Episode 8.4 Anticoagulation Use In Pregnancy
In this episode, we discuss prevention of thromboembolism during pregnancy and the postpartum period and the evidence for which patients should receive anticoagulation. Plus, the history of heparin and four tips for preventing embolism.
00:00:02 Four Tips for Preventing Thromboembolism in Obstetrics
00:08:39 Anticoagulation Recommendations in Pregnancy
00:22:29 Assessing Anticoagulation Recommendations in Pregnancy
00:35:22 Evaluating Anticoagulation Protocols in Pregnancy
00:42:22 Debating Risks of Obstetric Anticoagulation
00:59:28 History of Heparin
Links Discussed
Risks of Venous Thromboembolism After Cesarean Sections: A Meta-Analysis
Caprini Score for Venous Thromboembolism
Postpartum Heparin Thromboprophylaxis
Evaluation of a Risk-Stratified, Heparin-Based, Obstetric Thromboprophylaxis Protocol
A More Selective vs a Standard Risk-Stratified, Heparin-Based, Obstetric Thromboprophylaxis Protocol
Rate of Wound Complications With Enoxaparin Use Among Women at High Risk for Postpartum Thrombosis
Transcript
Announcer: 0:02
This is Thinking About OB-GYN with your hosts Antonia Roberts and Howard Herrell.
Howard: 0:17
Antonia.
Antonia: 0:18
Howard.
Howard: 0:19
What are we thinking about on today’s episode?
Antonia: 0:21
Well, we’re going to talk about prevention of thromboembolism with anticoagulants like enoxaparin or heparin and some of the controversies about who should receive those drugs and who shouldn’t, and we’ll also hopefully have time for the history of the discovery of heparin as our historical tidbit. But to complete this whole theme, we also have a quick four tips for this episode, which is four tips for preventing thromboembolism.
Howard: 0:49
Wow. So this will be one of the rare episodes where we really only talk about one thing.
Antonia: 0:54
Yeah, we’re going to try to really milk it. So our first tip is to only use prophylactic anticoagulants for patients at high risk for thromboembolism, and I should clarify this four tips regards obstetric patients, so only patients at high risk for thromboembolism during the antepartum and peripartum periods, as indicated.
Howard: 1:18
Okay. Well, that first tip is way more complicated than it sounds, I think. So, essentially, we’re going to spend the rest of the episode talking about and understanding which patients actually benefit from prophylactic treatment and which are potentially harmed by it, or at least not helped by it. And honestly, right now out there it’s the wild wild west where different programs and different hospitals are creating a bunch of different risk-based strategies to determine who they should anticoagulate, and they’re not really supported by much evidence or good evidence. So we’re going to spend most of the rest of the episode defining tip number one.
Antonia: 1:51
Okay, so we’ll just for now move on. But we’re going to circle back to number one. But we’ll move on to number two, which is use sequential compression devices, or SCDs, for patients during and after a cesarean.
Howard: 2:04
Okay, yeah so, and honestly, even this tip is more controversial than it might seem to at first blush or to anybody who’s so used to doing this.
Howard: 2:13
It’s certainly become the standard of care, but the benefit of this intervention has been like a lot of things extrapolated from non-pregnant patients, which we’ll see that a lot in our discussion about anticoagulation in general during the rest of the episode, and that’s not usually a fair thing to do.
Howard: 2:30
So we know that patients who have a cesarean delivery are at four times roughly the risk of venous thromboembolism compared to patients who give birth vaginally, and so the recommendation of using these pneumatic compression devices has become universal because we know that’s a risk, right, but this is level C recommendation, level C evidence, which means it’s based on expert opinion only. It’s not based on data. There are some epidemiologic studies that have been done after the time in which we adopted this, which seemed to show a decrease in the rate of thromboembolism for cesarean delivery, but the difference is so minimal and it’s really unclear if you can attribute the diminishment in some of these studies to these devices specifically. But at this point it is a consensus standard of care to use them and, if anything, it might mitigate the hypotensive effects of the spinal if you have them on during that part of the procedure. So there is some benefit there perhaps.
Antonia: 3:28
Okay. Well, our third tip has to do with prevention, which is not to recommend bed rest, or even modified bed rest, for any obstetric condition.
Howard: 3:37
Right. Well, ambulation is the key, and the main harm from any type of activity restriction, and bed rest in particular, is the increasing risk of thromboembolism that’s associated with that. So those patients they could wear sequential compression devices on their legs, but patients aren’t great at doing that for those sorts of situations where they might be in the hospital for days or weeks prior to then maybe undergoing a prolonged induction of labor followed by a cesarean delivery. That’s the cascade of events that often leads to the increased risk of thromboembolism and causes them. And while it’s unclear whether SEDs or, frankly, even Lovenox, makes a significant difference in changing those rates of thromboembolism in those at-rest patients, it is clear that recommending that they confine themselves to the bed just makes things worse for no benefit.
Antonia: 4:27
Yeah, there just are not any obstetric indications or benefits for anyone from activity restriction or bed rest.
Antonia: 4:36
Even if someone is in the hospital let’s say they’ve had preterm premature membrane rupture or they have a severe hypertensive disorder or I don’t know, they have mono-mono twins or they have quadruplets just anything crazy you can think of they should still be getting out of bed and walking, sitting in their chair and doing physical activities that decrease their risk of thromboembolism and maintain their muscle tone and kind of their bone density a little bit and their general conditioning. There’s simply no reason to make someone be stuck in bed for a prolonged period of time. Even a severe preeclamptic might be stuck in the bed for 24 hours or longer due to being given magnesium sulfate, which can make them somewhat sedated. So again, there’s a great opportunity to use sequential compression devices, but then also stop the magnesium, maybe by 12 hours if they’re not progressing into delivery or then, if they have delivered, maybe 12 hours postpartum. In most cases that’s going to be all they need, even the severe preeclamptics, and just get them out of bed.
Howard: 5:53
Agreed Ambulation is the key. We have this tradition of bed rest and activity restriction. That is just not scientific.
Antonia: 6:01
Okay, and then long-time listeners are already going to know what tip number four is. I’m sure they can guess. But that is reduce your cesarean rate.
Howard: 6:10
It’s like a broken record around here. Of course, you could give these patients who have cesareans anticoagulation, and we’re going to spend the rest of this episode discussing the benefit of who should be receiving it and who should not, what the pros and cons are. But one of the reasons why people are interested in anticoagulating mostly C-sections is because that rate is higher. So should 2% of women be getting anticoagulation? That’s about the average that a lot of us have traditionally done, or is it 4% or 8% or 16% or even 80% of patients getting anticoagulation around the time of delivery? You see those ranges reported across the country for rates of anticoagulation in major hospitals all over the map. But since cesarean delivery increases your risk so much, one thing that we know you can definitely do that will diminish your risk of thromboembolism on your unit.
Howard: 7:04
While you’re debating all of these various strategies we’re going to talk about is you could reduce the number of patients who get a cesarean and because it’s fourfold increased risk, you’ll reduce your rate in that population of patients.
Howard: 7:17
You’ll reduce the rate of thromboembolism by 75%, and we know that as many as half of cesareans in the United States are unnecessary, so the rate of thromboembolism is about 26 per 10,000 cesareans in the US, and we, of course, have a cesarean delivery rate of around 34 percent.
Howard: 7:38
Now the rate of thromboembolism is about 7.2 per 10,000 for patients who have vaginal deliveries, and these data, if you look them up, they’re all over the place. They vary widely, mostly depending upon how DVT is diagnosed and then what population, what risk level are you studying? But the numbers I’m giving you from an article from CHEST from 2016. But anyway, if the cesarean delivery rate were 20% instead of 34%, let’s say, and you can see then, that that would lead to a substantial reduction in the rate of thromboembolism in that segment of the population and a much greater reduction in the rate of thromboembolism than people are hoping to find or implement with some of these protocols we’re going to talk about. So, first, do no harm, don’t do unnecessary surgery, and people won’t have as much risk for not just thromboembolism but a lot of other complications.
Antonia: 8:31
All right. Well, that was a great four tips that’ll take us right into the meat of our discussion, so let’s get into it.
Howard: 8:39
Well, I think first we should talk about exactly what ACOG says about anticoagulation for pregnant and postpartum patients. So for patients with inherited thrombophilias or personal history of DVT, there are some prescriptive recommendations that ACOG makes, and these are the easy cases. These are the ones we know this works and we all agree on.
Antonia: 9:00
Yeah, so that’s discussed in practice. Bulletin number 197, which is inherited thrombophilias in pregnancy, they recommend screening patients for these inherited thrombophilias well, firstly by history, but then by blood tests only if it’s going to affect management and if they have a suspicious history for it. So that might be the case in someone who has themselves previously had a venous thromboembolism. Or maybe they have a first degree relative that has a specific known mutation that we would treat in pregnancy if our patient also had that mutation. And remember there are multiple mutations that are associated with DVTs that we wouldn’t necessarily treat, at least by themselves, without the patient also having actually had a DVT. Remember that outside of pregnancy oftentimes if someone gets a DVT, they don’t necessarily need to get this blood work done to see if they carry these mutations, because outside of pregnancy it’s not necessarily relevant to know that information.
Howard: 10:07
Right, remember we do things because it affects the management. So most medical and even surgical patients, they’ll receive the same anticoagulation treatment and protocol, regardless of why they had a DVT, and we also have to appreciate that patients with low-risk thrombophilias, with no other risk factors, well, they don’t require anticoagulation while pregnant. They also don’t normally require postpartum anticoagulation unless there are other risk factors that are present. So this could include cesarean delivery or obesity or immobilization or something like that.
Antonia: 10:40
Yeah, there could be various reasons why someone who is low risk might know that even though they never had a DVT, they still carry these mutations. Maybe, for example, they’ve done some kind of expanded direct-to-consumer genetic test like some kind of 23andMe thing. Or maybe they were going through maybe infertility treatments and they got massively expanded carrier screening. Or sometimes they just ask to get tested for things and get tested, maybe inappropriately.
Antonia: 11:15
But these low-risk thrombophilias that are defined in the ACOG practice bulletin would include the factor V Leiden heterozygous and also the prothrombin heterozygous mutation, as well as protein C or protein S deficiencies. And there’s plenty that are not even don’t even warrant being mentioned by the ACOG. So these are just the ones that are mentioned by them. So if the patient has one of these mutations and a first-degree relative who has had a blood clot, then that raises this patient’s risk just a little bit and for that patient we would anticoagulate them, but only postpartum that’s the highest risk time to get a DVT. If they themselves have had a personal history of a blood clot, even though they have just a low-risk mutation, that raises their risk further and for them we would anticoagulate in pregnancy as well. So this is all outlined in the ACOG 197 practice bulletin.
Howard: 12:17
Yeah. And then there are the group of high-risk thrombophilias, which would include factor V-laden and prothrombin gene homozygous mutations, or if you have the double whammy of both a factor V-laden and a prothrombin heterozygote mutation, even if they’re the lower-risk versions, or an antithrombin deficiency. So patients with those mutations would receive anticoagulation during pregnancy and the postpartum period, even if they’ve not had a previous personal history of thromboembolism.
Antonia: 12:47
So it follows if your OB patient has a first degree relative with one of those mutations, then screening your patient would also be appropriate because even if they have not had a DVT themselves, if they carry that mutation it would affect your management. There’s a nice table in the practice bulletin that explains who should get the lower dose versus the intermediate, versus the higher doses of the Lovenox or the alternative anticoagulant based on their level of risk. And I think generally we are pretty good about screening for these kinds of personal and family histories at the initial OB visits because that’s also what ACOG recommends and a lot of times it’s just on our standard. New intake forms some kind of question like have you ever had a blood clot or a family member with a blood clot. So we’re generally aware if the patient or their relative has ever had a thromboembolism and we’re usually taking that into account during their pregnancy and usually pretty good at ordering probably too many tests If anything. We might err on ordering things that are not included in baseline. Whatever that is plus or minus.
Howard: 13:56
One thing I think people do get wrong is exactly what, as you said, what thrombophilia tests need to be ordered and in whom. So if there is a first degree family history, then just check for the known mutation. What did mom or dad have? But if someone has a personal history of thromboembolism, we know we’re likely going to treat that, but knowing about the genetic abnormality may affect the duration and the dosages. So, in addition to the thrombophilia as we just mentioned, well, that would include testing for antiphospholipid antibody syndrome, for example.
Howard: 14:37
But it doesn’t include what we often see too much of and that’s testing for the methyl tetrahydrofolate reductase mutation and specifically the C677T polymorphism. So being neither a heterozygote or a homozygote for that mutation has any association at all with thromboembolism or with any other negative pregnancy outcome for that matter, and patients should not be tested for it. These mutations are quite common among patients of European extraction, but they’re not associated with an independent increased risk of thromboembolism in pregnancy or outside of pregnancy, nor is hyperhomocystinemia, and there’s no evidence that supplemental folic acid for these patients with these mutations, even if they’re homozygotes, makes any difference in outcomes whatsoever.
Antonia: 15:21
Another one I’ve seen tested a lot is the PAI1 gene and there’s a blurb in the practice bulletin about these other thrombophilias that lists basically what we’ve been talking about in different variants of the factor V Leidin or protein Z, and it just says these appear to exert little independent risk for VTE. So the overall insufficient evidence. So just not something we should be routinely be testing for. But in this case we have some directive advice about how to treat that 2% of patients who either have had a history of a DVT themselves or have had a first degree relative with that history, and then the patient personally does carry one of those mutations. So that’s not really a controversy, that patient population. For sure they should be anticoagulated. But now we really want to discuss the more gray areas and that is patients who may have other lesser or acquired risk factors for thromboembolism but they don’t otherwise fit into that category. So they haven’t had a DVT before and they don’t, to our knowledge, have one of those mutations.
Howard: 16:36
Right. So that gets us to practice bulletin 196, thromboembolism in pregnancy, which actually covers much of the same ground, but this new ground as well.
Antonia: 16:45
Yeah, so 196 and 197 go together a little bit and there is a little bit of overlap, but just maybe slightly different focus. So 196 describes the various anticoagulants that could be used during pregnancy and who should get them. So that’s what we’re going to focus in right now. They, of course, recommend that women be assessed for the things that we just discussed, so reinforces that part of practice Bulletin 197, the family history, personal history, etc. And they conclude that there’s insufficient evidence to guide clinical decision-making regarding routine pharmacologic thromboprophylaxis during and after pregnancy. And they call for large-scale, high-quality research. And they mention that what we do know is largely extrapolated from non-pregnant populations and studies on people that are not obstetric patients and therefore it’s up to the individual facility or even the individual provider to customize their plans for their patients based on these other risk factors.
Howard: 17:53
Yeah, and that’s where we get into trouble, perhaps because some of the risk tools that exist out there could lead to a large portion of OB patients getting anticoagulated, particularly in the postpartum period and, as I mentioned, as many as 80% of patients. Depending on which of these criteria you might use, they could meet the criteria to be treated when you use these screening tools, but we don’t know that any of these tools are valid for the pregnant population of patients or postpartum patients, and we also don’t understand if the risks that a person might have obesity, surgery, all those things, whether they’re synergistic, do they augment one another and combine? Are they additive, are they multiplicative or are they neutral? And that’s a really important point when you’re using these scoring systems, because you quickly accrue a bunch of risk but you actually might not be at increased risk.
Antonia: 18:49
Yeah, and for all we know, some of those risk factors could even cancel each other out. We just don’t know. So, for example, is being obese and smoking a risk factor worse than just being obese alone? Or just smoking alone for a pregnant patient? Or what if one patient has class three obesity but she doesn’t smoke, but then this other patient has class one obesity but they smoke two packs a day? Which one’s worse, and how would each additional other kind of risk factor interact with these two risk factors? And they probably don’t affect all patients the same way. There’s probably all kinds of other environmental and genetic confounders in there. So scoring systems that linearly try to give points for each identified risk factor, as if each of them are on equal footing, and just linearly add up may be very inaccurate.
Howard: 19:48
Yeah, there’s some evidence. That’s exactly true. One lesson of our podcast over the last few years is that things that just make sense are often not validated by science, and so we seek for empiric data rather than theory, and in this case, there just aren’t high quality trials that have shown that giving the patients that you’re describing besides the ones we discussed already, that let’s call it 2% that giving them anticoagulants during pregnancy or the postpartum period, actually saves lives or reduces the rates of thromboembolism or complication rates. And recall that giving an anticoagulant is not a harmless or innocuous procedure. We’re often giving patients anticoagulants who just had major abdominal surgery and the consequence of that might be bleeding. That causes its own set of problems as well as, in its own way, increasing mortality and morbidity.
Antonia: 20:40
They do say in this ACOG practice bulletin that no widely accepted scoring system has been prospectively validated in obstetric populations, but at the same time they do discuss some risk assessment tools for VTE in pregnancy. So specifically, the Royal College of Obstetricians and Gynecologists in the UK uses some risk factors to stratify patients as low or intermediate or high risk, and the American College of Chest Physicians also has an assessment tool to identify women at increased risk of postpartum thromboembolism.
Howard: 21:16
Yeah, so that first one you mentioned from the Royal College is actually the Green Top Guideline, number 37A I mentioned before. I love these Green Top Guidelines and they have these online. You can read them. If you’ve never seen them, just Google and take a look at them.
Howard: 21:30
But basically they say that in this one that all women with class three obesity, which is a BMI greater than 40, should receive prophylactic Lovenox for 10 days, or women with two or more persisting risk factors should receive prophylactic Lovenox for 10 days, and then they list those risk factors in a table and, apart from known thrombophilias that we’ve already discussed, this table includes things like cancer, heart failure, systemic lupus, type 1 diabetes, nephrotic syndrome, sickle cell disease plus any obesity age range of 35, sickle cell disease plus any obesity age rate of 35, parity of greater than or equal to 3, smoking, varicose veins, paraplegia, multiple pregnancies, preeclampsia, cesarean delivery, labor longer than 24 hours, stillbirth, preterm birth, ivf pregnancy, postpartum hemorrhage, operative vaginal delivery, postpartum sterilization, history of hyperemesis, admission longer than three days with bed rest, infection, including ETI, and more than four hours of travel. Recently I sound like a pharmaceutical commercial.
Antonia: 22:32
Yeah, like almost anything anyone could have possibly done it seems like is listed as a risk factor and some of those are interesting and they do, for some of them, list what limited evidence there is to suggest that association. But for most of those are interesting and they do, for some of them, list what limited evidence there is to suggest that association. But for most of those factors there’s no quantification, there’s like a little reference that then you have to go pull up some paper from the 80s or something, so it doesn’t actually give a weight to what that association is. And if, really, if you go through that list, how is that not at least 99% of OB patients falling into one of those, at least one of those categories? I think I would easily fit at least a few of those. So I guess if I was going to somehow happen to have a baby in the UK in the future, they’d probably want to put me on 10 days of blood thinners, which just seems crazy because I don’t think anyone here would think about that for a second.
Howard: 23:30
Right, and essentially all they’ve done is they’ve listed everything they can find that’s been associated with increased risk of blood clot, but that doesn’t mean that there’s any evidence that giving anticoagulants reduces that risk, and that’s the problem. And that list is that’s more widespread than the list about who gets aspirin to reduce preeclampsia, which we talked about. Even that’s on shaky practice, so they’re just picking anything that seems at risk.
Antonia: 23:58
They threw it on the table and God forbid we have a DVT, yeah, and I also love these green top guidelines in general. There’s a lot of them. They’re so readable and so digestible. They really lay things out in a nice format. So it’s nice short bullet points that you can read and there’s plain language where they explain their rationale and it’s even visually pleasing, like the font and the color and the layout. It’s even visually pleasing, like the font and the color and the layout. It’s nice to look at, it’s pleasant and it makes me think like how elegant these British obstetricians are.
Antonia: 24:33
But for this one in particular, just the rationale is it seems to be basically a zero tolerance attitude towards DVTs. They don’t want anyone to ever get a DVT, no matter what the cost is, no matter how many hemorrhages or hematomas. Give everyone these blood thinner shots, just so that no one dare get a DVT. So I guess one way to theoretically achieve that is to either anticoagulate everybody or just anyone that has even one slight risk factor. And it asserts that, for practical purposes at least, these risk factors are all equal and additive, although we know that physiologically that’s not true and that’s never been established, especially in pregnancy but even outside of pregnancy. And it recommends that as a rule, you just give anticoagulation to every C-section patient.
Antonia: 25:37
I think they said unless it’s completely scheduled, no labor, completely uncomplicated, but in any other case, c-section automatically means anticoagulation, and that even includes someone who maybe they had a hemorrhage in their C-section automatically means anticoagulation, and that even includes someone who maybe they had a hemorrhage in their C-section. And then you spent 90 minutes trying to obtain hemostasis because they lost over a liter of blood and now you’re supposed to anticoagulate them because they had a C-section and a hemorrhage. Well, there’s a good chance that now they might have more hemorrhage and have worse issues from that than they would have if you didn’t give them blood thinners. So I just wonder where does this aggressive attitude come from? Like was this from just a few OB-GYNs in the UK that happened to be involved in making these green top guidelines that they just really hate DVTs and this is their mission. Or maybe it’s some kind of widespread cultural thing. Maybe they thought it would be cost effective in their health system, and I don’t know if other countries take this extreme of an approach, but it seems quite extreme.
Howard: 26:35
Well, we have a few British listeners, I know, so maybe they’ll send us some comments about this. But yeah, and, but don’t hate on the British too much. Yeah, and, but don’t hate on the British too much.
Howard: 26:47
We do the same thing attitudinally with preterm labor or all the things that we talk about, where we’re just going to throw everything at it despite any no evidence saying because it might help one person in 10,000 or 100,000. The same attitude of DBT pulmonary embolism are bad things. We don’t want any woman to ever die from it. We don’t have any proof that it helps. There’s not any proof that it doesn’t necessarily, so let’s just do it for everybody and there you go. So again, my attitude is we should have some evidence that things are helpful before we do them, particularly when they cause harm. The difference here is given somebody procardia as a tocolytic, it’s hard to argue that it actually hurts anything right. But giving somebody a blood thinner, as you said, might kill them. So this is a little bit more evidence I think is needed Now.
Howard: 27:32
The ACOG practice, the other thing they cite, the other tool they cite is from 2016. And this was the National Partnership for Maternal Safety Consensus Bundle on venous thromboembolism, and this is part of the safety bundle which encourages the four R’s that a lot of us we’ve seen. We know these. They stand for readiness, recognition, prevention, response, reporting systems and this is the basis of a lot of our safety bundles around thromboembolism in US hospitals. This bundle recommends universal use of pneumatic compression devices for women who are not already on pharmacologic thromboprophylaxis, but it admits that for women with other risk factors the available recommendations are in conflict and that which risk factors are important is unknown. But it does suggest that women with Caprini scores of five or higher or for patients who’ve had a cesarean, that might be a population that you could consider pharmacologic thromboprophylaxis.
Antonia: 28:31
I am quite familiar with the Caprini scoring, especially during my time in military medicine. We used it a lot and we can link to the Caprini score calculator so our listeners, if they want to navigate to the website, they can play around with it. But it is actually harder than you would think to get a Caprini score of five or higher for most of our patients and of course it implies that’s added to the cesarean and not necessarily counting this as a point in and of itself. But recall, the Caprini score was created by, I think. But recall, the Caprini score was created by, I think, a cardiothoracic surgeon. He validated it on his patient population, so it has not actually been validated in OB patients.
Antonia: 29:13
There are OB modifications to the Caprini that other people have come up with. So that is also something that I’ve used in the past and it was built into our health record system and it prompted you to put in these scores for every patient. But especially for using it for OB patients, there’s a lot of assumption, with very little proof, that this is what is helpful for the patients. So for a typical patient who does not have a history of thromboembolism or a known thrombophilic mutation, then maybe you’ll get a point for swollen legs, a point for BMI greater than 25. And then, depending on which modification you’re using, maybe a point for being pregnant.
Antonia: 29:58
And even if you still count the cesarean as being the major abdominal surgery, you still wouldn’t anticoagulate that many patients because you still wouldn’t get to the caprini five. Certainly, and certainly not as many as the royal college scoring tool would suggest that you do. But even if you did have someone that reached that cutoff of five, I remember one of one of the modifications that I used to use that said surgery greater than 45 minutes. So anytime like an intern got to sew the skin it would make the whole thing 45 minutes or longer and sometimes that would put them over the cutoff. Well, even then, it’s not evidence-based that this OB patient with the Caprini 5 now is going to have a better outcome because you’re giving her Lovenox.
Howard: 30:44
Yeah, and that’s the important part is that zero of these systems have been validated in pregnant patients, and the vagueness about which scoring system we should use in the ACOG bulletin and which risk factors are important and how to weigh them, and no real formal recommendation.
Howard: 31:02
Well, all that has led to a lot of individual institutions or academic institutions or junior faculty who are interested in a research project or all this sort of thing, developing their own risk factors and trying to think through what might make sense and their own assessment tools and their own scoring systems or, as you said, modifying a Caprini score or some other something like that. And in general terms, it led to very large amounts of anticoagulation use in our patients, who may not be benefiting from it and may only, in fact, be harmed by it. And it also has this impact that we talk a lot about, about a whole generation of learners and residents thinking this is normal, especially if you’re not saying, hey, there’s zero evidence that this works, but we’re doing it and you see that effect. So we’re in the age of anticoagulation right now, where everybody going to graduate from residency program in the next few years thinks it’s normal to have 8%, 16%, 80% of their patients anticoagulated. So please tell your residents. There’s zero evidence that what you’re doing is effective.
Antonia: 32:05
Yeah, and it’s in the information void that’s been created by a lack of quality studies, along with this ACOG recommendation, that it seems to be telling us that we should just come up with our own systems, that a lot of programs around the country have been experimenting with this and probably in some cases harming patients with overutilization and promoting hematomas and hemorrhages. But hopefully, if they are experimenting around, they’re at least doing either a clinical trial where they publish their results or some kind of actual quality improvement, like internal analysis of outcomes, so that at least someone but hopefully the broader scientific community could learn from this experimentation. And that actually has been the case at University of Alabama, Birmingham, and so, thanks to them, now we do have a couple of studies that can provide some further insight into this.
Howard: 33:00
So the first study that they put out, their group was published in October of 2021 in the Green Journal, and this is a retrospective cohort study over a five-year period from 2013 to 2018. And during that time they created a protocol. So they’re comparing their pre-protocol experience to their post-protocol experience with their patient. Now their protocol took them from I think it was 1.2% of their patients at baseline. So that’s the 2% we were talking about before, plus or minus. So they were anticoagulating 1.2%. And then they created this protocol that got them to anticoagulate 15.6% of their patients and that involved nearly 25,000 deliveries with this implementation of this protocol.
Antonia: 33:43
Yeah, and if you think about those numbers, that 1.2% of patients that they were initially anticoagulating really represents that fraction that we were talking about of the highest risk patients. They’ve had a BVT before, they’re homozygous factor V Leiden, something like that that we all definitely agree should receive the anticoagulation. At UAB in this study when they implemented a protocol to be more broad, they also included patients with things like BMI over 40, low-risk thrombophilia, family history thromboembolism, age over 40, preeclampsia with severe features, endometritis, peripartum hysterectomy and then a few specific medical conditions like heart failure, coronary artery disease, lupus, nephrotic syndrome, other chronic kidney disease, diabetes or at least the more severe kinds of diabetes, or sickle cell disease or recent surgery, including cesarean. So they were brought more trending towards the UK.
Howard: 34:44
Right, it’s not exactly the Royal College criteria that we discussed, but it includes all the usual suspects of risk factors that many people talk about, a little bit more aggressive than the Caprini system would get you the modified Caprini. Some notable differences are that it doesn’t include smoking or postpartum hemorrhage, which are much more common than many of the other risk factors, but it gets all the usual suspects, and that led again to nearly 16% of their patients receiving anticoagulation. They also stratified the anticoagulation they received, so for patients with a BMI of greater than 40, for example, they received a higher dose of anticoagulation.
Antonia: 35:21
All right. So why don’t you tell us what they found?
Howard: 35:24
Well, they found zero difference in the rate of venous thromboembolism before and after the protocol was implemented. They actually had more pulmonary emboli and fewer deep vein thromboses post-protocol than they did before the protocol. But overall there was no difference in the diagnosis of venous thromboembolism in total after increasing their rate of anticoagulation significantly. And of course those other differences probably weren’t significant. But if anything there were more PEs right, like the thing you actually really want to avoid. But overall there was zero difference in the two arms.
Howard: 36:01
But they did find three times the rate of superficial wound hematomas and it was really no difference in other outcomes of concern, like maternal death, obviously, or ICU admission or need for transfusion, though all three of those were non-statistically significantly increased after implementation of the protocol. That might suggest that a larger study appropriately powered to find a difference in serious bleeding complications would show harm of that nature from the increased use of anticoagulation and not just simple things like wound hematomas and the infections that follow. Now I’ll put a link to that study as well as a really good editorial that accompanied that publication back then in the Green Journal. That discusses a lot of what we’re talking about and put some perspective on this.
Antonia: 36:48
All right.
Antonia: 36:48
So obviously that was a negative study.
Antonia: 36:51
Well later in, just a few months ago, in June 2024, that same group, with maybe a slight different mix of authors, published a follow-up study to this in JAMA where they revised their protocol a bit to be more selective about who received the anticoagulation, but otherwise they had a similarly designed protocol and this included 5,000 new patients under this stricter criteria.
Antonia: 37:18
And then they compared that to 12,000 patients from their previous very broad protocol. So in this JAMA paper they discussed how ACOG recommends postpartum heparin-based chemoprophylaxis, so that could include Lovenox according to a risk stratified algorithm. And it’s an interesting way for them to read that ACOG guideline, because the guideline doesn’t actually say that, but people have interpreted it in that way and they have taken it to mean that they should develop their own risk stratified algorithm. And so that’s what they did initially with their Brad criteria that led to the negative study. And then that’s what they did again here with their stricter criteria in the hopes that they could at least reduce those excess wound hematomas and maybe potentially zero in on some higher risk patients that might show a benefit this time, where previously they didn’t show a benefit.
Howard: 38:12
Yeah. So what’d they find? Well, they went from 16% of patients receiving anticoagulation in that first round of the protocol that we just discussed down to now 8% of their patients receiving anticoagulation. And remember, before either of these protocols they were only anticoagulating 1.2% of patients. And before we get into the results of this new study, I would also point out that it’s very interesting that they continued doing this anticoagulation protocol despite doing a study that showed it wasn’t beneficial and in fact was harmful.
Howard: 38:46
This is an example that we see all the time in our literature of a sunk cost fallacy. A lot of folks were investing a lot of time in finding some benefit or validating some protocol, thinking that ACOG is demanding that we make such a protocol. So when you don’t find what you wanted to see originally and in fact you find some harm from what you were doing and no benefit, then what you should do is stop doing it. If you find no benefit and only harm, stop doing the intervention. But what they did was perhaps try to reduce the harm while focusing on this smaller and perhaps higher risk patient population. The assumption was that by going too broadly, the benefit perhaps was diluted out by including too many patients. Overall, they were never going to get a DVT to begin with and so you just harm them without showing any benefit accrual.
Howard: 39:39
And there is some sense to that, except that only works if you actually decrease the numbers of thromboembolism in your original study and you were weighing that small decreased risk against the risk of intervention. But that’s not what they found. In other words, if you think about this in terms of a receiver-operator curve listeners love receiver-operator curves then the first study might have been too far to the right and by moving the threshold for intervention a little bit maybe you still include the benefits that you accrued without receiving the excess harm exposure. But the original study didn’t show any benefits, only harm. So if you weren’t going to tell me to have read the study already before we recorded the podcast, I would have predicted that the second study would also show no benefit but probably would show fewer hematomas, because that’s all they’ve done is moved a little bit down the receiver operator curve. But I can’t expect them to find benefit when they found no benefit in the first study.
Antonia: 40:42
Well, you are cheating a little bit because you’ve already read the study. But yes, they did decrease the risk of wound hematoma from 7% down to 3% by narrowing down their patient criteria. And again, they still found no difference in the rate of thromboembolism, which remained at a 0.1%. And I get it. I’m sure it’s very frustrating because women do die from DVTs and PEs, and the Royal College guideline made a highlight to say that almost all maternal DVT PE deaths are in women that have some additional risk factors. And so then they draw the conclusion that it must have been preventable.
Antonia: 41:27
If they had just been given this blood thinner, they wouldn’t have gotten the DVT and died. And then here this study it’s not showing any decrease when you just try to give the blood thinners to more people with risk factors. So they’re going to all this trouble, recruiting all these patients and doing everything that goes into doing a big clinical study to try to reduce this rate below 0.1%, this rate below 0.1%, but really, short of handing out blood thinners to every single patient, they’re not getting there. They are showing that they cannot decrease the rate below 0.1%. So it almost seems like if you really wanted to reduce the rate of DVT less than 0.1%. You just have to go to some really extreme measures that are not practical, like admit every pregnant patient for their whole pregnancy to be on a heparin IV drip, Then they wouldn’t get a DVT. But you, just you can’t do that.
Howard: 42:24
Well, you’re making an assumption. We need a study that shows that that’s true Common sense. It defies common sense, and that’s the point of empiric data.
Howard: 42:31
Yeah, yeah common sense, and that’s the point of empiric data. Yeah, yeah. Well, there’s a perhaps no one likes to say this, but there is perhaps an irreducible risk. We told you, though, in tip number four, a way that you can reduce a segment of your patient’s risk, and that’s by not doing C-sections right. So there are things we can do, like let’s pay attention to that 75% reduction by getting vaginal deliveries rather than cesareans.
Howard: 42:54
This doesn’t seem to be one of them.
Howard: 42:56
And recall, in the original study, the pre-protocol back before they started any of this, the pre-protocol rate of thrombin and blood sugar was 0.1% and the wound hematoma rate was similar in the pre-protocol group originally than it is in this new study. So maybe the harm equals out, but just recall, they showed no benefit from it, and all they really did in the second version of this was reduce the harm back to something more comparable to the original harm, and even that’s not fair to say, because it could be that 10 years later they have less wound hematomas in general. This isn’t a randomized controlled trial with a placebo arm, because they have changed their surgical technique a little bit, or they’re more routinely closing the subcutaneous layer or something like that. So we still don’t know that this means no harm. They just reduced it a little bit. They just reduced it a little bit. So the way to interpret this study is there’s no benefit to having 8% or 16% of your patients anticoagulated, compared to just having 1.2% of your patients anticoagulated when you take these studies in their totality.
Antonia: 44:05
And that’s at least from the UAB group. But those are not the only studies. So back in 2011, there was a paper in the Green Journal where, among over 2,500 women with cesareans, 68% of them met their set criteria for getting thromboprophylaxis. So that’s quite a lot of them, and not all of them ended up receiving it, because a portion of those patients were left to the discretion of the physician. So they might have held it, maybe if they had concern for excessive bleeding or something like that. But when they, in this study, compared the ones who got it strictly according to protocol as opposed to those who didn’t get it for some reason, even though they met the criteria, they found no difference in the rate of thromboembolism. But they did find a near doubling in the rate of thromboembolism. But they did find a near doubling of the rate of wound separation and a triple rate of re-hospitalization related to wound complications in the group that received the anticoagulation.
Howard: 45:07
Right and again, with no decrease in the risk of being thromboembolism in the patients who receive the cesareans, even the higher risk portion of them, no benefit.
Antonia: 45:17
Yeah, exactly so. Systematic review and analysis of the guidelines and recommendations was published in the British Journal of Obstetrics and Gynecology in 2018. And that was before these new papers from UAB and they concluded that the incidence of venous thromboembolism was being overestimated and, if anything, the harms were being underestimated from getting Lovenox or heparin. And they also estimated that the number needed to treat to prevent one death from a PE after cesarean in the first week after surgery, if we were to assume 70% protective benefit from the anticoagulation, would be 360,000 women, and that number would include nearly a thousand women having a major hemorrhage as a side effect of the medication. Even if that number 360,000 was true which it may still not be true if we’re actually overestimating the incidence and underestimating the harms- Right.
Howard: 46:22
And if you expose a thousand women after cesarean to hemorrhage, this unexpected hemorrhage or especially occult hemorrhage that you’re going to have with internal bleeding, you’re going to have some deaths. And we’re seeing an increased risk of death from concealed hemorrhage, meaning this post-cesarean intra-abdominal hemorrhage. We don’t pay attention to it, as well as the vaginal bleeding that happens after deliveries. We’re seeing that in the California perinatal data, where they reviewed the maternal deaths and now this is a leading category of death is concealed hemorrhage.
Howard: 46:57
Even the new studies from UAB, they’re not powered to see deaths from pulmonary embolism, that’s true, but they’re also not powered to see deaths from hemorrhage as a side effect of anticoagulation. But if we accept the premise that anticoagulation should result in 70% fewer thromboembolisms in general, they are powered to challenge the assumption that the model that you just described is based on and they didn’t see a reduction in the rate of thromboembolism. In other words, okay, we’re not going to see the rates decrease at PE, there’s just not enough of them in the study. But we should see 70% fewer DVTs in the subset that’s powered to see that and they found no difference. So therefore, since we find no difference in DVT, we really shouldn’t assume that there would be a reduction in fatal pulmonary embolism either. Truth is, the number needed to treat is likely higher than the 360,000, but even with that number needed to treat, you’re going to expose patients to hemorrhage and you’re going to have as many, or perhaps more, deaths due to hemorrhage, as you might say, from a pulmonary embolism.
Antonia: 48:02
So the bottom line is we should be having a rate of obstetric anticoagulation of about one to two percent, based on a personal history of DBT or PE or the existence of a high-risk thrombophilic gene mutation in that patient. And just as a side note, even if the number needed to treat was much lower than what we said let’s say 1 in 4,000, you would still need such a massive trial to see the benefit that it really would be hard for that to be feasibly done. In the oncology literature it is much lower, it’s about one in 50, but we know that OB patients simply are not the same as cancer patients.
Howard: 48:46
Well, sure, but the number needed to treat isn’t one in 50, that’s for sure, and it isn’t one in 4,000.
Howard: 48:54
There’s no reason to think that either, and I think that’s the main problem, and I would encourage folks to read this review that you mentioned in the British Journal, which shows that the number needed to treat is really significantly higher than we would assume. But yes, if you assume that the number needed to treat was one in 50, you’re going to make all sorts of mistakes. When you just model out the theoretic benefits, and even if it’s 1 in 4,000, well, you might be able to argue that the benefit of preventing a DBT or a PE or whatever you’re arguing, outweighs the risks of the hemorrhage. But pregnant patients aren’t non-pregnant cancer patients, and obstetric surgery is not gynecologic surgery, nor is it general surgery, and the conclusion that you would bring from that is completely false. It’s not supported by evidence, and these UAB articles, though they’re not prospective trials, they are the best evidence I’m aware of right now, and it agrees with the prior evidence that we have on this issue.
Antonia: 49:52
Yeah, so if the real number needed to treat is around 1 in 360,000 or even higher, we could never feasibly do a study in the real world that could show that degree of benefit.
Howard: 50:05
Well, there are plenty of things that are statistically significant but clinically insignificant. But if something is clinically significant, it will always be statistically significant as well. Right? So it’s true that you may not have a study powered sufficiently to show a statistically significant benefit. That’s called a type 2 error. But you’re not allowed to then just make up a benefit when you haven’t shown anything. If there was a trend to reduction in PE or DVT and you extrapolated out how many numbers you would need, or whatever, that would be one argument. You could say that you could argue for a type two error.
Howard: 50:44
But you can’t claim clinical benefit from something that hasn’t been proven and you can design a bigger and better study. Or you can use surrogate outcomes or other markers that occur more commonly to try to show some benefit. In this case, that would be showing a decrease in DVT. DVT is essentially the surrogate marker for pulmonary embolism and subsequent maternal mortality, what we’re really worried about. So we know that a fraction of patients with a DVT will go on to experience pulmonary embolism and then about 2% of those patients will die.
Howard: 51:16
So your study may not be powered to show that mortality benefit, but if it’s at least powered to show a decrease in DVT, then that could still have some credibility. But if you can’t even show a difference in the surrogate outcome of a DVT that you’re powered to find Well, you can’t go on and claim clinical significance in these other categories without statistical significance there. So I hear these arguments about this going around right and you can say that something’s not clinically significant even though it’s statistically significant. We do that all the time. But you can’t argue that something is not statistically significant but is clinically significant and you’re just making that out of whole cloth really.
Antonia: 52:00
That’s the harsh truth. Different facilities may selectively just pick and choose the criteria that seem to make the most sense to them. So one example is let’s say someone has kidney disease or preeclampsia and they’re spilling a lot of protein in their urine and in that protein we know they’re losing antithrombin 3. So it would theoretically make sense to put those people on anticoagulation because it would make sense that they’re at higher risk of DVT.
Howard: 52:30
Yeah, and I can be a little sympathetic to arguments like that. Right, absence of evidence isn’t evidence of absence. And of course, there is no individual study that anybody can cite that says that patients with antithrombin 3 deficiency due to massive proteinuria don’t benefit from anticoagulation. So maybe they do, but still, all of these studies that we’ve been discussing, well, they include all of those patients and one’s exactly like that that have these high-risk sorts of things and they’ve not shown benefit, right, so the hole hasn’t shown benefit.
Howard: 53:03
Could there be subgroups or parts in the hole that do? And I think if you want to do that, that’s fine, but what you should do is enroll that patient in a clinical trial. What you should do is enroll that patient in a clinical trial and someone should design a trial, maybe through the MFU network, that is, a multi-center trial that starts recruiting these super high risk and exotic things and randomize them to receiving anticoagulation and do it over four or five years and let’s see if there’s a benefit for these most at-risk patients. But we just don’t experiment on patients, particularly when we’ve just cited all the relevant literature about this topic and it says that these patients are only harmed by anticoagulants. There’s no end to the what-ifs of the world and the coulda shoulda wouldas that this is ultimately just anecdotal practice in medicine and not evidence-based practice in medicine.
Antonia: 53:50
Okay. Well, what if someone wanted to argue that with that UAB study they brought the rate of wound hematomas down to essentially their initial baseline with that second restricted criteria? So maybe that’s the criteria we should be using, because we at least didn’t increase the harm, even if we didn’t catch the benefit. I’m sure there’s people that are just can’t wrap their minds around any other logic that you’re treating more people, so you’ve got to be benefiting someone because they have the risk factors and we know risk factors increase their risk. So there’s got to be some benefit and there’s not any more harm. So why not at least use that more restricted UAB criteria?
Howard: 54:38
Yeah.
Howard: 54:38
And this, again, is the argumentation that gives us tocolytics and all the other things that we talk about. I don’t think that’s what the authors of this study were intending folks to take away from their work. They concluded that folks who have been suggesting this type of approach and guidelines need to rewrite. They say that the guidelines should be changed right Because they’re wrong. But it doesn’t matter. If something is just not harmful, say Procardia, it matters whether or not ultimately it is beneficial. And the failure of the first study at UAB and the second study at UAB is not the hematomas, it’s the fact that it didn’t decrease the rate of DVT at all. So you don’t use it because it doesn’t work. Sometimes you pay a consequence in patient harm for using a thing that doesn’t work, and sometimes you don’t, or at least initially. But again, that’s what we talk about a lot on this podcast. This is the story of DES all over again. Des didn’t work, it didn’t decrease the rate of miscarriage and it didn’t look like it was harming anyone too badly either. So people kept using it until they found out a generation later that in fact was harming.
Howard: 55:45
The same arguments about study size that relate to the potential number needed to treat are true for number needed to harm.
Howard: 55:52
In fact they’re greater.
Howard: 55:54
For the issue of harms in this case, I think there’s little doubt that the number needed to harm meaning causing bleeding after a cesarean that’s missed and leads to BIC and death that that number needed to harm is much lower than any theoretic number needed to treat, and it’s not captured in any studies either, because they’re not powered to see that.
Howard: 56:16
So if the argument is that, oh, it’s underpowered to see the benefit if they’d included 100,000 people we might’ve saved a life, or something like that Well the burden is on that person making that argument to prove it and to demonstrate that with scientific evidence. But you have to appreciate that the same thing’s true too that it wasn’t powered to find the death from the hemorrhage, and so they have to prove that the harm wasn’t going to occur at that scale as well, because the study’s underpowered for both things. The number needed to harm is lower from what we understand than the number needed to treat would be, and harm is not a wound separation, harm is someone bleeding to death. You don’t get to pick out the type 2 errors for the interventions and possible benefit in a study without discussing the type 2 errors for the potential harms from the intervention.
Antonia: 57:03
All right, well, lastly, what about? We talked about how C-section is one of the biggest risk factors, besides the personal history of the DVT. So what about going through that major surgery and activating the coagulation cascade? Or even if they have a hemorrhage or a cesarean hysterectomy? I guess it’s the same kind of question over again. So why not just pick out something really high risk, like one of those types of patients, and then selectively anticoagulate only the patients with the massive hemorrhage and the peripartum hysterectomy?
Howard: 57:45
Right, it’s the same question, it’s the same answer, except in this case the irony is you’re wanting to give an anticoagulant whose biggest risk factor is causing concealed intra-abdominal hemorrhage and potential maternal mortality from that to a patient who just experienced major intra-abdominal hemorrhage, and you’re doing it for an unproven and theoretic benefit that it might prevent her from getting a blood clot. So, yeah, some of these patients are the ones who most likely benefit from anticoagulation, but in this case they’re the ones most likely to be harmed by the anticoagulation. So maybe give that patient some tranexamic acid and not an anticoagulant. It makes sense, right, that you pick the highest risk patient to focus on, but you pick the patient most likely to focus on, but you pick the patient most likely to benefit and the patient most likely to be harmed, not you.
Howard: 58:39
By the way, these are questions that we received already, so we’re putting these out here and positing them because it does make sense theoretically and you can talk yourself into the potential benefit but recognize that you’re also having an even greater burden of potential harm.
Antonia: 58:56
So I don’t think we’re going to convince you to give anticoagulation to anybody extra, are we?
Howard: 59:01
Well, someone can certainly do a study in these subgroups like you mentioned and massive proteinuria or something like that, and you could convince me very easily. If you find some benefit, even from surrogate markers, even just show a reduction in DVT, then yeah, we could do that. But the burden of proof is on the person who wants to do the intervention and I’m not obligated to experiment on patients without scientific evidence or informed consent.
Antonia: 59:25
All right. Well, we don’t have much time left at all. Do you still want to do a little historical tidbit?
Howard: 59:30
We should finish it. So we’ve been talking about anticoagulation.
Antonia: 59:39
Let’s do the history of heparin real quick. Let’s do heparin so. Anticoagulation is one of the big medical breakthroughs that has made giving birth drastically safer than it ever was in the past. Only 2% of pregnant patients with pulmonary embolisms die today because we have the ability to anticoagulate them and, of course, we prevent many patients from developing them to begin with. So why don’t you tell us a little more about the heparin?
Howard: 59:59
Well, I won’t tell a whole patient story because we don’t have the time, but just the history of heparin. So recall that insulin was discovered by a medical student and an orthopedic surgeon in 1921. And heparin was also discovered by a medical student, a second-year med student and a physiologist, in 1916. So Jay McLean was a second-year med student at Johns Hopkins and William Henry Howell was the physiologist. Now.
Howard: 1:00:25
Dogs, as you recall, played a big role in the story of insulin and they do here as well. Insulin was isolated from the pancreas of dogs and heparin was isolated from the liver of dogs, and that do here as well. Insulin was isolated from the pancreas of dogs and heparin was isolated from the liver of dogs. And that’s why it’s called heparin, because hepar is the root word for liver in Greek. But it didn’t enter into clinical trials for use until 1935. So they discovered this in 1916, but it took until 1930s for it to become clinically studied. And then finally in Europe a Swedish company began manufacturing it for IV use in 1936. And in 1937, connott Medical Research Laboratories began making it available in North America, and Connott was part of the University of Toronto where insulin was also discovered and first manufactured. And today that has become part of Sanofi.
Antonia: 1:01:13
So this heparin was the first anticoagulant available as a medicine, right.
Howard: 1:01:18
The other old one was warfarin or coumadin, but that didn’t come about until the 1940s and 1950s really.
Howard: 1:01:24
So those are the first two therapeutic classes that we had.
Howard: 1:01:29
But as far as its impact on maternal mortality, right around 1930 or so we begin having widespread use of ergotamines to control bleeding, and then heparin becomes available in the middle of the decade along with sulfa-based antibiotics. So all of a sudden you see this tremendous decline in perinatal and maternal mortality, starting in mid-1920s, late 1920s down till about 1950. And you also see this dramatic increase in hospital births from about 55% to 90% by the end of that time period. And all this happened even before oxytocin was available. So penicillin was discovered during this time and heparin was part of this original trifecta of interventions that really made a huge difference. We developed the means to treat hemorrhage with ergotamines and anticoagulant, to treat thromboembolism with heparin and antibiotics, sulfa and then penicillin drugs to treat infection, and those three things were available in the hospital and so birth really made this dramatic shift from outside the hospital to the hospital during about a 10-year time period and along with that the largest and most dramatic drop in maternal mortality, chiefly because of these three interventions.
Antonia: 1:02:49
All right. Well, that was a good tidbit, some good perspective there. Well, we should wrap up for today and we’ll be back in a couple of weeks.
Announcer: 1:03:00
Thanks for listening. Find us online at thinkingaboutobgyn.com. Be sure to subscribe. Look for new episodes every two weeks.
