Tag Archives: pulmonary embolism

The IVC Filter In Trauma: Why?

The inferior vena cava (IVC) filter has been around in one form or another for over 40 years. One would think that we would have figured everything about it out by now. But no!  The filter has evolved through a number of iterations and form factors over the years. The existing studies, in general, give us piecemeal information on the utility and safety of the device.

One of the major innovations with this technology came with the development of a removable filter. Take a look at the product below. Note the hook at the top and the (relatively) blunt tips of the feet. This allows a metal sheath to be slipped over the filter while in place in the IVC. The legs collapse, and the entire thing can be removed via the internal jugular vein.

ivc-filter-complications1

The availability of the removable filter led the American College of Chest Physicians to recommend their placement in patients with known pulmonary embolism (PE) or proximal deep venous thrombosis (DVT) in patients with contraindications to anticoagulation. Unfortunately, this has been generalized by some trauma professionals over the years to include any trauma patients at high risk for DVT or PE, but who don’t actually have them yet.

One would think that, given the appearance of one of these filters, they would be protective and clots would get caught up in the legs and be unable to travel to the lungs as a PE. Previous studies have taught us that this is not necessarily the case. Plus, the filter can’t stop clots that originate in the upper extremities from becoming an embolism. And there are quite a few papers that have demonstrated the short- and long-term complications, including clot at and below the filter as well as post-phlebitic syndrome in the lower extremities.

A study from Boston University reviewed their own experience retrospectively over a 9 year period. This cohort study looked at patients with and without filters, matching them for age, sex, race, and injury severity. The authors specifically looked at mortality, and used four study periods during the 9 year interval.

Here are the factoids:

  • Over 18,000 patients were admitted during the study period, resulting in 451 with an IVC filter inserted and 1343 matched controls
  • The patients were followed for an average of 4 years after hospitalization
  • Mortality was identical between patients with filters vs the matched controls

dvt-study

  • There was still no difference in mortality, even if the patients with the filter had DVT or PE present when it was inserted
  • Only 8% ever had their “removable” filter removed (!)

Bottom line: Hopefully, it’s becoming obvious to all that the era of the IVC filter has come and gone. There are many studies that show the downside of placement. And there are several (including this one) that show how forgetful we are about taking them out when no longer needed. And, of course, they are expensive. But the final straw is that they do not seem to protect our patients like we thought (hoped?) they would. It’s time to reconsider those DVT/PE protocols and think really hard about whether we should be inserting IVC filters in trauma patients at all.

Tomorrow: a look at trends in filter insertion and retrieval.

Related post:

Reference: Association Between Inferior Vena Cava Filter Insertion
in Trauma Patients and In-Hospital and Overall Mortality. JAMA Surg, online ahead of print, September 28, 2016.

IVC Filters: Another Nail In The Coffin?

IVC filter insertion has been one of our tools for preventing pulmonary embolism for decades. Or so we thought. Its popularity has swung back and forth over the years, and has been in the waning stage now quite some time. This pendulum like motion offers an opportunity to study effectiveness when coupled with some of the large datasets that are now available to us.

IVC filters have been used in two ways: prophylactically in patients at high risk for pulmonary embolism (PE) who cannot be anticoagulated for some reason, and therapeutically once a patient has already suffered one. Over the years, guidelines have changed, and have frequently been in conflict. Currently, the American College of Chest Physicians does not recommend IVC filters in trauma patients, while the Eastern Association for the Surgery of Trauma promote their use in certain subsets.

A Pennsylvania group performed a large, retrospective review of three databases, the  Pennsylvania Trauma Outcome Study (462K patients), the National Trauma Data Bank (5.8M patients), and the National Inpatient Sample. All were patients with an emergent trauma-related admission.

Here are the most interesting factoids:

  • About 2% of all patients underwent IVC filter insertion, and 94% were inserted prophylactically
  • About 90% of patients with a prophylactic filter had at least one predictor for PE, which means that the remaining 10% had none (!)
  • Conversely, about 86% of patients who developed a PE had at least one risk factor, meaning that 14% had no recognized risk factors (!!)
  • The use of IVC filters peaked in 2006-2008 at 2-4%, then falling steadily over the following 5-7 years to less than 1%
  • PE rates peaked in 2008, then declined by 30% in the PTOS sample and stayed steady in the NTDB

Bottom line: The use of IVC filters peaked in 2008 and has been in decline since then. But interestingly, the rates of PE and fatal PE have been steady to declining, depending on the data set. Obviously, there are always some shortcomings for studies like this. Remember, IVC filters are intended to prevent fatal PE. It is possible that some fatal PEs were not identified in these databases. Furthermore, it was not possible to obtain any information on the use of chemical prophylaxis in these patients. 

Overall, there has been no increase in PE and fatal PE rates over the time period that IVC filter usage has been decreasing. This suggests that these devices have not had their intended effect. Trauma professionals need to very seriously consider the specific indications in any patient they are considering for insertion. They may not have the protective effect you think.

Related posts:

Reference: Vena Cava Filter Use in Trauma and Rates of Pulmonary Embolism, 2003-2015. JAMA Surg 152(8):724-732, 2017.

Enoxaparin And anti-Xa Levels: Who Cares? Part 3

Today is the final installment in a series about the use of anti-Factor Xa levels to titrate enoxaparin dosing to prevent venous thromboembolism (VTE). This is another study that tries to show that “hitting the number” actually makes a difference in patient care. You decide.

This study identified a subset of patients at high-risk for VTE based on a commonly used and very good risk screening tool, the Risk Assessment Profile (RAP). It takes some 17 factors into account to arrive at a numerical score. In this paper, the authors chose a score of 10 or greater to denote high risk. The patients were all seriously injured, and were in the trauma ICU of this established Level I trauma center.

This retrospective study excluded non-ICU patients, ones who did not receive enoxaparin or anti-Xa levels, and two patients with DVT on admission. This brought the number of eligible patients from 621 to 127 (the treatment group). They then narrowed the field down to the high-risk treatment group by excluding patients with a RAP score < 10. Now we are down to 86. But then 30 more (35%) were excluded because they did not undergo duplex ultrasound screening, leaving only 56 to study (!).

The control group was a “similar” historical cohort from a two year period from 2009 to 2012. You can tell that this group is getting a little stale, because the only patients included were those who received unfractionated heparin for prophylaxis (remember those days?). Of the 106 patients in the control group, 20 (28%) were reported as have VTE. However, it included 6 patients with DVT on admission, which were excluded in the study group. This makes the DVT rate look higher in the control group. It also included 2 upper extremity DVT and 1 septic pelvic venous thrombosis. Excluding all of these brings the historical VTE rate down to only 10%. Remember this.

So let’s get on to the factoids:

  • Only 35% of the 127 patient treatment group “hit the number” for anti-Xa (0.2-0.4 IU/ml) after three 30mg doses of enoxaparin
  • An additional 25% managed to achieve the desired anti-Xa level after dose adjustment, but 51 patients (40%) never did get there
  • There were 10 VTE events in the 127 treatment group patients, 9 of whom had high RAP scores, giving them a 7.8% rate of VTE
  • Nine of the 10 VTE patients occurred in patients with low anti-Xa levels
  • The authors compared their 7.1% DVT rate with the 21% in their historical controls, concluding that titrating anti-Xa levels reduced this rate. They did not include PE for some reason, and do not claim a statistical difference. They admit that the study was underpowered to detect differences in VTE. There is no significant difference in VTE rates in the study or control groups.

Bottom line: This is the last paper on the topic. I promise. At least for a while. Here’s what we know:

  • VTE is a problem in trauma patients, particularly seriously injured ones
  • We are not very good at sticking to a prophylaxis or screening regimen (note how many patients are excluded in all of these studies)
  • We can’t seem to generate the numbers to conduct a good study that can detect differences in what we do
  • It’s difficult to “hit the number” for anti-Xa using standard enoxaparin dosing
  • We don’t even know if it makes a difference if we do “hit the number”. VTE rates seem to be the same regardless.

So we are struggling to make a lab test look right to adjust enoxaparin dosing, and we don’t even know if it makes a difference. Will somebody put a good, multi-center study together and help us to figure all of this out?

Related posts:

Reference: Anti-Xa-guided enoxaparin thromboprophylaxis reduces rate of deep venous thromboembolism in high-risk trauma patients. J Trauma 81(6):1101-1108, 2016.

Enoxaparin And anti-Xa Levels: Who Cares? Part 1.5

Oops, I’ve got to backtrack a little. I just ran across a newly published study from the authors mentioned in Part 1 of this series a few days back. I pointed out some of the issues that surfaced as they tried to “hit the numbers” for factor anti-Xa levels in patients from their hospital. Here’s a breakdown of the new study.

First, I love the beginning of the title:

“If some is good, more is better”

Really?

Recognizing that 30% of patients had low anti-Xa trough levels when given the standard 30mg bid dosing regimen for enoxaparin, the authors engaged in some fancy predictive and statistical models to come up with a new one. A good portion of the methods section of the paper is devoted to explaining the machinations of exactly how they did this.

They used a patient dataset that was a little fresher than from Part 1. Three years of data from 2011 to 2014 were reviewed, and 275 patients were used to generate the new models. They selected one of seven candidates, based on a combination of simplicity and fewer supranormal levels of anti-Xa. They used this model to guide dosing to the next 145 patients. Here is the new regimen:

Weight Dose (q 12 hrs)
50-60 kg 30 mg
61-99 kg 40 mg
> 100 kg 50 mg

And here are the factoids:

  • Of the 275 patients used to create the model, 70% were subtherapeutic. (This is exactly the same number as in the first paper, but a different number of patients. Hmm.)
  • With the new dosing regimen in place, only 21% were subtherapeutic
  • Patients with supratherapeutic anti-Xa levels increased from 2 to 5% using the new routine
  • VTE was the same, at about 3-4%
  • Four patients developed VTE on the new regimen, and 3 of them had therapeutic anti-Xa levels (!)

Bottom line: A lot of modeling and statistical work went into the production of this paper. I still wonder why the number of patients included over 3 years is so low for such a busy center. But the authors certainly showed that they could improve the rate at which they “hit the number.” But how important is this, really?

The concluding sentence of the abstract reads, “further studies are needed to determine whether such dosing decreases venous thromboembolism rates.” Perhaps we should figure that out before continuing to spend lots of time playing with dosing changes and blood tests.

Reference: If some is good, more is better: an enoxaparin dosing strategy to improve pharmacologic venous thromboembolism prophylaxis. J Trauma 81(6):1095-1100, 2016.

Enoxaparin And anti-Xa Levels: Who Cares? Part 2

In my last post, I reviewed a study that looked at monitoring factor anti-Xa for the purpose of just “hitting the number.” Not very convincing. Today, I’ll review one that studied a reasonable outcome, the actual occurrence of VTE in patients.

This was another small, prospective study at a busy Level I trauma center. The outcomes that were analyzed included LOS, transfusion requirement, hematocrit on discharge, and diagnosis of deep venous thrombosis (DVT) or pulmonary embolism (PE). Only the last two of these make sense, especially for this small study. (205 patients in two 10 month periods).

At this center, all trauma patients are started on enoxaparin, regardless of injury severity. And all patients have sequential compression devices applied unless contraindicated by their injuries. Patients were included if the were administered 3 consecutive enoxaparin doses and had a trough anti-Xa level measured an hour before the fourth dose. If the trough was less than 0.1 IU/ml, dosing was adjusted until it rose to > 0.2 IU/ml. Outcomes were compared to historical controls from the prior year.

Here are the factoids:

  • A total of 87 study patients were enrolled in 10 months.  However, this represents only about 15% of trauma admissions to the center. Why were so few eligible for inclusion?
  • 84% of study patients did not “hit the number” with 30mg bid dosing (again!)
  • They were compared to 118 control patients who received enoxaparin during the same 10 month period, a year earlier
  • Screening by duplex ultrasound was only done for “clinical suspicion” of DVT or PE. No routine screening. And we know how reliable clinical suspicion can be.
  • 84% of patients were not at their anti-Xa goal when the first trough was done. Most of these patients needed 40mg bid to “hit the number.”
  • DVT and PE occurrences were “significantly lower” in the dose adjusted group compared to historical controls (1.1% vs 7.6%). Now this is a difference between only 1 adjusted patient and 9 controls, and the p value barely made it at 0.046.
  • Proximal DVT occurred  in no adjusted patients vs 2 controls (not significant)
  • PE occurred in no adjusted patients and 1 control (not significant)
  • Distal DVT occurred in 1 adjusted patient and 6 controls (not significant

Bottom line: This is yet another (very) small study. It also demonstrates why you must read the study, not just the abstract! The study group was a fraction of all of the patient admitted, even though all patients supposedly received prophylaxis. The attending physicians decided when to start dosing, and this varied from 0 to 4 days. Screening was ordered only if there was some kind of clinical suspicion for DVT or PE, and the details were not spelled out. 

For all these reasons, there are many, many opportunities for bias. But probably the most important problem is the statistics. I always worry when the p value for a numerical difference barely reaches 0.05, especially when the actual numbers look to be far apart. It is usually an indicator of small study size.

But in this case, the breakdown of VTE location is critical. The sums of the distal, proximal, and pulmonary occurrences show a p value difference just under 0.05. But when you compare study vs control for each, the bulk of the numbers are due to distal DVT.  The literature does not convincingly support prophylaxis for distal DVT, and we do not even treat it at my center. We continue surveillance to make sure it doesn’t creep up into the popliteal arteries.

This is yet another weak study trying to make the case for anti-Xa monitoring that doesn’t pass muster. Again, we see that 30mg bid doesn’t “hit the number” without adjustment. But we also haven’t shown that hitting that magic number of 0.2 IU/ml (peak or trough) by adjusting the dose makes a difference either.

But we continue to try. In my next post, we’ll look at another recently published study on the same topic.

Related posts:

Reference: Association between enoxaparin dosage adjusted by anti-factor Xa trough level and clinically evident venous thromboembolism after trauma. Jama Surg. Published online ahead of print July 6, 2016.