closure. The choice of administration route was based on individual surgeon preference, and both protocols reflect institutional standards for blood conservation in knee arthroplasty.

Surgical drains were not used in any cases in this study, in accordance with our institutional protocol, which does not routinely employ postoperative drainage systems for knee arthroplasty.

Anesthetic management during the procedures followed standard institutional protocols. The choice of anesthesia, either general anesthesia or neuraxial anesthesia, was based on patient factors and anesthesiologist preference. Intraoperative monitoring followed standard ASA guidelines, and anesthetic management was aimed at maintaining hemodynamic stability and optimizing oxygen delivery. Perioperative fluid and blood management were at the discretion of the attending anesthesiologist.

Data collection

Patient data were obtained through medical record review using the hospital information technology system. Siriraj Hospital is a 2,000-bed university-affiliated tertiary care center located in Bangkok, Thailand. Demographic and clinical information collected included age, sex, body mass index (BMI), American Society of Anesthesiologists (ASA) physical status classification, comorbidities, surgical details, preoperative hemoglobin and hematocrit values.

Detailed information regarding each transfusion episode was collected, including the pre-transfusion hemoglobin and/or hematocrit, vital signs, and clinical signs and symptoms prior to transfusion (e.g., active bleeding, tachycardia, hypotension, pallor, altered consciousness), and transfusion details (e.g., location, timing, number of units administered, and transfusion indication). Perioperative urine and drain outputs were also documented.

Transfusion appropriateness criteria

The appropriateness of transfusion in this study was evaluated according to a structured, evidence-informed algorithm based on the acute blood loss, hemoglobin/ hematocrit thresholds, clinical risk factors for ischemia, and clinical signs or symptoms of acute anemia^18,19.

In patients with acute blood loss, transfusion was considered appropriate when the estimated acute loss exceeded 20% of the total blood volume, irrespective of hemoglobin or hematocrit ‎levels.

In patients without acute blood loss, the appropriateness of transfusion was evaluated in a stepwise manner (Fig. 1). Initially, transfusion was considered appropriate when the Hb level was below 10 g/dL or the Hct was below 30% in the presence of clinical risk factors for ischemia, which included age over 65 years, a history of coronary artery disease, heart failure, cerebrovascular disease, or valvular heart disease.

Fig. 1 [Images not available. See PDF.]

Algorithm for assessing the appropriateness of packed red blood cell transfusion.

In patients without ischemic risk factors, age-specific transfusion thresholds were applied. Transfusion was considered appropriate only when Hb or Hct fell below the following thresholds and was accompanied by at least one clinical sign or symptom of acute anemia: Hb < 7 g/dL or Hct < 24% in patients aged < 40 years, Hb < 8 g/dL or Hct < 27% in patients aged 40–65 years, and Hb < 10 g/dL or Hct < 30% in patients aged > 65 years.

Clinical signs and symptoms of acute anemia included syncope, tachycardia, dyspnea, angina, oxygen saturation < 90%, and neurological alterations. In the absence of such symptoms, transfusion was considered inappropriate, even if laboratory thresholds were met. Transfusions not meeting these criteria defined above were classified as inappropriate.

Sample size calculation

Sample size was calculated using the specific formula: n = Z_α/2² p(1-p)/d², where p represents the estimated proportion of appropriated red blood cell transfusions (p = 0.25) regarding prior studies^13,15. The significance level was set at α = 0.05, with the corresponding critical value Z (0.975) = 1.96, and a margin of error (d) of 0.05. The resulting sample size was 289 transfusion episodes.

Statistical analysis

Descriptive statistics were used to summarize patient and transfusion characteristics. Categorical variables were presented as frequencies and percentages. Continuous variables were reported as means with standard deviations (SD), 95% ‎confidence intervals (CI), or medians with interquartile ranges (IQR), depending on data distribution.

Univariable analysis using chi-square or Fisher’s exact tests was conducted to identify potential predictor variables associated with inappropriate transfusion. Variables with p < 0.20 were entered into a multivariable Poisson regression model with a log link and robust standard errors. In addition to statistical criteria, variable selection was also guided by clinical relevance and previously published literature to enhance robustness, interpretability, and generalizability of the model^{12,13,15,16,20,21}.

Results were reported as adjusted risk ratios (RR) with corresponding 95% CI. All statistical tests were two-tailed, and a p-value < 0.05 was considered statistically significant. Data was analyzed using IBM SPSS Statistics version 30.0 (IBM Corp, Armonk, NY, USA), and STATA SE-Standard Edition version 14.1.

Results

Patient characteristics

Between August 2015 to December 2018, a total of 2,983 patients underwent knee arthroplasty in our institution. Of these, 275 patients (9.22%) received at least one unit of packed red blood cells (PRC) and were included in the analysis. The mean patient age was 71.5 years ‎(95% CI 70.5–72.5), with a range of 44 to 89 years. The majority of patients were female (86.9%), and most patients‎ were classified as ASA physical status II (65.4%), followed by ASA III (31.6%). Common comorbidities included hypertension (‎76.0%), dyslipidemia (32.0%), and diabetes mellitus (30.2%). Additionally, 60 patients (21.8%) were taking antiplatelet agents preoperatively, 2.2% had been on anticoagulants and were withheld appropriately prior to surgery. Table 1 summarizes detailed demographic and clinical characteristics, types of arthroplasty, and anesthetic techniques.

Table 1. Baseline characteristics of transfused knee arthroplasty patients.

	N = 275
Age (years)	71.5± 8.4
Sex: male	36 (13.1)
BMI (kg/m2)	26.0 ± 4.1
ASA physical status
I	8 (2.9)
II	180 (65.5)
III	87 (31.6)
Preoperative hemoglobin (g/dL)
Male	12.1 ± 1.7
Female	11.4 ± 1.4
Preoperative hematocrit (%)
Male	37.0 ± 4.7
Female	35.4 ± 4.0
Comorbidities
Hypertension	209 (76.0)
Dyslipidemia	88 (32.0)
Diabetic mellitus	83 (30.2)
Chronic kidney disease	49 (17.8)
Coronary artery disease	28 (10.2)
Cerebrovascular disease	21 (7.6)
Cardiac arrhythmia	16 (5.8)
Cardiomyopathy	6 (2.2)
Valvular heart disease	4 (1.5)
Preoperative type and screen	23 (8.4)
Preoperative cross-matching	150 (54.5)
Type of arthroplasty
TKA, unilateral	190 (69.1)
TKA, bilateral	52 (18.9)
UKA, unilateral	7 (2.5)
UKA, bilateral	26 (9.5)
Anesthetic technique
Neuraxial anesthesia	239 (86.9)
General anesthesia	36 (13.1)
Peripheral nerve block supplement	151 (54.9)

Data are presented as mean ± standard deviation or number (%), as appropriate.

ASA = American Society of Anesthesiologists, TKA = total knee arthroplasty‎, UKA = unicompartmental knee arthroplasty.

Transfusion characteristics

A total of 333 transfusion episodes, comprising 353 units of PRC, were evaluated. All transfusions were administered in the postoperative period (100%) and consisted of allogeneic blood (100%). The majority of transfusions occurred in the general ward (99.4%), while only two episodes (0.6%) were given in the postanesthetic care unit (PACU).

Most transfusion episodes (94.0%) involved transfusion of a single unit, and 20 episodes (6.0%) involved ‎two units per episode. The median number of units transfused per episode was 1 (IQR [1, 1]). At the individual patient level, 212 patients (77.1%) received a single unit, while 63 patients (22.9%) received two or more units. The highest total number of units transfused to a single patient was seven. Figure 2 illustrates the distribution of PRC units per patient.

Fig. 2 [Images not available. See PDF.]

Distribution of total packed red blood cell units transfused per patient.

Appropriateness of PRC transfusion

Among the 333 transfusion episodes, 258 episodes ‎‎(77.5%) were classified as appropriate based on the predefined criteria, whereas 75 episodes (22.5%) were considered inappropriate (Table 2).

Table 2. Appropriateness of packed red blood cells transfusion episodes.

Transfusion characteristic	Number of episodes (n)	Percentage (%)
Total PRCs transfusion episodes	333	100.0
Appropriate transfusions	258	77.5
Hb < 10 g/dL or Hct < 30% with clinical risk of ischemia	204	61.3
Acute blood loss > 20% of total blood volume	54	16.2
Hb/Hct below threshold with symptoms of anemia	2	0.6
Inappropriate transfusions	75	22.5
Hb ≥ 10 g/dL or Hct ‎≥ 30%	31	9.3
Hb/Hct below general threshold but exceeding age-specific criteria	25	7.5
Hb/Hct below age-specific criteria but no documented signs/symptoms of anemia	17	5.1

Hb = hemoglobin, Hct = hematocrit, PRC = packed red blood cells.

Among appropriate transfusions, the most common indication was hemoglobin < 10 g/dL or hematocrit < 30% in patients with clinical risk factors for ischemia‎ (n = 204, 79.0%). Acute blood loss exceeding 20% of total blood volume accounted for 54 episodes (20.9%) and only two episodes (0.6%) were attributed to symptomatic anemia in patients with hemoglobin or hematocrit levels below the age-specific thresholds.

Inappropriate transfusions most frequently occurred in patients with hemoglobin levels ≥ 10 g/dL or hematocrit levels ≥ 30% (n = 31, 41.3%), followed by transfusions in patients with hemoglobin/hematocrit values exceeding the age-specific thresholds (n = 25, 33.3%). Seventeen episodes (25.3%) involved transfusions in patients with low hemoglobin or hematocrit levels but without any documented clinical signs or symptoms of acute anemia.

Factors associated with inappropriate transfusion

In the univariate analysis, inappropriate transfusion was significantly associated with younger age (age < 65 years, p < 0.001), absence of clinical risk factors for ischemia (p < 0.001), preoperative hemoglobin < 10 g/dL (p = 0.006), and preoperative crossmatching (p = 0.019). Additionally, the absence of diabetes mellitus (p = 0.045) and coronary artery disease (p = 0.037) were also significantly associated (Table 3). The univariable analysis showed no statistically significant correlation between the type of arthroplasty (bilateral TKA) and the inappropriate transfusion (p = 0.748). A total of twelve variables with p < 0.20 in univariate analysis were included in the multivariable regression model.

Table 3. Univariable analysis of factors associated with inappropriate transfusion.

Variable	Transfusion			p-value
	All (n = 333)	Appropriate (n = 260; 78.1%)	Inappropriate (n = 73; 21.9%)
Age (years)				< 0.001
< 65 years	68	20 (29.4%)	48 (70.6%)
≥ 65 years	265	240 (90%)	25 (9.4%)
Sex				0.638
Female	284	223 (78.5%)	61 (21.5%)
Male	49	37 (75.5%)	12 (24.5%)
Body mass index (kg/m2)	25.87 ± 4.00	25.71 ± 4.06	26.45 ± 3.73	0.143
ASA classification				0.414
I-II	224	172 (76.8%)	52 (23.2%)
III	109	88 (80.7%)	21 (19.3%)
Clinical risk factors for ischemia				< 0.001
No	62	11 (17.7%)	51 (82.3%)
Yes	271	249 (91.9%)	22 (8.1%)
No comorbidities	42	25 (59.5%)	17 (40.5%)	0.002
Comorbidities
Hypertension				0.141
No	83	60 (72.3%)	23 (27.7%)
Yes	250	200 (80.0%)	50 (20.0%)
Dyslipidemia				0.136
No	227	172 (75.8%)	55 (24.2%)
Yes	106	88 (83.0%)	18 (17.0%)
Diabetic mellitus				0.045
No	233	175 (75.1%)	58 (24.9%)
Yes	100	85 (85.0%)	15 (15.0%)
Chronic kidney disease				0.066
No	272	207 (76.1%)	65 (23.9%)
Yes	61	53 (86.9%)	8 (13.1%)
Coronary artery disease				0.037
No	297	227 (76.4%)	70 (23.6%)
Yes	36	33 (91.7%)	3 (8.3%)
Cerebrovascular disease				0.134
No	305	235 (77.0%)	70 (23.0%)
Yes	28	25 (89.3%)	3 (10.7%)
Cardiac arrhythmia				0.183
No	307	237 (77.2%)	70 (22.8%)
Yes	26	23 (88.5%)	3 (11.5%)
Cardiomyopathy				0.190
No	327	254 (77.7%)	73 (22.3%)
Yes	6	6 (100%)	0
Valvular heart disease				0.580
No	329	256 (77.8%)	73 (22.2%)
Yes	4	4 (100%)	0
Preoperative antiplatelet	74	63 (85.1%)	11 (14.9%)	0.096
Preoperative steroid	5	2 (40.0%)	3 (60.0%)	0.072
Preoperative hemoglobin level				0.006
< 10 g/dL	52	33 (63.5%)	19 (36.5%)
≥ 10 g/dL	281	227 (80.8%)	54 (19.2%)
Preoperative cross matching	188	138 (73.4%)	50 (26.6%)	0.019
Type of surgery (bilateral TKA)	73	58 (79.5%)	15 (20.5%)	0.748
Anesthetic technique				0.381
Neuraxial anesthesia	286	221 (77.3%)	65 (22.7%)
General anesthesia	47	39 (83.0%)	8 (17.0%)
Peripheral nerve block supplement	177	139 (78.5%)	38 (21.5%)	0.831

Comparison of demographic and clinical variables between appropriate and inappropriate transfusion episodes using bivariate statistical analysis. Variables with p < 0.20 were considered for multivariable analysis.

ASA = American Society of Anesthesiologists, TKA = total knee arthroplasty‎.

In the multivariable Poisson regression analysis, the absence of clinical risk factors for ischemia and a preoperative hemoglobin level < 10 g/dL were independently associated with inappropriate transfusion, with adjusted risk ratio (RR) of 10.85 (95% CI, 5.43–21.70; p < 0.001) and 1.41 (95% CI, 1.01–1.96; p = 0.042), respectively (Table 4).

Table 4. Multivariate analysis of risk factors for inappropriate blood transfusion.

Variable	Univariate RR (95% CI)	p-value	Multivariate RR (95% CI)	p-value
Age < 65 years	7.48 (5.00-11.21)	< 0.001	1.02 (0.57–1.80)	0.955
Sex
Male	1		1
Female	0.88 (0.51–1.51)	0.634	0.31 (0.14–0.70)	0.004
Body mass index (kg/m2)	1.04 (0.99–1.08)	0.137	0.99 (0.95–1.03)	0.740
Clinical risk factors for ischemia
No	10.13 (6.67–15.38)	< 0.001	10.85 (5.43–21.70)	< 0.001
Yes	1		1
No comorbidities	2.10 (1.36–3.25)	0.001	1.15 (0.59–2.24)	0.691
Hypertension
No	1.39 (0.90–2.12)	0.135	0.83 (0.45–1.52)	0.544
Yes	1		1
Dyslipidemia
No	1.43 (0.88–2.31)	0.147	1.04 (0.67–1.63)	0.855
Yes	1		1
Diabetic mellitus
No	1.66 (0.99–2.79)	0.055	1.26 (0.79–2.01)	0.335
Yes	1		1
Chronic kidney disease
No	1.82 (0.92–3.60)	0.084	1.85 (0.80–4.32)	0.153
Yes	1		1
Coronary artery disease
No	2.83 (0.94–8.53)	0.065	1.27 (0.42–3.90)	0.671
Yes	1		1
Cerebrovascular disease
No	2.14 (0.72–6.37)	0.171	1.00 (0.31–3.20)	0.997
Yes	1		1
Preoperative antiplatelet	0.62 (0.34–1.12)	0.112	1.20 (0.73–1.97)	0.477
Preoperative hemoglobin level
< 10 g/dL	1.90 (1.23–2.93)	0.004	1.41 (1.01–1.96)	0.042
≥ 10 g/dL	1		1
Preoperative cross matching ‎	1.68 (1.08–2.61)	0.023	1.07 (0.75–1.53)	0.698

Adjusted risk ratios (RR) and 95% confidence intervals (CI) derived from Poisson regression with robust standard errors to identify independent predictors of inappropriate transfusion.

Figure 3 illustrates the receiver operating characteristic (ROC) curve for the absence of clinical risk factors for ischemia (age over 65 years, coronary artery disease, heart failure, cerebrovascular disease, or ‎valvular heart disease‎) and for each individual variable. Among these, the absence of clinical risk factors for ischemia demonstrated the highest predictor value for inappropriate transfusion, with an area under the ROC curve (AUROC) of 0.828, 95% CI: 0.774–0.883). Age below 65 years also showed good predictive value (AUROC 0.790; 95% CI: 0.733–0.847). In contrast, other individual variables demonstrated poor predictive utility, with AUROCs approximating 0.5, indicating minimal discriminatory performance.

Fig. 3 [Images not available. See PDF.]

Receiver operating characteristic (ROC) curves for predictors of inappropriate transfusion. Area under the ROC curve (AUROC) for predictors of inappropriate transfusion: no clinical risk factors for ischemia = 0.828 (95% CI 0.774–0.883); age < 65 years = 0.790 (95% CI 0.733–0.847); Preoperative hemoglobin level < 10 g/dL = 0.57 (95% CI 0.51–0.62); no cerebrovascular disease (CVD) = 0.528 (95% CI 0.498–0.557); no coronary artery disease (CAD) = 0.543 (95% CI 0.512–0.574); no valvular heart disease (VHD) = 0.508 (95% CI 0.500–0.515); and no history of heart failure = 0.512 (95% CI 0.502–0.521).

Discussion

This retrospective cohort study evaluated the appropriateness of packed red blood cells (PRC) transfusions in patients undergoing knee arthroplasty and identified clinical factors associated with inappropriate transfusion practices. Among 2,983 patients who underwent knee arthroplasty during the study period, 275 patients (9.22%) received at least one unit of PRC. Our analysis of 333 transfusion episodes revealed that 77.5% were appropriate based on predefined clinical criteria, while a significant portion (22.5%) may be considered inappropriate.

The observed transfusion rate aligns with previously reported data in elective orthopedic populations, which typically range from 5% to 20%^{1, 2, 3–4}. The demographic profile of the transfused population, predominantly elderly and female, is consistent with global trends in knee arthroplasty recipients²². Most transfusions occurred postoperatively in the general ward, with the administration of a single unit being the most common practice. This finding reflects recent efforts to adopt more conservative transfusion strategies in surgical patients, in line with restrictive transfusion thresholds recommended by current guidelines^{23, 24–25}.

Most appropriate transfusions were administered in patients with clinical risk factors for ischemia ‎and low hemoglobin < 10 g/dL or hematocrit < 30%,, consistent with recommendations by the American Association of Blood Banks (AABB), NICE, and international guidelines^{23, 24, 25, 26, 27–28}. However, symptomatic anemia was an infrequent documented indication, which may reflect either limited documentation practices or under recognition of clinical symptoms.

Importantly, our study reveals a considerable proportion of inappropriate transfusions (22.5%) ‎which is similar to those reported from several studies, ranging from 21.4% to 28.8% across various patient scenarios^13,15,29,30. Although our study’s rate of inappropriate transfusion is lower than the 59.3% reports by an international multidisciplinary expert panel³¹, the variation may reflect differences in assessment criteria or institutional practices. Our findings reinforce the ongoing challenge of optimizing transfusion decision-making, even in the presence of well-established evidence-based guidelines^{32, 33–34}.

Our results showed inappropriate transfusions were most frequently administered to patients with hemoglobin ≥ 10 g/dL or hematocrit ≥ 30%, contrary to widely accepted restrictive transfusion practices³⁵. Other inappropriate episodes also occurred in patients having hemoglobin‎ or hematocrit levels‎ exceeding age-specific thresholds without additional clinical justification. Additionally, some transfusions were given to patients with low hemoglobin or hematocrit values in the absence of documented symptoms or clinical signs of anemia such as ongoing bleeding, hemodynamic instability, or ischemic symptoms, underscoring the need for better clinical assessment and documentation prior to transfusion decisions. Furthermore, our analyses showed no significant association between the type of knee arthroplasty (bilateral TKA) and inappropriate transfusion. This lack of association may, at least in part, be explained by the institutional implementation of standardized blood conservation strategies, including routine tranexamic acid administration and the avoidance of surgical drains, which likely minimized differences between procedure types.

Multivariable analysis identified two independent predictors of inappropriate transfusion: the absence of clinical risk factors for ischemia (adjusted RR 11.70; 95% CI, 5.77–23.75; p < 0.001) and a preoperative hemoglobin level < 10 g/dL (adjusted RR 1.41; 95% CI, 1.01–1.96; p = 0.042). This finding regarding the absence of risk factors for ischemia suggests a tendency to transfuse patients without a clear clinical need, potentially leading to overuse in low-risk individuals. More notably, the paradoxical finding that a low preoperative hemoglobin level is also an independent predictor of inappropriate transfusion warrants specific attention. While a low hemoglobin level might intuitively be a reason to transfuse, our results suggest that these transfusions may be given without meeting established clinical criteria, such as the presence of symptoms of anemia. This indicates a potential practice of “anticipatory” or non-symptom-driven transfusions. These findings can lead to opportunities for quality improvement. Specifically, enhancing education on transfusion indications, reinforcing adherence to restrictive thresholds —particularly for patients with preoperative hemoglobin levels of 7–9 g/dL—is critical. Incorporating clinical decision support tools into electronic health records may help mitigate inappropriate transfusions.

Our findings are consistent with previous literature demonstrating that inappropriate transfusions are often driven by non-evidence-based practices, including reliance on absolute laboratory thresholds without consideration of clinical context^34,36. The results advocate for improved adherence to restrictive transfusion guidelines and the incorporation of individualized patient risk profiles to guide transfusion decisions.

This study has several strengths. It uses a clearly defined and evidence-based framework to evaluate transfusion appropriateness and the comprehensive analysis of both patient- and transfusion-related variables. However, several limitations must be acknowledged.

First, the retrospective design is subject to inherent biases, including limits the ability to capture all relevant clinical factors influencing transfusion decisions. Specifically, inconsistent or missing documentation of clinical signs and symptoms prior to transfusion may have led to misclassification of transfusion appropriateness. For example, symptoms such as dyspnea or pallor may not have been adequately recorded, leading to potential underestimation of clinically appropriate transfusions. This limitation highlights the need for future prospective studies to utilize a more structured and standardized clinical data collection tool at the bedside to ensure comprehensive documentation of clinical triggers for transfusion. Additionally, early postoperative hematoma formation was not routinely documented and therefore could not be analyzed as a potential factor influencing transfusion decisions.

Second, although pre-transfusion hemoglobin or hematocrit values were collected and used as surrogate transfusion triggers, this study did not evaluate postoperative hemoglobin or hematocrit decline. We acknowledge this as a limitation, as postoperative hematologic trends could provide additional insight into blood loss, help contextualize transfusion decisions more accurately, and potentially enhance future transfusion guidelines and decision-making algorithms. Third, the findings are derived from a single institution, which may limit generalizability. Finally, we did not assess transfusion-related outcomes (e.g., complications, hospital length of stay), which would offer additional context on the clinical impact of inappropriate transfusions. Despite these limitations, our findings reinforce the need for improved transfusion practices and implementation of PBM principles in surgical populations. Future studies should focus on prospective validation of transfusion appropriateness tools and assess the effectiveness of targeted interventions, such as real-time audit-and-feedback systems or clinician education, in reducing unnecessary transfusions.

Conclusion

In summary, while most PRC transfusions in knee arthroplasty patients were appropriate, some transfusions were not aligned with established clinical practices. The absence of clinical risk factors for ischemia strongly predicted an inappropriate transfusion. These findings underscore the importance of individualized risk assessment and adherence to evidence-based transfusion thresholds. Institutional strategies to enhance clinician awareness, documentation, and decision-making may help reduce inappropriate transfusions and optimize perioperative patient outcomes. Implementing educational interventions and clinical decision support tools may further reduce unnecessary transfusions and improve patient outcomes.

Acknowledgements

The author would like to thank Ms. Arporn Pimtong for her help with coordination in this study.

Author contributions

MW were responsible for the conceptualization, methodology, and project administration of the study. SA were involved in data curation. MW and SA conducted the formal analysis and interpretation of the data. MW and SA contributed to the original draft and to the review and editing of the manuscript. All authors read and approved the final version of the manuscript.

Funding

The study was supported by the Faculty of Medicine Siriraj Hospital, Mahidol University. Grant number was R016333013. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

The datasets used in this study are available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

This study was conducted according to the ethical standards established by the 1964 Declaration of Helsinki. The study was approved by the Siriraj Institutional Review Board of the Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand (Si 726/2019, Chairperson Prof. Chairat Shayakul, M.D.) on 22 October 2019. Informed consent was obtained from all individual participants included in the study.

Abbreviations

ASA

American Society of Anesthesiologists

BMI

Body mass index

PACU

Postanesthetic care unit

PBM

Patient blood management

PRC

Packed red blood cell

TKA

Total knee arthroplasty

TXA

Tranexamic acid

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Friedman, R; Homering, M; Holberg, G; Berkowitz, SD. Allogeneic blood transfusions and postoperative infections after total hip or knee arthroplasty. JBJS; 2014; 96, 4 pp. 272-278. [DOI: https://dx.doi.org/10.2106/JBJS.L.01268]

2. Hart, A et al. Blood transfusion in primary total hip and knee arthroplasty. Incidence, risk factors, and thirty-day complication rates. J. Bone Joint Surg. Am.; 2014; 96, 23 pp. 1945-1951. [DOI: https://dx.doi.org/10.2106/JBJS.N.00077] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25471908]

3. Danninger, T. et al. Blood transfusions in total hip and knee arthroplasty: an analysis of outcomes. Sci. World J.. 2014 (1), 623460 (2014).

4. Noticewala, MS et al. Predicting need for allogeneic transfusion after total knee arthroplasty. J. Arthroplasty; 2012; 27, 6 pp. 961-967. [DOI: https://dx.doi.org/10.1016/j.arth.2011.10.008] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22608172]

5. Gómez-Ramirez, S; Jericó, C; Muñoz, M. Perioperative anemia: prevalence, consequences and pathophysiology. Transfus. Apher Sci.; 2019; 58, 4 pp. 369-374. [DOI: https://dx.doi.org/10.1016/j.transci.2019.06.011] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31416710]

6. Baron, DM et al. Preoperative anaemia is associated with poor clinical outcome in non-cardiac surgery patients. Br. J. Anaesth.; 2014; 113, 3 pp. 416-423.1:STN:280:DC%2BC2cjivFWitw%3D%3D [DOI: https://dx.doi.org/10.1093/bja/aeu098] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24829444]

7. Browne, JA; Adib, F; Brown, TE; Novicoff, WM. Transfusion rates are increasing following total hip arthroplasty: risk factors and outcomes. J. Arthroplasty; 2013; 28, 8 Suppl pp. 34-37. [DOI: https://dx.doi.org/10.1016/j.arth.2013.03.035] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23896359]

8. Fernandez, MC; Gottlieb, M; Menitove, JE. Blood transfusion and postoperative infection in orthopedic patients. Transfusion; 1992; 32, 4 pp. 318-322.1:STN:280:DyaK383mtlCntg%3D%3D [DOI: https://dx.doi.org/10.1046/j.1537-2995.1992.32492263444.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/1585435]

9. Webster, NR. Stranger danger—mortality after transfusions. Br. J. Anaesth.; 2017; 118, 3 pp. 280-282.1:STN:280:DC%2BC1c3ntlGjsQ%3D%3D [DOI: https://dx.doi.org/10.1093/bja/aew405] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28203728]

10. Jans, O; Kehlet, H; Johansson, PI. Transfusion-related mortality after primary hip arthroplasty–an analysis of mechanisms and confounders. Vox Sang; 2012; 103, 4 pp. 301-308.1:STN:280:DC%2BC38rms1aktQ%3D%3D [DOI: https://dx.doi.org/10.1111/j.1423-0410.2012.01612.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22519366]

11. Vamvakas, EC; Blajchman, MA. Transfusion-related mortality: the ongoing risks of allogeneic blood transfusion and the available strategies for their prevention. Blood; 2009; 113, 15 pp. 3406-3417.1:CAS:528:DC%2BD1MXkslCkur0%3D [DOI: https://dx.doi.org/10.1182/blood-2008-10-167643] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19188662]

12. Kong, Y et al. Appropriateness of red blood cell use in China in the last thirteen years: a systematic review. Heliyon; 2019; 5, 3 e01408. [DOI: https://dx.doi.org/10.1016/j.heliyon.2019.e01408] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30976687][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6441792]

13. Mauka, WI et al. Risk factors for inappropriate blood requisition among hospitals in Tanzania. PLoS One; 2018; 13, 5 e0196453. [DOI: https://dx.doi.org/10.1371/journal.pone.0196453] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29771998][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5957429]

14. Chou, CW; Xu, R; Yang, L; Huang, W. Perioperative red blood cell transfusion for patients undergoing elective non-cardiac surgery: an audit at a Chinese tertiary hospital. Transfus. Apher Sci.; 2014; 51, 3 pp. 99-103. [DOI: https://dx.doi.org/10.1016/j.transci.2014.08.012] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25189106]

15. Colomina, MJ; de Miguel, M; Pelavski, A; Castellá, D. Appropriateness of red blood cell use in orthopedic surgery and traumatology: analysis of transfusion practice. Eur. J. Orthop. Surg. Traumatol.; 2012; 22, 2 pp. 129-135. [DOI: https://dx.doi.org/10.1007/s00590-011-0816-8]

16. Barr, PJ et al. The appropriateness of red blood cell use and the extent of overtransfusion: right decision?. Right amount? Transfusion; 2011; 51, 8 pp. 1684-1694. [DOI: https://dx.doi.org/10.1111/j.1537-2995.2011.03130.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21470238]

17. Nunez, JH et al. Routine pretransfusion testing before primary total hip or knee arthroplasty are an expensive and wasteful routine. Systematic review and meta-analysis. Arch. Orthop. Trauma. Surg.; 2024; 144, 4 pp. 1585-1595. [DOI: https://dx.doi.org/10.1007/s00402-024-05243-3] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38416137]

18. Simancas-Racines, D et al. Quality of clinical practice guidelines about red blood cell transfusion. J. Evid. Based Med.; 2019; 12, 2 pp. 113-124. [DOI: https://dx.doi.org/10.1111/jebm.12330] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30511477]

19. Abbas, K; Askari, R; Hafeez, K. Transfusion practice in orthopedic patients: do we really need it?. JPMA: J. Pak. Med. Assoc.; 2014; 64, 12 S144. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25989764]

20. Etminan, M. et al. To adjust or not to adjust: the role of different covariates in cardiovascular observational studies. Am Heart J2021, 23762–23767 (2021).

21. Sauerbrei, W. et al. State of the Art in selection of variables and functional forms in multivariable analysis-outstanding issues. Diagn Progn Res43, 2563 (2020).

22. Kurtz, S et al. Projections of primary and revision hip and knee arthroplasty in the united States from 2005 to 2030. J. Bone Joint Surg. Am.; 2007; 89, 4 pp. 780-785. [DOI: https://dx.doi.org/10.2106/00004623-200704000-00012] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17403800]

23. Yazer, MH; Triulzi, DJ. AABB red blood cell transfusion guidelines: something for almost everyone. JAMA; 2016; 316, 19 pp. 1984-1985. [DOI: https://dx.doi.org/10.1001/jama.2016.10887] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27732707]

24. Carson, JL et al. Clinical practice guidelines from the AABB: red blood cell transfusion thresholds and storage. JAMA; 2016; 316, 19 pp. 2025-2035. [DOI: https://dx.doi.org/10.1001/jama.2016.9185] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27732721]

25. Tobian, AA; Heddle, NM; Wiegmann, TL; Carson, JL. Red blood cell transfusion: 2016 clinical practice guidelines from AABB. Transfusion; 2016; 56, 10 pp. 2627-2630. [DOI: https://dx.doi.org/10.1111/trf.13735] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27739152]

26. 2020 surveillance of blood transfusion (NICE guideline NG24) (National Institute for Health and Care Excellence (NICE), 2020).

27. Padhi, S. et al. Blood transfusion: summary of NICE guidance. BMJ2015, 351h5832 (2015).

28. Goodnough, LT; Levy, JH; Murphy, MF. Concepts of blood transfusion in adults. Lancet; 2013; 381, 9880 pp. 1845-1854. [DOI: https://dx.doi.org/10.1016/S0140-6736(13)60650-9] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23706801]

29. Díaz, MQ et al. Appropriate use of red blood cell transfusion in emergency departments: a study in five emergency departments. Blood Transfus.; 2017; 15, 3 pp. 199-206. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27416566][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5448824]

30. Spradbrow, J et al. Evaluating appropriate red blood cell transfusions: a quality audit at 10 Ontario hospitals to determine the optimal measure for assessing appropriateness. Transfusion; 2016; 56, 10 pp. 2466-2476.1:CAS:528:DC%2BC28Xhs1Krsr3M [DOI: https://dx.doi.org/10.1111/trf.13737] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27465485]

31. Shander, A et al. Appropriateness of allogeneic red blood cell transfusion: the international consensus conference on transfusion outcomes. Transfus. Med. Rev.; 2011; 25, 3 pp. 232-246. [DOI: https://dx.doi.org/10.1016/j.tmrv.2011.02.001] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21498040]

32. Desai, N; Schofield, N; Richards, T. Perioperative patient blood management to improve outcomes. Anesth. Analg; 2018; 127, 5 pp. 1211-1220. [DOI: https://dx.doi.org/10.1213/ANE.0000000000002549] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29064875]

33. Filipescu, D et al. Perioperative patient blood management Programme. Multidisciplinary recommendations from the patient blood management initiative group. Rom J. Anaesth. Intensive Care; 2017; 24, 2 pp. 139-157. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29090267][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5642854]

34. Frank, SM et al. Variability in blood and blood component utilization as assessed by an anesthesia information management system. Anesthesiology; 2012; 117, 1 pp. 99-106. [DOI: https://dx.doi.org/10.1097/ALN.0b013e318255e550] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22531332]

35. National Blood Authority. Patient Blood Management Guidelines: Module 2 Perioperative Australia: National Blood Authority (2012). https://www.blood.gov.au/pbm-module-2.

36. Goodnough, L. T. & Shander, A. Patient blood manage. Anesthesiology116 (6), 1367–1376 (2012).