Better in their words…

VINAY PRASAD MD MPH AUG 10, 2026

For all the excitement about finding and detecting misinformation, it’s amazing the media does not recognize that many statistics it routinely touts are essentially that. 

• Closing USAID will cost 14-20 million lives
• Had red states adopted covid vaccines at the rate of blue states, we would have saved 200,000 lives
• Vaccine mandates saved 447,202 lives. 

These are all fake numbers. The people they trick are people who don’t know how these numbers are generated, and the readers of the paper who enter each article with a preconceived bias.

Take the Lancet USAID paper:

The paper first generates a relationship between USAID funding and improvements in mortality over time. This is nothing more than a crude observational association. No serious scientist would consider this to be causal. i.e that the association captures the effect of the funding versus not having the funding. 

Next, it extrapolates the crude relationship to estimate impact of funding cuts.

The authors don’t consider the effects of using the diverted money for different (perhaps wiser) purposes, and the downstream value of that. Alternatively, what if funding cuts lead to less taxes or less national debt— what are the downstream consequences of those choices— which may sway future political movements. The authors have no way to consider the impact of these alternatives. 

Furthermore, the authors don’t consider the fact that State department has taken over some the meritorious duties of USAID and that DOGE believes the funding cuts targeted dubious political causes and naked graft. 

Even basic details about USAID appear to be ignored. What proportion of the funding actually goes to an intervention versus overhead? And is the age adjusted mortality rate is measured consistently and accurately? In some countries, it may be poorly measured, or measure differently over time.

The number of arguable choices in a paper like this is on the order of thousands. If I had a series of meetings with the statistical modelers where they told me all of the things they were doing, assumptions they were making, and I pushed back on each one of them, I could imagine spending hundreds or thousands of hours of time, and still being no closer to an unbiased estimate. 

Using methods like this you could correlate the total number of Marvel movies with gains in global life expectancy. 

Scientists estimate that 4 new Marvel movies a year will save 3 million lives.

It isn’t science, but an exercise in imagination. 

The fundamental problem with these papers is no one knows whether a $1 in USAID spending saves 0.00001 lives or 0 lives or even -.00001 lives. 

No one knows if 1 more marvel movie lessens or worsens mortality— what if it draws people into air conditioned theaters during a European heat wave? Or if it removes a gang member from defending a street corner? One can create all sorts of domino theories of how it might help. 

No one knows if a covid vaccine reduces all cause mortality by 0.99999 HR or does nothing 1.0000. Observational data is useless because people who seek vaccination are different; here are the randomized data pooled. HR 1.03. How does that become 400,000 lives saved? (PS: I put no stock in this estimate either bc/ these data are threadbare— so thin, nothing should be concluded)

Some questions in life are answerable:

How many people would have lived if the Titanic didn’t hit any iceberg?

But there are many other questions in life where the answer is make believe.

How many people would have died had Truman NOT dropped 2 atomic bombs (how long would Japan have continued to fight)?

How many people would have died had Julius Caesar not been killed?

These are great questions for science fiction writers, but poor questions for scientists.

Imagine if the New York Times said this:

Scientists say had Julius Caesar lived there would be 499,242,123 more italians and 4,567,123 fewer Englishmen. 

You would be in disbelief.

Estimates like the USAID figure are also inherently biased by political winds. It is no secret the Lancet is a woke medical journal. (see the editorials) It favors the point of view of liberals. Liberals believe that closing USAID was bad, and that, had everyone done what St. Anthony Fauci said, 2 million lives would be saved. The papers begin with this conclusion and work backwards, creating stories that justify the headlines.

I worry that we are not far away from a major medical journal saying this:

• Scientists say 100,000,000 lives would be saved if everyone voted Democrat.
• A new Lancet paper shows that AOC defeating Pete Buttigieg in the primary will save 426,000 lives
• A new Science article shows that expanding the Supreme court will save 1,200,000 lives.
• Scientists say open borders save 2,372,000 lives a year

Believing made up numbers like this is not much better than believing there is a microchip inside a covid-19 vaccine, or that wearing a tin foil hat will ward off brain cancer. The only difference is the proponents are firmly inside the establishment, and proudly touting their figures.

It is a dangerous world when science pretends it has the answer to political questions.

You’re currently a free subscriber to Vinay Prasad’s Observations and Thoughts. For the full experience, upgrade your subscription.