Support for the infection cause of sudden infant death
A recent study supports the idea that infections contribute to sudden unexpected infant deaths (“Impact of COVID-19 Pandemic Interventions on Sudden Unexpected Death in Infancy Incidence in France,” Scherdel et al, J Pediatr, Oct. 2024):
We found a significant seasonal pattern of overall monthly SUDI [sudden unexpected death in infancy] incidence, with a peak observed periodically from November to February… The monthly SUDI incidence decreased significantly from the pre-pandemic to NPI [non-pharmaceutical interventions] periods (adjusted incidence rate ratio 0.83 [95% confidence interval .72-.96]). In particular, the monthly incidence of SUDI cases with a viral or bacterial identification decreased, while no significant difference was found for SUDI cases without a viral or bacterial identification.
The study is limited by its small sample size (750 pre-pandemic SUDI cases, 248 during NPI periods) and observational nature. Still, it makes sense that isolating and taking other protective measures when possible, when you have a newborn, would reduce infectious disease transmission — with sometimes life-or-death consequences. We don’t have strong social norms about protecting pregnant/postpartum women and newborns from infection. But there used to be lot of cross-cultural confinement practices that did just this.
***
Two items about mass screenings for low-prevalence problems in prenatal care
A recent letter to the editor claimed that confirmation of non-lethal fetal abnormalities from detailed first-trimester scans in the second trimester is <50%, with false-positives outnumbering true positives (“First-Trimester Ultrasound Screening in Routine Obstetric Practice,” Alessandro Ghidini, Obstetrics & Gynecology 144(4):p e105, October 2024). Informed consent about the risk of false-positive findings is lacking, and we don’t know how many women terminate pregnancies on the basis of anomalies detected on first-trimester scans that represent probable false positives. This fits the usual pattern of such programs, in which the common (baby is fine) overwhelms the rare (there’s a problem) — so under conditions of persistent inferential uncertainty and secondary screening/intervention harms, such programs risk massive societal net harm.
It’s not just false hits we have to worry about from these programs, but also misses. Maryland saw mother-to-infant HIV transmission increase from 0 cases in 2021 to 6 in 2022 (“Increase in Cases of Perinatal HIV Transmission in Maryland in 2022,” Griffith et al, Pediatrics, Oct. 2024). The prevention protocol includes testing all pregnant women, providing antiretroviral treatment (ART) to positive women and ART prophylaxis to their infants, and scheduling C-sections for moms with detectable virus at birth. This protocol successfully reduced vertical HIV transmission in the US to .9% in 2019. Until it didn’t. What happened?
A root cause analysis of the cases identified risk factors including delayed entry into perinatal and HIV care, premature birth, maternal adherence challenges in the setting of substance use and other adverse social determinants of health, and failure to diagnose maternal HIV infection in a timely way... To achieve and then sustain the elimination of perinatal HIV transmission, the constancy of systems that eliminate barriers for all pregnant people to access testing, prevention, and treatment is critical.
So it sounds like at least part of the problem was insufficient early testing access. This might point to interesting contrasts between the UK and the US. Though the numbers here are small, as these are low-prevalence problems (so all this comes with a big grain of salt), it appears that, while the US was losing ground in this battle, the UK was gaining…
In the UK, HIV transmission dropped in the wake of Covid; it dropped in 2020, the drop sustained in 2021, and the government planned to keep the momentum going. As part of its plan to end HIV transmission, in April 2022, the UK piloted opt-out emergency department testing for HIV, hepatitis B, and hepatitis C in high-HIV prevalence areas, in cases where people were already getting blood tests as part of their care. This may have played a role in the UK’s successful elimination of vertical transmission of hepatitis B.
In short, if delayed entry into prenatal care contributes to vertical transmission of HIV via delayed testing and diagnosis — i.e., people are not coming in for the kind of care you want to give them — then you test people where they are going for care. In the US, that’s likely to be urgent care centers for people with catastrophic health insurance, and emergency departments for people with Medicare, Medicaid, or no insurance. Following the UK’s opt-out testing example in these settings might help the US reverse the trend. But someone would have to pay for the testing… A state or federal reimbursement system in high-prevalence areas would be the obvious solution.
***
Anemia in pregnancy: diagnosis, treatment — and iatrogenic risks
The September/October issue of Contemporary OB/GYN features a thorough peer-reviewed article entitled “Ensuring successful treatment of anemia in pregnancy” by Jenny Mei, Elizabeta Nemeth, and Gabriela Dellapiana (p. 12-16). The article does not directly challenge the controversial August U.S. Preventive Services Task Force conclusion — that there is insufficient evidence to recommend routine iron deficiency screening and iron deficiency anemia supplementation in pregnancy. But it does highlight the CDC’s contrasting recommendation of universal iron supplementation in pregnancy, barring genetic disorders such as hemochromatosis.
The authors also explain why the problem is common (i.e., hemodilution of pregnancy, increased iron needs for making more blood and a baby, and deficient diets) — affecting around 30% of pregnant American women by the third trimester in one study. They argue that assessing and treating its underlying causes likely benefits both maternal and child health, concluding “Universal screening, workup, and treatment with iron supplementation is recommended for pregnant patients with anemia.”
This raises the question: How could practicing ob/gyns implement the article’s recommended universal screening for anemia causes without first conducting universal screening for anemia itself? The answer is, badly. Without routine anemia screening, only patients who bother spending the time and energy to come in and actively report symptoms like fatigue — which are often dismissed (by patients and healthcare practitioners alike) as common complaints — might receive screening and diagnosis. This means many anemic women likely go undiagnosed and untreated. Especially, ironically, those who are too tired to make a fuss about being tired.
So universal screening for anemia in pregnancy makes sense — assuming it’s safe to treat the most common cause, iron deficiency. However, there are two possible sorts of iatrogenic risks.
The first is unknown possible risks, such as risks of hypertensive disorders and hemorrhage. Available evidence does not establish whether iron supplementation in anemic pregnant women substantially decreases those risks, substantially increases them, or has no effect. As I wrote previously, studies reviewed in the recent U.S. Preventive Services Task Force report found some substantial possible risks:
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).
Although the compatability/confidence intervals here skew right, it’s not clear why improved iron status would increase women’s risk of hypertensive disorders or hemorrhage. And there could be no effect. We don’t know.
The second sort of iatrogenic risk is also unknown in the sense that the evidence for it is very limited, but the causal story is clearer and more plausible. Deena Mousa wrote recently in The Works in Progress Newsletter that treating iron-deficiency anemia in malaria-endemic areas, such as sub-Saharan Africa, may net harm both mothers and babies:
In 2003, researchers launched a randomized control trial to measure the health impact of iron and folic acid supplementation among preschool aged children on Pemba island in Zanzibar. The intervention was set up to measure how much health outcomes among children improved as a result of supplementation. However, the study was abruptly shut down when the intervention group, which received a daily iron dose, were 11 percent more likely to be hospitalized and 12 percent more likely to die from severe illness compared with those who did not receive treatment.
Mousa explains that a parallel study in Nepal found no harm from iron supplementation, malarial parasites depend on iron, and other research has found a protective effect of pediatric iron deficiency in Tanzania.
The public health solution in these areas is twofold: prevent malaria and treat anemia. But, if the risks of iron supplementation in malaria-endemic areas are real (evidence is limited), then the sequencing of interventions is critical to avoid unintended harm from well-meaning efforts. This is an example of why we must think about causality first and last in doing science.
***
Debate about fetal growth charts
A few recent publications join a running debate about using international versus customized fetal growth charts. You might think that classifications like “large for gestational age” (LGA) and “small for gestational age” (SGA) — which can affect pregnancy/birth/postpartum care quite a bit — would account for things like the baby’s gender (boys are bigger) and the mother’s height (bigger mom, bigger baby). But mostly they don’t, even though it looks like they’re more accurate when they do. Tall women seem to be especially affected.
This debate highlights a deeper problem with integrating what we know about race/ethnicity into medical care: It’s obviously stupid to use an international standard for fetal growth estimation, when different populations vary substantially in size. But it’s also increasingly taboo in many subcultures to recognize this, much less to integrate it into medical care. And, even when we do integrate it, at some point, someone gets left out of the evidence-based process. Because we can’t get the data we need to do it right on all groups all the time. We can just roughly do better fetal growth estimation corrected for race/ethnicity or some more politically acceptable proxy thereof (like nationality). Minorities will always risk falling through the cracks by falling outside majority norms.
***
Researchers minimize possible risks of paternal preconception metformin use
A few years ago, there was a media storm surrounding a paper that found “Preconception paternal metformin treatment is associated with major birth defects, particularly genital birth defects in boys,” as well as a slightly lower proportion of male offspring (“Preconception Antidiabetic Drugs in Men and Birth Defects in Offspring : A Nationwide Cohort Study,” Wensink et al, Ann Intern Med, May 2022, 175(5):665-673). This month, a new analysis claimed to repudiate these findings. The journal simultaneously published an accompanying editorial extolling its reassuring virtues.
But its authors misinterpreted their findings, overstepping the bounds of what the evidence shows. The way they did this promotes a story of pharmaceutical safety that aligns with the interests of powerful social and political networks — much like other researchers have used similar methods to draw overconfident conclusions with the same sociopolitical result in the context of denying possible risks of antidepressants in pregnancy.
What they claimed versus what they found
In “Paternal metformin use and risk of congenital malformations in offspring in Norway and Taiwan: population based, cross national cohort study,” Meng et al report results of a population-based cohort study using data from Norway and Taiwan, countries renowned for having good national health databases (BMJ 2024;387:e080127). They conclude:
The findings suggest that paternal use of metformin during the period of sperm development is not associated with congenital malformations in offspring, including organ specific malformations. Metformin can therefore continue to be considered a suitable initial oral agent for managing glucose levels in men with type 2 diabetes mellitus who plan on having children.
But their unadjusted analyses actually replicated prior findings that there may be cause for concern. They found:
Increased risks of any congenital malformation associated with paternal metformin use were observed in the unadjusted analysis and attenuated with increasing control of confounding. The relative risks of any malformations with paternal metformin use were 1.29 (95% confidence interval 1.07 to 1.55) in Norway and 1.08 (0.99 to 1.17) in Taiwan in the unadjusted analysis and 1.20 (0.94 to 1.53) and 0.93 (0.80 to 1.07), respectively, in the analysis restricted to fathers with type 2 diabetes mellitus. In the overlap propensity score weighting analysis restricted to fathers with type 2 diabetes mellitus, the relative risks were 0.98 (0.72 to 1.33) in Norway and 0.87 (0.74 to 1.02) in Taiwan, resulting in a pooled estimate of 0.89 (0.77 to 1.03).
In other words, the unadjusted Norwegian data analysis suggests a statistically significant increase in congenital malformations associated with paternal metformin use, with the estimated effect size compatible with 7-55% increased risk. This reflects a substantial possible risk increase. In the Taiwanese data, the unadjusted analysis suggests a possible 1% decrease to 17% risk increase.
Invalid subgroup restriction
Limiting the sample to fathers with type 2 diabetes, as the subsequent analyses do, lacks sound theoretical basis. The authors say metformin may decrease sperm quality in men with type 2 diabetes by lowering testosterone; but metformin’s testosterone-lowering effects aren’t limited to men with type 2 diabetes. If we don’t know that only this subgroup is affected by this posited causal mechanism of the possible harm at issue — and there’s no evidence cited that suggests that it is — then it doesn’t make sense to restrict the analysis to it. This is therefore an invalid subgroup analysis that should not have been conducted.
It’s also dangerous. Even subgroup analyses in large clinical trials risk misleading researchers by reducing statistical power while increasing “variance and the play of chance.” And this is a set of cohort studies, not a large clinical trial.
So we should ignore these invalid subgroup analyses. But even if we don’t, the unadjusted subgroup analysis also still finds evidence of substantial possible risk associated with preconception paternal metformin use. It estimates a possible 6% decrease to 53% increase in the Norwegian data, and a possible 20% decrease to 7% increase in the Taiwanese. These ranges show that the subgroup effect estimates contain substantial variability and uncertainty, especially around possible risk in Norway.
When it comes to differences between the two countries, the researchers didn’t ask obvious questions about possible systematic differences between these populations and these samples. Instead, their analyses have the effect of leveraging these differences to progressively disappear the substantial possible risk estimate their unrestricted, unadjusted Norwegian analysis found. After the invalid subgroup restriction, the next step in this disappearance was the use of propensity score weighting analysis — which still could not exclude a possible 33% risk increase in the Norwegian data. The final step was pooling the two countries’ estimates, which suggested a possible 23% risk decrease to 3% increase.
After multi-step statistical pyrotechnics incorporating invalid subgroup restriction at each stage, the effect basically flipped from one that troubled the acceptance of medical management of metabolic diseases, to one that supported it. The authors say this pattern of findings supports the idea that confounding, not causal association, explains the link between preconception paternal metformin use and congenital malformations. But their evidence does not establish this claim.
In further support of this claim, they performed sibling comparisons, which are also problematic…
Overconfidence in sibling comparisons
The authors write “Sibling matched comparisons were conducted to account for genetic and lifestyle factors.” But “All sibling pairs of each father were identified in the study, and only those pairs discordant for both metformin use and outcomes contributed to the estimated within pair association.”
This does not make sense: Fathers who used metformin before conceiving a child at one time and not at another are likely to be different in salient ways from fathers whose metformin use remained stable. Otherwise, they probably wouldn’t have changed the medication. In other words, dads who took metformin before one pregnancy and not before another may have also been different in other potentially relevant ways, such as diet, lifestyle, inflammatory status, and (obviously) parity — which could have numerous potentially relevant effects such as more processed food consumption due to less time for cooking.
As I wrote previously, these sorts of sibling comparison analyses overstate the method’s power by underappreciating the ways in which people change over time. They also often don’t account for whether siblings are full or half, casting doubt on claims that they account for genetic factors.
Lack of causal diagramming
The authors argue their analyses show confounding rather than causation drives the association between paternal preconception metformin use and congenital malformations. But they didn’t do the causal diagramming necessary to back up such a claim. Nor to defend against charges that, by correcting for confounds without thinking through causality first, they may have introduced more bias than they accounted for.
What the accompanying editorial says, and what it misses
The BMJ ran an editorial with the study claiming “Latest big study is reassuring for potential fathers and their partners.”
Nothing could be farther from the truth. This latest study actually replicates previous research suggesting substantial possible risks of malformations associated with preconception paternal metformin use. Just because its authors can make the effect disappear doing invalid subgroup analyses and adjusting for confounders, doesn’t mean it’s not there.
There are alternatives to beating the Big Pharma drum, and an editorial like this would seem to have been the place to highlight them. Diet and exercise can lead type 2 diabetes to remit. Such lifestyle changes have myriad other health benefits and no known risks — a superior profile to any drug regimen and one that a vast array of evidence suggests likely generalizes to offspring.
But it doesn’t pay to say that modern medicine is doing metabolic health wrong, from preventive medicine to diabetes management. Such a notion challenges consensus as well as powerful interests.
It would be interesting to see a randomized trial comparing a preconception diet and exercise intervention with metformin. But who would fund it? Randomized trials are expensive, and industry involvement in influential ones is the norm. Why would a pharmaceutical company that can pit its product against placebo and make billions instead try pitting it against a treatment with a better safety and possibly better efficacy profile that could generate results cutting into that profit?
They should do it for the children. But that’s not how power works.
***
Statistical significance testing misuse hides possible D&C harms
Authors of a recent study on cervical dilation and curettage (D&C) and future preterm delivery misinterpreted their results, wrongly claiming they show the procedure doesn’t increase this risk. In fact, consistent with earlier studies, their analysis suggested that it may increase this risk — in addition to increasing women’s future risk of delivering a small for gestational age (SGA) baby. Being born SGA carries numerous possible risks of additional complications, and as such is considered “the second major cause for perinatal diseases and death, following premature birth.”
The authors write:
Among the 1087 women meeting the inclusion criteria during the study period, 852 (78.4 %) underwent first-trimester curettage with cervical dilatation, while 235 (21.6 %) opted for curettage only. No significant maternal or neonatal different outcomes were noted between the study groups including preterm delivery (5.5 % vs. 3.5 %, p = 0.16), fertility treatments, placental complications, and mode of delivery. However, deliveries following D&C were associated with higher rates of small for gestational age neonates (7.6 % vs. 3.8 %, p = 0.04). Multivariate analysis revealed that cervical dilation before curettage was not significantly linked to preterm delivery [adjusted odds ratio 0.64 (0.33–1.26), p = 0.20].
Conclusion
The use of cervical dilatation during a curettage procedure for first trimester pregnancy loss, does not confer additional risk of preterm delivery. — “Impact of first-trimester mechanical cervical dilatation during curettage on maternal and neonatal outcomes: A retrospective comparative study,” Kalifa et al, Eur J Obstet Gynecol Reprod Biol, Sept. 2024, 300:1-5.
This study focused on women with pregnancy loss, but D&C (aka suction or surgical abortion) was also the most common form of abortion before medical or chemical abortion (aka mifepristone plus misoprostol)’s ascendance. So an important question is whether these results generalize to abortion, or if there’s relevant confounding — with the women opting for D&C systematically different from the women opting for curettage only, or from the women opting for abortion.
The larger point is that there is persistent uncertainty about possible fertility and other harms of abortion, about which abortion providers systematically misinform pregnant women. It’s an oft-repeated myth that we know that abortion doesn’t hurt women’s mental health or future fertility. There are substantial possible risks, but it’s basically impossible in the current, hyperpolarized discourse for anyone to credibly tell women that. This undermines informed consent, ironically leading women’s health professionals deeply invested in protecting women’s autonomy to jeopardize it.