Utilization, effectiveness, and safety of tranexamic acid use in hip fracture surgery: A population-based study

Abstract

Objective

To assess the effect of tranexamic acid (TXA) use in hip fracture surgery.

Methods

A retrospective cohort study was performed using the Premier Healthcare database. A propensity score matching approach was applied to assess associations between TXA use and blood transfusion, perioperative complications, length of stay (LOS), and hospitalization cost.

Results

In 153,169 patients, TXA use was associated with a 17% decrease in odds of blood transfusion, no increase in the risk of perioperative complications, 16% shorter LOS, and minimal effects on hospitalization cost.

Conclusion

Our results are in support of a wider use of TXA in hip fracture surgery.

Level of evidence

Level III.

1. Introduction

Morbidity and mortality following hip fractures and subsequent surgery in the elderly remain a significant public health issue worldwide, with one-year mortality rates between 12% and 36%.1, 2, 3 Total blood loss in hip fractures, including intraoperative and hidden blood loss, may be as high as 1500 mL.⁴ While the incidence of hip fractures in the elderly has declined since the 1980s, comorbidities among these geriatric patients have increased, making them more susceptible to adverse events from blood loss.¹ The perioperative rate of blood transfusion in patients undergoing hip fracture surgery ranges between 20% and 60%.⁵ Blood loss and blood transfusion have been strongly associated with perioperative complications including infection, prolonged hospital stay, and increased cost.^4,6

One potential strategy for decreasing perioperative blood loss and blood transfusion is the use of tranexamic acid (TXA). TXA is an anti-fibrinolytic agent that blocks the lysine binding site of plasminogen. Studies across various surgical subspecialties suggest that TXA reduces perioperative blood loss and transfusion without an increased risk of thrombosis.7, 8, 9, 10 Despite its widespread use in total hip and knee arthroplasty, spine, and cardiac surgery, TXA is not routinely used in the United States among patients requiring hip fracture surgery. However, there has been increasing interest in its use in this patient population. Existing studies suggest that TXA reduces blood loss and transfusion in hip fracture patients with no evidence of increased rates of venous thromboembolic complications.11, 12, 13, 14, 15, 16, 17 The objectives of this large population-based study were to investigate: 1) the current utilization patterns of TXA in hip fracture surgery in the United States, and 2) the effectiveness of TXA use in reducing the odds of blood transfusions in the postoperative period and other associated complications using the nationwide Premier Healthcare claims database. We hypothesized that TXA use in hip fracture surgery would lead to reduced odds of transfusion without increased adverse effects.

2. Materials and methods

Data source. Data for this study were extracted from the Premier Healthcare database (Premier Healthcare Solutions, Inc., Charlotte, NC).¹⁸ This database contains administrative claims data on approximately 1 out of 5 hospital discharges in the United States. Data collected includes International Classification of Disease – 9th revision (ICD-9) codes, Current Procedural Terminology (CPT) codes, and complete inpatient billing items. This study was considered exempt from our institution's Institutional Review Board review.

Cohort selection. Our study cohort was selected based on ICD-9 codes for a femoral neck, intertrochanteric, or subtrochanteric fracture (ICD-9 diagnosis codes 820.00–09; 820.10–19; 820.21–2; 820.31–2; 820.8–9) and an ICD-9 procedure code for hemiarthroplasty, total hip arthroplasty, or an appropriate fixation procedure (ICD-9 procedure codes 00.70–7; 00.85–7; 81.5–3; 81.40; 78.55; 79.15; 79.25; 79.35; 79.55; 79.65; 79.85; 79.95) as previously used.¹⁹ We initially included procedures that were performed between 2006 and 2016 and found low (<0.1%) TXA use in earlier years. Therefore, for trend analysis and multivariable analysis, two separate cohorts were utilized, 2006–2016 and 2014–2016, respectively. The latter cohort was decided based on the first year with TXA use exceeding the 1% threshold; here, exclusion criteria included unknown gender (n = 242), unknown discharge status (n = 166), procedures classified as something other than “elective”, “emergency” or “urgent” (n = 7444), outpatient procedure (n = 163), and procedure performed at a hospital performing <30 procedures during the study period (n = 587) to ensure sufficient sample size per hospital.

Variables. Patient demographics information included age, gender, and race (white, black, other). Healthcare related variables included type of health insurance (commercial, Medicaid, Medicare, uninsured, unknown), hospital location (rural, urban), hospital bed size (<300, 300–499, ≥500 beds), teaching status of hospital, and annual hip fracture surgery volume. Procedure related variables included year of procedure, urgency of procedure (elective, urgent/emergent), use of peripheral nerve block, fracture type (femoral neck, intertrochanteric, subtrochanteric), type of surgery (internal fixation, hip hemiarthroplasty, total hip arthroplasty), type of anesthesia (general, combined neuraxial and general, unknown/missing), and surgery within 48 h of admission. Comorbidity burden was assessed using the Quan adaptation of the Deyo-Charlson comorbidity index.²⁰ Anticoagulant use was also assessed (aspirin, other anti-platelet agent, heparin, warfarin, other anticoagulant agent).

The main effect of interest was the utilization of TXA, which was defined using medication billing items. This was recorded as either “parenteral” or “miscellaneous,” with the latter representing intra-articular use. Dose distributions of TXA (1000 mg, 2000 mg, or ≥3000 mg) were reported. Primary outcomes were need for blood transfusion, perioperative complications during the hospital stay (acute myocardial infarction, acute renal failure, venous thromboembolism), LOS, and cost of hospitalization. Complications were defined using ICD-9 codes.

Statistical analysis. Given low use of TXA in earlier years, separate cohorts were used for trend analysis and multivariable analysis, 2006–2016 and 2014–2016, respectively. Univariate associations were measured between study variables and TXA use using standardized differences instead of p-values given the large sample size. A 1:3 propensity score match was performed using the 30 individual Elixhauser comorbidities.²¹ Standardized differences were also reported for the matched cohort to ensure balance of the individual covariates after the propensity score matching procedure. The chi-square test and t-test were used for categorical and continuous variables, respectively. Mixed-effects models were used to assess the association between TXA use and outcomes in the matched cohort. Odds ratios (OR) and 95% confidence intervals (CI) were reported for binary outcomes. Percent change and 95% CI were reported for continuous outcomes, and we applied the gamma distribution with a log link function (within PROC GLIMMIX) as these variables were highly skewed. All analyses were performed using SAS v9.4 statistical software (SAS Institute, Cary, NC).

3. Results

Baseline patient characteristics. The baseline patient characteristics in the unmatched and matched TXA and control groups (those without TXA use) are summarized in Table 1. Overall, covariate balance was improved after matching as demonstrated by standardized differences <10%, a commonly accepted threshold. In the matched cohort, there were 3319 hip fracture patients (70.2% female, 29.8% male) who received TXA and 9957 patients (71.0% female, 29.0% male) who did not receive TXA. The Charlson-Deyo comorbidity index was similar between the two groups, with approximately half of patients having a comorbidity index of 1 or higher. Within the TXA group, 94.9% received TXA via an intravenous (IV) route, and the majority of patients received a TXA dose of either 1000 mg (41.4%) or 2000 mg (43.0%). Within the matched cohort, 84% patients were treated at an urban hospital, and 62.9% of patients were treated at a non-teaching hospital.

Table 1.

Population characteristics of matched and unmatched samples with standardized differences between TXA and no TXA groups.

	UNMATCHED FULL COHORT					MATCHED SAMPLE
	TXA (n = 3812)		No TXA (n = 149,357)		Standardized Difference	TXA (n = 3319)		No TXA (n = 9957)		Standardized Difference
	n	%	n	%		n	%	n	%
TXA ROUTE
Intravenous	3583	94.0	–	–		3149	94.9	–	–
Unspecified	166	4.4	–	–		138	4.2	–	–
TXA DOSE			–	–				–	–
Not Specified or N/A	229	6.0	–	–		170	5.1	–	–
1000 mg	1514	39.7	–	–		1374	41.4	–	–
2000 mg	1664	43.7	–	–		1427	43.0	–	–
≥3000 mg	405	10.6	–	–		348	10.5	–	–
PATIENT DEMOGRAPHICS
Agea	79	69–86	82	72–88		80	70–87	80	70–87
Gender
Female	2653	69.6	103,898	69.6	0.071	2331	70.2	7071	71.0	1.720
Male	1159	30.4	45,459	30.4	0.071	988	29.8	2886	29.0	1.720
Race/Ethnicity
White	3305	86.7	128,054	85.7	2.794	2869	86.4	8556	85.9	1.484
Black	143	3.8	5971	4.0	1.277	127	3.8	363	3.6	0.953
Other	364	9.5	15,332	10.3	2.399	323	9.7	1038	10.4	2.302
HEALTHCARE RELATED
Insurance Type
Commercial	428	11.2	11,870	7.9	11.159	332	10.0	1097	11.0	3.308
Medicaid	117	3.1	4969	3.3	1.465	100	3.0	303	3.0	0.176
Medicare	3101	81.3	127,159	85.1	10.159	2756	83.0	8150	81.9	3.115
Uninsured	49	1.3	1749	1.2	1.039	41	1.2	121	1.2	0.183
Unknown	117	3.1	3610	2.4	3.994	90	2.7	286	2.9	0.975
Hospital Location
Rural	652	17.1	20,004	13.4	10.335	534	16.1	1589	16.0	0.356
Urban	3160	82.9	129,353	86.6	10.335	2785	83.9	8368	84.0	0.356
Hospital Size
Small (<300 beds)	1595	41.8	62,450	41.8	2.051	1356	40.9	4105	41.2	0.755
Medium (300–499 beds)	1146	30.1	46,314	31.0	2.054	1012	30.5	2992	30.0	0.962
Large (≥500 beds)	1071	28.1	40,593	27.2	0.059	951	28.7	2860	28.7	0.155
Hospital Teaching Status
Non-Teaching	2443	64.1	93,426	62.6	3.185	2096	63.2	6263	62.9	0.520
Teaching	1369	35.9	55,931	37.4	3.185	1223	36.8	3694	37.1	0.520
# of Annual Procedures Per Hospitala	141	87–215	142	96–215		146	93–215	146	72–215
PROCEDURE RELATED
Year of Procedure
2014	890	23.3	52,598	35.2	26.307	805	24.3	2334	23.4	1.909
2015	1408	36.9	53,038	35.5	2.965	1218	36.7	3661	36.8	0.146
2016	1514	39.7	43,721	29.3	22.105	1296	39.0	3962	39.8	1.521
Type of Procedure
Elective	680	17.8	6200	4.2	44.840	394	11.9	1261	12.7	2.419
Emergency/Urgent	3132	82.2	143,157	95.8	44.840	2925	88.1	8696	87.3	2.419
Peripheral Nerve Block Use	226	5.9	7771	5.2	3.166	188	5.7	591	5.9	1.160
Fracture Location
Intertrochanteric	621	16.3	56,562	37.9	50.060	586	17.7	1616	16.2	3.802
Subtrochanteric	114	3.0	6396	4.3	6.905	107	3.2	291	2.9	1.746
Femoral Neck	2942	77.2	73,993	49.5	59.873	2496	75.2	7680	77.1	4.527
Multiple	135	3.5	12,406	8.3	20.287	130	3.9	370	3.7	1.048
Surgery Type
Internal Fixation	971	25.5	92,714	62.1	79.380	951	28.7	2581	25.9	6.136
Hemiarthroplasty	1839	48.2	47,969	32.1	33.345	1697	51.1	5184	52.1	1.869
Arthroplasty	1002	26.3	8674	5.8	58.099	671	20.2	2192	22.0	4.406
Anesthesia Type
General	2615	68.6	108,541	72.7	8.952	2341	70.5	7050	70.8	0.596
General + Neuraxial	331	8.7	10,951	7.3	4.979	257	7.7	790	7.9	0.710
Unknown/Missing	866	22.7	29,865	20.0	6.646	721	21.7	2117	21.3	1.125
Procedure within 48 h of Admission	3749	98.3	146,963	98.4	0.394	3258	98.2	9796	98.4	1.696
COMORBIDITY RELATED
Charlson-Deyo Comorbidity Index
0	1957	51.3	64,604	43.3	16.243	1651	49.7	4945	49.7	0.161
1	812	21.3	31,689	21.2	0.206	704	21.2	2150	21.6	0.931
2	545	14.3	24,139	16.2	5.192	504	15.2	1503	15.1	0.252
3+	498	13.1	28,925	19.4	17.161	460	13.9	1359	13.6	0.612
ANTICOAGULANT USE
Aspirin	705	18.5	9069	6.1	38.541	480	14.5	1447	14.5	0.200
Other Anti-Platelet	15	0.4	376	0.3	2.500	14	0.4	51	0.5	1.326
Heparin	1410	37.0	68,008	45.5	17.424	1323	39.9	3901	39.2	1.397
Warfarin	204	5.4	6080	4.1	6.047	169	5.1	556	5.6	2.189
Other Anticoagulant	267	7.0	6321	4.2	12.060	213	6.4	661	6.6	0.894
>1 of Above	1046	27.4	52,434	35.1	16.594	971	29.3	2927	29.4	0.309
None	165	4.3	7069	4.7	1.945	149	4.5	414	4.2	1.63

^aMedian and interquartile range instead of n and %.

The majority of patients in our cohort had femoral neck fractures, making up 75.2% of the TXA group and 77.1% of the control group. Intertrochanteric femur fractures were the second most common type of hip fracture, making up 17.7% of the TXA group and 16.2% of the control group. Hip hemiarthroplasty was the most common surgical treatment (51.1% in the TXA group, 52.1% in the control group), followed by internal fixation (28.7% in the TXA group, 25.9% in the control group). Total hip arthroplasty accounted for the remaining 20.2% of patients in the TXA group and 22.0% in the control group.

Trends in blood transfusions and TXA use. TXA was used only in a minority of hip fracture patients in the cohort (Fig. 1). The overall incidence of TXA use in the full, unmatched cohort was 2.49% (3812/153,169). Prior to 2014, there was nearly no use of TXA; the incidence of TXA use mildly increased from 0% in 2006 to 3.2% in 2016. In contrast, the incidence of blood transfusion in the same cohort of patients decreased from 33.1% in 2006 to 20.6% in 2016, with an initial increase between 2006 and 2009.

Fig. 1.

Trends in TXA use and blood transfusion from 2006 to 2016.

Blood transfusions and perioperative complications. The unadjusted incidence of blood transfusions and perioperative medical complications between the TXA and control groups are outlined in Table 2, with adjusted odds ratios and percentage changes summarized in Table 3. The rate of blood transfusion was 16.2% and 19.0% in the TXA and control groups, respectively, with an odds ratio of 0.83 (95% CI 0.74–0.92) in favor of TXA use. There were no significant differences in the incidence of complications including acute MI (OR 0.89, 95% CI 0.56–1.40), acute renal failure (OR 0.99, 95% CI 0.83–1.19), and venous thromboembolism (OR 1.13, 95% CI 0.70–1.81). TXA use was associated with shorter LOS (% change −16.2; 95% CI -17.9, −14.5) and a minimal change in cost of hospitalization (% change +2.8%; 95% CI + 0.9%, +4.8%).

Table 2.

Outcomes in the TXA and no TXA groups in the matched sample.

	MATCHED SAMPLE
	TXA (n = 3319)		No TXA (n = 9957)
	n	%	n	%
CONTINUOUS OUTCOMES
Length of Staya	3	3–6	4	3–6
Cost of Hospitalizationa	$15,684	$12,384-$20,844	$16,015	$12,525-$21,063
BINARY OUTCOMES
Blood Transfusion	537	16.2	1888	19.0
Acute Myocardial Infarction	24	0.7	81	0.8
Acute Renal Failure	161	4.9	486	4.9
Thromboembolism	24	0.7	64	0.6
Combined Complicationsb	195	5.9	597	6.0

^aMedian and interquartile range instead of n and %.

^bCombination of acute myocardial infarction, acute renal failure, and thromboembolism.

Table 3.

Odds ratios and percent changes for continuous variables comparing patients who received TXA versus those who did not receive TXA.

	TXA Use (ref = No TXA Use)
	OR (95% CI)
CONTINUOUS OUTCOMES
Length of Stay	−16.2% (−17.9%; −14.5%)*
Cost of Hospitalization	2.8% (0.9%; 4.8%)*
BINARY OUTCOMES
Blood Transfusion	0.83 (0.74; 0.92)*
Acute Myocardial Infarction	0.89 (0.56; 1.40)
Acute Renal Failure	0.99 (0.83; 1.19)
Thromboembolism	1.13 (0.70; 1.81)
Combined Complications**	0.98 (0.83; 1.16)

*p < 0.05.

**Combination of acute myocardial infarction, acute renal failure, and thromboembolism.

4. Discussion

In this first large scale population-based study on TXA use in hip fracture surgery, we found that TXA continues to be used in a very small minority of hip fracture patients despite increasing literature on its safety and efficacy in orthopedic surgery. Hip fractures incur an estimated annual cost of over $5.4 billion in the United States and reducing morbidity and mortality in hip fracture patients remains a worldwide public health issue.²² We found that TXA usage was associated with a 17% decrease in blood transfusions and a 16% decrease in LOS with no significant increase in the risk of perioperative medical complications. This is consistent with findings from prior smaller studies, which have demonstrated a beneficial effect of TXA administration on reducing blood loss and blood transfusions.12, 13, 14, 15, 16, 17

Numerous studies have examined the efficacy and safety of TXA use in major elective orthopedic surgery, such as joint arthroplasty and spine surgery.23, 24, 25, 26 The general consensus from these studies is strongly in favor of TXA use due to its association with a decrease in blood loss, blood transfusion, and cost.^27,28 An allogeneic blood transfusion has been estimated to increase costs by $1731 per hospital admission.²⁹ Additionally, hip fracture patients who receive allogeneic blood transfusions have been found to have a higher risk of postoperative urinary tract infection.⁶

The safety of TXA has also been extensively studied. Even though several smaller studies have reported increased risk of venous thromboembolic events associated with TXA use, larger studies and meta-analyses have found no such association.^13,30, 31, 32 Despite encouraging results from these prior studies, however, TXA is currently not routinely used in patients undergoing hip fracture surgery in the United States. In this study, we found a small upward trend in TXA administration - increasing from nearly no usage in 2006 to 3.2% in 2016. Although this remains a small minority of hip fracture patients, the upward trend is encouraging.

It is important to note that hip fractures are not all equal in terms of injury pattern or surgical treatment. Extracapsular hip fractures are associated with greater total blood loss than intracapsular hip fractures.⁴ As such, the effect of TXA may vary in different hip fractures, as well as in different surgical procedures.

Extracapsular hip fractures are commonly treated with an intramedullary nail or sliding hip compression screw. In a single-center, randomized controlled trial of 72 patients, Tengberg et al. examined the effect of TXA in patients with extracapsular hip fractures undergoing operative fixation with a short intramedullary nail.¹⁷ Their protocol consisted of a preoperative dose of 1 g of TXA followed by a 24-h postoperative infusion of 3 g of TXA. TXA reduced total blood loss by 600 mL and also reduced the risk of blood transfusion with a number needed to treat (NNT) of 35.8, without a significant increase in venous thromboembolic events at 90 days postoperatively.

These findings were consistent with another randomized controlled trial of 100 patients with intertrochanteric femur fractures who received 2 doses of IV TXA – 1 dose preoperatively and 1 dose 3 h postoperatively.¹⁵ The relative risk of blood transfusion was found to be 0.5 with the use of TXA and no VTE events were reported.¹⁵ Similarly, Baruah et al. found that a single preoperative dose of TXA reduced total blood loss in a study of 60 patients with extracapsular hip fractures treated with a dynamic hip screw plate.¹⁶

Intracapsular hip fractures are treated differently depending on fracture displacement. Valgus impacted and non-displaced femoral neck fractures are most commonly treated with percutaneous pinning, whereas displaced femoral neck fractures are treated with hip hemiarthroplasty or total hip arthroplasty. In a retrospective cohort study of 271 patients undergoing hip hemiarthroplasty for intracapsular hip fractures, Lee et al. found that a preoperative dose of TXA was associated with a three times decreased odds of requiring blood transfusion with a NNT of 8.¹⁴ There was no difference in the incidence of VTE complications, but this study was limited by selection bias as TXA administration was left up to surgeon preference. Furthermore, low demand and sick patients were excluded, and therefore the study cohort was not fully representative of a geriatric hip fracture population with medical comorbidities.

The optimal dosing regimen of TXA to maximize efficacy while minimizing complications has not yet been determined. Prior studies have been inconsistent in the dosing regimen of TXA. Dosing regimens have included a single preoperative dose, a preoperative dose followed by a second dose 3 hours later, a preoperative dose followed by continuous intraoperative infusion, and a preoperative dose followed by a 24-hour continuous infusion postoperatively.11, 12, 13, 14, 15, 16, 17 No study to date has compared the effect of different doses and dosing regimens of TXA in hip fracture patients.

Lastly, the optimal route of administration of TXA remains undetermined. While multiple prior studies have focused on IV TXA, topical TXA has also been a topic of interest in recent literature. However, there remains limited data on the efficacy and safety of topical TXA compared to IV TXA in hip fracture patients. In one randomized controlled trial of 60 patients with intracapsular hip fractures undergoing hemiarthroplasty, Emara et al. demonstrated that both topical and IV TXA reduced blood loss and the need for blood transfusion, but IV TXA administration was associated with a 30% increased risk of VTE events compared to topical TXA.¹³ Therefore, the authors recommended the use of topical TXA over IV TXA. However, the generalizability of this study was limited by its patient selection criteria. Patients with preoperative anemia of a hemoglobin level <11 gm/dL in females and <12 gm/dL in males were excluded, thus making their cohort not fully representative of an elderly hip fracture population. Additionally, patients on preoperative anticoagulation were also excluded. Given these limitations and the lack of evidence in the current literature, further studies are needed to better examine the efficacy and safety of topical TXA versus IV TXA in hip fracture patients undergoing surgery.

The strengths of this study are its large sample size and study of real-world (non-selected and non-randomized) practice with a dataset that encompasses approximately 20–25% of all hospital discharges in the United States. This contributes to the generalizability of our findings. Propensity score matching further allowed us to account for possible individual and hospital level confounding factors in the association between TXA and postoperative outcomes. There are also several limitations to this study. First, this is a retrospective study and subject to inherent inaccuracies in data collection of all database studies. However, these should be independent of TXA use. Second, our cohort included both extracapsular and intracapsular hip fractures. Blood loss related to the fracture itself is different between extracapsular and intracapsular hip fractures, making these two groups of patients different at baseline.⁴

In conclusion, TXA is used in a minority of hip fracture patients in the United States. Despite increasing usage of TXA in shoulder, hip and knee arthroplasty, the incidence of TXA use in hip fractures has only increased from 0% to 3.2% from 2006 to 2016. TXA use was associated with a 17% decrease in blood transfusions and a 16% decrease in LOS with no significant increase in the incidence of VTE complications. Although optimal dosage and frequency are under continued investigation, TXA use appears to be safe and efficacious in hip fracture patients and should be considered in the perioperative care for all patients in this surgical cohort.

Ethical review committee statement

This study was performed in accordance with the ethical standards in the 1964 Declaration of Helsinki. This study was carried out in accordance with relevant regulations of the US Health Insurance Portability and Accountability Act (HIPAA). Details that might disclose the identity of the subjects under study were omitted. Institutional ethical review board approval was obtained for this study.

Declaration of competing interest

Each author certifies that he or she has no commercial associations (e.g., consultancies, stock ownership, equity interest, patent/licensing arrangements, etc.) that might pose a conflict of interest in connection with the submitted article.