Tranexamic acid (TXA) administration reduces bleeding and the risk of transfusion in patients undergoing lower-extremity total joint arthroplasty (TJA)1,2. The use of TXA has been supported by clinical practice guidelines for TJA3. However, there is a lack of data concerning the safety of TXA in patients with prior arterial thrombotic or prothrombotic conditions. Previous research has found that high-risk patients receiving TXA for TJA and arthroplasty for femoral neck fracture are not at a greater risk for complications2,4. However, those analyses did not include patients with intertrochanteric (IT) fracture.

The efficacy of TXA to reduce blood loss and not increase thromboembolic events in patients with IT fracture has been assessed previously5-9. However, those studies did not separately assess patients at high risk for these events, and in fact, this patient population was often excluded. The lack of data regarding TXA use for this patient population is concerning. There is a growing population of high-risk patients receiving TXA during their procedures10, exacerbated by an aging population with comorbidities who more frequently require anticoagulation11.

Therefore, with the current study, we aimed to investigate the safety of TXA administration in high-risk patients receiving surgical repair of IT fractures. For this study, safety was assessed by identifying differences in mortality, deep venous thrombosis (DVT), pulmonary embolism (PE), myocardial infarction (MI), and stroke within 90 days of surgery between patients who did versus did not receive TXA. We hypothesized that there would not be a difference in those outcomes between high-risk patients who received versus did not receive TXA.

After receiving institutional review board approval, we utilized International Classification of Diseases (ICD)-9 and ICD-10 codes to identify patients who underwent surgery for IT fracture between January 1, 2015 and December 31, 2019. Patients were excluded if they were identified to have periprosthetic fractures of the femur, pathologic fractures, or polytrauma. Also excluded were patients <18 years of age, those with a history of ipsilateral hip surgery with hardware, patients with hip fractures treated with percutaneous cannulated screws or femoral neck fractures, or American Society of Anesthesiologists (ASA) class-5 patients. In all, 1,147 patients who underwent surgery for IT fracture were identified (Fig. 1).

As previously defined2, patients designated as high risk were those who had a history of ≥1 of the following before surgery: DVT, PE, MI, stroke, a prothrombotic condition (Factor V Leiden, protein C deficiency, protein S deficiency, or antiphospholipid antibody syndrome), atrial fibrillation, atrial flutter, a coronary artery bypass graft, or a coronary artery stent. Of the 1,147 total patients, 564 (49%) were designated high-risk.

The primary independent variable for this study was the receipt of TXA during surgery. For all hip fracture patients over the study period who received TXA, the median dose was 2 g of intravenous (IV) TXA administered intraoperatively in 2 divided doses on incision and closure. Control variables were determined a priori12 on the basis of previous work demonstrating associations with the outcomes of interest13-17. The following patient variables were included: age, sex, body mass index (BMI), Charlson Comorbidity Index, ASA class, and preoperative tobacco use. Additionally, patient preoperative hemoglobin level, time between admission and surgery, hospital site (masked as A, B, C, and D), year of surgery, receipt of blood transfusion during surgery or postoperatively, and the type of postoperative anticoagulants provided were collected. Our single health system consists of 3 tertiary care academic medical centers and a network of 16 community hospitals. Postoperative anticoagulants were grouped into 4 categories: (1) oral, (2) parenteral, (3) antiplatelet, or (4) P2Y12 inhibitors.

With regard to risk factors, there was little difference between the patients who received and those who did not receive TXA. The patients who received TXA were less likely to have a history of a coronary artery stent (3.5% versus 8.6%; p = 0.002). No other differences were noted (Table I).

Patients identified as high-risk are described in Table II. Those receiving TXA were more likely to have received care at site B (21.5% versus 2.2%) and less likely to have received care at site A (1.9% versus 13.5%) (p < 0.001). Additionally, those who received TXA were more likely to have received care in more recent years. Patients identified as low-risk are described in Table III. Among these low-risk patients, those receiving TXA were more likely to have received care at site B (15.3% versus 4.1%) and less likely to have received care at site A (3.2% versus 12.2%) or site D (28.0% versus 40.6%) (p < 0.001). In addition, those who received TXA were more likely to have received care in more recent years, had a slightly higher ASA class, and were more likely to have undergone surgery >48 hours after admission (13.8% versus 5.1%; p = 0.001).

Looking specifically at outcomes, among the high-risk patients, there was a smaller proportion who experienced DVT in the TXA group compared with the no-TXA group (0.0% versus 4.2%, respectively; p = 0.009). We noted no other differences between high-risk patients who did or did not receive TXA. In the low-risk cohort, there were no differences noted between the TXA and no-TXA groups in terms of mortality and 90-day medical complications.

Logit models were used to calculate propensity scores for TXA administration (administered versus not administered) for the high-risk cohort and the entire population. Nearest-neighbor matching was used to match patients based on TXA receipt (1:1 matching) with a caliper of 0.0618-20. Matching variables included patient age, sex, BMI, ASA class, Charlson Comorbidity Index, tobacco use, receipt of blood transfusion, postoperative anticoagulants, preoperative hemoglobin level, and the time between admission and surgery. The only difference between the 2 cohorts was that a risk category (high versus low) was included for the entire population cohort. After propensity matching, the standardized mean differences were less than the guideline maximum differences of 25%19, with 87% in the high-risk group and 100% in the all-patient group demonstrating a standardized mean difference of <10% (see Table IV). Additionally, c-statistics were close to 0.5 (0.65 for the high-risk group and 0.61 for the entire population; see Appendix Supplemental Table 1), indicating successful balance21. Differences in the occurrence of mortality, a DVT, PE, MI, or stroke within 90 days of surgery were assessed using the average treatment effect on the treated. Results were reported as adjusted odds ratios (ORs). An OR and 95% confidence interval (CI) exclusive of 1.0 were required for results to be considered significant. Stata/MP 16.1 (StataCorp) was used to conduct the analysis.

No funding was used for this research.

After matching, no differences were identified for any outcome between those who received versus did not receive TXA in the high-risk group (Fig. 2, Table V). Specifically, there were no differences in mortality (OR, 0.97 [95% CI, 0.90, 1.05]), DVT (OR, 0.97 [95% CI, 0.93, 1.00]), PE (OR, 1.00 [95% CI, 0.95, 1.05]), MI (OR, 1.04 [95% CI, 0.98, 1.10]), or stroke (OR, 1.00 [95% CI, 0.95, 1.05]).

Similarly, for the all-patients cohort, there were no identified differences between those who received versus did not receive TXA (Fig. 2, Table V). Specifically, there were no differences in mortality (OR, 0.97 [95% CI, 0.92, 1.02]), DVT (OR, 0.98 [95% CI, 0.96, 1.01]), PE (OR, 1.00 [95% CI, 0.97, 1.02]), MI (OR, 1.03 [95% CI, 1.00, 1.06]), or stroke (OR, 1.02 [95% CI, 0.98, 1.04]).

The results of our study revealed no association between TXA administration and mortality, DVT, PE, MI, or stroke within the high-risk patient cohort (n = 282 matched patients). Additionally, TXA was not associated with differences in mortality or serious postoperative complications within the entire population (n = 632 matched patients). As a result, this study provides support for TXA administration’s safety profile for all patients presenting for surgical repair of IT fractures.

Several small randomized controlled trials (77 to 122 patients) investigated the safety of TXA in hip fracture procedures6,7,9,22. However, these previous works either did not focus on or excluded patients at higher risk for thromboembolic complications, including those with a history of DVT, PE, MI, stroke, or a coronary stent6-8,22. Thus, the several meta-analyses summarizing the safety of TXA in hip fracture surgeries are limited to existing studies that have not focused on high-risk patients23,24.

Other prior retrospective work also evaluated the safety of TXA in hip fractures but likewise excluded high-risk patients. For example, in a study using the Premier Healthcare Database, a propensity-matched cohort (1:3) of 9,957 patients, of whom 2,002 were patients with IT fracture, revealed no differences in venous thromboembolism, renal failure, or MI when comparing those receiving TXA and those who did not25. While their work provides meaningful evidence toward the safety of TXA, their population did not focus on, nor subcategorize separately, high-risk patients. Furthermore, they did not evaluate outcomes past hospital discharge.

The current study provides evidence from a large single health system concerning TXA safety for high-risk patients with IT fracture. On the basis of the sample size of high-risk patients in this study, we estimate an ability to detect a 0.20 effect size with 90% power. As a result, our findings align with previous research indicating the safety of TXA in patients (high-risk or not) receiving TXA for IT fractures. Importantly, and consistent with practice guidelines for TJA, our practice uses a median dose of 2 g IV3.

While the current study was focused on safety, it is important to note that multiple previous studies have investigated the efficacy of TXA in patients with hip fracture. For example, in a population of patients with IT fracture undergoing surgical repair, Tian et al. randomized 100 patients to a dose of 10 mg/kg of IV TXA preoperatively and a second dose 5 hours postoperatively versus no TXA8. Their results revealed that patients receiving TXA had less total blood loss (182 mL) and were less likely to need a blood transfusion. Again, high-risk patients were excluded from this trial. Similarly, a meta-analysis of IV TXA efficacy in hip fracture surgeries revealed less blood loss (−273 mL; p < 0.0001) and a lower risk of blood transfusion (relative risk [RR], 0.66; p = 0.003)26. Hence, our study’s focus was not on the efficacy of TXA administration with respect to transfusion requirements, but on the remaining pertinent clinical questions regarding its safety in this nonelective, highly morbid patient population. Although there was a 3% lower rate of blood transfusion in both the high and low-risk TXA groups compared with the no-TXA group, this difference was not significant.

The current study is not without limitations. First, it relies on accurate coding within the institution’s electronic medical record (EMR), which could be inaccurate. While this is a common limitation for large, retrospective studies, it is worth noting the potential for bias25. Next, we did not explore the duration or timing of conditions qualifying patients for high-risk status. It is plausible that differences in these characteristics could contribute to outcomes, and they should be evaluated further in future studies. Similarly, we used several conditions aggregated to define high risk. This could potentially mask the associations between specific conditions and outcomes of interest. However, because of the limited sample size, we cannot specifically assess associations at this granular a level. We encourage additional research and the use of larger data sets to evaluate further if any differences in specific high-risk conditions are differentially associated with outcomes. Next, because of the increasing use of TXA in orthopaedic surgery over time, we could not balance our model by surgical year. This limitation has been noted by other authors as well25. Furthermore, we were unable to balance the model to include hospital site, and as a result, patient clustering may also bias the results. However, the 4 sites are geographically distant and have separate orthopaedic surgery departments.

In conclusion, in this propensity-matched study of high-risk patients receiving TXA during surgical repair of IT fractures, we found no evidence of increased mortality risk or other serious adverse outcomes. These results held for both the high-risk and entire-population cohorts.

Supporting material provided by the authors is posted with the online version of this article as a data supplement at jbjs.org (http://links.lww.com/JBJS/H29).