What coronavirus modeling can teach us about breastfeeding modeling


As Dr. Anthony Fauci explained last week:

What we do is that every time we get more data, you feed it back in and relook at the model. Is the model really telling you what is actually going on?Models are as good as the assumptions you put into them, and as we get more data, then you put it in and that might change.

Consider a model that predicts the number of lives lost to COVID-19 in the US each day. According to the Washington Post article that quoted Dr. Fauci:

One bit of data added to the … models was the actual number of deaths over the past few days. The graph presented on Monday, for example, included an estimate that there would be about 850 deaths on April 1. Data … indicates that the number was closer to 890 — though even that figure includes some uncertainty. The model released on Wednesday night estimated there would be 899 deaths on April 1.

The accuracy of the model is validated by how closely it approximates what happens in the real world. Models make predictions and if those predictions fail to occur, the model is revised so it more closely complies with reality.

Mathematical models must be revised regularly by incorporating real world data.

That’s how mathematical modeling works. The model should be revised regularly based on real world information.

That’s why it’s a very serious problem that the mathematical modeling of the benefits of breastfeeding is NEVER revised to incorporate real world data.

For years breastfeeding researchers like Melissa Bartick, MD have constructed mathematical models to predict how we would decrease infant mortality, severe morbidity and healthcare expenditures if more women would breastfeeding. Back in 2010, Dr. Bartick published The Burden of Suboptimal Breastfeeding in the United States: A Pediatric Cost Analysis.

She claimed:

If 90% of US families could comply with medical recommendations to breastfeed exclusively for 6 months, the United States would save $13 billion per year and prevent an excess 911 deaths, nearly all of which would be in infants ($10.5 billion and 741 deaths at 80% compliance).

Bartick notes that the model was based on at least 14 separate assumptions about associations between breastfeeding and various conditions and illnesses. It is not unreasonable to make assumptions, but it is indefensible to refuse to revisit those assumptions when predictions are not confirmed by real world evidence.

For example, the latest coronavirus predictions of new cases and deaths have been recently revised downward based on the fact that cases and deaths are decreasing faster than the earlier models predicted. Earlier models, like all mathematical models, were based on assumptions. That’s okay as long as you revise those assumptions based on what actually happens.

But breastfeeding researchers NEVER revise their assumptions, even when they are demonstrably false.

In the past 10 years, Dr. Bartick and other breastfeeding researchers have been unable to show that even one term infant death has been prevented by increasing breastfeeding rates, let alone hundreds. They have been unable to show that even a single dollar has been saved by increased breastfeeding rates, let alone billions of dollars per year.

Did they revise their models?

No, they actually doubled down.

In September 2017, they released a “Breastfeeding Savings Calculator” based on the 2010 model. As far as I can determine, they didn’t revise a single aspect of that original model. This despite the fact that there were 7 years of data to work with, 7 years that had FAILED to produce the benefits the model predicted in 2010. That’s inexcusable.

Nearly every claim on the benefits of breastfeeding circulating today is still based on the 2010 model. It’s the equivalent of continuing to use the coronavirus projections from November 2019 even though they did not accurately capture what is currently happening in April 2020.

As Dr. Fauci noted, mathematical models must be continually revised so they accurately reflect what is actually happening. Breastfeeding researchers never revise their models; that’s why they are completely wrong.