Episode 6.1 Antidepressants in Pregnancy and Cephalosporins in Patients with Penicillin Allergy
In this episode, we discuss current evidence about potential cross-reactivity for patients with penicillin allergy who receive a cephalosporin. Then we discuss the safety of antidepressant medications in pregnancy.
Drugs and Birth Defects
|ACE/ARB||fetal renal tubular dysplasia and neonatal renal failure, oligohydramnios, intrauterine IUGR, lack of cranial ossification|
|EtOH||FAS (growth restriction before and after birth, mental retardation, midfacial hypoplasia, renal and cardiac defects). Consumption of >6 drinks a day associated with 40% risk of FAS|
|Carbamazepine||neural tube defects, fingernail hypoplasia, microcephaly, developmental delay, IUGR|
|Cocaine||bowel atresias; congenital malformations of the heart, limbs, face, and GU tract; microcephaly; IUGR; cerebral infarctions|
|DES (diethylstilbestrol)||clear cell adenocarcinoma of the vagina or cervix, vaginal adenosis, abnormalities of the cervix and uterus or testes, possible infertility|
|Lead||increased spontaneous abortion rate; stillbirths|
|Lithium||congenital heart disease (Ebstein’s Anomaly)|
|Methotrexate||increased spontaneous abortion rate|
|Organic Mercury (from Whole Foods)||cerebral atrophy, microcephaly, mental retardation, spasticity, seizures, blindness|
|Phenytion||IUGR, mental retardation, microcephaly, dysmorphic craniofacial features, cardiac defects, fingernail hypoplasia|
|Radiation||microcephaly, mental retardation. Medical diagnostic radiation deliverying < 0.05 Gy to the fetus has no risk|
|Streptomycin and kanamycin||hearing loss; CN 8 dmg|
|tetracycline||permanent yellow-brown discoloration of deciduous teeth; hypoplasia of tooth enamel|
|thalidomide||bilateral limb deficiencies, anotia and microtia, cardiac and GI anomalies|
|Trimethadione and paramethadione||cleft lip or palate, cardiac defects, mricrocephaly, mental retardation|
|Valproic Acid||neural tube defects (spina bifida), minor craniofacial defects|
|Vitamin A and derivatives||increased spontaneous abortion rate, microtia, thymic agenesis, cardiovascular defects, craniofacial dysmorphism, micropththalmia, cleft lip or palate, mental retardation|
|Warfarin||nasal hypoplasia and stippled bone epiphyses, developmental delay, IUGR, ophthalmologic abnormalities|
[00:00:18] Howard: Antonia?
Howard: What are we thinking about on today’s episode?
[00:00:21] Antonia: Well, we’re finally gonna talk about the safety of antidepressants during pregnancy, and I guess we should point out that we’re beginning season six of this podcast.
[00:00:31] Howard: Yeah, that’s right. This is the 66th episode, and somehow we still have an endless amount of material to do.
[00:00:39] Antonia: Well, we’re always open to suggestions and we have a lot of good things planned for this season. But first, what’s the thing we do for no reason?
[00:00:48] Howard: Okay. Well, how about avoiding the use of cephalosporins in patients with a reported penicillin allergy?
[00:00:53] Antonia: Okay. Yeah, this is a great one. So we all learned in medical school that there is a cross reactivity between cephalosporins and penicillin-like antibiotics.
So because of that, if somebody reports a penicillin allergy, then there’s a tendency for their treating providers, especially surgeons, to avoid cephalosporins and use a different prophylactic antibiotic instead, or even to treat an infection. For example, if we have someone who’s gonna undergo a cesarean delivery and they say they’re allergic to penicillin, then it would be common for their provider to use Clindamycin and Gentamicin instead for fear that they will cause an allergic reaction by giving them Ancef. And this is a lot more likely to happen if that provider has not discussed the allergy with the patient beforehand. And this of course can be as simple as asking, well, what’s your reaction to penicillin? Or maybe, have you ever had Ancef or Rocephin or Keflex or something like that.
And even if the patient doesn’t know any of the answers, then sometimes the provider might even be able to look back at old records and see if the patient had gotten Ancef or a similar antibiotic in the past.
[00:02:04] Howard: Yeah, I had to do that for a patient today. So, but in some cases, I will say the aminoglycoside doesn’t get used, and Clindamycin alone is submitted in the case of a cesarean delivery. And it’s interesting, I’ve seen this happen and I think there’s some practical reasons why, and I think it might be related to Google. If you just do a Google search on which antibiotic you should use for a cesarean in a patient who has a penicillin allergy, instead of looking at the ACOG practice bulletin then you’ll find an old publication from Canada from 2010 that suggests just Clindamycin is all you need. But of course our current regimen in the United States is to add the aminoglycoside.
[00:02:41] Antonia: Yeah. And I know that even when we do routinely where I’ve practiced, order the aminoglycoside with the clindamycin, but that tends to take longer to get sent up from the pharmacy.
I think one reason might be that it’s weight-based dosing, whereas most other typical prophylactic antibiotics we use are not weight-based, and sometimes you just can’t wait for however long. Sometimes it’s an hour or more for them to send the gentamicin up. If you need to do a delivery, sometimes you just have to go forth and then give them the antibiotic after the fact, in which case it’s gonna be much less effective.
But yeah, it is interesting to think about how people obtain and discover knowledge now, especially medical knowledge. In the age of Google and artificial intelligence like chat GPT, it’s very tempting probably for younger physicians to just do a quick internet search on their phone when they just have a free second or two.
And then if that search produces a finding that appears to be relevant to their question, especially if there’s a reference to a scientific paper. Then there can be this search satisfying bias, and then they’ll just stop looking and they’ll just take that answer they found. But neither Google nor chat GPT are very good at understanding the relevance or the completeness of a particular finding, especially if it’s only derived from a single article like that.
[00:04:06] Howard: Yeah, that definitely happens. And it’s a brave new world of understanding how good these artificial intelligence, generative engines, and even just Google are, but people are reliant on them. But the thing is though, neither regimen adding Clindamycin, for example, and Gentamicin to do a C-section, provides the same coverage that a cephalosporin would.
So, really the goal is still to use a cephalosporin whenever you can, not just in obstetrics, but this is a big issue in total joints in orthopedics. So the main thing is we need to understand which patients with penicillin allergies are able to have a cephalosporin and which cannot have a cephalosporin, and use cephalosporin every time we can.
[00:04:46] Antonia: Yeah, that’s right. Even if we did always get the gentamicin up before the surgery starts and we appropriately administered it, it wouldn’t be as good. And the good news is that almost everyone with a self-reported penicillin allergy still will be a candidate for the cephalosporin, which is superior.
[00:05:03] Howard: Yeah, and I think that’s more true than either you or I realize. So there’s a new paper published in May 17th of this year in the Journal of Allergy and Clinical Immunology entitled Cephalosporins Can be Safely Administered to Patients with Verified Penicillin Allergy. This is published from a research group working out of Yale.
They point out that the true incidence of penicillin allergy is really quite rare. A lot of folks claim to have a penicillin allergy, usually related to some, purported reaction as a child, but the incidence of true allergy is exceedingly rare. But because of those reported allergies, as we said, people avoid the antibiotics and there’s a payoff.
There’s significantly elevated rates of surgical site infections, which is obviously tied to healthcare costs and even total mortality due to the avoidance – the unnecessary avoidance of cephalosporins.
Now before you raise your eyebrows at that last point, there is a retrospective cohort study in 2019 by Dr. Benjamin Harris and colleagues out of Duke, and they looked at the efficacy of cephalosporins versus non cephalosporins, specifically for cesarean surgical prophylaxis. And they found that the rates of surgical site infections were 7% after cephalosporin use versus 15% even when the appropriate alternative antibiotics were used.That’s more than double the risk.
So while you may not find a study that translates exactly over to mortality rates, obviously those are rare and you need a lot of numbers to get there. You can extrapolate that if you have double the risk of SSI due to using something other than the best antibiotic choice, then you’re gonna see an increase in severe morbidity and mortality on a large enough number– large enough scale.
Now, the current 2022 practice parameter guidelines says that patients with non-anaphylactic and unverified penicillin allergies should receive cephalosporins without testing. And that patients with a history of anaphylaxis can be given cephalosporins with a structurally dissimilar R1 side chain without doing any testing.
But the truth is people aren’t following even this current guideline.
[00:06:56] Antonia: Yeah, they’re definitely not and I think, well, this is a new paper. I think I would have to just print this out and carry this with me everywhere I go to actually get everyone else to say, sure, anaphylaxis to penicillin well let’s just go ahead and give Ancef now.
Right, so what cephalosporins are the ones that have the dissimilar R1 side chains?
[00:07:18] Howard: Well, that’s the most important question. I think a lot of us don’t know that. We were taught in medical school about the beta-lactam ring and not the side chains. But I’ll put a link to a paper that discusses this exact issue from 2019.
It turns out it’s the R1 side chain that’s responsible for the cross reactivity that we’ve all been taught to worry about between penicillin and cephalosporins. We used to think that because these antibiotics were made from the same mold, the penicillium species, that the penicillins and cephalosporins that have this common beta-lactam ring from this mold is why you’d have this cross reactivity.
But it turns out that’s not true. The key cephalosporin that does not share a similar side chain or one side chain is cefazolin.
[00:07:58] Antonia: Okay. That is Ancef. So, according to the current practice guidelines then, and this paper, we can give Ancef to someone even if they’ve had a true anaphylactic reaction to penicillin.
[00:08:13] Howard: That’s exactly right. And so that’s just very convenient for us. So that is the antibiotic that we mostly use. If you’re an orthopedic surgeon or an OBGYN, or many other specialties that use cefazolin as your primary antibiotic, that should change your practice immediately.
[00:08:28] Antonia: Absolutely. So it’s almost like, if we’re asking patients about their penicillin allergies, it’s not for the purpose of should we avoid Ancef or not. It’s more like, should we avoid penicillin for your GBS or should we avoid maybe some other cephalosporin like Rocephin for another reason.
But for prophylaxis for C-section, we can still use Ancef. We just need to know that they’re not specifically allergic to Ancef.
[00:08:55] Howard: That’s right. And other cephalosporins do have similar R1 side chains. So this isn’t true of all cephalosporins by any means, but this is true of cefazolin.
[00:09:03] Antonia: All right. Well I’m pretty sure so far, nobody’s following this guideline.
At least I, I don’t think a lot of people that I know. Well, that’s why we have this segment. Yeah, exactly. So even if they were to have a reaction when they’re given Ancef, because we know from this paper now that they can have Ancef, even if they have a penicillin allergy. Even if we gave them Ancef and they had a reaction in that case, it would more likely be a separate, unrelated reaction, to Ancef and not a cross reaction from penicillin. Is that right?
[00:09:36] Howard: Yeah, exactly. Mind blown. Yeah, so of course people will assume that if this happens, it could have been avoided because of the history of the penicillin allergy. And, oh, why did you use Ancef? You should have known better.
But that’s the point of this new paper that came out in May from the Yale Research Group. They actually expand our knowledge of this topic considerably. Paper says that even in patients with a verified documented anaphylactic penicillin allergy cephalosporins are safe to use.
And they didn’t limit that to Ancef. They were including all of them. So this was based upon a chart review over many years of patients who had had allergic events, and they found 50 patients who had truly verified penicillin allergies. And these patients received 77 unique cephalosporin antibiotic courses over that time period, and it just didn’t matter.
[00:10:18] Antonia: This will also affect our treatment of GBS positive patients with a penicillin allergy, as well as our other OB and GYN pre-op patients. Because for GBS positive, if we can’t use penicillin, then Ancef would be the next choice. And since Ancef is the drug of choice, then even when patients say they have had an anaphylactic reaction to penicillin, if we withhold Ancef from them, then we are violating the guidelines.
[00:10:47] Howard: Yeah. And we know this happens all the time out of conservatism about these antibiotics. I’ll put a link to a paper also from 2021 that shows in a population of women undergoing cesarean delivery, that those who reported a penicillin allergy of any type obviously, were much less likely to receive their prophylactic antibiotics within the recommended 60 minutes prior to incision.
And I think, like you said earlier, this is because of the difficulty in administering the gentamicin in particular. I think that people sort of realize at the last minute that, oh no, I need to switch the antibiotic up. The nurse calls ’em and says, Hey, she has a penicillin allergy. You’ve ordered Ancef.
They’re like, oh gosh, well give Clindamycin and Gentamicin, but they don’t order it soon enough. Maybe if they Google it, they only order the clindamycin, so they either don’t get the right antibiotic or they do order the gentamicin and the pharmacy has to dose it based on their weight and they have to get it down there.
It’s not normally there, so it just doesn’t get administered, and patients are getting gaps in coverage.
[00:11:39] Antonia: And all of that would be totally unnecessary if people would just give the right antibiotic in the first place.
Howard: That’s right. Yeah.
Antonia: Yeah, and just take the penicillin allergy for what it’s worth in this case, which is irrelevant.
[00:11:52] Howard: Definitely. So we’ll put links to those papers so you can show them to your nurses and your pharmacist.
[00:11:55] Antonia: All right. Well, let’s talk about antidepressants in pregnancy. This is not a new topic by any means, but I don’t think we’ve really talked about it on this podcast before. So we’re gonna do so now. It’s no secret that perinatal depression is a major and growing problem in the US.
As is depression and other mood disorders just in general, even in non-pregnant people. When women with preexisting mood disorders who are already on meds become pregnant, it seems that they are very commonly told to stop their medications for fear of causing problems with the pregnancy or fetus. Or that they don’t know, they haven’t gotten in with any doctor for a while and they just self-discontinue out of their own fears and not knowing. Then they’ve been off it for three months by the time they see the OB. But, this stopping of their mood medication often leads to destabilization of the mood disorder, and that can cause worsening function and sometimes even catastrophic depression or psychosis.
There is a lot of overlap between mental health disorders and maternal deaths, and in some cases, harm to the children as well. But this is a common grievance we have in obstetrics where non-obstetricians will just say without really any thought that their patients should stop their medications because they’re pregnant.
And this seems to be more common with psychotropic meds because people tend to treat mental health conditions differently than they do, say diabetes or hypertension. So we should talk about the safety of these drugs now and address this.
[00:13:36] Howard: Alright, let’s do it. I will say that in Tennessee, our maternal mortality review committee has deemed over the last few years that 27% of pregnancy related deaths were due to mental health conditions.
So it makes it one of the leading causes, one of the top three causes, of maternal death in the state of Tennessee at least. And the national numbers are not much different. Well, when we think about drug safety in pregnancy, I think we have to answer four main questions. First, does the medication cause a birth defect?
Secondly, does it increase the risk of maternal complications like hypertension or diabetes or preterm labor? And does it increase the risk of fetal complications during the pregnancy like growth restriction or oligohydramnios or something like that. Third, does the medication have some addiction issue or a neonatal abstinence syndrome associated with it so that the newborn may have some problems after birth due to withdrawal? And fourth, are there longer term complications associated with the offspring, the child, when they become an adult, or later in life, if they were exposed to one of these medications, like for example, an increased risk of cancer, or perhaps in this case a difference in the rates of their own mental health problems or autism or something like that.
[00:14:44] Antonia: Okay. Yeah, that is a good way of thinking about it. So let’s give some examples of medications that we know are problematic in terms of this rubric you’ve just outlined.
[00:14:55] Howard: This is good for medical students.
[00:14:56] Antonia: Yeah. So here’s some drugs that we know cause birth defects. And we can have a list. A lot of these, the specific birth defects that are caused by each drug, will tend to overlap.
So I won’t list all of them, but we’ll put a list on the website for your exam review or whatever, or if you’re curious. So Warfarin– the blood thinner, ACE inhibitors– which is a blood pressure med, alcohol, cocaine, lithium–which is a mood stabilizer, mercury causes birth defects– I don’t know exactly what mercury is used to treat?
[00:15:29] Howard: Nothing in modern medicine, but it used to be used to treat syphilis and other things.
[00:15:33] Antonia: Okay, well there you go. Don’t use it then for pregnant women. Carbamazepine and valproic acid, which are both seizure medications. Then vitamin A and retinoids, which you use for acne and certain chemotherapy drugs for cancer or even for rheumatoid arthritis, for example. Certain antibiotics like doxycycline or fluoroquinolones, and those last two ones specifically do cause less severe defects.
They’re more minor issues like tooth discoloration or some issues with the muscle tendons, and even those happen rarely with exposure to the medication, but still they’re strong enough of an association to avoid those antibiotics since there’s usually plenty of good alternative antibiotics that are less harmful. But all of these meds I listed can definitely cause serious and permanent damage to the baby, and they usually affect multiple major organ systems.
So like I said, we’ll post a list of specific effects for each med if someone’s studying for something.
[00:16:32] Howard: Yeah. And then there’s historic drugs like thalidomide, which can cause bilateral limb deficiencies and a lot of other medications that are maybe no longer FDA approved, things like that. So, then the second class of things was something that would increase the risk of neonatal or maternal complications.
And a lot of the drugs you mentioned, they do that as well, not just birth defects, but you know, fetal growth restrictions can be caused by phenytoin or cocaine or ACE inhibitors or angiotensin receptor blockers, carbamazepine, Coumadin, and lots of things, a lot of the chemotherapy drugs. So they may not cause a birth defect early in the first trimester, but even second and third trimester exposures can cause problems as well.
[00:17:08] Antonia: Yeah with the blood pressure meds, for example, there’s multiple theoretic mechanisms for this. There’s the relative placental hypoperfusion from the desired effect of lowering the mom’s blood pressure. Although this actually was disproven in a recent trial about treating hypertension in pregnancy to below 140/90. But this can still be seen, for example, if you use a blood pressure medicine in a non-hypertensive woman, like if she’s taking beta blockers for heart palpitations or for anxiety, then she might be at higher risk of having fetal growth restriction. And I think also ACE inhibitors can cause oligohydramnios later in the pregnancy due to fetal renal dysfunction.
[00:17:51] Howard: Yeah. And I’ll say all these aren’t like margin calls. These are all things that are fairly clearly easy to see in data, even if they’re rare. Yeah, yeah. Like we’re not like questioning these.
[00:18:00] Antonia: Yeah. And the third area of causing either an addiction or a neonatal abstinence withdrawal syndrome is probably gonna be a little bit more controversial. I’m sure we’ll get into that. But the easy ones here are opioids and benzodiazepines and really any medicine that is known to be associated with addiction in the mother.
I have also seen a case of neonatal abstinence syndrome that came from suspected high-dose gabapentin, which I hadn’t really seen described before, but that’s what the
[00:18:32] Howard: Yeah, we’re seeing that in adults too, that gabapentin is more addictive than we thought it was. Yeah. Okay. Well, the fourth area are these long-term things that cause problems well after birth.
So the classic example here of course is DES, which is associated with vaginal cancer and other malformations that occur well after birth when the offspring as an adult and then other, what we call endocrine disruptors, potentially like 17-hydroxyprogesterone, which we’ve talked about, which might apparently be associated with an increased risk of cancer in the offspring and even cannabis as we’ve discussed before.
[00:19:00] Antonia: Okay, so aside from using lithium as a mood stabilizer, where would the other antidepressant meds fall into these four categories.
[00:19:13] Howard: Yeah, we’ve gotta answer that. But maybe before we get into that, I would like to point out another way of thinking about this problem that I think is important.
We’ve made this mistake I’m gonna describe several times in the history of obstetrics. The mistake of blaming a drug as being the source of a problem, when actually the disease itself is the problem. So, for example, do women who take anti-epileptic medications, do they have an increased risk of birth defects because of the medicine or because of their epilepsy? Or maybe both contribute. And given the way that we historically have studied drugs in pregnancy and lactation, meaning, not with randomized controlled, placebo controlled trials, it’s very difficult to tease out the effects that are due to the disease itself rather than those that are due to the treatment for the disease.
Especially if there’s a possibility of mixed effect, which there certainly can be.
[00:19:58] Antonia: Yeah. So, historically, we’ve avoided any controlled safety trials of drugs in pregnancy and lactation, and most of what we know comes from really the lowest quality type of scientific evidence, which is retrospective databases.
And these can have all sorts of problems, including selection bias about who’s registered and who’s not. Say a doctor who’s had 15 women do well on one particular drug, may not choose to enroll those patients in the database, because these are usually like voluntary enrollments. But then when he sees a birth defect in patient number 16 and then he wonders, was this due to the medication, he might contact the manufacturer or some other agency and then enroll that patient into the registry ‘cause it sticks in his mind to do so. And we have a lot of theoretical understanding about the safety of drugs from measuring whether or not they cross the placenta or into breast milk or from animal studies, but the animal studies rarely correspond to human studies. When you finally can replicate that animal study in humans, in the drug trial process, they rarely correlate.
If a drug does not cross the placenta or breast milk, then we know it won’t directly affect the baby, but even if it does cross the placenta, that doesn’t mean it’s bad. And potentially if it doesn’t cross, that doesn’t necessarily mean that it’s completely innocent either.
[00:21:22] Howard: Yeah. So all sorts of problems here and all this makes it very difficult to draw what you might call a definitive conclusion that a lawyer might be happy with on the safety of a particular medication. And it’s difficult because also there is an entire industry of attorneys if we’re being honest, who look to sue drug manufacturers for a living, and doctors perhaps who prescribe the drugs whenever they find a birth defect or something like that.
And they are a source of a ton of misinformation on social media, on the internet. They also fund and pay for a lot of poor and shoddy research to help support some of the claims that you’ve probably heard about birth defects related to certain medications.
[00:22:00] Antonia: Okay. I get how drug companies can fund studies that might tend to favor their product, but how would attorneys fund studies that tend to favor their lawsuits?
[00:22:10] Howard: Well, they can hire certain epidemiologists or other researchers as expert witnesses in cases in these large, sometimes hundreds of millions or billion dollar class action claims, on the condition that if they can produce a scientific opinion based on a study they’ve done or something that validates their lawsuit, then they’ll get paid well.
So I’ve written about one such researcher who is a perinatal epidemiologist in Canada. I’ll put a link to that. But you could even think about Andrew Wakefield. He published the paper about autism and the MMR vaccine many years ago that’s since been withdrawn, but which is the source of a lot of skepticism about childhood vaccines.
And at the same time that he was writing, that he was being paid by a law firm who was planning to sue the manufacturer of the MMR vaccine. So there was a major conflict of interest and this happens all the time. Another article I’ll put a link to discusses whether or not antidepressants are associated with autism and I’ll just summarize it and say that the data that’s claimed such an association is false.
And that article will explain in detail where the data comes from and how the Canadian, I was mentioning, the promoter of the idea, was paid by a law firm.
[00:23:17] Antonia: Alright, well. Let’s answer the first question. Do antidepressants cause birth defects? And specifically, I think we can focus on just SSRIs and SNRIs and maybe Wellbutrin, which is a dopamine norepinephrine reuptake inhibitor.
The birth injury help center website, which is a law firm, claims there are studies showing that SSRIs cause spina bifida, heart defects and other birth defects as well.
[00:23:44] Howard: Well, luckily lawyers aren’t scientists or perhaps unluckily for us. They’ll cling to isolated case reports or registries or databases or anything that shows some apparent asymmetry in the rates of problems among women who are registered as having taken a medicine versus those who did not.
And as an example of this, let me talk about Paxil and heart defects. That’s one of the SSRIs. In 2009, a plaintiff was awarded $2.5 million in Philadelphia after his mother used Paxil for anxiety while she was pregnant. This led to some 800 such lawsuits about a variety of birth defects and somewhere around a billion dollars in settlements.
Now, Paxil today is listed as a Category D drug under the old system that everybody learns of course. And everyone has learned in OB/GYN to never give Paxil during pregnancy to any pregnant woman or anybody who could become pregnant. And GlaxoSmithKline the company that had to pay up that billion dollars, they definitely want you to stop prescribing Paxil to anybody certainly after 2009. The problem, of course, is that it almost certainly does not cause cardiac birth defects, even before that case was settled in 2008. There was a meta-analysis of nine studies available at the time about paroxetine and cardiac malformations that found that first trimester exposure does not appear to be associated with an increased risk of cardiac malformation.
[00:25:03] Antonia: So if there’s such good data showing it doesn’t cause a problem. Then how did that plaintiff win the lawsuit, and why did that company settle all those other lawsuits?
[00:25:11] Howard: So yeah, the big problem was not whether Paxil actually causes cardiac birth defects. The problem is the FDA had warned GlaxoSmithKline that there might be an increased risk of cardiac defects in 2005, and GlaxoSmithKline didn’t really do anything about that when they received that warning.
They didn’t warn doctors not to prescribe it to pregnant women or bring up any precautions based upon the FDA warning they’d received. And so that was the big smoking gun that made them look guilty as sin. What hurt them in court was not whether or not it actually caused the birth defect. But, how the drug maker acted in response to a potential warning that it might cause a birth defect.
[00:25:49] Antonia: So what happened in 2005 with that FDA warning?
[00:25:54] Howard: Well, it was actually GlaxoSmith Klein who had first seen a possible signal in their own internal data, indicating an increased risk of cardiovascular ventricular septal defects based upon some unpublished retrospective data that they had been acquiring in post-marketing surveillance.
Then there was a study presented in 2005 by Diav Citron and colleagues who were not associated with the company, which was held out to be a prospective comparative study that showed a slight increase in cardiovascular defects with paroxetine. So there was a 1.9% risk compared to a baseline 1% risk.
This led to health officials in Canada and the US issuing warnings, but the company really didn’t actively, as I said, they didn’t tell people not to use it in pregnancy after they saw that.
[00:26:35] Antonia: So they sort of dug themselves a hole because they had internal data, whether it was accurate or not, that Paxil might cause these heart defects.
And then similar external data that was public. And then ultimately they were punished by juries for not acting more aggressively on this information. But 1% to 1.9% seems like it could be a big enough jump that most pregnant women would want to avoid it if they were told this could increase the risk by 0.9% from baseline.
[00:27:07] Howard: Definitely, almost double. Yeah. But that being said, it’s very unlikely that this association is true. So previous studies had found no difference between Paxil and placebo for cardiac defects, and there had been three peer reviewed studies prior to that that showed no association, and the subsequent published studies really didn’t support it.
The one major exception to this was the systematic review published in 2016 by Anit Barard. This is the Canadian researcher I mentioned a few seconds ago, who claimed that antidepressants cause autism and dozens of other claims. This person’s incredibly biased and is known to work for attorneys who sue drug makers, and offer expert testimony.
So she published a meta-analysis that claimed definitively paroxetine used during the first trimester is associated with an increased risk of congenital malformations and cardiac malformations. But actually, even this study did not show statistical significance. This author likes to talk about trends in data as being significant, and so she’ll hold up non-significant data, as significant, and then say it’s her opinion.
In fact, in this article I wrote about it, which I’ll put a link to, there is this wonderful analysis written by a judge who threw out her testimony in a lawsuit against another antidepressant, I believe it was Zoloft, and didn’t allow her to be an expert witness because he felt that the claim that non statistically significant trends in data indicated truthful findings was not legitimate data science and could find no other statistician who agreed with the use of these methodologies. But despite all that, for nearly 20 years she’s published these sorts of studies and they cause a lot of harm. This is the definition of bias and conflict of interest.
And in this case, as in most cases of sort of bad medical literature, the bias is driven by personal financial gain for the authors. Maybe just a passion project, maybe a crusade, who knows? But this is not scientific research.
[00:28:52] Antonia: Okay well that still puts us in a tough spot. The data doesn’t truly support a risk.
The theory doesn’t support it either, but there’s already been a precedent of compensation for a suspected, but not proven, harmful association. And if we prescribe that medication despite that precedent, because we’re just going by the data, then we still have no leg to stand on. If our patient does get unlucky enough to have a baby with a cardiac defect or whatever the problem is that they think could be attributable. So we basically have to document like crazy that we discussed all the possible risks, even risks and quotation marks, and offered every other alternative, and still concluded with the patient as the final decision maker that the benefits outweigh the risks or they just, they wanna keep using that medication specifically.
So I think the bigger harm here is that if patients hear about any of these problems with Paxil and then we say, “okay well try this other drug, Zoloft, it’s in the same drug class, but it won’t hurt your baby”. They may be understandably suspicious and not wanna do anything and would rather just stop the medication completely.
It would almost require showing patients just how different the chemical structures are of all the different SSRIs, to at least give them some kind of visual of how they are not the same drug, even though they act on the same receptor.
[00:30:17] Howard: Well, and even if they don’t think it, all of the websites, when you Google safety or you Google this question, they aggregate the SSRIs together. So they will tell you that all antidepressants or all SSRIs are associated with cardiac defects based upon Paxil.
[00:30:32] Antonia: Mm-hmm. And as an aside, the only reason I’ve never had to have that conversation with a patient about Paxil in pregnancy is that I’ve never had a patient who was on Paxil at any time in her life before pregnancy.
And here’s a little anecdotal side note that some of the psychiatrists I know seem to think that it’s one of the less effective SSRIs for treating depression. Like if someone’s already on it, that’s one thing, but if they’re just starting one, out of all the options that at least these particular psychiatrists I talked with would never start with Paxil.
So maybe those prior legal settlements have played into this now for them to be saying that now, and it’s just prescribed less at baseline ’cause of its marred history. But most patients I meet with depression, who are on an antidepressant, are on something that is appropriate for pregnancy and does not have any big sensational lawsuits on its reputation like Paxil does.
But unfortunately, just like with vaccines and autism, antidepressants and birth defects do seem to have this whole industry of low quality research. So where is the good research on this?
[00:31:38] Howard: Well, yeah, that’s the problem. I mean, you already described the difficulties here.
At least there’s no good data that comes about in the way we study other drugs in other patient populations. When you see this sort of low quality data and then you see mixed data where some studies say that Paxil might cause a defect and many more say that it doesn’t, and then those studies are, themselves derived from low quality reporting registries with all sorts of biases and other poorly controlled and poorly sourced data sets. Then you’re going to decide to agree with the studies which show an effect, or to agree with those which do not show an effect probably based upon what you already believe.
That’s where the bias comes in. But outside of a prospective randomized trial, it’s hard to make any definitive claims about teratogenicity. The problem is that women who take antidepressants in pregnancy, and therefore get included in a registry, are not the same as women who do not take antidepressants in pregnancy and therefore are used as a control group.
To really study this, you need to randomize a large number of pregnant women, thousands, with depression or anxiety, to the drug versus placebo.
[00:32:47] Antonia: Right. Because you need to control for the underlying condition., before you can isolate what the medication alone is doing or not doing. And women with a psychiatric diagnosis that would warrant antidepressant treatment may have certain genetic predilections or maybe higher rates of substance use or certain other tendencies like abnormal sleep or abnormal food intake or weight fluctuations or any number of other things that all could affect the pregnancy separate from medication.
And maybe when their depression flares up, they forget to take their prenatals or they don’t eat right, or they have to use more sleep aids or maybe more caffeine during the day. And recall when we talked about caffeine exposure and miscarriage rates actually being a sign of a genetic polymorphism, like a common denominator that leads to both the higher rate of miscarriage and to the higher rate of caffeine consumption.
There’s no doubt that similar genetic polymorphisms would exist for other things like depression, for example. And only through randomization can you balance for those unknown variables.
[00:33:57] Howard: Yeah, that’s a good point. And I think that there’s a bias that people have about, they understand that when you talk about diabetes, they maybe understand it when you talk about seizure disorders or blood pressure.
But why would just having depression cause birth defects. And so that caffeine thing is a good example, that there may be underlying genetic causes of both that we are unaware of, and that’s why you have to do the trials in the right way to answer those questions. The caffeine thing is brand new, like for decades we didn’t understand it and now we finally do, that caffeine is actually not the problem. But there’s also a lack of confirmation of antidepressant exposure in clinical studies just because somebody says they took it, doesn’t mean they did. The data analysis always assumes 100% compliance with the prescribed antidepressant, but we know that even in clinical trials, about 40% of women don’t take the antidepressants as prescribed, and this rate’s even higher in exposure databases.
Plus we have no consistent way of knowing whether or not self-reporting to other illicit exposures is accurate. Pregnant women in particular, are, and have historically been socially pressured to not admit to tobacco, caffeine, illicit medication, self-medicating with a friend’s medicines, things like that.
But we know that people with untreated mental health problems tend to have a very high rate of self-medication with other substances, legal and illegal. So the combination of a high rate of use and pressure to not report invalidates a lot of this data in a very significant way. We also know that women tend to have incredibly poor recall about exposures they had during a pregnancy, even without significant bias by the time they’re six months or more postpartum.
And that’s often the minimum time after pregnancy when the defect has been discovered or something like that, that they’re asked to self-report in these studies, what they might have been exposed to while they were pregnant. And this has been objectively shown.
[00:35:43] Antonia: So these database exposure studies have significant and perhaps deal breaking limitations.
So particularly with heterogeneous findings and contradictory information, we probably should take any findings from these kinds of sources as preliminary and unconfirmed.
[00:36:02] Howard: Exactly. Although I will say that there isn’t heterogeneous findings in these databases for things like methotrexate and lithium.
So, real effects do show up. The problem is with studies like this where you have people taking trends in data as fact, when five other studies say there’s nothing and that’s our problem here. But, if you don’t do the appropriate studies and have the appropriate controls, then you’ll soon believe that all diseases are caused by antidepressants or all you know, or any other medication for that matter.
So what we’re saying of antidepressant medications, by the way, is also true of all the other medications that you would study. We have the same problems in how we look at ’em for all these types of medications. So you really have to look for data that is consistently showing an effect.
[00:36:46] Antonia: Okay, well, we’ve described the problems in our knowledge quite a bit, but let’s try to get back into answering the original question
[00:36:56] Howard: Well, it’s hard to talk about the question without understanding our limited knowledge of it, but I’ll put a link to a really good review about on the use of antidepressant medications in pregnancy that highlights the issues we’ve been discussing, but also describes the data in various studies. Up to the point that this paper was published.
So this review does a good job of showing where data are in agreement and explaining where other data is confounded. And you have to look at the literature as a whole and not cherry pick individual reports or even individual studies. So I’ll quote from them in their summary and they say, quote, the extent literature indicates that congenital malformations are not associated with prenatal exposure to tricyclic antidepressants or SSRIs alone, or atypical antidepressants.
They also highlight studies showing that birth effects appear to be more common among pregnant women with depression, including among women who stopped taking the medication well before they got pregnant. So the idea is very important if you have severe enough depression or another mental health issue that you need to take a medication, then you likely already are at an increased risk of having a child with birth defects, whether you take the medicine or not, and maybe that’s an underlying genetic issue, or maybe that’s an underlying environmental exposure problem with other things.
They also do a good job of going through the various studies about Paxil and overall they admit that the link to heart defects is possible, but unproven and that tons of studies show no such linkage. I’ll put a link to another 2022 review that includes the more recent literature published since that paper was done, and the authors of that paper conclude definitively that perinatal SSRI exposure does not increase risk of major malformations after accounting for underlying maternal illness, which is the important part.
[00:38:34] Antonia: It’s a shame that if underlying depression increases the risk of defects, that treating that depression with medication doesn’t seem to reduce that risk, but at least it doesn’t seem to elevate the risk either. But why don’t we move on from fetal harms to pregnancy harms. What do you see as the biggest claims about antidepressants on this question?
[00:38:59] Howard: Again, they reported lots of mixed data and it generally seems to focus on the outcomes of lower birth weight or earlier gestational age at delivery. I’ll point out the same trend here as for the teratogenicity question, that very few studies looking at these outcomes control for the mental health status of the mother, and there’s also an inconsistent dose effect associated with any pregnancy outcome and antidepressant use.
Overall though, it’s the same story with highly conflicting data from various studies, which means that you probably shouldn’t trust it. The mother’s underlying mental health disorder can absolutely affect her baby’s birth weight, and also how early she delivers, just like it can affect early fetal development. And I’ll also give a spoiler on one of the next questions. Mother’s mental health can have drastic effects on early infant health, safety, bonding, and brain development. Back to your question, remember also that just because a particular study might show that the newborns of mothers who took antidepressants were say, 50 or 100 grams smaller than the newborns of babies whose mothers did not take antidepressants, doesn’t mean that that’s clinically significant.
Smoking during pregnancy has a similar or greater effect that’s been consistently shown. While these antidepressant effects have not been consistently shown and may be related to other confounders, like smoking.
[00:40:11] Antonia: Okay. Well then let’s just move on to the next question about effects on the newborn. The first and scariest one on the internet is persistent pulmonary hypertension of the newborn, or even in the absence of that, just respiratory distress at birth ’cause that often still requires NICU or some kind of breathing assistance. And then the other issue is neonatal withdrawal syndrome, and we do know that serotonin has a potentially inhibitory effect on respiration. So knowing that, researchers have looked for associations and found a potential link between respiratory distress and newborns and their mothers taking SSRIs.
And additionally, serotonin has a role in vasoconstriction, which is part of the pathology of serotonin syndrome. And so it’s been hypothesized that this also could increase the risk of pulmonary arterial hypertension in the newborn. In 2006, there was a publication in the New England Journal indicating that women who took SSRIs had an odds ratio of six for having a newborn with persistent pulmonary hypertension.
A 2010 study showed that the odds ratio is closer to two for both SSRIs and SNRIs. But that earlier publication from 2006 still led to the FDA putting a black box warning on antidepressants regarding the persistent pulmonary hypertension and that’s led to probably most of the concern about their use.
[00:41:46] Howard: Yeah, definitely. And I think this is a true and physiologically feasible association. Women who take these medications, or at least women with depression or anxiety likely are at perhaps double the risk of having a child with persistent pulmonary hypertension. But again, the same issue has occurred where the mental health and other exposures of the mother were not controlled for.
We’ve talked our way into this basic science understanding that you just described, and then found data that matched our basic science understanding rather than try to disprove the hypothesis through a trial. One might also speculate that pulmonary hypertension is related to a thickening of the muscularis layer of the pulmonary vessels due to chronically elevated stress hormones in the mother, which is more common in women who suffer from depression or anxiety.
So I can make up these pathophysiological narratives all day long. The question is how to appropriately study it. Fortunately, this only affects about one in a thousand deliveries, but how much of this is related to the medication versus how much of this is related to the disease is still an unanswered question.
[00:42:48] Antonia: It seems that the absolute risk is very small for pulmonary hypertension or respiratory distress that would require NICU admission. One in a thousand is what we tend to quote for blood clots like deep vein thrombosis on birth control pills, or the risk of an IUD perforating the uterus at the time of placement.
This risk with antidepressants has to be weighed against the potential risk of untreated depression or anxiety in the mother, which itself may be also associated with persistent pulmonary hypertension or even more serious outcomes including self-harm or harming the infant.
[00:43:24] Howard: Yes. The NICU data is also mixed, and some studies have shown that the increased rate of admission goes away when controlling for maternal mental health status.
This would argue that the underlying depression, rather than its treatment, is influencing the outcomes. It’s also quite possible that pediatricians and neonatologists are biased by seeing the medication on the chart and therefore have a lower tendency to admit patients for observations because of this. There’s reasonable evidence for that.
Part of the reason is it’s quite popular for folks to write case reports anytime a baby has some issue, if the mother happened to have been taking an antidepressant medication. So there are case reports out there for things like jaundice and jitteriness, infantile hypertrophic pyloric stenosis, necrotizing enterocolitis, tachyarrhythmias, hypothermia, long QTC intervals, and several things like that.
And this has led to sort of an anticipation for when you see an antidepressant being used that this antidepressant neonatal syndrome will be there. And so it creates sort of a bias among pediatricians and neonatologists who have been trained to look for these things that may not even be real.
[00:44:25] Antonia: But outside of those case reports, there’s no evidence that antidepressants are associated with any of those things you just listed.
Antonia: Okay. Well, how about the withdrawal syndrome? That seems to be part of what you just described, and I’ve even heard of newborns being started on clonidine, for example, for neonatal abstinence syndrome, where the only known medication was an SSRI.
[00:44:50] Howard: I’ve seen ’em started on opiates for it.
Howard: So, yeah, I think that just as any adult might be irritable or feel some effects of discontinuing their SSRIs, in the same way this may happen with newborns. So it’s estimated that about 10% of newborns will have some mild jitteriness, but that stopping the SSRI earlier to prevent this isn’t clinically valuable.
So if someone is starting a medication for the newborn, particularly something like clonidine or as I said, even morphine, because they believe there is an abstinence syndrome related to SSRI use, then this is almost certainly unnecessary and inappropriate. Or there’s another medication the patient’s withdrawing from that the doctor just doesn’t know about.
[00:45:30] Antonia: Okay, and you already mentioned the possible concerns with a mother with unmanaged depression, trying to take care of herself while also taking care of her child or children. The early newborn period is tough on anyone, even when there are no problems and it goes as well as possible, it’s still tough. And it’s also such a critical time in terms of maternal postpartum complications and how fragile and helpless the baby is too. But it doesn’t stop there after those first few weeks. And one of really the most terrifying consequences of mental illness is when the mother harms herself or her kids. Those cases usually seem to make national news and when they happen it’s not even necessarily in the newborn period.
So, well, these stick out to me. I don’t know if it’s just ’cause I don’t know. These tend to stick out to me whenever I see a headline like this. So I remember a story of a woman in New York who was known to be mentally struggling who drowned her three children who were, the three of them were between three months old and seven years old.
And then another very horrifying case in Massachusetts where the kids were between eight months old and five years old. And that one was particularly tough because the mother was already on multiple antidepressants and sedatives, but clearly it wasn’t working. And there’s plenty of other stories like this, and I recommend that you just don’t Google it if you don’t wanna ruin your whole day.
But all of this is to say, I think that optimizing the mother’s wellbeing and functioning for after pregnancy, should be enough reason by itself, for a mother with depression to be on antidepressants during pregnancy. Because if you wait until after pregnancy to start these, you’re already well behind the curve. It takes several weeks just to get the right drug level to see if there even is an effect.
Is it a good enough effect? And then you still may need to adjust the dose or switch to a different drug. And it’s harder to do things like therapy sessions when you’re sleep deprived and you’re feeding a newborn every two hours. So it’s best to already be on a good level, find the right medication and be on it before.
[00:47:43] Howard: Well, and let’s, yeah, let’s underscore the problem. As I said at the beginning, in Tennessee data at least, 27% of pregnancy related deaths are related to mental health problems. And another 23% of pregnancy associated deaths, which is a bigger number, are related to mental health problems. And then you have the substance abuse numbers.
So we’re talking about hundreds and hundreds of women across the United States every year who are dying in pregnancy or in the first year after birth due to poorly treated mental health conditions. And we’re weighing that against theoretic and unproven risks of these medications.
[00:48:16] Antonia: Yeah, exactly. So let’s get to our last question about the long-term neurobehavioral issues like autism or different rates of depression or mental health problems for the child.
I don’t think there’s any claims about cancer from antidepressants yet.
[00:48:31] Howard: No. But autism is definitely a big one. So there’s been a lot of recent literature about this, some of which we discussed on this podcast previously. I will say a new 2022 summary says that there’s no increased risk of neurodevelopmental disorders, and we discussed a recent paper showing no difference in the rates of autism after antidepressant exposure.
The problem with things like depression or mood disorders in the offspring is that of course, children who have poorly adjusted mothers are likely to be in an environment and share genetics and environmental exposures and stressors, and then learn emotional cues from the mother and both parents that will lead them to have similar
mood or emotional maladjustments. And that will be much more prevalent in women, by the way, who are undertreated. But to some extent, even adequately treated women will still have flare ups and bad days and certain genes that affect their interactions with their children, perhaps in a greater frequency or in different ways than in mothers who don’t have depression.
So it’s incredibly complex to control for these sorts of effects in clinical studies. We do know that newborns develop a lot of emotional responses, particularly as it relates to anxiety by looking at their mothers in the first year of life. If the mother is maladjusted and crying for various reasons or having anxiety, panic attacks, or feeling emotionally numb and unable to engage with the baby, then the child will pick up on those behaviors and internalize them.
And in this sense, it’s likely highly beneficial to the newborn’s long-term developmental outcomes, for the mother to be on a medication, which helps to improve and stabilize her mood. And as you said, that treatment needs to start many times before delivery for it to be effective at the time of birth.
[00:50:08] Antonia: Okay. I think that answers our four questions about antidepressants. The theme among all the answers is poorly controlled retrospective literature with tons of confounders, including the mother’s underlying mental health status, substance abuse, genetic background, and social status, and probably other things too.
And in the studies that do adjust for these factors, many of the observed effects of antidepressants seem to go away.
[00:50:34] Howard: Yeah, I think it’s fair to tell women that there are some uncertainties. But the magnitude of any potential negative effect of these medicines to them and their babies is almost certainly significantly outweighed by the magnitude of benefit that a mother who needs these medications would derive from taking them. Having poor mental health during pregnancy and afterwards is a potentially fatal condition, and the endogenous stress hormones associated with poor mental health are likely associated with lots of the short and long-term adverse outcomes that we’ve been discussing.
[00:51:04] Antonia: Yep. That list is way too long to get into, but there is something like a 10 year reduction in someone’s lifespan when they have depression. There’s all kinds of associations with heart disease, diabetes, dementia, chronic pain among other things. Although that’s not to say that there are any clear cut one-way associations.
It’s probably bidirectional for most of those, but I think it’s safe to say that depression affects everything about a person’s health and their life, so treating it should always be a priority.
[00:51:36] Howard: Well, the number one causes of preventable death in women that we treat of reproductive age are overdoses and substance abuse, often related to poorly treated mental health problems or self-medication, and then suicides and homicides in situations where there’s intimate partner violence and things like that.
So we think a lot about cancer screenings and those sorts of things, but actually mental health is the most significant thing we should be doing for patients of reproductive age, and then that affects these other outcomes too. If you have good mental health, you probably eat well, sleep well, exercise, you probably don’t smoke or abuse alcohol and other substances and all these other things that are huge contributors to deaths and morbidity from all sorts of other diseases that we think about.
So mental health is a key focus that OBGYNs should have for their patients of reproductive age.
[00:52:24] Antonia: Yeah. So obviously there does need to be better quality data, but these issues are so important, we can’t wait decades and in the meantime, until we get the data we hoped for, just tell people just to be safe, don’t use it.
[00:52:39] Howard: Right, and not take mental health problems seriously. We wouldn’t tell a diabetic to stop her insulin or someone with severe hypertension to stop their blood pressure medicine. But people are way too nonchalant about telling somebody with sometimes profound mental health problems to stop the medications that are life-saving for them.
[00:52:56] Antonia: Yep. So did you write a new Howardism post about tennis.
[00:53:02] Howard: I know that’s unusual these days. I spend my time on this podcast, but I did write a new Howardism. I’ll put a link to anybody who’s interested in tennis, but it’s about medicine, so,
[00:53:12] Antonia: Well I read it. It was actually very good.
Howard: So do you play tennis?
Antonia: I do not. Well, I’ve tried to and I was so bad at it that
[00:53:19] Howard: I’m not gonna claim to play it either. But I do try.
[00:53:25] Antonia: Okay, well let’s wrap it up for the day. The thinking about OBGYN website will have links to things, papers, and things we talked about today, and that Howardism, and we’ll be back again in a couple weeks.