Introduction

An increase in human life expectancy has resulted in an increased in aging population, thereby leading to a dramatic increase in the number of patients with knee diseases. Knee osteoarthritis not only affects the quality of life of patients but also increases the burden on medical resources. Total knee arthroplasty (TKA) is often recommended for patients with severe knee pain and disability caused by conditions such as osteoarthritis, rheumatoid arthritis, or post-traumatic arthritis [1, 2]. The goal of TKA is to relieve pain, improve function, and restore mobility in patients with severe knee osteoarthritis or other knee conditions that have not responded to non-surgical treatments [3, 4]. Although TKA is an efficient and generally successful surgical method that provides considerable benefits to patients with severe knee problems, it is associated with certain risks. Surgical risks include, but are not limited to, infection, bleeding, thrombosis, and possible damage to peripheral nerves or blood vessels. Although rare, these complications can seriously affect the surgical outcomes and patient recovery [5].

Studies have shown that 18%–67% of patients require allogeneic blood transfusion after TKA to prevent complications such as myocardial ischaemia and cerebral infarction caused by severe anaemia [4, 6]. TKA has a high perioperative blood loss, ranging 300–1,000 mL and up to 2 L in some cases [7]. Anaemia and blood transfusion requirements after TKA are caused by many factors. First, TKA is a direct cause of anaemia. Blood loss is a direct result of tibiofibular dissection, bone-surface contouring, and soft-tissue dissection. In addition, patients may experience hidden post-operative blood loss, such as surgical site oozing and haematoma formation, which may not be immediately reflected but may significantly impact the patient’s overall haemoglobin (Hb) level [8]. The knee joint is the largest synovial joint in the body. TKA is an extremely difficult procedure that requires the removal of a large number of diseased synovial membranes and osteotomy during surgery, resulting in intraoperative blood loss exceeding that in conventional orthopaedic surgery [9]. Simultaneously, post-operative hidden blood loss is also occurs easily [10].

Although measures are taken to minimise blood loss, some bleeding is inevitable, and patients may require blood transfusions to replace the lost blood [11]. Surgeons typically perform a complete blood count (CBC) on the first post-operative day to assess the presence of post-operative anaemia. In a study involving 628 patients who underwent primary TKA, 956 CBC tests were performed, costing $116,804 [12]. Given the increasing number of TKA surgeries and the cost of post-operative laboratory testing, reducing the number of post-operative CBC tests has the potential to save hundreds of millions of dollars annually for the U.S. National Health Care System [13]. Although transfusion is necessary in some cases, the transfusion process is not risk-free; adverse events that may occur include transfusion-associated acute lung injury, transfusion-associated circulatory overload, transmission of infectious diseases, and immune-mediated reactions [14]. These adverse events may not only prolong a patient's hospital stay and increase medical costs but, in extreme cases, may endanger a patient’s life.

The operation duration taken for primary TKA is one of the most important metrics used to evaluate the quality of orthopaedic surgeries. Appropriate operation duration has several potential benefits for patients [15]. The longer patients undergo TKA, the higher their risk of blood loss, blood clots, and infection [16, 17], because longer surgical procedures result in greater blood exposure and loss and the inflammatory response caused by surgical trauma is more intense, leading to a greater risk of blood loss. Additionally, prolonged surgery may result in increased tissue and vessel damage, further increasing blood loss. However, speeding the surgery may result in less attention being paid to details. For example, intraoperative haemostasis is inadequate. Furthermore, to shorten the operation duration, the surgeon may adopt a more aggressive surgical approach, which can increase the risk of damage to important blood vessels. Therefore, a shorter operation duration does not always mean reduced transfusion risk but may increase the probability of perioperative transfusion.

To the best of our knowledge, only a few studies have independently evaluated the effect of operation duration on transfusion events in patients undergoing primary TKA after adjusting for other covariates. Moreover, these studies have small sample sizes and large biases. Thus, in this study, we investigated the relationship between operation duration and blood transfusion risk in patients undergoing primary TKA, thereby helping surgeons develop a safe and effective surgical strategy. We hypothesised that the longer a patient underwent primary TKA, the greater the blood loss and the higher the risk of perioperative transfusion.

Materials and methods

Data source and study design

This study retrospectively analysed data of 2,622 patients who underwent primary TKA at a single centre in Singapore between January 2013 and June 2014. The exclusion criteria of this study were as follows: patients with Hb levels < 8 g/dL (N = 5), > 85 years old (N = 16), who had undergone revision TKA (N = 22), and who were rehospitalised within 30 days (N = 17). Sixty patients met the exclusion criteria and were excluded from the final statistical analyses (Fig. 1).

Fig. 1
figure 1

Study flowchart

Full size image

The patients were categorised into three groups based on operation duration: < 90 min, 90–120 min, and > 120 min. Using these different cut-offs allows for the simple risk stratification of patients. At the same time, this can clearly show the impact of prolonged surgery duration on transfusion risk in patients undergoing primary TKA. Some investigators have also used these operation duration cut-offs to stratify patients undergoing other orthopaedic surgeries [18,19,20,21]. This study was approved by the Beijing Ditan Hospital of the Capital Medical University Ethics Committee. The Ethics Committee waived the informed consent requirement because the data are available publicly via the “DATADRYAD” and were analysed anonymously. The Ethics Committee waived the requirement for informed consent.

Definition of the variables

The outcome variable was the number of blood transfusion events after the primary TKA. A blood transfusion event was defined as a serum Hb level < 8.0 g/dL. Blood transfusion is also performed for patients with symptoms of anaemia or any organ dysfunction related with anaemia and with serum Hb level < 10.0 g/dL [22,23,24]. The exposure variable was the operation duration (min). Covariates in our study included patient demographics such as age, sex (male/female), race (Chinese/other), body mass index (BMI), smoking status (no/yes), and pre-operative Hb levels; comorbidity information, such as individual components of the revised risk cardiac index including cerebrovascular accident (CVA, no/yes), ischaemic heart disease (IHD, no/yes), congestive heart failure (CHF, no/yes), diabetes mellitus (DM, no/yes). pre-operative creatinine level > 2 mg/dL (no/yes), operation details such as American Society of Anesthesiologists Physical Status (ASA-PS) score, operation procedure (unilateral or bilateral), type of anaesthesia (general or regional), and day of week the surgery was performed.

The perioperative period was defined as two weeks before surgery and two weeks after the primary TKA. In most cases, patients were admitted to the hospital on the day of surgery. Particularly, all antiplatelet drugs, except aspirin, should be discontinued before TKA. To prevent thrombosis, each patient received a daily subcutaneous injection of 40 mg of low-molecular-weight heparin after surgery, which was discontinued after discharge. All patients underwent a standardised post-operative care protocol. Discharge criteria were defined as the ability to climb a few flights of stairs, walk a few meters with the aid of an assistive device after surgery, and being able to achieve flexion of the operated knee close to 90°.

Statistical analysis

All analyses were performed using R software (version 4.1.3, http://www.r-project.org). For continuous variables, data are presented as mean enstandard deviation or the interquartile range and use one-way analysis of variance or Kruskal–Wallis U test for differences between two groups. Categorical variables were presented as percentages (n, %). The chi-square and Fisher’s exact tests were used to compare differences between the groups. A univariate logistic regression model was used to explore the potential variables that were risk factors associated with transfusion after primary TKA. A multivariate logistic regression model was used to calculate the odds ratio (OR) and 95% confidence interval (CI) to estimate the relationship between TKA operation duration and blood transfusion risk after adjusting for other covariates. In the crude model, no adjustment was done for any covariables. In model 1, we adjusted for age, sex, and race. In model 2, we adjusted for age, sex, race, smoking, DM, IHD, CHF, CVA, and creatinine level. In model 3, we adjusted for surgical details such as ASA score, operation procedure, anaesthesia type, surgical date, and of all covariates adjusted in model 2. Furthermore, to explore whether the relationship was stable when operation duration was transformed from a continuous variable to a categorical variable, we divided the patients into two groups (TKA operation durations 90 and 120 min).

Subgroup and interaction analyses were performed using stratified logistic regression models by age (< 50 years, 50–70 years, > 70 years), sex (male/female), race (Chinese/other), BMI (< 25/25–30/ > 30 kg/m2), Hb level (< 11/11–13/ > 13 g/dL), ASA status, anaesthesia type, and procedure description. This approach ensured that the stability of the relationship between operation duration, transfusion risk across different patient subgroups, and other variables affecting the relationship could be examined. To account for the non-linear relationship between operation duration and blood transfusion risk, we also used a generalised additive model and smooth curve fitting (restricted cubic spline, RCS) to address non-linearity. All statistical tests were two-sided, and a P value < 0.05 was considered statistically significant.

Results

General characteristics of the study population

Table 1 shows the baseline characteristics of the participants. A total of 2,562 patients were included in our study. Among these, 136 (5.61%) experienced perioperative blood transfusion events and 1,579 (61.63%), 762 (29.74%), and 221 (8.63%) patients had an operation duration < 90, 90–120, and > 120 min, respectively. The mean age was 66.14 ± 8.01 years, 609 (23.77%) were male, and 2,152 (84.00%) were Chinese. The BMI was 27.81 ± 5.65 kg/m2. The intergroup differences in age, sex, diabetes mellitus, anaesthesia type, procedure description, date of surgery, and transfusion were significant.

Table 1 Demographic baseline characteristics of all participants
Full size table

Results of the univariate logistic regression analysis of the risk of blood transfusion in patients who underwent TKA

As show in Table 2, we found that age (OR = 1.013; 95% CI = 1.01–1.06, P = 0.003), BMI (OR = 0.94; 95%CI = 0.90–0.98, P = 0.004), ASA status (OR = 4.22; 95%CI = 1.61–13.2, P = 0.008), Hb level (OR = 0.60; 95%CI = 0.53–0.67, P < 0.001), CHF (OR = 5.41; 95%CI = 1.76–13.9, P = 0.005), creatinine > 2 mg/dL (OR = 8.6; 95%CI = 2.97–22.2, P < 0.001), anaesthesia type (OR = 1.91; 95%CI = 1.35–2.70, P = 0.001), and procedure description (OR = 3.31; 95%CI = 2.09–5.10, P < 0.001) were risk factors for blood transfusion events (P < 0.05).

Table 2 the univariate logistic regression analysis of the risk of blood transfusion in patients under going TKA
Full size table

Relationship between operation duration and blood transfusion risk

The results of the multivariate logistic regression analysis (as is show in the Table 3) showed that longer operation duration was associated with an increased risk of blood transfusion. This association was consistent in our crude model (OR = 2.202; 95%CI = 1.588–3.018, P < 0.001), model 1 (OR = 2.334; 95%CI = 1.677–3.21, P < 0.001), and model 2 (OR = 2.414; 95%CI = 1.697–3.396, P < 0.001). In the fully adjusted model 3, the positive association between operation duration and transfusion-risk events remained stable (OR = 1.87; 95% CI = 1.174–2.933, P = 0.007).

Table 3 The relationship between operation time and blood transfusion events in patients undergoing TKA
Full size table

Furthermore, we transformed the operation duration from a continuous to a categorical variable (grouped according to 90 and 120 min) for sensitivity analysis (Table 3). Compared with the transfusion-risk events with an operation duration < 90 min, that with an operation duration > 120 min increased 2.141-fold in the fully adjusted model (OR = 2.141; 95%CI = 1.035–4.265, P = 0.035). The trend test showed that operation duration (transformed from a continuous to a categorical variable) was also positively associated with blood transfusion risk (P < 0.001).

Subgroup and interaction analyses between operation duration and blood transfusion risk

As shown in Table 4, the subgroup and interaction analyses showed that in certain special subgroups, such as age > 50 years, Chinese sex, BMI < 30 kg/m2, Hb level > 11 g/dL, ASA status 2 and 3, general anaesthesia, and unilateral TKA, the transfusion-risk events increased with an increase in operation duration (P < 0.05). Furthermore, the interaction tests showed that these variables had no significant effect on the relationship between the two variables ( P > 0.05).

Table 4 Subgroup analysis and interaction analysis of operation time and blood transfusion in patients under going TKA
Full size table

Non-linear relationship between operation duration and blood transfusion risk in patients who underwent TKA

We observed a non-normal distribution of operation duration among patients (Fig. 2a). The operation durations were 95.55 ± 36.93 and 83.86 ± 26.29 min for blood transfused and non-blood transfused patients. Additionally, a clear difference existed in the operation duration between the transfusion and non-transfusion groups (Fig. 2b).

Fig. 2
figure 2

Distribution of operation duration (a); A violin-plot of the operation duration of the transfused and non-transfused groups (b); The non-adjusted relationship between operation duration and blood transfusion events (c); The adjusted relationship between operation duration and blood transfusion events (d)

Full size image

Using smooth curve-fitting analysis and the RCS, our study demonstrated a non-linear (P = 0.0197) and “J”-shape relationship between operation duration and blood transfusion risk in the crude model (Fig. 2c). The non-linear (P = 0.030) and “J”-shape relationship still existed after adjustment for age, sex, and race (Fig. 2d). Furthermore, we found a “threshold effect” in this relationship. The inflection point for operation duration was approximately 73.2 min, and the risk of blood transfusion increased when the operating duration was less or greater than the threshold.

Discussion

We performed a secondary analysis based on a previously published dataset to investigate the risk factors associated with blood transfusions in patients undergoing primary TKA. We found that patients who underwent primary TKA had a high risk of perioperative blood transfusion (5.61%). Age, BMI, ASA status, Hb level, OSA, CHF, creatinine level > 2 mg/dL, and anaesthesia type were risk factors for blood transfusion. Furthermore, the duration of primary TKA was an independent predictor of perioperative transfusion after adjusting for other covariates. Compared to those with an operation duration < 90 min, patients with an operation duration > 120 min had a 2.141-fold increased risk of blood transfusion. Importantly, we found a threshold effect on this relationship: perioperative transfusion rates were the lowest when the operation duration approached 73.2 min. In other words, perioperative transfusion risk increased when the duration of surgery was less than or greater than 73.2 min.

Primary TKA is one of the most common orthopaedic procedures performed in the lower extremities. However, perioperative transfusion requirements are high owing to large joint cavities, procedural complexity, and intraoperative bleeding, which can increase patient suffering and costs [25, 26]. Perioperative blood transfusion is not just a medical procedure; it involves blood matching, screening, and monitoring, all of which increase the cost of medical institutions. At the same time, blood transfusion itself also has certain risks that may cause transfusion-related complications, such as transfusion reactions, infections, or allergic reactions, thus prolonging the hospitalisation duration and rehabilitation period of patients and increasing the medical burden. In countries with aging populations and advanced public health systems, the number of primary TKA operations performed each year is increasing exponentially, and more effort is urgently required to identify modifiable risk factors for perioperative transfusion to reduce patient suffering and medical costs [27].

Previous studies have shown that the prolonged operation duration is a predictor of all-cause mortality, surgical site infection, and perioperative blood transfusion after various orthopaedic procedures ranging from shoulder arthroscopy to primary TKA [21, 28, 29]. In a study of 11,806 patients who underwent primary TKA, Michael et al. [18]. reported that patients with an operative duration > 120 min were 3.2 times more likely to receive a transfusion. Similarly, data from China suggested that an operation duration > 85 min significantly increased blood transfusion risk in patients undergoing primary TKA [30]. Although the results of these studies demonstrated that prolonged surgery significantly increases transfusion requirements, shortening the operation duration for the sake of efficiency may also result in adverse events, because longer procedure times mean more blood exposure and loss during surgery, and the inflammatory response caused by surgery is more intense. These factors work together to increase bleeding risk and need for blood transfusion. Additionally, prolonged surgery may result in increased tissue and vessel damage, further increasing blood loss.

However, some studies have shown that shorter operation durations may lead to increased perioperative transfusion rates in patients undergoing primary TKA. Young et al. [19] found that patients did not benefit from a shorter operation duration and that an operation duration < 40 min increased the risk of greater invisible blood loss. This effect is attributed to inadequate intraoperative haemostasis, which may result from the orthopaedic surgeon’s excessive pursuit of speed during the surgery. Moreover, to shorten the operation duration, the surgical team may adopt a more aggressive surgical method, which can not only reduce the operation duration but also increase the risk of bleeding because it operates on a larger area of blood vessels and muscle tissue. The duration of surgery is related to a variety of variables, such as surgeon’s competence, patients’ conditions, and procedure complexity. Therefore, it is necessary to develop personalised surgical strategies and determine the appropriate operation duration to reduce the risk of perioperative blood transfusions.

The present study also suggests that increased blood transfusion requirements are associated with other well-established factors such as age, pre-operative anaemia, congestive heart failure, and bilateral primary TKA. Patients with advanced age and chronic complications are at a high risk of perioperative blood transfusion due to compensatory dysfunction, malnutrition, and low blood volume. Anaemia has a strong impact on the perioperative blood transfusion events [31]. A study from Singapore showed that the presence of pre-operative anaemia was an independent predictor of perioperative transfusion, with an OR of blood transfusion of 4.13 for mild anaemia and 9.13 for moderate to severe anaemia [27]. Compared to unilateral primary TKA, bilateral surgery appears to increase the risk of blood loss owing to the longer operation duration and larger exposed area [32]. Considering that both patients’ underlying conditions and procedure description are related to the operation duration, these variables may also increase transfusion-event incidence by prolonging surgery duration.

Notably, we found that in some subgroups, the risk of perioperative transfusion events did not always correlate with operation duration. Interaction tests showed that the type of general anaesthesia used had a significant effect on this relationship. Chapman et al. [33] found that regional anaesthesia significantly reduced post-operative complications, hospital stay, and perioperative transfusion compared with that by general anaesthesia. An analysis from the National Joint Registry reported that regional anaesthesia was beneficial in reducing the length of stay and complications after total hip replacement and TKA and noted that regional anaesthesia could be used as a reference standard for patients undergoing these procedures [34]. However, the findings of other studies are not completely consistent with our results. According to a study by Riku et al. [35], spinal anaesthesia was associated with a higher incidence of post-operative pain and vomiting than general anaesthesia, and no difference existed in perioperative blood transfusion. Thus, the relationship between transfusion events and anaesthesia type needs further exploration.

Our study has several advantages. First, to the best of our knowledge, this is the first study to assess the association between the operation duration for primary TKA and transfusion risk. Second, we used multiple logistic regression models and subgroup analyses to explore the relationships after adjusting for other covariates. Importantly, we found a non-linear relationship with a “J” shape; further, a threshold effect on the relationship was observed, with transfusion risk lowest at 73.2 min of surgery duration. Third, to the best of our knowledge, this study is the first to demonstrate that excessive pursuit of speed and short surgery duration may lead to greater vascular and muscle tissue damage and inadequate haemostasis, resulting in increased transfusion risk, thus providing evidence to support clinical decisions.

Our study has a few limitations. First, the study was retrospective and used clinical data from a single centre from 2013 to 2014. Primary TKA performed a decade ago is different from that performed today in terms of surgical technique and drain or tourniquet strategy, which may also have affected the operation duration and transfusion risk. Second, some potential factors affecting blood loss were not included; for example, all patients received standardised perioperative treatment: 3 g tranexamic acid, intravenously, prior to TKA, 2 g tranexamic acid, intraoperatively, into the knee joint cavity, and no drainage tube was placed. Therefore, this covariate was not included in this study. However, we believe that our study provides valuable insights into the factors influencing blood transfusion risk in patients undergoing primary TKA, including the significant relationship identified between operation duration and transfusion events.

Conclusion

Our study found that the surgical duration was significantly associated with transfusion risk in patients undergoing primary TKA. Importantly, we found a “J-shaped” relationship and threshold effect between the two variable. The patients had the lowest risk of perioperative transfusion events after primary TKA when the operation duration was 73.2 min; a shorter operation duration implies irregular surgical procedures and incomplete intraoperative haemostasis, which lead to increased perioperative blood loss and blood transfusion. We believe that these results will be useful for clinical decision-making. However, the findings need further validation in a larger cohort.