Introduction

Blood loss in patients undergoing THA can be substantial and induce postoperative complications [20]. Surgical site bleeding and hematomas may cause nerve compression, prolonged wound drainage, infection, extended hospital stay, and extended rehabilitation [21, 22]. Anemia may aggravate vascular diseases [4], and homologous blood transfusions carry a small risk of infection, immunologic reactions, and fluid overload and may increase costs [19]. Surgical bleeding depends on the magnitude of surgery, the procedure, and patient-specific factors.

In THA, deep vein thrombosis (DVT) and other thrombin-driven events may originate during surgery and in a few patients nonfatal or fatal pulmonary embolisms (PE) develop. General chemical prophylaxis therefore has been recommended [10, 27]. Type of antithrombotic drug, dosage, and timing of the first dose may influence bleeding and development of thrombosis. Low-molecular-weight heparins (LMWHs) are widely used antithrombotics because of their favorable efficacy-to-safety profiles [9, 23], but the best timing of the first dose remains controversial [24, 28]. Preoperative initiation 12 hours before surgery has been based on the premise that DVT starts during surgery and that preoperative initiation is necessary to optimize antithrombotic effectiveness [17, 27]. In contrast, the premise for prophylaxis started after surgery has been to avoid the potential for increased bleeding complications. In clinical trials on antithrombotic regimens, venographically detected DVT has been a primary end point and bleeding a secondary underpowered outcome. Owing to various bleeding definitions, these trials have been criticized for underestimating the risk of bleeding and related complications [6, 7, 14].

We therefore asked if (1) there is a clinically important difference in total blood loss in THA between preoperative or postoperative start of thromboprophylaxis, and (2) there is a difference between the two regimens in transfusion requirements, incidence of bleeding events, and other complications detected up to 6 months after surgery.

Patients and Methods

After this study was approved by the regional ethics committee, we prospectively randomized 80 patients 50 years or older who underwent cemented THA for primary osteoarthritis between March and June 2008. During that same time, we treated a total of 104 patients with primary cemented THAs for osteoarthritis. Exclusion criteria were allergy to LMWH, bleeding disorders, renal failure, hepatic disease, active treatment for malignancy, ongoing antithrombotic treatment, history of DVT or PE, and patients experiencing major operations, traumas, stroke, or cardiac infarction the last 3 months before surgery. All patients were routinely hospitalized the day before surgery. We excluded 10 patients from enrollment owing to antithrombotic treatment, five patients with a history of DVT or PE, and two patients with liver disease. Seven patients refused to participate in the study. This left 80 patients for study. None of the 80 patients was lost to followup 6 months after surgery, and data collection was completed for all participants.

We performed a power analysis based on two earlier studies in which a significant reduction in the number of patients who had transfusions showed a 30% reduction in total blood loss [3, 15]. We considered this reduction clinically relevant. The effect size was based on blood loss and transfusion requirements in two earlier studies [3, 15]. In a prospective controlled study on patients who had THAs, Johansson et al. [15], found a 27% reduction in total blood loss (355 mL) reduced (p = 0.009) the number of patients who had transfusions. In a retrospective study [3], we found a 30% reduction in total blood loss (370 mL) reduced (p = 0.006) the number of patients who received transfusions by 28%. We believe a reduced risk for being exposed to blood transfusions is clinically relevant. With 80% power (alpha = 0.05), at least 37 patients were required in each group. We randomized a total of 80 patients to compensate for patients who might withdraw consent.

There were no differences between patients who received preoperative or postoperative start of thromboprophylaxis in terms of demographics (Table 1). We found no difference in operative times and length of hospitalization (Table 2), and preoperative laboratory values also were similar (Table 3).

Table 1 Patient characteristics
Full size table
Table 2 Surgery time, blood loss, and days of hospitalization in preoperative and postoperative groups
Full size table
Table 3 Preoperative and postoperative hemoglobin and hematocrit values
Full size table

In the hospital’s written patient information, patients are advised to stop antiplatelet medication (ie, NSAIDs and high-dose aspirin) 1 week before surgery. A complete record of the patients’ medications during the study period was recorded.

We assigned patients to either 5000 IU dalteparin (Fragmin®; Pharmacia and Upjohn, Stockholm, Sweden) subcutaneously or placebo (saline) injected 12 hours before surgery. All patients had 5000 IU dalteparin subcutaneously 6 hours after surgery and each day until the 35th postoperative day. Randomization was prepared in blocks of 20. Treatment group assignment was concealed by the hospital staff. The syringes with 5000 IU dalteparin and placebo with the same volume in each syringe were prepared by a study nurse who otherwise was not engaged in the study, according to randomized strata. The injection was blinded to the investigator, hospital staff, and the patient. The study blinding was broken after all patients had completed 6 months of followup.

All patients received spinal anesthesia without hypotensive effect with 5 mg/mL bupivacaine (Marcain®; AstraZeneca, Södertälje, Sweden) injected at the lumbar level. Cephalothin (Keflin®; EuroCept Pharmaceuticals BV, Kortenhoef, The Netherlands) 2 g was administered within 30 minutes of the arthroplasty. An equivalent dose subsequently was given 3 hours, 9 hours, and 15 hours after surgery as prophylaxis against infection. Voluven® and Ringer’s acetate (Fresenius KABI, Bad Homburg, Germany) were used as plasma substitutes.

The operation was performed with the patient in the lateral position, using a standardized posterior approach where only the piriformis muscle was detached and with capsular repair at the end of the procedure. All procedures were performed by two surgeons with at least one being experienced in performing THAs. All patients received stem and cups (Exeter®; Stryker Orthopaedics, Mahwah, NJ, USA) embedded in Simplex® tobramycin bone cement (Stryker Howmedica, Limerick, UK).

Postoperative analgesia was administered according to a standard protocol consisting of paracetamol + codeine sulfate (Paralgin forte®; Weifa AS, Oslo, Norway) and ketobemidone (Ketorax®; Jenahexal Pharma, Jena, Germany). Closed postoperative drainage was used for 24 hours. All patients were mobilized on the first postoperative day, and a program for simple self-administrated exercises was provided by the physiotherapists during hospitalization. Walking with the use of crutches was advised 6 to 8 weeks after surgery. Regular outpatient physiotherapy was not recommended until 2 months after surgery. We did not allow concomitant mechanical prophylaxis against DVT.

Hemoglobin and hematocrit were measured during surgery, before the first postoperative injection of dalteparin, and on postoperative Days 1, 3, and 6. We recorded the number of blood transfusions and plasma substitutes. The primary outcome was the volume of blood loss measured by weighing sponges and drapes (1 mg = 1 mL), volume in suction drains during surgery, and wound drains until removal 24 hours postoperatively [20]. We also recorded the number of patients who received transfusions, consumption of units of allogeneic leukodepleted erythrocyte concentrate, and decrease in hemoglobin concentration postoperatively in the two groups. We used a standard protocol with transfusion thresholds where a hemoglobin level less than 8 g/dL triggered transfusion and patients with a level greater than 10 g/dL did not receive a transfusion. Hemoglobin level on its own may be a poor indicator of tissue hypoxia, and the decision to transfuse patients with hemoglobin between 8 and 10 g/dL will, to some extent, be influenced by other parameters such as concomitant disease, weight, age, and others [1]. RBCs were given in 300-mL units, and autologous blood was not used. We evaluated all patients on a daily basis during hospitalization for possible bleeding events, such as hematoma, ongoing excessive bleeding, prolonged wound drainage (greater than 7 days), infections, and other complications. Overall surgical complications were classified according to Dindo et al. [8].

If patients showed any clinical sign of thromboembolic events, such as respiratory distress, chest pain, unstable hemodynamics, and a swollen, red, painful leg, we performed objective tests, including ECG, blood gases, plain chest radiography, venography, and spiral CT, after a clinical examination. Routine ultrasound screening, venography, or CT was not performed. A clinical research file (CRF) (Appendix 1; supplemental materials are available with the online version of CORR) was completed on a daily basis during hospitalization and at 6 months’ followup.

The data are presented as mean, range, and 1 SD or 95% CI. Patient characteristics, blood loss, hemoglobin, hematocrit, fluid volume, and operation time were compared between the two groups using Student’s t-test. We used the Mann-Whitney U test to compare the number of blood transfusions. Chi-square and Fisher’s exact tests were used to compare frequencies. We used SPSS Statistics Version 17 (IBM Inc, Chicago, IL, USA) for all analyses.

Results

The total volumes of blood loss during surgery and until drain removal were similar (p = 0.460) in the preoperative and postoperative prophylaxis groups: 1081 mL ± 424 mL versus 1023 mL ± 238 mL, respectively (Table 2). Blood loss until the first postoperative injection of dalteparin (p = 0.202) or after the first postoperative injection and until removal of drains (p = 0.471) also was similar. We observed the same decrease in hemoglobin in the two groups (Table 3). Decreases in hemoglobin greater than 2.0 g/dL were measured for 82.5% versus 90% of patients during surgery and until the first postoperative injection and 92% versus 90% of patients during surgery and the day after surgery. There were no differences in hematocrit between the two groups at any time. Neither hemoglobin nor hematocrit had recovered to preoperative levels on Postoperative Day 6.

We found no difference in the total amount of transfusion requirements among the groups. More (p = 0.099) patients in the preoperative group received blood transfusions during hospitalization (12 of 40 versus five of 40). Both groups received a similar (p = 1.000) number of RBC units until the first postoperative injection of dalteparin (two with two units of packed red cells and two with one unit in both groups) (Table 4). Altogether, 27 units of RBCs were transfused in the preoperative versus 11 units in the postoperative group (p = 0.071). The volumes of colloids and fluids were the same.

Table 4 Transfusion, units of RBCs, and other fluids
Full size table

In the preoperative group, four patients had bleeding-related events (Table 5). Three patients had wound hematomas, of which one was evacuated 5 days after surgery. One patient had wound drainage leading to prolonged hospitalization, but there were no positive bacteriologic cultures indicating infection. In the postoperative group, one patient had excessive bleeding after surgery, which was treated by surgical hemostasis. Four patients experienced hematomas and two had wound drainage without a positive culture that resulted in prolonged hospitalization. In the preoperative group, two patients had clinically suspected venous thrombosis not venographically confirmed. One patient with chest pain was examined by the cardiologist without obtaining any specific diagnosis. None had suspected PE during 6 months of followup. In the postoperative group, one patient had clinical and radiographically (spiral CT) confirmed PE 6 days after surgery and was treated according to protocol. The number of patients with other complications was low. In the preoperative group, one patient experienced a deep infection 3 months after surgery, and one patient dislocated the hip 2 months after surgery. In the postoperative group, one patient was admitted to another hospital with ileus, which spontaneously resolved. According to the classification of Dindo et al. [8] of surgical complications, three versus seven complications were Grade 1, 13 versus seven were Grade 2, and three versus zero were Grade 3 in the preoperative group versus the postoperative group.

Table 5 Complications recorded during surgery and until 6-month followup
Full size table

Discussion

Pharmacologic thromboprophylaxis is recommended in major orthopaedic surgery but potentially may increase bleeding and transfusion requirements, which makes its use controversial [18]. LMWH has been associated with increased operative blood loss and transfusions [29], and to reduce bleeding and its side effects, the first injection has been postponed until after surgery. However, the scientific basis for such a change in practice is uncertain and needs further attention. In this double-blind, randomized study of patients undergoing THA, we compared preoperative with postoperative start of 5000 IU dalteparin. We asked if (1) there is a clinically important difference in total blood loss in THA between these two approaches of thromboprophylaxis, and (2) there is a difference between the two regimens in transfusion requirements, incidence of bleeding events, and other complications detected up to 6 months after surgery.

There are some limitations to this study. First, our sample size was small, but it was powered to detect a difference in total blood loss as the primary outcome of the study and with clinical relevance related to transfusion [3, 15]. Second, a proportion of blood loss after surgery is hidden. The volume of hematomas is difficult to estimate clinically or by ultrasonography, although several methods and indices for calculation of hidden blood loss have been proposed [20, 26]. However, any underestimation of such masked blood loss should be equally distributed randomly in groups. We recorded hemoglobin, hematocrit, and volume of fluids transfused at fixed times during and after surgery and found these recordings consistent across the two groups and followed the same pattern. Third, our transfusion guidelines leave a gray zone between upper and lower transfusion thresholds where transfusion decisions are based on numerous factors, including preoperative and postoperative hemoglobin levels, age, BMI, additional comorbidities, physicians preferences, and others [1, 2, 25]. However, as there were no differences in any parameters between the two groups, we consider the decision to transfuse to be equal in the two groups. Fourth, we did not use predefined classifications of bleeding events. Variability in reporting of these events makes it difficult to compare between trials. Therefore, we rather described them in clinical terms, and there were no major differences between our two groups.

Trials of thromboprophylactic agents have shown wide variation in bleeding definitions and recording of bleeding events, and together with lack of statistical power, this may have resulted in misleading interpretation of the findings for bleeding [7, 12, 14]. This inconsistency also makes it difficult to draw conclusions from meta-analyses and make recommendations. In The North American Fragmin Trial, 2500 IU dalteparin was given either 1 hour before or 6 hours after surgery and compared with warfarin initiated 12 to 24 hours postoperatively [13]. Different surgical procedures were included, ie, primary THA and revisions. Fewer radiographic DVTs were recorded for both dalteparin regimens compared with warfarin. Predefined bleeding events were similar in all groups, but the proportion of patients receiving transfusions was greater for the dalteparin groups, particularly for those receiving dalteparin preoperatively. Consequently, a 6-hour postoperative dalteparin regimen was recommended even if the study was underpowered to assess the trial-specified bleeding. In a retrospective study, we found a reduction in total blood loss during THA when the first dose of dalteparin was postponed from 12 hours before to 6 hours after surgery [3], but the clinical importance of this reduction was questioned. In the current study in which the same drug and dose were compared, we could not find differences in total blood loss during and after surgery or in decrease in hemoglobin or hematocrit at any time until discharge after approximately 1 week. Other parameters that could influence blood loss, such as operative time, type of surgery, age, and BMI, were similar in the two study groups and strengthen our observations.

Our overall incidence of bleeding complications was greater than reported by others [18], which could reflect the method of collecting and classifying data. Some researchers use a decrease in hemoglobin greater than 2 g/dL in the definition of major bleeding [11], and the rate of major bleeding frequently is reported as a safety outcome in trials of thromboprophylaxis [9, 13, 16, 23]. We found such a decrease in hemoglobin for the majority of patients until the first postoperative injection (83% versus 90%) and the day after surgery (93% versus 90%), which indicates a decrease in hemoglobin is a poor parameter for defining major bleeding in patients undergoing THA.

Many patients undergoing major orthopaedic surgery receive blood transfusions (Table 6). A transfusion frequency of 30% to 40% has been reported in a publication regarding new anticoagulants [16], and a review on transfusion decision-making reported between 16% and 50% of patients who had THAs received transfusions [1]. We found similar percentages: 30% with preoperative and 12.5% with postoperative thromboprophylaxis. Our study was not powered to show differences in transfusions and we observed no differences in total units transfused, frequency of transfusions, or number of patients who had transfusions. Parameters known to affect transfusion requirements, such as preoperative and postoperative hemoglobin level, American Society of Anesthesiologists physical status classification, weight, and age [1], were similar in both groups and should not influence the results. However, fewer patients received blood transfusions and the number of RBC units transfused was less in the postoperative group.

Table 6 Comparison of blood loss and transfusion requirements
Full size table

Thrombosis formation begins at the time of surgery in THA, and it follows that efforts to prevent the formation of thrombi should begin as early as possible. The timing of initiation of pharmacologic prophylaxis is a clinical decision that should consider the risk of venous thromboembolism and bleeding associated with antithrombotic therapy. We found no differences in blood loss when 5000 IU dalteparin was initiated 12 hours before or 6 hours after primary THA. However, we observed a trend toward fewer transfusions with postoperative start.