Date: March 4th, 2021
Guest Skeptic: Dr. Anand Swaminathan is an Assistant Professor of Emergency Medicine at St. Joseph’s Regional Medical Center in Paterson, NJ. Managing editor of EM:RAP and Associate Editor at REBEL EM.
Case: A 70-year-old man presents with epistaxis. He’s got a history of atrial fibrillation and hypertension. His medications include apixaban, metoprolol and perindopril. He states that the nosebleed started that morning all of a sudden. Vitals are; a blood pressure 145/73 mmHg, heart rate 64 beats/minute, and oxygen saturation 99% on room air. You apply direct pressure for 10 minutes, but the bleeding continues. You administer phenylephrine topically, reapply pressure and, consider the use of tranexamic acid (TXA).
Background: Epistaxis is a common Emergency Department (ED) complaint with over 450,000 visits per year and a lifetime incidence of 60% (Gifford 2008, Pallin 2005). The majority of refractory hemorrhages are seen in the elderly and in more than two-thirds of the time no cause for the epistaxis is identified.
Standard anterior epistaxis treatment consists of holding pressure, use of local vasoconstrictors, topical application of silver nitrate and placement of an anterior nasal pack. The most common vasoconstrictor used in the US is oxymetazoline. Emergency physicians have multiple tools in the toolbox to address this condition.
We covered the topic of epistaxis on SGEM#53: Sunday Bloody Sunday. That episode we discussed 11 questions concerning epistaxis. It is a great review on the management of nosebleeds. The episode included the Dundee protocol for adult epistaxis management from 2012. I searched and could not find an updated version.
ED patients with epistaxis often fail conservative management and end up with anterior nasal packs which are uncomfortable. This is even more common in the group of patients who are taking antiplatelet agents or anticoagulants.
In recent years, TXA has been added to many physicians’ armamentarium based on a few relatively small studies. We reviewed two of these randomized control trials (RCTs) on the SGEM including the 2018 publication looking at using TXA in patients taking antiplatelet drugs. (SGEM#210).
Some of these RCTs looking at TXA for epistaxis have also been reviewed on REBEL EM and other FOAMed sites. The results were encouraging. Topical TXA has minimal safety concerns and is relatively inexpensive. However, the studies did have a number of limitations including, being single centered, relatively small sample sizes and a lack of blinding.
The SGEM structured critical review was skeptical of TXA for treating nosebleeds and had a conservative bottom line:
Despite some limitations in this un-blinded trial, topical tranexamic acid appears to improve some patient important outcomes in patients who are taking antiplatelet medications who present with epistaxis.
Clinical Question: Does the use of topical, intranasal TXA reduce the need for application of anterior nasal packing in ED patients with epistaxis who fail conservative management?
- Population: Patients older than 18 years of age presenting with persistent epistaxis after local pressure and/or ice was applied to the bridge of the nose for at least 10 minutes. If bleeding persisted (continued presence of blood on the upper lip after wiping emanating from the nares) patients were treated with cotton wool dental roll soaked with a topical vasoconstrictor and inserted into the affected nostril for 10 minutes. If the bleeding persisted after the removal of the vasoconstrictor dental roll, they were enrolled in the trial.
- Exclusions: Hemodynamic unstable patients, epistaxis due to trauma, out-of-hospital packing, allergy to TXA, ENT thought should be excluded, nasopharyngeal malignancy, pregnancy, hemophilia, and inability or unwillingness to provide consent.
- Intervention: Tranexamic acid (TXA) 200 mg in 2 ml applied to a cotton wool dental roll (could packing be repeated once). Dental roll held in place for 10 minutes with pressure. The treatments were done by EM consultants, junior house officers, or emergency nurse practitioners.
- Control: Cotton wool dental roll soaked in sterile water. Dental roll held in place for 10 minutes with pressure
- Primary Outcome: Use of anterior nasal packing (of any type) during the index ED visit regardless of any other additional treatments (intention to treat analysis)
- Secondary Outcomes: Hospital admission, need for blood transfusion, recurrent epistaxis, thrombotic events, hospital reattendance within 1 week.
Authors’ Conclusions: “In patients presenting to an ED with atraumatic epistaxis that is uncontrolled with simple first aid measures, topical tranexamic acid applied in the bleeding nostril on a cotton wool dental roll is no more effective than placebo at controlling bleeding and reducing the need for anterior nasal packing.”
Quality Checklist for Randomized Clinical Trials:
Key Result: No statistical difference in patients getting anterior packing between TXA and placebo
- Primary Outcome: Use of anterior packing
- Placebo 41.3% vs TXA 43.7% required packing (2.4% difference)
- Odds Ratio (OR) 1.107 (95% CI; 0.769 to 1.594; p=0.59)
- Secondary Outcomes:
- No statistical difference in any of the secondary outcomes including recurrent epistaxis, hospital admission or blood transfusion
- Also, no statistical difference with a per-protocol analysis
This is the largest trial and only double-blind study looking at topical TXA for epistaxis. The authors should be congratulated for successfully completing this multicentered study and getting it published. There are always limitations that need to be considered, even in very well-done studies. We will discuss five issues.
1. Patient Population: This is one of the major limitations to this study. It was a very select/narrow group of patients. There were 2,622 initially eligible but 2,126 (81%) were excluded. Almost 1/3 were excluded because a research nurse was unavailable. We understand that this is often a reality of doing research in an ED that is open 24/7/365. However, 29% of eligible patients were excluded for the reasons listed as “other”. What does that mean? Could it have introduced some subjectivity into excluding some patients but not others resulting in some selection bias?
Also, when you drill down these are patients that have “failed” standard therapy of at least 10 minutes of pressure and/or ice on the bridge of their nose followed by another 10 minutes of anterior packing soaked in a vasoconstricting drug. In the US, the most common vasoconstrictor is oxymetazoline. It is unclear if one agent is superior to the other and thus, it’s unclear how this affects application of this evidence.
The authors also excluded another 6% of patients if the emergency department was considered too busy. Was this objectively determined and if so how; wait times, length of stay, volume of patients, acuity of patients, or another metric? In addition, who made the decision that the department was too busy, EM consultant, junior house officer, nurse practitioner or the research assistant?
Patients “expected” to be seen by the ENT in-patient team for specialist treatment were also excluded. They also did not include pregnant patients which is an example of women being under-represented in clinical trials. All these issues lead to a very narrow cohort of patients with epistaxis and some of these could have introduced an element of selection bias.
2. Anticoagulation: A super majority of patients included in the study (65%) were taking anticoagulant medication. There were no additional details given in the manuscript as to what type of anticoagulants the patients were using: DOACs (factor Xa inhibitors or direct thrombin inhibitors), vitamin K antagonists (warfarin), or low molecular weight heparin (enoxaparin). Also, we are unsure if the term anticoagulation includes antiplatelet drugs (ASA, clopidogrel, and ticagrelor)? The high proportion of select patients on anticoagulation with epistaxis who failed conservative treatment and packing with a topical vasoconstrictor could explain why no statistical difference was found between the intervention group (TXA) and the placebo group.
3. Dosing: The initial dose of TXA used was 200mg. The clinician could repeat the dose once if the bleeding persisted. This is different from the 500mg dose used in the previous TXA trials. It is unclear if the larger amount of TXA that the applicator was soaked in resulted in a larger amount of TXA reaching the mucosal tissue due to absorption by the applicator. What is known is that the previous studies using 500mg demonstrated efficacy while this study using a starting dose of 200mg with a maximum dose of 400mg did not show superiority of TXA over placebo.
4. Power: The investigators powered the study to find a 10% absolute difference between the TXA group and the placebo group. This was based on the assumption that 95% of the patients would require anterior packing. However, the result was only half the patients required anterior packing.
If the authors knew a priori that only 42.5% of the patients would need packing that would change their power calculation. Using the same goal of detecting a 10% absolute decrease with TXA compared to placebo and assuming a corrected X2 test result powered at 90% and with a significance level of 5% they would have needed around 1,000 patients. This does not mean the study was “under-powered”. We do not know if a larger sample size would have demonstrated a statistical difference.
It’s not correct to go back and retrospectively apply a power calculation. The results are the results, and they did not demonstrate a statistical superiority of TXA over placebo for treating these select patients with epistaxis. There is a good editorial that explains why it is not useful to do this kind of post-mortem on a study. (Jiroutek and Turner 2017)
5. End Points: Their primary outcome was need for anterior packing. This was different than the primary outcome in prior studies (stopping bleeding). We assume the packing was placed because bleeding was not controlled but it’s possible there were other indications.
Looking at their secondary outcomes they had an admission rate ~44%. This seemed very high to us and did not reflect our care in the US or Canada. However, discussion with the authors reveals that this is likely due to the UK “4-hour rule”.
There were a number of other important secondary outcomes that they considered like the need for blood transfusion, recurrent epistaxis, thrombotic events and hospital reattendance within one week. However, they did not consider other potential important outcomes such as ED length of stay or patient satisfaction.
Compare the Authors’ Conclusions to the SGEM’s Conclusions: We would have tweaked their conclusions to say “In select patients (majority taking anticoagulant medication) presenting to an ED with atraumatic epistaxis that is uncontrolled with simple first aid measures and the application of a topical vasoconstrictor, 200mg-400mg of TXA applied in the bleeding nostril on a cotton wool dental roll is no more effective than placebo at controlling bleeding and reducing the need for anterior nasal packing”.
SGEM Bottom Line: Evidence on the use of TXA in epistaxis is inconsistent and it depends on the patient population and treatment protocol whether or not it demonstrates efficacy.
Case Resolution: You decide to use the protocol that has been shown to be successful in a few other RCTs. You soak a small gauze with 500mg of TXA and place it in the affected nostril. Ten minutes later, the patient’s bleeding has stopped. You watch them for another 30 minutes and then discharge the patient home with clear directions on when to return to the emergency department.
Clinical Application: Each new study needs to be considered individually and in context of the other relevant literature. Three other RCTs have been published on this topic. All three of these trials demonstrated efficacy in treating patients with epistaxis. They too have limitations: single center, unclear what care was provided prior to the intervention, smaller sample size and being unblinded. They also used a higher initial dose of 500mg of TXA instead of 200mg TXA (Zahed 2013, Zahed 2017, and Akkan2019).
This study definitely calls into question the routine use of TXA in this patient with epistaxis. Each clinician will need to interpret the literature for themselves, reflect upon their clinical experience and ask the patients what they would prefer. Essential, we would encourage clinicians to practice evidence-based medicine.
What Do I Tell My Patient? We have a medicine called TXA that may help stop your nose bleed faster and get you out of the emergency department quicker. It requires that I place this piece of cotton soaked in the medication up your nose and leave it there for about ten minutes. No major side effects have been identified in any of the research studies. Would you like to try it and see if it stops your bleeding?
Remember to be skeptical of anything you learn, even if you heard it on the Skeptics’ Guide to Emergency Medicine.