Experts Have Many Reasons To Stop Publicly Discrediting Themselves

Right reasons and wrong reasons alike! That’s what makes this so weird. . .

Jerusalem Demsas wrote a really interesting, important article for The Atlantic about a maternal mortality crisis that appeared to be nonexistent.

The short version: for a while, a bunch of people were saying that the maternal mortality rate in the United States had increased a terrifying amount, in some cases doubling, between 1999 and 2019. This was seen as an awful indictment of the U.S. healthcare system, and especially its treatment of members of more vulnerable groups.

Except: this simply didn’t happen. It was all a misunderstanding. 

Even before we get to the argument I want to make in this piece, this is a really interesting science-gone-wrong story in its own right. Debunking credit goes to Saloni Dattani of Our World in Data (who is also worth following on Twitter).

As Demsas sums it up:

To address past concerns about underreporting maternal deaths, Dattani explains, states gradually updated their reporting standards to be more inclusive. The old definition of maternal mortality focused on deaths during childbirth or closely following birth; the new one expanded to include deaths during pregnancy or the first six weeks after the end of pregnancy. States also added a checkbox on death certificates indicating whether a woman had been pregnant at the time of death or within a year of her death. The reported maternal-mortality ratio on average doubled after the checkbox implementation. Because individual states changed their standards at different times over the course of a decade and a half, the national maternal-death count seemed to keep rising.

So while maternal mortality is always moving around, of course, and in some American contexts is higher than it should be in 2024, this idea of a terrifying upsurge was driven almost entirely by an artifact of data-collection techniques.