Mass Screenings for Low-Prevalence Problems in Pregnancy
Mass Screenings for Low-Prevalence Problems in Pregnancy
Gestational diabetes screening illustrates problems with this dangerous structure
The dangerous structure has come for my unborn child.
Everybody has to let their guard down sometime. Sometimes you run out of attention, energy, or time. Sometimes you can’t borrow or bootstrap the right expertise to do a proper job evaluating the evidence yourself. And this is when it gets you: Mass screenings for low-prevalence problems can cause massive societal damages. In healthcare, they do this one patient at a time. In prenatal care, they do it one mother and (usually) one baby at a time.
One of the ways they do it is by investing finite resources in routinized care — ticking boxes, chasing metrics, making and sorting huge numbers of false positives — instead of putting more of those same resources into targeted investigations in the course of patient-centered care. It’s the same problem structurally and conceptually as in intelligence failures in security, and efforts to address quality problems in science with unvalidated heuristics like “trust markers”: we need to use the information that we already have better, not expend finite resources trawling for trash.
This is a post about how one of my perennial obsessions (this dangerous structure) manifests in prenatal care, to which I am currently subjected as an increasingly pregnant lady. Previously, I’ve written and spoken about how mass screenings of all pregnant women for HIV and hepatitis even in low-prevalence settings is a great idea. These screenings are examples of this potentially dangerous type of program that actually work well. They succeed because they don’t produce huge numbers of false positives; secondary screenings can disambiguate true from false positives; and those follow-up screenings don’t (in well-resourced settings) cause substantial harm, including by taking finite resources away from something else. Rather, they accrue net benefits to the same people who pay the costs of the screenings, improving maternal and child health.
There are also some prenatal screenings that don’t work as well on these terms, but that are increasingly mandatory even where they are nominally voluntary. Anatomy scans (in-depth ultrasound screenings) for birth defects, gestational diabetes screenings, and gestational weight gain monitoring are arguably all examples.
Gestational diabetes screening is the worst offender. It creates huge numbers of false positives that are difficult to disambiguate, and diagnosis may net harm pregnant women without benefitting them or their kids. At the same time, it also risks taking finite resources (like attention) away from more targeted screenings.
The broad theory here is that uncertainty aversion links overdiagnosis and underdiagnosis. Increasing quantification tracks with increased bureaucratic demands for proof and certainty. In this context, mass screenings for low-prevalence problems make it easier for busy clinicians and other experts (and precarious researchers) to run and act on the results of defensible, standardized screening tests — than they do for the same groups to review complex patient histories, crime/intelligence reports, or scientific literature, and spend time thinking about, discussing, and experimenting to solve the implicated puzzles (see, e.g., slide 27 of my Dec. 2023 CCC talk). My own experience shows how this can play out in prenatal care, with gestational diabetes screening I didn’t want or (I think) need taking finite resources from tailored medical care that might have prevented harm to my unborn child.
The overdiagnosis here is in pressure to get screening after screening (after screening) for gestational diabetes, despite its lack of proven benefit and possible risks. The underdiagnosis is in missed monitoring for thyroid problems and anemia, problems which pose substantial possible risks, and are more readily diagnosable and treatable than GD. My experiences highlight the problematic focus on metrics over health that characterizes contemporary medicine. I also critique the surveillance culture driving these mass screenings, and correct some misinterpretations in recent medical literature along the way. Through a review of evidence on preventive health measures like the Mediterranean diet plus exercise, particularly for metabolic concerns such as gestational diabetes, I advocate for more meaningful, preventive care instead of overdiagnosis — and, at the same time, more targeted screenings combatting underdiagnosis.
Missed Targets: How Mass Screenings Distract from Individualized Care and Harm Pregnant Women
In May, in early pregnancy, I asked a gynecologist’s office to check my thyroid and iron. I didn’t do it because I had a problem, or even because I was planning on being good about getting regular monitoring. I did it because I got hit by a bike, the medic told me I had to go see my gyn as soon as possible, and I figured… Well, something more or less literally hit me out of the blue, and I might as well go with it. Some people read horoscopes. Apparently, I believe in bikes and bossy medics.
Everything was fine. Except the gyn ordered gestational diabetes screenings without my knowledge or consent. There are three sorts of ways to do them: glucose tolerance tests (which create huge numbers of false positives but are the standard because they are easiest and minimize false negatives), fasting and post-prandial blood sugar testing (which produces far fewer false positives, but is more invasive and resource-intensive), and HBA1C screenings (which lack sufficient evidentiary basis for diagnosing GD).
The gyn ordered a few flavors of HBA1C screenings, which also turned up no problems. They just made me mad. And curious. Why did she do it? Was there an insurance reimbursement per screening test? Having no shame, I asked.
Prenatal care services in Germany, she said, are reimbursed with a flat quarterly fee that includes all examinations (except for specific indicated sonographies), regardless of how often a pregnant woman comes in. There was no profit motive. She just wanted to benefit her patients by screening all pregnant women for early or pre-existing metabolic disorders. She argued this enabled initiation of early endocrinological management and follow-up, significantly improving outcomes for both mother and child. And that it left the door open to discussing individual monitoring together and allowing women who desire out-of-hospital births and midwife-led care to have that option.
These claims rely on a number of assumptions that have insufficient evidentiary basis. As we’ll see, it’s not entirely clear what the management of gestational diabetes should look like. Nor that casting an ever-wider net for its diagnosis net benefits moms or babies. As for valuing discussion and patient choice — conducting nonstandard medical tests on pregnant women without their knowledge doesn’t exactly scream informed consent.
A More Unfortunate Discovery
Pregnancy in the modern Western world means a lot of routinized surveillance — most of which I strive to avoid because it’s without sufficient evidentiary basis, can cause iatrogenic harm (e.g., overdiagnosis), stresses me out, and takes precious time and energy. So maybe it is my fault that I didn’t get checked again for thyroid problems and anemia until last week, when my midwife drew blood for the fasting blood sugar screening (more gestational diabetes testing) — and then asked what else I would like checked.
I immediately suggested these screenings, since I’m high-risk for these problems on the basis of personal and family history, and have had nebulous symptoms that could be from pregnancy, other stuff, or these things. Both tests came back abnormal.
My fasting blood sugar was great at 87 — the lower bound of the prediabetic range is 100. But the midwife still wants further testing. And she thought the other values were nothing to worry about.
I thought differently.
Untreated low thyroid in early pregnancy causes permanent neurodevelopmental damage (1, 2, 3); the effect size appears to be large for severe hypothyroidism and small, perhaps on the order of four IQ points, for mild cases like mine. In early life, neurodevelopmental harm from even mild and transient prenatal hypothyroidism can manifest as a language or cognitive developmental delay before attenuating — but not before altering gene expression, impairing fetal brain development. It’s also associated with large for gestational age kids.
Sadly, this could explain some things:
After I moved back from Leipzig to Berlin, my gyn did an ultrasound and misdated the pregnancy by 10 days according to my calculations.
Everyone from my mom to a stranger in the park commented on the size of my belly as soon as I started showing, worrying or joking it must be twins.
A more expert sonographer corrected the due date by 7 days, but noted a small cerebellum she wanted to watch.
When I took my son swimming over the summer, the water ticked off my Raynaud’s (aka creepy fingers) — an excruciatingly painful vasoconstrictive form of cold intolerance.
I love being pregnant. Colors and smells are more intense, food never tastes better than when you’re eating what you crave, and you get to enjoy this once-in-a-lifetime intimacy with another person — someone you take care of by taking care of yourself. But I was wondering why something had changed. I was no longer hungry every three hours, food didn’t taste as amazing, everything in life was a bit harder, and (I was just noticing) I didn’t get the supermodel hair and skin this time around, either.
There is now no way to know when my thyroid level dropped — somewhere between the ends of May and September — and no way to undo any damage that was done. Both times, I asked it be checked because I have multiple risk factors, and the symptoms are nebulous ones. Things like fatigue, mental fogginess, or thinking your stomach looks too big. The requested screening was a targeted one based on specific information. Any damage done was fully preventable.
But everyone was busy doing, thinking about doing, and talking about doing mass screenings for low-prevalence problems, instead. With HIV and hepatitis out of the way, it was time for the anatomy scan. With that done, it was time for the gestational diabetes (GD) screening that is technically voluntary in Germany, but seems to be de facto mandatory, reflecting recently increasing pressure on healthcare providers to do it. (I was able to refuse it when I was pregnant with my son four years ago; that doesn’t seem to be possible this time.)
On one hand, getting blood drawn for the GD screening was the occasion for my overdue thyroid and iron checks. People sometimes make this argument for routinized medical care — that it gets people into contact more with healthcare providers, and then they incidentally get better care for other problems.
On the other hand, the evidence doesn’t establish a benefit of either the screening or the treatments that might result from a GD diagnosis. While targeted screening, diagnosis, and treatment of high-risk groups for treatable problems like thyroid and vitamin deficiencies improve outcomes by preventing damage, mass screening for GD may do net harm by making pregnant women feel sick, anxious, and surveilled — without helping them or their kids actually be healthier.
So why do experts seem to often operate under this misconception that mass screenings are better than targeted ones? Why is it so hard to find healthcare providers who listen to me? And what’s with my sense that these problems are intimately intertwined?
Doctor Defensive and the Dreary Dismissal of Women’s Health Problems
When I asked my regular doctor for help with the low thyroid and anemia my bloodwork showed, I got a lecture on how my anemia — though clinically diagnosable on labs (low hemoglobin, low hematocrit) — was not that bad. He would not, he said repeatedly without being asked, help me treat it; it was out of the question. He would not give me “an infusion or something like this.” Something I hadn’t asked for and wouldn’t want. Though it may (or may not) warrant treatment: anemia in pregnancy carries series risks including preterm birth and small for gestational age babies.
I thought of the sign in the waiting room disparaging Googling — “Nachdenken ist wie Googeln, nur krasser” (“reflecting is like Googling, just way more hardcore”). And refrained from reminding the good doctor that he had seen me the previous week for that worsening Raynaud’s, a form of cold intolerance — a symptom of both hypothyroidism and anemia. I refrained, too, from pointing out that my anemia didn’t actually correlate with low iron stores (my ferritin was pretty good — I already take a few iron supplements); but, rather, with low thyroid.
“I’m hypothyroid,” I interjected, simply. “I need medication to protect the baby.”
“I’m the doctor,” he said. “You’re not hypothyroid.”
“Here is the reference range,” I said, pointing at the highlighting on the printed bloodwork results I had brought in. “And here is my value.” It fell below.
He prescribed thyroid medication, informing me it was very important for the baby. I thanked him profusely and left. When I wrote later asking for a script for vitamins so I could get insurance reimbursement, he refused.
Everyone in Berlin is looking for a doctor. General practitioners here, as elsewhere, are overworked and unhappy. It shows.
It’s not just in medicine. Even the most concrete, unambiguous empirical findings can somehow become interpretive matters of discretion when psychosocial dynamics including threat, emotion, and time pressure set the terms of the conversation. That’s how bias works.
There is no exit from these forces. They are not exceptional; this is not a bad apples problem, but rather a species-wide one. It runs throughout medicine, security, education, tech, and other professions. We are just stupid monkeys, and not the nicest ones.
Left to my own devices to reflect on (and Google) the anemia, I was reminded of one of the medical news tidbits I hadn’t had time or mind to cover yet…
Iron Deficiency Screening and Anemia Treatment in Pregnancy: The Jury Is Out?
In late August, the U.S. Preventive Services Task Force found insufficient evidence to recommend routine iron deficiency screening and iron deficiency anemia supplementation during pregnancy. Although the evidence they reviewed showed routine prenatal iron supplementation reduced iron deficiency and iron deficiency anemia, and was not associated with (statistically) significant maternal harms, they concluded “evidence on maternal and infant health outcomes is limited or indicates no benefit.”
On one hand, this is insane. As a number of experts immediately told Undark, iron-deficiency anemia is common among reproductive-aged women, affecting around 1/3 of reproductive-aged American women. It’s generally more common in pregnant women, affecting around 20% of pregnant women in Europe — and more than double that (45.2% by one estimate) in low and middle income countries.
Mass screening for something this common is not screening for a low-prevalence problem. The best reason to not do mass screening for low iron status in pregnant women is that maybe you just want to supplement everyone anyway. The worst reason is that evidence doesn’t establish benefit. As emeritus professor of clinical translation research at the University of Pittsburgh and retired hematologist Margaret Ragni told Undark, “ ‘Iron deficiency is associated with a really poor quality of life… These poor women really could feel so much better.’ ”
It’s not just that iron deficiency and the anemia it can eventually cause can severely impact functioning and quality of life, leading to cognitive problems, exercise intolerance, and depression. Or that we know exactly why in causal terms: iron is part of the blood protein hemoglobin (“heme” = iron, “globin” = transport) that carries oxygen to all the tissues in the body. This is why anemia can cause such a wide range of problems, from fatigue and forgetfulness, to shortness of breath. You literally can’t move, think, or catch your breath if your blood can’t carry oxygen where it needs to go, because it’s missing the essential protein for doing that job. (Hypothyroidism can cause the same symptoms for different reasons, weakening the respiratory muscles, decreasing lung function, and suppressing bone marrow production of red blood cells — one of multiple causal pathways possible in the noted covariance of anemia and thyroid disorders. This may explain why, after four days of thyroid medication, I could breathe easy again for the first time in months.)
Nor is it just that iron deficiency, anemia, and hypothyroidism can cause ambiguous symptoms easily misattributed to other factors, like needing more sleep, stress, or being pregnant (“baby brain”). In other words, maybe you are feeling weak because you are a reproducing female.
No, it’s not just the sexism or the disproportionate effects that sexism has on women’s functioning, validating stereotypes. If we are actually less capable because we are ill because doctors dismiss instead of diagnosing and treating problems which are more common in women, this creates empirical evidence of our lesser merits.
No: In pregnancy, the stakes are higher than whether you can get through the day, with anemia (like hypothyroidism) associated with adverse outcomes like preterm birth. And here we come again to a bittersweet intersection of statistical methods and social power…
The U.S. PSTF iron review arguably came to the right conclusion for the wrong reason: There is, many experts argue, sufficient evidence linking iron deficiency with poor quality of life and treating it with improvements, to prioritize doing that for patients including pregnant women. (This is a matter of interpretation.) But there remain troubling questions about the safety of routine iron deficiency anemia supplementation during pregnancy…
Statistical Significance Testing Misuse in the U.S. PSTF Iron Review
As is common in current medical literature, the U.S. Preventive Services Task Force misinterpreted statistical significance test results in its August review of the evidence on iron deficiency screening and anemia supplementation during pregnancy. In sum, they dismissed the evidence as limited or indicating no benefit for iron supplementation on hypertensive disorders of pregnancy (95% CI .75-2.06), C-section (95% CI .90–1.14), preterm birth (95% CI .81–1.04), infant low birth weight (95% CI .79–1.14), and small for gestational age (95% CI .67–1.31). But these compatability/confidence intervals really contain a great deal of uncertainty. For one thing, they suggest a substantial possible decrease in preterm birth risk of up to 19%.
There are other serious methodological problems with the review, like its reliance on a sum total of one trial that used a low dose of iron (20 mg daily) and didn’t find a quality of life effect. Studying women’s quality of life is not a medical research priority. But if it were, there’s already enough evidence that raising iron status improves it, that it arguably wouldn’t be ethical to run more trials like this.
At the same time, one could argue that we don’t have enough evidence on the net costs and benefits of iron supplementation in pregnancy to recommend it for all women. The available data suggests that maybe screening for deficiency should be universal, but supplementation should be tailored on the basis of women’s other risks. In particular, there appears to be a chance that supplementation substantially increases risks of hypertensive disorders of pregnancy and hemorrhage.
For hypertensive disorders, iron supplementation may decrease risk by up to 25% — or increase it up to 106%. Similarly, one reviewed study, from Ireland, found it may decrease antepartum (prebirth) hemorrhage risk by up to 78% — or increase it by up to 612% (95% CI .22-7.12). Another reviewed study, from Iran, found it may decrease postpartum hemorrhage risk by up to 85% — or increase it up to 635% (95% CI .15-7.35).
The possible hemorrhage effects are particularly large — in both directions. Though these wide confidence intervals skew right (toward a positive association), one might think it’s more plausible that iron supplementation decreases than increases hemorrhage risk: hemoglobin makes blood thicker, increasing its viscosity. For the same reason, one might also be wary of the possibility that treating anemia by improving iron status could increase clot risks.
However, the literature suggests cause for concern about the opposite problem: reactive thrombocytosis (too many platelets) in iron-deficiency anemia itself increases thrombosis risk. So maybe supplementing iron deficiency in anemic pregnancy women increases hemorrhage risk by treating that reactive thrombosis.
Overall, these wide compatability/confidence intervals reflect uncertainty about whether there is an effect at all. They are compatible with substantial possible benefits as well as substantial possible risks. We should really know more about the effects of iron supplementation in pregnancy. In the meantime, it arguably makes sense to screen pregnant women for iron deficiency anemia, and suggest supplementing to women who are deficient and symptomatic — or discuss other treatments for women with different kinds of anemia. There could be no risks and substantial benefits. There could be real risks along with real benefits. We could tell people what the evidence says and let them choose, though doing that is always already fraught with interpretation.
The larger point here lies in the contrast: We get some universal or near-universal mass screenings for low-prevalence problems (HIV, hepatitis, anatomy scans for birth defects, gestational diabetes) and not universal screenings or ongoing monitoring for problems that are just as prevalent or more prevalent (prenatal iron status, anemia, and thyroid). Why?
The Politics of Screening: Policing Women's Health Instead of Promoting Well-being
I sympathize with experts who dislike having their authority questioned: When it comes to my body, I, too, consider myself the expert, and dislike having my authority questioned. And I think other people are the experts on their own bodies, too.
This is one of the reasons why mass screening for gestational diabetes really bothers me: False positives are common, affecting 57% of women diagnosed in accordance with new guidelines, according to one recent analysis. And result in pregnant women being subjected to a lot of onerous surveillance and intervention for no proven benefit. This surveillance typically involves regular blood sugar monitoring, keeping a food diary, and talking to experts to see what they think about what you’re eating. This looks very much like a regime of social control dressed up as science. Especially when we contrast thyroid and anemia screenings with GD ones.
Pregnant women in countries where HIV and hepatitis prevalence are extremely low are universally screened for them, but not for thyroid and iron deficiency problems that are relatively common. This despite the fact that some researchers have long argued that thyroid screening in pregnancy should be universal. The benefits of treating subclinical or frank hypothyroidism — including fewer miscarriages and preterm births, and higher IQ offspring — clearly outweigh the costs of screening in well-resourced contexts.
But most guidelines call for risk-based (targeted) screening instead. This is often misinterpreted clinically as screening once instead of routinely in high-risk groups — as if one test magically resolved all uncertainty around the risk. Such, after all, is the promise of screening tests: They’re supposed to neatly pack away our uncertainty (but don’t)…
I know multiple women in their 40s, when thyroid problems are most common, who suffered miscarriages potentially as a result of this practice. And others with autoimmunity, which also often tracks with thyroid problems, who were simply put on prophylactic thyroid medication; apparently some clinicians calculate it’s safer to take a small supplement to buffer against possible developing hypothyroidism when you’re high-risk, than to chance it costing a pregnancy between checks.
So why do rigorous, increasingly universal prenatal HIV, hepatitis, anatomy scan (birth defect), and gestational diabetes screening regimes contrast with these relatively lax and heterogeneous iron status, anemia, and thyroid screening practices? Might this have something to do with power?
Women presenting with vague symptoms like fatigue, poor focus, cold intolerance, poor sleep, exercise intolerance, and shortness of breath are often dismissed as having non-empirical medical problems. When we are unwell, classically, it’s in our heads: hysteria. Knowing this, we may not bother asking for help, or not with too many, seemingly unrelated problems.
But when we are too fat (or, rarely, too thin), we are reviled. Disciplined. And can’t hide.
To be a woman is to have to wear your reproductive choices on your sleeve — or in your maternity pants, as the case may be. Public surveillance of women’s bodies for conformity versus noncompliance with social norms is as old as social norms.
The disproportionate interest in screening for gestational diabetes when vastly more pregnant women actively suffer from low iron, anemia, and/or hypothyroidism — and more children in well-resourced settings are permanently harmed in utero by low maternal thyroid than high maternal blood sugar — is consistent with what the broader culture values about women: not quality of life or functioning, but compliance with patriarchal norms. Why medicalize well-being, when you can police what women eat?
Routinized Medicine: Box-ticking, Metrics, and the Loss of Patient-Centered Care
This is not to say that the people operating the healthcare system on the ground are doing GD screenings (or failing to do needed iron, anemia, and thyroid checks) because they are sexist. Probably the people who spend their professional lives doing prenatal care are doing it because they like women? Still, they can implement routinized, quantitative, professionalized process that unintentionally enact misogynistic assumptions and that have unintended consequences that hurt women — undermining precisely the values they want to uphold and hurting the people they want to help.
But basically these people are just busy, like everyone else. Busy, in this case, checking boxes and putting stickers in booklets, because that’s how the system works. In Germany, the maternity booklet is called, somewhat poetically, the Mutterpass — Mother Passport. I like to think this reflects a cultural understanding of pregnancy as a journey fraught with danger, on which routinizing care can help mitigate risks. And it can.
It maximizes the population-level benefits of the successful mass screenings it incorporates. For things like HIV and hepatitis (where secondary screening can sort true/false positives with minimal harms), this produces net benefits at a societal level. But for the mass screenings that generate overdiagnosis without proven benefits, like gestational diabetes and weight gain monitoring, it’s not clear that patients or society benefits. As in other contexts, evidentiary standards for these screenings could be improved to be more in line with accepted scientific evidentiary standards:
Proponents bear the burden of proof to establish that new interventions will do more good than harm.
Independent reviewers evaluate the evidence of claimed costs and benefits.
Relevant data must be public, including information about its production, storage, analysis, and interpretation.
Structure programs to yield better efficacy data.
Despite the absence of these standards, prenatal care among many other fields continues gravitating toward more mass screenings. This leaves fewer resources to pursue targeted ones — gearing healthcare system interactions toward satisfying bureaucratic ends (metrics) instead of listening and solving problems (health). This is not what healthcare providers are supposed to be doing. It’s probably not what they want to be doing. And it probably drives away some people who don’t want to be doing this.
Of Potato Soup and Power
When I was finishing college, I briefly shadowed a gerontologist in exchange for access to her research library. One of her patients, a relatively young African-American gentleman with type 2 diabetes, was shocked when his blood sugar tested sky-high. He had eaten potato soup for dinner the night before — in his mind, a vegetable. No one — including his doctor, a compassionate, knowledgeable, and popular physician — had taken the time to educate him on dietary and lifestyle choices that could help him manage his condition. Or even, as we now know, cure it.
The doctor still didn’t have time to do the missed education on more than just the potato soup and a maybe brief mention of glycemic index. The patient thanked her and went about his day. I decided medicine was perhaps not for me.
Critics often denounce this kind of failure as a consequence of perverse incentives. Health insurance companies pay busy doctors to get patients in and out, billing for discrete diagnoses (no uncertainty allowed) and treatments (take some insulin courtesy of Medicare; go look up nutrition science on your own). Pharmaceutical companies have financial interests in keeping people sick; you can’t patent a healthy lifestyle and make people pay monthly fees for it, for life, to survive.
When metrics become targets, as Goodhart’s law says, they cease to be good measures. They create perverse incentives, like for scientists to “publish or perish” — contributing to the cascade of mediocre or bad scientific literature instead of producing “less research, better research, and research done for the right reasons,” in Altman’s famous phrase. Or for hospitals to discharge new babies “exclusively breastfeeding” — causing increases in common and preventable harm from resultant accidental starvation including hospital readmissions for starvation complications and possibly permanent neurodevelopmental damage. Or for doctors to manage lifestyle diseases with billable tests and pharmaceuticals.
Metrics that have become targets like this produce something more than just money or power for dominant social and political networks: They produce apparent evidentiary bases for surveilling classes of people who might otherwise realize their larger aggregate interests in being supported rather than surveilled. Elevated blood sugar tests become the basis for more tests and ongoing monitoring. Common newborn harms resulting from common breastfeeding insufficiencies become the basis for more measurement and medicalization of infant feeding — requiring more instead of less expert assistance. Medical metrics identifying real problems become the basis for more medicalization of problems that might be said to be sociocultural in their origins — and, arguably, sociocultural in optimal solutions.
Still, it’s clear that people getting bigger and fatter is a sociocultural problem with medical implications. It shows up from gerontology to neonatology — with the same systemic neglect of preventive lifestyle interventions and patient education…
Houston, We Have Big Babies
“Babies are getting bigger. Pelvises aren’t keeping pace,” Belfast GP Kaitie Cairns recently Tweeted, summarizing Bonanni et al’s “Birthweights at term have increased globally: insights from a systematic review of 183 million births” (AJOG, Vol. 231, No. 4, p. 395-407, Oct. 2024).
Bonanni et al noted Kramer et al 2002 “identified maternal anthropometry changes” — specifically increased pre-pregnancy BMI and increased gestational weight gain — as well as “reduced cigarette smoking, and shifts in sociodemographic factors as significant factors contributing to this rise.” In addition, “Zaffarini and Mitteroecker underlined the impact of improving living conditions over time on the increase in fetal and adult body sizes.” These arguments aren’t mutually exclusive. What Kramer et al call BMI and gestational weight gain (metrics), Z&M narrate as a story unfolding over intergenerational time: “As the fetus is one generation ahead of the mother, the fetus is likely to experience better environmental conditions during development than the mother did, causing a disproportionately large fetus and an increased risk of obstructed labour.”
My family histories echo these arguments. My maternal grandmother could hold out her arm, and her mother could walk under it. On the other side, the first baby to be born in Florida to Romanian Jews who had fled starvation and violence in Iași had its skull crushed during an attempted instrumental delivery to save the mother’s life. This was an option of last resort before C-sections became widely available in the West — a 20th century development that mitigated the particular subset of the prisoner’s dilemma sometimes called the obstetrical dilemma, wherein big babies face off against little women in a battle for survival.
Probably, this particular infant was either LGA (large for gestational age) due to too much sugar/carbs in the New World diet and no knowledge of how to manage that new food availability, or there was simply too great a difference between food availability in America during its gestation and food availability in Romania during its mom’s sensitive developmental periods. These stories, too, are not mutually exclusive, as some researchers theorize that famine/feast exposure activate thrifty/wasteful genes to influence metabolism in ways that would be adaptive under original conditions — conditions that can drastically change. So early developmental scarcity could contribute to later metabolic disease in times of plenty (the Barker hypothesis), as mixed evidence from tragic natural experiments including the Dutch Hunger Winter suggests.
Bonanni et al’s review keys into conversations about gestational diabetes and weight gain, which in turn fit into larger conversations about metrics, surveillance, and social control. Surveillance culture and misaligned medical priorities, in turn, have potential adverse consequences for large numbers of people — people who may not understand or may even vehemently deny their collective interest in having a different regime. But without this larger political consciousness, we get more (exclusionary) surveillance instead of more (sweeping) support, in medicine as elsewhere. This serves powerful social and political interests while undermining equality and public health.
I am sorry, there is no short-form partisan takeaway to give you false hope that by attacking one side in a hyperpolarized discourse, we can solve this problem and grow two inches in two weeks (height) while losing thirty pounds in thirty days (except in case of pregnancy). But I can offer you, dear reader, the humble beginnings of a clickbait headline! And why it would be wrong.
Perverse incentives drive science translation on this topic just as they drive everything else. The audience for lengthier posts acknowledging unknowns and interrogating what they mean is a lot smaller than the audience for crap like this…
Top 10 Things to Eat During Pregnancy for Gestational Diabetes Prevention and Healthy Weight Gain!
broccoli and other cruciferous vegetables (but especially broccoli)
cranberry juice and other common berries
Stop clicking the links, you filthy clickbait reader you. They are just PubMed searches, and the evidence contained therein is generally weak and mixed. Drinking a lot of decaf coffee, for instance, may or may not increase miscarriage risk, as well as possibly improving metabolic parameters — or not.
Everything has risks. For example, eating a lot of cruciferous veg or other goitrogenic foods like soy could hurt your thyroid function, elevating risks of miscarriage and developmental problems including lower IQ in offspring. Especially in even moderately iodine-deficient areas like Germany and many other European countries. So much for the public health intervention risk aversion that kept fluoride out of Berlin’s water. Maybe they need more iodine in the food here.
In the big picture, one general problem with lists like this is, there’s such weak evidence for benefit or risk from such a wide variety of food items, often with implausibly large claimed effects in single studies, that we usually don’t know if there is really an effect, or not. But, as long as you don’t go overboard, this sort of thing probably isn’t a bad cheat sheet.
Another problem is that this kind of list focuses on diet to the exclusion of other things that seem like they might matter, like exercise, sleep, oxidative stress, perceived stress, and nocturnal light exposure. Yes, light exposure at night appears to increase gestational diabetes risk (1, 2, 3). And this is not sui generi: Shift workers, for example, also have notoriously heightened metabolic and cardiovascular along with sleep problem risks.
One of the possible mechanisms that might spring to mind for such an effect is the disruption of melatonin that nocturnal light exposure causes. Another is the disruption of Circadian rhythm/sleeping at night when you’re not on the night shift, as a result of that. Another has to do with further health degradations as a result of that.
These cascading mechanisms all have some suggestive but inconclusive support in the literature. Blue-blocking glasses may help with sleep disorders associated with nocturnal light exposure. If less melatonin contributes to GD risk, then you’d expect more melatonin to mitigate it. If it also contributes to the pathology of GD in a mechanistic way, then you might expect supplemental melatonin to mitigate the risks of GD once it develops. We don’t have randomized trial evidence on whether or not it does or not — in humans. (The latter seems to shake out for rats.)
But melatonin seems to play an important role in reducing oxidative stress in the placenta, may help prevent problems from heat stress by promoting a healthy microbiome, and is proposed as an adjuvant preeclampsia treatment. Basically there is a lot of hype around how it seems to matter and might be good to supplement in pregnancy, but everyone is too afraid of experimenting on pregnant women to run trials on this — even though a ton of pregnant women (one estimate says 4%) are already taking melatonin anyway.
In the bigger picture, exercise, sleep, and lower oxidative stress are in that bin of things that seem good for all sorts of health outcomes including many pregnancy outcomes. But that we struggle to assess in causal terms. You can’t randomize people to be able to exercise or not; nor to have insomnia — or be so exhausted they need to sleep 12 hours a night — or not. So even randomized trials on this stuff are not broadly generalizable, and observational studies are confounded.
Similarly, when you’re pregnant, you have to eat what you can sometimes, constrained by nausea, convenience, money, what your threenager is having, etc. So it makes more sense to pick what you like from a broader diet that might help and probably can’t hurt, as opposed to stuff from a shorter list that you might not be willing or able to eat. It’s just not quite as legible in the contemporary parlance of popular science.
That hasn’t stopped researchers like nutritional psychiatry rockstar Felice Jacka from trying to empirically prove and popularly translate what we know about diet and health to help people improve their quality of life. This is not a criticism of Jacka et al, who do some of the most sciencey food science out there. It is just a nod to how hard it is to make progress on all these fronts. Because, when it comes to Jacka’s pet intervention — the Mediterranean diet — a number of researchers still argue the jury is out on whether it moves metrics where we want to move them for pregnant women…
A Recent Review and Meta-analysis on Mediterranean Diet and Gestational Diabetes Misses the Bottom Line
In particular, Waugh et al recently argued there’s not enough evidence to say whether introducing the Mediterranean diet in pregnancy reduces gestational diabetes incidence (“Introduction of the Mediterranean diet in pregnancy and the incidence of gestational diabetes mellitus: A systematic review of randomised controlled trials and meta-analysis,” Eur J Obstet Gynecol Reprod Biol, Aug. 2024, 299:199-207).
This article has been gnawing at me for months. In the way of rabbit holes, there is a lot wrong with it — one tunnel of error leads to another. To correct each tunnel properly would take months. For one thing, a ton of relevant literature is missing. For another, the article makes a wrong inference from the takeaway of that literature. But, as usual, it’s not just about one bad article: science and medicine in this entire area is contradictory, myth-based, and focused on the wrong things.
The bottom line is that, sometimes, incomplete information is enough. We know eating a healthier diet (like a low-glycemic index Mediterranean one) and moving more improves a ton of health outcomes. This is true before, during, and after pregnancy. It’s true with respect to gestational diabetes and many other metabolic problems. Lack of evidence is not the reason why education on this is not available to everyone all the time in societies with growing (pun intended) diabesity problems.
Perverse incentives drive atomization — thinking about the problem as individual instead of systemic or structural. This neoliberal way of taking both the specific (here, patient-centered) and the collective (sociopolitical context and consciousness) out of the thing, in turn, feeds medicalization of cultural, social, and political problems (e.g., lost knowledge about typically healthier traditional cuisines, lack of “leisure” time or typically unpaid, female labor for shopping and cooking, and lack of walkable, bikeable cities and/or work-life rhythms that otherwise involve regular movement).
On one hand, individual lifestyle interventions still operate on this level, as opposed to addressing, e.g., food company corruption or zoning problems. On the other hand, they’re a lot less profitable for big corporations and seem to have better possible risk/benefit spreads than most medical model alternatives.
Complexities in Defining Some Terms — and Treating Some Problems
What are we talking about? The Mediterranean diet is one healthy, traditional, plant-based diet among many. But it’s one we happen to have particularly good evidence on. For instance, Jacka’s SMILES randomized controlled trial on a modified Mediterranean diet showed three months of dietary intervention versus social support resulted in major depression remission of 1/3 versus 8%. This is a massive effect size representing a possible benefit that comes without apparent risk of harm.
Gestational diabetes is what it sounds like, high blood sugar that develops during pregnancy and usually goes away after delivery. It is, unsurprisingly, a relatively common pregnancy complication in places with bad (modern Western processed) diets and population-level diabesity problems. By some estimates, GD affected around 8% of pregnant women in the U.S. in 2021, about the same in Europe, and around 15% globally — with a notably higher incidence in the Middle East and North Africa (around 27%). Advanced maternal age and pre-existing metabolic problems (e.g., overweight, obesity, PCOS) are risk factors. It carries risks to moms and babies, from the lethal (postpartum cardiovascular and metabolic risks; stillbirth) to the merely painful (large for gestational age/LGA infants requiring C-section births; big babies suffering shoulder dystocia or breaking bones during vaginal births).
But measurement debates characterize both ends of this process. At the product end, as it were, there are different standards for fetal growth. An even larger source of heterogeneity in data on gestational diabetes comes from the process end: There have been ongoing debates for some time about what test and threshold make most sense for diagnosing GD. For instance, women in Australia are subjected to less accurate testing and a lower diagnostic threshold than women in the UK and many other places — with disputed net effects.
Uncertain GD Treatment Risks and Benefits
Increasingly common, more conservative GD diagnosis and treatment regimes are stressful for the women affected, who are told they must conduct extensive blood sugar self-testing, attend additional appointments (which can cause things like heat exposure and loss of sleep that can adversely affect pregnancies — including by increasing blood sugar and gestational diabetes risk), and have their pregnancies reclassified as high-risk. This designation often changes things like who can provide what care (e.g., home birth may no longer be possible) and how their babies are treated (e.g., newborn separations for glucose monitoring). In addition, medical management of gestational diabetes includes a fair bit of relatively new terrain. This should make you nervous.
Results of the recent PERMIT randomized trial, for instance, proved no benefit from usual care for diabetic women in labor — under which nearly 1 in 6 women required an insulin drip (versus 0 in the permissive care group). The authors report no difference in maternal or neonatal outcomes, but a substantial possible difference in neonatal hypoglycemia rates — 25% in the usual care group versus 29% in permissive care (95% CI .60-2.17). (Medical management of neonatal hypoglycemia is itself contested terrain, with some researchers arguing for prophylactic dextrose gel and others pointing to uncertainty about its long-term neurodevelopmental effects.)
To take another example, metformin is a standard diabetes medication that happens to lower testosterone — which plays crucial roles in fetal development in terms of both growth and differentiation, and may affect cognition and IQ. That this is perhaps not a good idea is tentatively born out by research showing evidence of possible harm in metformin-exposed kids. When moms take it in pregnancy, it may increase the risk of SGA (small for gestational age) newborns (which implies other serious risks) — and raise long-term BMI z-scores among exposed offspring (a possible proxy for metabolic health). In addition, “Preconception paternal metformin treatment is associated with major birth defects, particularly genital birth defects in boys,” although the mechanism is unknown.
So at least one subgroup of cases of gestational diabetes, diagnosed by the one-step glucose tolerance test that makes more false positives than alternative testing regimes, and subsequently treated with metformin, looks like a possible case study of the common sort of overdiagnosis wherein a lower threshold for a lifestyle disease generates more diagnoses and so more treatment, including unnecessary treatment, causing possible harms for no proven net benefits. Other examples of this pattern include diseases like hypertension, diabetes, high cholesterol, and osteoporosis, as explored in critiques like Welch et al’s Overdiagnosed.
Instead of leaning on the precautionary principle or comparing these unknown effects to the known positive ones of relatively well-established and apparently risk-free interventions like the Mediterranean diet plus exercise, the usual spin is that the contemporary drug interventions don’t provably do harm even though there are plausible mechanisms for it. Or that we don’t know, so it seems fine. Or that the harm they may do is surely offset by the benefit (assuming one particular comparator among many possibilities).
For instance, one recent article noted we see some possible harms, but concluded “Nevertheless, metformin’s low cost, ease of administration, and global reach make it a reasonable intervention in a population affected by rising rates of obesity and diabetes in pregnancy.” This is the preferred narrative of powerful social and political networks including experts favoring medicalizing surveillance over lifestyle preventive regimes, and drug companies with financial interests in that approach.
Similarly Uncertain GLP-1 Agonist Exposure Risks
One might expect GLP-1 agonists (Ozempic etc.) to be another such case study, but (1) it doesn’t look like it’s being used (yet) for GD prevention or treatment, and (2) it’s not as clear from the evidence on prenatal exposure (in groups with metabolic problems putting them at risk of GD) so far what’s going on here. However, there a lot of evidence on how these relatively new drug interventions on pregnant women seem to improve their health, though who knows what they do long-term to their kids. Is this volume a product of bias and corruption? Is it just an accidental consequence of fertility improvement from health improvement, creating a windfall of preconception-exposed babies who, according to many studies, seem fine?
Why choose? Some researchers find evidence of harm to exposed pregnancies (e.g., increased miscarriage and congenital malformation risks). Some don’t — but misinterpreted their results…
Take, most recently, Ilari Kuitunen’s review synthesizing results from two studies on GLP-1 agonists in early pregnancy. Kuitunen misinterprets statistical significance test results in both studies to indicate there is no effect, when there are substantial possible effects.
The first (Cesta et al 2024), a retrospective register study with N = 938 type 2 diabetic mom pregnancies from four Nordic countries, the U.S., and Israel, found possible increased and decreased major congenital and cardiac malformation risks. These malformations are rare, so the study was underpowered to detect these effects within a narrow band; we got wide confidence/compatability intervals instead. These wide intervals do not mean, as the review article title says, “Exposure to GLP-1 receptor agonists in early pregnancy did not increase the risk of congenital malformations.” This is typical statistical significance testing misuse of the type you can’t read a medical journal digest without spotting.
But they do mean, comparing GLP-1 receptor agonists with insulin, that there may have been an up to 28% risk reduction or up to 26% risk increase in major malformations (RR=.95, 95% CI .72-1.26), and an up to 58% risk reduction or up to 12% risk increase in cardiac malformations (RR=.68, 95% CI .42-1.12). And there may have been no effect at all.
Kuitunen could have emphasized the possible teratogenicity of insulin. But emphasized the relative safety of GLP-1 agonists instead — a safety which this evidence does not establish.
Still, this analysis is valuable because it raises difficult questions that are also relevant to risk assessment in GD treatment…
The Art of Risk Assessment
We should be asking what the best comparator is. Is it insulin therapy, metformin, or inositol for diabetes? Mediterranean diet plus exercise for improving metabolic health? No treatment for obesity? Something else?
This type of analytical choice shapes the observations and results. It reflects an interpretive process moored in sociocultural assumptions about what medicine is and does (pills, injections, blood tests) — or isn’t and doesn’t do (diet and exercise education and support, preventive interventions), as well as in sociopolitical context (pharmaceutical funding for careers, absence of analogous funding for lifestyle interventions; structural and social regard for diseases and treatments as real and foundational, versus the lifestyle luxuries of cooking and exercise that depend at least to some extent on knowledge and time). Such choices abound in medical research, and are rarely adequately justified.
The second study Kuitunen synthesized results from (Dao et al 2024) was another register study, this one involving teratological databases in Germany, Israel, Italy, Switzerland, and the UK. Comparing GLP-1 agonist users to women using other antidiabetes meds, they again estimated no certain effect — but found a quite substantial possible risk increase (CI .16-5.82). Comparing them to obese women, they again estimated no certain effect, but a substantial possible risk increase (95% CI .11-2.75).
Again, Kuitunen misinterpreted these results, writing “This review identified two register-based cohort studies that did not indicate an increased risk of congenital malformations following exposure to GLP-1 receptor agonists in the first trimester.” This evidence doesn’t establish that. The intervals estimated here don’t establish risk or benefit. Both are possible.
Dao et al themselves argue:
In the GLP1-RA group, three major birth defects comprising one congenital heart defect, one congenital anomaly of the kidney and one case of multiple anomalies were observed, suggesting that these anomalies appear to be aetiologically distinct.
But one could just as easily argue that any one anomaly might be so rare that it could take a larger sample to accurately estimate a drug’s risk of causing that particular birth defect. Instead they conclude “This study offers further reassurance in cases of inadvertent exposure to GLP1-RA during the first trimester of pregnancy.”
On one hand, maybe that reassurance is needed. After all, GLP-1 agonist use is increasing rapidly, and it looks like a lot of women who got pregnant while using these drugs had abortions (18% versus 6% of the diabetes reference group and 8% of the overweight/obese reference group). Maybe some of them were scared of birth defects. Maybe they needn’t have been.
On the other hand, there’s a lot of uncertainty in the available evidence, which is itself limited to short-term outcomes. As with metformin, there’s cause for concern that in utero GLP-1 agonist exposure could inhibit fetal growth — or not. We don’t know.
As Scott Alexander recently wrote, Ozempic could cure all diseases. This could include treating or preventing fertility, pregnancy/birth complications, and future offspring risks otherwise associated with maternal metabolic disease. These benefits could loom even larger compared to alternatives like insulin or metformin treatment for gestational diabetes.
Or these apparent effects could result from confounding, and we could learn in a few decades that it actually carries long-term risks to kids exposed in utero. We just don’t know.
You know what we know is safe for moms and babies long-term? A healthy diet and exercise. And, for overweight/obese women, eating less while moving more.
Loads of Ignored Evidence
The literature Waugh et al claimed to review deals with the effect of the Mediterranean diet in pregnancy on gestational diabetes risk. There are vast swaths of evidence on the Mediterranean diet. It’s recommended by cardiologists, endocrinologists, neurologists, and psychiatrists for its relatively well-established health benefits. It would be surprising if the Mediterranean diet helped everyone but pregnant women with myriad cardiovascular, metabolic, and inflammation-linked risks except gestational diabetes.
Fine, Waugh et al don’t claim to review this whole literature. Just RCTs on Mediterranean diet in pregnancy. They say they found one trial that randomized women to a Mediterranean diet treatment group, or a standard UK diet. Then they say they threw in two more trials that randomized women to either Mediterranean diet as the control, or that plus extra extra-virgin olive oil (aka olive oil or EVOO) and pistachios as the treatment (p. 4, first two paragraphs).
After defining their focus so narrowly as to justify ignoring huge swaths of evidence, Waugh et al then misinterpreted what two out of three of these trials did. Two of them (St. Carlos and ESTEEM) pitted the Mediterranean diet against a standard diet or usual care. Only one (Zhao et al’s) compared the Mediterranean diet supplemented with EVOO and pistachios with the Mediterranean diet. But St. Carlos also supplemented their Mediterranean diet group with EVOO and pistachios. So there are really three different operationalizations in these three trials of randomizing pregnant women to the Mediterranean diet to see if it prevents GD.
Trials cost a lot of time, effort, and money to run. There is also, at least in theory, a higher bar for ethical treatment of pregnant women in human subjects research, because they (and their offspring) are particularly vulnerable. The fact that some researchers decided to use the Mediterranean diet as a control group should at least tell you something about the strength of belief in its benefits, and maybe about the strength of the evidence that it could help and couldn’t hurt.
Bracketing their operationalization misinterpretations, let’s just look at how Waugh et al interpreted their results. They reported the Mediterranean diet versus olive oil and pistachios trials found substantial reduction in gestational diabetes, and so did the standard UK versus Mediterranean diet trial. How substantial? They report a 95% compatability/confidence interval of .57-.88 for the former, and .55-.90 for the latter. That is, the Mediterranean diet with EVOO and pistachios decreased GD risk by 12% to 43% compared with the Mediterranean diet. They also say “Of note the intervention groups in these two studies were also given the advice to walk for 30 minutes per day.” Compared with the standard UK diet, they say the Mediterranean diet decreased GD risk by 10% to 45%.
Yet, then they concluded by dismissing this evidence as “scarce,” stating “A large multi-centre randomised controlled trial is needed to definitively determine the impact of the Mediterranean diet in pregnancy on the incidence of gestational diabetes.”
It’s not clear to me why it would be ethical or otherwise worthwhile to randomize pregnant women to standard diets just to better prove what multiple randomized trials and a vast array of other evidence already suggests: that the Mediterranean diet plus exercise may improve people’s health quite a lot.
What These Three Studies Actually Did and Found
St. Carlos
St. Carlos researchers randomized 874 Spanish women to the Mediterranean diet with extra virgin olive oil and pistachio supplementation versus a standard but fat-limited diet. The intervention significantly reduced gestational diabetes (95% CI .56-.95), as well as gestational weight gain, need for insulin treatment, premature birth risk, emergency C-section, perineal trauma, and small and large for gestational age newborns. This reflects a possible decreased GD risk of 5-44%.
It sliced, it diced, it seemed to more or less translate clinically in a follow-up interventional study with 932 women who “received a motivational lifestyle interview with emphasis on daily consumption of EVOO and nuts” and wound up with a lower gestational diabetes rate than the trial’s control group (.73-.93) and lower risk of delivering LGA infants. Diabetics also showed lower weight gain and risk of excessive weight gain without significant increase in insufficient weight gain. The authors concluded “an early MedDiet nutritional intervention reduces GDM incidence and maternal-foetal adverse outcomes and should be universally applied as 1st line therapy.” This really means education on the Mediterranean diet should be offered to pregnant women early and often, since the therapy here is really prevention (preventing gestational diabetes and weight gain problems).
ESTEEM
ESTEEM researchers “randomised [1,252] inner-city pregnant women with metabolic risk factors (obesity, chronic hypertension, or hypertriglyceridaemia)” to Mediterranean diet versus usual care (i.e., a horrible UK diet). Their randomized subjects, drawn from 4 London and 1 Birmingham maternity clinic, were 69% obese. (The UK obesity rate is around 26%.) They found the treatment reduced gestational diabetes (95% CI .47-.91) and gestational weight gain (95% CI -2.2 to -.2), but didn’t improve other maternal/child outcomes. This reflects a possible GD risk reduction of 9% to 53%, and a possible gestational weight gain reduction of .2-2.2 kg (a great example of a result that is statistically but not practically significant).
Possibly the diet didn’t reduce complications because the subject pool was selected to be so unhealthy to begin with (see, e.g., that obesity rate more than 2x the national one), and/or because caloric restriction in overweight/obese subgroups might be necessary to improve some outcomes. Despite this peculiarity of their sample, in a meta-analysis pooling their findings with similar trials, they found similar results with no heterogeneity.
Zhao et al
Zhao et al arguably ran the only ethical trial in this group, randomizing 560 pregnant women to either the Mediterranean diet as the control group, or the Mediterranean diet plus EVOO and pistachios. (It was also the most recent trial included, perhaps reflecting growing evidence and/or consensus.) Giving women this supplement paid off with a statistically significant albeit clinically small GD reduction, with GD “diagnosed in 51 (20.4%) women in the control group and 34 (13.6%) women in the IG.” (Sorry, no easy access to full-text means no 95% CI interpretation here.)
A Broader View of the Literature
Zooming out to the literature on “Mediterranean diet pregnancy,” the Cochrane Library returns 101 trial results. Narrowing the search to “Mediterranean diet gestational diabetes,” the CL returns 36 trial results.
Other meta-analysts still wound up looking at single-digit RCTs. For instance, Liu et al looked at five such trials — and found substantial effects for (e.g.) reducing GD (95% CI .51-.79) and gestational weight gain (-1.21 to -.28)), but may or may not have affected neonatal unit admissions (95% CI .47-1.20). (This last result suggests the intervention may have increased NICU admissions up to 20% — or decreased them up to 53%.)
This limited selection is still more than Waugh et al’s three trials. Maybe their focus on these three trials resulted from legitimate search method differences or analytical choices. I don’t know.
While Griffith et al’s Cochrane review of reviews and Waugh et al’s most recent meta-analysis come down on the side of needing more evidence, other recent meta-analysis authors looked at the literature and concluded (like me) that we know enough already. Liu et al were among them. Similarly, Zhang et al found in a 2022 meta-analysis of four randomized trials that the Mediterranean diet was associated with an 18-48% decreased GD risk (95% CI .52-.82) and lower gestational weight gain (95% CI -.26 to -.05). But — and here they used the same imprecise language as Liu et al — that it “had no obvious effect on” health outcomes we really care about (referring to non-statistically significant results).
The compatability/confidence intervals around preeclampsia, preterm birth, and NICU admission in Zhang et al’s meta-analysis were too wide to say whether there was an effect or which way it went. It may have reduced preeclampsia risk by up to 48%, increased it by up to 111%, or had no effect. It may have decreased preterm birth risk by up to 80%, increased it by 55%, or had no effect. And it may have decreased NICU admissions by up to 57%, increased them by up to 19%, or had no effect.
On the skeptic side, Griffith et al’s Cochrane review is consistent with Waugh et al’s read. Looking at a small subset of the literature, they concluded there’s not enough evidence to say the Mediterranean diet works for gestational diabetes prevention.
But what if the incomplete information we already have is enough? Let’s say what we really want is to improve health, or pregnancy outcomes — not just the number of women who could be diagnosed with GD in mass screening. Other researchers judge there is quite enough evidence to suggest diet and exercise interventions help with that.
When Incomplete Information Is Enough
The Mediterranean diet looks like a possible panacea for both “too low” and “too high” pregnancy weight gains: It helps prevent both small for gestational age (a feared possible consequence of too little pregnancy weight gain), and LGA babies. More pre-pregnancy adherence to the Mediterranean diet is also associated with lower risks of both gaining too much (or too little) baby weight — and keeping it on.
This isn’t just a matter of aesthetics or social control: too much or too little gestational weight gain may increase risks to moms and babies, including preterm birth. Unsurprisingly, too little weight gain correlates with increased risks of small for gestational age kids and decreased risks for LGA or macrosomia, while too much correlates with decreased risks of SGA and increased risks for LGA, macrosomia, and C-section.
Meanwhile, Teede et al’s 2021 review and meta-analysis found that evidence from 117 randomized trials (involving over 34,000 pregnancies) suggested:
antenatal diet and physical activity–based lifestyle interventions were associated with less gestational weight gain. Structured diet, physical activity, and diet with physical activity were all associated with improved maternal outcomes; only diet was associated with improved neonatal outcomes.
Arguably, if there’s a ton of suggestive evidence that the Mediterranean diet in pregnancy may help improve maternal and child health, and no evidence that it causes harm, then it would be unethical to run Waugh et al’s proposed trial. Unless…
Unless we take seriously the uncertainties surrounding possible associated harms. But there’s not a strong evidentiary basis for doing that. Googling for “mediterranean diet iatrogenesis” comes up with links like “Association of a Mediterranean Diet Pattern With Adverse Pregnancy Outcomes Among US Women” (reporting the diet is associated with 21% lower risk of developing any adverse pregnancy outcome), “Adherence to Mediterranean diet associated with health-related quality of life in children and adolescents: a systematic review” (finding better health-related quality of life is associated with the diet in kids), and “Short-term effects of Mediterranean diet on nutritional status in adults affected by Osteogenesis Imperfecta: a pilot study” (showing a restricted Mediterranean diet improved nutritional status and dietary quality in adults with the rare connective tissue disorder osteogenesis imperfecta). PubMed, Cochrane, and other medical database search results are similar…
But for this you want Google, in case there are Reddit and Quora pages with interesting anecdata, or grassroots patient advocacy groups alleging harm. There aren’t. There just don’t seem to be any indications that the Mediterranean diet can hurt people, whereas there seems to be a lot of evidence that it might help prevent and treat a lot of problems, from pregnancy complications and metabolic diseases, to depression and neurodegenerative disorders.
Ethical Requirement of Equipoise
This suggests that Waugh et al’s proposed trial seeking better evidence on the Mediterranean diet for preventing GD would violate the ethical requirement of equipoise. A standard component of human subjects research design, equipoise requires that there is uncertainty about which treatment group is likely to benefit the patient. The breadth and depth of evidence that the Mediterranean diet benefits health is too vast for such a trial to meet this requirement. We may not know exactly what effect size (if any) the Mediterranean diet has on likelihood of GD diagnosis. But we know enough to know it likely improves health and doesn’t hurt.
In Waugh et al’s words:
This systematic review and meta-analysis suggests that following the Mediterranean diet, in particular its EVOO and pistachio elements, in pregnancy may reduce the incidence of GDM. There may be other benefits to adopting the Mediterranean diet in pregnancy such as potential reduction in maternal weight gain and a potential reduction in the frequency of UTIs, emergency delivery, perineal trauma, preterm deliveries before 37 weeks, LGA >90th centile and SGA <10th centile.
So there is a possible substantial benefit and no apparent risks to counseling pregnant women to follow this diet. That makes it unethical to offer education on it to some participants while withholding it from others. The purpose of trials is not to firm up compatability/confidence intervals in the scientific literature. It’s to benefit people.
Zhao et al surmounted this problem by randomizing their control group to the Mediterranean diet, which should indeed already be standard of care, and their treatment group to supplemental EVOO and pistachios. There are lots of possible variations on this theme of doing for everyone what we have good enough evidence to suspect works without risking harm — while experimenting to see what works better.
After all, there’s plenty of uncertainty in what works to prevent and treat GD…
Legitimate Unknowns
According to Griffith et al’s 2020 overview of Cochrane reviews, “No interventions to prevent GDM in 11 systematic reviews were of clear benefit or harm.” This includes dietary interventions. But they claimed a combination of diet and exercise was of possible benefit, as were supplementation with myo-inositol, vitamin D, and metformin. In particular, diet plus exercise interventions possibly reduced GD risk by up to 29% — and possibly had no effect (95% CI: 0.71 to 1.01). Metformin for obese pregnant women possibly reduced GD risk up to 39% — and possibly increased it up to 19% (95% CI 0.61-1.19).
Griffith et al misinterpreted a few of the reviewed results, which indicated benefit (not possible benefit): The myo-inositol supplementation results (95% CI 0.29-0.64) estimated GD risk reduction of 36%-71%; the effect was statistically significant. Similarly, vitamin D supplementation reduced the risk of GDM by 3-73% (95% CI 0.27-0.97). Though in this case, it’s arguable that a 3% benefit is within the margin of error and could really mean no effect.
So it would make sense to counsel women at risk of developing GD to take myo-inositol and extra vitamin D, and eat a Mediterranean diet and exercise for prevention. Healthcare just isn’t very geared toward prevention. And there is a sea of medical literature to critically review to glean this takeaway.
It would also make sense to consider what you would do in the event of a GD diagnosis, before testing for it. Other mass screenings in pregnancy don’t have this problem: If a patient has HIV, hepatitis, or chlamydia, you treat it and protect the baby from transmission. If there’s not enough thyroid hormone, you supplement it. If there’s too much, you treat the hyperthyroidism, but don’t overtreat it — too little thyroid even at a subclinical level appears to be far more dangerous for fetuses than too much. What do you do for GD?
The standard and most evidence-based treatment is to monitor blood sugar, make a diet log, and talk about it. There are other options — inositol (which I already take), metformin (which I wouldn’t take because of possible risks to the baby and poor tolerability), insulin (which I wouldn’t use because of its risks and tolerability)… In short, none of them has a great evidence base for treating GD, and I wouldn’t do anything I’m not already doing if I were diagnosed with it.
Except — if I’m told I have to, in order to stay in my healthcare providers’ good graces — stick my fingers, log blood sugar four times daily, log what I eat, and talk to someone about it. Literally bleeding myself and justifying my dietary choices to someone who’s supposed to help me police them.
I don’t think I should have to suffer to meet someone else’s surveillance requirements to rule out a diagnosis of a disease that there’s nothing I would opt to do differently to treat, if I were diagnosed with it. This is the definition of a medical test that shouldn’t be run.
But pregnant women (apparently) don’t get to choose what tests and interventions we want and don’t want. At least sometimes, our healthcare providers make those choices for us; we merely try to influence them. When our preferences mismatch, we ultimately have to decide whether we want their help giving birth, or not. Or, in the case of many women over generations of nonconsensual birth interventions, whether we want to make a fuss or just move on after being etherized or gassed (Twilight birth), cut (episiotomy), or whatnot. In this context, a little gratuitous bleeding seems downright civilized.
Most non-human primates give birth unassisted. But most women need help. So we remain, by necessity, a highly cooperative species. Even when we would rather piss off.
Anyway, maybe I could learn more about the Mediterranean diet, do it better, and build back up to exercising more again once my thyroid level stabilizes and my anemia mends. What’s the deal with these extra-virgin olive oil and nut supplement studies? What are the mechanisms whereby these supplements might benefit health?
Grabbing Your Health By the Nuts
We always have to wonder about causality, and on this front there are some good reasons to suspect Mediterranean diets work — though there are also reasons to suspect many different versions of them might work just as well.
The first is polyphenols, compounds with strong antioxidant activity. Extra-virgin olive oil is the oil highest in them. Pecans, pistachios, and walnuts are the nuts highest in them.
Some researchers think high polyphenol intake from extra-virgin olive oil is what makes it improve insulin sensitivity. Others suspect other (not mutually exclusive) mechanisms like dietary fat quality are also implicated. They tend to single out two of its polyphenols as particularly powerful in this context, oleuropein and elenolic acid. But, as with that infamous red grape polyphenol resveratrol (which has anyway been debunked as an anti-aging panacea), it’s not clear that normal human consumption would get you an active dose of these things.
Another possibility is that it’s not so much that olive oil and nuts are great for you, as that they keep you from eating other crap that’s worse. Diets are zero-sum (if you eat one thing, you don’t have room for another), and so we have to seriously consider the possibility that simply being full on relatively healthy fats keeps people from going for the ice-cream and chips, because fats promote feeling full (satiety). Olive oil is a fat. Nuts are fatty. Pistachios are especially fatty. But, to the extent that they help promote health by making you too full to eat other crap, it wouldn’t matter what kind of nuts you ate. And there is indeed some evidence that peanuts (which are way cheaper, and yes, technically not nuts) are just as good for metabolic health as other nuts. (And hey, they also have resveratrol!)
Some researchers believe in pistachios, others favor walnuts, and both sorts of studies return the same sorts of results. For instance, Nakaki et al found Mediterranean diet and mindfulness-based stress reduction can both help prevent fetal stunting (small for gestational age, SGA), at least one mechanism appears to involve placental volume, and the diet group saw greater benefits than the stress reduction group. Their diet group got “monthly individual and group educational sessions, and free provision of extra-virgin olive oil and walnuts.”
So maybe there is something special about this oil and these nuts, or about oil and nuts (or nuts and soy, or nuts/soy/legumes) generally. Or maybe it’s just better for you to not eat the other crap you might eat, were you not feeling full from all these fats. One of the larger conversations here is about whether it’s really fat that makes people fat, or sugar that makes people have blood sugar problems.
American science writer Gary Taubes has perhaps most famously argued the latter case — an argument that might help explain consistently, curiously mixed findings on the health impact of high-fat dairy including ice-cream. Maybe eating an occasional ice-cream benefits metabolic health by keeping you from eating other crap. Or maybe there is just something to be said for eating what you want often enough to feel satisfied; maybe such satisfaction has its own momentum in revamping the willpower to make healthier choices most of the time.
So there are some possible causal explanations for Mediterranean diets’ apparent health benefits, even though there is plenty we don’t know.
Misguided Focus: When Surveillance Replaces Health Promotion in Prenatal Care
Why did Waugh et al draw a wrong inference from the existing evidence, which is suggestive that the Mediterranean diet (especially with supplemental extra-virgin olive oil and pistachios) can substantially lower gestational diabetes risks and otherwise improve the health of pregnant women and their children? Possibly they punted and said “more research is needed” because this conclusion was more novel than just agreeing with much of the literature (publication bias). Or maybe they really worried about needing more replications to establish effects, given the many uncertainties contained in any given estimate. Or maybe they just thought they were supposed to say that, because scientists are supposed to wag their fingers and shake their doubting heads.
They could be right. Griffith et al’s Cochrane review basically concluded the same thing. Several other recent meta-analysis authors and I interpret the underlying evidence differently. Maybe we’re wrong, or maybe it’s just a matter of interpretation.
One thing I don’t think is in dispute, though: The preponderance of the best available evidence suggests that pregnant women could have better lives and healthier pregnancies if they knew earlier how to live and eat healthier. They could be less surveilled by a medical apparatus that increasingly requires blood sugar monitoring and diet journaling, and more overjoyed at eating for two while this strange and wonderful pleasure — that you are, in the most literal, physical way, taking care of someone else by taking care of yourself — lasts all too briefly.
These narrower points of focus — on gestational diabetes classification instead of health, and pregnant women instead of people — reflect much broader social and political dynamics of surveillance culture. By classifying people into small groups and policing them, societal power structures are able to exert more dominance and control over them. This is what screening and treating pregnant women for gestational diabetes, instead of offering population-wide dietary intervention services to promote health, does.
I don’t know any pregnant women who have had to install continuous glucose monitors in their arms, who wouldn’t rather have been counseled on the Mediterranean diet before pregnancy or during early pregnancy. Yet, while there is a decent evidence base on the Mediterranean diet for gestational diabetes prevention, it has other health benefits for mother and child, and it’s fairly easy to implement — this knowledge is not often translated into clinical practice.
This likely results from completely benign causes, like busy healthcare practitioners prioritizing other problems and discussions. It shouldn’t necessarily fall on prenatal care specialists to do the entire work traditional cultures used to do on a population scale, of teaching people how to eat real food.
Some people argue that pregnant women should be subjected to more surveillance: It’s more cost-effective — two for the price of one. Society has a stake in protecting its young. And pregnant women already have more healthcare system contact anyway, so it’s possible to reach them in this critical window, with potentially lasting positive impacts.
These arguments neglect fairness. If we have a way of helping basically everyone be happier and healthier, then why aren’t we telling everybody about it? Why no Mediterranean diet intervention public health programs, perhaps available to all but incentivized for some?
Surely there are also possible arguments that other subgroups would save society even more money by taking better care of themselves and avoiding more medical costs (e.g., the elderly and homeless). But society should value everyone’s health equally. Fairness requires equality before the law and equal access to basic resources like roads, schools, justice, and food aid. Why would it not require the same sort of equality in access to health information and medical services?
Disobedient, Thick (and Skinny) Women and Our History of Necessity
It’s not just gestational diabetes. Research and clinical practice around pregnancy weight gain, too, is in a state of scientific and cultural disarray not atypical of health issues in complex modern societies. The guidelines vary by country.
They are widely disobeyed. Women, it seems, mostly don’t do what we’re supposed to do. Again, lack of clinical translation of weight targets by trimester and weight class is part of the problem. Doubtless, cultural beliefs about normal and appropriate behavior also play a role.
Japanese women don’t gain enough, and it’s supposedly hurting their babies — and their population. However, many of them start pregnancy underweight. It’s widely recognized that there are’t evidence-based guidelines for how much underweight women should gain in pregnancy. Probably not the best idea to add insufficient pregnancy weight gain to an underweight baseline.
On the other end of the spectrum, the majority of American women gain too much. The problem is global, with dual epidemics of malnutrition and diabesity in poor neighborhoods and countries worldwide. Giant corporations manufacture cheap crap specifically designed to taste great without being too filling — and that doesn’t give your body what it needs, so you eat more and get sicker. Some countries have better legislative frameworks for preventing particularly egregious abuses in this realm than others. Some countries have poorer populations with worse malnutrition problems than others.
UCL Professor of Anthropology and Paediatric Nutrition Jonathan Wells suggests these dual trends of malnutrition and obesity have created a new obstetrical dilemma of maternal stunting and overweight/obesity/gestational diabetes contributing to more obstructed labor and LGA babies. The original such dilemma stemmed from walking upright shrinking birth canals even while infant brain sizes increased — a widely contested theory. Wells argues for a competing hypothesis:
long-term ecological trends have likely played a role in changes in both pelvic dimensions and offspring brain size. One such trend was the rise of agriculture about 11,500 years ago in the Levant, which led to a shift from high-protein diet common among foragers to one replete with cereals. A high-carb diet is associated with both increased birth weight and shorter stature in the mother, and short stature is linked to smaller and flatter pelvises.
A low-glycemic index Mediterranean diet, in other words, would probably solve both old and new obstetrical dilemmas. The trouble is, you would ideally need to eat it well before pregnancy to prevent stunting, and possibly also to optimize fetal growth.
This line of thinking is consistent with the Maternal Nutritional Buffering model favored by researchers including Chris Kuzawa, John D. MacArthur Professor & Faculty Fellow at the Institute for Policy Research at Northwestern University. I suspect this Guardian gloss is not what anyone really said because there is just so much evidence on how the Mediterranean diet can help mitigate gestational diabetes risks like LGA, but for what the game of journalistic telephone is worth: A woman's diet in early life has more impact on her baby's birth weight than the food she eats as an adult, researchers say.
What the researchers actually said in their article’s lay translation was:
We propose that strategies to improve fetal macronutrient delivery will be most effective if they modify the pregnancy metabolism of mothers by targeting nutrition prior to conception and even during early development, as a complement to the conventional focus on bolstering macronutrient intake during pregnancy itself.
In other words, making a healthy diet a population-level priority may make more sense than targeting pregnant women with dietary interventions to improve pregnancy birth, maternal postpartum, and infant/child health outcomes. For one thing, this would hit future moms during their pre-pregnancy sensitive developmental periods (in utero, as infants/young children, and during puberty). This could prevent stunting and improve metabolic health, with intergenerational effects.
Of course, this doesn’t mean there’s no point in offering sound preventive medicine during pregnancy. Just that it might work better to offer it to everyone, all the time.
University of Rhode Island Anthropology Professor Holly Dunsworth proposed another competing hypothesis to the obstetrical dilemma, EGG theory, which says that when women hit their metabolic rate limits, the baby comes out (h/t Julia Rohrer). In other words, when you can’t eat enough to support your baby’s exponentially increasing energy needs growth, you go into labor. In this story, there is no obstetrical dilemma. The baby is evicted due to the mother’s metabolic constraints — not the head getting too big to go through the birth canal.
Wells noted this theory can’t explain the high and increasing frequency of obstructed labor with its attendant death and long-term injury risks — but malnutrition plus diabesity can. (Undarked/Wells don’t use the term “diabesity,” but it’s in common use in medicine, and aptly describes the combined risks of gestational diabetes and maternal obesity — which double the risk of LGA infants.)
If Dunsworth’s is a story of (unscientific) science having constructed a misogynistic myth to explain why birth can be hard for humans — and demolishing that myth, then Wells’s might be seen as one of poverty combining with corporate corruption and intergenerational knowledge loss. These strains may not explain all the same outcomes, but are also not mutually exclusive. We can have our misogyny cake and eat our corporate corruption, too!
Dunsworth’s point is that “there’s not much science in [the] science” underpinning the old obstetrical dilemma, to invoke Sander Greenland. Wells’s is that obstructed labor is increasingly common and mortally risky, and that the misogyny story doesn’t explain this puzzle. But modern lifestyles with less movement and poorer-quality calories, creating dual malnutrition and diabesity epidemics since the late 1970s, may.
This story jives with the evidence on the Mediterranean diet and exercise preventing GD and promoting apparently healthy pregnancy weight gain. Does this suggest that, in the good old days, we knew better how to live and eat? Is Wells’s poverty/corporate corruption story also necessarily one about lost or disused cultural knowledge?
No, and it’s important that we recognize this. Because traditional cultural knowledge about healthy eating isn’t exactly infallible, either.
Does This Make Me Look Fat, Or Did We Dismantle the Patriarchy Already?
“The reduction of food intake during pregnancy is part of many cultural and religious traditions around the world,” observed Powell et al, reporting “Lack of head sparing following third-trimester caloric restriction among Tanzanian Maasai.” They studied 141 mother-infant pairs, quantifying dietary changes and neonatal measurements. They found moms reduced their calories substantially, from 1601 kcal/day in earlier pregnancy to 799 kcal/day in the third trimester — mostly from cutting carbs. And infants, in turn, measured low birth weight (<2500 g) nearly a third of the time, compared to 12% of urban Tanzanian newborns from the same time. They also displayed much smaller head circumferences than expected — a negative correlate of long-term health problems. (I remember reading in the eHRAF years ago that some Maasai restrict almost entirely to a broth diet in the home stretch of pregnancy, but couldn’t find the reference in my notes from when I had a free trial. Please send it to me if you happen to know.)
Outside of these cultural contexts, women in the rest of the world probably aren’t at risk of stunting their babies from going on a well-balanced diet during pregnancy. But low-carb diets in the first trimester may substantially cut women’s intake of folic acid intake in places where starchy stuff like flour tends to be fortified, increasing their pregnancies’ risk of developing potentially fatal neural tube defects. Supplementing mitigates this risk.
Low-carb diets are commonly recommended for women with gestational diabetes, and overweight/obese women are often counseled about low-glycemic index diets to lower their risk of delivering LGA babies. Still, these interventions tend to target later gestational ages than the first trimester. We don’t seem to have much evidence on their first trimester risks or benefits.
When I get pregnant, my body demands carbs, and I wouldn’t dare deny it. For one thing, maybe there is some signal in the hunger noise; other common pregnancy cravings include things like meat, which has better-absorbed heme iron the body could need. For another thing, if I didn’t eat carbs in early pregnancy, I would puke.
But society has long demanded things of women that make us sick. A generation ago, it used to be normal for modern Western physicians to counsel women to try to avoid or minimize weight gain in pregnancy as much as possible. As far as I can tell, no one has gone back and tried to estimate the prevalence of or possible harms associated with this norm. Any attempt to match such advice with adverse outcomes like low birthweight, SGA, or preterm babies would be plagued by recall and other biases anyway.
Point being, we don’t need to romanticize the past or traditional cultural knowledge, in order to use the best available evidence (which isn’t complete or perfect) to improve on the present. But we also don’t need to continue and modernize harmful sociocultural traditions that police women’s bodies, either. And that’s what gestational diabetes screening arguably does.
By targeting pregnant women instead of the whole population, it maintains longstanding traditions of surveilling and controlling what women eat during pregnancy — instead of addressing population-level metabolic health problems. By subjecting them to invasive mandatory screenings with extremely large numbers of false positives that are painfully difficult to accurately disambiguate from true positives, it creates a lot of pregnant women who are more worried about their health — but don’t provably have healthier pregnancies or babies.
The uncertainty of any net gain in this payoff is compounded by the fact that women not diagnosed with GD might also benefit from the diet and lifestyle education/intervention that is the safest and most effective available treatment. So the screening and diagnosis steps are superfluous. Mass preventive intervention is more efficient (not to mention more respectful and ethical), yielding more possible gains for fewer harms.
Targeting women with obvious metabolic problems for heightened surveillance and more medical intervention is a different matter — but, again, the evidentiary basis for available interventions here is not great. You really don’t want to be in the business of treating pregnant women with lifestyle diseases that could have been treated before pregnancy. You want to be in the business of improving population health before people have babies.
This is not, however, as good a literal business as billing insurance for as many patient visits in as little time as possible, or selling them drugs. Nor is it possible for healthcare professionals working with relevant populations to change the fact that society generally just doesn’t invest in population-level preventive interventions like it does in Big Pharma.
But it should be possible for them to think critically about what we really know. Do we really know that a number on a screening test or a scale means health or illness? Or could it rather be calorie quality and movement that causally driver better and worse outcomes for moms and babies?
What If Pregnancy Weight Gain Guidelines Are Wrong?
Pregnancy weight gain guidelines are insufficiently evidence-based on several fronts. Weight checks are perhaps the least evidence-based among prenatal mass screenings. For one thing, there is insufficient evidence on which to base them for underweight women (who may need to gain more). For another, there may be not only costs to being overweight in pregnancy — but also benefits. There may be a reason that women listening to their bodies’ hunger cues seem to normally gain “too much” baby weight. Maybe “too much” weight gain from quality calories in the context of an active lifestyle actually confer net benefits. Maybe, just maybe, women’s bodies are right, and the guidelines are wrong.
Take, for instance, the association between a nasty but common type of birth injury and heavier body mass index. Shalabna et al recently found lower rates of obstetric anal sphincter injuries (OASIS) “in obese (OR 0.2, 95 %CI 0.04–0.9) and in overweight (OR 0.3, 95 %CI 0.1–0.99) women.” In other words, obesity confers a 10-96% OASIS risk reduction; overweight may or may not reduce risk, with estimates ranging from 1-90% risk reduction.
So maybe women who gain “too much” pregnancy weight are at least in part responding to hunger cues based in part in some biological advantages that may accrue from being heavier, such as protection against some birth injuries and greater fat stores to draw on for lactation. Maybe the problem isn’t the weight per se, but other factors, like calorie quality and lifestyle. Maybe we just see “too much” gestational weight gain associated with some adverse pregnancy outcomes because it’s confounded with poor-quality diet and sedentary lifestyles.
Point being, overweight does not necessarily mean unhealthy, either for the general population or for pregnant women. Correlations between weight and metabolic/cardiovascular outcomes are imperfect. It could well be that the average modern Western pregnant woman gains “too much” weight because she eats when she’s hungry, and current guidelines wrongly pathologize that.
There are many lingering uncertainties about gestational weight gain. Most importantly for women, we don’t know enough to know what the moral of the story is here. It could be that, as women, we want to game our hunger cues, eating less earlier (unless we’re underweight) and more later (when it’s more the baby’s weight gain than our own). For instance, this might optimize babies’ neurodevelopmental outcomes by helping women go into labor later, in week 41. (Some evidence suggests so; other evidence says no.) Alternately, it could be that we would rather do it the other way around — eating more in response to hunger cues earlier to optimize nutritional status for developing brain and other structures, and less responsively to the same cues later, when more weight gain accrues to the baby — making larger babies with higher risks of problems like obstructed labor and shoulder dystocia.
Either way, such a focus on what women can do to manage risk by eating a certain way gives short shrift to the limitations of our control…
The Blame (the Woman) Game
The focus on pregnant women’s metabolic health and weight gain makes a number of assumptions about women’s power to control these things. But part of running the pregnancy program is having another person’s genetic code in your body, and having that code issue commands involved in growing a new human being. That code may have its own ideas.
For instance, Demetriou et al found smaller than gestational age newborns had decreased expression of IGF2, a paternally expressed growth hormone linked to in utero growth (press; publication). They situate that finding in the context of the parental conflict hypothesis that dad’s genes upvote bigger babies to promote his genome’s success, while mom’s genes downvote them to protect her through the risks of pregnancy and birth.
How does this square with the increasing birth weight trend? Could there be increasing selection for men whose genes make bigger babies, or an interplay of parental genes where the balance gets tipped more (possibly for immunological reasons) in his genome’s favor? It would seem counter-intuitive if higher standards of living in many respects caused women to prefer heftier, hardier males.
But it could be a by-product of women becoming choosier in their mates, increasingly opting out of marriage and motherhood, and/or delaying first-time motherhood to older and older ages. Women with greater nominal reproductive self-determination may suffer net reproductive losses compared to previous generations (having fewer lifetime children than desired), while also being able to select mates they deem to be better (having fewer children with men whose genes would have given them smaller babies). However, there’s no a priori reason to suspect that the men these choosier women deem to be better mates, are really giving them babies who are genetically programmed to be born bigger.
A simpler, but not mutually exclusive, explanation is that people are eating more and moving less, while food quality is increasingly heterogeneous and empty calories more readily available. So people are fatter. So women are fatter. They’re fatter before, during, and after pregnancy. In pregnancy, this spells greater metabolic/cardiovascular disease risks for mom and baby alike, and greater risks of birthing a big baby.
But still, it would be simplistic to say you are what you eat in pregnancy. You’re also your baby-daddy’s genes, your grandma’s diet, and (sometimes) a lot of bossy doctors. You can eat more or less. You can eat a special (low glycemic index, high EVOO and nuts, goitrogenic food-low Mediterranean) diet or not. But some things will still be beyond your control — with genetics and earlier environmental factors shaping pregnancy outcomes including gestational weight gain, whether women (and our doctors) like it, or not.
Turns out, how big your bump will get is just like a lot of other stuff in life: uncertain and subject to a lot of contingencies outside your control. Tell that to the stranger who says it must be twins.
Chance Has Many Faces
We are all here, as leading British statistician David Spiegelhalter recently wrote in The Guardian, because of confluences of highly improbable events. Our cognitive capabilities, metabolic health, and other facets of ourselves are influenced by a lot of factors. Maternal thyroid status is just one.
In struggling to get a handle on the situation and my distress over it, I was attracted to the gross oversimplification of saying mild prenatal hypothyroidism costs a kid four IQ points. The truth is that some children will be affected more. Some less. And with differences in exposure timing, severity, duration, and compensation through other factors like fetal thyroid function, we’ll miss heterogeneity in how they’re affected.
Science cannot say whether or how much my healthcare providers’ focus on mass prenatal screenings that I did not expect to benefit me individually, to the exclusion of targeted ones I needed repeated regularly, has harmed my unborn child. That didn’t stop my gyn from trying to reassure me that there was no evidence of damage.
This displays nullism — confusing no proof of A for proof of not-A. This is a common logical and statistical error in the medical and scientific literature that I’ve written about previously (see further readings here). Here as in other contexts, the inability to prove that harm occurred does not equate to proving that no harm occurred. In reality, we simply don’t know. Yet, this uncertainty is often waved away.
Healthcare providers may be disconnected from the downstream consequences of their actions. They may never see the children who, years later, are diagnosed with neurodevelopmental disorders linked to inadequate prenatal care, such as untreated hypothyroidism, or dangerous current guidelines like those promoting exclusive breastfeeding. It’s convenient—professionally, legally, and psychologically—to believe that current practices don’t cause harm simply because it hasn’t been definitively proven. But, as Altman and Bland observed, absence of evidence isn’t evidence of absence.
I wish what I wished when I realized I had accidentally starved my newborn son following bad “exclusive breastfeeding” guidelines and medical advice: That this child had the perfect mother he deserved — a woman who would not have not made this stupid mistake. A woman who would have known to insist on the right monitoring at the right time.
But there is no perfect mother. Everyone gets one of us. We all have our problems. We all make mistakes.
I try to reassure myself there are also a lot of places to pick up a few IQ points here and there: Prenatal iodine supplementation beginning no later than the first trimester, continuing folic acid or folate supplementation through the entire pregnancy (not just the first trimester as is often recommended), and extra choline supplementation in the third trimester all seem to give kids a cognitive boost, based on randomized trial and other evidence.
It’s obviously self-serving to reassure myself that no one is perfect, lots of variables matter, and kids are resilient. But then, taking care of yourself as a parent is one of those selfishnesses that enables all the other caring.
Dangerous Structure — Or Dangerous Distraction?
On one hand, this has been a massive distraction. I should’ve been working on something else in my increasingly limited time before a newborn needs round-the-clock care. But, like all of us, I am being divided and conquered by my own, relatively small and inconsequential life fires. Such distraction at a global scale threatens civilization with death by a thousand cuts.
On the other hand, I clearly wasn’t distracted enough — in the sense that I didn’t do enough work, early and often enough, to manage my own risks optimally for this pregnancy. I was too… Something. Busy moving? Trusting in healthcare providers instead of doing my own research and being my own advocate for the appropriate heightened, targeted screening? Sick and tired, but blind to the problem, assuming I was just pregnant, using what energy I had to give my threenager happy summertime memories?
Mostly, I thought I was doing well. But my unborn child deserved better. There is little that hurts more in life than failing your children.
We are frail monkeys who have to delegate trust and cooperate in order to survive. We are also stupid monkeys who make mistakes and struggle to recognize our own biases. And we are also reflective monkeys who like to think about these processes, if only to better understand why it is that the troop acts the way it does.
Why the affinity for standardization, quantification, reliance on metrics and checklist protocols — as if complex problem-solving involving critical thinking could be rote? How to counter the typical combination of bias and perverse incentives that magnifies this affinity? Where to pitch more evidence-based, harm prevention-oriented processes of risk assessment and mitigation that make (concentrated interests) less money but more (public interest) sense?
Maybe someone else had better pitch measured reforms while I make demands. Saying the quiet part out loud is more my strong suit. I demand a world where people have autonomy, including to refuse unnecessary testing (medical, security, education), eat (breathe, sleep, communicate) without being subjected to surveillance, and still get the help they need fulfilling basic needs (healthcare, housing, work). I demand a randomized trial comparing preventive lifestyle intervention à la SMILES (consisting of education and support for the Mediterranean diet modified for pregnancy, plus exercise) with the currently ascendant, diagnosis- and surveillance-oriented (autonomy-infringing, punitive) gestational diabetes regimes which produce huge numbers of false positives without proven net benefits.
I demand a doctor who can know that I read (and have a few of my own articles indexed in) PubMed. An impossible guarantee that my unborn baby’s brain was unharmed. A Youtube tutorial on the Mediterranean diet for pregnant women by Felice Jacka. And, when my blood sugar testing is done, a piece of lasagna.