Countering “breast is best” propaganda
Evidence suggests common and preventable harm to newborns results from accidental starvation under current “exclusive breastfeeding” guidance. This includes a substantial possible increased risk of permanent neurodevelopmental disorders such as autism associated with newborn jaundice. The solution is straightforward: supplementing with formula can prevent jaundice caused by underfeeding, and help treat it by speeding up bilirubin excretion. To avoid widespread harm, it’s crucial to feed infants with adequate supplemental milk until breastfeeding is well-established, especially in environments with access to clean water and formula.
Unfortunately, these crucial facts were missing from this year’s World Breastfeeding Week (Aug. 1-7) messaging.
In “Five Things Infant Formula Companies Don’t Want You to Know,” Matt Kristoffersen of Yale School of Public Health presented what he claimed were “science-based facts” that formula companies may prefer to hide. However, these so-called facts were themselves rooted in myths about the benefits of exclusive breastfeeding (e.g., that it has proven benefits that it does not), ignoring the potential risks.
Sue Haddon, a UK Infant Feeding Alliance reformer, debunked some of these myths on X/Twitter under the title “Five things breastfeeding advocates don’t want you to know.”
***
Debunking antidepressants myths
Investigative journalist Maryanne Demasi and leading medical methodologist Peter Gøtzsche recently co-authored an essay, “Debunking the debunkers on the ‘myths’ of antidepressants” (Maryanne Demasi, reports, Aug. 5). While their arguments are well-grounded in evidence, they also risk falling into some of the same traps as the mainstream public health messages they critique.
Promoting the essay, Gøtzsche wrote on X/Twitter:
Psychologist Natalina Salmaso claimed to “debunk” five myths that make people hesitant to take antidepressants. We showed that her article is full of omissions and falsehoods. None of the “myths” are myths. People should not take depression pills.
This is broad and proscriptive statement. It encompasses many medications. Many practicing psychiatrists, like Scott Alexander, offer different takes (i.e., multiplicities of treatments, presentations, preferences, and reactions mean different people may want to try different things). Sure, one could argue they have conflicts of interest (e.g., wanting to not believe they hurt people by prescribing these drugs).
Still, different people really may respond differently to various treatments, and what works for one person may not work for another. Narrower warnings would seem to be more defensible (e.g., people who want to make babies in the next quarter should probably not take SSRIs/SSNIs to prevent possible preconception/pregnancy harms). Admonitions to regulatory agencies would be more defensible still (e.g., fluozetine/ “Prozac” should probably never have been approved due to its limited efficacy data versus evidence of harm).
A more nuanced approach to communicating the relevant science to individuals considering taking antidepressants would involve providing information that allows them to make informed decisions based on the best available evidence, especially in light of their own circumstances, perspectives, and preferences. Instead of offering blanket advice, it’s more respectful (and probably more accurate) to help people understand the specifics and complexities of their options. It also better honors the limitations with which we all approach the (often ambiguous, incomplete, exegetical, conflict-fraught, and otherwise suboptimal) evidence as human beings.
But that’s just my take on the open question: Should outsiders (methodologists, investigative reporters, social critics, risk literacy consultants) contesting subject-area experts’ paternalistic prescriptions really be in the business of offering their own such prescriptions — instead of trying to ensure people have the information they need to make their own decisions? If part of the point of being an outsider is that you can see some biases in the dominant discourse, this sort of consensus-contester role should ideally be played with the knowledge that we’re all in the soup, we all have biases, and there is no authority without its own reliances on trust, vulnerability to interests and errors, and psychosocial perspectives. That may generally mean not replacing one simplistic rule with another, but rather, helping people make their own decisions based on better information — and better understandings of it.
Alternately, one could argue science and science communication have a duty to warn vulnerable people to not do obviously ineffective and potentially seriously harmful things, and taking antidepressants is one of those things. Just because it’s currently debated, doesn’t mean the evidence is really all that debatable. Rather, it could be debated because the (contested) consensus is wrong. In which case, we should say so.
***
Statistical significance testing misuse and possible harms/benefits from high-volume IV fluids in labor
“It is biologically plausible that a relationship between volume of IV fluids and primary postpartum hemorrhage (PPH) exists,” note Bruce et al in framing their 2021-22 retrospective cohort study on term, singleton pregnancies in Sydney, Australia (“Association between intravenous fluids during labor and primary postpartum hemorrhage: A retrospective cohort study,” Acta Obstet Gynecol Scand, Aug. 2024; 103(8):1541-1549). They found evidence of a possible such relationship, but dismissed it due to statistical significance testing misuse — raising familiar concerns about rampant such misuse.
Their analysis found an association between high-volume as compared to low-volume IV fluids and C-section (adjusted odds ratio 95% compatability/confidence interval [CI] 1.4-2.8). It also found an association between high-volume IV fluids and primary postpartum hemorrhage ≥500 mL (unadjusted odds ratio 95% CI 1.12-2.09), but the association dropped significance in the adjusted logistic regression (95% CI 0.72-1.44).
There are methodological reasons to be wary of adjusting away raw findings without thinking through causality first, as the authors appear to have done here. Anyway, interpretation of the full adjusted interval still suggests an up to 44% increase in PPH risk associated with high-volume as compared to low-volume IV fluids. However, it also includes an up to 28% risk decrease, and we don’t know if there is a real effect or not.
The point is not that high-volume IV fluids worsen birth outcomes. There’s a lot of research on IV fluids during labor, and much of it has produced conflicting evidence. But it’s worth noting there are a number of relevant randomized trials (so better-quality evidence than this cohort study). This recent review and meta-analysis of relevant RCTs, for example, found the opposite effect:
IV fluids at an infusion rate of 250 mL/h during labor in nulliparous women decreased the cesarean delivery rate, increased the vaginal delivery rate, shortened the first and second-stage labor duration, decreased the incidence of prolonged labor as compared with 125 mL/h. These findings suggest enhanced labor progression and a lower risk of labor complications with higher infusion rates.
Which raises the question: If there’s a bunch of higher-quality evidence like this already addressing the possibility that IV fluids in labor modulate risks including PPH, then why run a new cohort study on the question in the first place? (The usual answer to this kind of question is Altman’s: perverse incentives to “publish or perish” motivated scientists to do bad medical research. Unpaid reviewers and ignorant editors rubber-stamped what then became someone’s latest CV line.)
****
Measurement error in NICU vitamin supplementation
Recent research suggests a serious problem with vitamin supplementation of nasogastric tube-fed preterm infants. Researchers were concerned the premies might not be getting the breastmilk fat or vitamins with which they were being supplemented. Testing showed substantial percentages of supplemented vitamins indeed stuck to the tubing.
Feeding simulations and analysis with chromatography with mass spectrometry showed:
When vitamins were administered according to standard but without breastmilk, 100% of the vitamin D and 40% of the multivitamins were lost. The vitamins adhered to the material, mainly in the nasogastric tube… We suggest using direct oral administration whenever possible.
(“Unpredictable supplementation of vitamin D to infants in the neonatal intensive care unit: An experimental study,” Albinsson et al, Acta Paediatr, July 2024 e-publication ahead of print).
This finding could have broader implications in infant feeding research. For instance, Colaizy et al’s recent MILK trial (accidentally) found extremely preterm infants fed preterm infant formula had a survival advantage over those fed donor breastmilk. Both groups’ milk would have included fortifiers. But it’s possible the donor milk group’s vitamins stuck to the tubing, whereas the formula milk group’s stayed in solution, and the subsequent comparative insufficient nutrition contributed to the mortality difference. (Though it’s also possible the likely caloric difference alone is what drove the breastmilk-fed group to both gain less weight, and die more.)
At a minimum, this research suggests that more attention needs to be paid to how vitamins are administered in neonatal care to ensure that infants receive the full benefits of supplementation. More broadly, it raises questions about the “breast is best” ideology, which may sometimes obscure the reality that, in certain situations, formula can be the better option for infant health.
***
Measurement error in U.S. maternal mortality statistics
In March, the CDC announced 2021 U.S. maternal mortality rates rose based on data from the National Vital Statistics System, and claimed they weren’t just counting deaths to women who happen to be pregnant or recently pregnant:
Consistent with previous reports, the number of maternal deaths does not include all deaths occurring to pregnant or recently pregnant women, but only deaths with the underlying cause of death assigned to International Statistical Classification of Diseases, 10th Revision code numbers A34, O00–O95, and O98–O99.
The rise was large:
In 2021, 1,205 women died of maternal causes in the United States compared with 861 in 2020 and 754 in 2019 (2). The maternal mortality rate for 2021 was 32.9 deaths per 100,000 live births, compared with a rate of 23.8 in 2020 and 20.1 in 2019 (Table).
In April, Joseph et al charged:
The high and rising rates of maternal mortality in the United States are a consequence of changes in maternal mortality surveillance, with reliance on the pregnancy checkbox leading to an increase in misclassified maternal deaths.
Reanalyzing all US deaths from 1999-2021, they identified maternal deaths using two approaches:
(1) per National Vital Statistics System methodology, as deaths in pregnancy or in the postpartum period, including deaths identified solely because of a positive pregnancy checkbox, and (2) under an alternative formulation, as deaths in pregnancy or in the postpartum period, with at least 1 mention of pregnancy among the multiple causes of death on the death certificate.
The latter formulation suggested the maternal mortality rate actually fell (negligibly) from 10.2 in 1999-2002 to 10.4 per 100,000 live births in 2018-2021. Deaths from the big maternal mortality killers — preeclampsia/eclampsia, postpartum hemorrhage, puerperal sepsis, venous complications, and embolism — decreased. Deaths among black and Hispanic mothers decreased.
At the same time, “deaths due to adherent placenta, renal and unspecified causes, cardiomyopathy, and preexisting hypertension increased. Maternal mortality increased among non-Hispanic White women… rates were [still] disproportionately higher among non-Hispanic Black women, with large disparities evident in several causes of death (eg, cardiomyopathy).” As a whole, this makes it look like U.S. maternity care probably improved over the last 20 years, while white American reproductive-aged women’s health deteriorated and racial disparities persisted but lessened.
These findings challenged the narrative of a U.S. maternal mortality crisis, prompting debate about the accuracy of these statistics and the potential for measurement error to distort public health messaging. The challenge ruffled some feathers.
Christopher M. Zahn, MD, FACOG, interim CEO and chief of clinical practice and health equity and quality of the American College of Obstetricians and Gynecologists, issued a statement accordingly: “Despite new manuscript, incontrovertible evidence proves the unacceptably high U.S. maternal mortality rate” (March 13, 2024). Zahn defended both the pregnancy checkbox and the crisis narrative, arguing “To reduce the U.S. maternal mortality crisis to an ‘overestimation’ is irresponsible and minimizes the many lives lost and the families that have been deeply affected.”
A recent letter to the editor responded, praising Joseph et al and rebuking Zahn:
As we hoped and suspected, our pregnancy and labor and delivery care of women is making a difference and is at least as good as other developed countries… We should not shoot the messenger and dismiss good news just because it doesn’t fit a narrative (“Maternal mortality in the United States: finally some good news,” Amen Ness, AJOG, Vol. 231, No. 2, E95, Aug. 2024).
In May, science writer and chart maker Saloni Dattani agreed with Joseph et al’s findings. “The rise in reported maternal mortality rates in the US is largely due to a change in measurement,” she wrote. “Maternal mortality rates appear to have risen in the last 20 years in the US. But this reflects a change in measurement rather than an actual rise in mortality.”
Yesterday The Atlantic ran a piece treating Dattani’s assessment as definitive. It didn’t appear to have contacted Joseph et al or the CDC for comment.
The CDC has yet to correct its March announcement.
***
Contesting mammogram screenings propaganda
Mass screenings for low-prevalence problems, such as mammograms for breast cancer, are often promoted as essential. However, they carry serious risks that are frequently overlooked. A recent Guardian article encouraged women to stay up to date with such screenings without mentioning the limits of their evidence base or their potential harms.
From Sarah Phillips’ July 11, 2024 “The experts: oncologists on the simple, doable, everyday things they do to try to prevent cancer”:
7. Keep up to date with screenings
“I’ve tried to be good about being up to date with my screenings: cervical, breast and bowel screening – I absolutely welcome all that,” says Price. “Only about 65% of women invited for breast screening in England currently attend. We’ve all got busy lives; the last thing we want to think about is our symptoms or a screening test which might find something. But remember, the chances are that it is going to be absolutely fine. The NHS does thousands of mammograms every day. There are a very small number that are actually positive (about nine in 1,000 tests). If they find something, it will probably be tiny and really treatable and curable. In some countries, there are no screening programmes. We are really lucky to have them, and we should just take the tests when invited.”
Compare this approach — simplistic, proscriptive, one-size-fits-all — with the Harding Center’s fact box approach. Instead of telling them what to do, it givespeople the best available information about possible risks and benefits so they can decide for themselves whether to undergo such screenings. This approach shows around one breast cancer death was prevented for 1,000 women screened, with no proven all-cause or all-cancer mortality benefit — while 100 of 1,000 screened women had false positives often leading to risky interventions (including 5 with non-progressive breast cancer unnecessarily having partial or complete mastectomy).
Simplified messaging that pushes universal screening can lead to overdiagnosis, causing unnecessary anxiety and potentially harmful, unnecessary treatments. In this context, leading evidence-based medicine reformer and obstetrician Susan Bewley has led the UK charge to inform women about their choices and stop the madness. Mandy Payne for evidence-based UK patient advocacy group HealthSense wrote a detailed 2019 critique of the sweeping age extension study of mammograms in England that Bewley had also critiqued in BMJ. (Previously, I reviewed Welch et al’s Overdiagnosed linking this case study to many other medical context cases that share the same mathematical structure, and gave a CCC talk in December on how probability theory implies these sorts of programs backfire under conditions of rarity, uncertainty, and secondary screening harms.)
Conventional wisdom encouraging mammogram and similar sorts of screenings echoes conventional wisdom promoting “breast is best” for all, SSRIs as first-line depression treatments, and other consensus health positions. It promotes these interventions as having proven benefits they do not have, while ignoring their risks. It encourages simplistic, one-size-fits-all thinking where realities are more complex and individual situations vary.
***
Rethinking “back to sleep”
Sensible Medicine is running an ongoing series by economist Elizabeth Fama questioning the evidentiary basis of the anti-SIDS “back to sleep” (aka “safe to sleep”) guidance. Today’s post (2/4) suggests possible harms from practice. This looks like another possible example of contemporary myth-based medicine supported by strong, simplistic consensus messaging.
***
The anti-science norms of public health messaging
The health science discourse is full of simple messages. But science is full of complex observations, analyses, interpretations, and disagreements about them all. Structural incentives partly drive this disconnect: Institutions design policies to promote goals. But what if “Science can never provide insights that allow the imposition of a universal blueprint to all,” as Elisabeth Paul et al recently argued (“Who is anti-science”? Public Health in Practice, Vol. 7, June 2024, p. 3)? Are science and public health innately in conflict?
Public health messaging often discourages critical thinking and individualized decision-making. Whether it’s about infant feeding or infant sleeping, popping antidepressants, rejecting the possibility of medical abortion reversal, or submitting to routine mammogram screenings or Covid vaccinations, the communication norms are decidedly authoritarian. In all of these cases, evidence is insufficient to establish the current consensus, and suggests it may be doing considerable preventable harm.
By contrast, though, there are cases where simple, directive public health messaging arguably makes sense. Infectious diseases physician Paul Offit and others make the case for measles vaccine mandates. HIV screening in all pregnant women, even in low-prevalence settings, also benefits more than it costs, since uncertainty is resolvable with minimal secondary screening harms, and preventing mother-to-child transmission is a big deal (TM).
These sorts of cases make me wonder whether Paul et al may have gone too far in proposing that “by trying to reduce ‘science’ to simple messages set in stone, scientists can become the worst enemies of science” (p. 1). Sometimes, science can be distilled into clear, actionable directives.
Yet, outside these sorts of cases lies a vast expanse of contested, complex, and unknown terrain that demands more open discussion governed by different norms—norms that prioritize autonomous decision-making over conformity. Democratic norms.
The challenge for science and science communication is that adhering to these higher standards often conflicts with institutional and professional interests as well as social norms. As Paul et al wrote:
… science is not about miracles, and scientific evidence of efficacy does not equate to effectiveness and broader impacts. Science is about questioning facts and depicting the whole complexity of the natural and social worlds (p. 2).
People want answers. We have questions.