Ask A Scientist: Ear Surgery (Tympanoplasty and Ossiculoplasty) Risk Vs. Benefit

A friend asks: "I have some progressive hearing loss, likely due to a hole in the upper part of one my tympanic membrane, along with scar tissue.  The nerve appears to be healthy and there is a conductive problem.  The hearing loss is getting worse (nobody knows why) and is worse in some frequencies than others.  I have a consult, but before I speak to the ENT surgeon, can you tell me about  the success rates, risks (especially more hearing loss), complications and so on for tympanoplasty and ossiculoplasty?"*

What?

The middle ear is the part we're concerned with.  The tympanic membrane ('eardrum') conducts sound to small bones (the malleus, incus, and stapes, or the 'ossicles') which then do complicated things with nerves.  Nobody has asked me about sensorineural hearing loss and anyways, it's generally irreversible.**

There are a number of more-or-less standardized tests used to measure and characterize hearing loss.  You can read about them here. (For a good time, read a bunch of papers wherein ENTs argue about the relative subjectivity of the audiogram, a boring-looking line graph.)   In this case, since the hearing loss has a mechanical (conductive) cause, the audiogram is in fact the test in question- the faintest decibel level a person can hear at a certain frequency.

Surgery in general

I have read a lot of very convincing papers about surgical success (see also a book called "The Checklist Manifesto"); they suggest that the two biggest predictors of a patient's outcome, aside from the particulars of the patient's condition, are how many of that surgery the medical center performs per year, and how many of that procedure the surgeon has personally performed.  (The use of checklists is, naturally, strongly recommended.)

Surgical studies in general frequently suffer from low N - that is, they don't have that many patients, so they lump everyone who got a certain surgery into one big group regardless of why they got the surgery specifically.  This makes it challenging to interpret the results.  This is also probably part of why surgeons, even more than other doctors, often rely on personal experience more than randomized controlled trials.

How To Fix Mechanical Hearing Loss?

You rebuild the part that had problems.

Tympanoplasty

The tympanic membrane, it turns out, can be reconstructed.  There are two main methods for major damage: using a piece of fascia (fibrous muscle covering) generally taken from behind the ear.

Here is a somewhat disturbing video of an endoscopic fascia tympanoplasty  (seriously, don't watch it; you don't want to see what goes on inside people).  Basically they take a little piece of fibrous gook, rough up the edges of the tympanic membrane, fill the ossicular space with gel gook, stuff the fibrous gook down into the ear, pack it in with more gel gook, and then it magically grows together and the gel slowly gets hydrolyzed or resorbed or vanishes in some way.

The other way is to take a little piece of cartilage and do more or less the same thing. Many studies in reputable journals suggest that cartilage tympanoplasty has a higher graft rate and a lower rate of surgical failure requiring a repeat of the surgery.

My understanding is that, because the entire ear canal ends up packed, one will be functionally deaf in that ear for quite some time, until everything heals/ dissolves.

* Complications in Tympanoplasty

Like any surgery, things can go wrong. There can be nerve damage if the surgeon isn't very, very careful.  They can damage the little bones inside the ear if they aren't very, very careful.  Extremely rarely, people react to stuff used in surgery, like surgical ointments.  How common are these?  It's hard to say.  But, without a doubt, the most common bad outcome is that it simply doesn't work.  Depending on the method and the surgeon, the tympanoplasty failure rate varies from 5% to 40%.  It's also nearly impossible to know which group one will fit into.  Obviously, methods that, in general, work more of the time are to be preferred, and surgeons with a high success rate are also to be preferred.

Ossiculoplasty

This can be loosely summarized as 'repairing or replacing some of the tiny bones in the inner ear - malleus, incus, stapes.'  There are many variations.  Also, a large percentage of trials are sponsored by the manufacturer, and are therefore probably crap immediately suspect.

Essentially, you can fix the bones with other bone from somewhere (the patient, or a cadaver, one guesses); you can replace some of the bones with a polymer, with titanium, or with a hydroxyapatite/titanium substance; you can replace all of the bones; you can use a  "titanium malleus replacement prosthesis (MRP) ... inserted into the external auditory canal".  You can do something with a footplate which is completely beyond me (MRP-to-footplate; TM-to-footplate; other-stuff-to-footplate).  You can do vein grafts (what these have to do with anything is also beyond me). 

The main problem is, it's apparently quite hard to figure out what's going on with itty bitty tiny bones inside the ear without opening it up.  So I ended up saying to my friend that I have no idea how to predict which procedure or procedures might be used, and there's an awful lot of them, and therefore I can supply very little information.

*I paraphrase, but this was the general idea.
** Irrelevant digression: my grad thesis project was on a set of proteins involved in hereditary progressive sensorineural hearing loss.  I've read about a lot of hard-of-hearing mice.

Ask A Scientist: Lab Politics Are Eating My PhD Project

Unearthed from the vault.  Apparently I wrote, but never published, this one!

***** 

Reader T. asks:

'My jerk of a boss wants to give away my project, which I spent many years on and which I've just gotten to work! To another lab, with whom I thought we were collaborating! And I'm worried the boss will kick me out. What the hell. What can I do???'

Dear T,

Your boss is a shortsighted, egomaniacal idiot. Does he, perhaps, lack tenure?

Ask A Scientist: Hair Loss (Alopecia Areata)

By request, I present a brief review of alopecia areata, also known as AAAAAGH!!!! My hair is falling out for no reason!!.

First, let me present you with a confusing medical fact: there is a poor distinction between a 'syndrome' and a 'disease'. Generally speaking, a syndrome is a collection of symptoms. Chronic fatigue syndrome is an example: it is a condition defined largely by how people feel, rather than a specific etiology (cause). A disease is (or should be) something with a well-known cause. Ebola virus is a disease. Wilson Disease, whose genetic cause is known, is a disease. AIDS is called 'immune deficiency syndrome,' but in fact the etiology, the HIV virus, is well known; it would, in my opinion, better be called a disease, but historical inertia prevents this.

At the same time, a lot of things with vague diagnoses are also 'diseases.' Graves' disease, for example, is caused by an autoimmune response to the thyroid. But what causes the immune problem to start with? Who knows.

Alopecia is something like this. It is defined as* hair loss with T-cell infiltration (a type of white blood cell).

Now let me give you my completely Mickey-Mouse-level understanding of autoimmunity:

T cells react to antigens- short bits of protein- but only when these bits of protein have been chewed up and 'presented' on the surface of other cells. So for T cells to be in the scalp, they have to be reacting to a secondary stimulus: another cell has to have, wrongly, recognized a hair protein as foreign. (There is a complicated process of selection which is supposed to kill off all your immune cells that would recognize and attack bits of your body. Obviously, every so often something goes wrong.)

B cells (another kind of white blood cell) start their life with an antibody stuck to their surface. (They are also selected to not attack one's own body.) When the antibody meets something it recognizes, it gloms onto it. Then it eats it along with its antibody. Then it chews it up and goes and shows it to a T cell. Lots of complicated things then happen, including that you get more T and B cells that specifically recognize that antigen. Some of the B cells become a new and exciting kind of B cell that secretes, or puts out, antibodies of various kinds.

Now back to alopecia.

According to several papers, up to 90% of alopecia patients have antibodies to hair follicles or hair proteins in their blood. This means that a B cell has to have been stimulated to produce antibody.

I don't know where the original antigen came from. When cells die in a 'pre-programmed' way (this is called apopotosis; it's natural when cells come to the end of their lifetimes or meet various kinds of trauma or... etc.) they blob off into little bits which are supposed to be eaten up by the body's internal vacuum system. Maybe some of them escape. Who knows.

The main problem in autoimmune diseases is, there shouldn't be anything there to react to bits-o'-body. But there is. As our understanding of what to do to the immune system to make it better is largely at a throwing-rocks level, the treatments for autoimmune diseases are, almost universally, immune suppression.

A relatively recent article in NEJM* suggests that injected steroids can benefit many patients. Small volumes are injected all over the affected area; in a few weeks (possibly due to hair life cycle?) hair may start to grow. Side effects, as with all steroids, include skin effects; usually the dose is too small to get many systemic effects. This treatment can be done with or without minoxidil. Side effects commonly include unwanted facial hair growth, either because it gets rubbed off onto a pillow, or because the face is, after all, near the scalp.

Several studies, which I am too lazy to link, suggest that topical steroids have little effect, but people do it anyways because it's cheap and relatively low-risk.

A meta-analysis* has also been done on all known double-blind studies. They conclude that no treatment is particularly effective and one might as well do nothing, or something equally depressing. However, meta-analyses are always tricky to interpret. They are comparing dozens of unrelated studies by criteria that may not have even been measured in the original study. In this one, they are defining clinical success as 50% hair regrowth. The authors note that patient satisfaction might have been a better, if less scientific, measure.

Interestingly, persons with other immune conditions- eczema, allergies, thyroid problems- are more likely to experience alopecia areata at some point. Some people theorize that this connection is indicative of a common genetic cause: something is slightly wonky in the MHC alleles (if you don't know what these are, trust me, you don't want to) and it manifests itself as various small problems. This makes sense, but is hard to test because who wants to fund hair loss? That's right. Nobody.

*Subscription on most articles; contact me if you want a copy.

Ask A Scientist: Genetic Testing For Rare Diseases

A (Jewish, Ashkenazi) friend recently asked, 'Should I have genetic testing for the Ashkenazi-Jew genetic syndromes?'

My advice to people, from a strictly practical level, is to consider the cost vs. the likelihood. If desired, one's personal making-me-crazy level may be added in.

Things like Tay-Sachs, while scary, tragic, and fatal, were actually quite rare even before genetic testing: 100 new cases per year in the entire country. The carrier rate is 1/27 for people of exclusively Ashkenazi (Eastern European) Jewish heritage; if two such people reproduce, there is a 1/1000 chance their child would have Tay-Sachs. (Laurie points out in the comments that two known carriers have a 25% chance, of course.)

The other recessive "Jewish" diseases occur even less frequently. The carrier rate for Canavan, for example, is somewhere near 1/100, so two Ashkenazi Jews have about a 1/10,000 risk for a child with Canavan.

In my particular case: my father is not Ashkenazi. Therefore my basal rate for Tay-Sachs carrier-ness is half: 1/54. The carrier rate in the general population is 1/200. My husband is also not Ashkenazi. Between the two of us, it's about 1/11,000. For contrast, in the general population, it would be 1/40,000.

In 1970, there were about 3.8 million births (and 100 cases) That's 1 Tay-Sachs case for 50,000 births. And? Nobody tests the general population for Tay-Sachs.

If 1/10,000 is an unacceptable risk to you- the testing is a good idea. In addition, many of the "Jewish" diseases are fatal, and most parents choose to terminate.

If one's risks of genetic diseases are elevated- family history, prior history in pregnancies, etc.- it might be a better idea to test. As there are a lot of things that can go wrong with genetics, however, there are many diseases for which there exists no test.

But bear in mind that all risk is relative. There are 21 yearly vehicle-accident fatalities per 100,000 drivers in this country. That's a 1/4700 yearly chance of dying in a car wreck.

Ask a Scientist: What's the deal with plastic baby bottles?

A kind reader asks: "Have you heard anything about plastic baby bottles being bad for baby?"

The main concern with plastic baby bottles is that they may leach Bisphenol A, which is what's called an endocrine disrupter or xeno-estrogen. That is, it looks enough like estrogen that proteins in the body bind to it as if it were estrogen, but since it's not the response isn't quite right. This same general principle is kind of part of what makes birth control pills work- disrupting a hormonal feedback loop- except then it's on purpose.

BPA certainly does bad things to rats at both high and low doses. There aren't a lot of decent studies on what it does to humans because a) humans are hard to study; b) reporting of exposure is notoriously ineffective; c) it's unethical to feed people harmful things; and d) it's often very hard to see a weak effect and so researchers would need to, say, take weekly blood samples from a lot of people to get long-term effects.

But in rats, it appears to do a variety of small things- not like instant cancer, just small effects: enlarged prostates, changes in some blood proteins, various hormonal disruptions, effects on the mammary gland, maybe developmental changes. Before you become too distressed, let me remind you that rats aren't people or we wouldn't need to do clinical trials, and nobody's going to keel over immediately from any of these things. On the other hand, bad effects in rats are usually at least a fairly decent gauge of what to avoid. These effects are nothing that's going to do someone instant harm, but nothing that sounds great.

Fortunately, there are many kinds of BPA-free plastic and glass bottles now, which it makes it much easier to avoid in baby bottles.

But. The problem is that this isn't the only place you'll find BPA. It's in some can linings- so that tomato sauce you just bought may have BPA in it. Nalgene bottles are another well-known culprit, but they sell several kinds of BPA-free bottles now. Some are even colors. Maybe your plastic spatula, your plastic-lined travel mug. It's at low levels in some drinking water. It does go from the mother's body into breastmilk, though at low levels; this extraordinarily long and boring review will tell you all about it if you're curious (email me if you want a copy of the PDF).

Nonetheless, take heart! It happens that the safest plastics are numbers 1 and 2, which are also the most recyclable. Here's a good summary of which plastics leach what- the recycling numbers are extremely handy. A quick rundown: 1 and 2 are safe and recyclable; 4 and 5 are safe but less recyclable; avoid 3, 6, and 7. Plastic wrap is PVC (3); I avoid that too.

After writing this I was so alarmed that I went and checked all our plastic goods. Happily, most of it is #2 plastic, along with the ketchup bottle and the juice bottle. Unhappily, much of it is not labeled. And then I threw away the Nalgene bottle.

Seriously, I don't think it's time to get really upset. But it might be best to cut out plastic when possible, and try not to worry about it the rest of the time. And plastic wrap and processed plastic-packaged foods are not so good. But we knew that already, right?

On Cloning (But Not Sheep)

Agarose gel of PCR products (DNA fragments). Lanes: 1, ladder; 2-5, pieces-o-gene.

Since this is supposed to be a blog about science...

When people hear 'cloning', their minds often turn to Dolly the Sheep, Carbon Copy the Cat, or Glowy the Glofish and his 5,000 cousins. (There is a variety of other successfully cloned livestock, apparently.)

Cloning and genetically modified organisms manage to inspire great hysteria. "Defects in clones are common,"- true, and the phenotype is usually DEAD- "and cloning scientists warn that even small imbalances in clones could lead to hidden food safety problems in clones' milk or meat," a 'Center for Food Safety' squeaks at us. (I would like to meet these scientists.)

On the other hand, animal studies involving cloning are the backbone of much biology. Want to know what your mammalian gene does? Eventually, you're probably going to knock it out in a mouse. How? You take embryonic mouse cells, genetically modify them so the gene in question is terminally screwed up, culture a bunch of cell clones (i.e. you put the cells on a plate and they grow into clonal, or identical, populations), transfer some cells back to an early-stage mouse embryo, and stick the whole deal back in a mouse.

But I'm not going to talk about any of that.

No, let's talk about PCR, which revolutionized biology in ways that cloning may never achieve. Making a million copies of DNA bits, rapidly and cheaply, made everyone's lives a great deal easier. Back in the day, one could get a PhD- a whole PhD!- for cloning and sequencing a single gene. Nowadays, I can do that in a week or less, for under $100.

Next: How PCR Works (More Than You Ever Wanted To Know)

Don't Believe People Who Profit From Lying (Or: Are Oats That Great?)

Ever eat that round cereal with a hole in the middle? Made of oats?

They've started trumpeting a 1998 study entitled "Cholesterol-lowering benefits of a whole grain oat ready-to-eat cereal". This study bases its conclusions entirely on this table, which is all about cornflakes vs. oaty cereal:



Now, with the help of this handy tool, I've calculated all the P-values:
Cornflakes before vs. after, 0.35; Oats b/a: 0.15; Cornflakes before/ oats before: 0.61; Cornflakes after/ oats after: 0.07. So there are no significant differences in this study, even for before and after oat consumption. Cheery oat cereals may reduce your cholesterol 4%- and that's only LDL, not total, but this study shows no statistical significance.

There are plenty of other studies showing that 5+ grams of oat fiber, or other soluble fibers, a day can help lower cholesterol. Which is all very well and good. My point is merely, this study does NOT demonstrate this effect. Nonsignificant differences might as well not exist. Don't believe everything you read!

So... all the same, eating more fruits and veggies and whole grains is good for your health. Who knew? (And if someone could convince my FIL, who has gout- gout! GOUT! - to stop eating meat at every meal, that'd be great too. I'm afraid he's going to meet his reward before he has grandkids.)

Ask a Scientist: Folate and Neuroblastoma

Aurelia asked rather a while ago (sorry, Aurelia!) if I'd supply a short review of folic acid and neuroblastoma incidence. Well, I'm sitting in lab waiting for something to cool off, so here goes.

Folate supplementation in pregnancy is known to reduce neural tube defects up to 70 or 90%, though only if taken in the first 28 days of pregnancy, after which the neural tube is formed and fused, i.e. it's too late. So if you're planning to become pregnant, either eat a lot of spinach or take your vitamins.

A Canadian Motherrisk study is one of the stronger pieces of evidence for folate preventing pediatric cancers. They survey NB cases before and after mandatory flour supplementation took effect, and find that the rate declines from "1.57 cases per 10,000 births before to 0.62 case per 10,000 births after folic acid fortification (P < .0001)". However, another author points out in a commentary that they would only be expected to catch about 40% of the NBs that would eventually arise. (I didn't check their calculations, but let's assume that's correct.) .62/.40 = 1.55 cases per 10,000 births. In other words, that implies no effect.

On the other hand, a recent meta-analysis looked at a bunch of studies on prenatal vitamins and cancers, and concluded that vitamins reduce NB risk by 50%. (Now, meta-analyses are notoriously flawed, because each paper is looking at something different, asking different questions of the patients, and using different samples.) So this could be true, but it could not. Who knows.

A large 'case-control' study- still interview-based, and I add quote marks because they didn't divide the mothers into groups and give half vitamins, they simply asked questions and then divided up the groups- did find an association between prenatal vitamin usage and decreased risk of brain tumor.

This study showed the same thing, but slightly more convincingly because they sample several hundred NB patients, comparing vitamin users to non-users. This German study, on the other hand, is rubbish. They do a case-control study of pediatric cancer incidence and conclude that vitamins cause cancer! Aaaaah! Eat your meat and potatoes. But. For the 'pain reliever use and bone tumor' association, for example, they have 93 (98.9% of respondents) who had cancer and didn't use pain relievers. But they only had one who did. From this they conclude that taking pain relievers lowers your risk of bone cancer by 70%, i.e. that the odds ratio is 0.33. Ah, no.

This is what's known as 'insufficient statistical power', i.e. 'Your sample's waaaay too small.' They are looking at a rare condition in 94 patients and it just so happens that only one mother took asprin. Well, to really study this, you need either a huge cohort of people who did and didn't take asprin, or a large set of bone cancer patients along with complete medication information for each mother. Neither is precisely ideal, but that's human studies for you. All those pesky ethics rules.

As far as I personally would judge, folate probably does reduce neuroblastoma risk. It makes sense, even though the mechanism isn't known, because folate clearly has something to do with proper neural development, migration, and signaling (to close the neural tube). To give a little perspective, we've used aspirin for hundreds of years until recently discovering it's a COX inhibitor. A mechanism isn't necessary to know something works.

Overall: The evidence that folate might cause cancer is weak. Prenatal vitamin use is strongly associated with lower rates of birth defects in many studies, is associated with a 90% reduction in neural tube defects, and may be associated with reduced risk for some pediatric cancers.

By the way: Methylenetetrahydrofolate reductase is apparently abbreviated 'MTHFR.'

Ask A Scientist: Male-Mediated Teratogenesis (and Bupropion)

A correspondent asks: "What are the risks and possible side effects of long-term usage of bupropion in terms of sperm genetic abnormalities?"  (In other words, "Do I need to worry that my husband is taking Wellbutrin?" and "Does Wellbutrin harm sperm?  Does it affect fertility?  Does it cause genetic problems?")

Well, speaking from a scientific standpoint, there's no evidence that bupropion causes DNA damage in any way; bupropion doesn't even cause so-called 'congenital' defects. (Congenital usually means not genetic, but rather caused by environmental or unknown factors.) So probably, bupropion does nothing to sperm.

To broaden the question, what is known about teratogenesis- physical but not genetic changes- caused by sperm abnormalities?

This is a very hard question to study. Any male-mediated effect usually has weaker effects than when female-mediated: a mother's dioxin exposure is more harmful than a father's.* For example, moderate male lead exposure doesn't appear to be associated with birth defects in future offspring, but maternal exposure causes neural tube defects and other nasty things.

Why? The sperm:egg volume ratio is something like 1:1000, so the metabolic organelles are almost all maternally contributed. Only DNA damage (whether direct or 'epigenetic', i.e., non-permanent) in sperm would have an effect on fetal development. That is: if there's something non-genetic wrong with the egg, or if that particular egg is wonky in some way, it can have a huge effect. Not so much with the sperm, because it's mostly a bundle of DNA with a tail attached. Or think of it this way: the father is contributing 23 chromosomes and maybe a mitochondrion, and the mother is contributing everything else. For nine months. Which one do you think matters more? So teratogens have much more severe effects in the female than the male.

(To return briefly to bupropion: lead is a LOT nastier than bupropion. I wouldn't worry about ADs. Really.)

(Also, to digress: women who are breastfeeding often fear drinking. But calculate it out: if your BAC is 0.1%, which is equal to really drunk, then the milk has 0.1% alcohol in it too. Therefore it is 0.2 proof. Like a beer diluted 50 times. This is 2/5 of a cup of beer in a gallon of water. Yeah! That's a lot!!! Similarly, the plasma concentration of the drug is usually equal to the milk concentration, i.e. the breastfeeding child gets some very low percentage of the dose. This is why doctors who categorically refuse to prescribe to breastfeeding women are, in my personal opinion, IDIOTS.)

What if something causes DNA damage in sperm but not eggs- what if there's a drug women use and it's fine, but it's bad for male reproduction? It's theoretically possible- the cell division processes that generate gametes (meiosis) are somewhat different. But it's not likely, because really they're quite similar. (If anyone has a counter-example, please send it.)

It's also possible that nonpermanent genetic changes, which are called epigenetic modifications, are caused by a lot of things. This is really, really hard to study too. If anyone really wants to know why, I can write about it later, but take my word for it.

It's widely thought that many miscarriages are caused by genetic abnormalities. But most drugs don't cause genetic abnormalities, for the simple reason that if they do, they'll raise your rate of cancer (among other things) and be pulled from the market. So sperm-contributed mutations are generally from chance, combined with an inherent error rate in DNA replication, and with the caveat that the error rate may increase with age.** Sometimes from environmental exposure to really nasty stuff, too.***

Interestingly enough, even thalidomide- the stereotypical Bad Scary Teratogenic Drug- caused only 20-30% malformations, and that only when taken between d34 and d50 (i.e. in the second month) by the mother. Motherisk points out that 'When asked, even women exposed to nonteratogenic drugs believe they have a 25% risk of having children with major malformations, apparently the size of the risk with thalidomide itself. This unrealistic perception leads pregnant women to avoid medications even when they clearly need them.' In other words: look at the evidence before freaking out about drugs; it's probably not as bad as you think.

All this aside, there are many chemicals thought to harm sperm - but note that many of their first-line effects are reduced sperm count and motility.

Overall, biologically speaking, there's no evidence that bupropion (Wellbutrin) causes harm to sperm in any way. This doesn't mean it absolutely doesn't under any circumstances, but viewing it scientifically, it is extremely unlikely that the drug is harmful to men or women or fetuses.

*This paper discusses increased risks, but doesn't differentiate well between maternal and paternal exposure- they have historically been quite difficult to assess and track in Vietnam. Take-home message: self-reporting is unreliable, and dioxin is not good for humans. Here's a more careful study. Correlation is not causation.
**This theory, which I so don't have time to go into, has also to do with why the cancer risk increases dramatically with age. In general, all systems degenerate entropically over time. Including the human body.
** But here's a really poorly researched paper on Scary! Bad! Everything. Anecdotal, non-prospective and non-tracking evidence is notoriously bad. I'm just saying.

Ask a Scientist: Listeria and Pregnancy

I know these are overwhelmingly about pregnancy. This is because my pregnant friends and acquaintances (now great, mighty and numerous) and my girlfriends who may one day be pregnant, keep asking things. And I believe that risks in pregnancy are, as Mark Twain said, greatly exaggerated.

Listeria! Not something you hear about a lot, because the CDC's most recently reported incidence- in 2005- was 0.3 per 100,000 (3 in a million).

A friend's midwife ('Not the kind with little bells and a faint smell of patchouli. These midwives are all mean and brisk and no-nonsense') told her that a pregnant woman's risk of listeriosis was 20 times the basal risk. Really?

Listeria is, in fact, super extra nasty if you're pregnant because it crosses the placental barrier and infects the fetus. This frequently results in miscarriage or fetal death.

[As a side note, most listeria infections in pregnancy occur in the 3rd trimester, when suppression of Th1-mediated immunity is at its maximum.]

The 20-fold figure is based on a single CDC study in 1986. They tracked all reported cases of listeria for one year in six health departments across the country, and came up with a rate of about 5 per million population overall. 27% were in pregnant women (pw); the authors record 67 perinatal cases, which at the time was 12.7/100,000 live births. The CDC reports an annual birth rate of 13.9 per 1000 population, so 1.1% of population is pregnant women (let's estimate, based on being pregnant 40/52 weeks). So if you're not pregnant, your incidence of listeria is 5 per million, and if you are, it's 127 per million pw. This a 25-fold increase in risk.

Data from France, where the overall incidence is about 3.5 per million, reported 24% of all cases occurred in pw. The birth rate is 11.99 per 1000- 0.92% of the population consists of pw; this works out to 2.6 cases per million nonpregnant population and 86 cases per million pw, or a 33-fold enhancement in risk.

A more recent seven-year CDC FoodNet survey ("Pregnancy-associated listeriosis in Hispanic women in FoodNet Sites", also here) found a rate of 3 per million population, and 16% were in pregnancy (2.5 per million non-pw, 43 per million pw)4. Of all cases, 7% of patients were Hispanic, but 28% of pregnancy-associated cases were in Hispanic women (eating queso fresco seems to be the cause of this increase). So that makes an 17-fold risk enhancement.

To further muddy the waters, of 75 cases reported in Finland over 9 years, only 4 (5%) were in pregnant women. On the other hand, in several reported many-case outbreaks, up to 85% of infections were in pregnant women. Except when they weren't: one in Boston had
14% infection in pregnant women. And then another review says maybe overall it's more like 35%? They think? And another study of isolated infections found 11% in pregnant women.

In case you need to feel alarmed about eating anything, Listeria infections or contamination have been found not only with cold-smoked fish (lox, etc.), feta, brie, other soft cheeses, and hot dogs, there have also been cases and outbreaks from coleslaw, pasteurized-but-somehow-contaminated chocolate milk, undercooked chicken, deli meat, hummus, melons, lettuce, packaged sandwiches, pasta salad, egg salad, mushrooms, dips, and pretty much anything else you could imagine.

So overall, the moral is, the most risk is from preserved fish, cheese from unpasteurized milk, and deli meats. Possibly also avoid vegetables, milk, hummus, Finland, and eating out if you're feeling particularly paranoid. The risk of listeriosis is increased substantially in pregnancy, and what do you know, it might really be 20-fold.

But don't be too alarmed. Remember even with the highest numbers, the total incidence of Listeria in pregnancy is only 0.01%.

Ask a Scientist: Metformin and PCOS and Pregnancy

What's PCOS? Polycystic ovary syndrome is characterized by, among other things, a lack of normal menstrual cycles, lack of ovulation, and cysts in the ovaries. It's not clear exactly what causes it.

Some people believe it has to do with gonadotropin-releasing hormone (GnRH), which is itself secreted by the hypothalamus. GnRH acts on the pituitary to control secretion of two of our favorite hormones, luteinizing hormone and follicle-stimulating hormone (LH and FSH). Some studies indicate that GnRH secretion in PCOS patients happens in fast pulses, and doesn't have normal feedback with estrogen/progesterone down the line. The end result is that PCOS patients end up with high LH and low FSH and so don't ovulate.

-Much of the following information is summarized from this review; email me if you want but can't get the pdf.

PCOS and insulin: PCOS is strongly linked to insulin resistance, where cells doesn't react correctly to insulin, and so insulin is produced at high levels as the body attempts to compensate. Although this effect is increased by obesity, even 'lean' women with PCOS have some insulin resistance. Pretty soon testosterone production increases in the ovaries and levels of the main testosterone-binding protein decrease. (This doesn't help any with the ovulation problems.) There are theories that this gets into a feedback loop with the pituitary and then you get even more testosterone-producing-stimulation badness.

Because of this, drugs that counter insulin resistance are frequently prescribed to manage PCOS in women who wish to conceive. Birth control pills are given to women who merely wish to manage symptoms. Metformin, hereafter M, is the best studied (and the one I was asked about). Troglizatone, the next-best-studied drug, has been withdrawn by the FDA due to liver toxicity.

Metformin and ovulation: M alone, in clinical trials, has caused ovulation in 34% of an obese population; when combined with clomiphene, 90% was achieved. This has been confirmed multiple times; the increase is about 4-fold over untreated, or a total of 46% in non-obese patients. Clomiphene is sometimes used alone, but M works better alone or in combination with clomiphene.

There are a few studies where gonadotropins are combined with M; multiple follicles (associated with multiple implantations and, if no reduction is performed, poor pregnancy outcomes) were reduced from an average of 4.5 to 2.5. Cycle cancellations due to too many follicles were also reduced. Another study found that M increased mono-ovulatory cycles and decreased hyperstimulation (These are good things.)

When M is compared to laparoscopic ovarian diathermy (in obesity only), which sounds kind of icky and apparently lacks randomized trials, ovulation is the same but M is better at allowing and maintaining pregnancy (overall: 18 vs. 13% of patients have pregnancies, 15 vs. 29% SAs).

Metformin and pregnancy loss: Spontaneous abortion (SA) is increased dramatically in PCOS. Although anovulation is a major factor in the infertility of PCOS, women with PCOS are also at high risk of first-trimester early pregnancy loss. Compared with an SA rate of 10-15% of clinically recognized pregnancies in controls, the comparable EPL rate of women with PCOS is 30-50%. M treatment seems to knock the risk back down to the basal level, if it is taken throughout pregnancy, and may reduce the risk of gestational diabetes ['GD developed in 4% of pregnancies versus 26% of previous pregnancies without M'].

Metformin and teratogenicity: In a small study of 126 infants, there were no teratogenic effects observed. Other small studies have come to similar conclusions. M is currently a Category B drug (along with Sudafed). Note, however, that in 126 infants one would not expect to see more than 3 congenital defects, and M would have to increase risk at least two-fold to be reliably observed.

Motherisk conveniently summarizes the animal data for us: high doses (600 mg/kg/day) in some animal or other have no effect, and clinical doses have no effect, and cultured embryos are susceptible to M toxicity. There may be a low incidence of malformation in rat embryos.

Type II diabetes is, like PCOS, characterized by insulin resistance and hyperinsulinemia and is associated with not only pregnancy loss and infant mortality, but also a sharply increased incidence of congenital malformations (about 11-fold). This makes it harder to tell if M does cause congenital defects because most of the population taking M has these 'confounding factors'.

Summary:

• M is often effective at increasing ovulation in PCOS.
• It has been combined with other drugs, or with gonadotropins, to successfully induce ovulation in a majority of study subjects.
• Usage through pregnancy, or at least the first trimester, sharply reduces the risk of SA in PCOS patients.
• In small studies, there is no teratogenicity in humans and minimal teratogenicity in animals. However, because of complications, it is difficult to tell if there is an effect.
  

Ask a Scientist: Antidepressants and Pregnancy (6)

Series: Antidepressants, Pregnancy, Risks

• Introduction

1. Spontaneous abortion
   
2. Premature birth
   
3. Congenital or teratogenic defects; that is, malformations in utero
• Cardiac defects
  
4. Antenatal/ postnatal adaptation problems
   
• Muscle stiffness
  
• Breathing distress at birth
  
• Neurological withdrawal symptoms
  
5. Breastfeeding risks
   
• Infant weight gain
• Infant serotonin levels
  
• Long-term neurologic development
  
6. Maternal risks of going without treatment
   
• Relapses
  
• Infant failure to thrive
  
• Parenting problems, attachment, child behavior

This information is drawn from published articles in medical journals, some of which are compiled here, and from public databases. (See also: summaries on Motherisk, FDA registries for pregnant women, this excellent Medscape article, and the FDA's index by drug name.) Information is also available through PubMed by searching various terms; I suggest starting with 'antidepressants and pregnancy'.

A list of generic and trade names is here.

*****
By the way, the pregnancy-and questions currently outnumber all others about ten to one. Who'd have guessed? Women are curious and underinformed!

6. Maternal risks and consequences of untreated depression

Aside from the obvious things about not being treated for depression (during or after pregnancy)- like not eating right, and never managing to wash the dishes, and being completely unable to take joy in anything, for example- it appears that being depressed affects children's growth and development. I find this entirely unsurprising: interaction with children, including how many words one speaks to them, has a huge impact on their development. (Here is where I would put references to 200 studies about this, if I had time.)

[Here is where I repeat that this is not medical advice, okay? These are my personal opinions.]

Each woman (possibly with her partner) decides what risks she is willing to take in pregnancy, and I'm all for that; the reason I wanted to write this is because so many doctors refuse to treat depression in pregnancy without evaluating the risks or letting the patient decide, and this in a situation where the risks appear to be relatively low (as opposed to, say, isoretinoin, which is associated with a roughly 25% incidence of major birth defects and a high rate of SA).

But especially once one is postpartum, I would argue there is NO reason to leave depression untreated. It is neither ennobling nor noble. It is merely dreadful.

And by the way, I have eliminated from consideration all studies that claim vitamins/ fish oil/ acupuncture will make all MDD patients, and especially women, feel all better. Because I don't do quackery. It exhausts me.

6a. Relapses

Relapses are very common in depression; one study, done over 1.5 years, cites a 40% relapse rate in the control group. They are investigating only one drug, but even in this isolated case, treatment reduces relapses to 26% and increases the time to relapse.
Another study cites a recurrence rate of 65% in controls vs. 27% treated with the same one drug. [And boy, are those control figures, um, depressing, by the way.]

These studies are both in cohorts with MDD. If you have MDD and it's untreated, guess what? It'll probably come back!

A series of small studies in pregnant women found that while half stopped or tried to stop taking ADs while pregnant, 75% of MDD patients "experienced depressive relapse". Of patients who discontinued or decreased AD dose, 61% went back to taking them.

6b. Infant failure to thrive
As previously mentioned, one study comparing 5 SSRIs and venlafaxine shows that maternal depression of >2 months decreases infant weight gain. This is confirmed by other studies, one longitudinal one which finds that 'infants of prenatally depressed mothers showed significantly more growth retardation than controls at all time points.'

A review mentions that extensive evidence shows that persistent deprssion is associated with higher cortisol levels and lower serotonin, and these stress effects are echoed in the infant's 'biochemical profile'. They also mention that prenatal depression is associated with both physical and behavioral complications throughout pregnancy, birth, and infancy. These results are supported by animal studies showing that maternal stress affects pregnancy outcome and offspring development.

To further sweeten the pot, an assortment of studies find that maternal depression and/or other psychiatric conditions are associated with missing prenatal checkups, prematurity, low birth rate, and more NICU stays: i.e., poorer self-and-fetus care during pregnancy, and poorer outcomes in birth.

6c. Parenting problems, attachment, child behavior

Here I'll quote because I couldn't put it any better:

"Depressive symptoms have also been found to be associated with negative parenting behaviors, insecure attachment, and subsequent internalizing behaviors in the child."

"Through children's early adolescence, maternal drug use [Ed: illegal, not prescription] is no more inimical for them than is maternal depression.... [on] mothers' everyday functioning [:] results showed that negative parenting behaviors were linked with multiple adverse child outcomes."

Child difficulty [to the parents] was associated with elevated levels of psychosocial stress, but only for some participants. Parental psychopathological symptoms during pregnancy should thus be considered as a risk factor for elevated and prolonged depression and elevated psychosocial stress in mothers and fathers across the transition to parenthood." So if you're stressed and depressed, your child may be even more of a handful. Eeeek.

"If depression persists into the postpartum period, it can have long-term consequences for both mother and baby. Mothers might go on to develop chronic mood disorders, and untreated postpartum depression can impair mother-infant attachment (level I evidence). Finally, being exposed to a chronically depressed mother can have cognitive, emotional, and behavioural consequences for a child."

Conclusions:

• Depression is miserable and loathsome.
• Untreated depression in pregnancy is associated with higher risk of poor outcomes.
• Infants with depressed mothers have weight-gain delays.
  
• Untreated depression after pregnancy is associated with developmental troubles and poor parenting.
• Parental depression interferes with children's physical and emotional development.

Cheers, and thanks for reading to the end of this exhaustingly long series. Whoooo.

Ask a Scientist: How Does Kool-Aid Clean Brass?/ Also, Randomness

I know I promised (the one person still reading) the last part of ADs and pregnancy today, but... um, I haven't written it yet. I was impeded by a) laziness; b) a Jews Do Iron Chef competition; and c) a lingering cold which features drippiness, lethargy, and a nasty taste, rather as though I'd snorted DTT.

And my procrastinating little self has a ton of labwork, a meeting, and a postal trip (perhaps figurative, but certainly literal) to do yet today. So, here's a less deep topic for this fine Monday morning.

Also, birthday loot:


Me to Mr. S: Good Scotch, shoes (not pictured on account o' he was wearing them).

Mr. S to me: 400 crossword puzzles.

Me to self: Port, silly pajamas.

Sister to me: scarf.

Sci-in-Law parental and grandparental units to us: Cash and cards. And a Superman paper airplane.

Mother to us: Hot water urn that doesn't leak on the cord. Not yet arrived on account o' neither of us plans in advance.




***
Irie asks: "My husband was in the Navy and told me that they used Kool-Aid to clean brass. How does that work?"

I think the first question is why on earth does the Navy use Kool-Aid? If there are any readers out there who know, please tell me; I'm fascinated.

Brass is an alloy of copper and zinc. Like most metals, it tarnishes: the metal combines with oxygen and/or water to form oxides, and it gets less shiny, or green, or brownish. If it's been touched, it will also accumulate a film of oil and minerals and dirt from the skin, which minerals etc. may hasten oxidation. Many metal oxides are insoluble or sparingly soluble in water. For example, if you put a tarnished penny in water, nothing happens: the tarnish doesn't dissolve off.

There are at least two easy at-home ways to shine metals: with a mild abrasive, or with acid. Abrasives such as baking soda or cleaning powders will simply rub off the layer of oxide, leaving the shiny metal visible beneath. So you can shine a penny with Comet, if you want. Or, if you're obsessive like me, you can shine your pots and pans with a baking soda paste.

The other popular way to clean pennies is to stick them in vinegar. Why? Because the acid reacts with the metal oxide and solubilizes it (by exchanging the hydroxide for the acid's counter-ion: Fe(OH)3 + 3H(CH3COOH) --> Fe(CH3COOH)3 + 3H2O, for example). This is also why you can clean pennies in soda (pop for you Yankees): it's acidic.

So why does Kool-Aid clean brass? As a powder, it's probably abrasive just like baking soda; as a liquid, it's quite acidic. (Irie notes that the Navy used it in the liquid form.) The ingredients, other than colors and flavors, are: citric acid, salt, cornstarch, calcium phosphate, and ascorbic acid. Tang also contains citric acid, which is why it can be used to clean various things including commodes. And in fact, non-Navy people do clean stuff with Kool-Aid. Personally, I go for vinegar. Or Comet.

Bottom line: The acid in Kool-Aid solubilizes the oxides that tarnish brass.

Fun random fact: You can mix Kool-Aid powder into your Rice Krispy treats to make tropical-punch-flavored red marshmallowy crunchy snacks. All together now: ewwww.

Ask a Scientist: Antidepressants and Pregnancy (5)

Series: Antidepressants, Pregnancy, Risks

• Introduction

1. Spontaneous abortion
   
2. Premature birth
   
3. Congenital or teratogenic defects; that is, malformations in utero
• Cardiac defects
  
4. Antenatal/ postnatal adaptation problems
   
• Muscle stiffness
  
• Breathing distress at birth
  
• Neurological withdrawal symptoms
  
5. Breastfeeding risks
   
• Infant weight gain
• Infant serotonin levels
• Long-term neurologic development
6. Maternal risks of going without treatment
   
• Relapses
  
• Infant failure to thrive
  
• Parenting problems, attachment, child behavior

This information is drawn from published articles in medical journals, some of which are compiled here, and from public databases. (See also: summaries on Motherisk, FDA registries for pregnant women, this excellent Medscape article, and the FDA's index by drug name.) Information is also available through PubMed by searching various terms; I suggest starting with 'antidepressants and pregnancy'.

A list of generic and trade names is here.


*****
5. Breastfeeding risks
Okay, folks, there are a LOT of articles about this. If you want to see them, search 'breastfeeding antidepressants' on Pubmed. As far as I can tell, they largely come to the same conclusions. So here's a summary of the first 50 or so.

Did you know there's a Journal of Human Lactation??? They have an excellent 2001 review of various ADs in lactation.

Motherisk has a lovely patient-oriented summary about drugs and breastfeeding. (Thanks, Aurelia.)

Summary:

Use of SSRIs and TCAs (except doxepin) is considered safe. Occasional adverse events are reported. They generally resolve once either breastfeeding is stopped, the schedule is altered to minimize infant exposure, or the mother stops taking the drug.

Of the SSRIs, sertraline and paroxetine have very low numbers of adverse events. Mild adverse events are associated with fluoxetine and citalopram (colic, crying, etc.).

Antipsychotics and lithium are not recommended in breastfeeding.

5a. Infant weight gain

A study comparing the 5 most common SSRIs and venlafaxine shows absolutely normal weights in all infants. They note, however, that maternal depression of >2 months DOES make the babies gain less weight.

Paroxetine in 27 mothers has no effect on weight gain or developmental milestones. A meta-analysis of SSRIs (other than fluoxetine) and TCAs also showed no effect on weight gain.

One study on 26 breastfed infants whose mothers took fluoxetine, however, showed an average 1-pound difference in weight gain over 6 months. The JHL review notes that none of the weights, even the lower ones, were below the national mean. Another study notes lower birth weights in fluoxetine usage, but only studies five exposed infants (i.e., not enough to say).

5b. Infant serotonin levels

Unsurprisingly, this isn't exactly studied a lot. One editorial (pdf) summarizes a few studies on the matter: rats exposed to fluoxetine had a 50% occupancy of their serotonin reuptake protein (5-HTT); however, apparently about 80% is necessary for an AD effect. (Also, nobody's quite sure exactly what 5-HTT does or how it relates to depression.)

A human study on fluoxetine says that "most infants may continue to breastfeed without experiencing meaningful changes in" serotonin uptake and metabolism. A small study that lumps together three ADs seems to suggest that newborn serotonin levels are changed by maternal AD use, but stabilize within one month. They go on to suggest that it could have terrible! developmental! consequences but I think they're full of beans (and hand-waving).

5c. Long-term neurologic development

You'll be shocked (shocked!) to learn that basically, nobody has a clue. In infants followed for up to a year, they're perfectly normal. After that, well, neurology is complicated and poorly understood anyways. The best that can be said is that SSRIs and other ADs don't appear to do anything either a) really nasty b) quick or c) much of the time.

Reported infant exposures in breastfeeding

The medical literature says that infant exposure of <10% is largely considered safe. In addition, some data indicate that breastfeeding can alleviate neonatal abstinence syndromes (here in, for example, methadone exposure. Think how much more innocuous sertraline is.).

These exposures and reported adverse effects in lactation are taken from the JHL article and a review in Drug Safety. Both studies note that changing dosage time and breastfeeding times can significantly reduce/ minimize infant exposure.

A modeling study notes that the total infant yearly dose (of fluoxetine) is generally about 1 or 1.5 of the mother's daily dose. This isn't very much. Another meta-analysis summarizes overall exposures for many psychoactive drugs.

Fluoxetine: Infants may receive up to 20% of maternal daily dose (MD). Some reports of extreme infant fussiness which resolve on maternal drug or breastfeeding discontinuation. Other studies indicate no complications in infants. Labelling recommends not using in breastfeeding, but is generally considered low-risk.

Paroxetine: Infant exposure ~3% MD.

Fluvoxamine: May be contraindicted in infants who are taking caffeine (i.e. for apnea). Infant breastmilk dose 0.75%-1.4% MD. Alternatively, undetectable in infant serum.

Citralopram: 0.4%-1.8% MD. No adverse effects reported.

Sertraline: 0.5-2% MD; sometimes undetectable in infant. No known adverse effects. Considered safe and well-studied.

Nefadozone: Adverse event in preterm infant receiving 0.45% of maternal dose; attributed to reduced hepatic function. Calculated to be safe if infant is confirmed to receive less than 10%.

Bupropion: ~0.02% MD; therefore considered very safe in lactating mothers. One case report of seizures; may be avoided if family history of seizure.

Venlafaxine: up to 7.6% MD; metabolized well by infants. No apparent effect on weight. Well-tolerated.

Conclusions:

• Being depressed affects infant weight gain.
• SSRIs and TCAs in breastmilk don't.
• There are no data on long-term neurological consequences.
• In any case, infant AD dosage through breastmilk is very low.
• On the other hand, infant/child exposure to maternal depression definitely has bad effects on weight gain, "emotional and behavioural development", maternal-infant bonding, cognitive development, and children's levels of fear and anxiety.
• Many ADs are well-studied in lactating mothers; few-to-no adverse events are reported. Most are widely considered to be safe.

Ask a Scientist: Antidepressants and Pregnancy (4)

Series: Antidepressants, Pregnancy, Risks

• Introduction

1. Spontaneous abortion
   
2. Premature birth
   
3. Congenital or teratogenic defects; that is, malformations in utero
• Cardiac defects
  
4. Antenatal/ postnatal adaptation problems
   
• Muscle stiffness
  
• Breathing distress at birth
  
• Neurological withdrawal symptoms
  
5. Breastfeeding risks
   
• Infant weight gain
  
• Long-term neurologic development
  
6. Maternal risks of going without treatment
   
• Relapses
  
• Infant failure to thrive
  
• Parenting problems, attachment, child behavior

[Disclaimer: I am not a medical professional; this is an academic summary of the evidence available and not a medical opinion. I do not offer medical advice. If you require medical opinions or treatment, please consult your physician.]

This information is drawn from published articles in medical journals, some of which are compiled here, and from public databases. (See also: summaries on Motherisk, FDA registries for pregnant women, this excellent Medscape article, and the FDA's index by drug name.) Information is also available through PubMed by searching various terms; I suggest starting with 'antidepressants and pregnancy'.

A list of generic and trade names is here.


*****
4. Antenatal/ postnatal adaptation problems

Certain post-birth symptoms are associated with maternal antidepressant- usually SSRI, but also venlafaxine- use in the last trimester. They are all categorized as 'neonatal withdrawal [or abstinence] syndrome'. These syndromes rarely lead to anything other than an unpleasant week with a cranky baby, and maybe a couple days in the NICU.

Muscle stiffness and/or convulsions are treated with support and/or low doses of SSRIs. A certain amount of breathing and/or neonatal distress is associated with maternal SSRI usage; it is rarely fatal except in infants with other complications, and is treated supportively.

4a. Muscle stiffness

A recent study looked at reported incidences of seizure-like symptoms in mothers taking SSRIs. They found 93 reported cases of these symptoms in a large adverse-event database and tried to eliminate other factors and drugs that might have caused the symptoms. Then they compared the incidence to the overall incidence in the adverse-event database. 74 cases were 'certain'; of those, 51 involved paroxetine use, 10 fluoxetine, 7 sertraline, and 6 citalopram.
Most of the incidents were neurological (see below); there were 11 total instances of hypertonia and convulsions.

Two case studies on venlafaxine report increased hypertonia (stiffness) in neonates; one treats with a low dose for a week, and the other recommends breastfeeding to mitigate withdrawal.

One case study reports convulsions associated with paroxetine.

4b. Breathing distress at birth (and other distress, ICU)

Third-trimester SSRI use is associated with babies more often having to go the NICU with breathing problems, neonatal lowered heart rate, or other unspecified distress. Various studies estimate the risk at various levels.

A large Canadian cohort of 75,000 mothers taking any of 6 SSRIs showed that 13.9% vs. 7.8% of controls had some neonatal respiratory distress (that is, had to spend some time in the NICU). A Finnish population study likewise showed that, of infants whose mothers bought SSRIs in the 3rd trimester, 15.7% vs. 11.2% for infants whose mothers bought SSRIs in the 1st trimester. A Motherisk study specifically on citralopram found an increased chance of having to go to a special-care nursery.

Another study, the inspiration for a recent mildly hysterical FDA alert, looked for infants with PPHN (a condition where there is high blood pressure in the lungs and therefore oxygen deficit) and then tried to see how many mothers used SSRIs. They found an increased risk of about 6-fold; that is, in their study, from 7/1000 to 37/1000. The problem with this study is they looked for a condition and then tried to check drug usage. It would be better to look among a large cohort of SSRI users (like in the Canadian study) and check for PPHN. Also, they only find 14 cases of PPHN with SSRI use. I find this not terribly credible due to poor design and small sample size.

4c. Neurological withdrawal symptoms

Many newborns with SSRI exposure do manifest neurological signs of a 'neonatal abstinence syndrome.' In addition to the higher incidence of respiratory distress (as above), many infants cry a lot and have trouble sleeping. This number is reported as high as 30% in a small study (60 infants). Of these, 8 had 'severe' and 10 'mild' symptoms. The average duration was about three days.

Another small study examines 76 mothers who took ADs and 90 control. 63% of exposed infants had some neurological symptoms. 75% of these infants resolved within 3 days; premature infants were more strongly affected by SSRI side effects. And yet one more study of 46 infants found much the same: about 30% of infants have mild respiratory distress or twitchiness. (They also found that clonazepam, a benzodiazapene, exacerbated the effect, probably due to metabolic effects.)

Numerous case reports have been published on neonatal withdrawal symptoms in maternal use of citralopram, fluoxetine, paroxetine, and I'm sure other drugs. The above case studies on venlafaxine also report irritability, etc.

One exceedingly crappy case report tries to correlate paroxetine levels in one newborn with- gasp!- pneumothorax. They do not measure maternal paroxetine levels, but cord-blood levels are four times above the normal therapeutic range. And then the infant is irritable in addition to intubated, but gets better within a week. Correlation is NOT causation here.

Conclusions:

• SSRIs cause some increase in post-natal problems.
• These manifest as mild breathing distress, NICU time, and longer hospital stays.
• Some infants will experience muscle stiffness and, occasionally, convulsions.
• Some incidents of severe respiratory distress are reported.
• Many 3rd-trimester-SSRI-exposed infants will be irritable for several days after birth, due to neurological withdrawal.
• If symptoms are severe, some doctors recommend treating with tapered SSRI doses.
• Breastfeeding may mitigate withdrawal.
  
• These effects can be worse in premature infants, who are predisposed to breathing distress.

• Motherrisk reminds us that "Among pregnant women, abrupt discontinuation of antidepressants has been associated with withdrawal symptoms, including nausea and vomiting, diarrhea, sweating, anxiety and panic attacks, mood swings, and suicidal thoughts."
  
• They recommend exposed infants be monitored for a week or so after birth.

• This sounds really alarming, but remember, this is an increase in nonfatal events.
  
• A review on AD withdrawal in adults tells us that "Most reactions are mild and short-lived and require no treatment other than patient reassurance."

Ask a Scientist: Antidepressants and Pregnancy (3)

Series: Antidepressants, Pregnancy, Risks

• Introduction

1. Spontaneous abortion
   
2. Premature birth
   
3. Congenital or teratogenic defects; that is, malformations in utero
• Cardiac defects
  
4. Antenatal adaptation problems
   
• Muscle stiffness
  
• Breathing distress at birth
  
• Neurological withdrawal symptoms
  
5. Breastfeeding risks
   
• Infant weight gain
  
• Long-term neurologic development
  
6. Maternal risks of going without treatment
   
• Relapses
  
• Infant failure to thrive
  
• Parenting problems, attachment, child behavior

[Disclaimer: I am not a medical professional; this is an academic summary of the evidence available and not a medical opinion. I do not offer medical advice. If you require medical opinions or treatment, please consult your physician.]

This information is drawn from published articles in medical journals, some of which are compiled here, and from public databases. (See also: summaries on Motherisk, FDA registries for pregnant women, this excellent Medscape article, and the FDA's index by drug name.) Information is also available through PubMed by searching various terms; I suggest starting with 'antidepressants and pregnancy'.

A list of generic and trade names is here.


*****
3. Congenital or teratogenic defects; that is, malformations in utero

• All defects; cardiac defects
  

There is a lot of very good data tracking birth defects (and post-natal events) in SSRIs and several other antidepressants. These studies have used large numbers of study subjects, and have characterized the various possible adverse effects. I would say that these data are reliable and believable.

These studies are often reported as odds ratios (ORs): the chance that X will happen under condition Y, divided by the rate of X without condition Y. Sometimes ORs are high: 2-fold, 4-fold. Remember that the basal rate is important; does it increase from 1% to 2% or 10% to 20%? Is the basal rate noisy; that is, is it small enough that there is doubt what the basal rate really is; could it be 1-3% in reality? So a high OR is not always cause for alarm. (Moral: Read articles carefully.)

It is believed that first-trimester exposure has the most impact on congenital (i.e. pre-birth) deformities; for an example, see this; this is why many of these studies focus on women who took ADs in the first trimester.

For comparison, the CDC runs a long-term birth defect database called the MACDP. Its latest report, from 2004, reports 1.2% cardiac defects and 2.3% all birth defects (including trisomies) among all births in the study.

The same large Finnish database mentioned in Part 1 investigated rate of birth defects among women with SSRI purchases in 1st, 2nd, 3rd, or all trimesters. About 1400 women bought SSRIs in the first trimester, being: citalopram, fluoxetine, paroxetine, sertraline, and fluvoxamine. Most SSRIs had no effect on birth defect incidence. There was a questionable association between 1st-trimester fluoxetine exposure and cardiac defects: 2.3% vs. a reported rate of 0.8% in Finland. After the authors adjusted for other factors, however(they don't specify, but probably things like smoking, family history, cocaine use, etc.), they found no association.

A study of 150 women on venlafaxine (Effexor) detected no defects above the basal level of 1-3% in the general population. A literature review of many other studies says that fluoxetine, sertraline (SSRIs), bupropion, and low doses of paroxetine are not associated with birth defects. A few isolated cases are reported with venlafaxine.

A large study/database (pdf) by Glaxo-Smith-Kline tracking reported defects (incident reporting), especially in bupropion use for all purposes, finds the following incidences (selected data):

All congenital defects:
Bupropion: 3.4%
All other ADs: 2.5%

1st trimester use (smaller cohorts), all defects:
Paroxetine: 4%
Citralopram: 3.2%
Trazodone: 5%
Venlafaxine: 1.9%
Sertraline: 1.4%

Bupropion is associated with a 1.9% incidence of cardiac defects (CDC: 1.2%).

A recent study on the Quebec pregnancy registry also reports that 7% of infants in 1403 women taking paroxetine had congenital defects, and 1.7% had cardiac defects.

CONCLUSIONS:

• SSRIs (except for paroxetine) are probably not associated with increased risk of birth defects, including cardiac defects. Neither is venlafaxine.
• Paroxetine and trazodone are associated with increased risk of birth defects.
  
• The Quebec study indicates that low doses of paroxetine are NOT associated with increased risk.
• Bupropion may slightly increase cardiac defects; however, the effect is small and therefore not certain.