Best Of AAST #8: Duplex Screening For DVT

To screen on not to screen, that is the question. If you do more testing, you will find more cases. But does it make a difference clinically? Sounds like some of the questions coming up in our current discussion of the Coronavirus. But that’s what we really need to know.

The group at Intermountain Medical Center in Salt Lake City performed a 2 ½ year randomized, prospective study of screening duplex ultrasound of the lower extremities vs no screening study. They used the Risk Assessment Profile (RAP) developed by Greenfield, first published in 2000. Any patient at moderate or higher risk for DVT (RAP score >5) was enrolled in the study. They were randomized into two groups: a screening group who received duplex scans on days 1, 3, 7, and then weekly, and a “no routine screening” group. All patients received chemoprophylaxis per the trauma service’s existing protocol.

The RAP score is a 17 factor scale that assigns a specific number of points based on underlying medical conditions, iatrogenic factors like central lines or transfusions, injury-related factors, and age.

Here are the factoids:

  • A total of 3,236 trauma patients were identified, and the 1,989 who were at moderate or higher risk for DVT were evenly randomized to screening vs no screening
  • There were no differences in age, sex, BMI, mechanism, ISS, or length of stay between the two groups
  • The incidence of DVT was 15% in the screened group vs 1.7% in the no screening group

The authors concluded that screening diagnoses more DVT, most of which is below the knee. And they also noted that screening identified DVT more often than clinical exam alone, but does not result in fewer PE or deaths. They suggest that more work needs to be done to identify exactly who benefits from duplex screening the most.

Here are my comments:

Finally, an easy to follow and well-designed study! But I think some of the results may be missing from the abstract. That section cuts off in the middle of some of the statistics, and there is no mention of the clot location or PE/mortality rates mentioned in the conclusion.

I also worry that a thousand patients in each group may not be enough. We are working with low incidence end points like PE and death, and this is an association study with many potential confounders/factors that may not have been recorded. I generally like to see the ability to detect a minimum of a 2x effect. So if the incidence of PE is 1.5%, I like to see the ability to detect a difference if the other group is 3%.

And speaking of study size. The RAP score was first described in 1997 and was a pilot study. They drew their conclusions from only 53 patients, and the only risk factor that they could show that was a statistically significant predictor of DVT was age. They concluded that surveillance of patients with RAP > 5 was warranted. This abstract builds upon this work, but is trying to say that maybe we don’t need to do duplex scans.

Here are my questions for the presenter and authors:

  • Is there some text missing from the end of the results section of the abstract? It seems to end unexpectedly, and some things are mentioned in the conclusions that are not in the results.
  • Why did you choose the RAP score? There are other risk assessment tools available out there. What is so special about RAP?
  • Is your sample size large enough to detect differences in incidence of PE or death? My back of the envelope calculations suggest at least 1,500 patients would be needed in each group.
  • How long did you follow patients to determine if they had PE or death? Until they were discharged? Later than that?  This makes a big difference in the eventual incidence of these outcomes.
  • Based on what you found, is there any value to treating asymptomatic proximal DVT? It sounds like you are saying that screening is not needed at all because PE and death are the same. Isn’t there value in treating proximal DVT if you find it?

This abstract certainly got me thinking! I am looking forward to the presentation and discussion of this abstract!

Reference: Head in the sand? The value of routine duplex ultrasound screening for venous thromboembolism in the trauma patient: a randomized Vanguard trial. AAST 2020, Oral Abstract #16.

Best Of The AAST #7: TXA And Thromboembolism

The use of tranexamic acid (TXA) in trauma patients has escalated dramatically since the CRASH-2 trial was published ten years ago. It has become a frequent addition to the massive transfusion protocols used by trauma centers. And we are now even seeing TXA given by prehospital provides when life-threatening bleeding is suspected.

This drug is popular because it is inexpensive (~$100/dose) and is thought to be safe. However, some trauma professionals have been concerned about thrombotic side effects since TXA is a finbrinolysis inhibitor.

The group at the Mayo Clinic performed a retrospective study of seven years of their own data to determine if the concern for thrombotic complications was warranted. They specifically evaluated in-hospital mortality and thrombotic events up to 28 days after injury.  They also scrutinized outcomes in patients who received only the bolus TXA injection, but not the infusion.

Here are the factoids:

  • A total of 848 patients were included in the study, but there was no information as to what the inclusion criteria were
  • Only 212 received TXA; the other 636 were considered the control group, and there were no differences in age, sex or mechanism of injury
  • Thrombotic events occurred in 13% of the TXA patients and only 6% in the control group, which was statistically significant
  • Specific thrombotic events in TXA vs non-TXA patients: DVT was 8.5% vs 3.5% (significant), pulmonary embolism was 3.8% vs 1.9% (NS), MI was 1.9 vs 0.4% (NS), stroke was 2.4% vs 1.1% (NS)
  • Thrombotic events occurred more frequently in patients who received both doses of TXA (23%) vs just the bolus (10%), and this was also significant
  • In-hospital mortality was 21% with TXA vs 10%, which was not significantly different, controlling for confounders

The authors concluded that TXA administration was associated with higher rates of thrombotic events. They went on to state that TXA should not be routinely given for trauma patients in the community setting.

Here are my comments:

This is a bold recommendation from this very small study. The CRASH-2 trial was randomized and placebo controlled, and analyzed their experience with 10,000 subjects in each arm. This retrospective study has only  212 TXA vs 636 control patients. Big difference.

The authors attempt to match the TXA patients with controls. They controlled for age, sex, mechanism, and ISS. But it does not appear that there was any control for injuries known to increase the risk of thrombotic events like spine and pelvic fractures.

And why look at a full 28 day interval for thrombotic events? I would expect most of these events to occur in the first few days. Including an entire month in the study allows thrombotic events from other causes to creep in.

Here are some questions for the presenter and authors:

  • Please comment on how the small numbers in your study may have an impact on the results.
  • What were the selection criteria for your 848 patients? Were they all of your trauma activation patients? If not, is there some selection bias possible?
  • DVT appears to be the driver for your “significant” number of thrombotic events. Yet the other events (MI, stroke, PE) were not significantly different. This seems counter-intuitive, since the DVT numbers themselves numbered only about 20 in each group. Please describe the statistics you used to derive this conclusion.
  • Did you look at the incidence of injuries that are known to increase the risk of thrombotic events in the two groups? If there was an excess of pelvic or spine fractures in the TXA group, this might not be picked up in your analyses and could skew your data.
  • Why did you include thrombotic events for a full 28 days after injury? This allows for later events caused by factors other than the TXA. Show us a redo of your analysis using 5 or 7 day thrombotic events.

These are interesting numbers, but I have to admit that I am skeptical. I’m not clear how community hospital administration of TXA makes it more likely associated with thrombotic events. I will definitely be listening intently to this presentation. And probably asking these questions.

Reference: Risk of thromboembolic events after the use of TXA in trauma patients. AAST 2020, Oral Abstract #15.

Best Of AAST #6: Timing Of Venous Thromboembolism Prophylaxis

Venous thromboembolism (VTE) and pulmonary embolism (PE) have caused major problems for trauma professionals for at least 50 years. Interestingly, despite advances in chemical and mechanical prophylaxis, the mortality rates for both have remained about the same.

The group at St. Joseph Mercy Hospital in Ann Arbor looked at the timing of start of VTE chemoprophylaxis. They were curious as to whether the start time made a difference in mortality. They reviewed a collaborative database with 12 years of data, tallying information for all trauma patients who were admitted for at least 48 hours.

Here are the factoids:

  • Over 89,000 patients were analyzed; 1.8% developed VTE and 1.9% died (?)
  • Delay in starting chemoprophylaxis increased the risk of VTE (see figure)
  • Delaying chemoprophylaxis beyond 48 hours was associated with increased mortality and increased incidence of VTE

The authors concluded that early initiation of chemoprophylaxis reduces mortality and thrombotic complications.

Here are my comments: Unfortunately, I’m not entirely clear about the details of the abstract. This frequently happens because the authors have to strain to fit all of their ideas in a finite amount of space.

First, it’s a large database study, so it’s difficult to ensure that all the factors you want to study have been included in it. Somebody else designed it years ago, so you get what you get.

I’m a little confused about the incidence of complications and death. They are both about the same number (1.8%). Typically, VTE incidence is a few percent and death from PE is less than 1%. The death number seems high, unless it includes some other type of death.

The VTE incidence vs time graph is very interesting, although the goodness of fit looks a little off toward the right side. It looks like it could easily be a little lower.

Finally, segregating time periods into two 24-hour periods (0-24 hours, 24-48 hours)and one 72-hour plus one (48-120+ hours) seems like it might bias your data. The longer that last period, the greater chance that each patient will develop VTE or die.

Overall, the numbers in Table 1 are noted to be statistically significant, but clinically they appear to be very similar.

Here are some questions for the presenter:

  • Please explain the mortality numbers (1.9%). What did these patients die of? A pulmonary embolism? Something unrelated? This number seems high, since it is equal to your VTE incidence.
  • Tell us about the risk adjustment you used to calculate mortality rates. What patient factors were available to you? Are there others that might have been helpful to have in the database?
  • What tool did you use to fit the curve in Figure 1? The right side looks considerably higher than the data bars would suggest. Please be sure to explain all of the statistical techniques you used, as they were not fully covered in the abstract.
  • What was the impact of cramming 3 days of data into your last cohort? Wouldn’t this be expected to yield higher incidences of VTE and death?

I agree that VTE prophylaxis is best started early, but I need a wee bit more information. I’m intrigued by the paper, but I think you will have to spend some time explaining how you designed the analysis so we can all understand.

Reference: Association of timing of initiation of pharmacologic venous thromboembolism prophylaxis with outcomes in trauma patients. AAST 2020, Oral Abstract #14.

Best Of AAST #5: MAP Goals For Spinal Cord Injury

The use of mean arterial pressure (MAP) goals in the management of spinal cord injury is commonplace. But hit the literature some time and try to find out what the ideal MAP is, or if they even make a difference. It’s very difficult to come up with really solid data.

The group at Dignity Health St. Joseph’s Hospital in Phoenix reviewed their own trauma registry over a 6 year period. They chose a specific MAP goal (85 torr) and sifted through all of their electronic health record data to see how consistently they achieved it, the pressor dose needed to do so, and an objective measure of neurologic improvement while in the hospital (ASIA impairment scale).

Here are the factoids:

  • There were 136 patients studied, with an average ISS of 24 and average length of stay (LOS) of 10 days
  • Each patient had an average of 157 MAP determinations, and MAP > 85 (MAP85) was achieved in about 72% of those readings
  • About 80% of patients required a pressor to maintain MAP85, with an average dose equivalent to 26mg of norepinephrine during their stay
  • Patients with an ASIA improvement of at least one level were at MAP85 79% of the time vs 68% for those that did not improve
  • Multivariate regression showed that MAP85 was the main factor associated with the higher ASIA scores

The authors concluded that MAP85 was an important predictor of neurologic improvement, and that increased vigilance in maintaining it would help optimize neurologic recovery.

Here are my comments: The ASIA Impairment Score is a detailed description of the degree of neurologic injury in patients with spinal cord injury. The worksheet used to document consists of two pages and requires an in-depth sensory and motor exam. This is then translated into a alphabetical grade from A (no sensory or motor function even in the sacral segments) to E (normal exam).

This is a very interesting study, but I always worry about the test instrument. In order to use this scoring system and have good inter-rater reliability, the people that administer the test must be specifically trained. Otherwise the results become muddled.

The last thing that I always think about in association studies like this is, how do you know you included all of the relevant factors? Are there potentially significant variables that you wish you had that just weren’t in your trauma registry?

Here are my comments and questions for the authors:

  • Tell us about the personnel who administered the ASIA assessment. Did every one of them have specific training to do it? This is important to ensure that the major conclusion in the study is valid.
  • It is hard to follow the change in ASIA score based on the patient’s initial exam. Please show us how many A’s became B (or higher), etc.
  • Was the amount of time that MAP goals were not met clinically significant? Using the length of stay and MAP determination numbers given, and assuming that the first three days were the most significant for recovery, each patient would have had their MAP measured every half hour during those three days. The patients who did not have an improved ASIA score had 50 measurements, on average, where MAP was < 85. But the patients who did improve still had 33 measurements below  goal. Does this 17 measurement difference really matter?

I’m hoping to firm up my appreciation for MAP85 while listening to your presentation!

Reference: Mean arterial pressure maintenance following spinal cord injury: does meeting the target matter? AAST 2020 Oral Abstract #8.

Best Of AAST #4: TBI and Antiplatelet / Antithrombotic Agents

More and more people are taking antiplatelet or antithrombotic agents for a variety of medical conditions. One of the dreaded side effects of these medications is undesirable bleeding, particularly after injury. This is especially true if the bleeding occurs inside the skull after any kind of head trauma.

Which agents, if any, lead to worse outcomes? The literature has been a bit inconsistent over the past 10 years. A group from HCA Healthcare reviewed the trauma registries from 90 hospitals, which I presume are in the HCA system. They included patients patients who suffered a ground level fall and were 65 years or older. They excluded those who had a significant injury to regions other than the head.

Here are the factoids:

  • Over, 33,000 patient records were reviewed, with an average age of 81
  • Nearly half were on single or multiple anti-thrombotic therapy (!)
  • The proportion of patients sustaining a “TBI” was roughly the same (21%) whether they were not on anti-thrombotic therapy or not
  • Apixaban and rivaroxiban were associated with lower rates of “TBI” (13-16%)
  • Clopidogrel was associated with a higher “TBI” rate (23%)
  • Patients requiring brain surgery  were more common in patients taking aspirin plus clopidogrel (2.9%) vs all the others (2%) and this was statistically significant
  • None of the treatment regimens were associated with higher mortality (roughly 2-3%)

The authors conclude that anti-thrombotic use in the elderly who suffer a ground level fall are not at risk for increased mortality and that they may have negligible impact on management.

My comments: The one thing that makes this abstract difficult to read is their use of the term TBI, which is why I put it in quotes above. I think that the authors are conflating this acronym with intracranial hemorrhage. It’s a bit confusing, because I think of TBI as a term that means the head was struck and either left a physical mark (bump on the outside or blood on the inside) or there was known or suspected loss of consciousness. They are apparently using  it to describe intracranial bleeding seen on CT.

And because this is a registry study, many of the patient-specific outcome details cannot be analyzed. Mortality and operative rates are very crude outcomes. What about some of the softer ones? Although the average GCS was stated to be 14.5, it would be interesting to know how many of these patients were able to return to their previous living situation, and how many were significantly impaired even though they didn’t die or need an operation.

Here are my questions for the presenter and authors:

  • How do you define a TBI in this study? Could it be just a concussion? Does it require some type of blood in the head? Assuming that there are lots of TBIs that occur without intracranial bleeding, including such patients in your analyses will skew the data toward lower incidence and will dilute out the patients with hemorrhage.
  • What was the length of your study? If it includes data that is older than six years or so, it may under-represent the use of some of the direct oral anticoagulant drugs (DOACs).
  • Are half of your elderly falls patients really on anti-thrombotic therapy? This is a shocking number, and seems to be high in my experience. Since your study was distributed across a large number of hospitals, it brings up the question of whether so many of our elders really need this medication.
  • Do you have any sense for how your various subgroups fared in terms of their discharge disposition? You conclude that the use of anti-thrombotic agents isn’t so bad, really. At least when it comes to needing brain surgery or dying. But are there other cognitive issues that are common that might encourage trauma professionals to continue to look at these drugs with a wary eye?

This is important work, and I am anticipating a great discussion after your presentation.

Reference: Antiplatelet and antiplatelet agents, alone and in combination, have minimal impact on traumatic brain injury (TBI) incidence, need for surgery, and mortality in ground level falls (GLFs): a multi-institutional analysis of 33,710 patients. AAST 2020 Oral Abstract # 7.